repository_name
stringclasses
316 values
func_path_in_repository
stringlengths
6
223
func_name
stringlengths
1
134
language
stringclasses
1 value
func_code_string
stringlengths
57
65.5k
func_documentation_string
stringlengths
1
46.3k
split_name
stringclasses
1 value
func_code_url
stringlengths
91
315
called_functions
listlengths
1
156
enclosing_scope
stringlengths
2
1.48M
scivision/gridaurora
gridaurora/__init__.py
chapman_profile
python
def chapman_profile(Z0: float, zKM: np.ndarray, H: float): return np.exp(.5*(1-(zKM-Z0)/H - np.exp((Z0-zKM)/H)))
Z0: altitude [km] of intensity peak zKM: altitude grid [km] H: scale height [km] example: pz = chapman_profile(110,np.arange(90,200,1),20)
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/__init__.py#L64-L73
null
from datetime import datetime, date from dateutil.parser import parse import numpy as np import logging from typing import Union def toyearmon(time: datetime) -> int: # %% date handle if isinstance(time, (tuple, list, np.ndarray)): logging.warning(f'taking only first time {time[0]}, would you like multiple times upgrade to code?') time = time[0] if isinstance(time, str): time = parse(time) elif isinstance(time, np.datetime64): time = time.astype(datetime) elif isinstance(time, (datetime, date)): pass else: raise TypeError(f'not sure what to do with type {type(time)}') ym = int(f'{time.year:d}{time.month:02d}') return ym def to_ut1unix(time: Union[str, datetime, float, np.ndarray]) -> np.ndarray: """ converts time inputs to UT1 seconds since Unix epoch """ # keep this order time = totime(time) if isinstance(time, (float, int)): return time if isinstance(time, (tuple, list, np.ndarray)): assert isinstance(time[0], datetime), f'expected datetime, not {type(time[0])}' return np.array(list(map(dt2ut1, time))) else: assert isinstance(time, datetime) return dt2ut1(time) def dt2ut1(t: datetime) -> float: epoch = datetime(1970, 1, 1) assert isinstance(t, datetime) return (t-epoch).total_seconds() def totime(time: Union[str, datetime, np.datetime64]) -> np.ndarray: time = np.atleast_1d(time) if isinstance(time[0], (datetime, np.datetime64)): pass elif isinstance(time[0], str): time = np.atleast_1d(list(map(parse, time))) return time.squeeze()[()]
scivision/gridaurora
gridaurora/calcemissions.py
calcemissions
python
def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br
Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L20-L59
[ "def sortelimlambda(lamb, ver, br):\n assert lamb.ndim == 1\n assert lamb.size == ver.shape[-1]\n# %% eliminate unused wavelengths and Einstein coeff\n mask = np.isfinite(lamb)\n ver = ver[..., mask]\n lamb = lamb[mask]\n br = br[:, mask]\n# %% sort by lambda\n lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic\n\n return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order\n", "def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path):\n with h5py.File(reactfn, 'r') as f:\n A = f['/metastable/A'][:]\n lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D!\n\n \"\"\"\n concatenate along the reaction dimension, axis=-1\n \"\"\"\n vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None],\n A[2:4] * rates.loc[..., 'no1d'].values[:, None],\n A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1)\n\n assert vnew.shape == (rates.shape[0], A.size)\n\n return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br)\n", "def getAtomic(rates, ver, lamb, br, reactfn):\n \"\"\" prompt atomic emissions (nm)\n 844.6 777.4\n \"\"\"\n with h5py.File(reactfn, 'r') as f:\n lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D!\n\n vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None],\n rates.loc[..., 'po3p5p'].values[..., None]), axis=-1)\n\n return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br)\n", "def getN21NG(rates, ver, lamb, br, reactfn):\n \"\"\"\n excitation Franck-Condon factors (derived from Vallance Jones, 1974)\n \"\"\"\n with h5py.File(str(reactfn), 'r', libver='latest') as f:\n A = f['/N2+1NG/A'].value\n lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F')\n franckcondon = f['/N2+1NG/fc'].value\n\n return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br)\n", "def getN2meinel(rates, ver, lamb, br, reactfn):\n with h5py.File(str(reactfn), 'r', libver='latest') as f:\n A = f['/N2+Meinel/A'].value\n lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F')\n franckcondon = f['/N2+Meinel/fc'].value\n # normalize\n franckcondon = franckcondon/franckcondon.sum() # special to this case\n\n return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br)\n", "def getN22PG(rates, ver, lamb, br, reactfn):\n \"\"\" from Benesch et al, 1966a \"\"\"\n with h5py.File(str(reactfn), 'r', libver='latest') as f:\n A = f['/N2_2PG/A'].value\n lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F')\n franckcondon = f['/N2_2PG/fc'].value\n\n return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br)\n", "def getN21PG(rates, ver, lamb, br, reactfn):\n\n with h5py.File(str(reactfn), 'r', libver='latest') as fid:\n A = fid['/N2_1PG/A'].value\n lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F')\n franckcondon = fid['/N2_1PG/fc'].value\n\n tau1PG = 1 / np.nansum(A, axis=1)\n \"\"\"\n solve for base concentration\n confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12\n Gattinger and Vallance Jones 1974\n confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001])\n \"\"\"\n\n consfac = franckcondon/franckcondon.sum() # normalize\n losscoef = (consfac / tau1PG).sum()\n N01pg = rates.loc[..., 'p1pg'] / losscoef\n\n scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab)\n\n vnew = scalevec[None, None, :] * N01pg.values[..., None]\n\n return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br)\n" ]
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! """ concatenate along the reaction dimension, axis=-1 """ vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getAtomic(rates, ver, lamb, br, reactfn): """ prompt atomic emissions (nm) 844.6 777.4 """ with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN21NG(rates, ver, lamb, br, reactfn): """ excitation Franck-Condon factors (derived from Vallance Jones, 1974) """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) """ solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001]) """ consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): """ ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu' """ tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br) def catvl(z, ver, vnew, lamb, lambnew, br): """ trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...] """ if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/calcemissions.py
getMetastable
python
def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br)
concatenate along the reaction dimension, axis=-1
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L62-L76
[ "def catvl(z, ver, vnew, lamb, lambnew, br):\n \"\"\"\n trapz integrates over altitude axis, axis = -2\n concatenate over reaction dimension, axis = -1\n\n br: column integrated brightness\n lamb: wavelength [nm]\n ver: volume emission rate [photons / cm^-3 s^-3 ...]\n \"\"\"\n if ver is not None:\n br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first!\n ver = np.concatenate((ver, vnew), axis=-1)\n lamb = np.concatenate((lamb, lambnew))\n else:\n ver = vnew.copy(order='F')\n lamb = lambnew.copy()\n br = np.trapz(ver, z, axis=-2)\n\n return ver, lamb, br\n" ]
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None """ Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php """ # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br def getAtomic(rates, ver, lamb, br, reactfn): """ prompt atomic emissions (nm) 844.6 777.4 """ with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN21NG(rates, ver, lamb, br, reactfn): """ excitation Franck-Condon factors (derived from Vallance Jones, 1974) """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) """ solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001]) """ consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): """ ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu' """ tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br) def catvl(z, ver, vnew, lamb, lambnew, br): """ trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...] """ if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/calcemissions.py
getAtomic
python
def getAtomic(rates, ver, lamb, br, reactfn): with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br)
prompt atomic emissions (nm) 844.6 777.4
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L79-L89
[ "def catvl(z, ver, vnew, lamb, lambnew, br):\n \"\"\"\n trapz integrates over altitude axis, axis = -2\n concatenate over reaction dimension, axis = -1\n\n br: column integrated brightness\n lamb: wavelength [nm]\n ver: volume emission rate [photons / cm^-3 s^-3 ...]\n \"\"\"\n if ver is not None:\n br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first!\n ver = np.concatenate((ver, vnew), axis=-1)\n lamb = np.concatenate((lamb, lambnew))\n else:\n ver = vnew.copy(order='F')\n lamb = lambnew.copy()\n br = np.trapz(ver, z, axis=-2)\n\n return ver, lamb, br\n" ]
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None """ Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php """ # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! """ concatenate along the reaction dimension, axis=-1 """ vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN21NG(rates, ver, lamb, br, reactfn): """ excitation Franck-Condon factors (derived from Vallance Jones, 1974) """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) """ solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001]) """ consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): """ ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu' """ tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br) def catvl(z, ver, vnew, lamb, lambnew, br): """ trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...] """ if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/calcemissions.py
getN21NG
python
def getN21NG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br)
excitation Franck-Condon factors (derived from Vallance Jones, 1974)
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L92-L101
[ "def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br):\n \"\"\"\n ver dimensions: wavelength, altitude, time\n\n A and lambda dimensions:\n axis 0 is upper state vib. level (nu')\n axis 1 is bottom state vib level (nu'')\n there is a Franck-Condon parameter (variable fc) for each upper state nu'\n \"\"\"\n tau = 1/np.nansum(Aein, axis=1)\n\n scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F')\n\n vnew = scalevec[None, None, :]*kin.values[..., None]\n\n return catvl(z, ver, vnew, lamb, lambnew, br)\n" ]
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None """ Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php """ # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! """ concatenate along the reaction dimension, axis=-1 """ vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getAtomic(rates, ver, lamb, br, reactfn): """ prompt atomic emissions (nm) 844.6 777.4 """ with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) """ solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001]) """ consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): """ ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu' """ tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br) def catvl(z, ver, vnew, lamb, lambnew, br): """ trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...] """ if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/calcemissions.py
getN21PG
python
def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br)
solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001])
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L125-L148
[ "def catvl(z, ver, vnew, lamb, lambnew, br):\n \"\"\"\n trapz integrates over altitude axis, axis = -2\n concatenate over reaction dimension, axis = -1\n\n br: column integrated brightness\n lamb: wavelength [nm]\n ver: volume emission rate [photons / cm^-3 s^-3 ...]\n \"\"\"\n if ver is not None:\n br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first!\n ver = np.concatenate((ver, vnew), axis=-1)\n lamb = np.concatenate((lamb, lambnew))\n else:\n ver = vnew.copy(order='F')\n lamb = lambnew.copy()\n br = np.trapz(ver, z, axis=-2)\n\n return ver, lamb, br\n" ]
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None """ Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php """ # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! """ concatenate along the reaction dimension, axis=-1 """ vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getAtomic(rates, ver, lamb, br, reactfn): """ prompt atomic emissions (nm) 844.6 777.4 """ with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN21NG(rates, ver, lamb, br, reactfn): """ excitation Franck-Condon factors (derived from Vallance Jones, 1974) """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): """ ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu' """ tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br) def catvl(z, ver, vnew, lamb, lambnew, br): """ trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...] """ if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/calcemissions.py
doBandTrapz
python
def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br)
ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu'
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L151-L166
[ "def catvl(z, ver, vnew, lamb, lambnew, br):\n \"\"\"\n trapz integrates over altitude axis, axis = -2\n concatenate over reaction dimension, axis = -1\n\n br: column integrated brightness\n lamb: wavelength [nm]\n ver: volume emission rate [photons / cm^-3 s^-3 ...]\n \"\"\"\n if ver is not None:\n br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first!\n ver = np.concatenate((ver, vnew), axis=-1)\n lamb = np.concatenate((lamb, lambnew))\n else:\n ver = vnew.copy(order='F')\n lamb = lambnew.copy()\n br = np.trapz(ver, z, axis=-2)\n\n return ver, lamb, br\n" ]
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None """ Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php """ # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! """ concatenate along the reaction dimension, axis=-1 """ vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getAtomic(rates, ver, lamb, br, reactfn): """ prompt atomic emissions (nm) 844.6 777.4 """ with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN21NG(rates, ver, lamb, br, reactfn): """ excitation Franck-Condon factors (derived from Vallance Jones, 1974) """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) """ solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001]) """ consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def catvl(z, ver, vnew, lamb, lambnew, br): """ trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...] """ if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/calcemissions.py
catvl
python
def catvl(z, ver, vnew, lamb, lambnew, br): if ver is not None: br = np.concatenate((br, np.trapz(vnew, z, axis=-2)), axis=-1) # must come first! ver = np.concatenate((ver, vnew), axis=-1) lamb = np.concatenate((lamb, lambnew)) else: ver = vnew.copy(order='F') lamb = lambnew.copy() br = np.trapz(ver, z, axis=-2) return ver, lamb, br
trapz integrates over altitude axis, axis = -2 concatenate over reaction dimension, axis = -1 br: column integrated brightness lamb: wavelength [nm] ver: volume emission rate [photons / cm^-3 s^-3 ...]
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/calcemissions.py#L169-L187
null
#!/usr/bin/env python from pathlib import Path import numpy as np import h5py from typing import Tuple import xarray """ inputs: spec: excitation rates, 3-D , dimensions time x altitude x reaction output: ver: a pandas DataFrame, wavelength x altitude br: flux-tube integrated intensity, dimension lamb See Eqn 9 of Appendix C of Zettergren PhD thesis 2007 to get a better insight on what this set of functions do. """ def calcemissions(rates: xarray.DataArray, sim) -> Tuple[xarray.DataArray, np.ndarray, np.ndarray]: if not sim.reacreq: return 0., 0., 0. ver = None lamb = None br = None """ Franck-Condon factor http://chemistry.illinoisstate.edu/standard/che460/handouts/460-Feb28lec-S13.pdf http://assign3.chem.usyd.edu.au/spectroscopy/index.php """ # %% METASTABLE if 'metastable' in sim.reacreq: ver, lamb, br = getMetastable(rates, ver, lamb, br, sim.reactionfn) # %% PROMPT ATOMIC OXYGEN EMISSIONS if 'atomic' in sim.reacreq: ver, lamb, br = getAtomic(rates, ver, lamb, br, sim.reactionfn) # %% N2 1N EMISSIONS if 'n21ng' in sim.reacreq: ver, lamb, br = getN21NG(rates, ver, lamb, br, sim.reactionfn) # %% N2+ Meinel band if 'n2meinel' in sim.reacreq: ver, lamb, br = getN2meinel(rates, ver, lamb, br, sim.reactionfn) # %% N2 2P (after Vallance Jones, 1974) if 'n22pg' in sim.reacreq: ver, lamb, br = getN22PG(rates, ver, lamb, br, sim.reactionfn) # %% N2 1P if 'n21pg' in sim.reacreq: ver, lamb, br = getN21PG(rates, ver, lamb, br, sim.reactionfn) # %% Remove NaN wavelength entries if ver is None: raise ValueError('you have not selected any reactions to generate VER') # %% sort by wavelength, eliminate NaN lamb, ver, br = sortelimlambda(lamb, ver, br) # %% assemble output dfver = xarray.DataArray(data=ver, coords=[('alt_km', rates.alt_km), ('wavelength_nm', lamb)]) return dfver, ver, br def getMetastable(rates, ver: np.ndarray, lamb, br, reactfn: Path): with h5py.File(reactfn, 'r') as f: A = f['/metastable/A'][:] lambnew = f['/metastable/lambda'].value.ravel(order='F') # some are not 1-D! """ concatenate along the reaction dimension, axis=-1 """ vnew = np.concatenate((A[:2] * rates.loc[..., 'no1s'].values[:, None], A[2:4] * rates.loc[..., 'no1d'].values[:, None], A[4:] * rates.loc[..., 'noii2p'].values[:, None]), axis=-1) assert vnew.shape == (rates.shape[0], A.size) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getAtomic(rates, ver, lamb, br, reactfn): """ prompt atomic emissions (nm) 844.6 777.4 """ with h5py.File(reactfn, 'r') as f: lambnew = f['/atomic/lambda'].value.ravel(order='F') # some are not 1-D! vnew = np.concatenate((rates.loc[..., 'po3p3p'].values[..., None], rates.loc[..., 'po3p5p'].values[..., None]), axis=-1) return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def getN21NG(rates, ver, lamb, br, reactfn): """ excitation Franck-Condon factors (derived from Vallance Jones, 1974) """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+1NG/A'].value lambdaA = f['/N2+1NG/lambda'].value.ravel(order='F') franckcondon = f['/N2+1NG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p1ng'], lamb, ver, rates.alt_km, br) def getN2meinel(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2+Meinel/A'].value lambdaA = f['/N2+Meinel/lambda'].value.ravel(order='F') franckcondon = f['/N2+Meinel/fc'].value # normalize franckcondon = franckcondon/franckcondon.sum() # special to this case return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'pmein'], lamb, ver, rates.alt_km, br) def getN22PG(rates, ver, lamb, br, reactfn): """ from Benesch et al, 1966a """ with h5py.File(str(reactfn), 'r', libver='latest') as f: A = f['/N2_2PG/A'].value lambdaA = f['/N2_2PG/lambda'].value.ravel(order='F') franckcondon = f['/N2_2PG/fc'].value return doBandTrapz(A, lambdaA, franckcondon, rates.loc[..., 'p2pg'], lamb, ver, rates.alt_km, br) def getN21PG(rates, ver, lamb, br, reactfn): with h5py.File(str(reactfn), 'r', libver='latest') as fid: A = fid['/N2_1PG/A'].value lambnew = fid['/N2_1PG/lambda'].value.ravel(order='F') franckcondon = fid['/N2_1PG/fc'].value tau1PG = 1 / np.nansum(A, axis=1) """ solve for base concentration confac=[1.66;1.56;1.31;1.07;.77;.5;.33;.17;.08;.04;.02;.004;.001]; %Cartwright, 1973b, stop at nuprime==12 Gattinger and Vallance Jones 1974 confac=array([1.66,1.86,1.57,1.07,.76,.45,.25,.14,.07,.03,.01,.004,.001]) """ consfac = franckcondon/franckcondon.sum() # normalize losscoef = (consfac / tau1PG).sum() N01pg = rates.loc[..., 'p1pg'] / losscoef scalevec = (A * consfac[:, None]).ravel(order='F') # for clarity (verified with matlab) vnew = scalevec[None, None, :] * N01pg.values[..., None] return catvl(rates.alt_km, ver, vnew, lamb, lambnew, br) def doBandTrapz(Aein, lambnew, fc, kin, lamb, ver, z, br): """ ver dimensions: wavelength, altitude, time A and lambda dimensions: axis 0 is upper state vib. level (nu') axis 1 is bottom state vib level (nu'') there is a Franck-Condon parameter (variable fc) for each upper state nu' """ tau = 1/np.nansum(Aein, axis=1) scalevec = (Aein * tau[:, None] * fc[:, None]).ravel(order='F') vnew = scalevec[None, None, :]*kin.values[..., None] return catvl(z, ver, vnew, lamb, lambnew, br) def sortelimlambda(lamb, ver, br): assert lamb.ndim == 1 assert lamb.size == ver.shape[-1] # %% eliminate unused wavelengths and Einstein coeff mask = np.isfinite(lamb) ver = ver[..., mask] lamb = lamb[mask] br = br[:, mask] # %% sort by lambda lambSortInd = lamb.argsort() # lamb is made piecemeal and is overall non-monotonic return lamb[lambSortInd], ver[..., lambSortInd], br[:, lambSortInd] # sort by wavelength ascending order
scivision/gridaurora
gridaurora/solarangle.py
solarzenithangle
python
def solarzenithangle(time: datetime, glat: float, glon: float, alt_m: float) -> tuple: time = totime(time) obs = EarthLocation(lat=glat*u.deg, lon=glon*u.deg, height=alt_m*u.m) times = Time(time, scale='ut1') sun = get_sun(times) sunobs = sun.transform_to(AltAz(obstime=times, location=obs)) return 90 - sunobs.alt.degree, sun, sunobs
Input: t: scalar or array of datetime
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/solarangle.py#L8-L21
[ "def totime(time: Union[str, datetime, np.datetime64]) -> np.ndarray:\n time = np.atleast_1d(time)\n\n if isinstance(time[0], (datetime, np.datetime64)):\n pass\n elif isinstance(time[0], str):\n time = np.atleast_1d(list(map(parse, time)))\n\n return time.squeeze()[()]\n" ]
from datetime import datetime import astropy.units as u from astropy.coordinates import get_sun, EarthLocation, AltAz from astropy.time import Time from . import totime
scivision/gridaurora
gridaurora/eFluxGen.py
maxwellian
python
def maxwellian(E: np.ndarray, E0: np.ndarray, Q0: np.ndarray) -> Tuple[np.ndarray, float]: E0 = np.atleast_1d(E0) Q0 = np.atleast_1d(Q0) assert E0.ndim == Q0.ndim == 1 assert (Q0.size == 1 or Q0.size == E0.size) Phi = Q0/(2*pi*E0**3) * E[:, None] * np.exp(-E[:, None]/E0) Q = np.trapz(Phi, E, axis=0) logging.info('total maxwellian flux Q: ' + (' '.join('{:.1e}'.format(q) for q in Q))) return Phi, Q
input: ------ E: 1-D vector of energy bins [eV] E0: characteristic energy (scalar or vector) [eV] Q0: flux coefficient (scalar or vector) (to yield overall flux Q) output: ------- Phi: differential number flux Q: total flux Tanaka 2006 Eqn. 1 http://odin.gi.alaska.edu/lumm/Papers/Tanaka_2006JA011744.pdf
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/eFluxGen.py#L14-L39
null
""" Michael Hirsch based on Strickland 1993 """ import logging from pathlib import Path import numpy as np import h5py from typing import Tuple pi = np.pi def fluxgen(E, E0, Q0, Wbc, bl, bm, bh, Bm, Bhf, verbose: int = 0) -> tuple: Wb = Wbc*E0 isimE0 = abs(E - E0).argmin() base = gaussflux(E, Wb, E0, Q0) diffnumflux = base.copy() low = letail(E, E0, Q0, bl, verbose) diffnumflux += low # intermediate result mid = midtail(E, E0, bm, Bm) diffnumflux += mid # intermediate result hi = hitail(E, diffnumflux, isimE0, E0, Bhf, bh, verbose) diffnumflux += hi if verbose > 0: diprat(E0, diffnumflux, isimE0) Q = np.trapz(diffnumflux, E, axis=0) if verbose > 0: print('total flux Q: ' + (' '.join('{:.1e}'.format(q) for q in Q))) return np.asfortranarray(diffnumflux), low, mid, hi, base, Q def letail(E: np.ndarray, E0: float, Q0: float, bl: float, verbose: int = 0) -> np.ndarray: # for LET, 1<b<2 # Bl = 8200. #820 (typo?) Bl = 0.4*Q0/(2*pi*E0**2) * np.exp(-1) # bl = 1.0 #1 low = Bl * (E[:, None]/E0)**-bl low[E[:, None] > E0] = 0. if verbose > 0: print('Bl: ' + (' '.join('{:0.1f}'.format(b) for b in Bl))) return low def midtail(E: np.ndarray, E0: np.ndarray, bm: float, Bm: float): # Bm = 1.8e4 #1.8e4 # bm = 3. #3 mid = Bm*(E[:, None]/E0)**bm mid[E[:, None] > E0] = 0. return mid def hitail(E: np.ndarray, diffnumflux: np.ndarray, isimE0: np.ndarray, E0: np.ndarray, Bhf: np.ndarray, bh: float, verbose: int = 0): """ strickland 1993 said 0.2, but 0.145 gives better match to peak flux at 2500 = E0 """ Bh = np.empty_like(E0) for iE0 in np.arange(E0.size): Bh[iE0] = Bhf[iE0]*diffnumflux[isimE0[iE0], iE0] # 4100. # bh = 4 #2.9 het = Bh*(E[:, None] / E0)**-bh het[E[:, None] < E0] = 0. if verbose > 0: print('Bh: ' + (' '.join('{:0.1f}'.format(b) for b in Bh))) return het def diprat(E0: np.ndarray, arc: np.ndarray, isimE0: np.ndarray): dipratio = np.empty_like(E0) for iE0 in np.arange(E0.size): idip = arc[:isimE0[iE0], iE0].argmin(axis=0) dipratio[iE0] = arc[idip, iE0]/arc[isimE0[iE0], iE0] print('dipratio: ' + (' '.join(f'{d:0.2f}' for d in dipratio))) # if not all(0.2<dipratio<0.5): # warn('dipratio outside of 0.2<dipratio<0.5') def gaussflux(E, Wb, E0, Q0): Qc = Q0/(pi**(3/2) * Wb*E0) return Qc * np.exp(-((E[:, None]-E0) / Wb)**2) def writeh5(h5fn: Path, Phi: np.ndarray, E, fp): if h5fn: with h5py.File(h5fn, 'w') as f: f.create_dataset('/diffnumflux', data=Phi) hE = f.create_dataset('/E', data=E) hE.attrs['Units'] = 'eV' f.create_dataset('/diffnumflux_params', data=fp)
scivision/gridaurora
gridaurora/eFluxGen.py
hitail
python
def hitail(E: np.ndarray, diffnumflux: np.ndarray, isimE0: np.ndarray, E0: np.ndarray, Bhf: np.ndarray, bh: float, verbose: int = 0): Bh = np.empty_like(E0) for iE0 in np.arange(E0.size): Bh[iE0] = Bhf[iE0]*diffnumflux[isimE0[iE0], iE0] # 4100. # bh = 4 #2.9 het = Bh*(E[:, None] / E0)**-bh het[E[:, None] < E0] = 0. if verbose > 0: print('Bh: ' + (' '.join('{:0.1f}'.format(b) for b in Bh))) return het
strickland 1993 said 0.2, but 0.145 gives better match to peak flux at 2500 = E0
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/eFluxGen.py#L90-L103
null
""" Michael Hirsch based on Strickland 1993 """ import logging from pathlib import Path import numpy as np import h5py from typing import Tuple pi = np.pi def maxwellian(E: np.ndarray, E0: np.ndarray, Q0: np.ndarray) -> Tuple[np.ndarray, float]: """ input: ------ E: 1-D vector of energy bins [eV] E0: characteristic energy (scalar or vector) [eV] Q0: flux coefficient (scalar or vector) (to yield overall flux Q) output: ------- Phi: differential number flux Q: total flux Tanaka 2006 Eqn. 1 http://odin.gi.alaska.edu/lumm/Papers/Tanaka_2006JA011744.pdf """ E0 = np.atleast_1d(E0) Q0 = np.atleast_1d(Q0) assert E0.ndim == Q0.ndim == 1 assert (Q0.size == 1 or Q0.size == E0.size) Phi = Q0/(2*pi*E0**3) * E[:, None] * np.exp(-E[:, None]/E0) Q = np.trapz(Phi, E, axis=0) logging.info('total maxwellian flux Q: ' + (' '.join('{:.1e}'.format(q) for q in Q))) return Phi, Q def fluxgen(E, E0, Q0, Wbc, bl, bm, bh, Bm, Bhf, verbose: int = 0) -> tuple: Wb = Wbc*E0 isimE0 = abs(E - E0).argmin() base = gaussflux(E, Wb, E0, Q0) diffnumflux = base.copy() low = letail(E, E0, Q0, bl, verbose) diffnumflux += low # intermediate result mid = midtail(E, E0, bm, Bm) diffnumflux += mid # intermediate result hi = hitail(E, diffnumflux, isimE0, E0, Bhf, bh, verbose) diffnumflux += hi if verbose > 0: diprat(E0, diffnumflux, isimE0) Q = np.trapz(diffnumflux, E, axis=0) if verbose > 0: print('total flux Q: ' + (' '.join('{:.1e}'.format(q) for q in Q))) return np.asfortranarray(diffnumflux), low, mid, hi, base, Q def letail(E: np.ndarray, E0: float, Q0: float, bl: float, verbose: int = 0) -> np.ndarray: # for LET, 1<b<2 # Bl = 8200. #820 (typo?) Bl = 0.4*Q0/(2*pi*E0**2) * np.exp(-1) # bl = 1.0 #1 low = Bl * (E[:, None]/E0)**-bl low[E[:, None] > E0] = 0. if verbose > 0: print('Bl: ' + (' '.join('{:0.1f}'.format(b) for b in Bl))) return low def midtail(E: np.ndarray, E0: np.ndarray, bm: float, Bm: float): # Bm = 1.8e4 #1.8e4 # bm = 3. #3 mid = Bm*(E[:, None]/E0)**bm mid[E[:, None] > E0] = 0. return mid def diprat(E0: np.ndarray, arc: np.ndarray, isimE0: np.ndarray): dipratio = np.empty_like(E0) for iE0 in np.arange(E0.size): idip = arc[:isimE0[iE0], iE0].argmin(axis=0) dipratio[iE0] = arc[idip, iE0]/arc[isimE0[iE0], iE0] print('dipratio: ' + (' '.join(f'{d:0.2f}' for d in dipratio))) # if not all(0.2<dipratio<0.5): # warn('dipratio outside of 0.2<dipratio<0.5') def gaussflux(E, Wb, E0, Q0): Qc = Q0/(pi**(3/2) * Wb*E0) return Qc * np.exp(-((E[:, None]-E0) / Wb)**2) def writeh5(h5fn: Path, Phi: np.ndarray, E, fp): if h5fn: with h5py.File(h5fn, 'w') as f: f.create_dataset('/diffnumflux', data=Phi) hE = f.create_dataset('/E', data=E) hE.attrs['Units'] = 'eV' f.create_dataset('/diffnumflux_params', data=fp)
scivision/gridaurora
gridaurora/plots.py
plotOptMod
python
def plotOptMod(verNObg3gray, VERgray): if VERgray is None and verNObg3gray is None: return fg = figure() ax2 = fg.gca() # summed (as camera would see) if VERgray is not None: z = VERgray.alt_km Ek = VERgray.energy_ev.values # ax1.semilogx(VERgray, z, marker='',label='filt', color='b') props = {'boxstyle': 'round', 'facecolor': 'wheat', 'alpha': 0.5} fgs, axs = fg.subplots(6, 6, sharex=True, sharey='row') axs = axs.ravel() # for convenient iteration fgs.subplots_adjust(hspace=0, wspace=0) fgs.suptitle('filtered VER/flux') fgs.text(0.04, 0.5, 'Altitude [km]', va='center', rotation='vertical') fgs.text(0.5, 0.04, 'Beam energy [eV]', ha='center') for i, e in enumerate(Ek): axs[i].semilogx(VERgray.loc[:, e], z) axs[i].set_xlim((1e-3, 1e4)) # place a text box in upper left in axes coords axs[i].text(0.95, 0.95, '{:0.0f}'.format(e)+'eV', transform=axs[i].transAxes, fontsize=12, va='top', ha='right', bbox=props) for i in range(33, 36): axs[i].axis('off') ax2.semilogx(VERgray.sum(axis=1), z, label='filt', color='b') # specific to energies ax = figure().gca() for e in Ek: ax.semilogx(VERgray.loc[:, e], z, marker='', label='{:.0f} eV'.format(e)) ax.set_title('filtered VER/flux') ax.set_xlabel('VER/flux') ax.set_ylabel('altitude [km]') ax.legend(loc='best', fontsize=8) ax.set_xlim((1e-5, 1e5)) ax.grid(True) if verNObg3gray is not None: ax1 = figure().gca() # overview z = verNObg3gray.alt_km Ek = verNObg3gray.energy_ev.values ax1.semilogx(verNObg3gray, z, marker='', label='unfilt', color='r') ax2.semilogx(verNObg3gray.sum(axis=1), z, label='unfilt', color='r') ax = figure().gca() for e in Ek: ax.semilogx(verNObg3gray.loc[:, e], z, marker='', label='{:.0f} eV'.format(e)) ax.set_title('UNfiltered VER/flux') ax.set_xlabel('VER/flux') ax.set_ylabel('altitude [km]') ax.legend(loc='best', fontsize=8) ax.set_xlim((1e-5, 1e5)) ax.grid(True) ax1.set_title('VER/flux, one profile per beam') ax1.set_xlabel('VER/flux') ax1.set_ylabel('altitude [km]') ax1.grid(True) ax2.set_xlabel('VER/flux') ax2.set_ylabel('altitude [km]') ax2.set_title('VER/flux summed over all energy beams \n (as the camera would see)') ax2.legend(loc='best') ax2.grid(True)
called from either readTranscar.py or hist-feasibility/plotsnew.py
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/plots.py#L260-L331
null
import logging from datetime import datetime from pathlib import Path import h5py import xarray from numpy.ma import masked_invalid # for pcolormesh, which doesn't like NaN from matplotlib.pyplot import figure, draw, close from matplotlib.colors import LogNorm from matplotlib.ticker import MultipleLocator from matplotlib.dates import SecondLocator, DateFormatter, MinuteLocator from typing import List import numpy as np import os import gridaurora.ztanh as ga import gridaurora.opticalmod as gao if os.name == 'nt': import pathvalidate else: pathvalidate = None # IEEE Transactions requires 600 dpi dymaj = 100 dymin = 20 def writeplots(fg, plotprefix, tind=None, odir=None, fmt='.png', anno=None, dpi=None, facecolor=None, doclose=True): try: if fg is None or odir is None: return # %% draw() # Must have this here or plot doesn't update in animation multiplot mode! # TIF was not faster and was 100 times the file size! # PGF is slow and big file, # RAW crashes # JPG no faster than PNG suff = nametime(tind) if anno: fg.text(0.15, 0.8, anno, fontsize='x-large') if pathvalidate is not None: cn = Path(odir).expanduser() / pathvalidate.sanitize_filename(plotprefix + suff + fmt) else: cn = Path(odir).expanduser() / (plotprefix + suff + fmt) print('write', cn) if facecolor is None: facecolor = fg.get_facecolor() fg.savefig(cn, bbox_inches='tight', dpi=dpi, facecolor=facecolor, edgecolor='none') if doclose: close(fg) except Exception as e: logging.error(f'{e} when plotting {plotprefix}') def nametime(tind): if isinstance(tind, int) and tind < 1e6: return '{:03d}'.format(tind) elif isinstance(tind, datetime): return tind.isoformat()[:-3] # -3 truncates to millisecond digits only (arbitrary) elif tind is not None: return str(tind) else: # is None return '' # %% def plotflux(E, E0, arc, base=None, hi=None, low=None, mid=None, ttxt='Differential Number Flux'): FMAX = 1e6 FMIN = 1e2 lblstr = ['{:.0f}'.format(e0) for e0 in E0] ax = figure().gca() if np.isscalar(E0) and mid is not None: ax.loglog(E, hi, 'k:') ax.loglog(E, low, 'k:') ax.loglog(E, mid, 'k:') ax.loglog(E, base, color='k') ax.loglog(E, arc, linewidth=2) ax.grid(True, which='both') ax.set_xlabel('Electron Energy [eV]') # ,fontsize=afs,labelpad=-2) ax.set_ylabel('Flux [cm$^{-2}$s$^{-1}$eV$^{-1}$sr$^{-1}$]') # ,fontsize=afs) ax.set_title(ttxt) # ax.tick_params(axis='both', which='both') ax.autoscale(True, tight=True) ax.set_ylim((1e2, FMAX)) ax.legend(lblstr, loc='best') # ,prop={'size':'large'}) # ax.set_xlim((1e2,1e4)) # sns.despine(ax=ax) if base is not None: ax = figure().gca() ax.loglog(E, base) ax.set_ylim((FMIN, FMAX)) # ax.set_xlim((1e2,1e4)) ax.set_title('arc Gaussian base function, E0=' + str(E0) + '[eV]' + '\n Wbc: width, Q0: height') ax.set_xlabel('Electron Energy [eV]') ax.set_ylabel('Flux [cm$^{-2}$s$^{-1}$eV$^{-1}$sr$^{-1}$]') ax.legend(lblstr) ax = figure().gca() ax.loglog(E, low) ax.set_ylim((FMIN, FMAX)) ax.set_title('arc low (E<E0). Bl: height, bh: slope') ax.set_xlabel('Electron Energy [eV]') ax.set_ylabel('Flux [cm$^{-2}$s$^{-1}$eV$^{-1}$sr$^{-1}$]') ax = figure().gca() ax.loglog(E, mid) ax.set_ylim((FMIN, FMAX)) ax.set_title('arc mid. Bm:height, bm: slope') ax.set_xlabel('Electron Energy [eV]') ax.set_ylabel('Flux [cm$^{-2}$s$^{-1}$eV$^{-1}$sr$^{-1}$]') ax = figure().gca() ax.loglog(E, hi) ax.set_ylim((FMIN, FMAX)) ax.set_title('arc hi (E>E0). Bhf: height, bh: slope') ax.set_xlabel('Electron Energy [eV]') ax.set_ylabel('Flux [cm$^{-2}$s$^{-1}$eV$^{-1}$sr$^{-1}$]') # %% def ploteigver(EKpcolor, zKM, eigenprofile, vlim=(None,)*6, sim=None, tInd=None, makeplot=None, prefix=None, progms=None): try: fg = figure() ax = fg.gca() # pcolormesh canNOT handle nan at all pcm = ax.pcolormesh(EKpcolor, zKM, masked_invalid(eigenprofile), edgecolors='none', # cmap=pcmcmap, norm=LogNorm(), vmin=vlim[4], vmax=vlim[5]) ax.set_xlabel('Energy [eV]') ax.set_ylabel(r'$B_\parallel$ [km]') ax.autoscale(True, tight=True) ax.set_xscale('log') ax.yaxis.set_major_locator(MultipleLocator(dymaj)) ax.yaxis.set_minor_locator(MultipleLocator(dymin)) # %% title if tInd is not None: mptitle = str(tInd) else: mptitle = '' mptitle += '$P_{{eig}}$' if sim: mptitle += ', filter: {}'.format(sim.opticalfilter) mptitle += str(sim.reacreq) ax.set_title(mptitle) # ,fontsize=tfs) # %% colorbar cbar = fg.colorbar(pcm, ax=ax) cbar.set_label('[photons cm$^{-3}$s$^{-1}$]', labelpad=0) # ,fontsize=afs) # cbar.ax.tick_params(labelsize=afs) # cbar.ax.yaxis.get_offset_text().set_size(afs) # %% ticks,lim ax.tick_params(axis='both', which='both', direction='out') ax.set_ylim(vlim[2:4]) # %% writeplots(fg, prefix, tInd, makeplot, progms) except Exception as e: logging.error('tind {} {}'.format(tInd, e)) def plotT(T, mmsl): ax1 = figure().gca() for c in ['filter', 'window', 'qe', 'atm']: ax1.plot(T.wavelength_nm, T[c], label=c) ax1.set_xlim(mmsl[:2]) ax1.set_title(f'{T.filename} Component transmittance') # ax2 = figure().gca() for s in ['sys', 'sysNObg3']: ax2.plot(T.wavelength_nm, T[s], label=s) ax2.set_title(f'{T.filename} System Transmittance') for a in (ax1, ax2): niceTax(a) def niceTax(a): a.set_xlabel('wavelength (nm)') a.set_ylabel('Transmittance (unitless)') # a.set_yscale('log') a.legend(loc='best') # a.set_ylim(1e-2,1) a.invert_xaxis() a.grid(True, which='both') def comparefilters(Ts): fg = figure() axs = fg.subplots(len(Ts), 1, sharex=True, sharey=True) for T, ax in zip(Ts, axs): try: ax.plot(T.wavelength_nm, T['filter'], label=T.filename) except ValueError: # just a plain filter assert T.ndim == 1 ax.plot(T.wavelength_nm, T, label=T.filename) forbidden = [630., 555.7, ] permitted = [391.4, 427.8, 844.6, 777.4] for l in forbidden: ax.axvline(l, linestyle='--', color='darkred', alpha=0.8) for l in permitted: ax.axvline(l, linestyle='--', color='darkgreen', alpha=0.8) ax.set_title(f'{T.filename}') fg.suptitle('Transmittance') ax.set_ylim((0, 1)) ax.set_xlim(T.wavelength_nm[[-1, 0]]) ax.set_xlabel('wavelength [nm]') def plotz(z: np.ndarray): dz = np.gradient(z, edge_order=2) # numpy>=1.9.1 dzmed = np.median(dz) ax = figure().gca() ax.plot(dz, z) ax.axvline(dzmed, color='r', linestyle='--', label='median') ax.set_xlabel('grid spacing [km]') ax.set_ylabel('altitude [km]') ax.set_title('$N_p=$'+str(z.shape[0])) ax.grid(True) ax.legend(loc='best') if __name__ == '__main__': from matplotlib.pyplot import show from argparse import ArgumentParser p = ArgumentParser(description='create continuously step sized grid') p.add_argument('-n', '--np', help='number of points in grid', type=int, default=300) p.add_argument('--zmin', help='bottom of grid', type=float, default=90) p.add_argument('--gmin', help='minimum grid spacing', type=float, default=1.5) p.add_argument('--gmax', help='max grid spacing', type=float, default=10.575) a = p.parse_args() zgrid = ga.setupz(a.np, a.zmin, a.gmin, a.gmax) plotz(zgrid) print(zgrid[-1]) show() # %% HIST def comparejgr2013(altkm, zenang, bg3fn, windfn, qefn): R = Path(__file__).parent with h5py.File(R / 'precompute/trans_jgr2013a.h5', 'r') as f: reqLambda = f['/lambda'][:] Tjgr2013 = f['/T'][:] optT = gao.getSystemT(reqLambda, bg3fn, windfn, qefn, altkm, zenang) ax = figure().gca() ax.semilogy(reqLambda, optT['sys'], 'b', label='HST') ax.semilogy(reqLambda, Tjgr2013, 'r', label='JGR2013') ax.set_xlabel('wavelength [nm]') ax.set_ylabel('T') ax.set_title('Comparision of Transmission models: HST vs. JGR2013') ax.grid(True) ax.legend(loc='best') ax.set_title('System Transmission + Atmospheric Absorption') ax.set_ylim(1e-10, 1) def plotAllTrans(optT, log): mutwl = optT.wavelength_nm fg = figure(figsize=(7, 5)) ax = fg.gca() ax.plot(mutwl, optT['sys'], label='optics') ax.plot(mutwl, optT['atm'], label='atmosphere') ax.plot(mutwl, optT['sys']*optT['atm'], label='total', linewidth=2) if log: ax.set_yscale('log') ax.set_ylim(bottom=1e-5) ax.set_xlabel('wavelength [nm]') ax.set_ylabel('Transmission [dimensionless]') ax.set_title('System Transmission') ax.grid(True, 'both') ax.invert_xaxis() ax.xaxis.set_major_locator(MultipleLocator(100)) ax.legend(loc='center', bbox_to_anchor=(0.3, 0.15)) return fg def plotPeigen(Peigen): # Peigen: Nalt x Nenergy if not isinstance(Peigen, xarray.DataArray): return fg = figure() ax = fg.gca() pcm = ax.pcolormesh(Peigen.energy_ev, Peigen.alt_km, Peigen.values) ax.autoscale(True, tight=True) ax.set_xscale('log') ax.set_xlabel('beam energy [eV]') ax.set_ylabel('altitude [km]') ax.set_title('Volume Emission Rate per unit diff num flux') fg.colorbar(pcm) def showIncrVER(tTC: np.ndarray, tReqInd: int, tctime: np.ndarray, ver: xarray.DataArray, tver: xarray.DataArray, titxt: str, makePlots: List[str]): saveplot = False z = ver.alt_km lamb = ver.wavelength # if 'spectra1d' in makePlots: # b = np.trapz(ver, z, axis=1) # integrate along z, looking up magnetic zenith # plotspectra(b, lamb) if 'eigtime' in makePlots: fg = figure(figsize=(11, 8), dpi=100, tight_layout=True) ax = fg.gca() pcm = ax.pcolormesh(tTC, z, tver.sum(axis=0), # sum over wavelength edgecolors='none', cmap=None, norm=None, vmin=0, vmax=1e3) ax.axvline(tTC[tReqInd], color='white', linestyle='--', label='Req. Time') ax.axvline(tctime['tstartPrecip'], color='red', linestyle='--', label='Precip. Start') ax.axvline(tctime['tendPrecip'], color='red', linestyle='--', label='Precip. End') titlemean = titxt + (r'\n VER/flux: $\lambda \in$' + str(lamb) + ' [nm]' + '\n geodetic lat:' + str(tctime['latgeo_ini']) + ' lon:' + str(tctime['longeo_ini']) + ' date: ' + tctime['dayofsim'].strftime('%Y-%m-%d')) # make room for long title fg.subplots_adjust(top=0.8) ax.set_title(titlemean, fontsize=9) ax.yaxis.set_major_locator(MultipleLocator(100)) ax.yaxis.set_minor_locator(MultipleLocator(20)) # ax.xaxis.set_major_locator(MinuteLocator(interval=10)) ax.xaxis.set_major_locator(MinuteLocator(interval=1)) ax.xaxis.set_minor_locator(SecondLocator(interval=10)) ax.xaxis.set_major_formatter(DateFormatter('%H:%M:%S')) ax.tick_params(axis='both', which='both', direction='out', labelsize=12) ax.autoscale(True, tight=True) cbar = fg.colorbar(pcm) cbar.set_label('VER/flux', labelpad=0) ax.set_xlabel('Time [UTC]') ax.set_ylabel('altitude [km]') if saveplot: sfn = ''.join(e for e in titxt if e.isalnum() or e == '.') # remove special characters fg.savefig('out/VER' + sfn + '.png', dpi=150, bbox_inches='tight') close(fg) if 'eigtime1d' in makePlots: fg = figure(figsize=(11, 8), dpi=100) ax = fg.gca() # fg.subplots_adjust(top=0.85) thistitle = titxt + ': {:d} emission lines\n VER/flux: geodetic lat: {} lon: {} {}'.format( ver.shape[0], tctime['latgeo_ini'], tctime['longeo_ini'], tTC[tReqInd]) ax.set_title(thistitle, fontsize=12) ax.set_xlabel('VER/flux') ax.set_ylabel('altitude [km]') for ifg, clamb in enumerate(lamb): ax.semilogx(ver.iloc[ifg, :], z, label=str(clamb)) ax.yaxis.set_major_locator(MultipleLocator(100)) ax.yaxis.set_minor_locator(MultipleLocator(20)) ax.grid(True) if ver.shape[0] < 20: ax.legend(loc='upper center', bbox_to_anchor=(1.05, .95), ncol=1, fancybox=True, shadow=True, fontsize=9) ax.tick_params(axis='both', which='both', direction='in', labelsize=12) ax.set_xlim(1e-9, 1e3) ax.set_ylim((z[0], z[-1])) if saveplot: sfn = ''.join(e for e in titxt if e.isalnum()) # remove special characters fg.savefig('out/VER' + sfn + '.png', dpi=150, bbox_inches='tight') close(fg) def plotspectra(br, optT: xarray.DataArray, E: float, lambminmax: tuple): spectraAminmax = (1e-1, 8e5) # for plotting spectrallines = (391.44, 427.81, 557.7, 630.0, 777.4, 844.6) # 297.2, 636.4,762.0, #for plotting lamb = optT.wavelength_nm def _plotspectrasub(ax, bf, txt): ax.set_yscale('log') ax.set_title('Auroral spectrum, ' + txt + f',integrated along flux tube: $E_0$ = {E:.0f} eV') ax.set_ylabel('optical intensity') ax.set_xlim(lambminmax) ax.set_ylim(spectraAminmax) ax.xaxis.set_major_locator(MultipleLocator(100)) # ax.invert_xaxis() for l in spectrallines: ax.text(l, bf[l]*1.7, '{:.1f}'.format(l), ha='center', va='bottom', fontsize='medium', rotation=60) # %% fg = figure() ax1, ax2 = fg.subplots(2, 1, sharex=True, figsize=(10, 8)) bf = br*optT['sysNObg3'] ax1.stem(lamb, bf) _plotspectrasub(ax1, bf, 'no filter') bf = br*optT['sys'] ax2.stem(lamb, bf) _plotspectrasub(ax2, bf, 'BG3 filter') ax2.set_xlabel('wavelength [nm]') return fg
scivision/gridaurora
gridaurora/opticalmod.py
opticalModel
python
def opticalModel(sim, ver: xarray.DataArray, obsAlt_km: float, zenithang: float): assert isinstance(ver, xarray.DataArray) # %% get system optical transmission T optT = getSystemT(ver.wavelength_nm, sim.bg3fn, sim.windowfn, sim.qefn, obsAlt_km, zenithang) # %% first multiply VER by T, THEN sum overall wavelengths if sim.opticalfilter == 'bg3': VERgray = (ver*optT['sys'].values[None, :]).sum('wavelength_nm') elif sim.opticalfilter == 'none': VERgray = (ver*optT['sysNObg3'].values[None, :]).sum('wavelength_nm') else: logging.warning(f'unknown OpticalFilter type: {sim.opticalfilter}' ' falling back to using no filter at all') VERgray = (ver*optT['sysNObg3'].values[None, :]).sum('wavelength_nm') return VERgray
ver: Nalt x Nwavelength
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/opticalmod.py#L7-L25
[ "def getSystemT(newLambda, bg3fn: Path, windfn: Path, qefn: Path,\n obsalt_km, zenang_deg, verbose: bool = False) -> xarray.Dataset:\n\n bg3fn = Path(bg3fn).expanduser()\n windfn = Path(windfn).expanduser()\n qefn = Path(qefn).expanduser()\n\n newLambda = np.asarray(newLambda)\n# %% atmospheric absorption\n if lowtran is not None:\n c1 = {'model': 5, 'h1': obsalt_km, 'angle': zenang_deg,\n 'wlshort': newLambda[0], 'wllong': newLambda[-1]}\n if verbose:\n print('loading LOWTRAN7 atmosphere model...')\n atmT = lowtran.transmittance(c1)['transmission'].squeeze()\n try:\n atmTcleaned = atmT.values.squeeze()\n atmTcleaned[atmTcleaned == 0] = np.spacing(1) # to avoid log10(0)\n fwl = interp1d(atmT.wavelength_nm, np.log(atmTcleaned), axis=0)\n except AttributeError: # problem with lowtran\n fwl = interp1d(newLambda, np.log(np.ones_like(newLambda)), kind='linear')\n else:\n fwl = interp1d(newLambda, np.log(np.ones_like(newLambda)), kind='linear')\n\n atmTinterp = np.exp(fwl(newLambda))\n if not np.isfinite(atmTinterp).all():\n logging.error('problem in computing LOWTRAN atmospheric attenuation, results are suspect!')\n# %% BG3 filter\n with h5py.File(bg3fn, 'r') as f:\n try:\n assert isinstance(f['/T'], h5py.Dataset), 'we only allow one transmission curve per file' # simple legacy behavior\n fbg3 = interp1d(f['/wavelength'], np.log(f['/T']), kind='linear', bounds_error=False)\n except KeyError:\n raise KeyError('could not find /wavelength in {}'.format(f.filename))\n\n try:\n fname = f['T'].attrs['name'].item()\n if isinstance(fname, bytes):\n fname = fname.decode('utf8')\n except KeyError:\n fname = ''\n# %% camera window\n with h5py.File(windfn, 'r') as f:\n fwind = interp1d(f['/lamb'], np.log(f['/T']), kind='linear')\n# %% quantum efficiency\n with h5py.File(qefn, 'r') as f:\n fqe = interp1d(f['/lamb'], np.log(f['/QE']), kind='linear')\n# %% collect results into DataArray\n\n T = xarray.Dataset({'filter': ('wavelength_nm', np.exp(fbg3(newLambda))),\n 'window': ('wavelength_nm', np.exp(fwind(newLambda))),\n 'qe': ('wavelength_nm', np.exp(fqe(newLambda))),\n 'atm': ('wavelength_nm', atmTinterp), },\n coords={'wavelength_nm': newLambda},\n attrs={'filename': fname})\n\n T['sysNObg3'] = T['window'] * T['qe'] * T['atm']\n T['sys'] = T['sysNObg3'] * T['filter']\n\n return T\n" ]
#!/usr/bin/env python import logging import xarray from .filterload import getSystemT
scivision/gridaurora
MakeIonoEigenprofile.py
main
python
def main(): p = ArgumentParser(description='Makes unit flux eV^-1 as input to GLOW or Transcar to create ionospheric eigenprofiles') p.add_argument('-i', '--inputgridfn', help='original Zettergren input flux grid to base off of', default='zettflux.csv') p.add_argument('-o', '--outfn', help='hdf5 file to write with ionospheric response (eigenprofiles)') p.add_argument('-t', '--simtime', help='yyyy-mm-ddTHH:MM:SSZ time of sim', nargs='+', default=['1999-12-21T00:00:00Z']) p.add_argument('-c', '--latlon', help='geodetic latitude/longitude (deg)', type=float, nargs=2, default=[65, -148.]) # p.add_argument('-m', '--makeplot', help='show to show plots, png to save pngs of plots', nargs='+', default=['show']) p.add_argument('-M', '--model', help='specify auroral model (glow,rees,transcar)', default='glow') p.add_argument('-z', '--zlim', help='minimum,maximum altitude [km] to plot', nargs=2, default=(None, None), type=float) p.add_argument('--isotropic', help='(rees model only) isotropic or non-isotropic pitch angle', action='store_true') p.add_argument('--vlim', help='plotting limits on energy dep and production plots', nargs=2, type=float, default=(1e-7, 1e1)) p = p.parse_args() if not p.outfn: print('you have not specified an output file with -o options, so I will only plot and not save result') # makeplot = p.makeplot if len(p.simtime) == 1: T = [parse(p.simtime[0])] elif len(p.simtime) == 2: T = list(rrule.rrule(rrule.HOURLY, dtstart=parse(p.simtime[0]), until=parse(p.simtime[1]))) # %% input unit flux Egrid = loadregress(Path(p.inputgridfn).expanduser()) Ebins = makebin(Egrid)[:3] EKpcolor, EK, diffnumflux = ekpcolor(Ebins) # %% ionospheric response model = p.model.lower() glat, glon = p.latlon if model == 'glow': ver, photIon, isr, phitop, zceta, sza, prates, lrates, tezs, sion = makeeigen(EK, diffnumflux, T, p.latlon, p.makeplot, p.outfn, p.zlim) writeeigen(p.outfn, EKpcolor, T, ver.z_km, diffnumflux, ver, prates, lrates, tezs, p.latlon) # %% plots # input doplot(p.inputgridfn, Ebins) # output sim = namedtuple('sim', ['reacreq', 'opticalfilter']) sim.reacreq = sim.opticalfilter = '' for t in ver: # TODO for each time # VER eigenprofiles, summed over wavelength ploteigver(EKpcolor, ver.z_km, ver.sum('wavelength_nm'), (None,)*6, sim, '{} Vol. Emis. Rate '.format(t)) # volume production rate, summed over reaction plotprodloss(prates.loc[:, 'final', ...].sum('reaction'), lrates.loc[:, 'final', ...].sum('reaction'), t, glat, glon, p.zlim) # energy deposition plotenerdep(tezs, t, glat, glon, p.zlim) elif model == 'rees': assert len(T) == 1, 'only one time with rees for now.' z = glowalt() q = reesiono(T, z, Ebins.loc[:, 'low'], glat, glon, p.isotropic) writeeigen(p.outfn, Ebins, T, z, prates=q, tezs=None, latlon=(glat, glon)) plotA(q, 'Volume Production Rate {} {} {}'.format(T, glat, glon), p.vlim) elif model == 'transcar': raise NotImplementedError('Transcar by request') else: raise NotImplementedError('I am not yet able to handle your model {}'.format(model)) # %% plots show()
three output eigenprofiles 1) ver (optical emissions) 4-D array: time x energy x altitude x wavelength 2) prates (production) 4-D array: time x energy x altitude x reaction 3) lrates (loss) 4-D array: time x energy x altitude x reaction
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/MakeIonoEigenprofile.py#L35-L112
[ "def ploteigver(EKpcolor, zKM, eigenprofile,\n vlim=(None,)*6, sim=None, tInd=None, makeplot=None, prefix=None, progms=None):\n try:\n fg = figure()\n ax = fg.gca()\n # pcolormesh canNOT handle nan at all\n pcm = ax.pcolormesh(EKpcolor, zKM, masked_invalid(eigenprofile),\n edgecolors='none', # cmap=pcmcmap,\n norm=LogNorm(),\n vmin=vlim[4], vmax=vlim[5])\n ax.set_xlabel('Energy [eV]')\n ax.set_ylabel(r'$B_\\parallel$ [km]')\n ax.autoscale(True, tight=True)\n ax.set_xscale('log')\n ax.yaxis.set_major_locator(MultipleLocator(dymaj))\n ax.yaxis.set_minor_locator(MultipleLocator(dymin))\n# %% title\n if tInd is not None:\n mptitle = str(tInd)\n else:\n mptitle = ''\n mptitle += '$P_{{eig}}$'\n if sim:\n mptitle += ', filter: {}'.format(sim.opticalfilter)\n mptitle += str(sim.reacreq)\n\n ax.set_title(mptitle) # ,fontsize=tfs)\n# %% colorbar\n cbar = fg.colorbar(pcm, ax=ax)\n cbar.set_label('[photons cm$^{-3}$s$^{-1}$]', labelpad=0) # ,fontsize=afs)\n # cbar.ax.tick_params(labelsize=afs)\n # cbar.ax.yaxis.get_offset_text().set_size(afs)\n# %% ticks,lim\n ax.tick_params(axis='both', which='both', direction='out')\n ax.set_ylim(vlim[2:4])\n# %%\n writeplots(fg, prefix, tInd, makeplot, progms)\n except Exception as e:\n logging.error('tind {} {}'.format(tInd, e))\n", "def doplot(fn: Path, bins: xarray.DataArray, Egrid: np.ndarray = None, debug: bool = False):\n # %% main plot\n ax = figure().gca()\n ax.bar(left=bins.loc[:, 'low'],\n height=bins.loc[:, 'flux'],\n width=bins.loc[:, 'high']-bins.loc[:, 'low'])\n ax.set_yscale('log')\n ax.set_xscale('log')\n ax.set_ylabel('flux [s$^{-1}$ sr$^{-1}$ cm$^{-2}$ eV$^{-1}$]')\n ax.set_xlabel('bin energy [eV]')\n ax.set_title(f'Input flux used to generate eigenprofiles, based on {fn}')\n\n# %% debug plots\n if debug:\n ax = figure().gca()\n bins[['low', 'high']].plot(logy=True, ax=ax, marker='.')\n ax.set_xlabel('bin number')\n ax.set_ylabel('bin energy [eV]')\n\n ax = figure().gca()\n bins['flux'].plot(logy=True, ax=ax, marker='.')\n ax.set_xlabel('bin number')\n ax.set_ylabel('flux [s$^{-1}$ sr$^{-1}$ cm$^{-2}$ eV$^{-1}$]')\n\n if Egrid is not None:\n ax = figure().gca()\n ax.plot(Egrid, marker='.')\n # ax.plot(Ematt,marker='.',color='k')\n ax.set_yscale('log')\n ax.set_ylabel('eV')\n ax.legend(['E1', 'E2', 'pr1', 'pr2'], loc='best')\n", "def loadregress(fn: Path):\n # %%\n Egrid = np.loadtxt(Path(fn).expanduser(), delimiter=',')\n# Ematt = asarray([logspace(1.7220248253079387,4.2082263059355824,num=Nold,base=10),\n# #[logspace(3.9651086925197356,9.689799159992674,num=33,base=exp(1)),\n# logspace(1.8031633895706722,4.2851520785250914,num=Nold,base=10)]).T\n# %% log-lin regression\n Enew = np.empty((Nnew, 4))\n Enew[:Nold, :] = Egrid\n for k in range(4):\n s, i = linregress(range(Nold), np.log10(Egrid[:, k]))[:2]\n Enew[Nold:, k] = 10**(np.arange(Nold, Nnew)*s+i)\n\n return Enew\n", "def makebin(Egrid: np.ndarray):\n E1 = Egrid[:, 0]\n E2 = Egrid[:, 1]\n pr1 = Egrid[:, 2]\n pr2 = Egrid[:, 3]\n\n dE = E2-E1\n Esum = E2+E1\n flux = flux0 / 0.5 / Esum / dE\n Elow = E1 - 0.5*(E1 - pr1)\n Ehigh = E2 - 0.5*(E2 - pr2)\n\n E = np.column_stack((Elow, Ehigh, flux))\n\n Ed = xarray.DataArray(data=E, dims=['energy', 'type'])\n Ed['type'] = ['low', 'high', 'flux']\n\n return Ed\n", "def writeeigen(fn: Path, Ebins, t, z, diffnumflux=None, ver=None, prates=None, lrates=None,\n tezs=None, latlon=None):\n if not fn:\n return\n\n fn = Path(fn).expanduser()\n\n if fn.suffix != '.h5':\n return\n\n print('writing to', fn)\n\n ut1_unix = to_ut1unix(t)\n\n with h5py.File(fn, 'w') as f:\n bdt = h5py.special_dtype(vlen=bytes)\n d = f.create_dataset('/sensorloc', data=latlon)\n d.attrs['unit'] = 'degrees'\n d.attrs['description'] = 'geographic coordinates'\n# %% input precipitation flux\n d = f.create_dataset('/Ebins', data=Ebins)\n d.attrs['unit'] = 'eV'\n d.attrs['description'] = 'Energy bin edges'\n d = f.create_dataset('/altitude', data=z)\n d.attrs['unit'] = 'km'\n\n d = f.create_dataset('/ut1_unix', data=ut1_unix)\n d.attrs['unit'] = 'sec. since Jan 1, 1970 midnight' # float\n\n if diffnumflux is not None:\n d = f.create_dataset('/diffnumflux', data=diffnumflux)\n d.attrs['unit'] = 'cm^-2 s^-1 eV^-1'\n d.attrs['description'] = 'primary electron flux at \"top\" of modeled ionosphere'\n# %% VER\n if isinstance(ver, DataArray):\n d = f.create_dataset('/ver/eigenprofile', data=ver.values, compression='gzip')\n d.attrs['unit'] = 'photons cm^-3 sr^-1 s^-1'\n d.attrs['size'] = 'Ntime x NEnergy x Nalt x Nwavelength'\n\n d = f.create_dataset('/ver/wavelength', data=ver.wavelength_nm)\n d.attrs['unit'] = 'Angstrom'\n# %% prod\n if isinstance(prates, DataArray):\n d = f.create_dataset('/prod/eigenprofile', data=prates.values, compression='gzip')\n d.attrs['unit'] = 'particle cm^-3 sr^-1 s^-1'\n if prates.ndim == 3:\n d.attrs['size'] = 'Ntime x NEnergy x Nalt'\n else: # ndim==4\n d.attrs['size'] = 'Ntime x NEnergy x Nalt x Nreaction'\n d = f.create_dataset('/prod/reaction', data=prates.reaction, dtype=bdt)\n d.attrs['description'] = 'reaction species state'\n# %% loss\n if isinstance(lrates, DataArray):\n d = f.create_dataset('/loss/eigenprofiles', data=lrates.values, compression='gzip')\n d.attrs['unit'] = 'particle cm^-3 sr^-1 s^-1'\n d.attrs['size'] = 'Ntime x NEnergy x Nalt x Nreaction'\n d = f.create_dataset('/loss/reaction', data=lrates.reaction, dtype=bdt)\n d.attrs['description'] = 'reaction species state'\n# %% energy deposition\n if isinstance(tezs, DataArray):\n d = f.create_dataset('/energydeposition', data=tezs.values, compression='gzip')\n d.attrs['unit'] = 'ergs cm^-3 s^-1'\n d.attrs['size'] = 'Ntime x Nalt x NEnergies'\n", "def glowalt() -> np.ndarray:\n # z = range(80,110+1,1)\n z = np.arange(30., 110+1., 1.)\n z = np.append(z, [111.5, 113., 114.5, 116.])\n z = np.append(z, np.arange(118, 150+2, 2.))\n z = np.append(z, np.arange(153, 168+3, 3.))\n z = np.append(z, np.arange(172, 180+4, 4.))\n z = np.append(z, np.arange(185, 205+5, 5))\n z = np.append(z, np.arange(211, 223+6, 6))\n z = np.append(z, np.arange(230, 244+7, 7))\n z = np.append(z, np.arange(252, 300+8, 8))\n z = np.append(z, np.arange(309, 345+9, 9))\n z = np.append(z, np.arange(355, 395+10, 10))\n z = np.append(z, np.arange(406, 428+11, 11))\n z = np.append(z, [440., 453, 467, 482, 498, 515, 533, 551])\n z = np.append(z, np.arange(570, 950+20, 20))\n\n return z\n" ]
#!/usr/bin/env python """ Computes Eigenprofiles of Ionospheric response to flux tube input via the following steps: 1. Generate unit input differential number flux vs. energy 2. Compute ionospheric energy deposition and hence production/loss rates for the modeled kinetic chemistries (12 in total) unverified for proper scaling, fitted exponential curve to extrapolate original Zettergren grid from 50eV-18keV up to 100MeV example: python MakeIonoEigenprofile.py -t 2013-01-31T09:00:00Z -c 65 -148 -o ~/data/eigen.h5 Michael Hirsch """ from argparse import ArgumentParser from gridaurora.loadtranscargrid import loadregress, makebin, doplot from gridaurora.writeeigen import writeeigen from gridaurora.zglow import glowalt from glowaurora.eigenprof import makeeigen, ekpcolor from glowaurora.plots import plotprodloss, plotenerdep from gridaurora.plots import ploteigver from reesaurora.rees_model import reesiono from reesaurora.plots import plotA from pathlib import Path from collections import namedtuple from matplotlib.pyplot import show from dateutil import rrule from dateutil.parser import parse import seaborn as sns # optional pretty plots sns.color_palette(sns.color_palette("cubehelix")) sns.set(context='talk', style='whitegrid') sns.set(rc={'image.cmap': 'cubehelix_r'}) # for contour if __name__ == '__main__': main()
scivision/gridaurora
gridaurora/ztanh.py
setupz
python
def setupz(Np: int, zmin: float, gridmin: float, gridmax: float) -> np.ndarray: dz = _ztanh(Np, gridmin, gridmax) return np.insert(np.cumsum(dz)+zmin, 0, zmin)[:-1]
np: number of grid points zmin: minimum STEP SIZE at minimum grid altitude [km] gridmin: minimum altitude of grid [km] gridmax: maximum altitude of grid [km]
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/ztanh.py#L9-L19
[ "def _ztanh(Np: int, gridmin: float, gridmax: float) -> np.ndarray:\n \"\"\"\n typically call via setupz instead\n \"\"\"\n x0 = np.linspace(0, 3.14, Np) # arbitrarily picking 3.14 as where tanh gets to 99% of asymptote\n return np.tanh(x0)*gridmax+gridmin\n" ]
#!/usr/bin/env python """ inspired by Matt Zettergren Michael Hirsch """ import numpy as np def _ztanh(Np: int, gridmin: float, gridmax: float) -> np.ndarray: """ typically call via setupz instead """ x0 = np.linspace(0, 3.14, Np) # arbitrarily picking 3.14 as where tanh gets to 99% of asymptote return np.tanh(x0)*gridmax+gridmin # def zexp(np,gridmin): # x0 = linspace(0, 1, np) # return exp(x0)**2+(gridmin-1)
scivision/gridaurora
gridaurora/ztanh.py
_ztanh
python
def _ztanh(Np: int, gridmin: float, gridmax: float) -> np.ndarray: x0 = np.linspace(0, 3.14, Np) # arbitrarily picking 3.14 as where tanh gets to 99% of asymptote return np.tanh(x0)*gridmax+gridmin
typically call via setupz instead
train
https://github.com/scivision/gridaurora/blob/c3957b93c2201afff62bd104e0acead52c0d9e90/gridaurora/ztanh.py#L22-L27
null
#!/usr/bin/env python """ inspired by Matt Zettergren Michael Hirsch """ import numpy as np def setupz(Np: int, zmin: float, gridmin: float, gridmax: float) -> np.ndarray: """ np: number of grid points zmin: minimum STEP SIZE at minimum grid altitude [km] gridmin: minimum altitude of grid [km] gridmax: maximum altitude of grid [km] """ dz = _ztanh(Np, gridmin, gridmax) return np.insert(np.cumsum(dz)+zmin, 0, zmin)[:-1] # def zexp(np,gridmin): # x0 = linspace(0, 1, np) # return exp(x0)**2+(gridmin-1)
jacebrowning/comparable
comparable/tools.py
match_similar
python
def match_similar(base, items): finds = list(find_similar(base, items)) if finds: return max(finds, key=base.similarity) # TODO: make O(n) return None
Get the most similar matching item from a list of items. @param base: base item to locate best match @param items: list of items for comparison @return: most similar matching item or None
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/tools.py#L40-L52
[ "def find_similar(base, items):\n \"\"\"Get an iterator of items similar to the base.\n\n @param base: base item to locate best match\n @param items: list of items for comparison\n @return: generator of similar items\n\n \"\"\"\n return (item for item in items if base.similarity(item))\n" ]
"""Functions to utilize lists of Comparable objects.""" def find_equal(base, items): """Get an iterator of items equal to the base. @param base: base item to find equality @param items: list of items for comparison @return: generator of equal items """ return (item for item in items if base.equality(item)) def match_equal(base, items): """Get the first item that is equivalent to the base. @param base: base item to find equality @param items: list of items for comparison @return: first equivalent item or None """ for item in find_equal(base, items): return item return None def find_similar(base, items): """Get an iterator of items similar to the base. @param base: base item to locate best match @param items: list of items for comparison @return: generator of similar items """ return (item for item in items if base.similarity(item)) def duplicates(base, items): """Get an iterator of items similar but not equal to the base. @param base: base item to perform comparison against @param items: list of items to compare to the base @return: generator of items sorted by similarity to the base """ for item in items: if item.similarity(base) and not item.equality(base): yield item def sort(base, items): """Get a sorted list of items ranked in descending similarity. @param base: base item to perform comparison against @param items: list of items to compare to the base @return: list of items sorted by similarity to the base """ return sorted(items, key=base.similarity, reverse=True)
jacebrowning/comparable
comparable/tools.py
duplicates
python
def duplicates(base, items): for item in items: if item.similarity(base) and not item.equality(base): yield item
Get an iterator of items similar but not equal to the base. @param base: base item to perform comparison against @param items: list of items to compare to the base @return: generator of items sorted by similarity to the base
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/tools.py#L55-L65
null
"""Functions to utilize lists of Comparable objects.""" def find_equal(base, items): """Get an iterator of items equal to the base. @param base: base item to find equality @param items: list of items for comparison @return: generator of equal items """ return (item for item in items if base.equality(item)) def match_equal(base, items): """Get the first item that is equivalent to the base. @param base: base item to find equality @param items: list of items for comparison @return: first equivalent item or None """ for item in find_equal(base, items): return item return None def find_similar(base, items): """Get an iterator of items similar to the base. @param base: base item to locate best match @param items: list of items for comparison @return: generator of similar items """ return (item for item in items if base.similarity(item)) def match_similar(base, items): """Get the most similar matching item from a list of items. @param base: base item to locate best match @param items: list of items for comparison @return: most similar matching item or None """ finds = list(find_similar(base, items)) if finds: return max(finds, key=base.similarity) # TODO: make O(n) return None def sort(base, items): """Get a sorted list of items ranked in descending similarity. @param base: base item to perform comparison against @param items: list of items to compare to the base @return: list of items sorted by similarity to the base """ return sorted(items, key=base.similarity, reverse=True)
jacebrowning/comparable
comparable/tools.py
sort
python
def sort(base, items): return sorted(items, key=base.similarity, reverse=True)
Get a sorted list of items ranked in descending similarity. @param base: base item to perform comparison against @param items: list of items to compare to the base @return: list of items sorted by similarity to the base
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/tools.py#L68-L76
null
"""Functions to utilize lists of Comparable objects.""" def find_equal(base, items): """Get an iterator of items equal to the base. @param base: base item to find equality @param items: list of items for comparison @return: generator of equal items """ return (item for item in items if base.equality(item)) def match_equal(base, items): """Get the first item that is equivalent to the base. @param base: base item to find equality @param items: list of items for comparison @return: first equivalent item or None """ for item in find_equal(base, items): return item return None def find_similar(base, items): """Get an iterator of items similar to the base. @param base: base item to locate best match @param items: list of items for comparison @return: generator of similar items """ return (item for item in items if base.similarity(item)) def match_similar(base, items): """Get the most similar matching item from a list of items. @param base: base item to locate best match @param items: list of items for comparison @return: most similar matching item or None """ finds = list(find_similar(base, items)) if finds: return max(finds, key=base.similarity) # TODO: make O(n) return None def duplicates(base, items): """Get an iterator of items similar but not equal to the base. @param base: base item to perform comparison against @param items: list of items to compare to the base @return: generator of items sorted by similarity to the base """ for item in items: if item.similarity(base) and not item.equality(base): yield item
jacebrowning/comparable
comparable/simple.py
Number.similarity
python
def similarity(self, other): numerator, denominator = sorted((self.value, other.value)) try: ratio = float(numerator) / denominator except ZeroDivisionError: ratio = 0.0 if numerator else 1.0 similarity = self.Similarity(ratio) return similarity
Get similarity as a ratio of the two numbers.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/simple.py#L44-L52
[ "def Similarity(self, value=None): # pylint: disable=C0103\n \"\"\"Constructor for new default Similarities.\"\"\"\n if value is None:\n value = 0.0\n return Similarity(value, threshold=self.threshold)\n" ]
class Number(_Simple): """Comparable positive number.""" threshold = 0.999 # 99.9% similar def __init__(self, value): super().__init__(value) if value < 0: raise ValueError("Number objects can only be positive") def equality(self, other): """Get equality using floating point equality.""" return float(self) == float(other)
jacebrowning/comparable
comparable/simple.py
Text.similarity
python
def similarity(self, other): ratio = SequenceMatcher(a=self.value, b=other.value).ratio() similarity = self.Similarity(ratio) return similarity
Get similarity as a ratio of the two texts.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/simple.py#L65-L69
[ "def Similarity(self, value=None): # pylint: disable=C0103\n \"\"\"Constructor for new default Similarities.\"\"\"\n if value is None:\n value = 0.0\n return Similarity(value, threshold=self.threshold)\n" ]
class Text(_Simple): """Comparable generic text.""" threshold = 0.83 # "Hello, world!" ~ "hello world" def equality(self, other): """Get equality using string comparison.""" return str(self) == str(other)
jacebrowning/comparable
comparable/simple.py
TextEnum.similarity
python
def similarity(self, other): ratio = 1.0 if (str(self).lower() == str(other).lower()) else 0.0 similarity = self.Similarity(ratio) return similarity
Get similarity as a discrete ratio (1.0 or 0.0).
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/simple.py#L78-L82
[ "def Similarity(self, value=None): # pylint: disable=C0103\n \"\"\"Constructor for new default Similarities.\"\"\"\n if value is None:\n value = 0.0\n return Similarity(value, threshold=self.threshold)\n" ]
class TextEnum(Text): """Comparable case-insensitive textual enumeration.""" threshold = 1.0 # enumerations must match
jacebrowning/comparable
comparable/simple.py
TextTitle._strip
python
def _strip(text): text = text.strip() text = text.replace(' ', ' ') # remove duplicate spaces text = text.lower() for joiner in TextTitle.JOINERS: text = text.replace(joiner, 'and') for article in TextTitle.ARTICLES: if text.startswith(article + ' '): text = text[len(article) + 1:] break return text
Strip articles/whitespace and remove case.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/simple.py#L100-L111
null
class TextTitle(Text): """Comparable case-insensitive textual titles.""" threshold = 0.93 # "The Cat and the Hat" ~ "cat an' the hat" ARTICLES = 'a', 'an', 'the' # stripped from the front JOINERS = '&', '+' # replaced with 'and' def __init__(self, value): super().__init__(value) self.stripped = self._strip(self.value) logging.debug("stripped %r to %r", self.value, self.stripped) @staticmethod def similarity(self, other): """Get similarity as a ratio of the stripped text.""" logging.debug("comparing %r and %r...", self.stripped, other.stripped) ratio = SequenceMatcher(a=self.stripped, b=other.stripped).ratio() similarity = self.Similarity(ratio) return similarity
jacebrowning/comparable
comparable/simple.py
TextTitle.similarity
python
def similarity(self, other): logging.debug("comparing %r and %r...", self.stripped, other.stripped) ratio = SequenceMatcher(a=self.stripped, b=other.stripped).ratio() similarity = self.Similarity(ratio) return similarity
Get similarity as a ratio of the stripped text.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/simple.py#L113-L118
[ "def Similarity(self, value=None): # pylint: disable=C0103\n \"\"\"Constructor for new default Similarities.\"\"\"\n if value is None:\n value = 0.0\n return Similarity(value, threshold=self.threshold)\n" ]
class TextTitle(Text): """Comparable case-insensitive textual titles.""" threshold = 0.93 # "The Cat and the Hat" ~ "cat an' the hat" ARTICLES = 'a', 'an', 'the' # stripped from the front JOINERS = '&', '+' # replaced with 'and' def __init__(self, value): super().__init__(value) self.stripped = self._strip(self.value) logging.debug("stripped %r to %r", self.value, self.stripped) @staticmethod def _strip(text): """Strip articles/whitespace and remove case.""" text = text.strip() text = text.replace(' ', ' ') # remove duplicate spaces text = text.lower() for joiner in TextTitle.JOINERS: text = text.replace(joiner, 'and') for article in TextTitle.ARTICLES: if text.startswith(article + ' '): text = text[len(article) + 1:] break return text
jacebrowning/comparable
comparable/base.py
equal
python
def equal(obj1, obj2): Comparable.log(obj1, obj2, '==') equality = obj1.equality(obj2) Comparable.log(obj1, obj2, '==', result=equality) return equality
Calculate equality between two (Comparable) objects.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L127-L132
[ "def equality(self, other):\n \"\"\"Compare two objects for equality.\n\n @param self: first object to compare\n @param other: second object to compare\n\n @return: boolean result of comparison\n\n \"\"\"\n # Compare specified attributes for equality\n cname = self.__class__.__name__\n for aname in self.attributes:\n try:\n attr1 = getattr(self, aname)\n attr2 = getattr(other, aname)\n except AttributeError as error:\n logging.debug(\"%s.%s: %s\", cname, aname, error)\n return False\n self.log(attr1, attr2, '==', cname=cname, aname=aname)\n eql = (attr1 == attr2)\n self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql)\n if not eql:\n return False\n\n return True\n", "def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913\n \"\"\"Log the objects being compared and the result.\n\n When no result object is specified, subsequence calls will have an\n increased indentation level. The indentation level is decreased\n once a result object is provided.\n\n @param obj1: first object\n @param obj2: second object\n @param sym: operation being performed ('==' or '%')\n @param cname: name of class (when attributes are being compared)\n @param aname: name of attribute (when attributes are being compared)\n @param result: outcome of comparison\n\n \"\"\"\n fmt = \"{o1} {sym} {o2} : {r}\"\n if cname or aname:\n assert cname and aname # both must be specified\n fmt = \"{c}.{a}: \" + fmt\n\n if result is None:\n result = '...'\n fmt = _Indent.indent(fmt)\n _Indent.more()\n else:\n _Indent.less()\n fmt = _Indent.indent(fmt)\n\n msg = fmt.format(o1=repr(obj1), o2=repr(obj2),\n c=cname, a=aname, sym=sym, r=result)\n logging.info(msg)\n" ]
"""Abstract base class and similarity functions.""" import logging from collections import OrderedDict from abc import ABCMeta, abstractmethod, abstractproperty # pylint: disable=W0611 class _Base(object): # pylint: disable=R0903 """Shared base class.""" def _repr(self, *args, **kwargs): """Return a __repr__ string from the arguments provided to __init__. @param args: list of arguments to __init__ @param kwargs: dictionary of keyword arguments to __init__ @return: __repr__ string """ # Remove unnecessary empty keywords arguments and sort the arguments kwargs = {k: v for k, v in kwargs.items() if v is not None} kwargs = OrderedDict(sorted(kwargs.items())) # Build the __repr__ string pieces args_repr = ', '.join(repr(arg) for arg in args) kwargs_repr = ', '.join(k + '=' + repr(v) for k, v in kwargs.items()) if args_repr and kwargs_repr: kwargs_repr = ', ' + kwargs_repr name = self.__class__.__name__ return "{}({}{})".format(name, args_repr, kwargs_repr) class Similarity(_Base): # pylint: disable=R0903 """Represents the similarity between two objects.""" def __init__(self, value, threshold=1.0): self.value = float(value) self.threshold = float(threshold) def __repr__(self): return self._repr(self.value, threshold=self.threshold) def __str__(self): return "{:.1%} similar".format(self.value) def __eq__(self, other): return abs(float(self) - float(other)) < 0.001 def __ne__(self, other): return not self == other def __lt__(self, other): return float(self) < float(other) def __gt__(self, other): return float(self) > float(other) def __bool__(self): """In boolean scenarios, similarity is True if the threshold is met.""" return self.value >= self.threshold def __float__(self): """In non-boolean scenarios, similarity is treated like a float.""" return self.value def __add__(self, other): return Similarity(self.value + float(other), threshold=self.threshold) def __radd__(self, other): return Similarity(float(other) + self.value, threshold=self.threshold) def __iadd__(self, other): self.value += float(other) return self def __sub__(self, other): return Similarity(self.value - float(other), threshold=self.threshold) def __rsub__(self, other): return Similarity(float(other) - self.value, threshold=self.threshold) def __isub__(self, other): self.value -= float(other) return self def __mul__(self, other): return Similarity(self.value * float(other), threshold=self.threshold) def __rmul__(self, other): return Similarity(float(other) * self.value, threshold=self.threshold) def __imul__(self, other): self.value *= float(other) return self def __abs__(self): return Similarity(abs(self.value), threshold=self.threshold) def __round__(self, digits): return Similarity(round(self.value, digits), threshold=self.threshold) class _Indent(object): """Indent formatter for logging calls.""" level = 0 @classmethod def more(cls): """Increase the indent level.""" cls.level += 1 @classmethod def less(cls): """Decrease the indent level.""" cls.level = max(cls.level - 1, 0) @classmethod def indent(cls, fmt): """Get a new format string with indentation.""" return '| ' * cls.level + fmt def similar(obj1, obj2): """Calculate similarity between two (Comparable) objects.""" Comparable.log(obj1, obj2, '%') similarity = obj1.similarity(obj2) Comparable.log(obj1, obj2, '%', result=similarity) return similarity class Comparable(_Base, metaclass=ABCMeta): """Abstract Base Class for objects that are comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. Both types of subclasses may also override the 'threshold' attribute to change the default similarity threshold. """ def __eq__(self, other): """Map the '==' operator to be a shortcut for "equality".""" return equal(self, other) def __ne__(self, other): return not self == other def __mod__(self, other): """Map the '%' operator to be a shortcut for "similarity".""" return similar(self, other) @abstractproperty def attributes(self): # pragma: no cover, abstract """Get an attribute {name: weight} dictionary for comparisons.""" return {} threshold = 1.0 # ratio for two objects to be considered "similar" @abstractmethod def equality(self, other): """Compare two objects for equality. @param self: first object to compare @param other: second object to compare @return: boolean result of comparison """ # Compare specified attributes for equality cname = self.__class__.__name__ for aname in self.attributes: try: attr1 = getattr(self, aname) attr2 = getattr(other, aname) except AttributeError as error: logging.debug("%s.%s: %s", cname, aname, error) return False self.log(attr1, attr2, '==', cname=cname, aname=aname) eql = (attr1 == attr2) self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql) if not eql: return False return True @abstractmethod def similarity(self, other): """Compare two objects for similarity. @param self: first object to compare @param other: second object to compare @return: L{Similarity} result of comparison """ sim = self.Similarity() total = 0.0 # Calculate similarity ratio for each attribute cname = self.__class__.__name__ for aname, weight in self.attributes.items(): attr1 = getattr(self, aname, None) attr2 = getattr(other, aname, None) self.log(attr1, attr2, '%', cname=cname, aname=aname) # Similarity is ignored if None on both objects if attr1 is None and attr2 is None: self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes are both None") continue # Similarity is 0 if either attribute is non-Comparable if not all((isinstance(attr1, Comparable), isinstance(attr2, Comparable))): self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes not Comparable") total += weight continue # Calculate similarity between the attributes attr_sim = (attr1 % attr2) self.log(attr1, attr2, '%', cname=cname, aname=aname, result=attr_sim) # Add the similarity to the total sim += attr_sim * weight total += weight # Scale the similarity so the total is 1.0 if total: sim *= (1.0 / total) return sim def Similarity(self, value=None): # pylint: disable=C0103 """Constructor for new default Similarities.""" if value is None: value = 0.0 return Similarity(value, threshold=self.threshold) @staticmethod def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913 """Log the objects being compared and the result. When no result object is specified, subsequence calls will have an increased indentation level. The indentation level is decreased once a result object is provided. @param obj1: first object @param obj2: second object @param sym: operation being performed ('==' or '%') @param cname: name of class (when attributes are being compared) @param aname: name of attribute (when attributes are being compared) @param result: outcome of comparison """ fmt = "{o1} {sym} {o2} : {r}" if cname or aname: assert cname and aname # both must be specified fmt = "{c}.{a}: " + fmt if result is None: result = '...' fmt = _Indent.indent(fmt) _Indent.more() else: _Indent.less() fmt = _Indent.indent(fmt) msg = fmt.format(o1=repr(obj1), o2=repr(obj2), c=cname, a=aname, sym=sym, r=result) logging.info(msg) class SimpleComparable(Comparable): # pylint: disable=W0223 """Abstract Base Class for objects that are directly comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. They may also override the 'threshold' attribute to change the default similarity threshold. """ @property def attributes(self): # pragma: no cover, abstract """A simple comparable does not use the attributes property.""" raise AttributeError() class CompoundComparable(Comparable): # pylint: disable=W0223 """Abstract Base Class for objects that are comparable by attributes. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. They may also override the 'threshold' attribute to change the default similarity threshold. """ def equality(self, other): """A compound comparable's equality is based on attributes.""" return super().equality(other) def similarity(self, other): """A compound comparable's similarity is based on attributes.""" return super().similarity(other)
jacebrowning/comparable
comparable/base.py
similar
python
def similar(obj1, obj2): Comparable.log(obj1, obj2, '%') similarity = obj1.similarity(obj2) Comparable.log(obj1, obj2, '%', result=similarity) return similarity
Calculate similarity between two (Comparable) objects.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L135-L140
[ "def similarity(self, other):\n \"\"\"Compare two objects for similarity.\n\n @param self: first object to compare\n @param other: second object to compare\n\n @return: L{Similarity} result of comparison\n\n \"\"\"\n sim = self.Similarity()\n total = 0.0\n\n # Calculate similarity ratio for each attribute\n cname = self.__class__.__name__\n for aname, weight in self.attributes.items():\n\n attr1 = getattr(self, aname, None)\n attr2 = getattr(other, aname, None)\n self.log(attr1, attr2, '%', cname=cname, aname=aname)\n\n # Similarity is ignored if None on both objects\n if attr1 is None and attr2 is None:\n self.log(attr1, attr2, '%', cname=cname, aname=aname,\n result=\"attributes are both None\")\n continue\n\n # Similarity is 0 if either attribute is non-Comparable\n if not all((isinstance(attr1, Comparable),\n isinstance(attr2, Comparable))):\n self.log(attr1, attr2, '%', cname=cname, aname=aname,\n result=\"attributes not Comparable\")\n total += weight\n continue\n\n # Calculate similarity between the attributes\n attr_sim = (attr1 % attr2)\n self.log(attr1, attr2, '%', cname=cname, aname=aname,\n result=attr_sim)\n\n # Add the similarity to the total\n sim += attr_sim * weight\n total += weight\n\n # Scale the similarity so the total is 1.0\n if total:\n sim *= (1.0 / total)\n\n return sim\n", "def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913\n \"\"\"Log the objects being compared and the result.\n\n When no result object is specified, subsequence calls will have an\n increased indentation level. The indentation level is decreased\n once a result object is provided.\n\n @param obj1: first object\n @param obj2: second object\n @param sym: operation being performed ('==' or '%')\n @param cname: name of class (when attributes are being compared)\n @param aname: name of attribute (when attributes are being compared)\n @param result: outcome of comparison\n\n \"\"\"\n fmt = \"{o1} {sym} {o2} : {r}\"\n if cname or aname:\n assert cname and aname # both must be specified\n fmt = \"{c}.{a}: \" + fmt\n\n if result is None:\n result = '...'\n fmt = _Indent.indent(fmt)\n _Indent.more()\n else:\n _Indent.less()\n fmt = _Indent.indent(fmt)\n\n msg = fmt.format(o1=repr(obj1), o2=repr(obj2),\n c=cname, a=aname, sym=sym, r=result)\n logging.info(msg)\n" ]
"""Abstract base class and similarity functions.""" import logging from collections import OrderedDict from abc import ABCMeta, abstractmethod, abstractproperty # pylint: disable=W0611 class _Base(object): # pylint: disable=R0903 """Shared base class.""" def _repr(self, *args, **kwargs): """Return a __repr__ string from the arguments provided to __init__. @param args: list of arguments to __init__ @param kwargs: dictionary of keyword arguments to __init__ @return: __repr__ string """ # Remove unnecessary empty keywords arguments and sort the arguments kwargs = {k: v for k, v in kwargs.items() if v is not None} kwargs = OrderedDict(sorted(kwargs.items())) # Build the __repr__ string pieces args_repr = ', '.join(repr(arg) for arg in args) kwargs_repr = ', '.join(k + '=' + repr(v) for k, v in kwargs.items()) if args_repr and kwargs_repr: kwargs_repr = ', ' + kwargs_repr name = self.__class__.__name__ return "{}({}{})".format(name, args_repr, kwargs_repr) class Similarity(_Base): # pylint: disable=R0903 """Represents the similarity between two objects.""" def __init__(self, value, threshold=1.0): self.value = float(value) self.threshold = float(threshold) def __repr__(self): return self._repr(self.value, threshold=self.threshold) def __str__(self): return "{:.1%} similar".format(self.value) def __eq__(self, other): return abs(float(self) - float(other)) < 0.001 def __ne__(self, other): return not self == other def __lt__(self, other): return float(self) < float(other) def __gt__(self, other): return float(self) > float(other) def __bool__(self): """In boolean scenarios, similarity is True if the threshold is met.""" return self.value >= self.threshold def __float__(self): """In non-boolean scenarios, similarity is treated like a float.""" return self.value def __add__(self, other): return Similarity(self.value + float(other), threshold=self.threshold) def __radd__(self, other): return Similarity(float(other) + self.value, threshold=self.threshold) def __iadd__(self, other): self.value += float(other) return self def __sub__(self, other): return Similarity(self.value - float(other), threshold=self.threshold) def __rsub__(self, other): return Similarity(float(other) - self.value, threshold=self.threshold) def __isub__(self, other): self.value -= float(other) return self def __mul__(self, other): return Similarity(self.value * float(other), threshold=self.threshold) def __rmul__(self, other): return Similarity(float(other) * self.value, threshold=self.threshold) def __imul__(self, other): self.value *= float(other) return self def __abs__(self): return Similarity(abs(self.value), threshold=self.threshold) def __round__(self, digits): return Similarity(round(self.value, digits), threshold=self.threshold) class _Indent(object): """Indent formatter for logging calls.""" level = 0 @classmethod def more(cls): """Increase the indent level.""" cls.level += 1 @classmethod def less(cls): """Decrease the indent level.""" cls.level = max(cls.level - 1, 0) @classmethod def indent(cls, fmt): """Get a new format string with indentation.""" return '| ' * cls.level + fmt def equal(obj1, obj2): """Calculate equality between two (Comparable) objects.""" Comparable.log(obj1, obj2, '==') equality = obj1.equality(obj2) Comparable.log(obj1, obj2, '==', result=equality) return equality class Comparable(_Base, metaclass=ABCMeta): """Abstract Base Class for objects that are comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. Both types of subclasses may also override the 'threshold' attribute to change the default similarity threshold. """ def __eq__(self, other): """Map the '==' operator to be a shortcut for "equality".""" return equal(self, other) def __ne__(self, other): return not self == other def __mod__(self, other): """Map the '%' operator to be a shortcut for "similarity".""" return similar(self, other) @abstractproperty def attributes(self): # pragma: no cover, abstract """Get an attribute {name: weight} dictionary for comparisons.""" return {} threshold = 1.0 # ratio for two objects to be considered "similar" @abstractmethod def equality(self, other): """Compare two objects for equality. @param self: first object to compare @param other: second object to compare @return: boolean result of comparison """ # Compare specified attributes for equality cname = self.__class__.__name__ for aname in self.attributes: try: attr1 = getattr(self, aname) attr2 = getattr(other, aname) except AttributeError as error: logging.debug("%s.%s: %s", cname, aname, error) return False self.log(attr1, attr2, '==', cname=cname, aname=aname) eql = (attr1 == attr2) self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql) if not eql: return False return True @abstractmethod def similarity(self, other): """Compare two objects for similarity. @param self: first object to compare @param other: second object to compare @return: L{Similarity} result of comparison """ sim = self.Similarity() total = 0.0 # Calculate similarity ratio for each attribute cname = self.__class__.__name__ for aname, weight in self.attributes.items(): attr1 = getattr(self, aname, None) attr2 = getattr(other, aname, None) self.log(attr1, attr2, '%', cname=cname, aname=aname) # Similarity is ignored if None on both objects if attr1 is None and attr2 is None: self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes are both None") continue # Similarity is 0 if either attribute is non-Comparable if not all((isinstance(attr1, Comparable), isinstance(attr2, Comparable))): self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes not Comparable") total += weight continue # Calculate similarity between the attributes attr_sim = (attr1 % attr2) self.log(attr1, attr2, '%', cname=cname, aname=aname, result=attr_sim) # Add the similarity to the total sim += attr_sim * weight total += weight # Scale the similarity so the total is 1.0 if total: sim *= (1.0 / total) return sim def Similarity(self, value=None): # pylint: disable=C0103 """Constructor for new default Similarities.""" if value is None: value = 0.0 return Similarity(value, threshold=self.threshold) @staticmethod def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913 """Log the objects being compared and the result. When no result object is specified, subsequence calls will have an increased indentation level. The indentation level is decreased once a result object is provided. @param obj1: first object @param obj2: second object @param sym: operation being performed ('==' or '%') @param cname: name of class (when attributes are being compared) @param aname: name of attribute (when attributes are being compared) @param result: outcome of comparison """ fmt = "{o1} {sym} {o2} : {r}" if cname or aname: assert cname and aname # both must be specified fmt = "{c}.{a}: " + fmt if result is None: result = '...' fmt = _Indent.indent(fmt) _Indent.more() else: _Indent.less() fmt = _Indent.indent(fmt) msg = fmt.format(o1=repr(obj1), o2=repr(obj2), c=cname, a=aname, sym=sym, r=result) logging.info(msg) class SimpleComparable(Comparable): # pylint: disable=W0223 """Abstract Base Class for objects that are directly comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. They may also override the 'threshold' attribute to change the default similarity threshold. """ @property def attributes(self): # pragma: no cover, abstract """A simple comparable does not use the attributes property.""" raise AttributeError() class CompoundComparable(Comparable): # pylint: disable=W0223 """Abstract Base Class for objects that are comparable by attributes. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. They may also override the 'threshold' attribute to change the default similarity threshold. """ def equality(self, other): """A compound comparable's equality is based on attributes.""" return super().equality(other) def similarity(self, other): """A compound comparable's similarity is based on attributes.""" return super().similarity(other)
jacebrowning/comparable
comparable/base.py
_Base._repr
python
def _repr(self, *args, **kwargs): # Remove unnecessary empty keywords arguments and sort the arguments kwargs = {k: v for k, v in kwargs.items() if v is not None} kwargs = OrderedDict(sorted(kwargs.items())) # Build the __repr__ string pieces args_repr = ', '.join(repr(arg) for arg in args) kwargs_repr = ', '.join(k + '=' + repr(v) for k, v in kwargs.items()) if args_repr and kwargs_repr: kwargs_repr = ', ' + kwargs_repr name = self.__class__.__name__ return "{}({}{})".format(name, args_repr, kwargs_repr)
Return a __repr__ string from the arguments provided to __init__. @param args: list of arguments to __init__ @param kwargs: dictionary of keyword arguments to __init__ @return: __repr__ string
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L12-L31
null
class _Base(object): # pylint: disable=R0903 """Shared base class."""
jacebrowning/comparable
comparable/base.py
Comparable.equality
python
def equality(self, other): # Compare specified attributes for equality cname = self.__class__.__name__ for aname in self.attributes: try: attr1 = getattr(self, aname) attr2 = getattr(other, aname) except AttributeError as error: logging.debug("%s.%s: %s", cname, aname, error) return False self.log(attr1, attr2, '==', cname=cname, aname=aname) eql = (attr1 == attr2) self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql) if not eql: return False return True
Compare two objects for equality. @param self: first object to compare @param other: second object to compare @return: boolean result of comparison
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L179-L203
[ "def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913\n \"\"\"Log the objects being compared and the result.\n\n When no result object is specified, subsequence calls will have an\n increased indentation level. The indentation level is decreased\n once a result object is provided.\n\n @param obj1: first object\n @param obj2: second object\n @param sym: operation being performed ('==' or '%')\n @param cname: name of class (when attributes are being compared)\n @param aname: name of attribute (when attributes are being compared)\n @param result: outcome of comparison\n\n \"\"\"\n fmt = \"{o1} {sym} {o2} : {r}\"\n if cname or aname:\n assert cname and aname # both must be specified\n fmt = \"{c}.{a}: \" + fmt\n\n if result is None:\n result = '...'\n fmt = _Indent.indent(fmt)\n _Indent.more()\n else:\n _Indent.less()\n fmt = _Indent.indent(fmt)\n\n msg = fmt.format(o1=repr(obj1), o2=repr(obj2),\n c=cname, a=aname, sym=sym, r=result)\n logging.info(msg)\n" ]
class Comparable(_Base, metaclass=ABCMeta): """Abstract Base Class for objects that are comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. Both types of subclasses may also override the 'threshold' attribute to change the default similarity threshold. """ def __eq__(self, other): """Map the '==' operator to be a shortcut for "equality".""" return equal(self, other) def __ne__(self, other): return not self == other def __mod__(self, other): """Map the '%' operator to be a shortcut for "similarity".""" return similar(self, other) @abstractproperty def attributes(self): # pragma: no cover, abstract """Get an attribute {name: weight} dictionary for comparisons.""" return {} threshold = 1.0 # ratio for two objects to be considered "similar" @abstractmethod @abstractmethod def similarity(self, other): """Compare two objects for similarity. @param self: first object to compare @param other: second object to compare @return: L{Similarity} result of comparison """ sim = self.Similarity() total = 0.0 # Calculate similarity ratio for each attribute cname = self.__class__.__name__ for aname, weight in self.attributes.items(): attr1 = getattr(self, aname, None) attr2 = getattr(other, aname, None) self.log(attr1, attr2, '%', cname=cname, aname=aname) # Similarity is ignored if None on both objects if attr1 is None and attr2 is None: self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes are both None") continue # Similarity is 0 if either attribute is non-Comparable if not all((isinstance(attr1, Comparable), isinstance(attr2, Comparable))): self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes not Comparable") total += weight continue # Calculate similarity between the attributes attr_sim = (attr1 % attr2) self.log(attr1, attr2, '%', cname=cname, aname=aname, result=attr_sim) # Add the similarity to the total sim += attr_sim * weight total += weight # Scale the similarity so the total is 1.0 if total: sim *= (1.0 / total) return sim def Similarity(self, value=None): # pylint: disable=C0103 """Constructor for new default Similarities.""" if value is None: value = 0.0 return Similarity(value, threshold=self.threshold) @staticmethod def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913 """Log the objects being compared and the result. When no result object is specified, subsequence calls will have an increased indentation level. The indentation level is decreased once a result object is provided. @param obj1: first object @param obj2: second object @param sym: operation being performed ('==' or '%') @param cname: name of class (when attributes are being compared) @param aname: name of attribute (when attributes are being compared) @param result: outcome of comparison """ fmt = "{o1} {sym} {o2} : {r}" if cname or aname: assert cname and aname # both must be specified fmt = "{c}.{a}: " + fmt if result is None: result = '...' fmt = _Indent.indent(fmt) _Indent.more() else: _Indent.less() fmt = _Indent.indent(fmt) msg = fmt.format(o1=repr(obj1), o2=repr(obj2), c=cname, a=aname, sym=sym, r=result) logging.info(msg)
jacebrowning/comparable
comparable/base.py
Comparable.similarity
python
def similarity(self, other): sim = self.Similarity() total = 0.0 # Calculate similarity ratio for each attribute cname = self.__class__.__name__ for aname, weight in self.attributes.items(): attr1 = getattr(self, aname, None) attr2 = getattr(other, aname, None) self.log(attr1, attr2, '%', cname=cname, aname=aname) # Similarity is ignored if None on both objects if attr1 is None and attr2 is None: self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes are both None") continue # Similarity is 0 if either attribute is non-Comparable if not all((isinstance(attr1, Comparable), isinstance(attr2, Comparable))): self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes not Comparable") total += weight continue # Calculate similarity between the attributes attr_sim = (attr1 % attr2) self.log(attr1, attr2, '%', cname=cname, aname=aname, result=attr_sim) # Add the similarity to the total sim += attr_sim * weight total += weight # Scale the similarity so the total is 1.0 if total: sim *= (1.0 / total) return sim
Compare two objects for similarity. @param self: first object to compare @param other: second object to compare @return: L{Similarity} result of comparison
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L206-L253
[ "def Similarity(self, value=None): # pylint: disable=C0103\n \"\"\"Constructor for new default Similarities.\"\"\"\n if value is None:\n value = 0.0\n return Similarity(value, threshold=self.threshold)\n", "def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913\n \"\"\"Log the objects being compared and the result.\n\n When no result object is specified, subsequence calls will have an\n increased indentation level. The indentation level is decreased\n once a result object is provided.\n\n @param obj1: first object\n @param obj2: second object\n @param sym: operation being performed ('==' or '%')\n @param cname: name of class (when attributes are being compared)\n @param aname: name of attribute (when attributes are being compared)\n @param result: outcome of comparison\n\n \"\"\"\n fmt = \"{o1} {sym} {o2} : {r}\"\n if cname or aname:\n assert cname and aname # both must be specified\n fmt = \"{c}.{a}: \" + fmt\n\n if result is None:\n result = '...'\n fmt = _Indent.indent(fmt)\n _Indent.more()\n else:\n _Indent.less()\n fmt = _Indent.indent(fmt)\n\n msg = fmt.format(o1=repr(obj1), o2=repr(obj2),\n c=cname, a=aname, sym=sym, r=result)\n logging.info(msg)\n" ]
class Comparable(_Base, metaclass=ABCMeta): """Abstract Base Class for objects that are comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. Both types of subclasses may also override the 'threshold' attribute to change the default similarity threshold. """ def __eq__(self, other): """Map the '==' operator to be a shortcut for "equality".""" return equal(self, other) def __ne__(self, other): return not self == other def __mod__(self, other): """Map the '%' operator to be a shortcut for "similarity".""" return similar(self, other) @abstractproperty def attributes(self): # pragma: no cover, abstract """Get an attribute {name: weight} dictionary for comparisons.""" return {} threshold = 1.0 # ratio for two objects to be considered "similar" @abstractmethod def equality(self, other): """Compare two objects for equality. @param self: first object to compare @param other: second object to compare @return: boolean result of comparison """ # Compare specified attributes for equality cname = self.__class__.__name__ for aname in self.attributes: try: attr1 = getattr(self, aname) attr2 = getattr(other, aname) except AttributeError as error: logging.debug("%s.%s: %s", cname, aname, error) return False self.log(attr1, attr2, '==', cname=cname, aname=aname) eql = (attr1 == attr2) self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql) if not eql: return False return True @abstractmethod def Similarity(self, value=None): # pylint: disable=C0103 """Constructor for new default Similarities.""" if value is None: value = 0.0 return Similarity(value, threshold=self.threshold) @staticmethod def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913 """Log the objects being compared and the result. When no result object is specified, subsequence calls will have an increased indentation level. The indentation level is decreased once a result object is provided. @param obj1: first object @param obj2: second object @param sym: operation being performed ('==' or '%') @param cname: name of class (when attributes are being compared) @param aname: name of attribute (when attributes are being compared) @param result: outcome of comparison """ fmt = "{o1} {sym} {o2} : {r}" if cname or aname: assert cname and aname # both must be specified fmt = "{c}.{a}: " + fmt if result is None: result = '...' fmt = _Indent.indent(fmt) _Indent.more() else: _Indent.less() fmt = _Indent.indent(fmt) msg = fmt.format(o1=repr(obj1), o2=repr(obj2), c=cname, a=aname, sym=sym, r=result) logging.info(msg)
jacebrowning/comparable
comparable/base.py
Comparable.Similarity
python
def Similarity(self, value=None): # pylint: disable=C0103 if value is None: value = 0.0 return Similarity(value, threshold=self.threshold)
Constructor for new default Similarities.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L255-L259
null
class Comparable(_Base, metaclass=ABCMeta): """Abstract Base Class for objects that are comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. Both types of subclasses may also override the 'threshold' attribute to change the default similarity threshold. """ def __eq__(self, other): """Map the '==' operator to be a shortcut for "equality".""" return equal(self, other) def __ne__(self, other): return not self == other def __mod__(self, other): """Map the '%' operator to be a shortcut for "similarity".""" return similar(self, other) @abstractproperty def attributes(self): # pragma: no cover, abstract """Get an attribute {name: weight} dictionary for comparisons.""" return {} threshold = 1.0 # ratio for two objects to be considered "similar" @abstractmethod def equality(self, other): """Compare two objects for equality. @param self: first object to compare @param other: second object to compare @return: boolean result of comparison """ # Compare specified attributes for equality cname = self.__class__.__name__ for aname in self.attributes: try: attr1 = getattr(self, aname) attr2 = getattr(other, aname) except AttributeError as error: logging.debug("%s.%s: %s", cname, aname, error) return False self.log(attr1, attr2, '==', cname=cname, aname=aname) eql = (attr1 == attr2) self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql) if not eql: return False return True @abstractmethod def similarity(self, other): """Compare two objects for similarity. @param self: first object to compare @param other: second object to compare @return: L{Similarity} result of comparison """ sim = self.Similarity() total = 0.0 # Calculate similarity ratio for each attribute cname = self.__class__.__name__ for aname, weight in self.attributes.items(): attr1 = getattr(self, aname, None) attr2 = getattr(other, aname, None) self.log(attr1, attr2, '%', cname=cname, aname=aname) # Similarity is ignored if None on both objects if attr1 is None and attr2 is None: self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes are both None") continue # Similarity is 0 if either attribute is non-Comparable if not all((isinstance(attr1, Comparable), isinstance(attr2, Comparable))): self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes not Comparable") total += weight continue # Calculate similarity between the attributes attr_sim = (attr1 % attr2) self.log(attr1, attr2, '%', cname=cname, aname=aname, result=attr_sim) # Add the similarity to the total sim += attr_sim * weight total += weight # Scale the similarity so the total is 1.0 if total: sim *= (1.0 / total) return sim @staticmethod def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913 """Log the objects being compared and the result. When no result object is specified, subsequence calls will have an increased indentation level. The indentation level is decreased once a result object is provided. @param obj1: first object @param obj2: second object @param sym: operation being performed ('==' or '%') @param cname: name of class (when attributes are being compared) @param aname: name of attribute (when attributes are being compared) @param result: outcome of comparison """ fmt = "{o1} {sym} {o2} : {r}" if cname or aname: assert cname and aname # both must be specified fmt = "{c}.{a}: " + fmt if result is None: result = '...' fmt = _Indent.indent(fmt) _Indent.more() else: _Indent.less() fmt = _Indent.indent(fmt) msg = fmt.format(o1=repr(obj1), o2=repr(obj2), c=cname, a=aname, sym=sym, r=result) logging.info(msg)
jacebrowning/comparable
comparable/base.py
Comparable.log
python
def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913 fmt = "{o1} {sym} {o2} : {r}" if cname or aname: assert cname and aname # both must be specified fmt = "{c}.{a}: " + fmt if result is None: result = '...' fmt = _Indent.indent(fmt) _Indent.more() else: _Indent.less() fmt = _Indent.indent(fmt) msg = fmt.format(o1=repr(obj1), o2=repr(obj2), c=cname, a=aname, sym=sym, r=result) logging.info(msg)
Log the objects being compared and the result. When no result object is specified, subsequence calls will have an increased indentation level. The indentation level is decreased once a result object is provided. @param obj1: first object @param obj2: second object @param sym: operation being performed ('==' or '%') @param cname: name of class (when attributes are being compared) @param aname: name of attribute (when attributes are being compared) @param result: outcome of comparison
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/base.py#L262-L292
[ "def more(cls):\n \"\"\"Increase the indent level.\"\"\"\n cls.level += 1\n", "def less(cls):\n \"\"\"Decrease the indent level.\"\"\"\n cls.level = max(cls.level - 1, 0)\n", "def indent(cls, fmt):\n \"\"\"Get a new format string with indentation.\"\"\"\n return '| ' * cls.level + fmt\n" ]
class Comparable(_Base, metaclass=ABCMeta): """Abstract Base Class for objects that are comparable. Subclasses directly comparable must override the 'equality' and 'similarity' methods to return a bool and 'Similarity' object, respectively. Subclasses comparable by attributes must override the 'attributes' property to define which (Comparable) attributes should be considered. Both types of subclasses may also override the 'threshold' attribute to change the default similarity threshold. """ def __eq__(self, other): """Map the '==' operator to be a shortcut for "equality".""" return equal(self, other) def __ne__(self, other): return not self == other def __mod__(self, other): """Map the '%' operator to be a shortcut for "similarity".""" return similar(self, other) @abstractproperty def attributes(self): # pragma: no cover, abstract """Get an attribute {name: weight} dictionary for comparisons.""" return {} threshold = 1.0 # ratio for two objects to be considered "similar" @abstractmethod def equality(self, other): """Compare two objects for equality. @param self: first object to compare @param other: second object to compare @return: boolean result of comparison """ # Compare specified attributes for equality cname = self.__class__.__name__ for aname in self.attributes: try: attr1 = getattr(self, aname) attr2 = getattr(other, aname) except AttributeError as error: logging.debug("%s.%s: %s", cname, aname, error) return False self.log(attr1, attr2, '==', cname=cname, aname=aname) eql = (attr1 == attr2) self.log(attr1, attr2, '==', cname=cname, aname=aname, result=eql) if not eql: return False return True @abstractmethod def similarity(self, other): """Compare two objects for similarity. @param self: first object to compare @param other: second object to compare @return: L{Similarity} result of comparison """ sim = self.Similarity() total = 0.0 # Calculate similarity ratio for each attribute cname = self.__class__.__name__ for aname, weight in self.attributes.items(): attr1 = getattr(self, aname, None) attr2 = getattr(other, aname, None) self.log(attr1, attr2, '%', cname=cname, aname=aname) # Similarity is ignored if None on both objects if attr1 is None and attr2 is None: self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes are both None") continue # Similarity is 0 if either attribute is non-Comparable if not all((isinstance(attr1, Comparable), isinstance(attr2, Comparable))): self.log(attr1, attr2, '%', cname=cname, aname=aname, result="attributes not Comparable") total += weight continue # Calculate similarity between the attributes attr_sim = (attr1 % attr2) self.log(attr1, attr2, '%', cname=cname, aname=aname, result=attr_sim) # Add the similarity to the total sim += attr_sim * weight total += weight # Scale the similarity so the total is 1.0 if total: sim *= (1.0 / total) return sim def Similarity(self, value=None): # pylint: disable=C0103 """Constructor for new default Similarities.""" if value is None: value = 0.0 return Similarity(value, threshold=self.threshold) @staticmethod
jacebrowning/comparable
comparable/compound.py
Group.equality
python
def equality(self, other): if not len(self) == len(other): return False return super().equality(other)
Calculate equality based on equality of all group items.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/compound.py#L42-L46
[ "def equality(self, other):\n \"\"\"A compound comparable's equality is based on attributes.\"\"\"\n return super().equality(other)\n" ]
class Group(CompoundComparable): # pylint: disable=W0223 """Comparable list of Comparable items.""" attributes = None # created dynamically def __init__(self, items): self.items = items names = ("item{0}".format(n + 1) for n in range(len(items))) self.attributes = {name: 1 for name in names} def __repr__(self): return self._repr(self.items) def __getattr__(self, name): """Allow self.items[<i>] to be accessed as self.item<i+1>.""" if name.startswith('item'): try: index = int(name[4:]) - 1 # "item<n>" -> <n>-1 return self[index] except ValueError: logging.debug("%s is not in the form 'item<n>'", name) except IndexError: logging.debug("item index %s is out of range", index) raise AttributeError def __len__(self): return len(self.items) def __getitem__(self, index): return self.items[index] def similarity(self, other): """Calculate similarity based on best matching permutation of items.""" # Select the longer list as the basis for comparison if len(self.items) > len(other.items): first, second = self, other else: first, second = other, self items = list(first.items) # backup items list length = len(items) sim = self.Similarity(0.0 if length else 1.0) # Calculate the similarity for each permutation of items cname = self.__class__.__name__ for num, perm in enumerate(permutations(items, length), start=1): first.items = perm aname = 'items-p{}'.format(num) self.log(first, second, '%', cname=cname, aname=aname) permutation_sim = super(Group, first).similarity(second) self.log(first, second, '%', cname=cname, aname=aname, result=permutation_sim) sim = max(sim, permutation_sim) logging.debug("highest similarity: %s", sim) first.items = items # restore original items list return sim
jacebrowning/comparable
comparable/compound.py
Group.similarity
python
def similarity(self, other): # Select the longer list as the basis for comparison if len(self.items) > len(other.items): first, second = self, other else: first, second = other, self items = list(first.items) # backup items list length = len(items) sim = self.Similarity(0.0 if length else 1.0) # Calculate the similarity for each permutation of items cname = self.__class__.__name__ for num, perm in enumerate(permutations(items, length), start=1): first.items = perm aname = 'items-p{}'.format(num) self.log(first, second, '%', cname=cname, aname=aname) permutation_sim = super(Group, first).similarity(second) self.log(first, second, '%', cname=cname, aname=aname, result=permutation_sim) sim = max(sim, permutation_sim) logging.debug("highest similarity: %s", sim) first.items = items # restore original items list return sim
Calculate similarity based on best matching permutation of items.
train
https://github.com/jacebrowning/comparable/blob/48455e613650e22412d31109681368fcc479298d/comparable/compound.py#L48-L75
[ "def Similarity(self, value=None): # pylint: disable=C0103\n \"\"\"Constructor for new default Similarities.\"\"\"\n if value is None:\n value = 0.0\n return Similarity(value, threshold=self.threshold)\n", "def log(obj1, obj2, sym, cname=None, aname=None, result=None): # pylint: disable=R0913\n \"\"\"Log the objects being compared and the result.\n\n When no result object is specified, subsequence calls will have an\n increased indentation level. The indentation level is decreased\n once a result object is provided.\n\n @param obj1: first object\n @param obj2: second object\n @param sym: operation being performed ('==' or '%')\n @param cname: name of class (when attributes are being compared)\n @param aname: name of attribute (when attributes are being compared)\n @param result: outcome of comparison\n\n \"\"\"\n fmt = \"{o1} {sym} {o2} : {r}\"\n if cname or aname:\n assert cname and aname # both must be specified\n fmt = \"{c}.{a}: \" + fmt\n\n if result is None:\n result = '...'\n fmt = _Indent.indent(fmt)\n _Indent.more()\n else:\n _Indent.less()\n fmt = _Indent.indent(fmt)\n\n msg = fmt.format(o1=repr(obj1), o2=repr(obj2),\n c=cname, a=aname, sym=sym, r=result)\n logging.info(msg)\n", "def similarity(self, other):\n \"\"\"A compound comparable's similarity is based on attributes.\"\"\"\n return super().similarity(other)\n" ]
class Group(CompoundComparable): # pylint: disable=W0223 """Comparable list of Comparable items.""" attributes = None # created dynamically def __init__(self, items): self.items = items names = ("item{0}".format(n + 1) for n in range(len(items))) self.attributes = {name: 1 for name in names} def __repr__(self): return self._repr(self.items) def __getattr__(self, name): """Allow self.items[<i>] to be accessed as self.item<i+1>.""" if name.startswith('item'): try: index = int(name[4:]) - 1 # "item<n>" -> <n>-1 return self[index] except ValueError: logging.debug("%s is not in the form 'item<n>'", name) except IndexError: logging.debug("item index %s is out of range", index) raise AttributeError def __len__(self): return len(self.items) def __getitem__(self, index): return self.items[index] def equality(self, other): """Calculate equality based on equality of all group items.""" if not len(self) == len(other): return False return super().equality(other)
timothydmorton/simpledist
simpledist/distributions.py
double_lorgauss
python
def double_lorgauss(x,p): mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi
Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L615-L664
null
from __future__ import absolute_import, division, print_function __author__ = 'Timothy D. Morton <tim.morton@gmail.com>' """ Defines objects useful for describing probability distributions. """ import numpy as np import matplotlib.pyplot as plt import logging from scipy.interpolate import UnivariateSpline as interpolate from scipy.integrate import quad import numpy.random as rand from scipy.special import erf from scipy.optimize import leastsq import pandas as pd from plotutils import setfig from .kde import KDE #figure this generic loading thing out; draft stage currently def load_distribution(filename,path=''): fns = pd.read_hdf(filename,path) store = pd.HDFStore(filename) if '{}/samples'.format(path) in store: samples = pd.read_hdf(filename,path+'/samples') samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0) cdf = interpolate(fns['vals'],fns['cdf'],s=0) attrs = store.get_storer('{}/fns'.format(path)).attrs keywords = attrs.keywords t = attrs.disttype store.close() return t.__init__() class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs) class Distribution_FromH5(Distribution): """Creates a Distribution object from one saved to an HDF file. File must have a `DataFrame` saved under [path]/fns in the .h5 file, containing 'vals', 'pdf', and 'cdf' columns. If samples are saved in the HDF storer, then they will be restored to this object; so will any saved keyword attributes. These appropriate .h5 files will be created by a call to the `save_hdf` method of the generic `Distribution` class. Parameters ---------- filename : string .h5 file where the distribution is saved. path : string, optional Path within the .h5 file where the distribution is saved. By default this will be the root level, but can be anywhere. kwargs Keyword arguments are passed to the `Distribution` constructor. """ def __init__(self,filename,path='',**kwargs): store = pd.HDFStore(filename,'r') fns = store[path+'/fns'] if '{}/samples'.format(path) in store: samples = store[path+'/samples'] self.samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0,k=1) #check to see if tabulated CDF is monotonically increasing d_cdf = fns['cdf'][1:] - fns['cdf'][:-1] if np.any(d_cdf < 0): logging.warning('tabulated CDF in {} is not strictly increasing. Recalculating CDF from PDF'.format(filename)) cdf = None #in this case, just recalc cdf from pdf else: cdf = interpolate(fns['vals'],fns['cdf'],s=0,k=1) Distribution.__init__(self,pdf,cdf,minval=minval,maxval=maxval, **kwargs) store = pd.HDFStore(filename,'r') try: keywords = store.get_storer('{}/fns'.format(path)).attrs.keywords for kw,val in keywords.iteritems(): setattr(self,kw,val) except AttributeError: logging.warning('saved distribution {} does not have keywords or disttype saved; perhaps this distribution was written with an older version.'.format(filename)) store.close() class Empirical_Distribution(Distribution): """Generates a Distribution object given a tabulated PDF. Parameters ---------- xs : array-like x-values at which the PDF is evaluated pdf : array-like Values of pdf at provided x-values. smooth : int or float Smoothing parameter used by the interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,xs,pdf,smooth=0,**kwargs): pdf /= np.trapz(pdf,xs) fn = interpolate(xs,pdf,s=smooth) keywords = {'smooth':smooth} Distribution.__init__(self,fn,minval=xs.min(),maxval=xs.max(), keywords=keywords,**kwargs) class Gaussian_Distribution(Distribution): """Generates a normal distribution with given mu, sigma. ***It's probably better to use scipy.stats.norm rather than this if you care about numerical precision/speed and don't care about the plotting bells/whistles etc. the `Distribution` class provides.*** Parameters ---------- mu : float Mean of normal distribution. sig : float Width of normal distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig,**kwargs): self.mu = mu self.sig = sig def pdf(x): return 1./np.sqrt(2*np.pi*sig**2)*np.exp(-(x-mu)**2/(2*sig**2)) def cdf(x): return 0.5*(1 + erf((x-mu)/np.sqrt(2*sig**2))) if 'minval' not in kwargs: kwargs['minval'] = mu - 10*sig if 'maxval' not in kwargs: kwargs['maxval'] = mu + 10*sig keywords = {'mu':self.mu,'sig':self.sig} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +/- %.2f' % (self.name,self.mu,self.sig) def resample(self,N,**kwargs): return rand.normal(size=int(N))*self.sig + self.mu class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi) def resample(self,N): """Returns a random sampling. """ return rand.random(size=N)*(self.maxval - self.minval) + self.minval ############## Double LorGauss ########### def double_lorgauss(x,p): """Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned. """ mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi def fit_double_lorgauss(bins,h,Ntry=5): """Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available. """ try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value)) class DoubleLorGauss_Distribution(Distribution): """Defines a "double LorGauss" distribution according to the provided parameters. Parameters ---------- mu,sig1,sig2,gam1,gam2,G1,G2 : float Parameters of `double_lorgauss` function. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig1,sig2,gam1,gam2,G1,G2,**kwargs): self.mu = mu self.sig1 = sig1 self.sig2 = sig2 self.gam1 = gam1 self.gam2 = gam2 self.G1 = G1 #self.L1 = L1 self.G2 = G2 #self.L2 = L2 def pdf(x): return double_lorgauss(x,(self.mu,self.sig1,self.sig2, self.gam1,self.gam2, self.G1,self.G2,)) keywords = {'mu':mu,'sig1':sig1, 'sig2':sig2,'gam1':gam1,'gam2':gam2, 'G1':G1,'G2':G2} Distribution.__init__(self,pdf,keywords=keywords,**kwargs) ######## DoubleGauss ######### def doublegauss(x,p): """Evaluates normalized two-sided Gaussian distribution Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) Returns ------- value : float or array-like Distribution evaluated at input value(s). If single value provided, single value returned. """ mu,sig1,sig2 = p x = np.atleast_1d(x) A = 1./(np.sqrt(2*np.pi)*(sig1+sig2)/2.) ylo = A*np.exp(-(x-mu)**2/(2*sig1**2)) yhi = A*np.exp(-(x-mu)**2/(2*sig2**2)) y = x*0 wlo = np.where(x < mu) whi = np.where(x >= mu) y[wlo] = ylo[wlo] y[whi] = yhi[whi] if np.size(x)==1: return y[0] else: return y def doublegauss_cdf(x,p): """Cumulative distribution function for two-sided Gaussian Parameters ---------- x : float Input values at which to calculate CDF. p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) """ x = np.atleast_1d(x) mu,sig1,sig2 = p sig1 = np.absolute(sig1) sig2 = np.absolute(sig2) ylo = float(sig1)/(sig1 + sig2)*(1 + erf((x-mu)/np.sqrt(2*sig1**2))) yhi = float(sig1)/(sig1 + sig2) + float(sig2)/(sig1+sig2)*(erf((x-mu)/np.sqrt(2*sig2**2))) lo = x < mu hi = x >= mu return ylo*lo + yhi*hi def fit_doublegauss_samples(samples,**kwargs): """Fits a two-sided Gaussian to a set of samples. Calculates 0.16, 0.5, and 0.84 quantiles and passes these to `fit_doublegauss` for fitting. Parameters ---------- samples : array-like Samples to which to fit the Gaussian. kwargs Keyword arguments passed to `fit_doublegauss`. """ sorted_samples = np.sort(samples) N = len(samples) med = sorted_samples[N/2] siglo = med - sorted_samples[int(0.16*N)] sighi = sorted_samples[int(0.84*N)] - med return fit_doublegauss(med,siglo,sighi,median=True,**kwargs) def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True): """Fits a two-sided Gaussian distribution to match a given confidence interval. The center of the distribution may be either the median or the mode. Parameters ---------- med : float The center of the distribution to which to fit. Default this will be the mode unless the `median` keyword is set to True. siglo : float Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is the "lower error bar." sighi : float Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is the "upper error bar." interval : float, optional The confidence interval enclosed by the provided error bars. Default is 0.683 (1-sigma). p0 : array-like, optional Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`). median : bool, optional Whether to treat the `med` parameter as the median or mode (default will be mode). return_distribution: bool, optional If `True`, then function will return a `DoubleGauss_Distribution` object. Otherwise, will return just the parameters. """ if median: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med,med+sighi]) qvals = np.array([q1,0.5,q2]) def objfn(pars): logging.debug('{}'.format(pars)) logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals)) return doublegauss_cdf(targetvals,pars) - qvals if p0 is None: p0 = [med,siglo,sighi] pfit,success = leastsq(objfn,p0) else: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med+sighi]) qvals = np.array([q1,q2]) def objfn(pars): params = (med,pars[0],pars[1]) return doublegauss_cdf(targetvals,params) - qvals if p0 is None: p0 = [siglo,sighi] pfit,success = leastsq(objfn,p0) pfit = (med,pfit[0],pfit[1]) if return_distribution: dist = DoubleGauss_Distribution(*pfit) return dist else: return pfit class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo) def resample(self,N,**kwargs): """Random resampling of the doublegauss distribution """ lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals def powerlawfn(alpha,minval,maxval): C = powerlawnorm(alpha,minval,maxval) def fn(inpx): x = np.atleast_1d(inpx) y = C*x**(alpha) y[(x < minval) | (x > maxval)] = 0 return y return fn def powerlawnorm(alpha,minval,maxval): if np.size(alpha)==1: if alpha == -1: C = 1/np.log(maxval/minval) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) else: C = np.zeros(np.size(alpha)) w = np.where(alpha==-1) if len(w[0]>0): C[w] = 1./np.log(maxval/minval)*np.ones(len(w[0])) nw = np.where(alpha != -1) C[nw] = (1+alpha[nw])/(maxval**(1+alpha[nw])-minval**(1+alpha[nw])) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) return C class PowerLaw_Distribution(Distribution): def __init__(self,alpha,minval,maxval,**kwargs): self.alpha = alpha pdf = powerlawfn(alpha,minval,maxval) Distribution.__init__(self,pdf,minval=minval,maxval=maxval) ######## KDE ########### class KDE_Distribution(Distribution): def __init__(self,samples,adaptive=True,draw_direct=True,bandwidth=None,**kwargs): self.samples = samples self.bandwidth = bandwidth self.kde = KDE(samples,adaptive=adaptive,draw_direct=draw_direct, bandwidth=bandwidth) if 'minval' not in kwargs: kwargs['minval'] = samples.min() if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() keywords = {'adaptive':adaptive,'draw_direct':draw_direct, 'bandwidth':bandwidth} Distribution.__init__(self,self.kde,keywords=keywords,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def resample(self,N,**kwargs): return self.kde.resample(N,**kwargs) class KDE_Distribution_Fromtxt(KDE_Distribution): def __init__(self,filename,**kwargs): samples = np.loadtxt(filename) KDE_Distribution.__init__(self,samples,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
fit_double_lorgauss
python
def fit_double_lorgauss(bins,h,Ntry=5): try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value))
Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L666-L754
[ "def double_lorgauss(x,p):\n \"\"\"Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian.\n\n Parameters\n ----------\n x : float or array-like\n Value(s) at which to evaluate distribution\n\n p : array-like\n Input parameters: mu (mode of distribution),\n sig1 (LH Gaussian width),\n sig2 (RH Gaussian width),\n gam1 (LH Lorentzian width),\n gam2 (RH Lorentzian width),\n G1 (LH Gaussian \"strength\"),\n G2 (RH Gaussian \"strength\").\n\n Returns\n -------\n values : float or array-like\n Double LorGauss distribution evaluated at input(s). If single value provided,\n single value returned. \n \"\"\"\n mu,sig1,sig2,gam1,gam2,G1,G2 = p\n gam1 = float(gam1)\n gam2 = float(gam2)\n\n G1 = abs(G1)\n G2 = abs(G2)\n sig1 = abs(sig1)\n sig2 = abs(sig2)\n gam1 = abs(gam1)\n gab2 = abs(gam2)\n\n L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - \n (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) +\n (gam2/gam1)*(4-G1-G2))\n L1 = 4 - G1 - G2 - L2\n\n\n #print G1,G2,L1,L2\n\n y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\\\n L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2)\n y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\\\n L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2)\n lo = (x < mu)\n hi = (x >= mu)\n\n return y1*lo + y2*hi\n", "def set_params(G1,G2):\n params = Parameters()\n params.add('mu',value=mu0)\n params.add('sig1',value=sig0)\n params.add('sig2',value=sig0)\n params.add('gam1',value=sig0/10)\n params.add('gam2',value=sig0/10)\n params.add('G1',value=G1)\n params.add('G2',value=G2)\n return params\n" ]
from __future__ import absolute_import, division, print_function __author__ = 'Timothy D. Morton <tim.morton@gmail.com>' """ Defines objects useful for describing probability distributions. """ import numpy as np import matplotlib.pyplot as plt import logging from scipy.interpolate import UnivariateSpline as interpolate from scipy.integrate import quad import numpy.random as rand from scipy.special import erf from scipy.optimize import leastsq import pandas as pd from plotutils import setfig from .kde import KDE #figure this generic loading thing out; draft stage currently def load_distribution(filename,path=''): fns = pd.read_hdf(filename,path) store = pd.HDFStore(filename) if '{}/samples'.format(path) in store: samples = pd.read_hdf(filename,path+'/samples') samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0) cdf = interpolate(fns['vals'],fns['cdf'],s=0) attrs = store.get_storer('{}/fns'.format(path)).attrs keywords = attrs.keywords t = attrs.disttype store.close() return t.__init__() class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs) class Distribution_FromH5(Distribution): """Creates a Distribution object from one saved to an HDF file. File must have a `DataFrame` saved under [path]/fns in the .h5 file, containing 'vals', 'pdf', and 'cdf' columns. If samples are saved in the HDF storer, then they will be restored to this object; so will any saved keyword attributes. These appropriate .h5 files will be created by a call to the `save_hdf` method of the generic `Distribution` class. Parameters ---------- filename : string .h5 file where the distribution is saved. path : string, optional Path within the .h5 file where the distribution is saved. By default this will be the root level, but can be anywhere. kwargs Keyword arguments are passed to the `Distribution` constructor. """ def __init__(self,filename,path='',**kwargs): store = pd.HDFStore(filename,'r') fns = store[path+'/fns'] if '{}/samples'.format(path) in store: samples = store[path+'/samples'] self.samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0,k=1) #check to see if tabulated CDF is monotonically increasing d_cdf = fns['cdf'][1:] - fns['cdf'][:-1] if np.any(d_cdf < 0): logging.warning('tabulated CDF in {} is not strictly increasing. Recalculating CDF from PDF'.format(filename)) cdf = None #in this case, just recalc cdf from pdf else: cdf = interpolate(fns['vals'],fns['cdf'],s=0,k=1) Distribution.__init__(self,pdf,cdf,minval=minval,maxval=maxval, **kwargs) store = pd.HDFStore(filename,'r') try: keywords = store.get_storer('{}/fns'.format(path)).attrs.keywords for kw,val in keywords.iteritems(): setattr(self,kw,val) except AttributeError: logging.warning('saved distribution {} does not have keywords or disttype saved; perhaps this distribution was written with an older version.'.format(filename)) store.close() class Empirical_Distribution(Distribution): """Generates a Distribution object given a tabulated PDF. Parameters ---------- xs : array-like x-values at which the PDF is evaluated pdf : array-like Values of pdf at provided x-values. smooth : int or float Smoothing parameter used by the interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,xs,pdf,smooth=0,**kwargs): pdf /= np.trapz(pdf,xs) fn = interpolate(xs,pdf,s=smooth) keywords = {'smooth':smooth} Distribution.__init__(self,fn,minval=xs.min(),maxval=xs.max(), keywords=keywords,**kwargs) class Gaussian_Distribution(Distribution): """Generates a normal distribution with given mu, sigma. ***It's probably better to use scipy.stats.norm rather than this if you care about numerical precision/speed and don't care about the plotting bells/whistles etc. the `Distribution` class provides.*** Parameters ---------- mu : float Mean of normal distribution. sig : float Width of normal distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig,**kwargs): self.mu = mu self.sig = sig def pdf(x): return 1./np.sqrt(2*np.pi*sig**2)*np.exp(-(x-mu)**2/(2*sig**2)) def cdf(x): return 0.5*(1 + erf((x-mu)/np.sqrt(2*sig**2))) if 'minval' not in kwargs: kwargs['minval'] = mu - 10*sig if 'maxval' not in kwargs: kwargs['maxval'] = mu + 10*sig keywords = {'mu':self.mu,'sig':self.sig} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +/- %.2f' % (self.name,self.mu,self.sig) def resample(self,N,**kwargs): return rand.normal(size=int(N))*self.sig + self.mu class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi) def resample(self,N): """Returns a random sampling. """ return rand.random(size=N)*(self.maxval - self.minval) + self.minval ############## Double LorGauss ########### def double_lorgauss(x,p): """Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned. """ mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi def fit_double_lorgauss(bins,h,Ntry=5): """Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available. """ try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value)) class DoubleLorGauss_Distribution(Distribution): """Defines a "double LorGauss" distribution according to the provided parameters. Parameters ---------- mu,sig1,sig2,gam1,gam2,G1,G2 : float Parameters of `double_lorgauss` function. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig1,sig2,gam1,gam2,G1,G2,**kwargs): self.mu = mu self.sig1 = sig1 self.sig2 = sig2 self.gam1 = gam1 self.gam2 = gam2 self.G1 = G1 #self.L1 = L1 self.G2 = G2 #self.L2 = L2 def pdf(x): return double_lorgauss(x,(self.mu,self.sig1,self.sig2, self.gam1,self.gam2, self.G1,self.G2,)) keywords = {'mu':mu,'sig1':sig1, 'sig2':sig2,'gam1':gam1,'gam2':gam2, 'G1':G1,'G2':G2} Distribution.__init__(self,pdf,keywords=keywords,**kwargs) ######## DoubleGauss ######### def doublegauss(x,p): """Evaluates normalized two-sided Gaussian distribution Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) Returns ------- value : float or array-like Distribution evaluated at input value(s). If single value provided, single value returned. """ mu,sig1,sig2 = p x = np.atleast_1d(x) A = 1./(np.sqrt(2*np.pi)*(sig1+sig2)/2.) ylo = A*np.exp(-(x-mu)**2/(2*sig1**2)) yhi = A*np.exp(-(x-mu)**2/(2*sig2**2)) y = x*0 wlo = np.where(x < mu) whi = np.where(x >= mu) y[wlo] = ylo[wlo] y[whi] = yhi[whi] if np.size(x)==1: return y[0] else: return y def doublegauss_cdf(x,p): """Cumulative distribution function for two-sided Gaussian Parameters ---------- x : float Input values at which to calculate CDF. p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) """ x = np.atleast_1d(x) mu,sig1,sig2 = p sig1 = np.absolute(sig1) sig2 = np.absolute(sig2) ylo = float(sig1)/(sig1 + sig2)*(1 + erf((x-mu)/np.sqrt(2*sig1**2))) yhi = float(sig1)/(sig1 + sig2) + float(sig2)/(sig1+sig2)*(erf((x-mu)/np.sqrt(2*sig2**2))) lo = x < mu hi = x >= mu return ylo*lo + yhi*hi def fit_doublegauss_samples(samples,**kwargs): """Fits a two-sided Gaussian to a set of samples. Calculates 0.16, 0.5, and 0.84 quantiles and passes these to `fit_doublegauss` for fitting. Parameters ---------- samples : array-like Samples to which to fit the Gaussian. kwargs Keyword arguments passed to `fit_doublegauss`. """ sorted_samples = np.sort(samples) N = len(samples) med = sorted_samples[N/2] siglo = med - sorted_samples[int(0.16*N)] sighi = sorted_samples[int(0.84*N)] - med return fit_doublegauss(med,siglo,sighi,median=True,**kwargs) def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True): """Fits a two-sided Gaussian distribution to match a given confidence interval. The center of the distribution may be either the median or the mode. Parameters ---------- med : float The center of the distribution to which to fit. Default this will be the mode unless the `median` keyword is set to True. siglo : float Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is the "lower error bar." sighi : float Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is the "upper error bar." interval : float, optional The confidence interval enclosed by the provided error bars. Default is 0.683 (1-sigma). p0 : array-like, optional Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`). median : bool, optional Whether to treat the `med` parameter as the median or mode (default will be mode). return_distribution: bool, optional If `True`, then function will return a `DoubleGauss_Distribution` object. Otherwise, will return just the parameters. """ if median: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med,med+sighi]) qvals = np.array([q1,0.5,q2]) def objfn(pars): logging.debug('{}'.format(pars)) logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals)) return doublegauss_cdf(targetvals,pars) - qvals if p0 is None: p0 = [med,siglo,sighi] pfit,success = leastsq(objfn,p0) else: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med+sighi]) qvals = np.array([q1,q2]) def objfn(pars): params = (med,pars[0],pars[1]) return doublegauss_cdf(targetvals,params) - qvals if p0 is None: p0 = [siglo,sighi] pfit,success = leastsq(objfn,p0) pfit = (med,pfit[0],pfit[1]) if return_distribution: dist = DoubleGauss_Distribution(*pfit) return dist else: return pfit class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo) def resample(self,N,**kwargs): """Random resampling of the doublegauss distribution """ lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals def powerlawfn(alpha,minval,maxval): C = powerlawnorm(alpha,minval,maxval) def fn(inpx): x = np.atleast_1d(inpx) y = C*x**(alpha) y[(x < minval) | (x > maxval)] = 0 return y return fn def powerlawnorm(alpha,minval,maxval): if np.size(alpha)==1: if alpha == -1: C = 1/np.log(maxval/minval) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) else: C = np.zeros(np.size(alpha)) w = np.where(alpha==-1) if len(w[0]>0): C[w] = 1./np.log(maxval/minval)*np.ones(len(w[0])) nw = np.where(alpha != -1) C[nw] = (1+alpha[nw])/(maxval**(1+alpha[nw])-minval**(1+alpha[nw])) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) return C class PowerLaw_Distribution(Distribution): def __init__(self,alpha,minval,maxval,**kwargs): self.alpha = alpha pdf = powerlawfn(alpha,minval,maxval) Distribution.__init__(self,pdf,minval=minval,maxval=maxval) ######## KDE ########### class KDE_Distribution(Distribution): def __init__(self,samples,adaptive=True,draw_direct=True,bandwidth=None,**kwargs): self.samples = samples self.bandwidth = bandwidth self.kde = KDE(samples,adaptive=adaptive,draw_direct=draw_direct, bandwidth=bandwidth) if 'minval' not in kwargs: kwargs['minval'] = samples.min() if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() keywords = {'adaptive':adaptive,'draw_direct':draw_direct, 'bandwidth':bandwidth} Distribution.__init__(self,self.kde,keywords=keywords,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def resample(self,N,**kwargs): return self.kde.resample(N,**kwargs) class KDE_Distribution_Fromtxt(KDE_Distribution): def __init__(self,filename,**kwargs): samples = np.loadtxt(filename) KDE_Distribution.__init__(self,samples,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
doublegauss
python
def doublegauss(x,p): mu,sig1,sig2 = p x = np.atleast_1d(x) A = 1./(np.sqrt(2*np.pi)*(sig1+sig2)/2.) ylo = A*np.exp(-(x-mu)**2/(2*sig1**2)) yhi = A*np.exp(-(x-mu)**2/(2*sig2**2)) y = x*0 wlo = np.where(x < mu) whi = np.where(x >= mu) y[wlo] = ylo[wlo] y[whi] = yhi[whi] if np.size(x)==1: return y[0] else: return y
Evaluates normalized two-sided Gaussian distribution Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) Returns ------- value : float or array-like Distribution evaluated at input value(s). If single value provided, single value returned.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L792-L824
null
from __future__ import absolute_import, division, print_function __author__ = 'Timothy D. Morton <tim.morton@gmail.com>' """ Defines objects useful for describing probability distributions. """ import numpy as np import matplotlib.pyplot as plt import logging from scipy.interpolate import UnivariateSpline as interpolate from scipy.integrate import quad import numpy.random as rand from scipy.special import erf from scipy.optimize import leastsq import pandas as pd from plotutils import setfig from .kde import KDE #figure this generic loading thing out; draft stage currently def load_distribution(filename,path=''): fns = pd.read_hdf(filename,path) store = pd.HDFStore(filename) if '{}/samples'.format(path) in store: samples = pd.read_hdf(filename,path+'/samples') samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0) cdf = interpolate(fns['vals'],fns['cdf'],s=0) attrs = store.get_storer('{}/fns'.format(path)).attrs keywords = attrs.keywords t = attrs.disttype store.close() return t.__init__() class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs) class Distribution_FromH5(Distribution): """Creates a Distribution object from one saved to an HDF file. File must have a `DataFrame` saved under [path]/fns in the .h5 file, containing 'vals', 'pdf', and 'cdf' columns. If samples are saved in the HDF storer, then they will be restored to this object; so will any saved keyword attributes. These appropriate .h5 files will be created by a call to the `save_hdf` method of the generic `Distribution` class. Parameters ---------- filename : string .h5 file where the distribution is saved. path : string, optional Path within the .h5 file where the distribution is saved. By default this will be the root level, but can be anywhere. kwargs Keyword arguments are passed to the `Distribution` constructor. """ def __init__(self,filename,path='',**kwargs): store = pd.HDFStore(filename,'r') fns = store[path+'/fns'] if '{}/samples'.format(path) in store: samples = store[path+'/samples'] self.samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0,k=1) #check to see if tabulated CDF is monotonically increasing d_cdf = fns['cdf'][1:] - fns['cdf'][:-1] if np.any(d_cdf < 0): logging.warning('tabulated CDF in {} is not strictly increasing. Recalculating CDF from PDF'.format(filename)) cdf = None #in this case, just recalc cdf from pdf else: cdf = interpolate(fns['vals'],fns['cdf'],s=0,k=1) Distribution.__init__(self,pdf,cdf,minval=minval,maxval=maxval, **kwargs) store = pd.HDFStore(filename,'r') try: keywords = store.get_storer('{}/fns'.format(path)).attrs.keywords for kw,val in keywords.iteritems(): setattr(self,kw,val) except AttributeError: logging.warning('saved distribution {} does not have keywords or disttype saved; perhaps this distribution was written with an older version.'.format(filename)) store.close() class Empirical_Distribution(Distribution): """Generates a Distribution object given a tabulated PDF. Parameters ---------- xs : array-like x-values at which the PDF is evaluated pdf : array-like Values of pdf at provided x-values. smooth : int or float Smoothing parameter used by the interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,xs,pdf,smooth=0,**kwargs): pdf /= np.trapz(pdf,xs) fn = interpolate(xs,pdf,s=smooth) keywords = {'smooth':smooth} Distribution.__init__(self,fn,minval=xs.min(),maxval=xs.max(), keywords=keywords,**kwargs) class Gaussian_Distribution(Distribution): """Generates a normal distribution with given mu, sigma. ***It's probably better to use scipy.stats.norm rather than this if you care about numerical precision/speed and don't care about the plotting bells/whistles etc. the `Distribution` class provides.*** Parameters ---------- mu : float Mean of normal distribution. sig : float Width of normal distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig,**kwargs): self.mu = mu self.sig = sig def pdf(x): return 1./np.sqrt(2*np.pi*sig**2)*np.exp(-(x-mu)**2/(2*sig**2)) def cdf(x): return 0.5*(1 + erf((x-mu)/np.sqrt(2*sig**2))) if 'minval' not in kwargs: kwargs['minval'] = mu - 10*sig if 'maxval' not in kwargs: kwargs['maxval'] = mu + 10*sig keywords = {'mu':self.mu,'sig':self.sig} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +/- %.2f' % (self.name,self.mu,self.sig) def resample(self,N,**kwargs): return rand.normal(size=int(N))*self.sig + self.mu class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi) def resample(self,N): """Returns a random sampling. """ return rand.random(size=N)*(self.maxval - self.minval) + self.minval ############## Double LorGauss ########### def double_lorgauss(x,p): """Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned. """ mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi def fit_double_lorgauss(bins,h,Ntry=5): """Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available. """ try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value)) class DoubleLorGauss_Distribution(Distribution): """Defines a "double LorGauss" distribution according to the provided parameters. Parameters ---------- mu,sig1,sig2,gam1,gam2,G1,G2 : float Parameters of `double_lorgauss` function. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig1,sig2,gam1,gam2,G1,G2,**kwargs): self.mu = mu self.sig1 = sig1 self.sig2 = sig2 self.gam1 = gam1 self.gam2 = gam2 self.G1 = G1 #self.L1 = L1 self.G2 = G2 #self.L2 = L2 def pdf(x): return double_lorgauss(x,(self.mu,self.sig1,self.sig2, self.gam1,self.gam2, self.G1,self.G2,)) keywords = {'mu':mu,'sig1':sig1, 'sig2':sig2,'gam1':gam1,'gam2':gam2, 'G1':G1,'G2':G2} Distribution.__init__(self,pdf,keywords=keywords,**kwargs) ######## DoubleGauss ######### def doublegauss_cdf(x,p): """Cumulative distribution function for two-sided Gaussian Parameters ---------- x : float Input values at which to calculate CDF. p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) """ x = np.atleast_1d(x) mu,sig1,sig2 = p sig1 = np.absolute(sig1) sig2 = np.absolute(sig2) ylo = float(sig1)/(sig1 + sig2)*(1 + erf((x-mu)/np.sqrt(2*sig1**2))) yhi = float(sig1)/(sig1 + sig2) + float(sig2)/(sig1+sig2)*(erf((x-mu)/np.sqrt(2*sig2**2))) lo = x < mu hi = x >= mu return ylo*lo + yhi*hi def fit_doublegauss_samples(samples,**kwargs): """Fits a two-sided Gaussian to a set of samples. Calculates 0.16, 0.5, and 0.84 quantiles and passes these to `fit_doublegauss` for fitting. Parameters ---------- samples : array-like Samples to which to fit the Gaussian. kwargs Keyword arguments passed to `fit_doublegauss`. """ sorted_samples = np.sort(samples) N = len(samples) med = sorted_samples[N/2] siglo = med - sorted_samples[int(0.16*N)] sighi = sorted_samples[int(0.84*N)] - med return fit_doublegauss(med,siglo,sighi,median=True,**kwargs) def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True): """Fits a two-sided Gaussian distribution to match a given confidence interval. The center of the distribution may be either the median or the mode. Parameters ---------- med : float The center of the distribution to which to fit. Default this will be the mode unless the `median` keyword is set to True. siglo : float Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is the "lower error bar." sighi : float Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is the "upper error bar." interval : float, optional The confidence interval enclosed by the provided error bars. Default is 0.683 (1-sigma). p0 : array-like, optional Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`). median : bool, optional Whether to treat the `med` parameter as the median or mode (default will be mode). return_distribution: bool, optional If `True`, then function will return a `DoubleGauss_Distribution` object. Otherwise, will return just the parameters. """ if median: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med,med+sighi]) qvals = np.array([q1,0.5,q2]) def objfn(pars): logging.debug('{}'.format(pars)) logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals)) return doublegauss_cdf(targetvals,pars) - qvals if p0 is None: p0 = [med,siglo,sighi] pfit,success = leastsq(objfn,p0) else: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med+sighi]) qvals = np.array([q1,q2]) def objfn(pars): params = (med,pars[0],pars[1]) return doublegauss_cdf(targetvals,params) - qvals if p0 is None: p0 = [siglo,sighi] pfit,success = leastsq(objfn,p0) pfit = (med,pfit[0],pfit[1]) if return_distribution: dist = DoubleGauss_Distribution(*pfit) return dist else: return pfit class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo) def resample(self,N,**kwargs): """Random resampling of the doublegauss distribution """ lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals def powerlawfn(alpha,minval,maxval): C = powerlawnorm(alpha,minval,maxval) def fn(inpx): x = np.atleast_1d(inpx) y = C*x**(alpha) y[(x < minval) | (x > maxval)] = 0 return y return fn def powerlawnorm(alpha,minval,maxval): if np.size(alpha)==1: if alpha == -1: C = 1/np.log(maxval/minval) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) else: C = np.zeros(np.size(alpha)) w = np.where(alpha==-1) if len(w[0]>0): C[w] = 1./np.log(maxval/minval)*np.ones(len(w[0])) nw = np.where(alpha != -1) C[nw] = (1+alpha[nw])/(maxval**(1+alpha[nw])-minval**(1+alpha[nw])) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) return C class PowerLaw_Distribution(Distribution): def __init__(self,alpha,minval,maxval,**kwargs): self.alpha = alpha pdf = powerlawfn(alpha,minval,maxval) Distribution.__init__(self,pdf,minval=minval,maxval=maxval) ######## KDE ########### class KDE_Distribution(Distribution): def __init__(self,samples,adaptive=True,draw_direct=True,bandwidth=None,**kwargs): self.samples = samples self.bandwidth = bandwidth self.kde = KDE(samples,adaptive=adaptive,draw_direct=draw_direct, bandwidth=bandwidth) if 'minval' not in kwargs: kwargs['minval'] = samples.min() if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() keywords = {'adaptive':adaptive,'draw_direct':draw_direct, 'bandwidth':bandwidth} Distribution.__init__(self,self.kde,keywords=keywords,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def resample(self,N,**kwargs): return self.kde.resample(N,**kwargs) class KDE_Distribution_Fromtxt(KDE_Distribution): def __init__(self,filename,**kwargs): samples = np.loadtxt(filename) KDE_Distribution.__init__(self,samples,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
doublegauss_cdf
python
def doublegauss_cdf(x,p): x = np.atleast_1d(x) mu,sig1,sig2 = p sig1 = np.absolute(sig1) sig2 = np.absolute(sig2) ylo = float(sig1)/(sig1 + sig2)*(1 + erf((x-mu)/np.sqrt(2*sig1**2))) yhi = float(sig1)/(sig1 + sig2) + float(sig2)/(sig1+sig2)*(erf((x-mu)/np.sqrt(2*sig2**2))) lo = x < mu hi = x >= mu return ylo*lo + yhi*hi
Cumulative distribution function for two-sided Gaussian Parameters ---------- x : float Input values at which to calculate CDF. p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width)
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L826-L847
null
from __future__ import absolute_import, division, print_function __author__ = 'Timothy D. Morton <tim.morton@gmail.com>' """ Defines objects useful for describing probability distributions. """ import numpy as np import matplotlib.pyplot as plt import logging from scipy.interpolate import UnivariateSpline as interpolate from scipy.integrate import quad import numpy.random as rand from scipy.special import erf from scipy.optimize import leastsq import pandas as pd from plotutils import setfig from .kde import KDE #figure this generic loading thing out; draft stage currently def load_distribution(filename,path=''): fns = pd.read_hdf(filename,path) store = pd.HDFStore(filename) if '{}/samples'.format(path) in store: samples = pd.read_hdf(filename,path+'/samples') samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0) cdf = interpolate(fns['vals'],fns['cdf'],s=0) attrs = store.get_storer('{}/fns'.format(path)).attrs keywords = attrs.keywords t = attrs.disttype store.close() return t.__init__() class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs) class Distribution_FromH5(Distribution): """Creates a Distribution object from one saved to an HDF file. File must have a `DataFrame` saved under [path]/fns in the .h5 file, containing 'vals', 'pdf', and 'cdf' columns. If samples are saved in the HDF storer, then they will be restored to this object; so will any saved keyword attributes. These appropriate .h5 files will be created by a call to the `save_hdf` method of the generic `Distribution` class. Parameters ---------- filename : string .h5 file where the distribution is saved. path : string, optional Path within the .h5 file where the distribution is saved. By default this will be the root level, but can be anywhere. kwargs Keyword arguments are passed to the `Distribution` constructor. """ def __init__(self,filename,path='',**kwargs): store = pd.HDFStore(filename,'r') fns = store[path+'/fns'] if '{}/samples'.format(path) in store: samples = store[path+'/samples'] self.samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0,k=1) #check to see if tabulated CDF is monotonically increasing d_cdf = fns['cdf'][1:] - fns['cdf'][:-1] if np.any(d_cdf < 0): logging.warning('tabulated CDF in {} is not strictly increasing. Recalculating CDF from PDF'.format(filename)) cdf = None #in this case, just recalc cdf from pdf else: cdf = interpolate(fns['vals'],fns['cdf'],s=0,k=1) Distribution.__init__(self,pdf,cdf,minval=minval,maxval=maxval, **kwargs) store = pd.HDFStore(filename,'r') try: keywords = store.get_storer('{}/fns'.format(path)).attrs.keywords for kw,val in keywords.iteritems(): setattr(self,kw,val) except AttributeError: logging.warning('saved distribution {} does not have keywords or disttype saved; perhaps this distribution was written with an older version.'.format(filename)) store.close() class Empirical_Distribution(Distribution): """Generates a Distribution object given a tabulated PDF. Parameters ---------- xs : array-like x-values at which the PDF is evaluated pdf : array-like Values of pdf at provided x-values. smooth : int or float Smoothing parameter used by the interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,xs,pdf,smooth=0,**kwargs): pdf /= np.trapz(pdf,xs) fn = interpolate(xs,pdf,s=smooth) keywords = {'smooth':smooth} Distribution.__init__(self,fn,minval=xs.min(),maxval=xs.max(), keywords=keywords,**kwargs) class Gaussian_Distribution(Distribution): """Generates a normal distribution with given mu, sigma. ***It's probably better to use scipy.stats.norm rather than this if you care about numerical precision/speed and don't care about the plotting bells/whistles etc. the `Distribution` class provides.*** Parameters ---------- mu : float Mean of normal distribution. sig : float Width of normal distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig,**kwargs): self.mu = mu self.sig = sig def pdf(x): return 1./np.sqrt(2*np.pi*sig**2)*np.exp(-(x-mu)**2/(2*sig**2)) def cdf(x): return 0.5*(1 + erf((x-mu)/np.sqrt(2*sig**2))) if 'minval' not in kwargs: kwargs['minval'] = mu - 10*sig if 'maxval' not in kwargs: kwargs['maxval'] = mu + 10*sig keywords = {'mu':self.mu,'sig':self.sig} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +/- %.2f' % (self.name,self.mu,self.sig) def resample(self,N,**kwargs): return rand.normal(size=int(N))*self.sig + self.mu class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi) def resample(self,N): """Returns a random sampling. """ return rand.random(size=N)*(self.maxval - self.minval) + self.minval ############## Double LorGauss ########### def double_lorgauss(x,p): """Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned. """ mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi def fit_double_lorgauss(bins,h,Ntry=5): """Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available. """ try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value)) class DoubleLorGauss_Distribution(Distribution): """Defines a "double LorGauss" distribution according to the provided parameters. Parameters ---------- mu,sig1,sig2,gam1,gam2,G1,G2 : float Parameters of `double_lorgauss` function. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig1,sig2,gam1,gam2,G1,G2,**kwargs): self.mu = mu self.sig1 = sig1 self.sig2 = sig2 self.gam1 = gam1 self.gam2 = gam2 self.G1 = G1 #self.L1 = L1 self.G2 = G2 #self.L2 = L2 def pdf(x): return double_lorgauss(x,(self.mu,self.sig1,self.sig2, self.gam1,self.gam2, self.G1,self.G2,)) keywords = {'mu':mu,'sig1':sig1, 'sig2':sig2,'gam1':gam1,'gam2':gam2, 'G1':G1,'G2':G2} Distribution.__init__(self,pdf,keywords=keywords,**kwargs) ######## DoubleGauss ######### def doublegauss(x,p): """Evaluates normalized two-sided Gaussian distribution Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) Returns ------- value : float or array-like Distribution evaluated at input value(s). If single value provided, single value returned. """ mu,sig1,sig2 = p x = np.atleast_1d(x) A = 1./(np.sqrt(2*np.pi)*(sig1+sig2)/2.) ylo = A*np.exp(-(x-mu)**2/(2*sig1**2)) yhi = A*np.exp(-(x-mu)**2/(2*sig2**2)) y = x*0 wlo = np.where(x < mu) whi = np.where(x >= mu) y[wlo] = ylo[wlo] y[whi] = yhi[whi] if np.size(x)==1: return y[0] else: return y def fit_doublegauss_samples(samples,**kwargs): """Fits a two-sided Gaussian to a set of samples. Calculates 0.16, 0.5, and 0.84 quantiles and passes these to `fit_doublegauss` for fitting. Parameters ---------- samples : array-like Samples to which to fit the Gaussian. kwargs Keyword arguments passed to `fit_doublegauss`. """ sorted_samples = np.sort(samples) N = len(samples) med = sorted_samples[N/2] siglo = med - sorted_samples[int(0.16*N)] sighi = sorted_samples[int(0.84*N)] - med return fit_doublegauss(med,siglo,sighi,median=True,**kwargs) def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True): """Fits a two-sided Gaussian distribution to match a given confidence interval. The center of the distribution may be either the median or the mode. Parameters ---------- med : float The center of the distribution to which to fit. Default this will be the mode unless the `median` keyword is set to True. siglo : float Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is the "lower error bar." sighi : float Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is the "upper error bar." interval : float, optional The confidence interval enclosed by the provided error bars. Default is 0.683 (1-sigma). p0 : array-like, optional Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`). median : bool, optional Whether to treat the `med` parameter as the median or mode (default will be mode). return_distribution: bool, optional If `True`, then function will return a `DoubleGauss_Distribution` object. Otherwise, will return just the parameters. """ if median: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med,med+sighi]) qvals = np.array([q1,0.5,q2]) def objfn(pars): logging.debug('{}'.format(pars)) logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals)) return doublegauss_cdf(targetvals,pars) - qvals if p0 is None: p0 = [med,siglo,sighi] pfit,success = leastsq(objfn,p0) else: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med+sighi]) qvals = np.array([q1,q2]) def objfn(pars): params = (med,pars[0],pars[1]) return doublegauss_cdf(targetvals,params) - qvals if p0 is None: p0 = [siglo,sighi] pfit,success = leastsq(objfn,p0) pfit = (med,pfit[0],pfit[1]) if return_distribution: dist = DoubleGauss_Distribution(*pfit) return dist else: return pfit class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo) def resample(self,N,**kwargs): """Random resampling of the doublegauss distribution """ lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals def powerlawfn(alpha,minval,maxval): C = powerlawnorm(alpha,minval,maxval) def fn(inpx): x = np.atleast_1d(inpx) y = C*x**(alpha) y[(x < minval) | (x > maxval)] = 0 return y return fn def powerlawnorm(alpha,minval,maxval): if np.size(alpha)==1: if alpha == -1: C = 1/np.log(maxval/minval) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) else: C = np.zeros(np.size(alpha)) w = np.where(alpha==-1) if len(w[0]>0): C[w] = 1./np.log(maxval/minval)*np.ones(len(w[0])) nw = np.where(alpha != -1) C[nw] = (1+alpha[nw])/(maxval**(1+alpha[nw])-minval**(1+alpha[nw])) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) return C class PowerLaw_Distribution(Distribution): def __init__(self,alpha,minval,maxval,**kwargs): self.alpha = alpha pdf = powerlawfn(alpha,minval,maxval) Distribution.__init__(self,pdf,minval=minval,maxval=maxval) ######## KDE ########### class KDE_Distribution(Distribution): def __init__(self,samples,adaptive=True,draw_direct=True,bandwidth=None,**kwargs): self.samples = samples self.bandwidth = bandwidth self.kde = KDE(samples,adaptive=adaptive,draw_direct=draw_direct, bandwidth=bandwidth) if 'minval' not in kwargs: kwargs['minval'] = samples.min() if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() keywords = {'adaptive':adaptive,'draw_direct':draw_direct, 'bandwidth':bandwidth} Distribution.__init__(self,self.kde,keywords=keywords,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def resample(self,N,**kwargs): return self.kde.resample(N,**kwargs) class KDE_Distribution_Fromtxt(KDE_Distribution): def __init__(self,filename,**kwargs): samples = np.loadtxt(filename) KDE_Distribution.__init__(self,samples,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
fit_doublegauss_samples
python
def fit_doublegauss_samples(samples,**kwargs): sorted_samples = np.sort(samples) N = len(samples) med = sorted_samples[N/2] siglo = med - sorted_samples[int(0.16*N)] sighi = sorted_samples[int(0.84*N)] - med return fit_doublegauss(med,siglo,sighi,median=True,**kwargs)
Fits a two-sided Gaussian to a set of samples. Calculates 0.16, 0.5, and 0.84 quantiles and passes these to `fit_doublegauss` for fitting. Parameters ---------- samples : array-like Samples to which to fit the Gaussian. kwargs Keyword arguments passed to `fit_doublegauss`.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L849-L868
[ "def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True):\n \"\"\"Fits a two-sided Gaussian distribution to match a given confidence interval.\n\n The center of the distribution may be either the median or the mode.\n\n Parameters\n ----------\n med : float\n The center of the distribution to which to fit. Default this\n will be the mode unless the `median` keyword is set to True.\n\n siglo : float\n Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is\n the \"lower error bar.\"\n\n sighi : float\n Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is\n the \"upper error bar.\"\n\n interval : float, optional\n The confidence interval enclosed by the provided error bars. Default\n is 0.683 (1-sigma).\n\n p0 : array-like, optional\n Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`).\n\n median : bool, optional\n Whether to treat the `med` parameter as the median or mode\n (default will be mode).\n\n return_distribution: bool, optional\n If `True`, then function will return a `DoubleGauss_Distribution` object.\n Otherwise, will return just the parameters.\n \"\"\"\n if median:\n q1 = 0.5 - (interval/2)\n q2 = 0.5 + (interval/2)\n targetvals = np.array([med-siglo,med,med+sighi])\n qvals = np.array([q1,0.5,q2])\n def objfn(pars):\n logging.debug('{}'.format(pars))\n logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals))\n return doublegauss_cdf(targetvals,pars) - qvals\n\n if p0 is None:\n p0 = [med,siglo,sighi]\n pfit,success = leastsq(objfn,p0)\n\n else:\n q1 = 0.5 - (interval/2)\n q2 = 0.5 + (interval/2)\n targetvals = np.array([med-siglo,med+sighi])\n qvals = np.array([q1,q2])\n def objfn(pars):\n params = (med,pars[0],pars[1])\n return doublegauss_cdf(targetvals,params) - qvals\n\n if p0 is None:\n p0 = [siglo,sighi]\n pfit,success = leastsq(objfn,p0)\n pfit = (med,pfit[0],pfit[1])\n\n if return_distribution:\n dist = DoubleGauss_Distribution(*pfit)\n return dist\n else:\n return pfit\n" ]
from __future__ import absolute_import, division, print_function __author__ = 'Timothy D. Morton <tim.morton@gmail.com>' """ Defines objects useful for describing probability distributions. """ import numpy as np import matplotlib.pyplot as plt import logging from scipy.interpolate import UnivariateSpline as interpolate from scipy.integrate import quad import numpy.random as rand from scipy.special import erf from scipy.optimize import leastsq import pandas as pd from plotutils import setfig from .kde import KDE #figure this generic loading thing out; draft stage currently def load_distribution(filename,path=''): fns = pd.read_hdf(filename,path) store = pd.HDFStore(filename) if '{}/samples'.format(path) in store: samples = pd.read_hdf(filename,path+'/samples') samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0) cdf = interpolate(fns['vals'],fns['cdf'],s=0) attrs = store.get_storer('{}/fns'.format(path)).attrs keywords = attrs.keywords t = attrs.disttype store.close() return t.__init__() class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs) class Distribution_FromH5(Distribution): """Creates a Distribution object from one saved to an HDF file. File must have a `DataFrame` saved under [path]/fns in the .h5 file, containing 'vals', 'pdf', and 'cdf' columns. If samples are saved in the HDF storer, then they will be restored to this object; so will any saved keyword attributes. These appropriate .h5 files will be created by a call to the `save_hdf` method of the generic `Distribution` class. Parameters ---------- filename : string .h5 file where the distribution is saved. path : string, optional Path within the .h5 file where the distribution is saved. By default this will be the root level, but can be anywhere. kwargs Keyword arguments are passed to the `Distribution` constructor. """ def __init__(self,filename,path='',**kwargs): store = pd.HDFStore(filename,'r') fns = store[path+'/fns'] if '{}/samples'.format(path) in store: samples = store[path+'/samples'] self.samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0,k=1) #check to see if tabulated CDF is monotonically increasing d_cdf = fns['cdf'][1:] - fns['cdf'][:-1] if np.any(d_cdf < 0): logging.warning('tabulated CDF in {} is not strictly increasing. Recalculating CDF from PDF'.format(filename)) cdf = None #in this case, just recalc cdf from pdf else: cdf = interpolate(fns['vals'],fns['cdf'],s=0,k=1) Distribution.__init__(self,pdf,cdf,minval=minval,maxval=maxval, **kwargs) store = pd.HDFStore(filename,'r') try: keywords = store.get_storer('{}/fns'.format(path)).attrs.keywords for kw,val in keywords.iteritems(): setattr(self,kw,val) except AttributeError: logging.warning('saved distribution {} does not have keywords or disttype saved; perhaps this distribution was written with an older version.'.format(filename)) store.close() class Empirical_Distribution(Distribution): """Generates a Distribution object given a tabulated PDF. Parameters ---------- xs : array-like x-values at which the PDF is evaluated pdf : array-like Values of pdf at provided x-values. smooth : int or float Smoothing parameter used by the interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,xs,pdf,smooth=0,**kwargs): pdf /= np.trapz(pdf,xs) fn = interpolate(xs,pdf,s=smooth) keywords = {'smooth':smooth} Distribution.__init__(self,fn,minval=xs.min(),maxval=xs.max(), keywords=keywords,**kwargs) class Gaussian_Distribution(Distribution): """Generates a normal distribution with given mu, sigma. ***It's probably better to use scipy.stats.norm rather than this if you care about numerical precision/speed and don't care about the plotting bells/whistles etc. the `Distribution` class provides.*** Parameters ---------- mu : float Mean of normal distribution. sig : float Width of normal distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig,**kwargs): self.mu = mu self.sig = sig def pdf(x): return 1./np.sqrt(2*np.pi*sig**2)*np.exp(-(x-mu)**2/(2*sig**2)) def cdf(x): return 0.5*(1 + erf((x-mu)/np.sqrt(2*sig**2))) if 'minval' not in kwargs: kwargs['minval'] = mu - 10*sig if 'maxval' not in kwargs: kwargs['maxval'] = mu + 10*sig keywords = {'mu':self.mu,'sig':self.sig} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +/- %.2f' % (self.name,self.mu,self.sig) def resample(self,N,**kwargs): return rand.normal(size=int(N))*self.sig + self.mu class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi) def resample(self,N): """Returns a random sampling. """ return rand.random(size=N)*(self.maxval - self.minval) + self.minval ############## Double LorGauss ########### def double_lorgauss(x,p): """Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned. """ mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi def fit_double_lorgauss(bins,h,Ntry=5): """Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available. """ try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value)) class DoubleLorGauss_Distribution(Distribution): """Defines a "double LorGauss" distribution according to the provided parameters. Parameters ---------- mu,sig1,sig2,gam1,gam2,G1,G2 : float Parameters of `double_lorgauss` function. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig1,sig2,gam1,gam2,G1,G2,**kwargs): self.mu = mu self.sig1 = sig1 self.sig2 = sig2 self.gam1 = gam1 self.gam2 = gam2 self.G1 = G1 #self.L1 = L1 self.G2 = G2 #self.L2 = L2 def pdf(x): return double_lorgauss(x,(self.mu,self.sig1,self.sig2, self.gam1,self.gam2, self.G1,self.G2,)) keywords = {'mu':mu,'sig1':sig1, 'sig2':sig2,'gam1':gam1,'gam2':gam2, 'G1':G1,'G2':G2} Distribution.__init__(self,pdf,keywords=keywords,**kwargs) ######## DoubleGauss ######### def doublegauss(x,p): """Evaluates normalized two-sided Gaussian distribution Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) Returns ------- value : float or array-like Distribution evaluated at input value(s). If single value provided, single value returned. """ mu,sig1,sig2 = p x = np.atleast_1d(x) A = 1./(np.sqrt(2*np.pi)*(sig1+sig2)/2.) ylo = A*np.exp(-(x-mu)**2/(2*sig1**2)) yhi = A*np.exp(-(x-mu)**2/(2*sig2**2)) y = x*0 wlo = np.where(x < mu) whi = np.where(x >= mu) y[wlo] = ylo[wlo] y[whi] = yhi[whi] if np.size(x)==1: return y[0] else: return y def doublegauss_cdf(x,p): """Cumulative distribution function for two-sided Gaussian Parameters ---------- x : float Input values at which to calculate CDF. p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) """ x = np.atleast_1d(x) mu,sig1,sig2 = p sig1 = np.absolute(sig1) sig2 = np.absolute(sig2) ylo = float(sig1)/(sig1 + sig2)*(1 + erf((x-mu)/np.sqrt(2*sig1**2))) yhi = float(sig1)/(sig1 + sig2) + float(sig2)/(sig1+sig2)*(erf((x-mu)/np.sqrt(2*sig2**2))) lo = x < mu hi = x >= mu return ylo*lo + yhi*hi def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True): """Fits a two-sided Gaussian distribution to match a given confidence interval. The center of the distribution may be either the median or the mode. Parameters ---------- med : float The center of the distribution to which to fit. Default this will be the mode unless the `median` keyword is set to True. siglo : float Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is the "lower error bar." sighi : float Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is the "upper error bar." interval : float, optional The confidence interval enclosed by the provided error bars. Default is 0.683 (1-sigma). p0 : array-like, optional Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`). median : bool, optional Whether to treat the `med` parameter as the median or mode (default will be mode). return_distribution: bool, optional If `True`, then function will return a `DoubleGauss_Distribution` object. Otherwise, will return just the parameters. """ if median: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med,med+sighi]) qvals = np.array([q1,0.5,q2]) def objfn(pars): logging.debug('{}'.format(pars)) logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals)) return doublegauss_cdf(targetvals,pars) - qvals if p0 is None: p0 = [med,siglo,sighi] pfit,success = leastsq(objfn,p0) else: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med+sighi]) qvals = np.array([q1,q2]) def objfn(pars): params = (med,pars[0],pars[1]) return doublegauss_cdf(targetvals,params) - qvals if p0 is None: p0 = [siglo,sighi] pfit,success = leastsq(objfn,p0) pfit = (med,pfit[0],pfit[1]) if return_distribution: dist = DoubleGauss_Distribution(*pfit) return dist else: return pfit class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo) def resample(self,N,**kwargs): """Random resampling of the doublegauss distribution """ lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals def powerlawfn(alpha,minval,maxval): C = powerlawnorm(alpha,minval,maxval) def fn(inpx): x = np.atleast_1d(inpx) y = C*x**(alpha) y[(x < minval) | (x > maxval)] = 0 return y return fn def powerlawnorm(alpha,minval,maxval): if np.size(alpha)==1: if alpha == -1: C = 1/np.log(maxval/minval) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) else: C = np.zeros(np.size(alpha)) w = np.where(alpha==-1) if len(w[0]>0): C[w] = 1./np.log(maxval/minval)*np.ones(len(w[0])) nw = np.where(alpha != -1) C[nw] = (1+alpha[nw])/(maxval**(1+alpha[nw])-minval**(1+alpha[nw])) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) return C class PowerLaw_Distribution(Distribution): def __init__(self,alpha,minval,maxval,**kwargs): self.alpha = alpha pdf = powerlawfn(alpha,minval,maxval) Distribution.__init__(self,pdf,minval=minval,maxval=maxval) ######## KDE ########### class KDE_Distribution(Distribution): def __init__(self,samples,adaptive=True,draw_direct=True,bandwidth=None,**kwargs): self.samples = samples self.bandwidth = bandwidth self.kde = KDE(samples,adaptive=adaptive,draw_direct=draw_direct, bandwidth=bandwidth) if 'minval' not in kwargs: kwargs['minval'] = samples.min() if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() keywords = {'adaptive':adaptive,'draw_direct':draw_direct, 'bandwidth':bandwidth} Distribution.__init__(self,self.kde,keywords=keywords,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def resample(self,N,**kwargs): return self.kde.resample(N,**kwargs) class KDE_Distribution_Fromtxt(KDE_Distribution): def __init__(self,filename,**kwargs): samples = np.loadtxt(filename) KDE_Distribution.__init__(self,samples,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
fit_doublegauss
python
def fit_doublegauss(med,siglo,sighi,interval=0.683,p0=None,median=False,return_distribution=True): if median: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med,med+sighi]) qvals = np.array([q1,0.5,q2]) def objfn(pars): logging.debug('{}'.format(pars)) logging.debug('{} {}'.format(doublegauss_cdf(targetvals,pars),qvals)) return doublegauss_cdf(targetvals,pars) - qvals if p0 is None: p0 = [med,siglo,sighi] pfit,success = leastsq(objfn,p0) else: q1 = 0.5 - (interval/2) q2 = 0.5 + (interval/2) targetvals = np.array([med-siglo,med+sighi]) qvals = np.array([q1,q2]) def objfn(pars): params = (med,pars[0],pars[1]) return doublegauss_cdf(targetvals,params) - qvals if p0 is None: p0 = [siglo,sighi] pfit,success = leastsq(objfn,p0) pfit = (med,pfit[0],pfit[1]) if return_distribution: dist = DoubleGauss_Distribution(*pfit) return dist else: return pfit
Fits a two-sided Gaussian distribution to match a given confidence interval. The center of the distribution may be either the median or the mode. Parameters ---------- med : float The center of the distribution to which to fit. Default this will be the mode unless the `median` keyword is set to True. siglo : float Value at lower quantile (`q1 = 0.5 - interval/2`) to fit. Often this is the "lower error bar." sighi : float Value at upper quantile (`q2 = 0.5 + interval/2`) to fit. Often this is the "upper error bar." interval : float, optional The confidence interval enclosed by the provided error bars. Default is 0.683 (1-sigma). p0 : array-like, optional Initial guess `doublegauss` parameters for the fit (`mu, sig1, sig2`). median : bool, optional Whether to treat the `med` parameter as the median or mode (default will be mode). return_distribution: bool, optional If `True`, then function will return a `DoubleGauss_Distribution` object. Otherwise, will return just the parameters.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L871-L937
null
from __future__ import absolute_import, division, print_function __author__ = 'Timothy D. Morton <tim.morton@gmail.com>' """ Defines objects useful for describing probability distributions. """ import numpy as np import matplotlib.pyplot as plt import logging from scipy.interpolate import UnivariateSpline as interpolate from scipy.integrate import quad import numpy.random as rand from scipy.special import erf from scipy.optimize import leastsq import pandas as pd from plotutils import setfig from .kde import KDE #figure this generic loading thing out; draft stage currently def load_distribution(filename,path=''): fns = pd.read_hdf(filename,path) store = pd.HDFStore(filename) if '{}/samples'.format(path) in store: samples = pd.read_hdf(filename,path+'/samples') samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0) cdf = interpolate(fns['vals'],fns['cdf'],s=0) attrs = store.get_storer('{}/fns'.format(path)).attrs keywords = attrs.keywords t = attrs.disttype store.close() return t.__init__() class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs) class Distribution_FromH5(Distribution): """Creates a Distribution object from one saved to an HDF file. File must have a `DataFrame` saved under [path]/fns in the .h5 file, containing 'vals', 'pdf', and 'cdf' columns. If samples are saved in the HDF storer, then they will be restored to this object; so will any saved keyword attributes. These appropriate .h5 files will be created by a call to the `save_hdf` method of the generic `Distribution` class. Parameters ---------- filename : string .h5 file where the distribution is saved. path : string, optional Path within the .h5 file where the distribution is saved. By default this will be the root level, but can be anywhere. kwargs Keyword arguments are passed to the `Distribution` constructor. """ def __init__(self,filename,path='',**kwargs): store = pd.HDFStore(filename,'r') fns = store[path+'/fns'] if '{}/samples'.format(path) in store: samples = store[path+'/samples'] self.samples = np.array(samples) minval = fns['vals'].iloc[0] maxval = fns['vals'].iloc[-1] pdf = interpolate(fns['vals'],fns['pdf'],s=0,k=1) #check to see if tabulated CDF is monotonically increasing d_cdf = fns['cdf'][1:] - fns['cdf'][:-1] if np.any(d_cdf < 0): logging.warning('tabulated CDF in {} is not strictly increasing. Recalculating CDF from PDF'.format(filename)) cdf = None #in this case, just recalc cdf from pdf else: cdf = interpolate(fns['vals'],fns['cdf'],s=0,k=1) Distribution.__init__(self,pdf,cdf,minval=minval,maxval=maxval, **kwargs) store = pd.HDFStore(filename,'r') try: keywords = store.get_storer('{}/fns'.format(path)).attrs.keywords for kw,val in keywords.iteritems(): setattr(self,kw,val) except AttributeError: logging.warning('saved distribution {} does not have keywords or disttype saved; perhaps this distribution was written with an older version.'.format(filename)) store.close() class Empirical_Distribution(Distribution): """Generates a Distribution object given a tabulated PDF. Parameters ---------- xs : array-like x-values at which the PDF is evaluated pdf : array-like Values of pdf at provided x-values. smooth : int or float Smoothing parameter used by the interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,xs,pdf,smooth=0,**kwargs): pdf /= np.trapz(pdf,xs) fn = interpolate(xs,pdf,s=smooth) keywords = {'smooth':smooth} Distribution.__init__(self,fn,minval=xs.min(),maxval=xs.max(), keywords=keywords,**kwargs) class Gaussian_Distribution(Distribution): """Generates a normal distribution with given mu, sigma. ***It's probably better to use scipy.stats.norm rather than this if you care about numerical precision/speed and don't care about the plotting bells/whistles etc. the `Distribution` class provides.*** Parameters ---------- mu : float Mean of normal distribution. sig : float Width of normal distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig,**kwargs): self.mu = mu self.sig = sig def pdf(x): return 1./np.sqrt(2*np.pi*sig**2)*np.exp(-(x-mu)**2/(2*sig**2)) def cdf(x): return 0.5*(1 + erf((x-mu)/np.sqrt(2*sig**2))) if 'minval' not in kwargs: kwargs['minval'] = mu - 10*sig if 'maxval' not in kwargs: kwargs['maxval'] = mu + 10*sig keywords = {'mu':self.mu,'sig':self.sig} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +/- %.2f' % (self.name,self.mu,self.sig) def resample(self,N,**kwargs): return rand.normal(size=int(N))*self.sig + self.mu class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi) def resample(self,N): """Returns a random sampling. """ return rand.random(size=N)*(self.maxval - self.minval) + self.minval ############## Double LorGauss ########### def double_lorgauss(x,p): """Evaluates a normalized distribution that is a mixture of a double-sided Gaussian and Double-sided Lorentzian. Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Input parameters: mu (mode of distribution), sig1 (LH Gaussian width), sig2 (RH Gaussian width), gam1 (LH Lorentzian width), gam2 (RH Lorentzian width), G1 (LH Gaussian "strength"), G2 (RH Gaussian "strength"). Returns ------- values : float or array-like Double LorGauss distribution evaluated at input(s). If single value provided, single value returned. """ mu,sig1,sig2,gam1,gam2,G1,G2 = p gam1 = float(gam1) gam2 = float(gam2) G1 = abs(G1) G2 = abs(G2) sig1 = abs(sig1) sig2 = abs(sig2) gam1 = abs(gam1) gab2 = abs(gam2) L2 = (gam1/(gam1 + gam2)) * ((gam2*np.pi*G1)/(sig1*np.sqrt(2*np.pi)) - (gam2*np.pi*G2)/(sig2*np.sqrt(2*np.pi)) + (gam2/gam1)*(4-G1-G2)) L1 = 4 - G1 - G2 - L2 #print G1,G2,L1,L2 y1 = G1/(sig1*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig1**2) +\ L1/(np.pi*gam1) * gam1**2/((x-mu)**2 + gam1**2) y2 = G2/(sig2*np.sqrt(2*np.pi)) * np.exp(-0.5*(x-mu)**2/sig2**2) +\ L2/(np.pi*gam2) * gam2**2/((x-mu)**2 + gam2**2) lo = (x < mu) hi = (x >= mu) return y1*lo + y2*hi def fit_double_lorgauss(bins,h,Ntry=5): """Uses lmfit to fit a "Double LorGauss" distribution to a provided histogram. Uses a grid of starting guesses to try to avoid local minima. Parameters ---------- bins, h : array-like Bins and heights of a histogram, as returned by, e.g., `np.histogram`. Ntry : int, optional Spacing of grid for starting guesses. Will try `Ntry**2` different initial values of the "Gaussian strength" parameters `G1` and `G2`. Returns ------- parameters : tuple Parameters of best-fit "double LorGauss" distribution. Raises ------ ImportError If the lmfit module is not available. """ try: from lmfit import minimize, Parameters, Parameter, report_fit except ImportError: raise ImportError('you need lmfit to use this function.') #make sure histogram is normalized h /= np.trapz(h,bins) #zero-pad the ends of the distribution to keep fits positive N = len(bins) dbin = (bins[1:]-bins[:-1]).mean() newbins = np.concatenate((np.linspace(bins.min() - N/10*dbin,bins.min(),N/10), bins, np.linspace(bins.max(),bins.max() + N/10*dbin,N/10))) newh = np.concatenate((np.zeros(N/10),h,np.zeros(N/10))) mu0 = bins[np.argmax(newh)] sig0 = abs(mu0 - newbins[np.argmin(np.absolute(newh - 0.5*newh.max()))]) def set_params(G1,G2): params = Parameters() params.add('mu',value=mu0) params.add('sig1',value=sig0) params.add('sig2',value=sig0) params.add('gam1',value=sig0/10) params.add('gam2',value=sig0/10) params.add('G1',value=G1) params.add('G2',value=G2) return params sum_devsq_best = np.inf outkeep = None for G1 in np.linspace(0.1,1.9,Ntry): for G2 in np.linspace(0.1,1.9,Ntry): params = set_params(G1,G2) def residual(ps): pars = (params['mu'].value, params['sig1'].value, params['sig2'].value, params['gam1'].value, params['gam2'].value, params['G1'].value, params['G2'].value) hmodel = double_lorgauss(newbins,pars) return newh-hmodel out = minimize(residual,params) pars = (out.params['mu'].value,out.params['sig1'].value, out.params['sig2'].value,out.params['gam1'].value, out.params['gam2'].value,out.params['G1'].value, out.params['G2'].value) sum_devsq = ((newh - double_lorgauss(newbins,pars))**2).sum() #print 'devs = %.1f; initial guesses for G1, G2; %.1f, %.1f' % (sum_devsq,G1, G2) if sum_devsq < sum_devsq_best: sum_devsq_best = sum_devsq outkeep = out return (outkeep.params['mu'].value,abs(outkeep.params['sig1'].value), abs(outkeep.params['sig2'].value),abs(outkeep.params['gam1'].value), abs(outkeep.params['gam2'].value),abs(outkeep.params['G1'].value), abs(outkeep.params['G2'].value)) class DoubleLorGauss_Distribution(Distribution): """Defines a "double LorGauss" distribution according to the provided parameters. Parameters ---------- mu,sig1,sig2,gam1,gam2,G1,G2 : float Parameters of `double_lorgauss` function. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,mu,sig1,sig2,gam1,gam2,G1,G2,**kwargs): self.mu = mu self.sig1 = sig1 self.sig2 = sig2 self.gam1 = gam1 self.gam2 = gam2 self.G1 = G1 #self.L1 = L1 self.G2 = G2 #self.L2 = L2 def pdf(x): return double_lorgauss(x,(self.mu,self.sig1,self.sig2, self.gam1,self.gam2, self.G1,self.G2,)) keywords = {'mu':mu,'sig1':sig1, 'sig2':sig2,'gam1':gam1,'gam2':gam2, 'G1':G1,'G2':G2} Distribution.__init__(self,pdf,keywords=keywords,**kwargs) ######## DoubleGauss ######### def doublegauss(x,p): """Evaluates normalized two-sided Gaussian distribution Parameters ---------- x : float or array-like Value(s) at which to evaluate distribution p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) Returns ------- value : float or array-like Distribution evaluated at input value(s). If single value provided, single value returned. """ mu,sig1,sig2 = p x = np.atleast_1d(x) A = 1./(np.sqrt(2*np.pi)*(sig1+sig2)/2.) ylo = A*np.exp(-(x-mu)**2/(2*sig1**2)) yhi = A*np.exp(-(x-mu)**2/(2*sig2**2)) y = x*0 wlo = np.where(x < mu) whi = np.where(x >= mu) y[wlo] = ylo[wlo] y[whi] = yhi[whi] if np.size(x)==1: return y[0] else: return y def doublegauss_cdf(x,p): """Cumulative distribution function for two-sided Gaussian Parameters ---------- x : float Input values at which to calculate CDF. p : array-like Parameters of distribution: (mu: mode of distribution, sig1: LH width, sig2: RH width) """ x = np.atleast_1d(x) mu,sig1,sig2 = p sig1 = np.absolute(sig1) sig2 = np.absolute(sig2) ylo = float(sig1)/(sig1 + sig2)*(1 + erf((x-mu)/np.sqrt(2*sig1**2))) yhi = float(sig1)/(sig1 + sig2) + float(sig2)/(sig1+sig2)*(erf((x-mu)/np.sqrt(2*sig2**2))) lo = x < mu hi = x >= mu return ylo*lo + yhi*hi def fit_doublegauss_samples(samples,**kwargs): """Fits a two-sided Gaussian to a set of samples. Calculates 0.16, 0.5, and 0.84 quantiles and passes these to `fit_doublegauss` for fitting. Parameters ---------- samples : array-like Samples to which to fit the Gaussian. kwargs Keyword arguments passed to `fit_doublegauss`. """ sorted_samples = np.sort(samples) N = len(samples) med = sorted_samples[N/2] siglo = med - sorted_samples[int(0.16*N)] sighi = sorted_samples[int(0.84*N)] - med return fit_doublegauss(med,siglo,sighi,median=True,**kwargs) class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo) def resample(self,N,**kwargs): """Random resampling of the doublegauss distribution """ lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals def powerlawfn(alpha,minval,maxval): C = powerlawnorm(alpha,minval,maxval) def fn(inpx): x = np.atleast_1d(inpx) y = C*x**(alpha) y[(x < minval) | (x > maxval)] = 0 return y return fn def powerlawnorm(alpha,minval,maxval): if np.size(alpha)==1: if alpha == -1: C = 1/np.log(maxval/minval) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) else: C = np.zeros(np.size(alpha)) w = np.where(alpha==-1) if len(w[0]>0): C[w] = 1./np.log(maxval/minval)*np.ones(len(w[0])) nw = np.where(alpha != -1) C[nw] = (1+alpha[nw])/(maxval**(1+alpha[nw])-minval**(1+alpha[nw])) else: C = (1+alpha)/(maxval**(1+alpha)-minval**(1+alpha)) return C class PowerLaw_Distribution(Distribution): def __init__(self,alpha,minval,maxval,**kwargs): self.alpha = alpha pdf = powerlawfn(alpha,minval,maxval) Distribution.__init__(self,pdf,minval=minval,maxval=maxval) ######## KDE ########### class KDE_Distribution(Distribution): def __init__(self,samples,adaptive=True,draw_direct=True,bandwidth=None,**kwargs): self.samples = samples self.bandwidth = bandwidth self.kde = KDE(samples,adaptive=adaptive,draw_direct=draw_direct, bandwidth=bandwidth) if 'minval' not in kwargs: kwargs['minval'] = samples.min() if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() keywords = {'adaptive':adaptive,'draw_direct':draw_direct, 'bandwidth':bandwidth} Distribution.__init__(self,self.kde,keywords=keywords,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def resample(self,N,**kwargs): return self.kde.resample(N,**kwargs) class KDE_Distribution_Fromtxt(KDE_Distribution): def __init__(self,filename,**kwargs): samples = np.loadtxt(filename) KDE_Distribution.__init__(self,samples,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Distribution.pctile
python
def pctile(self,pct,res=1000): grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))]
Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L136-L158
[ "def cdf(x):\n x = np.atleast_1d(x)\n y = np.atleast_1d(cdf_fn(x))\n y[np.where(x < self.minval)] = 0\n y[np.where(x > self.maxval)] = 1\n return y\n" ]
class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Distribution.save_hdf
python
def save_hdf(self,filename,path='',res=1000,logspace=False): if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close()
Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L162-L206
[ "def cdf(x):\n x = np.atleast_1d(x)\n y = np.atleast_1d(cdf_fn(x))\n y[np.where(x < self.minval)] = 0\n y[np.where(x > self.maxval)] = 1\n return y\n" ]
class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Distribution.plot
python
def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2)
Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L244-L294
null
class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Distribution.resample
python
def resample(self,N,minval=None,maxval=None,log=False,res=1e4): N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds]
Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L296-L357
[ "def cdf(x):\n x = np.atleast_1d(x)\n y = np.atleast_1d(cdf_fn(x))\n y[np.where(x < self.minval)] = 0\n y[np.where(x > self.maxval)] = 1\n return y\n" ]
class Distribution(object): """Base class to describe probability distribution. Has some minimal functional overlap with scipy.stats random variates (e.g. `ppf`, `rvs`) Parameters ---------- pdf : callable The probability density function to be used. Does not have to be normalized, but must be non-negative. cdf : callable, optional The cumulative distribution function. If not provided, this will be tabulated from the pdf, as long as minval and maxval are also provided name : string, optional The name of the distribution (will be used, for example, to label a plot). Default is empty string. minval,maxval : float, optional The minimum and maximum values of the distribution. The Distribution will evaluate to zero outside these ranges, and this will also define the range of the CDF. Defaults are -np.inf and +np.inf. If these are not explicity provided, then a CDF function must be provided. norm : float, optional If not provided, this will be calculated by integrating the pdf from minval to maxval so that the Distribution is a proper PDF that integrates to unity. `norm` can be non-unity if desired, but beware, as this will cause some things to act unexpectedly. cdf_pts : int, optional Number of points to tabulate in order to calculate CDF, if not provided. Default is 500. keywords : dict, optional Optional dictionary of keywords; these will be saved with the distribution when `save_hdf` is called. Raises ------ ValueError If `cdf` is not provided and minval or maxval are infinity. """ def __init__(self,pdf,cdf=None,name='',minval=-np.inf,maxval=np.inf,norm=None, cdf_pts=500,keywords=None): self.name = name self.pdf = pdf self.cdf = cdf self.minval = minval self.maxval = maxval if keywords is None: self.keywords = {} else: self.keywords = keywords self.keywords['name'] = name self.keywords['minval'] = minval self.keywords['maxval'] = maxval if norm is None: self.norm = quad(self.pdf,minval,maxval,full_output=1)[0] else: self.norm = norm if cdf is None and (minval == -np.inf or maxval == np.inf): raise ValueError('must provide either explicit cdf function or explicit min/max values') else: #tabulate & interpolate CDF. pts = np.linspace(minval,maxval,cdf_pts) pdfgrid = self(pts) cdfgrid = pdfgrid.cumsum()/pdfgrid.cumsum().max() cdf_fn = interpolate(pts,cdfgrid,s=0,k=1) def cdf(x): x = np.atleast_1d(x) y = np.atleast_1d(cdf_fn(x)) y[np.where(x < self.minval)] = 0 y[np.where(x > self.maxval)] = 1 return y self.cdf = cdf #define minval_cdf, maxval_cdf zero_mask = cdfgrid==0 one_mask = cdfgrid==1 if zero_mask.sum()>0: self.minval_cdf = pts[zero_mask][-1] #last 0 value if one_mask.sum()>0: self.maxval_cdf = pts[one_mask][0] #first 1 value def pctile(self,pct,res=1000): """Returns the desired percentile of the distribution. Will only work if properly normalized. Designed to mimic the `ppf` method of the `scipy.stats` random variate objects. Works by gridding the CDF at a given resolution and matching the nearest point. NB, this is of course not as precise as an analytic ppf. Parameters ---------- pct : float Percentile between 0 and 1. res : int, optional The resolution at which to grid the CDF to find the percentile. Returns ------- percentile : float """ grid = np.linspace(self.minval,self.maxval,res) return grid[np.argmin(np.absolute(pct-self.cdf(grid)))] ppf = pctile def save_hdf(self,filename,path='',res=1000,logspace=False): """Saves distribution to an HDF5 file. Saves a pandas `dataframe` object containing tabulated pdf and cdf values at a specfied resolution. After saving to a particular path, a distribution may be regenerated using the `Distribution_FromH5` subclass. Parameters ---------- filename : string File in which to save the distribution. Should end in .h5. path : string, optional Path in which to save the distribution within the .h5 file. By default this is an empty string, which will lead to saving the `fns` dataframe at the root level of the file. res : int, optional Resolution at which to grid the distribution for saving. logspace : bool, optional Sets whether the tabulated function should be gridded with log or linear spacing. Default will be logspace=False, corresponding to linear gridding. """ if logspace: vals = np.logspace(np.log10(self.minval), np.log10(self.maxval), res) else: vals = np.linspace(self.minval,self.maxval,res) d = {'vals':vals, 'pdf':self(vals), 'cdf':self.cdf(vals)} df = pd.DataFrame(d) df.to_hdf(filename,path+'/fns') if hasattr(self,'samples'): s = pd.Series(self.samples) s.to_hdf(filename,path+'/samples') store = pd.HDFStore(filename) attrs = store.get_storer('{}/fns'.format(path)).attrs attrs.keywords = self.keywords attrs.disttype = type(self) store.close() def __call__(self,x): """ Evaluates pdf. Forces zero outside of (self.minval,self.maxval). Will return Parameters ---------- x : float, array-like Value(s) at which to evaluate PDF. Returns ------- pdf : float, array-like Probability density (or re-normalized density if self.norm was explicity provided. """ y = self.pdf(x) x = np.atleast_1d(x) y = np.atleast_1d(y) y[(x < self.minval) | (x > self.maxval)] = 0 y /= self.norm if np.size(x)==1: return y[0] else: return y def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name, self.pctile(0.5), self.pctile(0.84)-self.pctile(0.5), self.pctile(0.5)-self.pctile(0.16)) def __repr__(self): return '<%s object: %s>' % (type(self),str(self)) def plot(self,minval=None,maxval=None,fig=None,log=False, npts=500,**kwargs): """ Plots distribution. Parameters ---------- minval : float,optional minimum value to plot. Required if minval of Distribution is `-np.inf`. maxval : float, optional maximum value to plot. Required if maxval of Distribution is `np.inf`. fig : None or int, optional Parameter to pass to `setfig`. If `None`, then a new figure is created; if a non-zero integer, the plot will go to that figure (clearing everything first), if zero, then will overplot on current axes. log : bool, optional If `True`, the x-spacing of the points to plot will be logarithmic. npoints : int, optional Number of points to plot. kwargs Keyword arguments are passed to plt.plot Raises ------ ValueError If finite lower and upper bounds are not provided. """ if minval is None: minval = self.minval if maxval is None: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to plot. (use minval, maxval kws)') if log: xs = np.logspace(np.log10(minval),np.log10(maxval),npts) else: xs = np.linspace(minval,maxval,npts) setfig(fig) plt.plot(xs,self(xs),**kwargs) plt.xlabel(self.name) plt.ylim(ymin=0,ymax=self(xs).max()*1.2) def resample(self,N,minval=None,maxval=None,log=False,res=1e4): """Returns random samples generated according to the distribution Mirrors basic functionality of `rvs` method for `scipy.stats` random variates. Implemented by mapping uniform numbers onto the inverse CDF using a closest-matching grid approach. Parameters ---------- N : int Number of samples to return minval,maxval : float, optional Minimum/maximum values to resample. Should both usually just be `None`, which will default to `self.minval`/`self.maxval`. log : bool, optional Whether grid should be log- or linear-spaced. res : int, optional Resolution of CDF grid used. Returns ------- values : ndarray N samples. Raises ------ ValueError If maxval/minval are +/- infinity, this doesn't work because of the grid-based approach. """ N = int(N) if minval is None: if hasattr(self,'minval_cdf'): minval = self.minval_cdf else: minval = self.minval if maxval is None: if hasattr(self,'maxval_cdf'): maxval = self.maxval_cdf else: maxval = self.maxval if maxval==np.inf or minval==-np.inf: raise ValueError('must have finite upper and lower bounds to resample. (set minval, maxval kws)') u = rand.random(size=N) if log: vals = np.logspace(log10(minval),log10(maxval),res) else: vals = np.linspace(minval,maxval,res) #sometimes cdf is flat. so ys will need to be uniqued ys,yinds = np.unique(self.cdf(vals), return_index=True) vals = vals[yinds] inds = np.digitize(u,ys) return vals[inds] def rvs(self,*args,**kwargs): return self.resample(*args,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Hist_Distribution.plothist
python
def plothist(self,fig=None,**kwargs): setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs)
Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L548-L560
null
class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def resample(self,N): """Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples. """ inds = rand.randint(len(self.samples),size=N) return self.samples[inds] def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Hist_Distribution.resample
python
def resample(self,N): inds = rand.randint(len(self.samples),size=N) return self.samples[inds]
Returns a bootstrap resampling of provided samples. Parameters ---------- N : int Number of samples.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L562-L571
null
class Hist_Distribution(Distribution): """Generates a distribution from a histogram of provided samples. Uses `np.histogram` to create a histogram using the bins keyword, then interpolates this histogram to create the pdf to pass to the `Distribution` constructor. Parameters ---------- samples : array-like The samples used to create the distribution bins : int or array-like, optional Keyword passed to `np.histogram`. If integer, ths will be the number of bins, if array-like, then this defines bin edges. equibin : bool, optional If true and ``bins`` is an integer ``N``, then the bins will be found by splitting the data into ``N`` equal-sized groups. smooth : int or float Smoothing parameter used by the interpolation function. order : int Order of the spline to be used for interpolation. Default is for linear interpolation. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,samples,bins=10,equibin=True,smooth=0,order=1,**kwargs): self.samples = samples if type(bins)==type(10) and equibin: N = len(samples)//bins sortsamples = np.sort(samples) bins = sortsamples[0::N] if bins[-1] != sortsamples[-1]: bins = np.concatenate([bins,np.array([sortsamples[-1]])]) hist,bins = np.histogram(samples,bins=bins,density=True) self.bins = bins bins = (bins[1:] + bins[:-1])/2. pdf_initial = interpolate(bins,hist,s=smooth,k=order) def pdf(x): x = np.atleast_1d(x) y = pdf_initial(x) w = np.where((x < self.bins[0]) | (x > self.bins[-1])) y[w] = 0 return y cdf = interpolate(bins,hist.cumsum()/hist.cumsum().max(),s=smooth, k=order) if 'maxval' not in kwargs: kwargs['maxval'] = samples.max() if 'minval' not in kwargs: kwargs['minval'] = samples.min() keywords = {'bins':bins,'smooth':smooth,'order':order} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.1f +/- %.1f' % (self.name,self.samples.mean(),self.samples.std()) def plothist(self,fig=None,**kwargs): """Plots a histogram of samples using provided bins. Parameters ---------- fig : None or int Parameter passed to `setfig`. kwargs Keyword arguments passed to `plt.hist`. """ setfig(fig) plt.hist(self.samples,bins=self.bins,**kwargs) def save_hdf(self,filename,path='',**kwargs): Distribution.save_hdf(self,filename,path=path,**kwargs)
timothydmorton/simpledist
simpledist/distributions.py
Box_Distribution.resample
python
def resample(self,N): return rand.random(size=N)*(self.maxval - self.minval) + self.minval
Returns a random sampling.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L606-L609
null
class Box_Distribution(Distribution): """Simple distribution uniform between provided lower and upper limits. Parameters ---------- lo,hi : float Lower/upper limits of the distribution. kwargs Keyword arguments passed to `Distribution` constructor. """ def __init__(self,lo,hi,**kwargs): self.lo = lo self.hi = hi def pdf(x): return 1./(hi-lo) + 0*x def cdf(x): x = np.atleast_1d(x) y = (x - lo) / (hi - lo) y[x < lo] = 0 y[x > hi] = 1 return y Distribution.__init__(self,pdf,cdf,minval=lo,maxval=hi,**kwargs) def __str__(self): return '%.1f < %s < %.1f' % (self.lo,self.name,self.hi)
timothydmorton/simpledist
simpledist/distributions.py
DoubleGauss_Distribution.resample
python
def resample(self,N,**kwargs): lovals = self.mu - np.absolute(rand.normal(size=N)*self.siglo) hivals = self.mu + np.absolute(rand.normal(size=N)*self.sighi) u = rand.random(size=N) hi = (u < float(self.sighi)/(self.sighi + self.siglo)) lo = (u >= float(self.sighi)/(self.sighi + self.siglo)) vals = np.zeros(N) vals[hi] = hivals[hi] vals[lo] = lovals[lo] return vals
Random resampling of the doublegauss distribution
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/distributions.py#L982-L995
null
class DoubleGauss_Distribution(Distribution): """A Distribution oject representing a two-sided Gaussian distribution This can be used to represent a slightly asymmetric distribution, and consists of two half-Normal distributions patched together at the mode, and normalized appropriately. The pdf and cdf are according to the `doubleguass` and `doubleguass_cdf` functions, respectively. Parameters ---------- mu : float The mode of the distribution. siglo : float Width of lower half-Gaussian. sighi : float Width of upper half-Gaussian. kwargs Keyword arguments are passed to `Distribution` constructor. """ def __init__(self,mu,siglo,sighi,**kwargs): self.mu = mu self.siglo = float(siglo) self.sighi = float(sighi) def pdf(x): return doublegauss(x,(mu,siglo,sighi)) def cdf(x): return doublegauss_cdf(x,(mu,siglo,sighi)) if 'minval' not in kwargs: kwargs['minval'] = mu - 5*siglo if 'maxval' not in kwargs: kwargs['maxval'] = mu + 5*sighi keywords = {'mu':mu,'siglo':siglo,'sighi':sighi} Distribution.__init__(self,pdf,cdf,keywords=keywords,**kwargs) def __str__(self): return '%s = %.2f +%.2f -%.2f' % (self.name,self.mu,self.sighi,self.siglo)
timothydmorton/simpledist
simpledist/kde.py
deriv
python
def deriv(f,c,dx=0.0001): return (f(c+dx)-f(c-dx))/(2*dx)
deriv(f,c,dx) --> float Returns f'(x), computed as a symmetric difference quotient.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/kde.py#L224-L230
null
from __future__ import absolute_import, division, print_function import numpy as np from scipy.stats import gaussian_kde import numpy.random as rand from scipy.integrate import quad class KDE(object): """An implementation of a kernel density estimator allowing for adaptive kernels. If the `adaptive` keyword is set to `False`, then this will essentially be just a wrapper for the `scipy.stats.gaussian_kde` class. If adaptive, though, it allows for different kernels and different kernel widths according to the "K-nearest-neighbors" algorithm as discussed `here <http://en.wikipedia.org/wiki/Variable_kernel_density_estimation#Balloon_estimators>`_. The `fast` option does the NN calculation using broadcasting arrays rather than a brute-force sort. By default the fast option will be used for datasets smaller than 5000. Parameters ---------- dataset : array-like Data set from which to calculate the KDE. kernel : {'tricube','ep','gauss'}, optional Kernel function to use for adaptive estimator. adaptive : bool, optional Flag whether or not to use adaptive KDE. If this is false, then this class will just be a wrapper for `scipy.stats.gaussian_kde`. k : `None` or int, optional Number to use for K-nearest-neighbor algorithm. If `None`, then it will be set to the `N/4`, where `N` is the size of the dataset. fast : `None` or bool, optional If `None`, then `fast = N < 5001`, where `N` is the size of the dataset. `fast=True` will force array calculations, which will use lots of RAM if the dataset is large. norm : float, optional Allows the normalization of the distribution to be something other than unity bandwidth : `None` or float, optional Passed to `scipy.stats.gaussian_kde` if not using adaptive mode. weights : array-like, optional Not yet implemented. draw_direct : bool, optional If `True`, then resampling will be just a bootstrap resampling of the input samples. If `False`, then resampling will actually resample each individual kernel (not recommended for large-ish datasets). kwargs Keyword arguments passed to `scipy.stats.gaussian_kde` if adaptive mode is not being used. """ def __init__(self,dataset,kernel='tricube',adaptive=True,k=None, fast=None,norm=1.,bandwidth=None,weights=None, draw_direct=False,**kwargs): self.dataset = np.atleast_1d(dataset) self.weights = weights self.n = np.size(dataset) self.kernel = kernelfn(kernel) self.kernelname = kernel self.bandwidth = bandwidth self.draw_direct = draw_direct if k: self.k = k else: self.k = self.n/4 self.norm=norm self.adaptive = adaptive self.fast = fast if adaptive: if self.fast==None: self.fast = self.n < 5001 if self.fast: #d1,d2 = np.meshgrid(self.dataset,self.dataset) #use broadcasting instead of meshgrid diff = np.absolute(self.dataset - self.dataset[:,np.newaxis]) diffsort = np.sort(diff,axis=0) self.h = diffsort[self.k,:] ##Attempt to handle larger datasets more easily: else: sortinds = np.argsort(self.dataset) x = self.dataset[sortinds] h = np.zeros(len(x)) for i in np.arange(len(x)): lo = i - self.k hi = i + self.k + 1 if lo < 0: lo = 0 if hi > len(x): hi = len(x) diffs = abs(x[lo:hi]-x[i]) h[sortinds[i]] = np.sort(diffs)[self.k] self.h = h else: self.gauss_kde = gaussian_kde(self.dataset,bw_method=bandwidth,**kwargs) def renorm(self,norm): """Change the normalization""" self.norm = norm def evaluate(self,points): if not self.adaptive: return self.gauss_kde(points)*self.norm points = np.atleast_1d(points).astype(self.dataset.dtype) k = self.k npts = np.size(points) h = self.h X,Y = np.meshgrid(self.dataset,points) H = np.resize(h,(npts,self.n)) U = (X-Y)/H.astype(float) result = 1./self.n*1./H*self.kernel(U) return np.sum(result,axis=1)*self.norm __call__ = evaluate def integrate_box(self,low,high,forcequad=False,**kwargs): """Integrates over a box. Optionally force quad integration, even for non-adaptive. If adaptive mode is not being used, this will just call the `scipy.stats.gaussian_kde` method `integrate_box_1d`. Else, by default, it will call `scipy.integrate.quad`. If the `forcequad` flag is turned on, then that integration will be used even if adaptive mode is off. Parameters ---------- low : float Lower limit of integration high : float Upper limit of integration forcequad : bool If `True`, then use the quad integration even if adaptive mode is off. kwargs Keyword arguments passed to `scipy.integrate.quad`. """ if not self.adaptive and not forcequad: return self.gauss_kde.integrate_box_1d(low,high)*self.norm return quad(self.evaluate,low,high,**kwargs)[0] def resample(self,size=None,direct=None): if direct is None: direct = self.draw_direct size=int(size) if not self.adaptive: return np.squeeze(self.gauss_kde.resample(size=size)) if direct: inds = rand.randint(self.n,size=size) return self.dataset[inds] else: if size is None: size = self.n indices = rand.randint(0,self.n,size=size) means = self.dataset[indices] h = self.h[indices] fuzz = kerneldraw(size,self.kernelname)*h return np.squeeze(means + fuzz) draw = resample def epkernel(u): x = np.atleast_1d(u) y = 3./4*(1-x*x) y[((x>1) | (x < -1))] = 0 return y def gausskernel(u): return 1/np.sqrt(2*np.pi)*np.exp(-0.5*u*u) def tricubekernel(u): x = np.atleast_1d(u) y = 35./32*(1-x*x)**3 y[((x > 1) | (x < -1))] = 0 return y def kernelfn(kernel='tricube'): if kernel=='ep': return epkernel elif kernel=='gauss': return gausskernel elif kernel=='tricube': return tricubekernel def kerneldraw(size=1,kernel='tricube',exact=False): if kernel=='tricube': fn = lambda x: 1./2 + 35./32*x - 35./32*x**3 + 21./32*x**5 - 5./32*x**7 u = rand.random(size=size) if not exact: xs = np.linspace(-1,1,1e4) ys = fn(xs) inds = np.digitize(u,ys) return xs[inds] else: #old way (exact) rets = np.zeros(size) for i in np.arange(size): f = lambda x: u[i]-fn(x) rets[i] = newton(f,0,restrict=(-1,1)) return rets def deriv(f,c,dx=0.0001): """ deriv(f,c,dx) --> float Returns f'(x), computed as a symmetric difference quotient. """ return (f(c+dx)-f(c-dx))/(2*dx) def fuzzyequals(a,b,tol=0.0001): return abs(a-b) < tol def newton(f,c,tol=0.0001,restrict=None): """ newton(f,c) --> float Returns the x closest to c such that f(x) = 0 """ #print(c) if restrict: lo,hi = restrict if c < lo or c > hi: print(c) c = random*(hi-lo)+lo if fuzzyequals(f(c),0,tol): return c else: try: return newton(f,c-f(c)/deriv(f,c,tol),tol,restrict) except: return None
timothydmorton/simpledist
simpledist/kde.py
newton
python
def newton(f,c,tol=0.0001,restrict=None): #print(c) if restrict: lo,hi = restrict if c < lo or c > hi: print(c) c = random*(hi-lo)+lo if fuzzyequals(f(c),0,tol): return c else: try: return newton(f,c-f(c)/deriv(f,c,tol),tol,restrict) except: return None
newton(f,c) --> float Returns the x closest to c such that f(x) = 0
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/kde.py#L235-L254
[ "def newton(f,c,tol=0.0001,restrict=None):\n \"\"\"\n newton(f,c) --> float\n\n Returns the x closest to c such that f(x) = 0\n \"\"\"\n #print(c)\n if restrict:\n lo,hi = restrict\n if c < lo or c > hi:\n print(c)\n c = random*(hi-lo)+lo\n\n if fuzzyequals(f(c),0,tol):\n return c\n else:\n try:\n return newton(f,c-f(c)/deriv(f,c,tol),tol,restrict)\n except:\n return None\n", "def deriv(f,c,dx=0.0001):\n \"\"\"\n deriv(f,c,dx) --> float\n\n Returns f'(x), computed as a symmetric difference quotient.\n \"\"\"\n return (f(c+dx)-f(c-dx))/(2*dx)\n", "def fuzzyequals(a,b,tol=0.0001):\n return abs(a-b) < tol\n", "f = lambda x: u[i]-fn(x)\n" ]
from __future__ import absolute_import, division, print_function import numpy as np from scipy.stats import gaussian_kde import numpy.random as rand from scipy.integrate import quad class KDE(object): """An implementation of a kernel density estimator allowing for adaptive kernels. If the `adaptive` keyword is set to `False`, then this will essentially be just a wrapper for the `scipy.stats.gaussian_kde` class. If adaptive, though, it allows for different kernels and different kernel widths according to the "K-nearest-neighbors" algorithm as discussed `here <http://en.wikipedia.org/wiki/Variable_kernel_density_estimation#Balloon_estimators>`_. The `fast` option does the NN calculation using broadcasting arrays rather than a brute-force sort. By default the fast option will be used for datasets smaller than 5000. Parameters ---------- dataset : array-like Data set from which to calculate the KDE. kernel : {'tricube','ep','gauss'}, optional Kernel function to use for adaptive estimator. adaptive : bool, optional Flag whether or not to use adaptive KDE. If this is false, then this class will just be a wrapper for `scipy.stats.gaussian_kde`. k : `None` or int, optional Number to use for K-nearest-neighbor algorithm. If `None`, then it will be set to the `N/4`, where `N` is the size of the dataset. fast : `None` or bool, optional If `None`, then `fast = N < 5001`, where `N` is the size of the dataset. `fast=True` will force array calculations, which will use lots of RAM if the dataset is large. norm : float, optional Allows the normalization of the distribution to be something other than unity bandwidth : `None` or float, optional Passed to `scipy.stats.gaussian_kde` if not using adaptive mode. weights : array-like, optional Not yet implemented. draw_direct : bool, optional If `True`, then resampling will be just a bootstrap resampling of the input samples. If `False`, then resampling will actually resample each individual kernel (not recommended for large-ish datasets). kwargs Keyword arguments passed to `scipy.stats.gaussian_kde` if adaptive mode is not being used. """ def __init__(self,dataset,kernel='tricube',adaptive=True,k=None, fast=None,norm=1.,bandwidth=None,weights=None, draw_direct=False,**kwargs): self.dataset = np.atleast_1d(dataset) self.weights = weights self.n = np.size(dataset) self.kernel = kernelfn(kernel) self.kernelname = kernel self.bandwidth = bandwidth self.draw_direct = draw_direct if k: self.k = k else: self.k = self.n/4 self.norm=norm self.adaptive = adaptive self.fast = fast if adaptive: if self.fast==None: self.fast = self.n < 5001 if self.fast: #d1,d2 = np.meshgrid(self.dataset,self.dataset) #use broadcasting instead of meshgrid diff = np.absolute(self.dataset - self.dataset[:,np.newaxis]) diffsort = np.sort(diff,axis=0) self.h = diffsort[self.k,:] ##Attempt to handle larger datasets more easily: else: sortinds = np.argsort(self.dataset) x = self.dataset[sortinds] h = np.zeros(len(x)) for i in np.arange(len(x)): lo = i - self.k hi = i + self.k + 1 if lo < 0: lo = 0 if hi > len(x): hi = len(x) diffs = abs(x[lo:hi]-x[i]) h[sortinds[i]] = np.sort(diffs)[self.k] self.h = h else: self.gauss_kde = gaussian_kde(self.dataset,bw_method=bandwidth,**kwargs) def renorm(self,norm): """Change the normalization""" self.norm = norm def evaluate(self,points): if not self.adaptive: return self.gauss_kde(points)*self.norm points = np.atleast_1d(points).astype(self.dataset.dtype) k = self.k npts = np.size(points) h = self.h X,Y = np.meshgrid(self.dataset,points) H = np.resize(h,(npts,self.n)) U = (X-Y)/H.astype(float) result = 1./self.n*1./H*self.kernel(U) return np.sum(result,axis=1)*self.norm __call__ = evaluate def integrate_box(self,low,high,forcequad=False,**kwargs): """Integrates over a box. Optionally force quad integration, even for non-adaptive. If adaptive mode is not being used, this will just call the `scipy.stats.gaussian_kde` method `integrate_box_1d`. Else, by default, it will call `scipy.integrate.quad`. If the `forcequad` flag is turned on, then that integration will be used even if adaptive mode is off. Parameters ---------- low : float Lower limit of integration high : float Upper limit of integration forcequad : bool If `True`, then use the quad integration even if adaptive mode is off. kwargs Keyword arguments passed to `scipy.integrate.quad`. """ if not self.adaptive and not forcequad: return self.gauss_kde.integrate_box_1d(low,high)*self.norm return quad(self.evaluate,low,high,**kwargs)[0] def resample(self,size=None,direct=None): if direct is None: direct = self.draw_direct size=int(size) if not self.adaptive: return np.squeeze(self.gauss_kde.resample(size=size)) if direct: inds = rand.randint(self.n,size=size) return self.dataset[inds] else: if size is None: size = self.n indices = rand.randint(0,self.n,size=size) means = self.dataset[indices] h = self.h[indices] fuzz = kerneldraw(size,self.kernelname)*h return np.squeeze(means + fuzz) draw = resample def epkernel(u): x = np.atleast_1d(u) y = 3./4*(1-x*x) y[((x>1) | (x < -1))] = 0 return y def gausskernel(u): return 1/np.sqrt(2*np.pi)*np.exp(-0.5*u*u) def tricubekernel(u): x = np.atleast_1d(u) y = 35./32*(1-x*x)**3 y[((x > 1) | (x < -1))] = 0 return y def kernelfn(kernel='tricube'): if kernel=='ep': return epkernel elif kernel=='gauss': return gausskernel elif kernel=='tricube': return tricubekernel def kerneldraw(size=1,kernel='tricube',exact=False): if kernel=='tricube': fn = lambda x: 1./2 + 35./32*x - 35./32*x**3 + 21./32*x**5 - 5./32*x**7 u = rand.random(size=size) if not exact: xs = np.linspace(-1,1,1e4) ys = fn(xs) inds = np.digitize(u,ys) return xs[inds] else: #old way (exact) rets = np.zeros(size) for i in np.arange(size): f = lambda x: u[i]-fn(x) rets[i] = newton(f,0,restrict=(-1,1)) return rets def deriv(f,c,dx=0.0001): """ deriv(f,c,dx) --> float Returns f'(x), computed as a symmetric difference quotient. """ return (f(c+dx)-f(c-dx))/(2*dx) def fuzzyequals(a,b,tol=0.0001): return abs(a-b) < tol def newton(f,c,tol=0.0001,restrict=None): """ newton(f,c) --> float Returns the x closest to c such that f(x) = 0 """ #print(c) if restrict: lo,hi = restrict if c < lo or c > hi: print(c) c = random*(hi-lo)+lo if fuzzyequals(f(c),0,tol): return c else: try: return newton(f,c-f(c)/deriv(f,c,tol),tol,restrict) except: return None
timothydmorton/simpledist
simpledist/kde.py
KDE.integrate_box
python
def integrate_box(self,low,high,forcequad=False,**kwargs): if not self.adaptive and not forcequad: return self.gauss_kde.integrate_box_1d(low,high)*self.norm return quad(self.evaluate,low,high,**kwargs)[0]
Integrates over a box. Optionally force quad integration, even for non-adaptive. If adaptive mode is not being used, this will just call the `scipy.stats.gaussian_kde` method `integrate_box_1d`. Else, by default, it will call `scipy.integrate.quad`. If the `forcequad` flag is turned on, then that integration will be used even if adaptive mode is off. Parameters ---------- low : float Lower limit of integration high : float Upper limit of integration forcequad : bool If `True`, then use the quad integration even if adaptive mode is off. kwargs Keyword arguments passed to `scipy.integrate.quad`.
train
https://github.com/timothydmorton/simpledist/blob/d9807c90a935bd125213445ffed6255af558f1ca/simpledist/kde.py#L131-L156
null
class KDE(object): """An implementation of a kernel density estimator allowing for adaptive kernels. If the `adaptive` keyword is set to `False`, then this will essentially be just a wrapper for the `scipy.stats.gaussian_kde` class. If adaptive, though, it allows for different kernels and different kernel widths according to the "K-nearest-neighbors" algorithm as discussed `here <http://en.wikipedia.org/wiki/Variable_kernel_density_estimation#Balloon_estimators>`_. The `fast` option does the NN calculation using broadcasting arrays rather than a brute-force sort. By default the fast option will be used for datasets smaller than 5000. Parameters ---------- dataset : array-like Data set from which to calculate the KDE. kernel : {'tricube','ep','gauss'}, optional Kernel function to use for adaptive estimator. adaptive : bool, optional Flag whether or not to use adaptive KDE. If this is false, then this class will just be a wrapper for `scipy.stats.gaussian_kde`. k : `None` or int, optional Number to use for K-nearest-neighbor algorithm. If `None`, then it will be set to the `N/4`, where `N` is the size of the dataset. fast : `None` or bool, optional If `None`, then `fast = N < 5001`, where `N` is the size of the dataset. `fast=True` will force array calculations, which will use lots of RAM if the dataset is large. norm : float, optional Allows the normalization of the distribution to be something other than unity bandwidth : `None` or float, optional Passed to `scipy.stats.gaussian_kde` if not using adaptive mode. weights : array-like, optional Not yet implemented. draw_direct : bool, optional If `True`, then resampling will be just a bootstrap resampling of the input samples. If `False`, then resampling will actually resample each individual kernel (not recommended for large-ish datasets). kwargs Keyword arguments passed to `scipy.stats.gaussian_kde` if adaptive mode is not being used. """ def __init__(self,dataset,kernel='tricube',adaptive=True,k=None, fast=None,norm=1.,bandwidth=None,weights=None, draw_direct=False,**kwargs): self.dataset = np.atleast_1d(dataset) self.weights = weights self.n = np.size(dataset) self.kernel = kernelfn(kernel) self.kernelname = kernel self.bandwidth = bandwidth self.draw_direct = draw_direct if k: self.k = k else: self.k = self.n/4 self.norm=norm self.adaptive = adaptive self.fast = fast if adaptive: if self.fast==None: self.fast = self.n < 5001 if self.fast: #d1,d2 = np.meshgrid(self.dataset,self.dataset) #use broadcasting instead of meshgrid diff = np.absolute(self.dataset - self.dataset[:,np.newaxis]) diffsort = np.sort(diff,axis=0) self.h = diffsort[self.k,:] ##Attempt to handle larger datasets more easily: else: sortinds = np.argsort(self.dataset) x = self.dataset[sortinds] h = np.zeros(len(x)) for i in np.arange(len(x)): lo = i - self.k hi = i + self.k + 1 if lo < 0: lo = 0 if hi > len(x): hi = len(x) diffs = abs(x[lo:hi]-x[i]) h[sortinds[i]] = np.sort(diffs)[self.k] self.h = h else: self.gauss_kde = gaussian_kde(self.dataset,bw_method=bandwidth,**kwargs) def renorm(self,norm): """Change the normalization""" self.norm = norm def evaluate(self,points): if not self.adaptive: return self.gauss_kde(points)*self.norm points = np.atleast_1d(points).astype(self.dataset.dtype) k = self.k npts = np.size(points) h = self.h X,Y = np.meshgrid(self.dataset,points) H = np.resize(h,(npts,self.n)) U = (X-Y)/H.astype(float) result = 1./self.n*1./H*self.kernel(U) return np.sum(result,axis=1)*self.norm __call__ = evaluate def resample(self,size=None,direct=None): if direct is None: direct = self.draw_direct size=int(size) if not self.adaptive: return np.squeeze(self.gauss_kde.resample(size=size)) if direct: inds = rand.randint(self.n,size=size) return self.dataset[inds] else: if size is None: size = self.n indices = rand.randint(0,self.n,size=size) means = self.dataset[indices] h = self.h[indices] fuzz = kerneldraw(size,self.kernelname)*h return np.squeeze(means + fuzz) draw = resample
ttm/socialLegacy
social/twiter.py
Twitter.searchTag
python
def searchTag(self,HTAG="#arenaNETmundial"): search = t.search(q=HTAG,count=100,result_type="recent") ss=search[:] search = t.search(q=HTAG,count=150,max_id=ss[-1]['id']-1,result_type="recent") #search = t.search(q=HTAG,count=150,since_id=ss[-1]['id'],result_type="recent") while seach: ss+=search[:] search = t.search(q=HTAG,count=150,max_id=ss[-1]['id']-1,result_type="recent")
Set Twitter search or stream criteria for the selection of tweets
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/twiter.py#L24-L32
null
class Twitter: """Simplified Twitter interface for Stability observance # function to set authentication: __init__() # function to set hashtag and other tweets selection criteria: searchTag() # function to search tweets: searchTag() # function to stream tweets: void """ def __init__(self,app_key=TWITTER_API_KEY, app_secret=TWITTER_API_KEY_SECRET, oauth_token=TWITTER_ACCESS_TOKEN, oauth_token_secret=TWITTER_ACCESS_TOKEN_SECRET): """Start twitter seach and stream interface""" self.t = Twython(app_key =app_key , app_secret =app_secret , oauth_token =oauth_token , oauth_token_secret=oauth_token_secret)
ttm/socialLegacy
social/utils.py
makeRetweetNetwork
python
def makeRetweetNetwork(tweets): G=x.DiGraph() G_=x.DiGraph() for tweet in tweets: text=tweet["text"] us=tweet["user"]["screen_name"] if text.startswith("RT @"): prev_us=text.split(":")[0].split("@")[1] #print(us,prev_us,text) if G.has_edge(prev_us,us): G[prev_us][us]["weight"]+=1 G_[prev_us][us]["weight"]+=1 else: G.add_edge(prev_us, us, weight=1.) G_.add_edge(prev_us, us, weight=1.) if us not in G_.nodes(): G_.add_node(us) return G,G_
Receives tweets, returns directed retweet networks. Without and with isolated nodes.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/utils.py#L2-L23
null
import networkx as x def makeRetweetNetwork(tweets): """Receives tweets, returns directed retweet networks. Without and with isolated nodes. """ G=x.DiGraph() G_=x.DiGraph() for tweet in tweets: text=tweet["text"] us=tweet["user"]["screen_name"] if text.startswith("RT @"): prev_us=text.split(":")[0].split("@")[1] #print(us,prev_us,text) if G.has_edge(prev_us,us): G[prev_us][us]["weight"]+=1 G_[prev_us][us]["weight"]+=1 else: G.add_edge(prev_us, us, weight=1.) G_.add_edge(prev_us, us, weight=1.) if us not in G_.nodes(): G_.add_node(us) return G,G_ class GDFgraph: """Read GDF graph into networkX""" def __init__(self,filename="../data/RenatoFabbri06022014.gdf"): with open(filename,"r") as f: self.data=f.read() self.lines=self.data.split("\n") columns=self.lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in self.lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) self.data_friendships=data_friendships self.data_friends=data_friends self.n_friends=len(data_friends[column_names[0]]) self.n_friendships=len(data_friendships[column_names2[0]]) self.makeNetwork() def makeNetwork(self): """Makes graph object from .gdf loaded data""" if "weight" in self.data_friendships.keys(): self.G=G=x.DiGraph() else: self.G=G=x.Graph() F=self.data_friends for friendn in range(self.n_friends): if "posts" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], posts=F["posts"][friendn]) elif "agerank" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn], agerank=F["agerank"][friendn]) else: G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn]) F=self.data_friendships for friendshipn in range(self.n_friendships): if "weight" in F.keys(): G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn],weight=F["weight"][friendshipn]) else: G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn])
ttm/socialLegacy
social/utils.py
GDFgraph.makeNetwork
python
def makeNetwork(self): if "weight" in self.data_friendships.keys(): self.G=G=x.DiGraph() else: self.G=G=x.Graph() F=self.data_friends for friendn in range(self.n_friends): if "posts" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], posts=F["posts"][friendn]) elif "agerank" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn], agerank=F["agerank"][friendn]) else: G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn]) F=self.data_friendships for friendshipn in range(self.n_friendships): if "weight" in F.keys(): G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn],weight=F["weight"][friendshipn]) else: G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn])
Makes graph object from .gdf loaded data
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/utils.py#L57-L85
null
class GDFgraph: """Read GDF graph into networkX""" def __init__(self,filename="../data/RenatoFabbri06022014.gdf"): with open(filename,"r") as f: self.data=f.read() self.lines=self.data.split("\n") columns=self.lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in self.lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) self.data_friendships=data_friendships self.data_friends=data_friends self.n_friends=len(data_friends[column_names[0]]) self.n_friendships=len(data_friendships[column_names2[0]]) self.makeNetwork()
ttm/socialLegacy
social/fb/gml2rdf.py
triplifyGML
python
def triplifyGML(dpath="../data/fb/",fname="foo.gdf",fnamei="foo_interaction.gdf", fpath="./fb/",scriptpath=None,uid=None,sid=None,fb_link=None,ego=True,umbrella_dir=None): c("iniciado tripgml") if sum(c.isdigit() for c in fname)==4: year=re.findall(r".*(\d\d\d\d).gml",fname)[0][0] B.datetime_snapshot=datetime.date(*[int(i) for i in (year)]) if sum(c.isdigit() for c in fname)==12: day,month,year,hour,minute=re.findall(r".*(\d\d)(\d\d)(\d\d\d\d)_(\d\d)(\d\d).gml",fname)[0] B.datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]) if sum(c.isdigit() for c in fname)==14: day,month,year,hour,minute,second=re.findall(r".*(\d\d)(\d\d)(\d\d\d\d)_(\d\d)(\d\d)(\d\d).gml",fname)[0] B.datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute,second)]) elif sum(c.isdigit() for c in fname)==8: day,month,year=re.findall(r".*(\d\d)(\d\d)(\d\d\d\d).gml",fname)[0] B.datetime_snapshot=datetime.date(*[int(i) for i in (year,month,day)]) B.datetime_snapshot_=datetime_snapshot.isoformat() B.fname=fname B.fnamei=fnamei B.name=fname.replace(".gml","_gml") if fnamei: B.namei=fnamei[:-4] B.ego=ego B.friendship=bool(fname) B.interaction=bool(fnamei) B.sid=sid B.uid=uid B.scriptpath=scriptpath B.fb_link=fb_link B.dpath=dpath B.fpath=fpath B.prefix="https://raw.githubusercontent.com/OpenLinkedSocialData/{}master/".format(umbrella_dir) B.umbrella_dir=umbrella_dir c("antes de ler") #fnet=S.fb.readGML(dpath+fname) # return networkx graph fnet=S.fb.readGML2(dpath+fname) # return networkx graph # return fnet c("depois de ler, antes de fazer rdf") fnet_=rdfFriendshipNetwork(fnet) # return rdflib graph if B.interaction: inet=S.fb.readGML(dpath+fnamei) # return networkx graph inet_=rdfInteractionNetwork(inet) # return rdflib graph else: inet_=0 meta=makeMetadata(fnet_,inet_) # return rdflib graph with metadata about the structure c("depois de rdf, escrita em disco") writeAllFB(fnet_,inet_,meta) # write linked data tree c("cabo")
Produce a linked data publication tree from a standard GML file. INPUTS: ====== => the data directory path => the file name (fname) of the friendship network => the file name (fnamei) of the interaction network => the final path (fpath) for the tree of files to be created => a path to the script that is calling this function (scriptpath) => the numeric id (uid) of the facebook user or group of the network(s) => the numeric id (sid) of the facebook user or group of the network (s) => the facebook link (fb_link) of the user or group => the network is from a user (ego==True) or a group (ego==False) OUTPUTS: ======= the tree in the directory fpath.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fb/gml2rdf.py#L4-L67
[ "def rdfFriendshipNetwork(fnet):\n tg=P.rdf.makeBasicGraph([[\"po\",\"fb\"],[P.rdf.ns.per,P.rdf.ns.fb]],\"Facebook friendship network from {} . Ego: {}\".format(B.name,B.ego))\n #if sum([(\"user\" in i) for i in fnet[\"individuals\"][\"label\"]])==len(fnet[\"individuals\"][\"label\"]):\n # # nomes falsos, ids espurios\n # anonymized=True\n #else:\n # anonymized=False\n B.fanon=False\n\n tkeys=list(fnet[\"individuals\"].keys())\n foo={\"uris\":[],\"vals\":[]}\n for tkey in tkeys:\n if tkey != \"groupid\":\n foo[\"uris\"]+=[eval(\"P.rdf.ns.fb.\"+trans(tkey))]\n foo[\"vals\"]+=[fnet[\"individuals\"][tkey]]\n if \"groupid\" in tkeys:\n B.groupuid=fnet[\"individuals\"][\"groupid\"][0]\n tkeys.remove(\"groupid\")\n else:\n B.groupuid=None\n iname= tkeys.index(\"name\")\n B.uid_names={}\n for vals_ in zip(*foo[\"vals\"]):\n vals_=list(vals_)\n cid=vals_[iname]\n foo_=foo[\"uris\"][:]\n take=0\n name_=\"{}-{}\".format(B.name,cid)\n B.uid_names[cid]=name_\n vals_=[el for i,el in enumerate(vals_) if i not in (iname,)]\n foo_= [el for i,el in enumerate(foo_) if i not in (iname,)]\n i=0\n ii=[]\n for val in vals_:\n if not val:\n ii+=[i]\n i+=1\n vals_=[val for i,val in enumerate(vals_) if i not in ii]\n\n# take+=1\n# if not vals_[isex-take]:\n# vals_=[el for i,el in enumerate(vals_) if i not in (isex-take,)]\n# foo_= [el for i,el in enumerate(foo_) if i not in (isex-take,)]\n# take+=1\n# if not vals_[ilocale-take]:\n# vals_=[el for i,el in enumerate(vals_) if i not in (ilocale-take,)]\n# foo_= [el for i,el in enumerate(foo_) if i not in (ilocale-take,)]\n ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name_)\n P.rdf.link([tg],ind,None,foo_,\n vals_)\n B.nfriends=len(foo[\"vals\"][0])\n #if anonymized:\n # B.fvars=[trans(i) for j,i in enumerate(tkeys) if j not in (ilabel,iname)]\n #else:\n # B.fvars=[trans(i) for i in tkeys]\n B.fvars=[trans(i) for j,i in enumerate(tkeys) if j not in (iname,)]\n\n friendships_=[fnet[\"relations\"][i] for i in (\"node1\",\"node2\")]\n c(\"escritos participantes\")\n i=1\n for uid1,uid2 in zip(*friendships_):\n uids=[r.URIRef(P.rdf.ns.fb.Participant+\"#\"+B.uid_names[i]) for i in (uid1,uid2)]\n P.rdf.link_([tg],uids[0],None,[P.rdf.ns.fb.friend],[uids[1]])\n if (i%1000)==0:\n c(i)\n i+=1\n P.rdf.G(tg[0],P.rdf.ns.fb.friend,\n P.rdf.ns.rdf.type,\n P.rdf.ns.owl.SymmetricProperty)\n B.nfriendships=len(friendships_[0])\n c(\"escritas amizades\")\n return tg\n", "def makeMetadata(fnet,inet):\n desc=\"facebook network from {} . Ego: {}. Friendship: {}. Interaction: {}.\".format(B.name,B.ego,B.friendship,B.interaction)\n tg2=P.rdf.makeBasicGraph([[\"po\",\"fb\"],[P.rdf.ns.per,P.rdf.ns.fb]],\"Metadata for the \"+desc)\n aname=B.name+\"_fb\"\n ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot,\n aname,\"Snapshot {}\".format(aname))\n ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot,\n aname,\"Snapshot {}\".format(aname))\n\n foo={\"uris\":[],\"vals\":[]}\n if ego:\n if B.uid:\n foo[\"uris\"].append(P.rdf.ns.fb.uid)\n foo[\"vals\"].append(B.uid)\n if B.sid:\n foo[\"uris\"].append(P.rdf.ns.fb.sid)\n foo[\"vals\"].append(B.sid)\n else:\n if B.uid:\n foo[\"uris\"].append(P.rdf.ns.fb.groupID)\n foo[\"vals\"].append(B.uid)\n if B.sid:\n foo[\"uris\"].append(P.rdf.ns.fb.groupSID)\n foo[\"vals\"].append(B.sid)\n if B.groupuid:\n foo[\"uris\"].append(P.rdf.ns.fb.groupID)\n foo[\"vals\"].append(B.groupuid)\n if B.fb_link:\n if type(B.fb_link) not in (type([2,3]),type((2,3))):\n foo[\"uris\"].append(P.rdf.ns.fb.fbLink)\n foo[\"vals\"].append(B.fb_link)\n else:\n for link in B.fb_link:\n foo[\"uris\"].append(P.rdf.ns.fb.fbLink)\n foo[\"vals\"].append(link)\n if B.friendship:\n B.ffile=\"{}{}/base/{}\".format(B.prefix,aname,B.fname)\n foo[\"uris\"]+=[P.rdf.ns.fb.originalFriendshipFile,\n P.rdf.ns.po.friendshipXMLFile,\n P.rdf.ns.po.friendshipTTLFile]+\\\n [ P.rdf.ns.fb.nFriends,\n P.rdf.ns.fb.nFriendships,\n P.rdf.ns.fb.fAnon ]+\\\n [P.rdf.ns.fb.friendAttribute]*len(B.fvars)\n B.frdf_file=\"{}{}/rdf/{}Friendship.owl\".format(B.prefix,aname,aname)\n foo[\"vals\"]+=[B.ffile,\n B.frdf_file,\n \"{}{}/rdf/{}Friendship.ttl\".format(B.prefix,aname,aname) ]+\\\n [B.nfriends,B.nfriendships,B.fanon]+list(B.fvars)\n\n if B.interaction:\n B.ifile=\"{}{}/base/{}\".format(B.prefix,aname,B.fnamei)\n foo[\"uris\"]+=[P.rdf.ns.fb.originalInteractionFile,\n P.rdf.ns.po.interactionXMLFile,\n P.rdf.ns.po.interactionTTLFile,]+\\\n [ P.rdf.ns.fb.nFriendsInteracted,\n P.rdf.ns.fb.nInteractions,\n P.rdf.ns.fb.iAnon ]+\\\n [ P.rdf.ns.fb.interactionFriendAttribute]*len(B.fvarsi)+\\\n [ P.rdf.ns.fb.interactionAttribute]*len(B.ivars)\n\n B.irdf_file=\"{}{}/rdf/{}Interaction.owl\".format(B.prefix,aname,aname)\n foo[\"vals\"]+=[B.ifile,\n B.irdf_file,\n \"{}{}/rdf/{}Interaction.ttl\".format(B.prefix,aname,aname),]+\\\n [B.nfriendsi,B.ninteractions,B.ianon]+list(B.fvarsi)+list(B.ivars)\n\n foo[\"uris\"]+=[\n P.rdf.ns.fb.ego,\n P.rdf.ns.fb.friendship,\n P.rdf.ns.fb.interaction,\n ]\n foo[\"vals\"]+=[B.ego,B.friendship,B.interaction]\n\n #https://github.com/OpenLinkedSocialData/fbGroups/tree/master/AdornoNaoEhEnfeite29032013_fb\n B.available_dir=\"https://github.com/OpenLinkedSocialData/{}tree/master/{}\".format(B.umbrella_dir,aname)\n B.mrdf_file=\"{}{}/rdf/{}Meta.owl\".format(B.prefix,aname,aname)\n P.rdf.link([tg2],ind,\"Snapshot {}\".format(aname),\n [P.rdf.ns.po.createdAt,\n P.rdf.ns.po.triplifiedIn,\n P.rdf.ns.po.donatedBy,\n P.rdf.ns.po.availableAt,\n P.rdf.ns.po.discorveryRDFFile,\n P.rdf.ns.po.discoveryTTLFile,\n P.rdf.ns.po.acquiredThrough,\n P.rdf.ns.rdfs.comment,\n ]+foo[\"uris\"],\n [B.datetime_snapshot,\n datetime.datetime.now(),\n B.name,\n B.available_dir,\n B.mrdf_file,\n \"{}{}/rdf/{}Meta.ttl\".format(B.prefix,aname,aname),\n \"Netvizz\",\n desc,\n ]+foo[\"vals\"])\n ind2=P.rdf.IC([tg2],P.rdf.ns.po.Platform,\"Facebook\")\n P.rdf.link_([tg2],ind,\"Snapshot {}\".format(aname),\n [P.rdf.ns.po.socialProtocol],\n [ind2],\n [\"Facebook\"])\n #for friend_attr in fg2[\"friends\"]:\n return tg2\n", "def writeAllFB(fnet,inet,mnet):\n aname=B.name+\"_fb\"\n fpath_=\"{}{}/\".format(B.fpath,aname)\n if B.friendship:\n P.rdf.writeAll(fnet,aname+\"Friendship\",fpath_,False,1)\n if B.interaction:\n P.rdf.writeAll(inet,aname+\"Interaction\",fpath_)\n # copia o script que gera este codigo\n if not os.path.isdir(fpath_+\"scripts\"):\n os.mkdir(fpath_+\"scripts\")\n shutil.copy(scriptpath,fpath_+\"scripts/\")\n # copia do base data\n if not os.path.isdir(fpath_+\"base\"):\n os.mkdir(fpath_+\"base\")\n shutil.copy(B.dpath+B.fname,fpath_+\"base/\")\n if B.interaction:\n shutil.copy(B.dpath+B.fnamei,fpath_+\"base/\")\n tinteraction=\"\"\"\\n{} individuals with metadata {}\nand {} interactions with metadata {} constitute the interaction \nnetwork in file:\n{}\n(anonymized: {}).\"\"\".format( B.nfriendsi,str(B.fvarsi),\n B.ninteractions,str(B.ivars),B.irdf_file,\n B.ianon)\n originals=\"{}\\n{}\".format(B.ffile,B.ifile)\n else:\n tinteraction=\"\"\n originals=B.ffile\n\n\n P.rdf.writeAll(mnet,aname+\"Meta\",fpath_,1)\n # faz um README\n with open(fpath_+\"README\",\"w\") as f:\n f.write(\"\"\"This repo delivers RDF data from the facebook\nfriendship network of {} collected around {}.\n{} individuals with metadata {}\nand {} friendships constitute the friendship network in file:\n{}\n(anonymized: {}).{}\nMetadata for discovery is in file:\n{}\nOriginal files:\n{}\nEgo network: {}\nFriendship network: {}\nInteraction network: {}\nAll files should be available at the git repository:\n{}\n\\n\"\"\".format(\n B.name,B.datetime_snapshot_,\n B.nfriends,str(B.fvars),\n B.nfriendships, B.frdf_file,\n# B.fanon,\n \"FALSE, but no id\",\n tinteraction,\n B.mrdf_file,originals,\n B.ego, B.friendship,B.interaction,B.available_dir\n ))\n" ]
import percolation as P, social as S, rdflib as r, builtins as B, re, datetime, os, shutil c=P.utils.check def trans(tkey): if tkey=="name": return "numericID" if tkey=="label": return "name" return tkey def rdfInteractionNetwork(fnet): tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"Facebook interaction network from {} . Ego: {}".format(B.name,B.ego)) tkeys=list(fnet["individuals"].keys()) if sum([("user" in i) for i in fnet["individuals"]["label"]])==len(fnet["individuals"]["label"]): # nomes falsos, ids espurios anonymized=True else: anonymized=False B.ianon=anonymized foo={"uris":[],"vals":[]} for tkey in tkeys: foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fnet["individuals"][tkey]] iname= tkeys.index("name") ilabel=tkeys.index("label") B.nfriendsi=len(foo["vals"][0]) if anonymized: B.fvarsi=[trans(i) for j,i in enumerate(tkeys) if j not in (ilabel,iname)] else: B.fvarsi=[trans(i) for i in tkeys] icount=0 #uid_names={} for vals_ in zip(*foo["vals"]): vals_=list(vals_) cid=vals_[iname] foo_=foo["uris"][:] if anonymized: if cid in B.uid_names.keys(): name_=B.uid_names[cid] else: anon_name=vals_[ilabel] name_="{}-{}".format(B.namei,anon_name) B.uid_names[cid]=name_ #anon_name=vals_[ilabel] #name_="{}-{}".format(B.name,anon_name) #uid_names[cid]=name_ vals_=[el for i,el in enumerate(vals_) if i not in (ilabel,iname)] foo_= [el for i,el in enumerate(foo_) if i not in (ilabel,iname)] elif not vals_[ilabel]: vals_=[el for i,el in enumerate(vals_) if i not in (ilabel,)] foo_= [el for i,el in enumerate(foo_) if i not in (ilabel,)] name_=cid #name_="po:noname-{}-{}-{}".format(cid,B.groupuid,B.datetime_snapshot) #c("{} --- {}".format(name_, vals_[ilabel])) #vals_=list(vals_) #vals_[ilabel]=name_ else: name_,label=[foo["vals"][i][icount] for i in (iname,ilabel)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name_) P.rdf.link([tg],ind,None,foo_, vals_) icount+=1 B.ivars=["node1","node2","weight"] interactions_=[fnet["relations"][i] for i in B.ivars] B.ninteractions=len(interactions_[0]) c("escritos participantes") i=1 for uid1,uid2,weight in zip(*interactions_): weight_=int(weight) if weight_-weight != 0: raise ValueError("float weights in fb interaction networks?") if anonymized: uid1=uid_names[uid1] uid2=uid_names[uid2] flabel="{}-{}".format(uid1,uid2) else: flabel="{}-{}-{}-{}".format(B.fname,B.datetime_snapshot_,uid1,uid2) ind=P.rdf.IC([tg],P.rdf.ns.fb.Interaction,flabel) uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+str(i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,None,[P.rdf.ns.fb.iFrom,P.rdf.ns.fb.iTo], uids,draw=False) P.rdf.link([tg],ind,None,[P.rdf.ns.fb.weight], [weight_],draw=False) if (i%1000)==0: c(i) i+=1 c("escritas amizades") return tg def rdfFriendshipNetwork(fnet): tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"Facebook friendship network from {} . Ego: {}".format(B.name,B.ego)) #if sum([("user" in i) for i in fnet["individuals"]["label"]])==len(fnet["individuals"]["label"]): # # nomes falsos, ids espurios # anonymized=True #else: # anonymized=False B.fanon=False tkeys=list(fnet["individuals"].keys()) foo={"uris":[],"vals":[]} for tkey in tkeys: if tkey != "groupid": foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fnet["individuals"][tkey]] if "groupid" in tkeys: B.groupuid=fnet["individuals"]["groupid"][0] tkeys.remove("groupid") else: B.groupuid=None iname= tkeys.index("name") B.uid_names={} for vals_ in zip(*foo["vals"]): vals_=list(vals_) cid=vals_[iname] foo_=foo["uris"][:] take=0 name_="{}-{}".format(B.name,cid) B.uid_names[cid]=name_ vals_=[el for i,el in enumerate(vals_) if i not in (iname,)] foo_= [el for i,el in enumerate(foo_) if i not in (iname,)] i=0 ii=[] for val in vals_: if not val: ii+=[i] i+=1 vals_=[val for i,val in enumerate(vals_) if i not in ii] # take+=1 # if not vals_[isex-take]: # vals_=[el for i,el in enumerate(vals_) if i not in (isex-take,)] # foo_= [el for i,el in enumerate(foo_) if i not in (isex-take,)] # take+=1 # if not vals_[ilocale-take]: # vals_=[el for i,el in enumerate(vals_) if i not in (ilocale-take,)] # foo_= [el for i,el in enumerate(foo_) if i not in (ilocale-take,)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name_) P.rdf.link([tg],ind,None,foo_, vals_) B.nfriends=len(foo["vals"][0]) #if anonymized: # B.fvars=[trans(i) for j,i in enumerate(tkeys) if j not in (ilabel,iname)] #else: # B.fvars=[trans(i) for i in tkeys] B.fvars=[trans(i) for j,i in enumerate(tkeys) if j not in (iname,)] friendships_=[fnet["relations"][i] for i in ("node1","node2")] c("escritos participantes") i=1 for uid1,uid2 in zip(*friendships_): uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+B.uid_names[i]) for i in (uid1,uid2)] P.rdf.link_([tg],uids[0],None,[P.rdf.ns.fb.friend],[uids[1]]) if (i%1000)==0: c(i) i+=1 P.rdf.G(tg[0],P.rdf.ns.fb.friend, P.rdf.ns.rdf.type, P.rdf.ns.owl.SymmetricProperty) B.nfriendships=len(friendships_[0]) c("escritas amizades") return tg def makeMetadata(fnet,inet): desc="facebook network from {} . Ego: {}. Friendship: {}. Interaction: {}.".format(B.name,B.ego,B.friendship,B.interaction) tg2=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"Metadata for the "+desc) aname=B.name+"_fb" ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) foo={"uris":[],"vals":[]} if ego: if B.uid: foo["uris"].append(P.rdf.ns.fb.uid) foo["vals"].append(B.uid) if B.sid: foo["uris"].append(P.rdf.ns.fb.sid) foo["vals"].append(B.sid) else: if B.uid: foo["uris"].append(P.rdf.ns.fb.groupID) foo["vals"].append(B.uid) if B.sid: foo["uris"].append(P.rdf.ns.fb.groupSID) foo["vals"].append(B.sid) if B.groupuid: foo["uris"].append(P.rdf.ns.fb.groupID) foo["vals"].append(B.groupuid) if B.fb_link: if type(B.fb_link) not in (type([2,3]),type((2,3))): foo["uris"].append(P.rdf.ns.fb.fbLink) foo["vals"].append(B.fb_link) else: for link in B.fb_link: foo["uris"].append(P.rdf.ns.fb.fbLink) foo["vals"].append(link) if B.friendship: B.ffile="{}{}/base/{}".format(B.prefix,aname,B.fname) foo["uris"]+=[P.rdf.ns.fb.originalFriendshipFile, P.rdf.ns.po.friendshipXMLFile, P.rdf.ns.po.friendshipTTLFile]+\ [ P.rdf.ns.fb.nFriends, P.rdf.ns.fb.nFriendships, P.rdf.ns.fb.fAnon ]+\ [P.rdf.ns.fb.friendAttribute]*len(B.fvars) B.frdf_file="{}{}/rdf/{}Friendship.owl".format(B.prefix,aname,aname) foo["vals"]+=[B.ffile, B.frdf_file, "{}{}/rdf/{}Friendship.ttl".format(B.prefix,aname,aname) ]+\ [B.nfriends,B.nfriendships,B.fanon]+list(B.fvars) if B.interaction: B.ifile="{}{}/base/{}".format(B.prefix,aname,B.fnamei) foo["uris"]+=[P.rdf.ns.fb.originalInteractionFile, P.rdf.ns.po.interactionXMLFile, P.rdf.ns.po.interactionTTLFile,]+\ [ P.rdf.ns.fb.nFriendsInteracted, P.rdf.ns.fb.nInteractions, P.rdf.ns.fb.iAnon ]+\ [ P.rdf.ns.fb.interactionFriendAttribute]*len(B.fvarsi)+\ [ P.rdf.ns.fb.interactionAttribute]*len(B.ivars) B.irdf_file="{}{}/rdf/{}Interaction.owl".format(B.prefix,aname,aname) foo["vals"]+=[B.ifile, B.irdf_file, "{}{}/rdf/{}Interaction.ttl".format(B.prefix,aname,aname),]+\ [B.nfriendsi,B.ninteractions,B.ianon]+list(B.fvarsi)+list(B.ivars) foo["uris"]+=[ P.rdf.ns.fb.ego, P.rdf.ns.fb.friendship, P.rdf.ns.fb.interaction, ] foo["vals"]+=[B.ego,B.friendship,B.interaction] #https://github.com/OpenLinkedSocialData/fbGroups/tree/master/AdornoNaoEhEnfeite29032013_fb B.available_dir="https://github.com/OpenLinkedSocialData/{}tree/master/{}".format(B.umbrella_dir,aname) B.mrdf_file="{}{}/rdf/{}Meta.owl".format(B.prefix,aname,aname) P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.discorveryRDFFile, P.rdf.ns.po.discoveryTTLFile, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, ]+foo["uris"], [B.datetime_snapshot, datetime.datetime.now(), B.name, B.available_dir, B.mrdf_file, "{}{}/rdf/{}Meta.ttl".format(B.prefix,aname,aname), "Netvizz", desc, ]+foo["vals"]) ind2=P.rdf.IC([tg2],P.rdf.ns.po.Platform,"Facebook") P.rdf.link_([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.socialProtocol], [ind2], ["Facebook"]) #for friend_attr in fg2["friends"]: return tg2 def writeAllFB(fnet,inet,mnet): aname=B.name+"_fb" fpath_="{}{}/".format(B.fpath,aname) if B.friendship: P.rdf.writeAll(fnet,aname+"Friendship",fpath_,False,1) if B.interaction: P.rdf.writeAll(inet,aname+"Interaction",fpath_) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(B.dpath+B.fname,fpath_+"base/") if B.interaction: shutil.copy(B.dpath+B.fnamei,fpath_+"base/") tinteraction="""\n{} individuals with metadata {} and {} interactions with metadata {} constitute the interaction network in file: {} (anonymized: {}).""".format( B.nfriendsi,str(B.fvarsi), B.ninteractions,str(B.ivars),B.irdf_file, B.ianon) originals="{}\n{}".format(B.ffile,B.ifile) else: tinteraction="" originals=B.ffile P.rdf.writeAll(mnet,aname+"Meta",fpath_,1) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} collected around {}. {} individuals with metadata {} and {} friendships constitute the friendship network in file: {} (anonymized: {}).{} Metadata for discovery is in file: {} Original files: {} Ego network: {} Friendship network: {} Interaction network: {} All files should be available at the git repository: {} \n""".format( B.name,B.datetime_snapshot_, B.nfriends,str(B.fvars), B.nfriendships, B.frdf_file, # B.fanon, "FALSE, but no id", tinteraction, B.mrdf_file,originals, B.ego, B.friendship,B.interaction,B.available_dir ))
ttm/socialLegacy
social/fb/fb.py
triplifyGML
python
def triplifyGML(fname="foo.gml",fpath="./fb/",scriptpath=None,uid=None,sid=None,extra_info=None): # aname=fname.split("/")[-1].split(".")[0] aname=fname.split("/")[-1].split(".")[0] if "RonaldCosta" in fname: aname=fname.split("/")[-1].split(".")[0] name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_="Ronald Scherolt Costa" elif "AntonioAnzoategui" in fname: aname=re.findall(".*/([a-zA-Z]*\d*)",fname)[0] name,year,month,day,hour,minute=re.findall(r".*/([a-zA-Z]*).*_(\d+)_(\d*)_(\d*)_(\d*)_(\d*)_.*",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat()[:-3] name_="Antônio Anzoategui Fabbri" elif re.findall(".*/[a-zA-Z]*(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("[a-zA-Z]*_",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) aname+="_fb" name+="_fb" c("started snapshot",aname) tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"the {} facebook ego friendship network") tg2=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"RDF metadata for the facebook friendship network of my son") snapshot=P.rdf.IC([tg2],P.rdf.ns.po.FacebookSnapshot, aname,"Snapshot {}".format(aname)) extra_uri=extra_val=[] if extra_info: extra_uri=[NS.po.extraInfo] extra_val=[extra_info] P.rdf.link([tg2],snapshot,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.onlineTranslateXMLFile, P.rdf.ns.po.onlineTranslateTTLFile, P.rdf.ns.po.translateXMLFile, P.rdf.ns.po.translateTTLFile, P.rdf.ns.po.onlineMetaXMLFile, P.rdf.ns.po.onlineMetaTTLFile, P.rdf.ns.po.metaXMLFilename, P.rdf.ns.po.metaTTLFilename, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, P.rdf.ns.fb.uid, P.rdf.ns.fb.sid ]+extra_uri, [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")[-1]), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.rdf".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "{}Translate.rdf".format(aname), "{}Translate.ttl".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.rdf".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "{}Meta.owl".format(aname), "{}Meta.ttl".format(aname), "Netvizz", "The facebook friendship network from {}".format(name_), uid, sid ]+extra_val) #for friend_attr in fg2["friends"]: c((aname,name_,datetime_snapshot)) fg2=x.read_gml(fname) c("read gml") for uid in fg2: c(uid) ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,"{}-{}".format(aname,uid)) if "locale" in fg2.node[uid].keys(): data=[fg2.node[uid][attr] for attr in ("id","label","locale","sex","agerank","wallcount")] uris=[NS.fb.gid, NS.fb.name, NS.fb.locale, NS.fb.sex, NS.fb.agerank,NS.fb.wallcount] else: data=[fg2.node[uid][attr] for attr in ("id","label","sex","agerank","wallcount")] uris=[NS.fb.gid, NS.fb.name, NS.fb.sex, NS.fb.agerank,NS.fb.wallcount] P.rdf.link([tg],ind, None,uris,data,draw=False) P.rdf.link_([tg],ind,None,[NS.po.snapshot],[snapshot],draw=False) #friends_=[fg2["friends"][i] for i in ("name","label","locale","sex","agerank")] #for name,label,locale,sex,agerank in zip(*friends_): # ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) # P.rdf.link([tg],ind,label,[P.rdf.ns.fb.uid,P.rdf.ns.fb.name, # P.rdf.ns.fb.locale,P.rdf.ns.fb.sex, # P.rdf.ns.fb.agerank], # [name,label,locale,sex,agerank]) c("escritos participantes") #friendships_=[fg2["friendships"][i] for i in ("node1","node2")] i=1 for uid1,uid2 in fg2.edges(): flabel="{}-{}-{}".format(aname,uid1,uid2) ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship,flabel) uids=[P.rdf.IC(None,P.rdf.ns.fb.Participant,"{}-{}".format(aname,i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,flabel,[NS.po.snapshot]+[NS.fb.member]*2, [snapshot]+uids,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}/{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") #shutil.copy(this_dir+"/../tests/rdfMyFNetwork2.py",fpath+"scripts/") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,False) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} ({}) collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,aname,datetime_snapshot, fg2.number_of_nodes(), "name, locale (maybe), sex, agerank and wallcount", fg2.number_of_edges()))
Produce a linked data publication tree from a standard GML file. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook user of which fname holds a friendship network => the numeric id (sid) of the facebook user of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fb/fb.py#L11-L167
null
import time, os, pickle, shutil, datetime, re import networkx as x, rdflib as r from splinter import Browser from bs4 import BeautifulSoup import percolation as P c=P.utils.check this_dir = os.path.split(__file__)[0] NS=P.rdf.ns a=NS.rdf.type def triplifyGML(fname="foo.gml",fpath="./fb/",scriptpath=None,uid=None,sid=None,extra_info=None): """Produce a linked data publication tree from a standard GML file. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook user of which fname holds a friendship network => the numeric id (sid) of the facebook user of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.""" # aname=fname.split("/")[-1].split(".")[0] aname=fname.split("/")[-1].split(".")[0] if "RonaldCosta" in fname: aname=fname.split("/")[-1].split(".")[0] name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_="Ronald Scherolt Costa" elif "AntonioAnzoategui" in fname: aname=re.findall(".*/([a-zA-Z]*\d*)",fname)[0] name,year,month,day,hour,minute=re.findall(r".*/([a-zA-Z]*).*_(\d+)_(\d*)_(\d*)_(\d*)_(\d*)_.*",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat()[:-3] name_="Antônio Anzoategui Fabbri" elif re.findall(".*/[a-zA-Z]*(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("[a-zA-Z]*_",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) aname+="_fb" name+="_fb" c("started snapshot",aname) tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"the {} facebook ego friendship network") tg2=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"RDF metadata for the facebook friendship network of my son") snapshot=P.rdf.IC([tg2],P.rdf.ns.po.FacebookSnapshot, aname,"Snapshot {}".format(aname)) extra_uri=extra_val=[] if extra_info: extra_uri=[NS.po.extraInfo] extra_val=[extra_info] P.rdf.link([tg2],snapshot,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.onlineTranslateXMLFile, P.rdf.ns.po.onlineTranslateTTLFile, P.rdf.ns.po.translateXMLFile, P.rdf.ns.po.translateTTLFile, P.rdf.ns.po.onlineMetaXMLFile, P.rdf.ns.po.onlineMetaTTLFile, P.rdf.ns.po.metaXMLFilename, P.rdf.ns.po.metaTTLFilename, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, P.rdf.ns.fb.uid, P.rdf.ns.fb.sid ]+extra_uri, [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")[-1]), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.rdf".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "{}Translate.rdf".format(aname), "{}Translate.ttl".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.rdf".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "{}Meta.owl".format(aname), "{}Meta.ttl".format(aname), "Netvizz", "The facebook friendship network from {}".format(name_), uid, sid ]+extra_val) #for friend_attr in fg2["friends"]: c((aname,name_,datetime_snapshot)) fg2=x.read_gml(fname) c("read gml") for uid in fg2: c(uid) ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,"{}-{}".format(aname,uid)) if "locale" in fg2.node[uid].keys(): data=[fg2.node[uid][attr] for attr in ("id","label","locale","sex","agerank","wallcount")] uris=[NS.fb.gid, NS.fb.name, NS.fb.locale, NS.fb.sex, NS.fb.agerank,NS.fb.wallcount] else: data=[fg2.node[uid][attr] for attr in ("id","label","sex","agerank","wallcount")] uris=[NS.fb.gid, NS.fb.name, NS.fb.sex, NS.fb.agerank,NS.fb.wallcount] P.rdf.link([tg],ind, None,uris,data,draw=False) P.rdf.link_([tg],ind,None,[NS.po.snapshot],[snapshot],draw=False) #friends_=[fg2["friends"][i] for i in ("name","label","locale","sex","agerank")] #for name,label,locale,sex,agerank in zip(*friends_): # ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) # P.rdf.link([tg],ind,label,[P.rdf.ns.fb.uid,P.rdf.ns.fb.name, # P.rdf.ns.fb.locale,P.rdf.ns.fb.sex, # P.rdf.ns.fb.agerank], # [name,label,locale,sex,agerank]) c("escritos participantes") #friendships_=[fg2["friendships"][i] for i in ("node1","node2")] i=1 for uid1,uid2 in fg2.edges(): flabel="{}-{}-{}".format(aname,uid1,uid2) ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship,flabel) uids=[P.rdf.IC(None,P.rdf.ns.fb.Participant,"{}-{}".format(aname,i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,flabel,[NS.po.snapshot]+[NS.fb.member]*2, [snapshot]+uids,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}/{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") #shutil.copy(this_dir+"/../tests/rdfMyFNetwork2.py",fpath+"scripts/") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,False) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} ({}) collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,aname,datetime_snapshot, fg2.number_of_nodes(), "name, locale (maybe), sex, agerank and wallcount", fg2.number_of_edges())) def triplifyGDFInteraction(fname="foo.gdf",fpath="./fb/",scriptpath=None,uid=None,sid=None,dlink=None): """Produce a linked data publication tree from GDF files of a Facebook interaction network. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook group => the string id (sid) of the facebook group of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.""" #aname=fname.split("/")[-1].split(".")[0]+"_fb" aname=fname.split("/")[-1].split(".")[0] if re.findall("[a-zA-Z]*_[0-9]",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: datetime_snapshot=datetime.datetime(2013,3,15).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",aname)) aname+="_fb" name=aname tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"The facebook interaction network from the {} file".format(fname)) # drop de agraph tg2=P.rdf.makeBasicGraph([["po"],[P.rdf.ns.per]],"Metadata for my facebook ego friendship network RDF files") # drop de agraph ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) foo={"uris":[],"vals":[]} if sid: foo["uris"].append(P.rdf.ns.fb.sid) foo["vals"].append(sid) if uid: foo["uris"].append(P.rdf.ns.fb.uid) foo["vals"].append(uid) if dlink: foo["uris"].append(P.rdf.ns.fb.link) foo["vals"].append(dlink) P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.rdfFile, P.rdf.ns.po.ttlFile, P.rdf.ns.po.discorveryRDFFile, P.rdf.ns.po.discoveryTTLFile, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, ]+foo["uris"], [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "Netvizz", "The facebook friendship network from {}".format(name_), ]+foo["vals"]) #for friend_attr in fg2["friends"]: fg2=readGDF(fname) tkeys=list(fg2["friends"].keys()) def trans(tkey): if tkey=="name": return "uid" if tkey=="label": return "name" return tkey foo={"uris":[],"vals":[]} for tkey in tkeys: if tkey=="groupid": P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.uid,], [fg2["friends"][tkey][0]]) if tkey: foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fg2["friends"][tkey]] print(tkeys) iname=tkeys.index("name") ilabel=tkeys.index("label") icount=0 name_label={} for vals_ in zip(*foo["vals"]): name,label=[foo["vals"][i][icount] for i in (iname,ilabel)] if not label: label="po:noname" vals_=list(vals_) vals_[ilabel]=label name_label[name]=label ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) P.rdf.link([tg],ind,label,foo["uris"], vals_,draw=False) icount+=1 friendships_=[fg2["friendships"][i] for i in ("node1","node2")] c("escritos participantes") i=1 for uid1,uid2 in zip(*friendships_): flabel="{}-{}".format(uid1,uid2) labels=[name_label[uu] for uu in (uid1,uid2)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship, flabel) #flabel,"Friendship "+flabel) ind1=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid1) ind2=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid2) uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+str(i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,"Friendship "+flabel,[P.rdf.ns.fb.member]*2, uids,labels,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 P.rdf.G(tg[0],P.rdf.ns.fb.friend, P.rdf.ns.rdf.type, P.rdf.ns.owl.SymmetricProperty) c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,1) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,datetime_snapshot, len(fg2["friends"]["name"]), "facebook numeric id, name, locale, sex and agerank", len(fg2["friendships"]["node1"]) )) def triplifyGDF(fname="foo.gdf",fpath="./fb/",scriptpath=None,uid=None,sid=None,dlink=None): """Produce a linked data publication tree from a standard GDF file. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook user of which fname holds a friendship network => the numeric id (sid) of the facebook user of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.""" #aname=fname.split("/")[-1].split(".")[0]+"_fb" aname=fname.split("/")[-1].split(".")[0] if re.findall("[a-zA-Z]*_[0-9]",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: datetime_snapshot=datetime.datetime(2013,3,15).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",aname)) aname+="_fb" name=aname tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"My facebook ego friendship network") # drop de agraph tg2=P.rdf.makeBasicGraph([["po"],[P.rdf.ns.per]],"Metadata for my facebook ego friendship network RDF files") # drop de agraph ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) foo={"uris":[],"vals":[]} if sid: foo["uris"].append(P.rdf.ns.fb.sid) foo["vals"].append(sid) if uid: foo["uris"].append(P.rdf.ns.fb.uid) foo["vals"].append(uid) if dlink: foo["uris"].append(P.rdf.ns.fb.link) foo["vals"].append(dlink) P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.rdfFile, P.rdf.ns.po.ttlFile, P.rdf.ns.po.discorveryRDFFile, P.rdf.ns.po.discoveryTTLFile, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, ]+foo["uris"], [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "Netvizz", "The facebook friendship network from {}".format(name_), ]+foo["vals"]) #for friend_attr in fg2["friends"]: fg2=readGDF(fname) tkeys=list(fg2["friends"].keys()) def trans(tkey): if tkey=="name": return "uid" if tkey=="label": return "name" return tkey foo={"uris":[],"vals":[]} for tkey in tkeys: if tkey=="groupid": P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.uid,], [fg2["friends"][tkey][0]]) if tkey: foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fg2["friends"][tkey]] print(tkeys) iname=tkeys.index("name") ilabel=tkeys.index("label") icount=0 name_label={} for vals_ in zip(*foo["vals"]): name,label=[foo["vals"][i][icount] for i in (iname,ilabel)] if not label: label="po:noname" vals_=list(vals_) vals_[ilabel]=label name_label[name]=label ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) P.rdf.link([tg],ind,label,foo["uris"], vals_,draw=False) icount+=1 friendships_=[fg2["friendships"][i] for i in ("node1","node2")] c("escritos participantes") i=1 for uid1,uid2 in zip(*friendships_): flabel="{}-{}".format(uid1,uid2) labels=[name_label[uu] for uu in (uid1,uid2)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship, flabel) #flabel,"Friendship "+flabel) ind1=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid1) ind2=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid2) uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+str(i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,"Friendship "+flabel,[P.rdf.ns.fb.member]*2, uids,labels,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 P.rdf.G(tg[0],P.rdf.ns.fb.friend, P.rdf.ns.rdf.type, P.rdf.ns.owl.SymmetricProperty) c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,1) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,datetime_snapshot, len(fg2["friends"]["name"]), "facebook numeric id, name, locale, sex and agerank", len(fg2["friendships"]["node1"]) )) def makeRDF(readgdf_dict,fdir="../data/rdf/"): # return rdflib graph from the data rd=readgdf_dict # ns=namespaces=pe.namespaces(["rdf","rdfs","xsd", # basic namespaces # ]) # for friend in range(len(rd["friends"]["name"])): # pass def readGDF(filename="../data/RenatoFabbri06022014.gdf"): """Made to work with my own network. Check file to ease adaptation""" with open(filename,"r") as f: data=f.read() lines=data.split("\n") columns=lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) return {"friendships":data_friendships, "friends":data_friends} #self.makeNetwork() class GDFgraph: """Read GDF graph into networkX""" def __init__(self,filename="../data/RenatoFabbri06022014.gdf"): with open(filename,"r") as f: self.data=f.read() self.lines=self.data.split("\n") columns=self.lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in self.lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) self.data_friendships=data_friendships self.data_friends=data_friends self.n_friends=len(data_friends[column_names[0]]) self.n_friendships=len(data_friendships[column_names2[0]]) self.makeNetwork() def makeNetwork(self): """Makes graph object from .gdf loaded data""" if "weight" in self.data_friendships.keys(): self.G=G=x.DiGraph() else: self.G=G=x.Graph() F=self.data_friends for friendn in range(self.n_friends): if "posts" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], posts=F["posts"][friendn]) elif "agerank" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn], agerank=F["agerank"][friendn]) else: G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn]) F=self.data_friendships for friendshipn in range(self.n_friendships): if "weight" in F.keys(): G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn],weight=F["weight"][friendshipn]) else: G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn]) def readFBPost(fpath=""): """Extract information from HTML page with a Facebook post""" html=open(fpath,"rb") soup = BeautifulSoup(html, "lxml") return soup class ScrapyBrowser: """Opens a browser for user to login to facebook. Such browser pulls data as requested by user.""" def __init__(self,user_email=None, user_password=None,basedir="~/.social/"): self._BASE_DIR=basedir.replace("~",os.path.expanduser("~")) if not os.path.isdir(self._BASE_DIR): os.mkdir(self._BASE_DIR) print("Opening *Scrappy* firefox browser. Please wait.") self.browser=browser=Browser(wait_time=2) url="http://facebook.com" browser.visit(url) if (not user_email) or (not user_password): input("\n\n==> Input user and password and login, please.\ and then press <enter>") else: browser.fill("email",user_email) browser.fill("pass",user_password) browser.find_by_value("Log In").click() def getFriends(self,user_id="astronauta.mecanico",write=True): """Returns user_ids (that you have access) of the friends of your friend with user_ids""" while user_id not in self.browser.url: self.browser.visit("http://www.facebook.com/{}/friends".format(user_id), wait_time=3) #self.go("http://www.facebook.com/{}/friends".format(user_id)) T=time.time() while 1: h1=self.browser.evaluate_script("document.body.scrollHeight") self.browser.execute_script("window.scrollTo(0, document.body.scrollHeight);") h2=self.browser.evaluate_script("document.body.scrollHeight") if h1 != h2: T=time.time() elif time.time()-T>10: break #links=self.browser.find_link_by_partial_href("hc_location=friends_tab") links=self.browser.find_by_css(".fcb") friends=[] for link in links: name=link.value user_id_=link.find_by_tag("a")["href"].split("/")[-1].split("?")[0] friends.append((user_id_,name)) tdict={} tdict["name"]=self.browser.find_by_id("fb-timeline-cover-name").value tdict["user_id"]=user_id tdict["friends"]=friends infos=self.browser.find_by_css("._3c_") mutual=0 for info in infos: if info.value=="Mutual Friends": if info.find_by_css("._3d0").value: tdict["n_mutual"]=info.find_by_css("._3d0").value mutual=1 if info.value=="All Friends": tdict["n_friends"]=info.find_by_css("._3d0").value if mutual==0: links=self.browser.find_by_css("._gs6") if "Mutual" in links.value: tdict["n_mutual"]=links.value.split(" ")[0] if write: if not os.path.isdir("{}/fb_ids/".format(self._BASE_DIR)): os.mkdir("{}/fb_ids/".format(self._BASE_DIR)) with open("{}fb_ids/{}.pickle".format(self._BASE_DIR,user_id),"wb") as f: pickle.dump(tdict,f) self.tdict=tdict return tdict
ttm/socialLegacy
social/fb/fb.py
triplifyGDFInteraction
python
def triplifyGDFInteraction(fname="foo.gdf",fpath="./fb/",scriptpath=None,uid=None,sid=None,dlink=None): #aname=fname.split("/")[-1].split(".")[0]+"_fb" aname=fname.split("/")[-1].split(".")[0] if re.findall("[a-zA-Z]*_[0-9]",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: datetime_snapshot=datetime.datetime(2013,3,15).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",aname)) aname+="_fb" name=aname tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"The facebook interaction network from the {} file".format(fname)) # drop de agraph tg2=P.rdf.makeBasicGraph([["po"],[P.rdf.ns.per]],"Metadata for my facebook ego friendship network RDF files") # drop de agraph ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) foo={"uris":[],"vals":[]} if sid: foo["uris"].append(P.rdf.ns.fb.sid) foo["vals"].append(sid) if uid: foo["uris"].append(P.rdf.ns.fb.uid) foo["vals"].append(uid) if dlink: foo["uris"].append(P.rdf.ns.fb.link) foo["vals"].append(dlink) P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.rdfFile, P.rdf.ns.po.ttlFile, P.rdf.ns.po.discorveryRDFFile, P.rdf.ns.po.discoveryTTLFile, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, ]+foo["uris"], [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "Netvizz", "The facebook friendship network from {}".format(name_), ]+foo["vals"]) #for friend_attr in fg2["friends"]: fg2=readGDF(fname) tkeys=list(fg2["friends"].keys()) def trans(tkey): if tkey=="name": return "uid" if tkey=="label": return "name" return tkey foo={"uris":[],"vals":[]} for tkey in tkeys: if tkey=="groupid": P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.uid,], [fg2["friends"][tkey][0]]) if tkey: foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fg2["friends"][tkey]] print(tkeys) iname=tkeys.index("name") ilabel=tkeys.index("label") icount=0 name_label={} for vals_ in zip(*foo["vals"]): name,label=[foo["vals"][i][icount] for i in (iname,ilabel)] if not label: label="po:noname" vals_=list(vals_) vals_[ilabel]=label name_label[name]=label ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) P.rdf.link([tg],ind,label,foo["uris"], vals_,draw=False) icount+=1 friendships_=[fg2["friendships"][i] for i in ("node1","node2")] c("escritos participantes") i=1 for uid1,uid2 in zip(*friendships_): flabel="{}-{}".format(uid1,uid2) labels=[name_label[uu] for uu in (uid1,uid2)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship, flabel) #flabel,"Friendship "+flabel) ind1=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid1) ind2=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid2) uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+str(i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,"Friendship "+flabel,[P.rdf.ns.fb.member]*2, uids,labels,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 P.rdf.G(tg[0],P.rdf.ns.fb.friend, P.rdf.ns.rdf.type, P.rdf.ns.owl.SymmetricProperty) c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,1) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,datetime_snapshot, len(fg2["friends"]["name"]), "facebook numeric id, name, locale, sex and agerank", len(fg2["friendships"]["node1"]) ))
Produce a linked data publication tree from GDF files of a Facebook interaction network. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook group => the string id (sid) of the facebook group of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fb/fb.py#L170-L326
[ "def readGDF(filename=\"../data/RenatoFabbri06022014.gdf\"):\n \"\"\"Made to work with my own network. Check file to ease adaptation\"\"\"\n with open(filename,\"r\") as f:\n data=f.read()\n lines=data.split(\"\\n\")\n columns=lines[0].split(\">\")[1].split(\",\")\n column_names=[i.split(\" \")[0] for i in columns]\n data_friends={cn:[] for cn in column_names}\n for line in lines[1:]:\n if not line:\n break\n if \">\" in line:\n columns=line.split(\">\")[1].split(\",\")\n column_names2=[i.split(\" \")[0] for i in columns]\n data_friendships={cn:[] for cn in column_names2}\n continue\n fields=line.split(\",\")\n if \"column_names2\" not in locals():\n for i, field in enumerate(fields):\n if field.isdigit(): field=int(field)\n data_friends[column_names[i]].append(field)\n else:\n for i, field in enumerate(fields):\n if field.isdigit(): field=int(field)\n data_friendships[column_names2[i]].append(field)\n return {\"friendships\":data_friendships,\n \"friends\":data_friends}\n", "def trans(tkey):\n if tkey==\"name\":\n return \"uid\"\n if tkey==\"label\":\n return \"name\"\n return tkey\n" ]
import time, os, pickle, shutil, datetime, re import networkx as x, rdflib as r from splinter import Browser from bs4 import BeautifulSoup import percolation as P c=P.utils.check this_dir = os.path.split(__file__)[0] NS=P.rdf.ns a=NS.rdf.type def triplifyGML(fname="foo.gml",fpath="./fb/",scriptpath=None,uid=None,sid=None,extra_info=None): """Produce a linked data publication tree from a standard GML file. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook user of which fname holds a friendship network => the numeric id (sid) of the facebook user of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.""" # aname=fname.split("/")[-1].split(".")[0] aname=fname.split("/")[-1].split(".")[0] if "RonaldCosta" in fname: aname=fname.split("/")[-1].split(".")[0] name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_="Ronald Scherolt Costa" elif "AntonioAnzoategui" in fname: aname=re.findall(".*/([a-zA-Z]*\d*)",fname)[0] name,year,month,day,hour,minute=re.findall(r".*/([a-zA-Z]*).*_(\d+)_(\d*)_(\d*)_(\d*)_(\d*)_.*",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat()[:-3] name_="Antônio Anzoategui Fabbri" elif re.findall(".*/[a-zA-Z]*(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("[a-zA-Z]*_",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gml",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) aname+="_fb" name+="_fb" c("started snapshot",aname) tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"the {} facebook ego friendship network") tg2=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"RDF metadata for the facebook friendship network of my son") snapshot=P.rdf.IC([tg2],P.rdf.ns.po.FacebookSnapshot, aname,"Snapshot {}".format(aname)) extra_uri=extra_val=[] if extra_info: extra_uri=[NS.po.extraInfo] extra_val=[extra_info] P.rdf.link([tg2],snapshot,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.onlineTranslateXMLFile, P.rdf.ns.po.onlineTranslateTTLFile, P.rdf.ns.po.translateXMLFile, P.rdf.ns.po.translateTTLFile, P.rdf.ns.po.onlineMetaXMLFile, P.rdf.ns.po.onlineMetaTTLFile, P.rdf.ns.po.metaXMLFilename, P.rdf.ns.po.metaTTLFilename, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, P.rdf.ns.fb.uid, P.rdf.ns.fb.sid ]+extra_uri, [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")[-1]), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.rdf".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "{}Translate.rdf".format(aname), "{}Translate.ttl".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.rdf".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "{}Meta.owl".format(aname), "{}Meta.ttl".format(aname), "Netvizz", "The facebook friendship network from {}".format(name_), uid, sid ]+extra_val) #for friend_attr in fg2["friends"]: c((aname,name_,datetime_snapshot)) fg2=x.read_gml(fname) c("read gml") for uid in fg2: c(uid) ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,"{}-{}".format(aname,uid)) if "locale" in fg2.node[uid].keys(): data=[fg2.node[uid][attr] for attr in ("id","label","locale","sex","agerank","wallcount")] uris=[NS.fb.gid, NS.fb.name, NS.fb.locale, NS.fb.sex, NS.fb.agerank,NS.fb.wallcount] else: data=[fg2.node[uid][attr] for attr in ("id","label","sex","agerank","wallcount")] uris=[NS.fb.gid, NS.fb.name, NS.fb.sex, NS.fb.agerank,NS.fb.wallcount] P.rdf.link([tg],ind, None,uris,data,draw=False) P.rdf.link_([tg],ind,None,[NS.po.snapshot],[snapshot],draw=False) #friends_=[fg2["friends"][i] for i in ("name","label","locale","sex","agerank")] #for name,label,locale,sex,agerank in zip(*friends_): # ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) # P.rdf.link([tg],ind,label,[P.rdf.ns.fb.uid,P.rdf.ns.fb.name, # P.rdf.ns.fb.locale,P.rdf.ns.fb.sex, # P.rdf.ns.fb.agerank], # [name,label,locale,sex,agerank]) c("escritos participantes") #friendships_=[fg2["friendships"][i] for i in ("node1","node2")] i=1 for uid1,uid2 in fg2.edges(): flabel="{}-{}-{}".format(aname,uid1,uid2) ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship,flabel) uids=[P.rdf.IC(None,P.rdf.ns.fb.Participant,"{}-{}".format(aname,i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,flabel,[NS.po.snapshot]+[NS.fb.member]*2, [snapshot]+uids,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}/{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") #shutil.copy(this_dir+"/../tests/rdfMyFNetwork2.py",fpath+"scripts/") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,False) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} ({}) collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,aname,datetime_snapshot, fg2.number_of_nodes(), "name, locale (maybe), sex, agerank and wallcount", fg2.number_of_edges())) def triplifyGDFInteraction(fname="foo.gdf",fpath="./fb/",scriptpath=None,uid=None,sid=None,dlink=None): """Produce a linked data publication tree from GDF files of a Facebook interaction network. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook group => the string id (sid) of the facebook group of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.""" #aname=fname.split("/")[-1].split(".")[0]+"_fb" aname=fname.split("/")[-1].split(".")[0] if re.findall("[a-zA-Z]*_[0-9]",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: datetime_snapshot=datetime.datetime(2013,3,15).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",aname)) aname+="_fb" name=aname tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"The facebook interaction network from the {} file".format(fname)) # drop de agraph tg2=P.rdf.makeBasicGraph([["po"],[P.rdf.ns.per]],"Metadata for my facebook ego friendship network RDF files") # drop de agraph ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) foo={"uris":[],"vals":[]} if sid: foo["uris"].append(P.rdf.ns.fb.sid) foo["vals"].append(sid) if uid: foo["uris"].append(P.rdf.ns.fb.uid) foo["vals"].append(uid) if dlink: foo["uris"].append(P.rdf.ns.fb.link) foo["vals"].append(dlink) P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.rdfFile, P.rdf.ns.po.ttlFile, P.rdf.ns.po.discorveryRDFFile, P.rdf.ns.po.discoveryTTLFile, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, ]+foo["uris"], [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "Netvizz", "The facebook friendship network from {}".format(name_), ]+foo["vals"]) #for friend_attr in fg2["friends"]: fg2=readGDF(fname) tkeys=list(fg2["friends"].keys()) def trans(tkey): if tkey=="name": return "uid" if tkey=="label": return "name" return tkey foo={"uris":[],"vals":[]} for tkey in tkeys: if tkey=="groupid": P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.uid,], [fg2["friends"][tkey][0]]) if tkey: foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fg2["friends"][tkey]] print(tkeys) iname=tkeys.index("name") ilabel=tkeys.index("label") icount=0 name_label={} for vals_ in zip(*foo["vals"]): name,label=[foo["vals"][i][icount] for i in (iname,ilabel)] if not label: label="po:noname" vals_=list(vals_) vals_[ilabel]=label name_label[name]=label ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) P.rdf.link([tg],ind,label,foo["uris"], vals_,draw=False) icount+=1 friendships_=[fg2["friendships"][i] for i in ("node1","node2")] c("escritos participantes") i=1 for uid1,uid2 in zip(*friendships_): flabel="{}-{}".format(uid1,uid2) labels=[name_label[uu] for uu in (uid1,uid2)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship, flabel) #flabel,"Friendship "+flabel) ind1=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid1) ind2=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid2) uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+str(i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,"Friendship "+flabel,[P.rdf.ns.fb.member]*2, uids,labels,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 P.rdf.G(tg[0],P.rdf.ns.fb.friend, P.rdf.ns.rdf.type, P.rdf.ns.owl.SymmetricProperty) c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,1) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,datetime_snapshot, len(fg2["friends"]["name"]), "facebook numeric id, name, locale, sex and agerank", len(fg2["friendships"]["node1"]) )) def triplifyGDF(fname="foo.gdf",fpath="./fb/",scriptpath=None,uid=None,sid=None,dlink=None): """Produce a linked data publication tree from a standard GDF file. INPUTS: => the file name (fname, with path) where the gdf file of the friendship network is. => the final path (fpath) for the tree of files to be created. => a path to the script that is calling this function (scriptpath). => the numeric id (uid) of the facebook user of which fname holds a friendship network => the numeric id (sid) of the facebook user of which fname holds a friendship network OUTPUTS: the tree in the directory fpath.""" #aname=fname.split("/")[-1].split(".")[0]+"_fb" aname=fname.split("/")[-1].split(".")[0] if re.findall("[a-zA-Z]*_[0-9]",fname): name,year,month,day,hour,minute=re.findall(".*/([a-zA-Z]*).*(\d\d\d\d)_(\d\d)_(\d\d)_(\d\d)_(\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day,hour,minute)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) elif re.findall("(\d)",fname): name,day,month,year=re.findall(".*/([a-zA-Z]*)(\d\d)(\d\d)(\d\d\d\d).*.gdf",fname)[0] datetime_snapshot=datetime.datetime(*[int(i) for i in (year,month,day)]).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",name)) else: datetime_snapshot=datetime.datetime(2013,3,15).isoformat().split("T")[0] name_=" ".join(re.findall("[A-Z][^A-Z]*",aname)) aname+="_fb" name=aname tg=P.rdf.makeBasicGraph([["po","fb"],[P.rdf.ns.per,P.rdf.ns.fb]],"My facebook ego friendship network") # drop de agraph tg2=P.rdf.makeBasicGraph([["po"],[P.rdf.ns.per]],"Metadata for my facebook ego friendship network RDF files") # drop de agraph ind=P.rdf.IC([tg2],P.rdf.ns.po.Snapshot, aname,"Snapshot {}".format(aname)) foo={"uris":[],"vals":[]} if sid: foo["uris"].append(P.rdf.ns.fb.sid) foo["vals"].append(sid) if uid: foo["uris"].append(P.rdf.ns.fb.uid) foo["vals"].append(uid) if dlink: foo["uris"].append(P.rdf.ns.fb.link) foo["vals"].append(dlink) P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.createdAt, P.rdf.ns.po.triplifiedIn, P.rdf.ns.po.donatedBy, P.rdf.ns.po.availableAt, P.rdf.ns.po.originalFile, P.rdf.ns.po.rdfFile, P.rdf.ns.po.ttlFile, P.rdf.ns.po.discorveryRDFFile, P.rdf.ns.po.discoveryTTLFile, P.rdf.ns.po.acquiredThrough, P.rdf.ns.rdfs.comment, ]+foo["uris"], [datetime_snapshot, datetime.datetime.now(), name, "https://github.com/ttm/{}".format(aname), "https://raw.githubusercontent.com/ttm/{}/master/base/{}".format(aname,fname.split("/")), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Translate.ttl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.owl".format(aname,aname), "https://raw.githubusercontent.com/ttm/{}/master/rdf/{}Meta.ttl".format(aname,aname), "Netvizz", "The facebook friendship network from {}".format(name_), ]+foo["vals"]) #for friend_attr in fg2["friends"]: fg2=readGDF(fname) tkeys=list(fg2["friends"].keys()) def trans(tkey): if tkey=="name": return "uid" if tkey=="label": return "name" return tkey foo={"uris":[],"vals":[]} for tkey in tkeys: if tkey=="groupid": P.rdf.link([tg2],ind,"Snapshot {}".format(aname), [P.rdf.ns.po.uid,], [fg2["friends"][tkey][0]]) if tkey: foo["uris"]+=[eval("P.rdf.ns.fb."+trans(tkey))] foo["vals"]+=[fg2["friends"][tkey]] print(tkeys) iname=tkeys.index("name") ilabel=tkeys.index("label") icount=0 name_label={} for vals_ in zip(*foo["vals"]): name,label=[foo["vals"][i][icount] for i in (iname,ilabel)] if not label: label="po:noname" vals_=list(vals_) vals_[ilabel]=label name_label[name]=label ind=P.rdf.IC([tg],P.rdf.ns.fb.Participant,name,label) P.rdf.link([tg],ind,label,foo["uris"], vals_,draw=False) icount+=1 friendships_=[fg2["friendships"][i] for i in ("node1","node2")] c("escritos participantes") i=1 for uid1,uid2 in zip(*friendships_): flabel="{}-{}".format(uid1,uid2) labels=[name_label[uu] for uu in (uid1,uid2)] ind=P.rdf.IC([tg],P.rdf.ns.fb.Friendship, flabel) #flabel,"Friendship "+flabel) ind1=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid1) ind2=P.rdf.IC(None,P.rdf.ns.fb.Participant,uid2) uids=[r.URIRef(P.rdf.ns.fb.Participant+"#"+str(i)) for i in (uid1,uid2)] P.rdf.link_([tg],ind,"Friendship "+flabel,[P.rdf.ns.fb.member]*2, uids,labels,draw=False) P.rdf.L_([tg],uids[0],P.rdf.ns.fb.friend,uids[1]) if (i%1000)==0: c(i) i+=1 P.rdf.G(tg[0],P.rdf.ns.fb.friend, P.rdf.ns.rdf.type, P.rdf.ns.owl.SymmetricProperty) c("escritas amizades") tg_=[tg[0]+tg2[0],tg[1]] fpath_="{}{}/".format(fpath,aname) P.rdf.writeAll(tg_,aname+"Translate",fpath_,False,1) # copia o script que gera este codigo if not os.path.isdir(fpath_+"scripts"): os.mkdir(fpath_+"scripts") shutil.copy(scriptpath,fpath_+"scripts/") # copia do base data if not os.path.isdir(fpath_+"base"): os.mkdir(fpath_+"base") shutil.copy(fname,fpath_+"base/") P.rdf.writeAll(tg2,aname+"Meta",fpath_,1) # faz um README with open(fpath_+"README","w") as f: f.write("""This repo delivers RDF data from the facebook friendship network of {} collected at {}. It has {} friends with metadata {}; and {} friendships. The linked data is available at rdf/ dir and was generated by the routine in the script/ directory. Original data from Netvizz in data/\n""".format( name_,datetime_snapshot, len(fg2["friends"]["name"]), "facebook numeric id, name, locale, sex and agerank", len(fg2["friendships"]["node1"]) )) def makeRDF(readgdf_dict,fdir="../data/rdf/"): # return rdflib graph from the data rd=readgdf_dict # ns=namespaces=pe.namespaces(["rdf","rdfs","xsd", # basic namespaces # ]) # for friend in range(len(rd["friends"]["name"])): # pass def readGDF(filename="../data/RenatoFabbri06022014.gdf"): """Made to work with my own network. Check file to ease adaptation""" with open(filename,"r") as f: data=f.read() lines=data.split("\n") columns=lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) return {"friendships":data_friendships, "friends":data_friends} #self.makeNetwork() class GDFgraph: """Read GDF graph into networkX""" def __init__(self,filename="../data/RenatoFabbri06022014.gdf"): with open(filename,"r") as f: self.data=f.read() self.lines=self.data.split("\n") columns=self.lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in self.lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) self.data_friendships=data_friendships self.data_friends=data_friends self.n_friends=len(data_friends[column_names[0]]) self.n_friendships=len(data_friendships[column_names2[0]]) self.makeNetwork() def makeNetwork(self): """Makes graph object from .gdf loaded data""" if "weight" in self.data_friendships.keys(): self.G=G=x.DiGraph() else: self.G=G=x.Graph() F=self.data_friends for friendn in range(self.n_friends): if "posts" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], posts=F["posts"][friendn]) elif "agerank" in F.keys(): G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn], agerank=F["agerank"][friendn]) else: G.add_node(F["name"][friendn], label=F["label"][friendn], gender=F["sex"][friendn], locale=F["locale"][friendn]) F=self.data_friendships for friendshipn in range(self.n_friendships): if "weight" in F.keys(): G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn],weight=F["weight"][friendshipn]) else: G.add_edge(F["node1"][friendshipn],F["node2"][friendshipn]) def readFBPost(fpath=""): """Extract information from HTML page with a Facebook post""" html=open(fpath,"rb") soup = BeautifulSoup(html, "lxml") return soup class ScrapyBrowser: """Opens a browser for user to login to facebook. Such browser pulls data as requested by user.""" def __init__(self,user_email=None, user_password=None,basedir="~/.social/"): self._BASE_DIR=basedir.replace("~",os.path.expanduser("~")) if not os.path.isdir(self._BASE_DIR): os.mkdir(self._BASE_DIR) print("Opening *Scrappy* firefox browser. Please wait.") self.browser=browser=Browser(wait_time=2) url="http://facebook.com" browser.visit(url) if (not user_email) or (not user_password): input("\n\n==> Input user and password and login, please.\ and then press <enter>") else: browser.fill("email",user_email) browser.fill("pass",user_password) browser.find_by_value("Log In").click() def getFriends(self,user_id="astronauta.mecanico",write=True): """Returns user_ids (that you have access) of the friends of your friend with user_ids""" while user_id not in self.browser.url: self.browser.visit("http://www.facebook.com/{}/friends".format(user_id), wait_time=3) #self.go("http://www.facebook.com/{}/friends".format(user_id)) T=time.time() while 1: h1=self.browser.evaluate_script("document.body.scrollHeight") self.browser.execute_script("window.scrollTo(0, document.body.scrollHeight);") h2=self.browser.evaluate_script("document.body.scrollHeight") if h1 != h2: T=time.time() elif time.time()-T>10: break #links=self.browser.find_link_by_partial_href("hc_location=friends_tab") links=self.browser.find_by_css(".fcb") friends=[] for link in links: name=link.value user_id_=link.find_by_tag("a")["href"].split("/")[-1].split("?")[0] friends.append((user_id_,name)) tdict={} tdict["name"]=self.browser.find_by_id("fb-timeline-cover-name").value tdict["user_id"]=user_id tdict["friends"]=friends infos=self.browser.find_by_css("._3c_") mutual=0 for info in infos: if info.value=="Mutual Friends": if info.find_by_css("._3d0").value: tdict["n_mutual"]=info.find_by_css("._3d0").value mutual=1 if info.value=="All Friends": tdict["n_friends"]=info.find_by_css("._3d0").value if mutual==0: links=self.browser.find_by_css("._gs6") if "Mutual" in links.value: tdict["n_mutual"]=links.value.split(" ")[0] if write: if not os.path.isdir("{}/fb_ids/".format(self._BASE_DIR)): os.mkdir("{}/fb_ids/".format(self._BASE_DIR)) with open("{}fb_ids/{}.pickle".format(self._BASE_DIR,user_id),"wb") as f: pickle.dump(tdict,f) self.tdict=tdict return tdict
ttm/socialLegacy
social/fb/fb.py
ScrapyBrowser.getFriends
python
def getFriends(self,user_id="astronauta.mecanico",write=True): while user_id not in self.browser.url: self.browser.visit("http://www.facebook.com/{}/friends".format(user_id), wait_time=3) #self.go("http://www.facebook.com/{}/friends".format(user_id)) T=time.time() while 1: h1=self.browser.evaluate_script("document.body.scrollHeight") self.browser.execute_script("window.scrollTo(0, document.body.scrollHeight);") h2=self.browser.evaluate_script("document.body.scrollHeight") if h1 != h2: T=time.time() elif time.time()-T>10: break #links=self.browser.find_link_by_partial_href("hc_location=friends_tab") links=self.browser.find_by_css(".fcb") friends=[] for link in links: name=link.value user_id_=link.find_by_tag("a")["href"].split("/")[-1].split("?")[0] friends.append((user_id_,name)) tdict={} tdict["name"]=self.browser.find_by_id("fb-timeline-cover-name").value tdict["user_id"]=user_id tdict["friends"]=friends infos=self.browser.find_by_css("._3c_") mutual=0 for info in infos: if info.value=="Mutual Friends": if info.find_by_css("._3d0").value: tdict["n_mutual"]=info.find_by_css("._3d0").value mutual=1 if info.value=="All Friends": tdict["n_friends"]=info.find_by_css("._3d0").value if mutual==0: links=self.browser.find_by_css("._gs6") if "Mutual" in links.value: tdict["n_mutual"]=links.value.split(" ")[0] if write: if not os.path.isdir("{}/fb_ids/".format(self._BASE_DIR)): os.mkdir("{}/fb_ids/".format(self._BASE_DIR)) with open("{}fb_ids/{}.pickle".format(self._BASE_DIR,user_id),"wb") as f: pickle.dump(tdict,f) self.tdict=tdict return tdict
Returns user_ids (that you have access) of the friends of your friend with user_ids
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fb/fb.py#L611-L655
null
class ScrapyBrowser: """Opens a browser for user to login to facebook. Such browser pulls data as requested by user.""" def __init__(self,user_email=None, user_password=None,basedir="~/.social/"): self._BASE_DIR=basedir.replace("~",os.path.expanduser("~")) if not os.path.isdir(self._BASE_DIR): os.mkdir(self._BASE_DIR) print("Opening *Scrappy* firefox browser. Please wait.") self.browser=browser=Browser(wait_time=2) url="http://facebook.com" browser.visit(url) if (not user_email) or (not user_password): input("\n\n==> Input user and password and login, please.\ and then press <enter>") else: browser.fill("email",user_email) browser.fill("pass",user_password) browser.find_by_value("Log In").click()
ttm/socialLegacy
social/fb/read.py
readGDF
python
def readGDF(filename="../data/RenatoFabbri06022014.gdf"): with open(filename,"r") as f: data=f.read() lines=data.split("\n") columns=lines[0].split(">")[1].split(",") column_names=[i.split(" ")[0] for i in columns] data_friends={cn:[] for cn in column_names} for line in lines[1:]: if not line: break if ">" in line: columns=line.split(">")[1].split(",") column_names2=[i.split(" ")[0] for i in columns] data_friendships={cn:[] for cn in column_names2} continue fields=line.split(",") if "column_names2" not in locals(): for i, field in enumerate(fields): if column_names[i] in ("name","groupid"): pass elif field.isdigit(): field=int(field) data_friends[column_names[i]].append(field) else: for i, field in enumerate(fields): if column_names2[i]=="name": pass elif field.isdigit(): field=int(field) data_friendships[column_names2[i]].append(field) return {"relations":data_friendships, "individuals":data_friends}
Made to work with gdf files from my own network and friends and groups
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fb/read.py#L164-L192
null
import networkx as x, percolation as P, re c=P.utils.check def readGML2(filename="../data/RenatoFabbri06022014.gml"): with open(filename,"r") as f: data=f.read() lines=data.split("\n") nodes=[] # list of dicts, each a node edges=[] # list of tuples state="receive" for line in lines: if state=="receive": if "node" in line: state="node" nodes.append({}) if "edge" in line: state="edge" edges.append({}) elif "]" in line: state="receive" elif "[" in line: pass elif state=="node": var,val=re.findall(r"(.*?) (.*)",line.strip())[0] if var=="id": var="name" val="user_{}".format(val) elif '"' in val: val=val.replace('"',"") else: val=int(val) nodes[-1][var]=val elif state=="edge": var,val=line.strip().split() edges[-1][var]=val else: c("SPURIOUS LINE: "+line) keys=set([j for i in nodes for j in i.keys()]) nodes_={} for key in keys: if key == "id": nodes_["name"]=[None]*len(nodes) i=0 for node in nodes: nodes_["name"][i]="user_{}".format(node[key]) i+=1 else: nodes_[key]=[None]*len(nodes) i=0 for node in nodes: if key in node.keys(): nodes_[key][i]=node[key] i+=1 c("para carregar as amizades") edges_={"node1":[None]*len(edges), "node2":[None]*len(edges)} i=0 for edge in edges: u1="user_{}".format(edge["source"]) u2="user_{}".format(edge["target"]) edges_["node1"][i]=u1 edges_["node2"][i]=u2 i+=1 return {"relations": edges_, "individuals": nodes_} gg=x.read_gml(filename) nodes=gg.nodes(data=True) nodes_=[i[1] for i in nodes] nodes__={} nkeys=[] c("para carregar os individuos") for node in nodes_: nkeys+=list(node.keys()) nkeys=set(nkeys) for key in nkeys: if key == "id": nodes__["name"]=[None]*len(nodes_) i=0 for node in nodes_: nodes__["name"][i]="user_{}".format(node[key]) i+=1 else: nodes__[key]=[None]*len(nodes_) i=0 for node in nodes_: if key in node.keys(): nodes__[key][i]=node[key] i+=1 c("para carregar as amizades") edges=gg.edges(data=True) edges_={"node1":[None]*len(edges), "node2":[None]*len(edges)} i=0 for edge in edges: u1="user_{}".format(edge[0]) u2="user_{}".format(edge[1]) edges_["node1"][i]=u1 edges_["node2"][i]=u2 i+=1 # if edges[0][2]: # edges_=[i[2] for i in edges] # edges__={} # ekeys=edges_[0].keys() # for key in ekeys: # edges__[key]=[] # for edge in edges_: # edges__[key]+=[edge[key]] return {"relations": edges_, "individuals": nodes__} def readGML(filename="../data/RenatoFabbri06022014.gml"): gg=x.read_gml(filename) nodes=gg.nodes(data=True) nodes_=[i[1] for i in nodes] nodes__={} nkeys=[] c("para carregar os individuos") for node in nodes_: nkeys+=list(node.keys()) nkeys=set(nkeys) for key in nkeys: if key == "id": nodes__["name"]=[None]*len(nodes_) i=0 for node in nodes_: nodes__["name"][i]="user_{}".format(node[key]) i+=1 else: nodes__[key]=[None]*len(nodes_) i=0 for node in nodes_: if key in node.keys(): nodes__[key][i]=node[key] i+=1 c("para carregar as amizades") edges=gg.edges(data=True) edges_={"node1":[None]*len(edges), "node2":[None]*len(edges)} i=0 for edge in edges: u1="user_{}".format(edge[0]) u2="user_{}".format(edge[1]) edges_["node1"][i]=u1 edges_["node2"][i]=u2 i+=1 # if edges[0][2]: # edges_=[i[2] for i in edges] # edges__={} # ekeys=edges_[0].keys() # for key in ekeys: # edges__[key]=[] # for edge in edges_: # edges__[key]+=[edge[key]] return {"relations": edges_, "individuals": nodes__} return gg
ttm/socialLegacy
social/tw.py
Twitter.searchTag
python
def searchTag(self,HTAG="#python"): self.t = Twython(app_key =self.app_key , app_secret =self.app_secret , oauth_token =self.oauth_token , oauth_token_secret =self.oauth_token_secret) search =self.t.search(q=HTAG,count=100,result_type="recent") ss=search[:] search = self.t.search(q=HTAG,count=150,max_id=ss[-1]['id']-1,result_type="recent") #search = t.search(q=HTAG,count=150,since_id=ss[-1]['id'],result_type="recent") while seach: ss+=search[:] search = self.t.search(q=HTAG,count=150,max_id=ss[-1]['id']-1,result_type="recent") self.ss=ss
Set Twitter search or stream criteria for the selection of tweets
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/tw.py#L293-L307
null
class Twitter: """Simplified Twitter interface for Stability observance # function to set authentication: __init__() # function to set hashtag and other tweets selection criteria: searchTag() # function to search tweets: searchTag() # function to stream tweets: void """ TWITTER_API_KEY = tw.tak TWITTER_API_KEY_SECRET = tw.taks TWITTER_ACCESS_TOKEN = tw.tat TWITTER_ACCESS_TOKEN_SECRET = tw.tats def __init__(self,app_key= None, app_secret= None, oauth_token= None, oauth_token_secret=None,): """Start twitter seach and stream interface""" if not app_key: self.app_key= self.TWITTER_API_KEY self.app_secret= self.TWITTER_API_KEY_SECRET self.oauth_token= self.TWITTER_ACCESS_TOKEN self.oauth_token_secret=self.TWITTER_ACCESS_TOKEN_SECRET else: self.app_key= app_key self.app_secret= app_secret self.oauth_token= oauth_token self.oauth_token_secret=oauth_token_secret def streamTag(self,HTAG="#python",aname=None): if not aname: aname=HTAG[1:]+"_tw" stream=MyStreamer(self.app_key , self.app_secret , self.oauth_token , self.oauth_token_secret) stream.putName(aname) self.stream=stream stream.statuses.filter(track=HTAG) def finishStream(self): self.stream.D.close()
ttm/socialLegacy
social/fsong.py
FSong.makePartitions
python
def makePartitions(self): class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g")
Make partitions with gmane help.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L29-L41
null
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makeImages(self): """Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both """ # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): """Plot graph with nodes (iterable) into filename """ if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato") def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeSong(self): """Render abstract animation """ self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation() def makeVisualSong(self): """Return a sequence of images and durations. """ self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4)) # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAudibleSong(self): """Use mass to render wav soundtrack. """ sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav") def makeAnimation(self): """Use pymovie to render (visual+audio)+text overlays. """ aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
ttm/socialLegacy
social/fsong.py
FSong.makeImages
python
def makeImages(self): # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png")
Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L42-L63
[ "def plotGraph(self,mode=\"plain\",nodes=None,filename=\"tgraph.png\"):\n \"\"\"Plot graph with nodes (iterable) into filename\n \"\"\"\n if nodes==None:\n nodes=self.nodes\n else:\n nodes=[i for i in self.nodes if i in nodes]\n for node in self.nodes:\n n_=self.A.get_node(node)\n if mode==\"plain\":\n nmode=1\n else:\n nmode=-1\n pos=\"{},{}\".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)])\n n_.attr[\"pos\"]=pos\n n_.attr[\"pin\"]=True\n color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]])\n n_.attr['fillcolor']= color\n n_.attr['fixedsize']=True\n n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5))\n n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5))\n n_.attr[\"label\"]=\"\"\n if node not in nodes:\n n_.attr[\"style\"]=\"invis\"\n else:\n n_.attr[\"style\"]=\"filled\"\n for e in self.edges:\n e.attr['penwidth']=3.4\n e.attr[\"arrowsize\"]=1.5\n e.attr[\"arrowhead\"]=\"lteeoldiamond\"\n e.attr[\"style\"]=\"\"\n if sum([i in nodes for i in (e[0],e[1])])==2:\n e.attr[\"style\"]=\"\"\n else:\n e.attr[\"style\"]=\"invis\"\n tname=\"{}{}\".format(self.basedir,filename)\n print(tname)\n self.A.draw(tname,prog=\"neato\")\n", "def setAgraph(self):\n self.A=x.to_agraph(self.network)\n self.A.graph_attr[\"viewport\"]=\"500,500,.03\"\n self.edges=self.A.edges()\n self.nodes=self.A.nodes()\n self.cm=p.cm.Reds(range(2**10)) # color table\n self.clustering=x.clustering(self.network)\n", "def makeLayout(self):\n ri=4\n rf=100\n nturns=3\n ii=n.linspace(0,nturns*2*n.pi,self.nm.N)\n rr=n.linspace(ri,rf,self.nm.N)\n self.xi=(rr*n.cos(ii))\n self.yi=(rr*n.sin(ii))\n" ]
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makePartitions(self): """Make partitions with gmane help. """ class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): """Plot graph with nodes (iterable) into filename """ if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato") def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeSong(self): """Render abstract animation """ self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation() def makeVisualSong(self): """Return a sequence of images and durations. """ self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4)) # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAudibleSong(self): """Use mass to render wav soundtrack. """ sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav") def makeAnimation(self): """Use pymovie to render (visual+audio)+text overlays. """ aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
ttm/socialLegacy
social/fsong.py
FSong.plotGraph
python
def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato")
Plot graph with nodes (iterable) into filename
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L72-L109
null
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makePartitions(self): """Make partitions with gmane help. """ class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g") def makeImages(self): """Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both """ # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeSong(self): """Render abstract animation """ self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation() def makeVisualSong(self): """Return a sequence of images and durations. """ self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4)) # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAudibleSong(self): """Use mass to render wav soundtrack. """ sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav") def makeAnimation(self): """Use pymovie to render (visual+audio)+text overlays. """ aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
ttm/socialLegacy
social/fsong.py
FSong.makeSong
python
def makeSong(self): self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation()
Render abstract animation
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L127-L133
[ "def makeVisualSong(self):\n \"\"\"Return a sequence of images and durations.\n \"\"\"\n self.files=os.listdir(self.basedir)\n self.stairs=[i for i in self.files if (\"stair\" in i) and (\"R\" in i)]\n self.sectors=[i for i in self.files if \"sector\" in i]\n self.stairs.sort()\n self.sectors.sort()\n filenames=[self.basedir+i for i in self.sectors[:4]]\n self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5])\n self.iS1=mpy.ImageSequenceClip(\n [self.basedir+self.sectors[2],\n self.basedir+self.sectors[3],\n self.basedir+self.sectors[2],\n self.basedir+self.sectors[3],\n self.basedir+self.sectors[2],\n self.basedir+self.sectors[3],\n self.basedir+self.sectors[2],\n self.basedir+self.sectors[3]],\n durations=[0.25]*8)\n self.iS2=mpy.ImageSequenceClip(\n [self.basedir+self.sectors[2],\n self.basedir+self.sectors[3],\n self.basedir+self.sectors[2],\n self.basedir+self.sectors[3],\n self.basedir+self.sectors[0]],\n durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível\n\n self.iS3=mpy.ImageSequenceClip(\n [self.basedir+\"BLANK.png\",\n self.basedir+self.sectors[0],\n self.basedir+self.sectors[1],\n self.basedir+self.sectors[1],\n self.basedir+self.sectors[1],\n self.basedir+self.sectors[0],\n self.basedir+self.sectors[0]],\n durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8]\n\n self.iS4=mpy.ImageSequenceClip(\n [self.basedir+self.sectors[2], # 1\n self.basedir+self.sectors[3], # .5\n self.basedir+self.sectors[5], # .5\n self.basedir+self.sectors[2], # .75\n self.basedir+self.sectors[0], #.25\n self.basedir+self.sectors[2], # 1\n self.basedir+self.sectors[0], # 2 8\n self.basedir+self.sectors[3], # 2 7\n self.basedir+self.sectors[0], # 2 -1\n self.basedir+\"BLANK.png\",# 2\n ],\n durations=[1,0.5,0.5,.75,\n .25,1., 2.,2.,2.,2.]) # [0,7,11,0]\n\n self.iS=mpy.concatenate_videoclips((\n self.iS0,self.iS1,self.iS2,self.iS3,self.iS4))\n", "def makeAudibleSong(self):\n \"\"\"Use mass to render wav soundtrack.\n \"\"\"\n sound0=n.hstack((sy.render(220,d=1.5),\n sy.render(220*(2**(7/12)),d=2.5),\n sy.render(220*(2**(-5/12)),d=.5),\n sy.render(220*(2**(0/12)),d=1.5),\n ))\n sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25),\n sy.render(220*(2**(7/12)),d=.25),\n sy.render(220*(2**(0/12)),d=.25),\n sy.render(220*(2**(7/12)),d=.25),\n sy.render(220*(2**(0/12)),d=.25),\n sy.render(220*(2**(7/12)),d=.25),\n sy.render(220*(2**(0/12)),d=.25),\n sy.render(220*(2**(7/12)),d=.25),\n ))\n sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75),\n sy.render(220*(2**(0/12)),d=.25),\n sy.render(220*(2**(7/12)),d=.75),\n sy.render(220*(2**(0/12)),d=.25),\n sy.render(220*(2**(-1/12)),d=2.0),\n ))\n sound3=n.hstack((n.zeros(44100),\n sy.render(220*(2**(-1/12)),d=.5),\n sy.render(220*(2**(8/12)),d=2.),\n sy.render(220*(2**(8/12)),d=.25),\n sy.render(220*(2**(8/12)),d=.25),\n sy.render(220*(2**(-1/12)),d=1.75),\n sy.render(220*(2**(-1/12)),d=.25),\n ))\n sound4=n.hstack((\n sy.render(220*(2**(0/12)),d=1.),\n sy.render(220*(2**(7/12)),d=.5),\n sy.render(220*(2**(11/12)),d=.5),\n sy.render(220*(2**(12/12)),d=.75),\n sy.render(220*(2**(11/12)),d=.25),\n sy.render(220*(2**(12/12)),d=1.),\n sy.render(220*(2**(8/12)),d=2.),\n sy.render(220*(2**(7/12)),d=2.),\n sy.render(220*(2**(-1/12)),d=2.),\n n.zeros(2*44100)\n ))\n\n sound=n.hstack((sound0,sound1,sound2,sound3,sound4))\n UT.write(sound,\"sound.wav\")\n", "def makeAnimation(self):\n \"\"\"Use pymovie to render (visual+audio)+text overlays.\n \"\"\"\n aclip=mpy.AudioFileClip(\"sound.wav\")\n self.iS=self.iS.set_audio(aclip)\n self.iS.write_videofile(\"mixedVideo.webm\",15,audio=True)\n print(\"wrote \"+\"mixedVideo.webm\")\n" ]
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makePartitions(self): """Make partitions with gmane help. """ class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g") def makeImages(self): """Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both """ # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): """Plot graph with nodes (iterable) into filename """ if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato") def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeVisualSong(self): """Return a sequence of images and durations. """ self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4)) # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAudibleSong(self): """Use mass to render wav soundtrack. """ sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav") def makeAnimation(self): """Use pymovie to render (visual+audio)+text overlays. """ aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
ttm/socialLegacy
social/fsong.py
FSong.makeVisualSong
python
def makeVisualSong(self): self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4))
Return a sequence of images and durations.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L134-L188
null
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makePartitions(self): """Make partitions with gmane help. """ class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g") def makeImages(self): """Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both """ # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): """Plot graph with nodes (iterable) into filename """ if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato") def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeSong(self): """Render abstract animation """ self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation() # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAudibleSong(self): """Use mass to render wav soundtrack. """ sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav") def makeAnimation(self): """Use pymovie to render (visual+audio)+text overlays. """ aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
ttm/socialLegacy
social/fsong.py
FSong.makeAudibleSong
python
def makeAudibleSong(self): sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav")
Use mass to render wav soundtrack.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L194-L239
null
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makePartitions(self): """Make partitions with gmane help. """ class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g") def makeImages(self): """Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both """ # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): """Plot graph with nodes (iterable) into filename """ if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato") def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeSong(self): """Render abstract animation """ self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation() def makeVisualSong(self): """Return a sequence of images and durations. """ self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4)) # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAnimation(self): """Use pymovie to render (visual+audio)+text overlays. """ aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
ttm/socialLegacy
social/fsong.py
FSong.makeAnimation
python
def makeAnimation(self): aclip=mpy.AudioFileClip("sound.wav") self.iS=self.iS.set_audio(aclip) self.iS.write_videofile("mixedVideo.webm",15,audio=True) print("wrote "+"mixedVideo.webm")
Use pymovie to render (visual+audio)+text overlays.
train
https://github.com/ttm/socialLegacy/blob/c0930cfe6e84392729449bf7c92569e1556fd109/social/fsong.py#L240-L246
null
class FSong: """Create song from undirected (friendship) network """ def __init__(self, network,basedir="fsong/",clean=False,render_images=False,render_images2=False,make_video=False): os.system("mkdir {}".format(basedir)) if clean: os.system("rm {}*".format(basedir)) self.basedir=basedir self.network=network self.makePartitions() if render_images: self.makeImages() self.make_video=make_video self.makeSong() if render_images2: self.makeImages2() self.makeSong2() def makeSong2(self): pass def makePartitions(self): """Make partitions with gmane help. """ class NetworkMeasures: pass self.nm=nm=NetworkMeasures() nm.degrees=self.network.degree() nm.nodes_= sorted(self.network.nodes(), key=lambda x : nm.degrees[x]) nm.degrees_=[nm.degrees[i] for i in nm.nodes_] nm.edges= self.network.edges(data=True) nm.E=self.network.number_of_edges() nm.N=self.network.number_of_nodes() self.np=g.NetworkPartitioning(nm,10,metric="g") def makeImages(self): """Make spiral images in sectors and steps. Plain, reversed, sectorialized, negative sectorialized outline, outline reversed, lonely only nodes, only edges, both """ # make layout self.makeLayout() self.setAgraph() # make function that accepts a mode, a sector # and nodes and edges True and False self.plotGraph() self.plotGraph("reversed",filename="tgraphR.png") agents=n.concatenate(self.np.sectorialized_agents__) for i, sector in enumerate(self.np.sectorialized_agents__): self.plotGraph("plain", sector,"sector{:02}.png".format(i)) self.plotGraph("reversed",sector,"sector{:02}R.png".format(i)) self.plotGraph("plain", n.setdiff1d(agents,sector),"sector{:02}N.png".format(i)) self.plotGraph("reversed",n.setdiff1d(agents,sector),"sector{:02}RN.png".format(i)) self.plotGraph("plain", [],"BLANK.png") def makeImages2(self): for i, node in enumerate(self.nm.nodes_): self.plotGraph("plain", [node],"lonely{:09}.png".format(i)) self.plotGraph("reversed",[node],"lonely{:09}R.png".format(i)) self.plotGraph("plain", self.nm.nodes_[:i],"stair{:09}.png".format(i)) self.plotGraph("reversed",self.nm.nodes_[:i],"stair{:09}R.png".format(i)) # plotar novamente usando somente vertices e depois somente arestas def plotGraph(self,mode="plain",nodes=None,filename="tgraph.png"): """Plot graph with nodes (iterable) into filename """ if nodes==None: nodes=self.nodes else: nodes=[i for i in self.nodes if i in nodes] for node in self.nodes: n_=self.A.get_node(node) if mode=="plain": nmode=1 else: nmode=-1 pos="{},{}".format(self.xi[::nmode][self.nm.nodes_.index(node)],self.yi[::nmode][self.nm.nodes_.index(node)]) n_.attr["pos"]=pos n_.attr["pin"]=True color='#%02x%02x%02x' % tuple([255*i for i in self.cm[int(self.clustering[n_]*255)][:-1]]) n_.attr['fillcolor']= color n_.attr['fixedsize']=True n_.attr['width']= abs(.1*(self.nm.degrees[n_]+ .5)) n_.attr['height']= abs(.1*(self.nm.degrees[n_]+.5)) n_.attr["label"]="" if node not in nodes: n_.attr["style"]="invis" else: n_.attr["style"]="filled" for e in self.edges: e.attr['penwidth']=3.4 e.attr["arrowsize"]=1.5 e.attr["arrowhead"]="lteeoldiamond" e.attr["style"]="" if sum([i in nodes for i in (e[0],e[1])])==2: e.attr["style"]="" else: e.attr["style"]="invis" tname="{}{}".format(self.basedir,filename) print(tname) self.A.draw(tname,prog="neato") def setAgraph(self): self.A=x.to_agraph(self.network) self.A.graph_attr["viewport"]="500,500,.03" self.edges=self.A.edges() self.nodes=self.A.nodes() self.cm=p.cm.Reds(range(2**10)) # color table self.clustering=x.clustering(self.network) def makeLayout(self): ri=4 rf=100 nturns=3 ii=n.linspace(0,nturns*2*n.pi,self.nm.N) rr=n.linspace(ri,rf,self.nm.N) self.xi=(rr*n.cos(ii)) self.yi=(rr*n.sin(ii)) def makeSong(self): """Render abstract animation """ self.makeVisualSong() self.makeAudibleSong() if self.make_video: self.makeAnimation() def makeVisualSong(self): """Return a sequence of images and durations. """ self.files=os.listdir(self.basedir) self.stairs=[i for i in self.files if ("stair" in i) and ("R" in i)] self.sectors=[i for i in self.files if "sector" in i] self.stairs.sort() self.sectors.sort() filenames=[self.basedir+i for i in self.sectors[:4]] self.iS0=mpy.ImageSequenceClip(filenames,durations=[1.5,2.5,.5,1.5]) self.iS1=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3]], durations=[0.25]*8) self.iS2=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[2], self.basedir+self.sectors[3], self.basedir+self.sectors[0]], durations=[0.75,0.25,0.75,0.25,2.]) # cai para sensível self.iS3=mpy.ImageSequenceClip( [self.basedir+"BLANK.png", self.basedir+self.sectors[0], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[1], self.basedir+self.sectors[0], self.basedir+self.sectors[0]], durations=[1,0.5,2.,.25,.25,1.75, 0.25]) # [-1,8] self.iS4=mpy.ImageSequenceClip( [self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[3], # .5 self.basedir+self.sectors[5], # .5 self.basedir+self.sectors[2], # .75 self.basedir+self.sectors[0], #.25 self.basedir+self.sectors[2], # 1 self.basedir+self.sectors[0], # 2 8 self.basedir+self.sectors[3], # 2 7 self.basedir+self.sectors[0], # 2 -1 self.basedir+"BLANK.png",# 2 ], durations=[1,0.5,0.5,.75, .25,1., 2.,2.,2.,2.]) # [0,7,11,0] self.iS=mpy.concatenate_videoclips(( self.iS0,self.iS1,self.iS2,self.iS3,self.iS4)) # Clip with three first images3 # each sector a sound # sweep from periphery to center # all, all inverted # sectors with inversions def makeAudibleSong(self): """Use mass to render wav soundtrack. """ sound0=n.hstack((sy.render(220,d=1.5), sy.render(220*(2**(7/12)),d=2.5), sy.render(220*(2**(-5/12)),d=.5), sy.render(220*(2**(0/12)),d=1.5), )) sound1=n.hstack((sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.25), )) sound2=n.hstack((sy.render(220*(2**(0/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(7/12)),d=.75), sy.render(220*(2**(0/12)),d=.25), sy.render(220*(2**(-1/12)),d=2.0), )) sound3=n.hstack((n.zeros(44100), sy.render(220*(2**(-1/12)),d=.5), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(8/12)),d=.25), sy.render(220*(2**(-1/12)),d=1.75), sy.render(220*(2**(-1/12)),d=.25), )) sound4=n.hstack(( sy.render(220*(2**(0/12)),d=1.), sy.render(220*(2**(7/12)),d=.5), sy.render(220*(2**(11/12)),d=.5), sy.render(220*(2**(12/12)),d=.75), sy.render(220*(2**(11/12)),d=.25), sy.render(220*(2**(12/12)),d=1.), sy.render(220*(2**(8/12)),d=2.), sy.render(220*(2**(7/12)),d=2.), sy.render(220*(2**(-1/12)),d=2.), n.zeros(2*44100) )) sound=n.hstack((sound0,sound1,sound2,sound3,sound4)) UT.write(sound,"sound.wav")
harlowja/failure
failure/finders.py
match_modules
python
def match_modules(allowed_modules): cleaned_allowed_modules = [ utils.mod_to_mod_name(tmp_mod) for tmp_mod in allowed_modules ] cleaned_split_allowed_modules = [ tmp_mod.split(".") for tmp_mod in cleaned_allowed_modules ] cleaned_allowed_modules = [] del cleaned_allowed_modules def matcher(cause): cause_cls = None cause_type_name = cause.exception_type_names[0] # Rip off the class name (usually at the end). cause_type_name_pieces = cause_type_name.split(".") cause_type_name_mod_pieces = cause_type_name_pieces[0:-1] # Do any modules provided match the provided causes module? mod_match = any( utils.array_prefix_matches(mod_pieces, cause_type_name_mod_pieces) for mod_pieces in cleaned_split_allowed_modules) if mod_match: cause_cls = importutils.import_class(cause_type_name) cause_cls = ensure_base_exception(cause_type_name, cause_cls) return cause_cls return matcher
Creates a matcher that matches a list/set/tuple of allowed modules.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/finders.py#L44-L73
null
# -*- coding: utf-8 -*- # Copyright (C) 2016 GoDaddy Inc. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); you may # not use this file except in compliance with the License. You may obtain # a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the # License for the specific language governing permissions and limitations # under the License. from __future__ import absolute_import import itertools from oslo_utils import importutils from oslo_utils import reflection from failure import _utils as utils class InvalidTypeError(TypeError): pass def ensure_base_exception(cause_type_name, cls): # Ensure source class is correct (ie that it has the right # root that **all** python exceptions must have); if not right then # it will be discarded. if not issubclass(cls, BaseException): raise InvalidTypeError( "Cause with type '%s' was regenerated as a non-exception" " base class '%s'" % (cause_type_name, reflection.get_class_name(cls))) else: return cls def match_classes(allowed_classes): """Creates a matcher that matches a list/tuple of allowed classes.""" cleaned_allowed_classes = [ utils.cls_to_cls_name(tmp_cls) for tmp_cls in allowed_classes ] def matcher(cause): cause_cls = None cause_type_name = cause.exception_type_names[0] try: cause_cls_idx = cleaned_allowed_classes.index(cause_type_name) except ValueError: pass else: cause_cls = allowed_classes[cause_cls_idx] if not isinstance(cause_cls, type): cause_cls = importutils.import_class(cause_cls) cause_cls = ensure_base_exception(cause_type_name, cause_cls) return cause_cls return matcher def combine_or(matcher, *more_matchers): """Combines more than one matcher together (first that matches wins).""" def matcher(cause): for sub_matcher in itertools.chain([matcher], more_matchers): cause_cls = sub_matcher(cause) if cause_cls is not None: return cause_cls return None return matcher
harlowja/failure
failure/finders.py
match_classes
python
def match_classes(allowed_classes): cleaned_allowed_classes = [ utils.cls_to_cls_name(tmp_cls) for tmp_cls in allowed_classes ] def matcher(cause): cause_cls = None cause_type_name = cause.exception_type_names[0] try: cause_cls_idx = cleaned_allowed_classes.index(cause_type_name) except ValueError: pass else: cause_cls = allowed_classes[cause_cls_idx] if not isinstance(cause_cls, type): cause_cls = importutils.import_class(cause_cls) cause_cls = ensure_base_exception(cause_type_name, cause_cls) return cause_cls return matcher
Creates a matcher that matches a list/tuple of allowed classes.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/finders.py#L76-L97
null
# -*- coding: utf-8 -*- # Copyright (C) 2016 GoDaddy Inc. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); you may # not use this file except in compliance with the License. You may obtain # a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the # License for the specific language governing permissions and limitations # under the License. from __future__ import absolute_import import itertools from oslo_utils import importutils from oslo_utils import reflection from failure import _utils as utils class InvalidTypeError(TypeError): pass def ensure_base_exception(cause_type_name, cls): # Ensure source class is correct (ie that it has the right # root that **all** python exceptions must have); if not right then # it will be discarded. if not issubclass(cls, BaseException): raise InvalidTypeError( "Cause with type '%s' was regenerated as a non-exception" " base class '%s'" % (cause_type_name, reflection.get_class_name(cls))) else: return cls def match_modules(allowed_modules): """Creates a matcher that matches a list/set/tuple of allowed modules.""" cleaned_allowed_modules = [ utils.mod_to_mod_name(tmp_mod) for tmp_mod in allowed_modules ] cleaned_split_allowed_modules = [ tmp_mod.split(".") for tmp_mod in cleaned_allowed_modules ] cleaned_allowed_modules = [] del cleaned_allowed_modules def matcher(cause): cause_cls = None cause_type_name = cause.exception_type_names[0] # Rip off the class name (usually at the end). cause_type_name_pieces = cause_type_name.split(".") cause_type_name_mod_pieces = cause_type_name_pieces[0:-1] # Do any modules provided match the provided causes module? mod_match = any( utils.array_prefix_matches(mod_pieces, cause_type_name_mod_pieces) for mod_pieces in cleaned_split_allowed_modules) if mod_match: cause_cls = importutils.import_class(cause_type_name) cause_cls = ensure_base_exception(cause_type_name, cause_cls) return cause_cls return matcher def combine_or(matcher, *more_matchers): """Combines more than one matcher together (first that matches wins).""" def matcher(cause): for sub_matcher in itertools.chain([matcher], more_matchers): cause_cls = sub_matcher(cause) if cause_cls is not None: return cause_cls return None return matcher
harlowja/failure
failure/finders.py
combine_or
python
def combine_or(matcher, *more_matchers): def matcher(cause): for sub_matcher in itertools.chain([matcher], more_matchers): cause_cls = sub_matcher(cause) if cause_cls is not None: return cause_cls return None return matcher
Combines more than one matcher together (first that matches wins).
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/finders.py#L100-L110
null
# -*- coding: utf-8 -*- # Copyright (C) 2016 GoDaddy Inc. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); you may # not use this file except in compliance with the License. You may obtain # a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the # License for the specific language governing permissions and limitations # under the License. from __future__ import absolute_import import itertools from oslo_utils import importutils from oslo_utils import reflection from failure import _utils as utils class InvalidTypeError(TypeError): pass def ensure_base_exception(cause_type_name, cls): # Ensure source class is correct (ie that it has the right # root that **all** python exceptions must have); if not right then # it will be discarded. if not issubclass(cls, BaseException): raise InvalidTypeError( "Cause with type '%s' was regenerated as a non-exception" " base class '%s'" % (cause_type_name, reflection.get_class_name(cls))) else: return cls def match_modules(allowed_modules): """Creates a matcher that matches a list/set/tuple of allowed modules.""" cleaned_allowed_modules = [ utils.mod_to_mod_name(tmp_mod) for tmp_mod in allowed_modules ] cleaned_split_allowed_modules = [ tmp_mod.split(".") for tmp_mod in cleaned_allowed_modules ] cleaned_allowed_modules = [] del cleaned_allowed_modules def matcher(cause): cause_cls = None cause_type_name = cause.exception_type_names[0] # Rip off the class name (usually at the end). cause_type_name_pieces = cause_type_name.split(".") cause_type_name_mod_pieces = cause_type_name_pieces[0:-1] # Do any modules provided match the provided causes module? mod_match = any( utils.array_prefix_matches(mod_pieces, cause_type_name_mod_pieces) for mod_pieces in cleaned_split_allowed_modules) if mod_match: cause_cls = importutils.import_class(cause_type_name) cause_cls = ensure_base_exception(cause_type_name, cause_cls) return cause_cls return matcher def match_classes(allowed_classes): """Creates a matcher that matches a list/tuple of allowed classes.""" cleaned_allowed_classes = [ utils.cls_to_cls_name(tmp_cls) for tmp_cls in allowed_classes ] def matcher(cause): cause_cls = None cause_type_name = cause.exception_type_names[0] try: cause_cls_idx = cleaned_allowed_classes.index(cause_type_name) except ValueError: pass else: cause_cls = allowed_classes[cause_cls_idx] if not isinstance(cause_cls, type): cause_cls = importutils.import_class(cause_cls) cause_cls = ensure_base_exception(cause_type_name, cause_cls) return cause_cls return matcher
harlowja/failure
failure/failure.py
WrappedFailure.check
python
def check(self, *exc_classes): if not exc_classes: return None for cause in self: result = cause.check(*exc_classes) if result is not None: return result return None
Check if any of exception classes caused the failure/s. :param exc_classes: exception types/exception type names to search for. If any of the contained failures were caused by an exception of a given type, the corresponding argument that matched is returned. If not then ``None`` is returned.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L77-L93
null
class WrappedFailure(utils.StrMixin, Exception): """Wraps one or several failure objects. When exception/s cannot be re-raised (for example, because the value and traceback are lost in serialization) or there are several exceptions active at the same time (due to more than one thread raising exceptions), we will wrap the corresponding failure objects into this exception class and *may* reraise this exception type to allow users to handle the contained failures/causes as they see fit... See the failure class documentation for a more comprehensive set of reasons why this object *may* be reraised instead of the original exception. :param causes: the :py:class:`~failure.Failure` objects that caused this this exception to be raised. """ def __init__(self, causes): super(WrappedFailure, self).__init__() self._causes = [] for cause in causes: if cause.check(type(self)) and cause.exception is not None: # NOTE(imelnikov): flatten wrapped failures. self._causes.extend(cause.exception) else: self._causes.append(cause) def __iter__(self): """Iterate over failures that caused the exception.""" return iter(self._causes) def __len__(self): """Return number of wrapped failures.""" return len(self._causes) def __bytes__(self): buf = six.BytesIO() buf.write(b'WrappedFailure: [') causes_gen = (six.binary_type(cause) for cause in self._causes) buf.write(b", ".join(causes_gen)) buf.write(b']') return buf.getvalue() def __unicode__(self): buf = six.StringIO() buf.write(u'WrappedFailure: [') causes_gen = (six.text_type(cause) for cause in self._causes) buf.write(u", ".join(causes_gen)) buf.write(u']') return buf.getvalue()
harlowja/failure
failure/failure.py
Failure.from_exc_info
python
def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb
Creates a failure object from a ``sys.exc_info()`` tuple.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L241-L291
[ "def extract_roots(exc_type):\n return to_tuple(\n reflection.get_all_class_names(exc_type, up_to=BaseException,\n truncate_builtins=False))\n", "def _extract_cause(cls, exc_val):\n \"\"\"Helper routine to extract nested cause (if any).\"\"\"\n # See: https://www.python.org/dev/peps/pep-3134/ for why/what\n # these are...\n #\n # '__cause__' attribute for explicitly chained exceptions\n # '__context__' attribute for implicitly chained exceptions\n # '__traceback__' attribute for the traceback\n #\n # See: https://www.python.org/dev/peps/pep-0415/ for why/what\n # the '__suppress_context__' is/means/implies...\n nested_exc_vals = []\n seen = [exc_val]\n while True:\n suppress_context = getattr(\n exc_val, '__suppress_context__', False)\n if suppress_context:\n attr_lookups = ['__cause__']\n else:\n attr_lookups = ['__cause__', '__context__']\n nested_exc_val = None\n for attr_name in attr_lookups:\n attr_val = getattr(exc_val, attr_name, None)\n if attr_val is None:\n continue\n nested_exc_val = attr_val\n if nested_exc_val is None or nested_exc_val in seen:\n break\n seen.append(nested_exc_val)\n nested_exc_vals.append(nested_exc_val)\n exc_val = nested_exc_val\n last_cause = None\n for exc_val in reversed(nested_exc_vals):\n f = cls.from_exception(exc_val, cause=last_cause,\n find_cause=False)\n last_cause = f\n return last_cause\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.from_exception
python
def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause)
Creates a failure object from a exception instance.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L294-L304
[ "def from_exc_info(cls, exc_info=None,\n retain_exc_info=True,\n cause=None, find_cause=True):\n \"\"\"Creates a failure object from a ``sys.exc_info()`` tuple.\"\"\"\n if exc_info is None:\n exc_info = sys.exc_info()\n if not any(exc_info):\n raise NoActiveException(\"No exception currently\"\n \" being handled\")\n # This should always be the (type, value, traceback) tuple,\n # either from a prior sys.exc_info() call or from some other\n # creation...\n if len(exc_info) != 3:\n raise ValueError(\"Provided 'exc_info' must contain three\"\n \" elements\")\n exc_type, exc_val, exc_tb = exc_info\n try:\n if exc_type is None or exc_val is None:\n raise ValueError(\"Invalid exception tuple (exception\"\n \" type and exception value must\"\n \" be provided)\")\n exc_args = tuple(getattr(exc_val, 'args', []))\n exc_kwargs = dict(getattr(exc_val, 'kwargs', {}))\n exc_type_names = utils.extract_roots(exc_type)\n if not exc_type_names:\n exc_type_name = reflection.get_class_name(\n exc_val, truncate_builtins=False)\n # This should only be possible if the exception provided\n # was not really an exception...\n raise TypeError(\"Invalid exception type '%s' (not an\"\n \" exception)\" % (exc_type_name))\n exception_str = utils.exception_message(exc_val)\n if hasattr(exc_val, '__traceback_str__'):\n traceback_str = exc_val.__traceback_str__\n else:\n if exc_tb is not None:\n traceback_str = '\\n'.join(\n traceback.format_exception(*exc_info))\n else:\n traceback_str = ''\n if not retain_exc_info:\n exc_info = None\n if find_cause and cause is None:\n cause = cls._extract_cause(exc_val)\n return cls(exc_info=exc_info, exc_args=exc_args,\n exc_kwargs=exc_kwargs, exception_str=exception_str,\n exc_type_names=exc_type_names, cause=cause,\n traceback_str=traceback_str,\n generated_on=sys.version_info[0:2])\n finally:\n del exc_type, exc_val, exc_tb\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.validate
python
def validate(cls, data): try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause)
Validate input data matches expected failure ``dict`` format.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L307-L338
null
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.matches
python
def matches(self, other): if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other
Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L351-L362
[ "def _matches(self, other):\n if self is other:\n return True\n return (self.exception_type_names == other.exception_type_names and\n self.exception_args == other.exception_args and\n self.exception_kwargs == other.exception_kwargs and\n self.exception_str == other.exception_str and\n self.traceback_str == other.traceback_str and\n self.cause == other.cause and\n self.generated_on == other.generated_on)\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.reraise_if_any
python
def reraise_if_any(failures, cause_cls_finder=None): if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures)
Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L436-L451
null
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.reraise
python
def reraise(self, cause_cls_finder=None): if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None)
Re-raise captured exception (possibly trying to recreate).
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L453-L482
[ "def iter_causes(self):\n \"\"\"Iterate over all causes.\"\"\"\n curr = self._cause\n while curr is not None:\n yield curr\n curr = curr._cause\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.check
python
def check(self, *exc_classes): for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None
Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L484-L496
[ "def cls_to_cls_name(cls):\n if isinstance(cls, type):\n cls_name = reflection.get_class_name(cls, truncate_builtins=False)\n else:\n cls_name = str(cls)\n return cls_name\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.pformat
python
def pformat(self, traceback=False): buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue()
Pretty formats the failure object into a string.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L513-L532
null
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.iter_causes
python
def iter_causes(self): curr = self._cause while curr is not None: yield curr curr = curr._cause
Iterate over all causes.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L534-L539
null
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure._extract_cause
python
def _extract_cause(cls, exc_val): # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause
Helper routine to extract nested cause (if any).
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L584-L620
null
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.from_dict
python
def from_dict(cls, data): data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data)
Converts this from a dictionary to a object.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L623-L629
[ "def from_dict(cls, data):\n \"\"\"Converts this from a dictionary to a object.\"\"\"\n data = dict(data)\n cause = data.get('cause')\n if cause is not None:\n data['cause'] = cls.from_dict(cause)\n return cls(**data)\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.to_dict
python
def to_dict(self, include_args=True, include_kwargs=True): data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data
Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output.
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L631-L650
null
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def copy(self, deep=False): """Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy). """ cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
harlowja/failure
failure/failure.py
Failure.copy
python
def copy(self, deep=False): cause = self._cause if cause is not None: cause = cause.copy(deep=deep) exc_info = utils.copy_exc_info(self.exc_info, deep=deep) exc_args = self.exception_args exc_kwargs = self.exception_kwargs if deep: exc_args = copy.deepcopy(exc_args) exc_kwargs = copy.deepcopy(exc_kwargs) else: exc_args = tuple(exc_args) exc_kwargs = exc_kwargs.copy() # These are just simple int/strings, so deep copy doesn't really # matter/apply here (as they are immutable anyway). exc_type_names = tuple(self._exc_type_names) generated_on = self._generated_on if generated_on: generated_on = tuple(generated_on) # NOTE(harlowja): use `self.__class__` here so that we can work # with subclasses (assuming anyone makes one). return self.__class__(exc_info=exc_info, exception_str=self.exception_str, traceback_str=self.traceback_str, exc_args=exc_args, exc_kwargs=exc_kwargs, exc_type_names=exc_type_names, cause=cause, generated_on=generated_on)
Copies this object (shallow or deep). :param deep: boolean indicating whether to do a deep copy (or a shallow copy).
train
https://github.com/harlowja/failure/blob/9ea9a46ebb26c6d7da2553c80e36892f3997bd6f/failure/failure.py#L652-L684
[ "def copy_exc_info(exc_info, deep=False):\n if exc_info is None:\n return None\n exc_type, exc_value, exc_tb = exc_info\n # NOTE(imelnikov): there is no need to copy the exception type, and\n # a shallow copy of the value is fine and we can't copy the traceback since\n # it contains reference to the internal stack frames...\n if deep:\n return (exc_type, copy.deepcopy(exc_value), exc_tb)\n else:\n return (exc_type, copy.copy(exc_value), exc_tb)\n" ]
class Failure(utils.StrMixin): """An immutable object that represents failure. Failure objects encapsulate exception information so that they can be re-used later to re-raise, inspect, examine, log, print, serialize, deserialize... For those who are curious, here are a few reasons why the original exception itself *may* not be reraised and instead a reraised wrapped failure exception object will be instead. These explanations are *only* applicable when a failure object is serialized and deserialized (when it is retained inside the python process that the exception was created in the the original exception can be reraised correctly without issue). * Traceback objects are not serializable/recreatable, since they contain references to stack frames at the location where the exception was raised. When a failure object is serialized and sent across a channel and recreated it is *not* possible to restore the original traceback and originating stack frames. * The original exception *type* can not *always* be guaranteed to be found, certain nodes can run code that is not accessible/available when the failure is being deserialized. Even if it was possible to use pickle safely (which it is not) it would not *always* be possible to find the originating exception or associated code in this situation. * The original exception *type* can not be guaranteed to be constructed in a *correct* manner. At the time of failure object creation the exception has already been created and the failure object can not assume it has knowledge (or the ability) to recreate the original type of the captured exception (this is especially hard if the original exception was created via a complex process via some custom exception ``__init__`` method). * The original exception *type* can not *always* be guaranteed to be constructed and/or imported in a *safe* manner. Importing *foreign* exception types dynamically can be problematic when not done correctly and in a safe manner; since failure objects can capture *any* exception it would be *unsafe* to try to import those exception types namespaces and modules on the receiver side dynamically (this would create similar issues as the ``pickle`` module has). TODO(harlowja): use parts of http://bugs.python.org/issue17911 and the backport at https://pypi.python.org/pypi/traceback2/ to (hopefully) simplify the methods and contents of this object... """ BASE_EXCEPTIONS = { # py2.x old/legacy names... 2: ('exceptions.BaseException', 'exceptions.Exception'), # py3.x new names... 3: ('builtins.BaseException', 'builtins.Exception'), } """ Root exceptions of all other python exceptions (as a string). See: https://docs.python.org/2/library/exceptions.html """ #: Expected failure schema (in json schema format). SCHEMA = { "$ref": "#/definitions/cause", "definitions": { "cause": { "type": "object", 'properties': { 'exc_args': { "type": "array", "minItems": 0, }, 'exc_kwargs': { "type": "object", "additionalProperties": True, }, 'exception_str': { "type": "string", }, 'traceback_str': { "type": "string", }, 'exc_type_names': { "type": "array", "items": { "type": "string", }, "minItems": 1, }, 'generated_on': { "type": "array", "items": { "type": "number", }, "minItems": 1, }, 'cause': { "type": "object", "$ref": "#/definitions/cause", }, }, "required": [ "exception_str", 'traceback_str', 'exc_type_names', 'generated_on', ], "additionalProperties": True, }, }, } def __init__(self, exc_info=None, exc_args=None, exc_kwargs=None, exception_str='', exc_type_names=None, cause=None, traceback_str='', generated_on=None): exc_type_names = utils.to_tuple(exc_type_names) if not exc_type_names: raise ValueError("Invalid exception type (no type names" " provided)") self._exc_type_names = exc_type_names self._exc_info = utils.to_tuple(exc_info, on_none=None) self._exc_args = utils.to_tuple(exc_args) if exc_kwargs: self._exc_kwargs = dict(exc_kwargs) else: self._exc_kwargs = {} self._exception_str = exception_str self._cause = cause self._traceback_str = traceback_str self._generated_on = utils.to_tuple(generated_on, on_none=None) @classmethod def from_exc_info(cls, exc_info=None, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a ``sys.exc_info()`` tuple.""" if exc_info is None: exc_info = sys.exc_info() if not any(exc_info): raise NoActiveException("No exception currently" " being handled") # This should always be the (type, value, traceback) tuple, # either from a prior sys.exc_info() call or from some other # creation... if len(exc_info) != 3: raise ValueError("Provided 'exc_info' must contain three" " elements") exc_type, exc_val, exc_tb = exc_info try: if exc_type is None or exc_val is None: raise ValueError("Invalid exception tuple (exception" " type and exception value must" " be provided)") exc_args = tuple(getattr(exc_val, 'args', [])) exc_kwargs = dict(getattr(exc_val, 'kwargs', {})) exc_type_names = utils.extract_roots(exc_type) if not exc_type_names: exc_type_name = reflection.get_class_name( exc_val, truncate_builtins=False) # This should only be possible if the exception provided # was not really an exception... raise TypeError("Invalid exception type '%s' (not an" " exception)" % (exc_type_name)) exception_str = utils.exception_message(exc_val) if hasattr(exc_val, '__traceback_str__'): traceback_str = exc_val.__traceback_str__ else: if exc_tb is not None: traceback_str = '\n'.join( traceback.format_exception(*exc_info)) else: traceback_str = '' if not retain_exc_info: exc_info = None if find_cause and cause is None: cause = cls._extract_cause(exc_val) return cls(exc_info=exc_info, exc_args=exc_args, exc_kwargs=exc_kwargs, exception_str=exception_str, exc_type_names=exc_type_names, cause=cause, traceback_str=traceback_str, generated_on=sys.version_info[0:2]) finally: del exc_type, exc_val, exc_tb @classmethod def from_exception(cls, exception, retain_exc_info=True, cause=None, find_cause=True): """Creates a failure object from a exception instance.""" exc_info = ( type(exception), exception, getattr(exception, '__traceback__', None) ) return cls.from_exc_info(exc_info=exc_info, retain_exc_info=retain_exc_info, cause=cause, find_cause=find_cause) @classmethod def validate(cls, data): """Validate input data matches expected failure ``dict`` format.""" try: jsonschema.validate( data, cls.SCHEMA, # See: https://github.com/Julian/jsonschema/issues/148 types={'array': (list, tuple)}) except jsonschema.ValidationError as e: raise InvalidFormat("Failure data not of the" " expected format: %s" % (e.message)) else: # Ensure that all 'exc_type_names' originate from one of # base exceptions, because those are the root exceptions that # python mandates/provides and anything else is invalid... causes = collections.deque([data]) while causes: cause = causes.popleft() try: generated_on = cause['generated_on'] ok_bases = cls.BASE_EXCEPTIONS[generated_on[0]] except (KeyError, IndexError): ok_bases = [] root_exc_type = cause['exc_type_names'][-1] if root_exc_type not in ok_bases: raise InvalidFormat( "Failure data 'exc_type_names' must" " have an initial exception type that is one" " of %s types: '%s' is not one of those" " types" % (ok_bases, root_exc_type)) sub_cause = cause.get('cause') if sub_cause is not None: causes.append(sub_cause) def _matches(self, other): if self is other: return True return (self.exception_type_names == other.exception_type_names and self.exception_args == other.exception_args and self.exception_kwargs == other.exception_kwargs and self.exception_str == other.exception_str and self.traceback_str == other.traceback_str and self.cause == other.cause and self.generated_on == other.generated_on) def matches(self, other): """Checks if another object is equivalent to this object. :returns: checks if another object is equivalent to this object :rtype: boolean """ if not isinstance(other, Failure): return False if self.exc_info is None or other.exc_info is None: return self._matches(other) else: return self == other def __eq__(self, other): if not isinstance(other, Failure): return NotImplemented return (self._matches(other) and utils.are_equal_exc_info_tuples(self.exc_info, other.exc_info)) def __ne__(self, other): return not (self == other) # NOTE(imelnikov): obj.__hash__() should return same values for equal # objects, so we should redefine __hash__. Failure equality semantics # is a bit complicated, so for now we just mark Failure objects as # unhashable. See python docs on object.__hash__ for more info: # http://docs.python.org/2/reference/datamodel.html#object.__hash__ __hash__ = None @property def exception(self): """Exception value, or ``None`` if exception value is not present. Exception value *may* be lost during serialization. """ if self._exc_info: return self._exc_info[1] else: return None @property def generated_on(self): """Python major & minor version tuple this failure was generated on. May be ``None`` if not provided during creation (or after if lost). """ return self._generated_on @property def exception_str(self): """String representation of exception.""" return self._exception_str @property def exception_args(self): """Tuple of arguments given to the exception constructor.""" return self._exc_args @property def exception_kwargs(self): """Dict of keyword arguments given to the exception constructor.""" return self._exc_kwargs @property def exception_type_names(self): """Tuple of current exception type **names** (in MRO order).""" return self._exc_type_names @property def exc_info(self): """Exception info tuple or ``None``. See: https://docs.python.org/2/library/sys.html#sys.exc_info for what the contents of this tuple are (if none, then no contents can be examined). """ return self._exc_info @property def traceback_str(self): """Exception traceback as string.""" return self._traceback_str @staticmethod def reraise_if_any(failures, cause_cls_finder=None): """Re-raise exceptions if argument is not empty. If argument is empty list/tuple/iterator, this method returns None. If argument is converted into a list with a single ``Failure`` object in it, that failure is reraised. Else, a :class:`~.WrappedFailure` exception is raised with the failure list as causes. """ if not isinstance(failures, (list, tuple)): # Convert generators/other into a list... failures = list(failures) if len(failures) == 1: failures[0].reraise(cause_cls_finder=cause_cls_finder) elif len(failures) > 1: raise WrappedFailure(failures) def reraise(self, cause_cls_finder=None): """Re-raise captured exception (possibly trying to recreate).""" if self._exc_info: six.reraise(*self._exc_info) else: # Attempt to regenerate the full chain (and then raise # from the root); without a traceback, oh well... root = None parent = None for cause in itertools.chain([self], self.iter_causes()): if cause_cls_finder is not None: cause_cls = cause_cls_finder(cause) else: cause_cls = None if cause_cls is None: # Unable to find where this cause came from, give up... raise WrappedFailure([self]) exc = cause_cls( *cause.exception_args, **cause.exception_kwargs) # Saving this will ensure that if this same exception # is serialized again that we will extract the traceback # from it directly (thus proxying along the original # traceback as much as we can). exc.__traceback_str__ = cause.traceback_str if root is None: root = exc if parent is not None: parent.__cause__ = exc parent = exc six.reraise(type(root), root, tb=None) def check(self, *exc_classes): """Check if any of ``exc_classes`` caused the failure. Arguments of this method can be exception types or type names (strings **fully qualified**). If captured exception is an instance of exception of given type, the corresponding argument is returned, otherwise ``None`` is returned. """ for cls in exc_classes: cls_name = utils.cls_to_cls_name(cls) if cls_name in self._exc_type_names: return cls return None @property def cause(self): """Nested failure *cause* of this failure. This property is typically only useful on 3.x or newer versions of python as older versions do **not** have associated causes. Refer to :pep:`3134` and :pep:`409` and :pep:`415` for what this is examining to find failure causes. """ return self._cause def __unicode__(self): return self.pformat() def pformat(self, traceback=False): """Pretty formats the failure object into a string.""" buf = six.StringIO() if not self._exc_type_names: buf.write('Failure: %s' % (self._exception_str)) else: buf.write('Failure: %s: %s' % (self._exc_type_names[0], self._exception_str)) if traceback: if self._traceback_str is not None: traceback_str = self._traceback_str.rstrip() else: traceback_str = None if traceback_str: buf.write(os.linesep) buf.write(traceback_str) else: buf.write(os.linesep) buf.write('Traceback not available.') return buf.getvalue() def iter_causes(self): """Iterate over all causes.""" curr = self._cause while curr is not None: yield curr curr = curr._cause def __getstate__(self): dct = self.to_dict() if self._exc_info: # Avoids 'TypeError: can't pickle traceback objects' dct['exc_info'] = self._exc_info[0:2] return dct def __setstate__(self, dct): self._exception_str = dct['exception_str'] if 'exc_args' in dct: self._exc_args = tuple(dct['exc_args']) else: # Guess we got an older version somehow, before this # was added, so at that point just set to an empty tuple... self._exc_args = () if 'exc_kwargs' in dct: self._exc_kwargs = dict(dct['exc_kwargs']) else: self._exc_kwargs = {} self._traceback_str = dct['traceback_str'] self._exc_type_names = dct['exc_type_names'] self._generated_on = dct['generated_on'] if 'exc_info' in dct: # Tracebacks can't be serialized/deserialized, but since we # provide a traceback string (and more) this should be # acceptable... # # TODO(harlowja): in the future we could do something like # what the twisted people have done, see for example # twisted-13.0.0/twisted/python/failure.py#L89 for how they # created a fake traceback object... exc_info = list(dct['exc_info']) while len(exc_info) < 3: exc_info.append(None) self._exc_info = tuple(exc_info[0:3]) else: self._exc_info = None cause = dct.get('cause') if cause is not None: cause = self.from_dict(cause) self._cause = cause @classmethod def _extract_cause(cls, exc_val): """Helper routine to extract nested cause (if any).""" # See: https://www.python.org/dev/peps/pep-3134/ for why/what # these are... # # '__cause__' attribute for explicitly chained exceptions # '__context__' attribute for implicitly chained exceptions # '__traceback__' attribute for the traceback # # See: https://www.python.org/dev/peps/pep-0415/ for why/what # the '__suppress_context__' is/means/implies... nested_exc_vals = [] seen = [exc_val] while True: suppress_context = getattr( exc_val, '__suppress_context__', False) if suppress_context: attr_lookups = ['__cause__'] else: attr_lookups = ['__cause__', '__context__'] nested_exc_val = None for attr_name in attr_lookups: attr_val = getattr(exc_val, attr_name, None) if attr_val is None: continue nested_exc_val = attr_val if nested_exc_val is None or nested_exc_val in seen: break seen.append(nested_exc_val) nested_exc_vals.append(nested_exc_val) exc_val = nested_exc_val last_cause = None for exc_val in reversed(nested_exc_vals): f = cls.from_exception(exc_val, cause=last_cause, find_cause=False) last_cause = f return last_cause @classmethod def from_dict(cls, data): """Converts this from a dictionary to a object.""" data = dict(data) cause = data.get('cause') if cause is not None: data['cause'] = cls.from_dict(cause) return cls(**data) def to_dict(self, include_args=True, include_kwargs=True): """Converts this object to a dictionary. :param include_args: boolean indicating whether to include the exception args in the output. :param include_kwargs: boolean indicating whether to include the exception kwargs in the output. """ data = { 'exception_str': self.exception_str, 'traceback_str': self.traceback_str, 'exc_type_names': self.exception_type_names, 'exc_args': self.exception_args if include_args else tuple(), 'exc_kwargs': self.exception_kwargs if include_kwargs else {}, 'generated_on': self.generated_on, } if self._cause is not None: data['cause'] = self._cause.to_dict(include_args=include_args, include_kwargs=include_kwargs) return data
ambitioninc/django-entity-event
entity_event/context_loader.py
get_context_hints_per_source
python
def get_context_hints_per_source(context_renderers): # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source
Given a list of context renderers, return a dictionary of context hints per source.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L21-L42
null
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
get_querysets_for_context_hints
python
def get_querysets_for_context_hints(context_hints_per_source): model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets
Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... }
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L45-L74
null
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
dict_find
python
def dict_find(d, which_key): # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result
Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L77-L94
[ "def dict_find(d, which_key):\n \"\"\"\n Finds key values in a nested dictionary. Returns a tuple of the dictionary in which\n the key was found along with the value\n \"\"\"\n # If the starting point is a list, iterate recursively over all values\n if isinstance(d, (list, tuple)):\n for i in d:\n for result in dict_find(i, which_key):\n yield result\n\n # Else, iterate over all key values of the dictionary\n elif isinstance(d, dict):\n for k, v in d.items():\n if k == which_key:\n yield d, v\n for result in dict_find(v, which_key):\n yield result\n" ]
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
get_model_ids_to_fetch
python
def get_model_ids_to_fetch(events, context_hints_per_source): number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch
Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... }
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L97-L119
[ "def dict_find(d, which_key):\n \"\"\"\n Finds key values in a nested dictionary. Returns a tuple of the dictionary in which\n the key was found along with the value\n \"\"\"\n # If the starting point is a list, iterate recursively over all values\n if isinstance(d, (list, tuple)):\n for i in d:\n for result in dict_find(i, which_key):\n yield result\n\n # Else, iterate over all key values of the dictionary\n elif isinstance(d, dict):\n for k, v in d.items():\n if k == which_key:\n yield d, v\n for result in dict_find(v, which_key):\n yield result\n" ]
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
fetch_model_data
python
def fetch_model_data(model_querysets, model_ids_to_fetch): return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() }
Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... }
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L122-L138
null
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
load_fetched_objects_into_contexts
python
def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value)
Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L141-L155
[ "def dict_find(d, which_key):\n \"\"\"\n Finds key values in a nested dictionary. Returns a tuple of the dictionary in which\n the key was found along with the value\n \"\"\"\n # If the starting point is a list, iterate recursively over all values\n if isinstance(d, (list, tuple)):\n for i in d:\n for result in dict_find(i, which_key):\n yield result\n\n # Else, iterate over all key values of the dictionary\n elif isinstance(d, dict):\n for k, v in d.items():\n if k == which_key:\n yield d, v\n for result in dict_find(v, which_key):\n yield result\n" ]
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
load_renderers_into_events
python
def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr
Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L158-L191
null
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style def load_contexts_and_renderers(events, mediums): """ Given a list of events and mediums, load the context model data into the contexts of the events. """ sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
ambitioninc/django-entity-event
entity_event/context_loader.py
load_contexts_and_renderers
python
def load_contexts_and_renderers(events, mediums): sources = {event.source for event in events} rendering_styles = {medium.rendering_style for medium in mediums if medium.rendering_style} # Fetch the default rendering style and add it to the set of rendering styles default_rendering_style = get_default_rendering_style() if default_rendering_style: rendering_styles.add(default_rendering_style) context_renderers = ContextRenderer.objects.filter( Q(source__in=sources, rendering_style__in=rendering_styles) | Q(source_group_id__in=[s.group_id for s in sources], rendering_style__in=rendering_styles)).select_related( 'source', 'rendering_style').prefetch_related('source_group__source_set') context_hints_per_source = get_context_hints_per_source(context_renderers) model_querysets = get_querysets_for_context_hints(context_hints_per_source) model_ids_to_fetch = get_model_ids_to_fetch(events, context_hints_per_source) model_data = fetch_model_data(model_querysets, model_ids_to_fetch) load_fetched_objects_into_contexts(events, model_data, context_hints_per_source) load_renderers_into_events(events, mediums, context_renderers, default_rendering_style) return events
Given a list of events and mediums, load the context model data into the contexts of the events.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_loader.py#L202-L226
[ "def get_context_hints_per_source(context_renderers):\n \"\"\"\n Given a list of context renderers, return a dictionary of context hints per source.\n \"\"\"\n # Merge the context render hints for each source as there can be multiple context hints for\n # sources depending on the render target. Merging them together involves combining select\n # and prefetch related hints for each context renderer\n context_hints_per_source = defaultdict(lambda: defaultdict(lambda: {\n 'app_name': None,\n 'model_name': None,\n 'select_related': set(),\n 'prefetch_related': set(),\n }))\n for cr in context_renderers:\n for key, hints in cr.context_hints.items() if cr.context_hints else []:\n for source in cr.get_sources():\n context_hints_per_source[source][key]['app_name'] = hints['app_name']\n context_hints_per_source[source][key]['model_name'] = hints['model_name']\n context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', []))\n context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', []))\n\n return context_hints_per_source\n", "def get_querysets_for_context_hints(context_hints_per_source):\n \"\"\"\n Given a list of context hint dictionaries, return a dictionary\n of querysets for efficient context loading. The return value\n is structured as follows:\n\n {\n model: queryset,\n ...\n }\n \"\"\"\n model_select_relateds = defaultdict(set)\n model_prefetch_relateds = defaultdict(set)\n model_querysets = {}\n for context_hints in context_hints_per_source.values():\n for hints in context_hints.values():\n model = get_model(hints['app_name'], hints['model_name'])\n model_querysets[model] = model.objects\n model_select_relateds[model].update(hints.get('select_related', []))\n model_prefetch_relateds[model].update(hints.get('prefetch_related', []))\n\n # Attach select and prefetch related parameters to the querysets if needed\n for model, queryset in model_querysets.items():\n if model_select_relateds[model]:\n queryset = queryset.select_related(*model_select_relateds[model])\n if model_prefetch_relateds[model]:\n queryset = queryset.prefetch_related(*model_prefetch_relateds[model])\n model_querysets[model] = queryset\n\n return model_querysets\n", "def get_model_ids_to_fetch(events, context_hints_per_source):\n \"\"\"\n Obtains the ids of all models that need to be fetched. Returns a dictionary of models that\n point to sets of ids that need to be fetched. Return output is as follows:\n\n {\n model: [id1, id2, ...],\n ...\n }\n \"\"\"\n number_types = (complex, float) + six.integer_types\n model_ids_to_fetch = defaultdict(set)\n\n for event in events:\n context_hints = context_hints_per_source.get(event.source, {})\n for context_key, hints in context_hints.items():\n for d, value in dict_find(event.context, context_key):\n values = value if isinstance(value, list) else [value]\n model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update(\n v for v in values if isinstance(v, number_types)\n )\n\n return model_ids_to_fetch\n", "def fetch_model_data(model_querysets, model_ids_to_fetch):\n \"\"\"\n Given a dictionary of models to querysets and model IDs to models, fetch the IDs\n for every model and return the objects in the following structure.\n\n {\n model: {\n id: obj,\n ...\n },\n ...\n }\n \"\"\"\n return {\n model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch))\n for model, ids_to_fetch in model_ids_to_fetch.items()\n }\n", "def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source):\n \"\"\"\n Given the fetched model data and the context hints for each source, go through each\n event and populate the contexts with the loaded information.\n \"\"\"\n for event in events:\n context_hints = context_hints_per_source.get(event.source, {})\n for context_key, hints in context_hints.items():\n model = get_model(hints['app_name'], hints['model_name'])\n for d, value in dict_find(event.context, context_key):\n if isinstance(value, list):\n for i, model_id in enumerate(d[context_key]):\n d[context_key][i] = model_data[model].get(model_id)\n else:\n d[context_key] = model_data[model].get(value)\n", "def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style):\n \"\"\"\n Given the events and the context renderers, load the renderers into the event objects\n so that they may be able to call the 'render' method later on.\n \"\"\"\n # Make a mapping of source groups and rendering styles to context renderers. Do\n # the same for sources and rendering styles to context renderers\n source_group_style_to_renderer = {\n (cr.source_group_id, cr.rendering_style_id): cr\n for cr in context_renderers if cr.source_group_id\n }\n source_style_to_renderer = {\n (cr.source_id, cr.rendering_style_id): cr\n for cr in context_renderers if cr.source_id\n }\n\n for e in events:\n for m in mediums:\n # Try the following when loading a context renderer for a medium in an event.\n # 1. Try to look up the renderer based on the source group and medium rendering style\n # 2. If step 1 doesn't work, look up based on the source and medium rendering style\n # 3. If step 2 doesn't work, look up based on the source group and default rendering style\n # 4. if step 3 doesn't work, look up based on the source and default rendering style\n # If none of those steps work, this event will not be able to be rendered for the mediun\n cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id))\n if not cr:\n cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id))\n if not cr and default_rendering_style:\n cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id))\n if not cr and default_rendering_style:\n cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id))\n\n if cr:\n e._context_renderers[m] = cr\n", "def get_default_rendering_style():\n default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None)\n if default_rendering_style:\n default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style)\n\n return default_rendering_style\n" ]
""" A module for loading contexts using context hints. """ from collections import defaultdict import six from django.conf import settings from django.db.models import Q try: # Django 1.9 from django.apps import apps get_model = apps.get_model except ImportError: # pragma: no cover # Django < 1.9 from django.db.models import get_model from manager_utils import id_dict from entity_event.models import ContextRenderer def get_context_hints_per_source(context_renderers): """ Given a list of context renderers, return a dictionary of context hints per source. """ # Merge the context render hints for each source as there can be multiple context hints for # sources depending on the render target. Merging them together involves combining select # and prefetch related hints for each context renderer context_hints_per_source = defaultdict(lambda: defaultdict(lambda: { 'app_name': None, 'model_name': None, 'select_related': set(), 'prefetch_related': set(), })) for cr in context_renderers: for key, hints in cr.context_hints.items() if cr.context_hints else []: for source in cr.get_sources(): context_hints_per_source[source][key]['app_name'] = hints['app_name'] context_hints_per_source[source][key]['model_name'] = hints['model_name'] context_hints_per_source[source][key]['select_related'].update(hints.get('select_related', [])) context_hints_per_source[source][key]['prefetch_related'].update(hints.get('prefetch_related', [])) return context_hints_per_source def get_querysets_for_context_hints(context_hints_per_source): """ Given a list of context hint dictionaries, return a dictionary of querysets for efficient context loading. The return value is structured as follows: { model: queryset, ... } """ model_select_relateds = defaultdict(set) model_prefetch_relateds = defaultdict(set) model_querysets = {} for context_hints in context_hints_per_source.values(): for hints in context_hints.values(): model = get_model(hints['app_name'], hints['model_name']) model_querysets[model] = model.objects model_select_relateds[model].update(hints.get('select_related', [])) model_prefetch_relateds[model].update(hints.get('prefetch_related', [])) # Attach select and prefetch related parameters to the querysets if needed for model, queryset in model_querysets.items(): if model_select_relateds[model]: queryset = queryset.select_related(*model_select_relateds[model]) if model_prefetch_relateds[model]: queryset = queryset.prefetch_related(*model_prefetch_relateds[model]) model_querysets[model] = queryset return model_querysets def dict_find(d, which_key): """ Finds key values in a nested dictionary. Returns a tuple of the dictionary in which the key was found along with the value """ # If the starting point is a list, iterate recursively over all values if isinstance(d, (list, tuple)): for i in d: for result in dict_find(i, which_key): yield result # Else, iterate over all key values of the dictionary elif isinstance(d, dict): for k, v in d.items(): if k == which_key: yield d, v for result in dict_find(v, which_key): yield result def get_model_ids_to_fetch(events, context_hints_per_source): """ Obtains the ids of all models that need to be fetched. Returns a dictionary of models that point to sets of ids that need to be fetched. Return output is as follows: { model: [id1, id2, ...], ... } """ number_types = (complex, float) + six.integer_types model_ids_to_fetch = defaultdict(set) for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): for d, value in dict_find(event.context, context_key): values = value if isinstance(value, list) else [value] model_ids_to_fetch[get_model(hints['app_name'], hints['model_name'])].update( v for v in values if isinstance(v, number_types) ) return model_ids_to_fetch def fetch_model_data(model_querysets, model_ids_to_fetch): """ Given a dictionary of models to querysets and model IDs to models, fetch the IDs for every model and return the objects in the following structure. { model: { id: obj, ... }, ... } """ return { model: id_dict(model_querysets[model].filter(id__in=ids_to_fetch)) for model, ids_to_fetch in model_ids_to_fetch.items() } def load_fetched_objects_into_contexts(events, model_data, context_hints_per_source): """ Given the fetched model data and the context hints for each source, go through each event and populate the contexts with the loaded information. """ for event in events: context_hints = context_hints_per_source.get(event.source, {}) for context_key, hints in context_hints.items(): model = get_model(hints['app_name'], hints['model_name']) for d, value in dict_find(event.context, context_key): if isinstance(value, list): for i, model_id in enumerate(d[context_key]): d[context_key][i] = model_data[model].get(model_id) else: d[context_key] = model_data[model].get(value) def load_renderers_into_events(events, mediums, context_renderers, default_rendering_style): """ Given the events and the context renderers, load the renderers into the event objects so that they may be able to call the 'render' method later on. """ # Make a mapping of source groups and rendering styles to context renderers. Do # the same for sources and rendering styles to context renderers source_group_style_to_renderer = { (cr.source_group_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_group_id } source_style_to_renderer = { (cr.source_id, cr.rendering_style_id): cr for cr in context_renderers if cr.source_id } for e in events: for m in mediums: # Try the following when loading a context renderer for a medium in an event. # 1. Try to look up the renderer based on the source group and medium rendering style # 2. If step 1 doesn't work, look up based on the source and medium rendering style # 3. If step 2 doesn't work, look up based on the source group and default rendering style # 4. if step 3 doesn't work, look up based on the source and default rendering style # If none of those steps work, this event will not be able to be rendered for the mediun cr = source_group_style_to_renderer.get((e.source.group_id, m.rendering_style_id)) if not cr: cr = source_style_to_renderer.get((e.source_id, m.rendering_style_id)) if not cr and default_rendering_style: cr = source_group_style_to_renderer.get((e.source.group_id, default_rendering_style.id)) if not cr and default_rendering_style: cr = source_style_to_renderer.get((e.source_id, default_rendering_style.id)) if cr: e._context_renderers[m] = cr def get_default_rendering_style(): default_rendering_style = getattr(settings, 'DEFAULT_ENTITY_EVENT_RENDERING_STYLE', None) if default_rendering_style: default_rendering_style = get_model('entity_event', 'RenderingStyle').objects.get(name=default_rendering_style) return default_rendering_style
ambitioninc/django-entity-event
entity_event/models.py
_unseen_event_ids
python
def _unseen_event_ids(medium): query = ''' SELECT event.id FROM entity_event_event AS event LEFT OUTER JOIN (SELECT * FROM entity_event_eventseen AS seen WHERE seen.medium_id=%s) AS eventseen ON event.id = eventseen.event_id WHERE eventseen.medium_id IS NULL ''' unseen_events = Event.objects.raw(query, params=[medium.id]) ids = [e.id for e in unseen_events] return ids
Return all events that have not been seen on this medium.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/models.py#L1191-L1206
null
from collections import defaultdict from datetime import datetime from operator import or_ from six.moves import reduce from cached_property import cached_property from django.contrib.postgres.fields import JSONField from django.core.serializers.json import DjangoJSONEncoder from django.db import models, transaction from django.db.models import Q from django.db.models.query import QuerySet from django.template.loader import render_to_string from django.template import Context, Template from django.utils.encoding import python_2_unicode_compatible from entity.models import Entity, EntityRelationship from entity_event.context_serializer import DefaultContextSerializer @python_2_unicode_compatible class Medium(models.Model): """ A ``Medium`` is an object in the database that defines the method by which users will view events. The actual objects in the database are fairly simple, only requiring a ``name``, ``display_name`` and ``description``. Mediums can be created with ``Medium.objects.create``, using the following parameters: :type name: str :param name: A short, unique name for the medium. :type display_name: str :param display_name: A short, human readable name for the medium. Does not need to be unique. :type description: str :param description: A human readable description of the medium. Encoding a ``Medium`` object in the database serves two purposes. First, it is referenced when subscriptions are created. Second the ``Medium`` objects provide an entry point to query for events and have all the subscription logic and filtering taken care of for you. Any time a new way to display events to a user is created, a corresponding ``Medium`` should be created. Some examples could include a medium for sending email notifications, a medium for individual newsfeeds, or a medium for a site wide notification center. Once a medium object is created, and corresponding subscriptions are created, there are three methods on the medium object that can be used to query for events. They are ``events``, ``entity_events`` and ``events_targets``. The differences between these methods are described in their corresponding documentation. A medium can use a ``RenderingStyle`` to use a configured style of rendering with the medium. Any associated ``ContextRenderer`` models defined with that rendering style will be used to render events in the ``render`` method of the medium. This is an optional part of Entity Event's built-in rendering system. If a rendering style is not set up for a particular source or source group, it will try to use the default rendering style specified in settings. A medium can also provided ``additional_context`` that will always be passed to the templates of its rendered events. This allows for medium-specific rendering styles to be used. For example, perhaps a medium wishes to display a short description of an event but does not wish to display the names of the event actors since those names are already displayed in other places on the page. In this case, the medium can always pass additional context to suppress rendering of names. """ # A name and display name for the medium along with a description for any # application display name = models.CharField(max_length=64, unique=True) display_name = models.CharField(max_length=64) description = models.TextField() time_created = models.DateTimeField(auto_now_add=True) # The rendering style determines the primary way the medium will try to render events. # If a context loader has been defined for this rendering style along with the appropriate # source, the renderer will be used. If a context renderer has not been set up with this # rendering style, it will try to use the default style configured in settings. rendering_style = models.ForeignKey('entity_event.RenderingStyle', null=True, on_delete=models.CASCADE) # These values are passed in as additional context to whatever event is being rendered. additional_context = JSONField(null=True, default=None, encoder=DjangoJSONEncoder) def __str__(self): """ Readable representation of ``Medium`` objects. """ return self.display_name @transaction.atomic def events(self, **event_filters): """ Return subscribed events, with basic filters. This method of getting events is useful when you want to display events for your medium, independent of what entities were involved in those events. For example, this method can be used to display a list of site-wide events that happened in the past 24 hours: .. code-block:: python TEMPLATE = ''' <html><body> <h1> Yoursite's Events </h1> <ul> {% for event in events %} <li> {{ event.context.event_text }} </li> {% endfor %} </ul> </body></html> ''' def site_feed(request): site_feed_medium = Medium.objects.get(name='site_feed') start_time = datetime.utcnow() - timedelta(days=1) context = {} context['events'] = site_feed_medium.events(start_time=start_time) return HttpResponse(TEMPLATE.render(context)) While the `events` method does not filter events based on what entities are involved, filtering based on the properties of the events themselves is supported, through the following arguments, all of which are optional. :type start_time: datetime.datetime (optional) :param start_time: Only return events that occurred after the given time. If no time is given for this argument, no filtering is done. :type end_time: datetime.datetime (optional) :param end_time: Only return events that occurred before the given time. If no time is given for this argument, no filtering is done :type seen: Boolean (optional) :param seen: This flag controls whether events that have marked as seen are included. By default, both events that have and have not been marked as seen are included. If ``True`` is given for this parameter, only events that have been marked as seen will be included. If ``False`` is given, only events that have not been marked as seen will be included. :type include_expired: Boolean (optional) :param include_expired: By default, events that have a expiration time, which has passed, are not included in the results. Passing in ``True`` to this argument causes expired events to be returned as well. :type actor: Entity (optional) :param actor: Only include events with the given entity as an actor. :type mark_seen: Boolean (optional) :param mark_seen: Create a side effect in the database that marks all the returned events as having been seen by this medium. :rtype: EventQuerySet :returns: A queryset of events. """ events = self.get_filtered_events(**event_filters) subscriptions = Subscription.objects.cache_related().filter( medium=self ) subscription_q_objects = [ Q( eventactor__entity__in=self.followed_by(sub.subscribed_entities()), source=sub.source ) for sub in subscriptions if sub.only_following ] subscription_q_objects.append( Q(source__in=[sub.source for sub in subscriptions if not sub.only_following])) events = events.cache_related().filter(reduce(or_, subscription_q_objects)) return events @transaction.atomic def entity_events(self, entity, **event_filters): """ Return subscribed events for a given entity. This method of getting events is useful when you want to see only the events relevant to a single entity. The events returned are events that the given entity is subscribed to, either directly as an individual entity, or because they are part of a group subscription. As an example, the `entity_events` method can be used to implement a newsfeed for a individual entity: .. code-block:: python TEMPLATE = ''' <html><body> <h1> {entity}'s Events </h1> <ul> {% for event in events %} <li> {{ event.context.event_text }} </li> {% endfor %} </ul> </body></html> ''' def newsfeed(request): newsfeed_medium = Medium.objects.get(name='newsfeed') entity = Entity.get_for_obj(request.user) context = {} context['entity'] = entity context['events'] = site_feed_medium.entity_events(entity, seen=False, mark_seen=True) return HttpResponse(TEMPLATE.render(context)) The only required argument for this method is the entity to get events for. Filtering based on the properties of the events themselves is supported, through the rest of the following arguments, which are optional. :type_entity: Entity :param entity: The entity to get events for. :type start_time: datetime.datetime (optional) :param start_time: Only return events that occurred after the given time. If no time is given for this argument, no filtering is done. :type end_time: datetime.datetime (optional) :param end_time: Only return events that occurred before the given time. If no time is given for this argument, no filtering is done :type seen: Boolean (optional) :param seen: This flag controls whether events that have marked as seen are included. By default, both events that have and have not been marked as seen are included. If ``True`` is given for this parameter, only events that have been marked as seen will be included. If ``False`` is given, only events that have not been marked as seen will be included. :type include_expired: Boolean (optional) :param include_expired: By default, events that have a expiration time, which has passed, are not included in the results. Passing in ``True`` to this argument causes expired events to be returned as well. :type actor: Entity (optional) :param actor: Only include events with the given entity as an actor. :type mark_seen: Boolean (optional) :param mark_seen: Create a side effect in the database that marks all the returned events as having been seen by this medium. :rtype: EventQuerySet :returns: A queryset of events. """ events = self.get_filtered_events(**event_filters) subscriptions = Subscription.objects.filter(medium=self) subscriptions = self.subset_subscriptions(subscriptions, entity) subscription_q_objects = [ Q( eventactor__entity__in=self.followed_by(entity), source=sub.source ) for sub in subscriptions if sub.only_following ] subscription_q_objects.append( Q(source__in=[sub.source for sub in subscriptions if not sub.only_following]) ) return [ event for event in events.filter(reduce(or_, subscription_q_objects)) if self.filter_source_targets_by_unsubscription(event.source_id, [entity]) ] @transaction.atomic def events_targets(self, entity_kind=None, **event_filters): """ Return all events for this medium, with who each event is for. This method is useful for individually notifying every entity concerned with a collection of events, while still respecting subscriptions and usubscriptions. For example, ``events_targets`` can be used to send email notifications, by retrieving all unseen events (and marking them as now having been seen), and then processing the emails. In code, this could look like: .. code-block:: python email = Medium.objects.get(name='email') new_emails = email.events_targets(seen=False, mark_seen=True) for event, targets in new_emails: django.core.mail.send_mail( subject = event.context["subject"] message = event.context["message"] recipient_list = [t.entity_meta["email"] for t in targets] ) This ``events_targets`` method attempts to make bulk processing of push-style notifications straightforward. This sort of processing should normally occur in a separate thread from any request/response cycle. Filtering based on the properties of the events themselves is supported, through the rest of the following arguments, which are optional. :type entity_kind: EntityKind :param entity_kind: Only include targets of the given kind in each targets list. :type start_time: datetime.datetime (optional) :param start_time: Only return events that occurred after the given time. If no time is given for this argument, no filtering is done. :type end_time: datetime.datetime (optional) :param end_time: Only return events that occurred before the given time. If no time is given for this argument, no filtering is done :type seen: Boolean (optional) :param seen: This flag controls whether events that have marked as seen are included. By default, both events that have and have not been marked as seen are included. If ``True`` is given for this parameter, only events that have been marked as seen will be included. If ``False`` is given, only events that have not been marked as seen will be included. :type include_expired: Boolean (optional) :param include_expired: By default, events that have a expiration time, which has passed, are not included in the results. Passing in ``True`` to this argument causes expired events to be returned as well. :type actor: Entity (optional) :param actor: Only include events with the given entity as an actor. :type mark_seen: Boolean (optional) :param mark_seen: Create a side effect in the database that marks all the returned events as having been seen by this medium. :rtype: List of tuples :returns: A list of tuples in the form ``(event, targets)`` where ``targets`` is a list of entities. """ events = self.get_filtered_events(**event_filters) subscriptions = Subscription.objects.filter(medium=self) event_pairs = [] for event in events: targets = [] for sub in subscriptions: if event.source != sub.source: continue subscribed = sub.subscribed_entities() if sub.only_following: potential_targets = self.followers_of( event.eventactor_set.values_list('entity__id', flat=True) ) subscription_targets = list(Entity.objects.filter( Q(id__in=subscribed), Q(id__in=potential_targets))) else: subscription_targets = list(subscribed) targets.extend(subscription_targets) targets = self.filter_source_targets_by_unsubscription(event.source_id, targets) if entity_kind: targets = [t for t in targets if t.entity_kind == entity_kind] if targets: event_pairs.append((event, targets)) return event_pairs def subset_subscriptions(self, subscriptions, entity=None): """ Return only subscriptions the given entity is a part of. An entity is "part of a subscription" if either: 1. The subscription is for that entity, with no sub-entity-kind. That is, it is not a group subscription. 2. The subscription is for a super-entity of the given entity, and the subscription's sub-entity-kind is the same as that of the entity's. :type subscriptions: QuerySet :param subscriptions: A QuerySet of subscriptions to subset. :type entity: (optional) Entity :param entity: Subset subscriptions to only those relevant for this entity. :rtype: QuerySet :returns: A queryset of filtered subscriptions. """ if entity is None: return subscriptions super_entities = EntityRelationship.objects.filter( sub_entity=entity).values_list('super_entity') subscriptions = subscriptions.filter( Q(entity=entity, sub_entity_kind=None) | Q(entity__in=super_entities, sub_entity_kind=entity.entity_kind) ) return subscriptions @cached_property def unsubscriptions(self): """ Returns the unsubscribed entity IDs for each source as a dict, keyed on source_id. :rtype: Dictionary :returns: A dictionary of the form ``{source_id: entities}`` where ``entities`` is a list of entities unsubscribed from that source for this medium. """ unsubscriptions = defaultdict(list) for unsub in Unsubscription.objects.filter(medium=self).values('entity', 'source'): unsubscriptions[unsub['source']].append(unsub['entity']) return unsubscriptions def filter_source_targets_by_unsubscription(self, source_id, targets): """ Given a source id and targets, filter the targets by unsubscriptions. Return the filtered list of targets. """ unsubscriptions = self.unsubscriptions return [t for t in targets if t.id not in unsubscriptions[source_id]] def get_filtered_events_queries(self, start_time, end_time, seen, include_expired, actor): """ Return Q objects to filter events table to relevant events. The filters that are applied are those passed in from the method that is querying the events table: One of ``events``, ``entity_events`` or ``events_targets``. The arguments have the behavior documented in those methods. :rtype: List of Q objects :returns: A list of Q objects, which can be used as arguments to ``Event.objects.filter``. """ now = datetime.utcnow() filters = [] if start_time is not None: filters.append(Q(time__gte=start_time)) if end_time is not None: filters.append(Q(time__lte=end_time)) if not include_expired: filters.append(Q(time_expires__gte=now)) # Check explicitly for True and False as opposed to None # - `seen==False` gets unseen notifications # - `seen is None` does no seen/unseen filtering if seen is True: filters.append(Q(eventseen__medium=self)) elif seen is False: unseen_ids = _unseen_event_ids(medium=self) filters.append(Q(id__in=unseen_ids)) # Filter by actor if actor is not None: filters.append(Q(eventactor__entity=actor)) return filters def get_filtered_events( self, start_time=None, end_time=None, seen=None, mark_seen=False, include_expired=False, actor=None): """ Retrieves events, filters by event level filters, and marks them as seen if necessary. :rtype: EventQuerySet :returns: All events which match the given filters. """ filtered_events_queries = self.get_filtered_events_queries(start_time, end_time, seen, include_expired, actor) events = Event.objects.filter(*filtered_events_queries) if seen is False and mark_seen: # Evaluate the event qset here and create a new queryset that is no longer filtered by # if the events are marked as seen. We do this because we want to mark the events # as seen in the next line of code. If we didn't evaluate the qset here first, it result # in not returning unseen events since they are marked as seen. events = Event.objects.filter(id__in=list(e.id for e in events)) events.mark_seen(self) return events def followed_by(self, entities): """ Define what entities are followed by the entities passed to this method. This method can be overridden by a class that concretely inherits ``Medium``, to define custom semantics for the ``only_following`` flag on relevant ``Subscription`` objects. Overriding this method, and ``followers_of`` will be sufficient to define that behavior. This method is not useful to call directly, but is used by the methods that filter events and targets. This implementation attempts to provide a sane default. In this implementation, the entities followed by the ``entities`` argument are the entities themselves, and their super entities. That is, individual entities follow themselves, and the groups they are a part of. This works as a default implementation, but, for example, an alternate medium may wish to define the opposite behavior, where an individual entity follows themselves and all of their sub-entities. Return a queryset of the entities that the given entities are following. This needs to be the inverse of ``followers_of``. :type entities: Entity or EntityQuerySet :param entities: The Entity, or QuerySet of Entities of interest. :rtype: EntityQuerySet :returns: A QuerySet of all the entities followed by any of those given. """ if isinstance(entities, Entity): entities = Entity.objects.filter(id=entities.id) super_entities = EntityRelationship.objects.filter( sub_entity__in=entities).values_list('super_entity') followed_by = Entity.objects.filter( Q(id__in=entities) | Q(id__in=super_entities)) return followed_by def followers_of(self, entities): """ Define what entities are followers of the entities passed to this method. This method can be overridden by a class that concretely inherits ``Medium``, to define custom semantics for the ``only_following`` flag on relevant ``Subscription`` objects. Overriding this method, and ``followed_by`` will be sufficient to define that behavior. This method is not useful to call directly, but is used by the methods that filter events and targets. This implementation attempts to provide a sane default. In this implementation, the followers of the entities passed in are defined to be the entities themselves, and their sub-entities. That is, the followers of individual entities are themselves, and if the entity has sub-entities, those sub-entities. This works as a default implementation, but, for example, an alternate medium may wish to define the opposite behavior, where an the followers of an individual entity are themselves and all of their super-entities. Return a queryset of the entities that follow the given entities. This needs to be the inverse of ``followed_by``. :type entities: Entity or EntityQuerySet :param entities: The Entity, or QuerySet of Entities of interest. :rtype: EntityQuerySet :returns: A QuerySet of all the entities who are followers of any of those given. """ if isinstance(entities, Entity): entities = Entity.objects.filter(id=entities.id) sub_entities = EntityRelationship.objects.filter( super_entity__in=entities).values_list('sub_entity') followers_of = Entity.objects.filter( Q(id__in=entities) | Q(id__in=sub_entities)) return followers_of def render(self, events): """ Renders a list of events for this medium. The events first have their contexts loaded. Afterwards, the rendered events are returned as a dictionary keyed on the event itself. The key points to a tuple of (txt, html) renderings of the event. :type events: list :param events: A list or queryset of Event models. :rtype: dict :returns: A dictionary of rendered text and html tuples keyed on the provided events. """ from entity_event import context_loader context_loader.load_contexts_and_renderers(events, [self]) return {e: e.render(self) for e in events} @python_2_unicode_compatible class Source(models.Model): """ A ``Source`` is an object in the database that represents where events come from. These objects only require a few fields, ``name``, ``display_name`` ``description``, and ``group``. Source objects categorize events based on where they came from, or what type of information they contain. Each source should be fairly fine grained, with broader categorizations possible through ``SourceGroup`` objects. Sources can be created with ``Source.objects.create`` using the following parameters: :type name: str :param name: A short, unique name for the source. :type display_name: str :param display_name: A short, human readable name for the source. Does not need to be unique. :type description: str :param description: A human readable description of the source. :type group: SourceGroup :param group: A SourceGroup object. A broad grouping of where the events originate. Storing source objects in the database servers two purposes. The first is to provide an object that Subscriptions can reference, allowing different categories of events to be subscribed to over different mediums. The second is to allow source instances to store a reference to a function which can populate event contexts with additional information that is relevant to the source. This allows ``Event`` objects to be created with minimal data duplication. Once sources are created, they will primarily be used to categorize events, as each ``Event`` object requires a reference to a source. Additionally they will be referenced by ``Subscription`` objects to route events of the given source to be handled by a given medium. """ name = models.CharField(max_length=64, unique=True) display_name = models.CharField(max_length=64) description = models.TextField() group = models.ForeignKey('entity_event.SourceGroup', on_delete=models.CASCADE) def __str__(self): """ Readable representation of ``Source`` objects. """ return self.display_name @python_2_unicode_compatible class SourceGroup(models.Model): """ A ``SourceGroup`` object is a high level categorization of events. Since ``Source`` objects are meant to be very fine grained, they are collected into ``SourceGroup`` objects. There is no additional behavior associated with the source groups other than further categorization. Source groups can be created with ``SourceGroup.objects.create``, which takes the following arguments: :type name: str :param name: A short, unique name for the source group. :type display_name: str :param display_name: A short, human readable name for the source group. Does not need to be unique. :type description: str :param description: A human readable description of the source group. """ name = models.CharField(max_length=64, unique=True) display_name = models.CharField(max_length=64) description = models.TextField() def __str__(self): """ Readable representation of ``SourceGroup`` objects. """ return self.display_name @python_2_unicode_compatible class Unsubscription(models.Model): """ Because django-entity-event allows for whole groups to be subscribed to events at once, unsubscribing an entity is not as simple as removing their subscription object. Instead, the Unsubscription table provides a simple way to ensure that an entity does not see events if they don't want to. Unsubscriptions are created for a single entity at a time, where they are unsubscribed for events from a source on a medium. This is stored as an ``Unsubscription`` object in the database, which can be created using ``Unsubscription.objects.create`` using the following arguments: :type entity: Entity :param entity: The entity to unsubscribe. :type medium: Medium :param medium: The ``Medium`` object representing where they don't want to see the events. :type source: Source :param source: The ``Source`` object representing what category of event they no longer want to see. Once an ``Unsubscription`` object is created, all of the logic to ensure that they do not see events form the given source by the given medium is handled by the methods used to query for events via the ``Medium`` object. That is, once the object is created, no more work is needed to unsubscribe them. """ entity = models.ForeignKey('entity.Entity', on_delete=models.CASCADE) medium = models.ForeignKey('entity_event.Medium', on_delete=models.CASCADE) source = models.ForeignKey('entity_event.Source', on_delete=models.CASCADE) def __str__(self): """ Readable representation of ``Unsubscription`` objects. """ s = '{entity} from {source} by {medium}' entity = self.entity.__str__() source = self.source.__str__() medium = self.medium.__str__() return s.format(entity=entity, source=source, medium=medium) class SubscriptionQuerySet(QuerySet): """ A custom QuerySet for Subscriptions. """ def cache_related(self): """ Cache any related objects that we may use :return: """ return self.select_related('medium', 'source', 'entity', 'sub_entity_kind') @python_2_unicode_compatible class Subscription(models.Model): """ Which types of events are available to which mediums is controlled through ``Subscription`` objects. By creating a ``Subscription`` object in the database, you are storing that events from a given ``Source`` object should be available to a given ``Medium`` object. Each ``Subscription`` object can be one of two levels, either an individual subscription or a group subscription. Additionally, each ``Subscription`` object can be one of two types of subscription, either a global subscription, or an "only following" subscription. ``Subscription`` objects are created using ``Subscription.objects.create`` which takes the following arguments: :type medium: Medium :param medium: The ``Medium`` object to make events available to. :type source: Source :param source: The ``Source`` object that represents the category of events to make available. :type entity: Entity :param entity: The entity to subscribe in the case of an individual subscription, or in the case of a group subscription, the super-entity of the group. :type sub_entity_kind: (optional) EntityKind :param sub_entity_kind: When creating a group subscription, this is a foreign key to the ``EntityKind`` of the sub-entities to subscribe. In the case of an individual subscription, this should be ``None``. :type only_following: Boolean :param only_following: If ``True``, events will be available to entities through the medium only if the entities are following the actors of the event. If ``False``, the events will all be available to all the entities through the medium. When a ``Medium`` object is used to query for events, only the events that have a subscription for their source to that medium will ever be returned. This is an extremely useful property that allows complex subscription logic to be handled simply by storing subscription objects in the database. Storing subscriptions is made simpler by the ability to subscribe groups of entities with a single subscription object. Groups of entities of a given kind can be subscribed by subscribing their super-entity and providing the ``sub_entity_kind`` argument. Subscriptions further are specified to be either an "only following" subscription or not. This specification controls what events will be returned when ``Medium.entity_events`` is called, and controls what targets are returned when ``Medium.events_targets`` is called. For example, if events are created for a new photo being uploaded (from a single source called, say "photos"), and we want to provide individuals with a notification in their newsfeed (through a medium called "newsfeed"), we want to be able to display only the events where the individual is tagged in the photo. By setting ``only_following`` to true the following code would only return events where the individual was included in the ``EventActor`` s, rather than returning all "photos" events: .. code-block:: python user_entity = Entity.objects.get_for_obj(user) newsfeed_medium = Medium.objects.get(name='newsfeed') newsfeed.entity_events(user) The behavior of what constitutes "following" is controlled by the Medium class. A default implementation of following is provided and documented in the ``Medium.followers_of`` and ``Medium.followed_by`` methods, but could be extended by subclasses of Medium. """ medium = models.ForeignKey('entity_event.Medium', on_delete=models.CASCADE) source = models.ForeignKey('entity_event.Source', on_delete=models.CASCADE) entity = models.ForeignKey('entity.Entity', related_name='+', on_delete=models.CASCADE) sub_entity_kind = models.ForeignKey( 'entity.EntityKind', null=True, related_name='+', default=None, on_delete=models.CASCADE ) only_following = models.BooleanField(default=True) objects = SubscriptionQuerySet.as_manager() def __str__(self): """ Readable representation of ``Subscription`` objects. """ s = '{entity} to {source} by {medium}' entity = self.entity.__str__() source = self.source.__str__() medium = self.medium.__str__() return s.format(entity=entity, source=source, medium=medium) def subscribed_entities(self): """ Return a queryset of all subscribed entities. This will be a single entity in the case of an individual subscription, otherwise it will be all the entities in the group subscription. :rtype: EntityQuerySet :returns: A QuerySet of all the entities that are a part of this subscription. """ if self.sub_entity_kind is not None: sub_entities = self.entity.sub_relationships.filter( sub_entity__entity_kind=self.sub_entity_kind).values_list('sub_entity') entities = Entity.objects.filter(id__in=sub_entities) else: entities = Entity.all_objects.filter(id=self.entity.id) return entities class EventQuerySet(QuerySet): """ A custom QuerySet for Events. """ def cache_related(self): """ Cache any related objects that we may use :return: """ return self.select_related( 'source' ).prefetch_related( 'source__group' ) def mark_seen(self, medium): """ Creates EventSeen objects for the provided medium for every event in the queryset. Creating these EventSeen objects ensures they will not be returned when passing ``seen=False`` to any of the medium event retrieval functions, ``events``, ``entity_events``, or ``events_targets``. """ EventSeen.objects.bulk_create([ EventSeen(event=event, medium=medium) for event in self ]) def load_contexts_and_renderers(self, medium): """ Loads context data into the event ``context`` variable. This method destroys the queryset and returns a list of events. """ from entity_event import context_loader return context_loader.load_contexts_and_renderers(self, [medium]) class EventManager(models.Manager): """ A custom Manager for Events. """ def get_queryset(self): """ Return the EventQuerySet. """ return EventQuerySet(self.model) def cache_related(self): """ Return a queryset with prefetched values :return: """ return self.get_queryset().cache_related() def mark_seen(self, medium): """ Creates EventSeen objects for the provided medium for every event in the queryset. Creating these EventSeen objects ensures they will not be returned when passing ``seen=False`` to any of the medium event retrieval functions, ``events``, ``entity_events``, or ``events_targets``. """ return self.get_queryset().mark_seen(medium) def load_contexts_and_renderers(self, medium): """ Loads context data into the event ``context`` variable. This method destroys the queryset and returns a list of events. """ return self.get_queryset().load_contexts_and_renderers(medium) @transaction.atomic def create_event(self, actors=None, ignore_duplicates=False, **kwargs): """ Create events with actors. This method can be used in place of ``Event.objects.create`` to create events, and the appropriate actors. It takes all the same keywords as ``Event.objects.create`` for the event creation, but additionally takes a list of actors, and can be told to not attempt to create an event if a duplicate event exists. :type source: Source :param source: A ``Source`` object representing where the event came from. :type context: dict :param context: A dictionary containing relevant information about the event, to be serialized into JSON. It is possible to load additional context dynamically when events are fetched. See the documentation on the ``ContextRenderer`` model. :type uuid: str :param uuid: A unique string for the event. Requiring a ``uuid`` allows code that creates events to ensure they do not create duplicate events. This id could be, for example some hash of the ``context``, or, if the creator is unconcerned with creating duplicate events a call to python's ``uuid1()`` in the ``uuid`` module. :type time_expires: datetime (optional) :param time_expires: If given, the default methods for querying events will not return this event after this time has passed. :type actors: (optional) List of entities or list of entity ids. :param actors: An ``EventActor`` object will be created for each entity in the list. This allows for subscriptions which are only following certain entities to behave appropriately. :type ignore_duplicates: (optional) Boolean :param ignore_duplicates: If ``True``, a check will be made to ensure that an event with the give ``uuid`` does not exist before attempting to create the event. Setting this to ``True`` allows the creator of events to gracefully ensure no duplicates are attempted to be created. There is a uniqueness constraint on uuid so it will raise an exception if duplicates are allowed and submitted. :rtype: Event :returns: The created event. Alternatively if a duplicate event already exists and ``ignore_duplicates`` is ``True``, it will return ``None``. """ kwargs['actors'] = actors kwargs['ignore_duplicates'] = ignore_duplicates events = self.create_events([kwargs]) if events: return events[0] return None def create_events(self, kwargs_list): """ Create events in bulk to save on queries. Each element in the kwargs list should be a dict with the same set of arguments you would normally pass to create_event :param kwargs_list: list of kwargs dicts :return: list of Event """ # Build map of uuid to event info uuid_map = { kwargs.get('uuid', ''): { 'actors': kwargs.pop('actors', []), 'ignore_duplicates': kwargs.pop('ignore_duplicates', False), 'event_kwargs': kwargs } for kwargs in kwargs_list } # Check for uuids uuid_set = set(Event.objects.filter(uuid__in=uuid_map.keys()).values_list('uuid', flat=True)) # Set a flag for whether each uuid exists for uuid, event_dict in uuid_map.items(): event_dict['exists'] = uuid in uuid_set # Build list of events to bulk create events_to_create = [] for uuid, event_dict in uuid_map.items(): # If the event doesn't already exist or the event does exist but we are allowing duplicates if not event_dict['exists'] or not event_dict['ignore_duplicates']: events_to_create.append(Event(**event_dict['event_kwargs'])) # Bulk create the events created_events = Event.objects.bulk_create(events_to_create) # Build list of EventActor objects to bulk create event_actors_to_create = [] for created_event in created_events: event_dict = uuid_map[created_event.uuid] if event_dict['actors'] is not None: for actor in event_dict['actors']: actor_id = actor.id if hasattr(actor, 'id') else actor event_actors_to_create.append(EventActor(entity_id=actor_id, event=created_event)) EventActor.objects.bulk_create(event_actors_to_create) return created_events @python_2_unicode_compatible class Event(models.Model): """ ``Event`` objects store information about events. By storing events, from a given source, with some context, they are made available to any ``Medium`` object with an appropriate subscription. Events can be created with ``Event.objects.create_event``, documented above. When creating an event, the information about what occurred is stored in a JSON blob in the ``context`` field. This context can be any type of information that could be useful for displaying events on a given Medium. It is entirely the role of the application developer to ensure that there is agreement between what information is stored in ``Event.context`` and what information the code the processes and displays events on each medium expects. Events will usually be created by code that also created, or knows about the ``Source`` object that is required to create the event. To prevent storing unnecessary data in the context, this code can define a context loader function when creating this source, which can be used to dynamically fetch more data based on whatever limited amount of data makes sense to store in the context. This is further documented in the ``Source`` documentation. """ source = models.ForeignKey('entity_event.Source', on_delete=models.CASCADE) context = JSONField(encoder=DjangoJSONEncoder) time = models.DateTimeField(auto_now_add=True, db_index=True) time_expires = models.DateTimeField(default=datetime.max, db_index=True) uuid = models.CharField(max_length=512, unique=True) objects = EventManager() def __init__(self, *args, **kwargs): super(Event, self).__init__(*args, **kwargs) # A dictionary that is populated with renderers after the contexts have been # properly loaded. When renderers are available, the 'render' method may be # called with a medium and optional observer self._context_renderers = {} def _merge_medium_additional_context_with_context(self, medium): """ If the medium has additional context properties, merge those together here in the main context before rendering. """ if medium.additional_context: context = self.context.copy() context.update(medium.additional_context) return context else: return self.context def render(self, medium, observing_entity=None): """ Returns the rendered event as a tuple of text and html content. This information is filled out with respect to which medium is rendering the event, what context renderers are available with the prefetched context, and which optional entity may be observing the rendered event. """ if medium not in self._context_renderers: raise RuntimeError('Context and renderer for medium {0} has not or cannot been fetched'.format(medium)) else: context = self._merge_medium_additional_context_with_context(medium) return self._context_renderers[medium].render_context_to_text_html_templates(context) def get_serialized_context(self, medium): """ Returns the serialized context of the event for a specific medium :param medium: :return: """ if medium not in self._context_renderers: raise RuntimeError('Context and renderer for medium {0} has not or cannot been fetched'.format(medium)) else: context = self._merge_medium_additional_context_with_context(medium) return self._context_renderers[medium].get_serialized_context(context) def __str__(self): """ Readable representation of ``Event`` objects. """ s = '{source} event at {time}' source = self.source.__str__() time = self.time.strftime('%Y-%m-%d::%H:%M:%S') return s.format(source=source, time=time) class AdminEvent(Event): """ A proxy model used to provide a separate interface for event creation through the django-admin interface. """ class Meta: proxy = True @python_2_unicode_compatible class EventActor(models.Model): """ ``EventActor`` objects encode what entities were involved in an event. They provide the information necessary to create "only following" subscriptions which route events only to the entities that are involved in the event. ``EventActor`` objects should not be created directly, but should be created as part of the creation of ``Event`` objects, using ``Event.objects.create_event``. """ event = models.ForeignKey('entity_event.Event', on_delete=models.CASCADE) entity = models.ForeignKey('entity.Entity', on_delete=models.CASCADE) def __str__(self): """ Readable representation of ``EventActor`` objects. """ s = 'Event {eventid} - {entity}' eventid = self.event.id entity = self.entity.__str__() return s.format(eventid=eventid, entity=entity) @python_2_unicode_compatible class EventSeen(models.Model): """ ``EventSeen`` objects store information about where and when an event was seen. They store the medium that the event was seen on, and what time it was seen. This information is used by the event querying methods on ``Medium`` objects to filter events by whether or not they have been seen on that medium. ``EventSeen`` objects should not be created directly, but should be created by using the ``EventQuerySet.mark_seen`` method, available on the QuerySets returned by the event querying methods. """ event = models.ForeignKey('entity_event.Event', on_delete=models.CASCADE) medium = models.ForeignKey('entity_event.Medium', on_delete=models.CASCADE) time_seen = models.DateTimeField(default=datetime.utcnow) class Meta: unique_together = ('event', 'medium') def __str__(self): """ Readable representation of ``EventSeen`` objects. """ s = 'Seen on {medium} at {time}' medium = self.medium.__str__() time = self.time_seen.strftime('%Y-%m-%d::%H:%M:%S') return s.format(medium=medium, time=time) @python_2_unicode_compatible class RenderingStyle(models.Model): """ Defines a rendering style. This is used to group together mediums that have similar rendering styles and allows context renderers to be used across mediums. """ name = models.CharField(max_length=64, unique=True) display_name = models.CharField(max_length=64, default='') def __str__(self): return '{0} {1}'.format(self.display_name, self.name) @python_2_unicode_compatible class ContextRenderer(models.Model): """ ``ContextRenderer`` objects store information about how a source or source group is rendered with a particular rendering style, along with information for loading the render context in a database-efficient manner. Of the four template fields: `text_template_path`, 'html_template_path', `text_template`, and `html_template`, at least one must be non-empty. Both a text and html template may be provided, either through a path to the template, or a raw template object. If both are provided, the template given in the path will be used and the text template will be ignored. This object is linked to a `RenderingStyle` object. This is how the context renderer is associated with various `Medium` objects. It also provides the `source` that uses the renderer. If a `source_group` is specified, all sources under that group use this context renderer for the rendering style. The `context_hints` provide the ability to fetch model IDs of an event context that are stored in the database. For example, if an event context has a `user` key that points to the PK of a Django `User` model, the context hints for it would be specified as follows: .. code-block:: python { 'user': { 'app_name': 'auth', 'model_name': 'User', } } With these hints, the 'user' field in the event context will be treated as a PK in the database and fetched appropriately. If one wishes to perform and prefetch or select_related calls, the following options can be added: .. code-block:: python { 'user': { 'app_name': 'auth', 'model_name': 'User', 'select_related': ['foreign_key_field', 'one_to_one_field'], 'prefetch_related': ['reverse_foreign_key_field', 'many_to_many_field'], } } Note that as many keys can be defined that have corresponding keys in the event context for the particular source or source group. Also note that the keys in the event context can be embedded anywhere in the context and can also point to a list of PKs. For example: .. code-block:: python { 'my_context': { 'user': [1, 3, 5, 10], 'other_context_info': 'other_info_string', }, 'more_context': { 'hello': 'world', } } In the above case, `User` objects with the PKs 1, 3, 5, and 10 will be fetched and loaded into the event context whenever rendering is performed. """ name = models.CharField(max_length=64, unique=True) # The various templates that can be used for rendering text_template_path = models.CharField(max_length=256, default='') html_template_path = models.CharField(max_length=256, default='') text_template = models.TextField(default='') html_template = models.TextField(default='') # The source or source group of the event. It can only be one or the other source = models.ForeignKey('entity_event.Source', null=True, on_delete=models.CASCADE) source_group = models.ForeignKey('entity_event.SourceGroup', null=True, on_delete=models.CASCADE) # The rendering style. Used to associated it with a medium rendering_style = models.ForeignKey('entity_event.RenderingStyle', on_delete=models.CASCADE) # Contains hints on how to fetch the context from the database context_hints = JSONField(null=True, default=None, encoder=DjangoJSONEncoder) class Meta: unique_together = ('source', 'rendering_style') def get_sources(self): return [self.source] if self.source_id else self.source_group.source_set.all() def __str__(self): return self.name def get_serialized_context(self, context): """ Serializes the context using the serializer class. """ return DefaultContextSerializer(context).data def render_text_or_html_template(self, context, is_text=True): """ Renders a text or html template based on either the template path or the stored template. """ template_path = getattr(self, '{0}_template_path'.format('text' if is_text else 'html')) template = getattr(self, '{0}_template'.format('text' if is_text else 'html')) if template_path: return render_to_string(template_path, context) elif template: return Template(template).render(Context(context)) else: return '' def render_context_to_text_html_templates(self, context): """ Render the templates with the provided context. Args: A loaded context. Returns: A tuple of (rendered_text, rendered_html). Either, but not both may be an empty string. """ # Process text template: return ( self.render_text_or_html_template(context, is_text=True).strip(), self.render_text_or_html_template(context, is_text=False).strip(), )
ambitioninc/django-entity-event
entity_event/models.py
EventQuerySet.mark_seen
python
def mark_seen(self, medium): EventSeen.objects.bulk_create([ EventSeen(event=event, medium=medium) for event in self ])
Creates EventSeen objects for the provided medium for every event in the queryset. Creating these EventSeen objects ensures they will not be returned when passing ``seen=False`` to any of the medium event retrieval functions, ``events``, ``entity_events``, or ``events_targets``.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/models.py#L866-L878
null
class EventQuerySet(QuerySet): """ A custom QuerySet for Events. """ def cache_related(self): """ Cache any related objects that we may use :return: """ return self.select_related( 'source' ).prefetch_related( 'source__group' ) def load_contexts_and_renderers(self, medium): """ Loads context data into the event ``context`` variable. This method destroys the queryset and returns a list of events. """ from entity_event import context_loader return context_loader.load_contexts_and_renderers(self, [medium])
ambitioninc/django-entity-event
entity_event/models.py
EventManager.create_event
python
def create_event(self, actors=None, ignore_duplicates=False, **kwargs): kwargs['actors'] = actors kwargs['ignore_duplicates'] = ignore_duplicates events = self.create_events([kwargs]) if events: return events[0] return None
Create events with actors. This method can be used in place of ``Event.objects.create`` to create events, and the appropriate actors. It takes all the same keywords as ``Event.objects.create`` for the event creation, but additionally takes a list of actors, and can be told to not attempt to create an event if a duplicate event exists. :type source: Source :param source: A ``Source`` object representing where the event came from. :type context: dict :param context: A dictionary containing relevant information about the event, to be serialized into JSON. It is possible to load additional context dynamically when events are fetched. See the documentation on the ``ContextRenderer`` model. :type uuid: str :param uuid: A unique string for the event. Requiring a ``uuid`` allows code that creates events to ensure they do not create duplicate events. This id could be, for example some hash of the ``context``, or, if the creator is unconcerned with creating duplicate events a call to python's ``uuid1()`` in the ``uuid`` module. :type time_expires: datetime (optional) :param time_expires: If given, the default methods for querying events will not return this event after this time has passed. :type actors: (optional) List of entities or list of entity ids. :param actors: An ``EventActor`` object will be created for each entity in the list. This allows for subscriptions which are only following certain entities to behave appropriately. :type ignore_duplicates: (optional) Boolean :param ignore_duplicates: If ``True``, a check will be made to ensure that an event with the give ``uuid`` does not exist before attempting to create the event. Setting this to ``True`` allows the creator of events to gracefully ensure no duplicates are attempted to be created. There is a uniqueness constraint on uuid so it will raise an exception if duplicates are allowed and submitted. :rtype: Event :returns: The created event. Alternatively if a duplicate event already exists and ``ignore_duplicates`` is ``True``, it will return ``None``.
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/models.py#L926-L988
[ "def create_events(self, kwargs_list):\n \"\"\"\n Create events in bulk to save on queries. Each element in the kwargs list should be a dict with the same set\n of arguments you would normally pass to create_event\n :param kwargs_list: list of kwargs dicts\n :return: list of Event\n \"\"\"\n # Build map of uuid to event info\n uuid_map = {\n kwargs.get('uuid', ''): {\n 'actors': kwargs.pop('actors', []),\n 'ignore_duplicates': kwargs.pop('ignore_duplicates', False),\n 'event_kwargs': kwargs\n\n }\n for kwargs in kwargs_list\n }\n\n # Check for uuids\n uuid_set = set(Event.objects.filter(uuid__in=uuid_map.keys()).values_list('uuid', flat=True))\n\n # Set a flag for whether each uuid exists\n for uuid, event_dict in uuid_map.items():\n event_dict['exists'] = uuid in uuid_set\n\n # Build list of events to bulk create\n events_to_create = []\n for uuid, event_dict in uuid_map.items():\n # If the event doesn't already exist or the event does exist but we are allowing duplicates\n if not event_dict['exists'] or not event_dict['ignore_duplicates']:\n events_to_create.append(Event(**event_dict['event_kwargs']))\n\n # Bulk create the events\n created_events = Event.objects.bulk_create(events_to_create)\n\n # Build list of EventActor objects to bulk create\n event_actors_to_create = []\n for created_event in created_events:\n event_dict = uuid_map[created_event.uuid]\n if event_dict['actors'] is not None:\n for actor in event_dict['actors']:\n actor_id = actor.id if hasattr(actor, 'id') else actor\n event_actors_to_create.append(EventActor(entity_id=actor_id, event=created_event))\n\n EventActor.objects.bulk_create(event_actors_to_create)\n\n return created_events\n" ]
class EventManager(models.Manager): """ A custom Manager for Events. """ def get_queryset(self): """ Return the EventQuerySet. """ return EventQuerySet(self.model) def cache_related(self): """ Return a queryset with prefetched values :return: """ return self.get_queryset().cache_related() def mark_seen(self, medium): """ Creates EventSeen objects for the provided medium for every event in the queryset. Creating these EventSeen objects ensures they will not be returned when passing ``seen=False`` to any of the medium event retrieval functions, ``events``, ``entity_events``, or ``events_targets``. """ return self.get_queryset().mark_seen(medium) def load_contexts_and_renderers(self, medium): """ Loads context data into the event ``context`` variable. This method destroys the queryset and returns a list of events. """ return self.get_queryset().load_contexts_and_renderers(medium) @transaction.atomic def create_events(self, kwargs_list): """ Create events in bulk to save on queries. Each element in the kwargs list should be a dict with the same set of arguments you would normally pass to create_event :param kwargs_list: list of kwargs dicts :return: list of Event """ # Build map of uuid to event info uuid_map = { kwargs.get('uuid', ''): { 'actors': kwargs.pop('actors', []), 'ignore_duplicates': kwargs.pop('ignore_duplicates', False), 'event_kwargs': kwargs } for kwargs in kwargs_list } # Check for uuids uuid_set = set(Event.objects.filter(uuid__in=uuid_map.keys()).values_list('uuid', flat=True)) # Set a flag for whether each uuid exists for uuid, event_dict in uuid_map.items(): event_dict['exists'] = uuid in uuid_set # Build list of events to bulk create events_to_create = [] for uuid, event_dict in uuid_map.items(): # If the event doesn't already exist or the event does exist but we are allowing duplicates if not event_dict['exists'] or not event_dict['ignore_duplicates']: events_to_create.append(Event(**event_dict['event_kwargs'])) # Bulk create the events created_events = Event.objects.bulk_create(events_to_create) # Build list of EventActor objects to bulk create event_actors_to_create = [] for created_event in created_events: event_dict = uuid_map[created_event.uuid] if event_dict['actors'] is not None: for actor in event_dict['actors']: actor_id = actor.id if hasattr(actor, 'id') else actor event_actors_to_create.append(EventActor(entity_id=actor_id, event=created_event)) EventActor.objects.bulk_create(event_actors_to_create) return created_events
ambitioninc/django-entity-event
entity_event/models.py
EventManager.create_events
python
def create_events(self, kwargs_list): # Build map of uuid to event info uuid_map = { kwargs.get('uuid', ''): { 'actors': kwargs.pop('actors', []), 'ignore_duplicates': kwargs.pop('ignore_duplicates', False), 'event_kwargs': kwargs } for kwargs in kwargs_list } # Check for uuids uuid_set = set(Event.objects.filter(uuid__in=uuid_map.keys()).values_list('uuid', flat=True)) # Set a flag for whether each uuid exists for uuid, event_dict in uuid_map.items(): event_dict['exists'] = uuid in uuid_set # Build list of events to bulk create events_to_create = [] for uuid, event_dict in uuid_map.items(): # If the event doesn't already exist or the event does exist but we are allowing duplicates if not event_dict['exists'] or not event_dict['ignore_duplicates']: events_to_create.append(Event(**event_dict['event_kwargs'])) # Bulk create the events created_events = Event.objects.bulk_create(events_to_create) # Build list of EventActor objects to bulk create event_actors_to_create = [] for created_event in created_events: event_dict = uuid_map[created_event.uuid] if event_dict['actors'] is not None: for actor in event_dict['actors']: actor_id = actor.id if hasattr(actor, 'id') else actor event_actors_to_create.append(EventActor(entity_id=actor_id, event=created_event)) EventActor.objects.bulk_create(event_actors_to_create) return created_events
Create events in bulk to save on queries. Each element in the kwargs list should be a dict with the same set of arguments you would normally pass to create_event :param kwargs_list: list of kwargs dicts :return: list of Event
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/models.py#L990-L1036
null
class EventManager(models.Manager): """ A custom Manager for Events. """ def get_queryset(self): """ Return the EventQuerySet. """ return EventQuerySet(self.model) def cache_related(self): """ Return a queryset with prefetched values :return: """ return self.get_queryset().cache_related() def mark_seen(self, medium): """ Creates EventSeen objects for the provided medium for every event in the queryset. Creating these EventSeen objects ensures they will not be returned when passing ``seen=False`` to any of the medium event retrieval functions, ``events``, ``entity_events``, or ``events_targets``. """ return self.get_queryset().mark_seen(medium) def load_contexts_and_renderers(self, medium): """ Loads context data into the event ``context`` variable. This method destroys the queryset and returns a list of events. """ return self.get_queryset().load_contexts_and_renderers(medium) @transaction.atomic def create_event(self, actors=None, ignore_duplicates=False, **kwargs): """ Create events with actors. This method can be used in place of ``Event.objects.create`` to create events, and the appropriate actors. It takes all the same keywords as ``Event.objects.create`` for the event creation, but additionally takes a list of actors, and can be told to not attempt to create an event if a duplicate event exists. :type source: Source :param source: A ``Source`` object representing where the event came from. :type context: dict :param context: A dictionary containing relevant information about the event, to be serialized into JSON. It is possible to load additional context dynamically when events are fetched. See the documentation on the ``ContextRenderer`` model. :type uuid: str :param uuid: A unique string for the event. Requiring a ``uuid`` allows code that creates events to ensure they do not create duplicate events. This id could be, for example some hash of the ``context``, or, if the creator is unconcerned with creating duplicate events a call to python's ``uuid1()`` in the ``uuid`` module. :type time_expires: datetime (optional) :param time_expires: If given, the default methods for querying events will not return this event after this time has passed. :type actors: (optional) List of entities or list of entity ids. :param actors: An ``EventActor`` object will be created for each entity in the list. This allows for subscriptions which are only following certain entities to behave appropriately. :type ignore_duplicates: (optional) Boolean :param ignore_duplicates: If ``True``, a check will be made to ensure that an event with the give ``uuid`` does not exist before attempting to create the event. Setting this to ``True`` allows the creator of events to gracefully ensure no duplicates are attempted to be created. There is a uniqueness constraint on uuid so it will raise an exception if duplicates are allowed and submitted. :rtype: Event :returns: The created event. Alternatively if a duplicate event already exists and ``ignore_duplicates`` is ``True``, it will return ``None``. """ kwargs['actors'] = actors kwargs['ignore_duplicates'] = ignore_duplicates events = self.create_events([kwargs]) if events: return events[0] return None
ambitioninc/django-entity-event
entity_event/context_serializer.py
DefaultContextSerializer.serialize_value
python
def serialize_value(self, value): # Create a list of serialize methods to run the value through serialize_methods = [ self.serialize_model, self.serialize_json_string, self.serialize_list, self.serialize_dict ] # Run all of our serialize methods over our value for serialize_method in serialize_methods: value = serialize_method(value) # Return the serialized context value return value
Given a value, ensure that it is serialized properly :param value: :return:
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_serializer.py#L29-L48
null
class DefaultContextSerializer(object): """ Default class for serializing context data """ def __init__(self, context): super(DefaultContextSerializer, self).__init__() self.context = context @property def data(self): """ Data property that will return the serialized data :return: """ # Create a serialized context dict serialized_context = self.serialize_value(self.context) # Return the serialized context return serialized_context def serialize_model(self, value): """ Serializes a model and all of its prefetched foreign keys :param value: :return: """ # Check if the context value is a model if not isinstance(value, models.Model): return value # Serialize the model serialized_model = model_to_dict(value) # Check the model for cached foreign keys for model_field, model_value in serialized_model.items(): model_state = value._state # Django >= 2 if hasattr(model_state, 'fields_cache'): # pragma: no cover if model_state.fields_cache.get(model_field): serialized_model[model_field] = model_state.fields_cache.get(model_field) else: # pragma: no cover # Django < 2 cache_field = '_{0}_cache'.format(model_field) if hasattr(value, cache_field): serialized_model[model_field] = getattr(value, cache_field) # Return the serialized model return self.serialize_value(serialized_model) def serialize_json_string(self, value): """ Tries to load an encoded json string back into an object :param json_string: :return: """ # Check if the value might be a json string if not isinstance(value, six.string_types): return value # Make sure it starts with a brace if not value.startswith('{') or value.startswith('['): return value # Try to load the string try: return json.loads(value) except: return value def serialize_list(self, value): """ Ensure that all values of a list or tuple are serialized :return: """ # Check if this is a list or a tuple if not isinstance(value, (list, tuple)): return value # Loop over all the values and serialize the values return [ self.serialize_value(list_value) for list_value in value ] def serialize_dict(self, value): """ Ensure that all values of a dictionary are properly serialized :param value: :return: """ # Check if this is a dict if not isinstance(value, dict): return value # Loop over all the values and serialize them return { dict_key: self.serialize_value(dict_value) for dict_key, dict_value in value.items() }
ambitioninc/django-entity-event
entity_event/context_serializer.py
DefaultContextSerializer.serialize_model
python
def serialize_model(self, value): # Check if the context value is a model if not isinstance(value, models.Model): return value # Serialize the model serialized_model = model_to_dict(value) # Check the model for cached foreign keys for model_field, model_value in serialized_model.items(): model_state = value._state # Django >= 2 if hasattr(model_state, 'fields_cache'): # pragma: no cover if model_state.fields_cache.get(model_field): serialized_model[model_field] = model_state.fields_cache.get(model_field) else: # pragma: no cover # Django < 2 cache_field = '_{0}_cache'.format(model_field) if hasattr(value, cache_field): serialized_model[model_field] = getattr(value, cache_field) # Return the serialized model return self.serialize_value(serialized_model)
Serializes a model and all of its prefetched foreign keys :param value: :return:
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_serializer.py#L50-L79
[ "def serialize_value(self, value):\n \"\"\"\n Given a value, ensure that it is serialized properly\n :param value:\n :return:\n \"\"\"\n # Create a list of serialize methods to run the value through\n serialize_methods = [\n self.serialize_model,\n self.serialize_json_string,\n self.serialize_list,\n self.serialize_dict\n ]\n\n # Run all of our serialize methods over our value\n for serialize_method in serialize_methods:\n value = serialize_method(value)\n\n # Return the serialized context value\n return value\n" ]
class DefaultContextSerializer(object): """ Default class for serializing context data """ def __init__(self, context): super(DefaultContextSerializer, self).__init__() self.context = context @property def data(self): """ Data property that will return the serialized data :return: """ # Create a serialized context dict serialized_context = self.serialize_value(self.context) # Return the serialized context return serialized_context def serialize_value(self, value): """ Given a value, ensure that it is serialized properly :param value: :return: """ # Create a list of serialize methods to run the value through serialize_methods = [ self.serialize_model, self.serialize_json_string, self.serialize_list, self.serialize_dict ] # Run all of our serialize methods over our value for serialize_method in serialize_methods: value = serialize_method(value) # Return the serialized context value return value def serialize_json_string(self, value): """ Tries to load an encoded json string back into an object :param json_string: :return: """ # Check if the value might be a json string if not isinstance(value, six.string_types): return value # Make sure it starts with a brace if not value.startswith('{') or value.startswith('['): return value # Try to load the string try: return json.loads(value) except: return value def serialize_list(self, value): """ Ensure that all values of a list or tuple are serialized :return: """ # Check if this is a list or a tuple if not isinstance(value, (list, tuple)): return value # Loop over all the values and serialize the values return [ self.serialize_value(list_value) for list_value in value ] def serialize_dict(self, value): """ Ensure that all values of a dictionary are properly serialized :param value: :return: """ # Check if this is a dict if not isinstance(value, dict): return value # Loop over all the values and serialize them return { dict_key: self.serialize_value(dict_value) for dict_key, dict_value in value.items() }
ambitioninc/django-entity-event
entity_event/context_serializer.py
DefaultContextSerializer.serialize_json_string
python
def serialize_json_string(self, value): # Check if the value might be a json string if not isinstance(value, six.string_types): return value # Make sure it starts with a brace if not value.startswith('{') or value.startswith('['): return value # Try to load the string try: return json.loads(value) except: return value
Tries to load an encoded json string back into an object :param json_string: :return:
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_serializer.py#L81-L100
null
class DefaultContextSerializer(object): """ Default class for serializing context data """ def __init__(self, context): super(DefaultContextSerializer, self).__init__() self.context = context @property def data(self): """ Data property that will return the serialized data :return: """ # Create a serialized context dict serialized_context = self.serialize_value(self.context) # Return the serialized context return serialized_context def serialize_value(self, value): """ Given a value, ensure that it is serialized properly :param value: :return: """ # Create a list of serialize methods to run the value through serialize_methods = [ self.serialize_model, self.serialize_json_string, self.serialize_list, self.serialize_dict ] # Run all of our serialize methods over our value for serialize_method in serialize_methods: value = serialize_method(value) # Return the serialized context value return value def serialize_model(self, value): """ Serializes a model and all of its prefetched foreign keys :param value: :return: """ # Check if the context value is a model if not isinstance(value, models.Model): return value # Serialize the model serialized_model = model_to_dict(value) # Check the model for cached foreign keys for model_field, model_value in serialized_model.items(): model_state = value._state # Django >= 2 if hasattr(model_state, 'fields_cache'): # pragma: no cover if model_state.fields_cache.get(model_field): serialized_model[model_field] = model_state.fields_cache.get(model_field) else: # pragma: no cover # Django < 2 cache_field = '_{0}_cache'.format(model_field) if hasattr(value, cache_field): serialized_model[model_field] = getattr(value, cache_field) # Return the serialized model return self.serialize_value(serialized_model) def serialize_list(self, value): """ Ensure that all values of a list or tuple are serialized :return: """ # Check if this is a list or a tuple if not isinstance(value, (list, tuple)): return value # Loop over all the values and serialize the values return [ self.serialize_value(list_value) for list_value in value ] def serialize_dict(self, value): """ Ensure that all values of a dictionary are properly serialized :param value: :return: """ # Check if this is a dict if not isinstance(value, dict): return value # Loop over all the values and serialize them return { dict_key: self.serialize_value(dict_value) for dict_key, dict_value in value.items() }
ambitioninc/django-entity-event
entity_event/context_serializer.py
DefaultContextSerializer.serialize_list
python
def serialize_list(self, value): # Check if this is a list or a tuple if not isinstance(value, (list, tuple)): return value # Loop over all the values and serialize the values return [ self.serialize_value(list_value) for list_value in value ]
Ensure that all values of a list or tuple are serialized :return:
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_serializer.py#L102-L116
null
class DefaultContextSerializer(object): """ Default class for serializing context data """ def __init__(self, context): super(DefaultContextSerializer, self).__init__() self.context = context @property def data(self): """ Data property that will return the serialized data :return: """ # Create a serialized context dict serialized_context = self.serialize_value(self.context) # Return the serialized context return serialized_context def serialize_value(self, value): """ Given a value, ensure that it is serialized properly :param value: :return: """ # Create a list of serialize methods to run the value through serialize_methods = [ self.serialize_model, self.serialize_json_string, self.serialize_list, self.serialize_dict ] # Run all of our serialize methods over our value for serialize_method in serialize_methods: value = serialize_method(value) # Return the serialized context value return value def serialize_model(self, value): """ Serializes a model and all of its prefetched foreign keys :param value: :return: """ # Check if the context value is a model if not isinstance(value, models.Model): return value # Serialize the model serialized_model = model_to_dict(value) # Check the model for cached foreign keys for model_field, model_value in serialized_model.items(): model_state = value._state # Django >= 2 if hasattr(model_state, 'fields_cache'): # pragma: no cover if model_state.fields_cache.get(model_field): serialized_model[model_field] = model_state.fields_cache.get(model_field) else: # pragma: no cover # Django < 2 cache_field = '_{0}_cache'.format(model_field) if hasattr(value, cache_field): serialized_model[model_field] = getattr(value, cache_field) # Return the serialized model return self.serialize_value(serialized_model) def serialize_json_string(self, value): """ Tries to load an encoded json string back into an object :param json_string: :return: """ # Check if the value might be a json string if not isinstance(value, six.string_types): return value # Make sure it starts with a brace if not value.startswith('{') or value.startswith('['): return value # Try to load the string try: return json.loads(value) except: return value def serialize_dict(self, value): """ Ensure that all values of a dictionary are properly serialized :param value: :return: """ # Check if this is a dict if not isinstance(value, dict): return value # Loop over all the values and serialize them return { dict_key: self.serialize_value(dict_value) for dict_key, dict_value in value.items() }
ambitioninc/django-entity-event
entity_event/context_serializer.py
DefaultContextSerializer.serialize_dict
python
def serialize_dict(self, value): # Check if this is a dict if not isinstance(value, dict): return value # Loop over all the values and serialize them return { dict_key: self.serialize_value(dict_value) for dict_key, dict_value in value.items() }
Ensure that all values of a dictionary are properly serialized :param value: :return:
train
https://github.com/ambitioninc/django-entity-event/blob/70f50df133e42a7bf38d0f07fccc6d2890e5fd12/entity_event/context_serializer.py#L118-L133
null
class DefaultContextSerializer(object): """ Default class for serializing context data """ def __init__(self, context): super(DefaultContextSerializer, self).__init__() self.context = context @property def data(self): """ Data property that will return the serialized data :return: """ # Create a serialized context dict serialized_context = self.serialize_value(self.context) # Return the serialized context return serialized_context def serialize_value(self, value): """ Given a value, ensure that it is serialized properly :param value: :return: """ # Create a list of serialize methods to run the value through serialize_methods = [ self.serialize_model, self.serialize_json_string, self.serialize_list, self.serialize_dict ] # Run all of our serialize methods over our value for serialize_method in serialize_methods: value = serialize_method(value) # Return the serialized context value return value def serialize_model(self, value): """ Serializes a model and all of its prefetched foreign keys :param value: :return: """ # Check if the context value is a model if not isinstance(value, models.Model): return value # Serialize the model serialized_model = model_to_dict(value) # Check the model for cached foreign keys for model_field, model_value in serialized_model.items(): model_state = value._state # Django >= 2 if hasattr(model_state, 'fields_cache'): # pragma: no cover if model_state.fields_cache.get(model_field): serialized_model[model_field] = model_state.fields_cache.get(model_field) else: # pragma: no cover # Django < 2 cache_field = '_{0}_cache'.format(model_field) if hasattr(value, cache_field): serialized_model[model_field] = getattr(value, cache_field) # Return the serialized model return self.serialize_value(serialized_model) def serialize_json_string(self, value): """ Tries to load an encoded json string back into an object :param json_string: :return: """ # Check if the value might be a json string if not isinstance(value, six.string_types): return value # Make sure it starts with a brace if not value.startswith('{') or value.startswith('['): return value # Try to load the string try: return json.loads(value) except: return value def serialize_list(self, value): """ Ensure that all values of a list or tuple are serialized :return: """ # Check if this is a list or a tuple if not isinstance(value, (list, tuple)): return value # Loop over all the values and serialize the values return [ self.serialize_value(list_value) for list_value in value ]