from enum import IntEnum from pathlib import Path import numpy as np from h5py import Dataset, File as H5File, Group class StatisticType(IntEnum): AVERAGE = 0 TOTAL = 1 MIN = 2 MAX = 3 class ImageMetadata: name: str group_path: str extents: np.ndarray origin: np.ndarray shape: tuple[int] def __init__(self, name: str, group_path: str): self.name = name self.group_path = group_path def get_dataset(self, hdf5File: H5File, time_index: int) -> Dataset: image_group: Group = hdf5File[self.group_path] image_ds: Dataset = image_group[f'time{time_index:06d}'] return image_ds def read(self, f: H5File): image_group: Group = f[self.group_path] # get attributes from the group extents_list = [] origin_list = [] if "ExtentX" in image_group.attrs: extents_list.append(image_group.attrs['ExtentX']) if "ExtentY" in image_group.attrs: extents_list.append(image_group.attrs['ExtentY']) if "ExtentZ" in image_group.attrs: extents_list.append(image_group.attrs['ExtentZ']) if "OriginX" in image_group.attrs: origin_list.append(image_group.attrs['OriginX']) if "OriginY" in image_group.attrs: origin_list.append(image_group.attrs['OriginY']) if "OriginZ" in image_group.attrs: origin_list.append(image_group.attrs['OriginZ']) self.extents = np.array(extents_list) self.origin = np.array(origin_list) self.shape = self.get_dataset(f, 0).shape class VariableInfo: var_index: int var_name: str # e.g. "C_cyt" stat_channel: int statistic_type: StatisticType # e.g. StatisticType.AVERAGE stat_var_name: str # e.g. "C_cyt_average" stat_var_unit: str # e.g. "uM" def __init__(self, stat_var_name: str, stat_var_unit: str, stat_channel: int, var_index: int): self.stat_var_name = stat_var_name self.stat_var_unit = stat_var_unit self.stat_channel = stat_channel self.var_index = var_index # stat_var_name is in the form of "C_cyt_average" so remove _average to get the variable name stat_type_raw = stat_var_name.split("_")[-1] self.statistic_type = StatisticType[stat_type_raw.upper()] self.var_name = stat_var_name.replace("_"+stat_type_raw, "") class PostProcessing: postprocessing_hdf5_path: Path times: np.ndarray variables: list[VariableInfo] statistics: np.ndarray # shape (times, vars, stats) where status is average=0, total=1, min=2, max=3 image_metadata: list[ImageMetadata] def __init__(self, postprocessing_hdf5_path: Path): self.postprocessing_hdf5_path = postprocessing_hdf5_path self.variables = [] self.image_metadata = [] def read(self) -> None: # read the file as hdf5 with H5File(self.postprocessing_hdf5_path, 'r') as f: # read dataset with path /PostProcessing/Times times_ds: Dataset = f['/PostProcessing/Times'] # read array from times dataset into a ndarray self.times = times_ds[()] # read attributes from group /PostProcessing/VariableStatistics # data is flat, so we can read the attributes directly, so name and units for each channel are separate # # key=comp_0_name, value=b'C_cyt_average' # key=comp_0_unit, value=b'uM' # key=comp_1_name, value=b'C_cyt_total' # key=comp_1_unit, value=b'molecules' # key=comp_2_name, value=b'C_cyt_min' # key=comp_2_unit, value=b'uM' # key=comp_3_name, value=b'C_cyt_max' # key=comp_3_unit, value=b'uM' # var_stats_grp = f['/PostProcessing/VariableStatistics'] # gather stat_var_name and stat_var_unit for each channel into dictionaries by channel var_name_by_channel: dict[int, str] = {} var_unit_by_channel: dict[int, str] = {} for k, v in var_stats_grp.attrs.items(): parts = k.split('_') channel = int(parts[1]) value = v.decode('utf-8') if parts[2] == "name": var_name_by_channel[channel] = value elif parts[2] == "unit": var_unit_by_channel[channel] = value # combine into a single list of VariableInfo objects, one for each channel self.variables = [VariableInfo(stat_var_name=var_name_by_channel[i], stat_var_unit=var_unit_by_channel[i], stat_channel=i, var_index=i//4) for i in range(len(var_name_by_channel))] # within /PostProcessing/VariableStatistics, there are datasets for each time point # PostProcessing/VariableStatistics # PostProcessing/VariableStatistics/time000000 # PostProcessing/VariableStatistics/time000001 # PostProcessing/VariableStatistics/time000002 # PostProcessing/VariableStatistics/time000003 # PostProcessing/VariableStatistics/time000004 # we can read the data for each time point into a list of ndarrays statistics_raw: np.ndarray = np.zeros((len(self.times), len(self.variables))) for time_index in range(len(self.times)): time_ds: Dataset = var_stats_grp[f'time{time_index:06d}'] statistics_raw[time_index, :] = time_ds[()] # reshape the statistics_raw into a 3D array (times, vars, stats) self.statistics = statistics_raw.reshape((len(self.times), len(self.variables)//4, 4)) # get list of child groups from /PostProcessing which are not Times or VariableStatistics # e.g. /PostProcessing/fluor image_groups = [k for k in f['/PostProcessing'].keys() if k not in ['Times', 'VariableStatistics']] # for each image group, read the metadata to allow reading later for image_group in image_groups: metadata = ImageMetadata(group_path=f"/PostProcessing/{image_group}", name=image_group) metadata.read(f) self.image_metadata.append(metadata) def read_image_data(self, image_metadata: ImageMetadata, time_index: int) -> np.ndarray: with H5File(self.postprocessing_hdf5_path, 'r') as f: image_ds: Dataset = image_metadata.get_dataset(hdf5File=f, time_index=time_index) return image_ds[()]