Spaces:
Sleeping
Sleeping
Update persist.py
Browse files- persist.py +40 -49
persist.py
CHANGED
|
@@ -11,7 +11,7 @@ from typing import Any
|
|
| 11 |
BASE_DIR = "./data/store"
|
| 12 |
os.makedirs(BASE_DIR, exist_ok=True)
|
| 13 |
|
| 14 |
-
# TTL validity map
|
| 15 |
VALIDITY_MAP = {
|
| 16 |
"result": {"days": 7},
|
| 17 |
"qresult": {"days": 7},
|
|
@@ -22,7 +22,7 @@ VALIDITY_MAP = {
|
|
| 22 |
}
|
| 23 |
|
| 24 |
# ==============================
|
| 25 |
-
#
|
| 26 |
# ==============================
|
| 27 |
def _ts():
|
| 28 |
return datetime.now().strftime("%Y_%m_%d_%H_%M_%S")
|
|
@@ -34,7 +34,6 @@ def _list_files():
|
|
| 34 |
return os.listdir(BASE_DIR)
|
| 35 |
|
| 36 |
def _latest(prefix: str, ext: str):
|
| 37 |
-
"""Return latest file with prefix + ext"""
|
| 38 |
files = [
|
| 39 |
f for f in _list_files()
|
| 40 |
if f.startswith(prefix + "_") and f.endswith("." + ext)
|
|
@@ -51,7 +50,8 @@ def save(name: str, data: Any, ftype: str) -> bool:
|
|
| 51 |
try:
|
| 52 |
if ftype == "csv":
|
| 53 |
if not isinstance(data, pd.DataFrame):
|
| 54 |
-
|
|
|
|
| 55 |
data.to_csv(path, index=False)
|
| 56 |
elif ftype == "json":
|
| 57 |
with open(path, "w", encoding="utf-8") as f:
|
|
@@ -63,72 +63,80 @@ def save(name: str, data: Any, ftype: str) -> bool:
|
|
| 63 |
with open(path, "wb") as f:
|
| 64 |
pickle.dump(data, f)
|
| 65 |
else:
|
| 66 |
-
|
|
|
|
|
|
|
| 67 |
return True
|
| 68 |
-
except Exception:
|
|
|
|
| 69 |
return False
|
| 70 |
|
| 71 |
# ==============================
|
| 72 |
# Load function
|
| 73 |
# ==============================
|
| 74 |
def load(name: str, ftype: str):
|
| 75 |
-
"""
|
| 76 |
-
Load file by:
|
| 77 |
-
1) full filename (name_YYYY_MM_DD_HH_MM_SS.ext)
|
| 78 |
-
2) base name + ftype (latest)
|
| 79 |
-
Returns False if file not present
|
| 80 |
-
"""
|
| 81 |
filename = _latest(name, ftype) if "." not in name else name
|
| 82 |
path = _path(filename)
|
| 83 |
if not os.path.exists(path):
|
|
|
|
| 84 |
return False
|
| 85 |
try:
|
| 86 |
if filename.endswith(".csv"):
|
| 87 |
-
|
| 88 |
-
|
| 89 |
with open(path, "r", encoding="utf-8") as f:
|
| 90 |
-
|
| 91 |
-
|
| 92 |
with open(path, "r", encoding="utf-8") as f:
|
| 93 |
-
|
| 94 |
-
|
| 95 |
with open(path, "rb") as f:
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
return False
|
| 100 |
|
| 101 |
# ==============================
|
| 102 |
# Exists with TTL
|
| 103 |
# ==============================
|
| 104 |
def exists(name: str, ftype: str) -> bool:
|
| 105 |
-
"""
|
| 106 |
-
Check if file exists and is within TTL defined in VALIDITY_MAP
|
| 107 |
-
Returns True/False
|
| 108 |
-
"""
|
| 109 |
filename = _latest(name, ftype)
|
| 110 |
if not filename:
|
|
|
|
| 111 |
return False
|
| 112 |
|
| 113 |
path = _path(filename)
|
| 114 |
if not os.path.exists(path):
|
|
|
|
| 115 |
return False
|
| 116 |
|
| 117 |
mtime = datetime.fromtimestamp(os.path.getmtime(path))
|
| 118 |
func_name = name.split("_")[0]
|
| 119 |
validity = VALIDITY_MAP.get(func_name)
|
| 120 |
if not validity:
|
| 121 |
-
|
|
|
|
| 122 |
|
| 123 |
now = datetime.now()
|
|
|
|
| 124 |
if "minutes" in validity:
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
|
| 131 |
-
return
|
| 132 |
|
| 133 |
# ==============================
|
| 134 |
# Utilities
|
|
@@ -140,20 +148,3 @@ def list_files(name=None, ftype=None):
|
|
| 140 |
if ftype:
|
| 141 |
files = [f for f in files if f.endswith("." + ftype)]
|
| 142 |
return files
|
| 143 |
-
'''
|
| 144 |
-
import pandas as pd
|
| 145 |
-
from hf_persistence import save, load, exists
|
| 146 |
-
|
| 147 |
-
# Save DataFrame
|
| 148 |
-
df = pd.DataFrame({"A":[1,2], "B":[3,4]})
|
| 149 |
-
save("intraday_RELIANCE", df, "csv")
|
| 150 |
-
|
| 151 |
-
# Check existence with TTL
|
| 152 |
-
if exists("intraday_RELIANCE", "csv"):
|
| 153 |
-
df_cached = load("intraday_RELIANCE", "csv")
|
| 154 |
-
|
| 155 |
-
# Save HTML
|
| 156 |
-
save("intraday_RELIANCE", "<h1>Hello</h1>", "html")
|
| 157 |
-
if exists("intraday_RELIANCE", "html"):
|
| 158 |
-
html_cached = load("intraday_RELIANCE", "html")
|
| 159 |
-
'''
|
|
|
|
| 11 |
BASE_DIR = "./data/store"
|
| 12 |
os.makedirs(BASE_DIR, exist_ok=True)
|
| 13 |
|
| 14 |
+
# TTL validity map
|
| 15 |
VALIDITY_MAP = {
|
| 16 |
"result": {"days": 7},
|
| 17 |
"qresult": {"days": 7},
|
|
|
|
| 22 |
}
|
| 23 |
|
| 24 |
# ==============================
|
| 25 |
+
# Helpers
|
| 26 |
# ==============================
|
| 27 |
def _ts():
|
| 28 |
return datetime.now().strftime("%Y_%m_%d_%H_%M_%S")
|
|
|
|
| 34 |
return os.listdir(BASE_DIR)
|
| 35 |
|
| 36 |
def _latest(prefix: str, ext: str):
|
|
|
|
| 37 |
files = [
|
| 38 |
f for f in _list_files()
|
| 39 |
if f.startswith(prefix + "_") and f.endswith("." + ext)
|
|
|
|
| 50 |
try:
|
| 51 |
if ftype == "csv":
|
| 52 |
if not isinstance(data, pd.DataFrame):
|
| 53 |
+
print(f"[SAVE FAILED] CSV requires pandas DataFrame for {filename}")
|
| 54 |
+
return False
|
| 55 |
data.to_csv(path, index=False)
|
| 56 |
elif ftype == "json":
|
| 57 |
with open(path, "w", encoding="utf-8") as f:
|
|
|
|
| 63 |
with open(path, "wb") as f:
|
| 64 |
pickle.dump(data, f)
|
| 65 |
else:
|
| 66 |
+
print(f"[SAVE FAILED] Unsupported file type: {ftype} for {filename}")
|
| 67 |
+
return False
|
| 68 |
+
print(f"[SAVE OK] {filename}")
|
| 69 |
return True
|
| 70 |
+
except Exception as e:
|
| 71 |
+
print(f"[SAVE FAILED] {filename} - Exception: {e}")
|
| 72 |
return False
|
| 73 |
|
| 74 |
# ==============================
|
| 75 |
# Load function
|
| 76 |
# ==============================
|
| 77 |
def load(name: str, ftype: str):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
filename = _latest(name, ftype) if "." not in name else name
|
| 79 |
path = _path(filename)
|
| 80 |
if not os.path.exists(path):
|
| 81 |
+
print(f"[LOAD FAILED] File does not exist: {filename}")
|
| 82 |
return False
|
| 83 |
try:
|
| 84 |
if filename.endswith(".csv"):
|
| 85 |
+
df = pd.read_csv(path)
|
| 86 |
+
elif filename.endswith(".json"):
|
| 87 |
with open(path, "r", encoding="utf-8") as f:
|
| 88 |
+
df = json.load(f)
|
| 89 |
+
elif filename.endswith(".html"):
|
| 90 |
with open(path, "r", encoding="utf-8") as f:
|
| 91 |
+
df = f.read()
|
| 92 |
+
elif filename.endswith(".pkl"):
|
| 93 |
with open(path, "rb") as f:
|
| 94 |
+
df = pickle.load(f)
|
| 95 |
+
else:
|
| 96 |
+
print(f"[LOAD FAILED] Unsupported file type: {filename}")
|
| 97 |
+
return False
|
| 98 |
+
print(f"[LOAD OK] {filename}")
|
| 99 |
+
return df
|
| 100 |
+
except Exception as e:
|
| 101 |
+
print(f"[LOAD FAILED] {filename} - Exception: {e}")
|
| 102 |
return False
|
| 103 |
|
| 104 |
# ==============================
|
| 105 |
# Exists with TTL
|
| 106 |
# ==============================
|
| 107 |
def exists(name: str, ftype: str) -> bool:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 108 |
filename = _latest(name, ftype)
|
| 109 |
if not filename:
|
| 110 |
+
print(f"[EXISTS] No file found for {name}.{ftype}")
|
| 111 |
return False
|
| 112 |
|
| 113 |
path = _path(filename)
|
| 114 |
if not os.path.exists(path):
|
| 115 |
+
print(f"[EXISTS] File not present: {filename}")
|
| 116 |
return False
|
| 117 |
|
| 118 |
mtime = datetime.fromtimestamp(os.path.getmtime(path))
|
| 119 |
func_name = name.split("_")[0]
|
| 120 |
validity = VALIDITY_MAP.get(func_name)
|
| 121 |
if not validity:
|
| 122 |
+
print(f"[EXISTS] File exists with no TTL: {filename}")
|
| 123 |
+
return True
|
| 124 |
|
| 125 |
now = datetime.now()
|
| 126 |
+
is_valid = True
|
| 127 |
if "minutes" in validity:
|
| 128 |
+
is_valid = (now - mtime) <= timedelta(minutes=validity["minutes"])
|
| 129 |
+
elif "hours" in validity:
|
| 130 |
+
is_valid = (now - mtime) <= timedelta(hours=validity["hours"])
|
| 131 |
+
elif "days" in validity:
|
| 132 |
+
is_valid = (now - mtime) <= timedelta(days=validity["days"])
|
| 133 |
+
|
| 134 |
+
if is_valid:
|
| 135 |
+
print(f"[EXISTS] File exists and valid: {filename}")
|
| 136 |
+
else:
|
| 137 |
+
print(f"[EXISTS] File expired: {filename}")
|
| 138 |
|
| 139 |
+
return is_valid
|
| 140 |
|
| 141 |
# ==============================
|
| 142 |
# Utilities
|
|
|
|
| 148 |
if ftype:
|
| 149 |
files = [f for f in files if f.endswith("." + ftype)]
|
| 150 |
return files
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|