Spaces:
No application file
No application file
File size: 19,065 Bytes
669d6a1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 | """
Robust cache key generation for financial ML data structures.
Handles numpy arrays, pandas DataFrames, and time-series data properly.
"""
import hashlib
import pickle
from pathlib import Path
from typing import Any, Optional, Tuple
import numpy as np
import pandas as pd
from loguru import logger
class CacheKeyGenerator:
"""Generate robust, collision-resistant cache keys for ML data structures."""
@staticmethod
def generate_key(func, args: tuple, kwargs: dict) -> str:
"""
Generate a robust cache key for a function call.
Args:
func: The function being cached
args: Positional arguments
kwargs: Keyword arguments
Returns:
MD5 hash string representing the unique call signature
"""
key_parts = [
func.__module__,
func.__qualname__,
]
# Process positional arguments
for i, arg in enumerate(args):
try:
key_part = CacheKeyGenerator._hash_argument(arg, f"arg_{i}")
key_parts.append(key_part)
except Exception as e:
logger.warning(f"Failed to hash argument {i} of type {type(arg)}: {e}")
# Fallback to string representation
key_parts.append(f"arg_{i}_{str(hash(str(arg)))}")
# Process keyword arguments (sorted for consistency)
for key, value in sorted(kwargs.items()):
try:
key_part = CacheKeyGenerator._hash_argument(value, key)
key_parts.append(f"{key}={key_part}")
except Exception as e:
logger.warning(
f"Failed to hash kwarg '{key}' of type {type(value)}: {e}"
)
# Fallback
key_parts.append(f"{key}={str(hash(str(value)))}")
# Combine all parts and hash
combined = "_".join(key_parts)
return hashlib.md5(combined.encode()).hexdigest()
@staticmethod
def _hash_argument(arg: Any, name: str) -> str:
"""Hash a single argument based on its type."""
try:
from sklearn.base import BaseEstimator
if isinstance(arg, BaseEstimator):
return CacheKeyGenerator._hash_sklearn_estimator(arg, name)
except ImportError:
pass # sklearn not available, continue with other types
if isinstance(arg, np.ndarray):
return CacheKeyGenerator._hash_numpy_array(arg, name)
elif isinstance(arg, pd.DataFrame):
return CacheKeyGenerator._hash_dataframe(arg, name)
elif isinstance(arg, pd.Series):
return CacheKeyGenerator._hash_series(arg, name)
elif isinstance(arg, (list, tuple)):
return CacheKeyGenerator._hash_sequence(arg, name)
elif isinstance(arg, dict):
return CacheKeyGenerator._hash_dict(arg, name)
elif isinstance(arg, (int, float, str, bool, type(None))):
return CacheKeyGenerator._hash_primitive(arg, name)
else:
# Fallback for unknown types
return CacheKeyGenerator._hash_generic(arg, name)
@staticmethod
def _hash_numpy_array(arr: np.ndarray, name: str) -> str:
"""Hash numpy array including shape, dtype, and content."""
# For large arrays, sample for performance
if arr.size > 10000:
# Hash shape, dtype, and a sample
sample = arr.flat[:: max(1, arr.size // 1000)] # Sample ~1000 points
content_hash = hashlib.md5(sample.tobytes()).hexdigest()[:8]
else:
# Hash full content for small arrays
content_hash = hashlib.md5(arr.tobytes()).hexdigest()[:8]
return f"{name}_arr_{arr.shape}_{arr.dtype}_{content_hash}"
@staticmethod
def _hash_dataframe(df: pd.DataFrame, name: str) -> str:
"""Hash pandas DataFrame including index, columns, dtypes, and content."""
parts = [
f"shape_{df.shape}",
f"cols_{hashlib.md5(str(tuple(df.columns)).encode()).hexdigest()[:8]}",
f"dtypes_{hashlib.md5(str(tuple(df.dtypes)).encode()).hexdigest()[:8]}",
]
# Hash index
if isinstance(df.index, pd.DatetimeIndex):
# For datetime index, hash start, end, and frequency
parts.append(f"idx_dt_{df.index[0]}_{df.index[-1]}_{len(df.index)}")
else:
idx_hash = hashlib.md5(str(tuple(df.index)).encode()).hexdigest()[:8]
parts.append(f"idx_{idx_hash}")
# Hash content (sample for large DataFrames)
if df.size > 10000:
# Sample rows for hashing
sample_rows = df.iloc[:: max(1, len(df) // 100)] # ~100 rows
content_hash = hashlib.md5(sample_rows.values.tobytes()).hexdigest()[:8]
else:
content_hash = hashlib.md5(df.values.tobytes()).hexdigest()[:8]
parts.append(f"data_{content_hash}")
return f"{name}_df_{'_'.join(parts)}"
@staticmethod
def _hash_series(series: pd.Series, name: str) -> str:
"""Hash pandas Series."""
parts = [
f"len_{len(series)}",
f"dtype_{series.dtype}",
]
# Hash index
if isinstance(series.index, pd.DatetimeIndex):
parts.append(f"idx_dt_{series.index[0]}_{series.index[-1]}")
else:
idx_hash = hashlib.md5(str(tuple(series.index)).encode()).hexdigest()[:8]
parts.append(f"idx_{idx_hash}")
# Hash values
if len(series) > 1000:
sample = series.iloc[:: max(1, len(series) // 100)]
content_hash = hashlib.md5(sample.values.tobytes()).hexdigest()[:8]
else:
content_hash = hashlib.md5(series.values.tobytes()).hexdigest()[:8]
parts.append(f"data_{content_hash}")
return f"{name}_series_{'_'.join(parts)}"
@staticmethod
def _hash_sequence(seq: Tuple[Any, ...] | list, name: str) -> str:
"""Hash list or tuple recursively."""
if len(seq) == 0:
return f"{name}_empty_seq"
# Hash each element
element_hashes = []
for i, item in enumerate(seq):
elem_hash = CacheKeyGenerator._hash_argument(item, f"{name}_{i}")
element_hashes.append(elem_hash)
combined = "_".join(element_hashes)
return hashlib.md5(combined.encode()).hexdigest()[:8]
@staticmethod
def _hash_dict(d: dict, name: str) -> str:
"""Hash dictionary recursively."""
if len(d) == 0:
return f"{name}_empty_dict"
# Sort keys for consistency
items_hash = []
for key, value in sorted(d.items()):
val_hash = CacheKeyGenerator._hash_argument(value, f"{name}_{key}")
items_hash.append(f"{key}={val_hash}")
combined = "_".join(items_hash)
return hashlib.md5(combined.encode()).hexdigest()[:8]
@staticmethod
def _hash_primitive(value: Any, name: str) -> str:
"""Hash primitive types."""
return f"{name}_{type(value).__name__}_{hash(value)}"
@staticmethod
def _hash_generic(obj: Any, name: str) -> str:
"""Fallback hashing for unknown types."""
try:
# Try to use object's __repr__
return f"{name}_{type(obj).__name__}_{hash(repr(obj))}"
except Exception:
# Last resort: use id
return f"{name}_{type(obj).__name__}_{id(obj)}"
@staticmethod
def _hash_sklearn_estimator(estimator: Any, name: str) -> str:
"""Hash sklearn estimator including nested estimators."""
try:
from sklearn.base import BaseEstimator
if not isinstance(estimator, BaseEstimator):
return CacheKeyGenerator._hash_generic(estimator, name)
# Use the enhanced estimator hashing from cv_cache
from .cv_cache import _hash_classifier
estimator_hash = _hash_classifier(estimator)
return f"{name}_estimator_{estimator_hash}"
except ImportError:
# Fallback if sklearn not available
return CacheKeyGenerator._hash_generic(estimator, name)
class TimeSeriesCacheKey(CacheKeyGenerator):
"""
Extended cache key generator with time-series awareness.
Useful for financial data where lookback periods matter.
"""
@staticmethod
def generate_key_with_time_range(
func,
args: tuple,
kwargs: dict,
time_range: Tuple[pd.Timestamp, pd.Timestamp] = None,
) -> str:
"""
Generate cache key that includes time range information.
Args:
func: Function being cached
args: Positional arguments
kwargs: Keyword arguments
time_range: Optional (start, end) timestamp tuple
Returns:
Cache key string
"""
base_key = CacheKeyGenerator.generate_key(func, args, kwargs)
if time_range is None:
# Try to extract time range from data
time_range = TimeSeriesCacheKey._extract_time_range(args, kwargs)
if time_range:
start, end = time_range
time_hash = f"time_{start}_{end}"
return f"{base_key}_{time_hash}"
return base_key
@staticmethod
def _extract_time_range(
args: tuple, kwargs: dict
) -> Tuple[pd.Timestamp, pd.Timestamp] | None:
"""
Attempt to extract time range from function arguments.
Looks for DataFrames with DatetimeIndex or explicit start/end parameters.
"""
# Check kwargs for explicit time parameters
if "start_date" in kwargs and "end_date" in kwargs:
return (
pd.Timestamp(kwargs["start_date"]),
pd.Timestamp(kwargs["end_date"]),
)
# Check for DataFrames with DatetimeIndex in args
for arg in args:
if isinstance(arg, pd.DataFrame) and isinstance(
arg.index, pd.DatetimeIndex
):
if len(arg.index) > 0:
return (arg.index[0], arg.index[-1])
elif isinstance(arg, pd.Series) and isinstance(arg.index, pd.DatetimeIndex):
if len(arg.index) > 0:
return (arg.index[0], arg.index[-1])
return None
# =============================================================================
# Integration with existing cacheable decorator
# =============================================================================
def create_robust_cacheable(
track_data_access: bool = False,
dataset_name: Optional[str] = None,
purpose: Optional[str] = None,
use_time_awareness: bool = False,
):
"""
Factory function to create robust cacheable decorators with data tracking.
Args:
track_data_access: Whether to track DataFrame accesses
dataset_name: Name of the dataset for tracking
purpose: One of 'train', 'test', 'validate', 'optimize', 'analyze'
use_time_awareness: Whether to use time-series aware cache keys
Returns:
Decorator function
"""
import time
from functools import wraps
from . import cache_stats, memory
from .cache_monitoring import get_cache_monitor
def decorator(func):
func_name = f"{func.__module__}.{func.__qualname__}"
cached_func = memory.cache(func)
seen_signatures = set()
monitor = get_cache_monitor()
@wraps(func)
def wrapper(*args, **kwargs):
nonlocal seen_signatures
# Track access time (ALWAYS do this first)
monitor.track_access(func_name)
# Generate cache key
cache_key = None
is_hit = False
computation_start = None
try:
if use_time_awareness:
cache_key = TimeSeriesCacheKey.generate_key_with_time_range(
func, args, kwargs
)
else:
cache_key = CacheKeyGenerator.generate_key(func, args, kwargs)
# Track hit/miss
try:
cached_func.check_call_in_cache(*args, **kwargs)
is_hit = True
cache_stats.record_hit(func_name)
logger.debug(f"Cache HIT for {func_name}")
except:
cache_stats.record_miss(func_name)
is_hit = False
computation_start = time.time() # Start timing for misses
logger.debug(f"Cache MISS for {func_name}")
# Add to seen_signatures for this session
seen_signatures.add(cache_key)
except Exception as e:
logger.warning(f"Cache key generation failed for {func_name}: {e}")
cache_stats.record_miss(func_name)
cache_key = None
is_hit = False
computation_start = time.time() # Start timing for error case
# Track data access if requested
if track_data_access:
try:
from .data_access_tracker import get_data_tracker
_track_dataframe_access(
get_data_tracker(), args, kwargs, dataset_name, purpose
)
except Exception as e:
logger.warning(f"Data tracking failed for {func_name}: {e}")
# Execute function
try:
if is_hit:
# For cache hits, just return cached result (no timing needed)
result = cached_func(*args, **kwargs)
else:
# For cache misses, time the computation
result = cached_func(*args, **kwargs)
if computation_start:
computation_time = time.time() - computation_start
monitor.track_computation_time(func_name, computation_time)
logger.debug(
f"Computation time for {func_name}: {computation_time:.3f}s"
)
return result
except (EOFError, pickle.PickleError, OSError) as e:
# Handle cache corruption
logger.warning(
f"Cache corruption for {func_name}: {type(e).__name__} - recomputing"
)
# Clear corrupted cache if possible
if cache_key is not None:
_clear_corrupted_cache(cached_func, args, kwargs, func_name)
# Execute function directly and track time
direct_start = time.time()
result = func(*args, **kwargs)
if computation_start: # Track time if it was originally a miss
computation_time = time.time() - direct_start
monitor.track_computation_time(func_name, computation_time)
logger.debug(
f"Direct computation time for {func_name}: {computation_time:.3f}s"
)
return result
except Exception as e:
# Other unexpected errors
logger.error(f"Unexpected cache error for {func_name}: {e}")
raise
# Add cache info method for debugging
def cache_info():
return {
"function_name": func_name,
"seen_signatures": len(seen_signatures),
"hits": cache_stats._stats.get(func_name, {}).get("hits", 0),
"misses": cache_stats._stats.get(func_name, {}).get("misses", 0),
}
wrapper.cache_info = cache_info
wrapper._afml_cacheable = True
return wrapper
return decorator
def _clear_corrupted_cache(cached_func, args, kwargs, func_name):
"""Helper to clear corrupted cache entries."""
try:
if hasattr(cached_func, "_get_cache_id"):
joblib_cache_key = cached_func._get_cache_id(*args, **kwargs)
cache_dir = Path(cached_func.store_backend.location)
# Remove files matching this cache key
removed_count = 0
for cache_file in cache_dir.rglob("*"):
if cache_file.is_file() and str(joblib_cache_key) in str(cache_file):
cache_file.unlink()
removed_count += 1
logger.debug(f"Removed corrupted file: {cache_file.name}")
if removed_count > 0:
logger.info(
f"Cleared {removed_count} corrupted cache files for {func_name}"
)
except Exception as clear_exc:
logger.warning(f"Failed to clear corrupted cache for {func_name}: {clear_exc}")
def _track_dataframe_access(tracker, args, kwargs, dataset_name, purpose):
"""Track DataFrame accesses for data hygiene monitoring."""
# Check all arguments for DataFrames with DatetimeIndex
for i, arg in enumerate(args):
if _is_trackable_dataframe(arg):
_log_dataframe_access(tracker, arg, dataset_name or f"arg_{i}", purpose)
for key, value in kwargs.items():
if _is_trackable_dataframe(value):
_log_dataframe_access(tracker, value, dataset_name or key, purpose)
def _is_trackable_dataframe(obj):
"""Check if object is a DataFrame with temporal index."""
return (
isinstance(obj, pd.DataFrame)
and isinstance(obj.index, pd.DatetimeIndex)
and len(obj) > 0
)
def _log_dataframe_access(tracker, df, name, purpose):
"""Log DataFrame access to tracker."""
tracker.log_access(
dataset_name=name,
start_date=df.index[0],
end_date=df.index[-1],
purpose=purpose or "unknown",
data_shape=df.shape,
)
# =============================================================================
# Final convenience exports
# =============================================================================
# Standard decorators (backward compatible)
robust_cacheable = create_robust_cacheable(use_time_awareness=False)
time_aware_cacheable = create_robust_cacheable(use_time_awareness=True)
# Data tracking decorators (new functionality)
data_tracking_cacheable = lambda dataset_name, purpose: create_robust_cacheable(
track_data_access=True,
dataset_name=dataset_name,
purpose=purpose,
use_time_awareness=False,
)
time_aware_data_tracking_cacheable = (
lambda dataset_name, purpose: create_robust_cacheable(
track_data_access=True,
dataset_name=dataset_name,
purpose=purpose,
use_time_awareness=True,
)
)
__all__ = [
"CacheKeyGenerator",
"TimeSeriesCacheKey",
"data_tracking_cacheable", # NEW
"robust_cacheable", # Backward compatible
"time_aware_cacheable", # Backward compatible
"time_aware_data_tracking_cacheable", # NEW
]
|