File size: 16,945 Bytes
1cd137c | 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 | #!/usr/bin/env python3
"""
Phase 1 — Data Quality / Cleaning / Validation / Provenance
Compliments Reference DB Pipeline
Authoritative Input:
https://huggingface.co/datasets/saraNour/compliments-brand/blob/main/source_of_truth/products.parquet
This phase:
1. Downloads the authoritative products.parquet from HuggingFace
2. Validates schema, row count, nulls, duplicates
3. Performs deterministic cleaning/standardization
4. Preserves raw values for traceability where cleaning modifies data
5. Analyzes UPC patterns, brand values, size fields
6. Documents provenance of every column
7. Drops 100% null columns with explicit documentation
8. Produces clean Phase 1 output + validation + statistics
Cleaning operations (Phase 1 = data quality foundation):
A. String normalization: whitespace trim, empty-to-null
B. Brand cleaning: whitespace/case normalization, preserve raw
C. UPC validation: format, nulls, duplicates, reused UPCs
D. external_id validation: nulls, duplicates
E. Title cleaning: whitespace normalization
F. Size fields: consistency audit (semantic parsing is Phase 2)
G. Null column handling: documented removal of 100% null columns
"""
import json
import re
from datetime import UTC, datetime
from pathlib import Path
import pandas as pd
from huggingface_hub import hf_hub_download
# Configuration
HF_REPO = "saraNour/compliments-brand"
HF_FILE = "source_of_truth/products.parquet"
HF_REPO_TYPE = "dataset"
OUTPUT_DIR = Path(__file__).resolve().parent.parent / "outputs"
VALIDATION_DIR = Path(__file__).resolve().parent.parent / "validation"
STATISTICS_DIR = Path(__file__).resolve().parent.parent / "statistics"
VERSION = "3.0.0"
TIMESTAMP = datetime.now(UTC).isoformat()
def log(msg: str) -> None:
print(f"[Phase1] {msg}")
# A. STRING NORMALIZATION
def normalize_whitespace(s):
"""Trim leading/trailing whitespace, collapse repeated internal whitespace."""
if pd.isna(s):
return s
s_str = str(s).strip()
s_str = re.sub(r"\s+", " ", s_str)
return s_str if s_str else None
def empty_to_null(s):
"""Convert empty/whitespace-only strings to None."""
if pd.isna(s):
return None
s_str = str(s).strip()
return None if s_str == "" else s_str
# B. BRAND CLEANING
def clean_brand(raw_brand):
"""
Clean brand string: trim whitespace, normalize case for downstream.
Returns (brand_clean, brand_raw).
"""
if pd.isna(raw_brand):
return None, None
raw = str(raw_brand)
cleaned = raw.strip()
# Normalize obvious case noise: "COMPLIMENTS" -> "Compliments"
# But preserve mixed case that might be intentional
if cleaned.upper() == cleaned and len(cleaned) > 1:
cleaned = cleaned.title()
return cleaned, raw
# C. UPC VALIDATION
def validate_upc(upc_val):
"""
Validate a single UPC value.
Returns dict with validation results.
"""
if pd.isna(upc_val):
return {"valid": False, "reason": "null"}
s = str(upc_val).strip()
if s == "" or s == "nan":
return {"valid": False, "reason": "empty"}
# Check for non-digit characters (allow decimal point for float representation)
# UPCs stored as floats may have .0 suffix
s_clean = s.replace(".0", "") if s.endswith(".0") else s
if not s_clean.isdigit():
return {"valid": False, "reason": f"non_digit_chars: {s}"}
# Check length (standard UPC is 12 digits, but variants exist)
if len(s_clean) < 6 or len(s_clean) > 14:
return {"valid": False, "reason": f"unusual_length: {len(s_clean)}"}
return {"valid": True, "cleaned": s_clean}
def audit_upcs(df):
"""Comprehensive UPC audit."""
upc = df["upc"]
# Validate each UPC
validations = upc.apply(validate_upc)
valid_mask = validations.apply(lambda x: x["valid"])
invalid_upcs = df[~valid_mask].copy()
# Count reused UPCs
upc_counts = upc.value_counts()
reused = upc_counts[upc_counts > 1]
# Analyze reused UPCs: do they map to different titles?
reused_analysis = []
for upc_val in reused.index:
if pd.isna(upc_val):
continue
subset = df[df["upc"] == upc_val]
titles = subset["title"].unique()
brands = subset["brand"].unique()
sizes = subset["size"].unique()
reused_analysis.append({
"upc": str(upc_val),
"count": int(reused[upc_val]),
"unique_titles": len(titles),
"titles": [str(t) for t in titles[:5]],
"unique_brands": len(brands),
"brands": [str(b) for b in brands],
"unique_sizes": len(sizes),
"sizes": [str(s) for s in sizes[:5]],
})
return {
"total_rows": len(df),
"null_count": int(upc.isna().sum()),
"unique_count": int(upc.nunique()),
"invalid_format_count": int((~valid_mask).sum()),
"invalid_format_examples": invalid_upcs[["external_id", "title", "upc"]].head(10).to_dict("records"),
"reused_upc_count": len(reused),
"reused_upc_total_rows": int(reused.sum()),
"reused_upc_examples": reused_analysis[:15],
}
# D. EXTERNAL_ID VALIDATION
def audit_external_ids(df):
"""Validate external_id field."""
ext = df["external_id"]
return {
"total_rows": len(df),
"null_count": int(ext.isna().sum()),
"unique_count": int(ext.nunique()),
"duplicate_count": int(ext.duplicated().sum()),
"pass": int(ext.duplicated().sum()) == 0 and ext.isna().sum() == 0,
}
# E. TITLE CLEANING
def clean_title(raw_title):
"""
Clean title: trim whitespace, collapse repeated spaces.
Returns (title_clean, title_raw).
"""
if pd.isna(raw_title):
return None, None
raw = str(raw_title)
cleaned = raw.strip()
cleaned = re.sub(r"\s+", " ", cleaned)
return cleaned if cleaned else None, raw
def audit_titles(df):
"""Audit title field quality."""
titles = df["title"]
# Check for leading/trailing whitespace
has_leading = titles.apply(lambda x: str(x) != str(x).strip() if pd.notna(x) else False).sum()
has_repeated_ws = titles.apply(lambda x: bool(re.search(r"\s{2,}", str(x))) if pd.notna(x) else False).sum()
empty_titles = titles.isna().sum() + (titles.apply(lambda x: str(x).strip() == "" if pd.notna(x) else False).sum())
# Duplicated titles
title_counts = titles.value_counts()
duplicated_titles = title_counts[title_counts > 1]
return {
"total_rows": len(titles),
"null_count": int(titles.isna().sum()),
"unique_count": int(titles.nunique()),
"leading_trailing_whitespace": int(has_leading),
"repeated_whitespace": int(has_repeated_ws),
"empty_titles": int(empty_titles),
"duplicated_title_count": len(duplicated_titles),
"duplicated_title_total_rows": int(duplicated_titles.sum()),
"duplicated_title_examples": {str(k): int(v) for k, v in list(duplicated_titles.head(10).items())},
}
# F. SIZE FIELDS AUDIT
def audit_sizes(df):
"""Audit size-related fields for consistency."""
size_str = df["size"]
size_amount = df["size_amount"]
size_unit = df["size_unit"]
size_unit_norm = df["size_unit_norm"]
size_qty = df["size_qty"]
# Check: when size_amount is null, is size_unit also null?
amount_null = size_amount.isna()
unit_null = size_unit.isna()
both_null = (amount_null & unit_null).sum()
amount_null_unit_not = (amount_null & ~unit_null).sum()
unit_null_amount_not = (~amount_null & unit_null).sum()
# Check: size_qty should usually be 1
qty_distribution = size_qty.value_counts().to_dict()
return {
"size_string": {
"null_count": int(size_str.isna().sum()),
"unique_count": int(size_str.nunique()),
},
"size_amount": {
"null_count": int(amount_null.sum()),
"null_pct": round(amount_null.sum() / len(df) * 100, 2),
},
"size_unit": {
"null_count": int(unit_null.sum()),
"null_pct": round(unit_null.sum() / len(df) * 100, 2),
"distribution": {str(k): int(v) for k, v in size_unit.value_counts().items()},
},
"size_unit_norm": {
"null_count": int(size_unit_norm.isna().sum()),
"distribution": {str(k): int(v) for k, v in size_unit_norm.value_counts().items()},
},
"consistency": {
"both_amount_and_unit_null": int(both_null),
"amount_null_unit_not_null": int(amount_null_unit_not),
"unit_null_amount_null_not": int(unit_null_amount_not),
},
"size_qty": {
"distribution": {str(k): int(v) for k, v in qty_distribution.items()},
},
}
# G. NULL COLUMN HANDLING
def identify_null_columns(df):
"""Identify columns that are 100% null."""
null_cols = []
for col in df.columns:
if df[col].isna().all():
null_cols.append(col)
return null_cols
# H. DUPLICATE AUDIT
def audit_duplicates(df):
"""Comprehensive duplicate audit."""
full_dupes = int(df.duplicated().sum())
ext_dupes = int(df["external_id"].duplicated().sum())
upc_dupes = int(df["upc"].duplicated().sum())
# Title + size + brand combinations
if "size" in df.columns:
combo = df["title"].fillna("") + "|" + df["size"].fillna("") + "|" + df["brand"].fillna("")
combo_dupes = int(combo.duplicated().sum())
else:
combo_dupes = 0
return {
"full_row_duplicates": full_dupes,
"external_id_duplicates": ext_dupes,
"upc_duplicates": upc_dupes,
"title_size_brand_duplicates": combo_dupes,
}
# MAIN
def main():
log("Starting Phase 1 (v2.0.0 — Data Quality Foundation)")
# Ensure output dirs exist
OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
VALIDATION_DIR.mkdir(parents=True, exist_ok=True)
STATISTICS_DIR.mkdir(parents=True, exist_ok=True)
# 1. LOAD
log(f"Downloading {HF_REPO}/{HF_FILE} ...")
path = hf_hub_download(HF_REPO, HF_FILE, repo_type=HF_REPO_TYPE)
df_raw = pd.read_parquet(path)
log(f"Loaded: {df_raw.shape[0]} rows, {df_raw.shape[1]} columns")
# 2. SCHEMA VALIDATION
log("Validating schema ...")
EXPECTED_COLUMNS = [
"upc", "external_id", "brand", "title", "price", "price_currency",
"size", "size_amount", "size_unit", "size_qty", "size_per_unit",
"size_unit_norm", "size_total", "image_url", "source", "source_url",
]
missing = [c for c in EXPECTED_COLUMNS if c not in df_raw.columns]
assert len(missing) == 0, f"Missing columns: {missing}"
assert len(df_raw) == 4440, f"Expected 4440 rows, got {len(df_raw)}"
log(" Schema: PASS")
# 3. AUDIT (before cleaning)
log("Auditing raw data ...")
null_info = {}
for col in df_raw.columns:
n = int(df_raw[col].isna().sum())
null_info[col] = {
"null_count": n,
"null_pct": round(n / len(df_raw) * 100, 2),
"is_100pct_null": n == len(df_raw),
}
dup_info_raw = audit_duplicates(df_raw)
size_info_raw = audit_sizes(df_raw)
# 4. CLEANING
log("Cleaning data ...")
df = df_raw.copy()
# A. String normalization on all string columns
log(" A. Normalizing whitespace on string columns ...")
string_cols = ["upc", "external_id", "brand", "title", "size",
"size_unit", "size_unit_norm", "image_url", "source", "source_url"]
cleaning_log = {}
for col in string_cols:
if col in df.columns:
before_nulls = int(df[col].isna().sum())
df[col] = df[col].apply(empty_to_null)
after_nulls = int(df[col].isna().sum())
new_nulls = after_nulls - before_nulls
if new_nulls > 0:
cleaning_log[col] = {"empty_to_null_count": new_nulls}
# B. Brand cleaning
log(" B. Cleaning brand field ...")
brand_clean_results = df["brand"].apply(clean_brand)
df["brand_clean"] = brand_clean_results.apply(lambda x: x[0])
log(f" Cleaned brands: {df['brand_clean'].nunique()} unique")
# C. Title cleaning
log(" C. Cleaning title field ...")
title_clean_results = df["title"].apply(clean_title)
df["title_clean"] = title_clean_results.apply(lambda x: x[0])
# D. Generate source_product_id (composite identifier)
log(" D. Generating source_product_id ...")
df["source_product_id"] = df.apply(
lambda row: f"{row['source']}|{row['upc']}|{row['external_id']}"
if pd.notna(row['source']) and pd.notna(row['upc']) and pd.notna(row['external_id'])
else f"{row['source']}|{row['external_id']}",
axis=1
)
log(f" source_product_id unique: {df['source_product_id'].nunique()}")
# E. Drop 100% null columns
null_100pct = [c for c, v in null_info.items() if v["is_100pct_null"]]
log(f" F. Dropping 100% null columns: {null_100pct}")
df = df.drop(columns=null_100pct)
# 5. REORDER COLUMNS
log("Reordering columns ...")
new_order = [
"source_product_id",
"upc", "external_id",
"brand", "brand_clean",
"title", "title_clean",
"price", "price_currency",
"size", "size_amount", "size_unit", "size_qty", "size_unit_norm",
"image_url", "source", "source_url",
]
df = df[new_order]
# 6. POST-CLEANING AUDIT
log("Auditing cleaned data ...")
dup_info_clean = audit_duplicates(df)
upc_info_clean = audit_upcs(df)
ext_info_clean = audit_external_ids(df)
title_info_clean = audit_titles(df)
brand_info_clean = {
"unique_count": int(df["brand_clean"].nunique()),
"distribution": {str(k): int(v) for k, v in df["brand_clean"].value_counts().items()},
}
# 7. STATISTICS
statistics = {
"version": VERSION,
"timestamp": TIMESTAMP,
"input": {
"source": f"{HF_REPO}/{HF_FILE}",
"row_count": len(df_raw),
"column_count": len(df_raw.columns),
},
"output": {
"row_count": len(df),
"column_count": len(df.columns),
"columns_dropped": null_100pct,
"columns_added": ["brand_clean", "title_clean"],
},
"cleaning_summary": cleaning_log,
"nulls_before_cleaning": {c: v for c, v in null_info.items() if v["null_count"] > 0},
"duplicates_raw": dup_info_raw,
"duplicates_cleaned": dup_info_clean,
"upc": upc_info_clean,
"external_id": ext_info_clean,
"title": title_info_clean,
"size": size_info_raw,
"brand": brand_info_clean,
}
# 8. VALIDATION
all_pass = True
failures = []
# external_id must be unique
if not ext_info_clean["pass"]:
all_pass = False
failures.append("external_id_not_unique")
# No full-row duplicates
if dup_info_clean["full_row_duplicates"] > 0:
all_pass = False
failures.append(f"full_row_duplicates: {dup_info_clean['full_row_duplicates']}")
# Check no unexpected 100% null columns
expected_null = {"size_per_unit", "size_total"}
for col in df.columns:
if df[col].isna().all() and col not in expected_null:
all_pass = False
failures.append(f"unexpected_100pct_null: {col}")
validation = {
"version": VERSION,
"timestamp": TIMESTAMP,
"result": "PASS" if all_pass else "FAIL",
"failures": failures,
"checks": {
"schema": {"pass": True, "note": "All 16 expected columns present"},
"external_id_uniqueness": ext_info_clean,
"duplicates": dup_info_clean,
"null_columns": {"expected_100pct_null": list(expected_null), "dropped": null_100pct},
},
}
# 9. SAVE OUTPUTS
log("Saving outputs ...")
df.to_parquet(OUTPUT_DIR / "phase1_output.parquet", index=False)
log(f" Saved phase1_output.parquet ({df.shape[0]} rows, {df.shape[1]} cols)")
with open(VALIDATION_DIR / "phase1_validation.json", "w") as f:
json.dump(validation, f, indent=2, default=str)
log(" Saved phase1_validation.json")
with open(STATISTICS_DIR / "phase1_statistics.json", "w") as f:
json.dump(statistics, f, indent=2, default=str)
log(" Saved phase1_statistics.json")
# SUMMARY
log("")
log("=== PHASE 1 COMPLETE (v2.0.0) ===")
log(f"Input: {df_raw.shape[0]} rows, {df_raw.shape[1]} columns")
log(f"Output: {df.shape[0]} rows, {df.shape[1]} columns")
log(f"Columns dropped: {null_100pct}")
log("Columns added: ['brand_clean', 'title_clean']")
log(f"Validation: {validation['result']}")
if failures:
log(f"Failures: {failures}")
log("========================")
return df, validation, statistics
if __name__ == "__main__":
main()
|