Download scripts/generate_data_engineering_fingerprint.py from jlai300/FitGround: direct link, hf CLI and curl.
- Browser
- Download file 4.58 kB
-
https://huggingface.co/datasets/jlai300/FitGround/resolve/main/scripts/generate_data_engineering_fingerprint.py
- Command line
-
hf download hf://datasets/jlai300/FitGround/scripts/generate_data_engineering_fingerprint.py
-
curl -L -o generate_data_engineering_fingerprint.py https://huggingface.co/datasets/jlai300/FitGround/resolve/main/scripts/generate_data_engineering_fingerprint.py
4.58 kB
| #!/usr/bin/env python3 | |
| """Generate data_engineering_v0.1_fingerprint.json when all gates pass.""" | |
| import hashlib | |
| import json | |
| import subprocess | |
| import sys | |
| from datetime import UTC, datetime | |
| from pathlib import Path | |
| sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "src")) | |
| from fitground.config import ( | |
| ARTIFACTS_DIR, | |
| EVAL_SHARD_COUNT, | |
| HF_REPO_ID, | |
| HF_SOURCE_REVISION, | |
| MANIFEST_PATH, | |
| TRAIN_SHARD_COUNT, | |
| ) | |
| def _sha256_file(path: Path) -> str: | |
| h = hashlib.sha256() | |
| with path.open("rb") as f: | |
| for chunk in iter(lambda: f.read(1 << 20), b""): | |
| h.update(chunk) | |
| return h.hexdigest() | |
| def _pytest_summary() -> dict[str, int]: | |
| r = subprocess.run( | |
| [sys.executable, "-m", "pytest", "-q", "--tb=no"], | |
| capture_output=True, | |
| text=True, | |
| cwd=Path(__file__).resolve().parents[1], | |
| ) | |
| line = [ln for ln in r.stdout.splitlines() if "passed" in ln] | |
| collected = passed = failed = 0 | |
| if line: | |
| import re | |
| m = re.search(r"(\d+) passed", line[-1]) | |
| if m: | |
| passed = int(m.group(1)) | |
| m2 = re.search(r"(\d+) failed", line[-1]) | |
| if m2: | |
| failed = int(m2.group(1)) | |
| m3 = re.search(r"(\d+) skipped", line[-1]) | |
| collected = passed + failed + (int(m3.group(1)) if m3 else 0) | |
| return {"pytest_collected": collected, "pytest_passed": passed, "pytest_failed": failed} | |
| def _git_commit() -> str | None: | |
| try: | |
| r = subprocess.run( | |
| ["git", "rev-parse", "HEAD"], | |
| capture_output=True, | |
| text=True, | |
| cwd=Path(__file__).resolve().parents[1], | |
| ) | |
| if r.returncode == 0: | |
| return r.stdout.strip() | |
| except OSError: | |
| pass | |
| return None | |
| def main() -> None: | |
| import pyarrow.parquet as pq | |
| schema_path = ARTIFACTS_DIR / "fit_clean_schema_v0.1.json" | |
| quality_path = ARTIFACTS_DIR / "quality_rules_v0.1.yaml" | |
| source_path = ARTIFACTS_DIR / "source_manifest.json" | |
| lock_candidates = [ | |
| ARTIFACTS_DIR / "uv.lock", | |
| Path(__file__).resolve().parents[1] / "uv.lock", | |
| ARTIFACTS_DIR / "requirements.lock.txt", | |
| Path(__file__).resolve().parents[1] / "requirements.lock.txt", | |
| ] | |
| lock_path = next((p for p in lock_candidates if p.exists()), None) | |
| if not MANIFEST_PATH.exists(): | |
| print(json.dumps({"error": "manifest missing", "path": str(MANIFEST_PATH)}, indent=2)) | |
| sys.exit(1) | |
| meta = pq.read_metadata(MANIFEST_PATH) | |
| train_rows = eval_rows = 0 | |
| # Read split counts from parquet via pandas for accuracy | |
| import pandas as pd | |
| df = pd.read_parquet(MANIFEST_PATH, columns=["source_split"]) | |
| train_rows = int((df["source_split"] == "train").sum()) | |
| eval_rows = int((df["source_split"] == "eval").sum()) | |
| total_rows = len(df) | |
| gates_ok = ( | |
| train_rows == 100_000 | |
| and eval_rows == 5_000 | |
| and total_rows == 105_000 | |
| and TRAIN_SHARD_COUNT == 406 | |
| and EVAL_SHARD_COUNT == 20 | |
| ) | |
| pytest = _pytest_summary() | |
| status = "FROZEN" if gates_ok and pytest["pytest_failed"] == 0 else "PARTIAL" | |
| fp = { | |
| "dataset_id": HF_REPO_ID, | |
| "dataset_revision": HF_SOURCE_REVISION, | |
| "train_shards": TRAIN_SHARD_COUNT, | |
| "eval_shards": EVAL_SHARD_COUNT, | |
| "train_rows": train_rows, | |
| "eval_rows": eval_rows, | |
| "total_rows": total_rows, | |
| "fit_clean_path": str(MANIFEST_PATH.relative_to(MANIFEST_PATH.parents[2])), | |
| "fit_clean_size": MANIFEST_PATH.stat().st_size, | |
| "fit_clean_sha256": _sha256_file(MANIFEST_PATH), | |
| "schema_version": "v0.1", | |
| "schema_sha256": _sha256_file(schema_path) if schema_path.exists() else None, | |
| "quality_rules_version": "v0.1", | |
| "quality_rules_sha256": _sha256_file(quality_path) if quality_path.exists() else None, | |
| "source_manifest_sha256": _sha256_file(source_path) if source_path.exists() else None, | |
| "pipeline_git_commit": _git_commit(), | |
| "python_version": sys.version.split()[0], | |
| "dependency_lock_sha256": _sha256_file(lock_path) if lock_path else None, | |
| "dependency_lock_path": str(lock_path) if lock_path else None, | |
| **pytest, | |
| "data_engineering_status": status, | |
| "generated_at": datetime.now(UTC).isoformat(), | |
| } | |
| out = ARTIFACTS_DIR / "data_engineering_v0.1_fingerprint.json" | |
| out.write_text(json.dumps(fp, indent=2)) | |
| print(json.dumps(fp, indent=2)) | |
| if status != "FROZEN": | |
| sys.exit(2) | |
| if __name__ == "__main__": | |
| main() | |