Domi31tls
1c44a26eb3
Anonymise toutes les références à des entités réelles (CHCB, Bayonne, Saint-Denis, Réunion, etc.) dans le code source, les configurations YAML, les scripts/outils, et les tests unitaires. Conserve les tests synthétiques (cases) intentionnels. - profile key chcb_strict → chuxx_strict - CHCB → CHUXX, Bayonne → Chicago, Saint-Denis → Springfield, Réunion → Province Bêta, 64100/97400 → 12345, FINESS → 999999999, préfixe tél 05.59.44 → 0X.XX.XX - renomme tools/test_chcb_leak.py → tools/test_force_term_leak.py Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
86 lines
2.7 KiB
Python
86 lines
2.7 KiB
Python
#!/usr/bin/env python3
|
|
"""Batch processing des 59 premiers OGC — script CLI pour test post-modifications."""
|
|
import sys
|
|
import time
|
|
import json
|
|
from pathlib import Path
|
|
from collections import Counter
|
|
|
|
sys.path.insert(0, str(Path(__file__).parent))
|
|
|
|
import anonymizer_core_refactored_onnx as core
|
|
from config_defaults import RUNTIME_DICTIONARIES_CONFIG_PATH
|
|
from eds_pseudo_manager import EdsPseudoManager
|
|
|
|
SRC = Path("/home/dom/Téléchargements/II-1 Ctrl_T2A_2025_CHUXX_DocJustificatifs (1)")
|
|
OUTDIR = SRC / "anonymise"
|
|
CONFIG = RUNTIME_DICTIONARIES_CONFIG_PATH
|
|
|
|
def main():
|
|
# Charger EDS-Pseudo
|
|
print("Chargement EDS-Pseudo...", flush=True)
|
|
ner = EdsPseudoManager()
|
|
ner.load()
|
|
assert ner.is_loaded(), "EDS-Pseudo non chargé"
|
|
print("EDS-Pseudo chargé.", flush=True)
|
|
|
|
# Lister les 59 premiers dossiers OGC
|
|
ogc_dirs = sorted(
|
|
[d for d in SRC.iterdir() if d.is_dir() and "_" in d.name and d.name[0].isdigit()],
|
|
key=lambda d: int(d.name.split("_")[0]),
|
|
)[:59]
|
|
print(f"Dossiers OGC: {len(ogc_dirs)}")
|
|
|
|
# Collecter tous les PDFs
|
|
pdfs = []
|
|
for d in ogc_dirs:
|
|
for pdf in sorted(d.glob("*.pdf")):
|
|
pdfs.append(pdf)
|
|
print(f"PDFs à traiter: {len(pdfs)}")
|
|
|
|
OUTDIR.mkdir(exist_ok=True)
|
|
ok = ko = 0
|
|
global_counts = Counter()
|
|
t0 = time.time()
|
|
|
|
for i, pdf in enumerate(pdfs, 1):
|
|
ogc = pdf.parent.name.split("_")[0]
|
|
print(f"[{i}/{len(pdfs)}] {pdf.name} (OGC {ogc})...", end=" ", flush=True)
|
|
try:
|
|
outputs = core.process_pdf(
|
|
pdf_path=pdf,
|
|
out_dir=OUTDIR,
|
|
make_vector_redaction=False,
|
|
also_make_raster_burn=True,
|
|
config_path=CONFIG,
|
|
use_hf=True,
|
|
ner_manager=ner,
|
|
ner_thresholds=None,
|
|
ogc_label=ogc,
|
|
)
|
|
# Compter les hits audit
|
|
audit_path = Path(outputs.get("audit", ""))
|
|
if audit_path.exists():
|
|
for line in audit_path.read_text().splitlines():
|
|
try:
|
|
h = json.loads(line)
|
|
global_counts[h["kind"]] += 1
|
|
except Exception:
|
|
pass
|
|
print("OK", flush=True)
|
|
ok += 1
|
|
except Exception as e:
|
|
print(f"ERREUR: {e}", flush=True)
|
|
ko += 1
|
|
|
|
elapsed = time.time() - t0
|
|
print(f"\n{'='*60}")
|
|
print(f"Terminé en {elapsed:.0f}s — OK: {ok}, Erreurs: {ko}")
|
|
print(f"Total PII détectés: {sum(global_counts.values())}")
|
|
print(f"\nDétail par type:")
|
|
for k, v in sorted(global_counts.items(), key=lambda x: -x[1]):
|
|
print(f" {k:30s} {v:6d}")
|
|
|
|
if __name__ == "__main__":
|
|
main()
|