anonymisation/scripts/merge_params.py

#!/usr/bin/env python3
"""Fusionne les fichiers de paramètres envoyés par les établissements.

Usage :
    python scripts/merge_params.py fichier1.json [fichier2.json ...]
    python scripts/merge_params.py --dir /chemin/vers/exports/

Fusionne les whitelist_phrases et blacklist_force_mask_terms de chaque
fichier JSON exporté par la GUI dans la config maîtresse (dictionnaires.yml).
"""
import argparse
import json
import sys
from pathlib import Path

try:
    import yaml
except ImportError:
    print("ERREUR : pyyaml requis (pip install pyyaml)")
    sys.exit(1)

CONFIG = Path(__file__).parent.parent / "config" / "dictionnaires.yml"


def merge_params(json_files: list, config_path: Path = CONFIG, dry_run: bool = False):
    """Fusionne les paramètres des fichiers JSON dans la config YAML."""
    if not config_path.exists():
        print(f"ERREUR : config introuvable : {config_path}")
        return

    cfg = yaml.safe_load(config_path.read_text(encoding="utf-8")) or {}

    # Charger les listes existantes
    existing_wl = set(cfg.get("whitelist_phrases", []))
    existing_bl = set(cfg.get("blacklist", {}).get("force_mask_terms", []))

    added_wl = set()
    added_bl = set()
    sources = []

    for jf in json_files:
        try:
            data = json.loads(Path(jf).read_text(encoding="utf-8"))
            src = f"{Path(jf).name} (v{data.get('version', '?')}, {data.get('date_export', '?')[:10]})"
            sources.append(src)

            for phrase in data.get("whitelist_phrases", []):
                if phrase and phrase.strip() and phrase.strip() not in existing_wl:
                    added_wl.add(phrase.strip())

            for term in data.get("blacklist_force_mask_terms", []):
                if term and str(term).strip() and str(term).strip() not in existing_bl:
                    added_bl.add(str(term).strip())

        except Exception as e:
            print(f"  ERREUR lecture {jf}: {e}")

    print(f"\nSources traitées : {len(sources)}")
    for s in sources:
        print(f"  - {s}")

    print(f"\nNouvelles phrases whitelist : {len(added_wl)}")
    for p in sorted(added_wl):
        print(f"  + {p}")

    print(f"\nNouveaux termes blacklist : {len(added_bl)}")
    for t in sorted(added_bl):
        print(f"  + {t}")

    if not added_wl and not added_bl:
        print("\nRien de nouveau à fusionner.")
        return

    if dry_run:
        print("\n(dry-run — aucune modification)")
        return

    # Appliquer
    cfg["whitelist_phrases"] = sorted(existing_wl | added_wl)
    if "blacklist" not in cfg:
        cfg["blacklist"] = {}
    cfg["blacklist"]["force_mask_terms"] = sorted(existing_bl | added_bl)

    config_path.write_text(
        yaml.dump(cfg, allow_unicode=True, default_flow_style=False, sort_keys=False),
        encoding="utf-8",
    )
    print(f"\nConfig mise à jour : {config_path}")
    print(f"  Whitelist : {len(cfg['whitelist_phrases'])} phrases")
    print(f"  Blacklist : {len(cfg['blacklist']['force_mask_terms'])} termes")


def main():
    parser = argparse.ArgumentParser(description="Fusionner les paramètres d'anonymisation")
    parser.add_argument("files", nargs="*", help="Fichiers JSON à fusionner")
    parser.add_argument("--dir", type=Path, help="Dossier contenant les fichiers JSON")
    parser.add_argument("--config", type=Path, default=CONFIG, help="Config YAML cible")
    parser.add_argument("--dry-run", action="store_true", help="Afficher sans modifier")
    args = parser.parse_args()

    json_files = list(args.files)
    if args.dir and args.dir.is_dir():
        json_files.extend(str(f) for f in args.dir.glob("*.json"))

    if not json_files:
        print("Aucun fichier JSON spécifié. Usage :")
        print("  python scripts/merge_params.py export1.json export2.json")
        print("  python scripts/merge_params.py --dir /chemin/exports/")
        return

    merge_params(json_files, config_path=args.config, dry_run=args.dry_run)


if __name__ == "__main__":
    main()