nacim/Presidio
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update app.py

Browse Source
This commit is contained in:
Nacim
2025-06-23 17:07:57 +02:00
committed by GitHub
parent 63f48af110
commit 6206d50490
1 changed files with 28 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

59
app.py
View File

@@ -19,7 +19,7 @@ analyzer = None
try: try:
logger.info("--- Presidio Analyzer Service Starting ---") logger.info("--- Presidio Analyzer Service Starting ---")
# 1. Charger la configuration depuis le fichier YAML # 1. Charger la configuration
CONFIG_FILE_PATH = os.environ.get("PRESIDIO_ANALYZER_CONFIG_FILE", "conf/default.yaml") CONFIG_FILE_PATH = os.environ.get("PRESIDIO_ANALYZER_CONFIG_FILE", "conf/default.yaml")
logger.info(f"Loading configuration from: {CONFIG_FILE_PATH}") logger.info(f"Loading configuration from: {CONFIG_FILE_PATH}")
with open(CONFIG_FILE_PATH, 'r', encoding='utf-8') as f: with open(CONFIG_FILE_PATH, 'r', encoding='utf-8') as f:
@@ -30,55 +30,52 @@ try:
logger.info("Creating NLP engine provider...") logger.info("Creating NLP engine provider...")
provider = NlpEngineProvider(nlp_configuration=config) provider = NlpEngineProvider(nlp_configuration=config)
# 3. Créer un registre et le peupler SEULEMENT avec les détecteurs français # 3. Créer et VIDER le registre pour garantir une base saine
logger.info("Creating and populating a French-first RecognizerRegistry...") logger.info("Creating a new RecognizerRegistry...")
registry = RecognizerRegistry() registry = RecognizerRegistry()
logger.info(f"Initial registry state supports: {registry.supported_languages}")
# Charger les recognizers personnalisés (définis sous la clé 'recognizers') # === CORRECTION DÉFINITIVE : VIDER LE REGISTRE ===
custom_recognizers_conf = config.get("recognizers", []) registry.recognizers.clear()
for recognizer_conf in custom_recognizers_conf: logger.info(f"Registry cleared. Now supports: {registry.supported_languages}")
# On s'assure de ne charger que les détecteurs prévus pour le français
if recognizer_conf.get('supported_language') == 'fr': # 4. Ajouter les détecteurs requis
patterns = [Pattern(name=p['name'], regex=p['regex'], score=p['score']) for p in recognizer_conf['patterns']] # Ajouter les détecteurs personnalisés pour le français
custom_recognizer = PatternRecognizer( for recognizer_conf in config.get("recognizers", []):
supported_entity=recognizer_conf['entity_name'], patterns = [Pattern(name=p['name'], regex=p['regex'], score=p['score']) for p in recognizer_conf['patterns']]
name=recognizer_conf['name'], registry.add_recognizer(PatternRecognizer(
supported_language='fr', supported_entity=recognizer_conf['entity_name'],
patterns=patterns, name=recognizer_conf['name'],
context=recognizer_conf.get('context') supported_language=recognizer_conf['supported_language'],
) patterns=patterns,
registry.add_recognizer(custom_recognizer) context=recognizer_conf.get('context')
logger.info(f"Loaded and registered custom FRENCH recognizer: {custom_recognizer.name}") ))
logger.info(f"Added custom recognizer: {recognizer_conf['name']}")
# Ajouter le support de base pour les entités françaises (PERSON, LOC) via Spacy # Ajouter le support des entités de base (PERSON, LOC) pour les deux langues
registry.add_recognizer(SpacyRecognizer(supported_language="en"))
registry.add_recognizer(SpacyRecognizer(supported_language="fr")) registry.add_recognizer(SpacyRecognizer(supported_language="fr"))
logger.info("Registered SpacyRecognizer for 'fr'.") logger.info("Added SpacyRecognizer for 'en' and 'fr'.")
logger.info(f"Registry state post-french setup. Supported languages: {registry.supported_languages}")
# 4. Créer l'AnalyzerEngine logger.info(f"Final registry state. Now supports: {registry.supported_languages}")
# 5. Créer l'AnalyzerEngine
logger.info("Initializing AnalyzerEngine...") logger.info("Initializing AnalyzerEngine...")
analyzer = AnalyzerEngine( analyzer = AnalyzerEngine(
nlp_engine=provider.create_engine(), nlp_engine=provider.create_engine(),
registry=registry, registry=registry,
supported_languages=config.get("supported_languages", ["en", "fr"]), supported_languages=config.get("supported_languages")
# === CORRECTION DÉFINITIVE ===
# On empêche le moteur de créer un SpacyRecognizer anglais par défaut
# en lui fournissant un nous-mêmes. Il l'ajoutera au registre.
default_recognizer=SpacyRecognizer(supported_language="en")
) )
analyzer.set_allow_list(config.get("allow_list", [])) analyzer.set_allow_list(config.get("allow_list", []))
logger.info("--- Presidio Analyzer Service Ready ---") logger.info("--- Presidio Analyzer Service Ready ---")
logger.info(f"Final registry languages: {registry.supported_languages}")
logger.info(f"Analyzer supported languages: {analyzer.supported_languages}")
except Exception as e: except Exception as e:
logger.exception("FATAL: Error during AnalyzerEngine initialization.") logger.exception("FATAL: Error during AnalyzerEngine initialization.")
analyzer = None analyzer = None
# Le reste du fichier Flask reste identique... # Le reste du fichier Flask est identique
@app.route('/analyze', methods=['POST']) @app.route('/analyze', methods=['POST'])
def analyze_text(): def analyze_text():
if not analyzer: return jsonify({"error": "Analyzer engine is not available."}), 500 if not analyzer: return jsonify({"error": "Analyzer engine is not available."}), 500