Update app.py

app.py

@@ -1,39 +1,78 @@
-from flask import Flask, request, jsonify, make_response
-from presidio_analyzer import AnalyzerEngine
 import os
 import logging
+import yaml
+from flask import Flask, request, jsonify, make_response
 
-# Configuration du logging
+from presidio_analyzer import AnalyzerEngine, RecognizerRegistry
+from presidio_analyzer.nlp_engine import NlpEngineProvider
+
+# Configuration du logging pour un meilleur débogage
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
 logger = logging.getLogger(__name__)
 
+# --- CHARGEMENT MANUEL ET EXPLICITE DE LA CONFIGURATION ---
+
+# Chemin vers le fichier de configuration, défini par la variable d'environnement
+CONFIG_FILE_PATH = os.environ.get("PRESIDIO_ANALYZER_CONFIG_FILE", "conf/default.yaml")
+logger.info(f"Loading configuration from: {CONFIG_FILE_PATH}")
+
+try:
+    with open(CONFIG_FILE_PATH, 'r') as f:
+        config = yaml.safe_load(f)
+    logger.info("Configuration file loaded successfully.")
+except FileNotFoundError:
+    logger.error(f"Configuration file not found at {CONFIG_FILE_PATH}. Exiting.")
+    config = {}
+except yaml.YAMLError as e:
+    logger.error(f"Error parsing YAML configuration file: {e}. Exiting.")
+    config = {}
+
+# Création du fournisseur de moteur NLP basé sur la configuration
+logger.info("Creating NLP engine provider...")
+nlp_engine_provider = NlpEngineProvider(nlp_configuration=config.get("nlp_engine_configuration"))
+nlp_engine = nlp_engine_provider.create_engine()
+logger.info(f"NLP engine created with models for: {nlp_engine.get_supported_languages()}")
+
+# Création du registre de recognizers basé sur la configuration
+logger.info("Creating and populating recognizer registry...")
+registry = RecognizerRegistry()
+registry.load_predefined_recognizers(languages=config.get("supported_languages", ["en"]))
+
+# Ajout des recognizers personnalisés définis dans le YAML
+custom_recognizers_conf = config.get("recognizers", [])
+for recognizer_conf in custom_recognizers_conf:
+    registry.add_pattern_recognizer(
+        name=recognizer_conf['name'],
+        patterns=recognizer_conf['patterns'],
+        context=recognizer_conf.get('context'),
+        supported_language=recognizer_conf['supported_language'],
+        supported_entity=recognizer_conf['entity_name']
+    )
+    logger.info(f"Loaded custom recognizer: {recognizer_conf['name']}")
+
+# --- FIN DU CHARGEMENT DE LA CONFIGURATION ---
+
 # Initialisation de l'application Flask
 app = Flask(__name__)
 
-# Initialisation du moteur Presidio Analyzer
-analyzer = None
-try:
-    logger.info("Initializing AnalyzerEngine...")
-    analyzer = AnalyzerEngine()
+# Initialisation du moteur Presidio Analyzer avec les composants que nous avons créés
+logger.info("Initializing AnalyzerEngine with custom configuration...")
+analyzer = AnalyzerEngine(
+    nlp_engine=nlp_engine,
+    registry=registry,
+    supported_languages=config.get("supported_languages", ["en"])
+)
+
+# On ajoute l'allow_list manuellement
+allow_list = config.get("allow_list", [])
+if allow_list:
+    registry.add_recognizer(DenyListRecognizer(supported_entity="GENERIC_PII", deny_list=allow_list))
+    logger.info(f"Loaded {len(allow_list)} terms into the allow list (deny list recognizer).")
+
 logger.info("AnalyzerEngine initialized successfully.")
 
-    # --- CORRECTION ICI ---
-    # La ligne de débogage est corrigée ou commentée.
-    # On va la commenter pour l'instant car elle n'est pas essentielle au fonctionnement.
-    # loaded_recognizers = [r.name for r in analyzer.registry.get_recognizers(language="fr")]
-    # logger.info(f"Loaded recognizers for 'fr': {loaded_recognizers}")
-
-except Exception as e:
-    # La ligne 'analyzer = None' était déjà là, mais on s'assure qu'elle est bien là.
-    analyzer = None
-    logger.exception("FATAL: Error initializing AnalyzerEngine.")
-
-
 @app.route('/analyze', methods=['POST'])
 def analyze_text():
-    if not analyzer:
-        return jsonify({"error": "Analyzer engine not initialized"}), 500
-    
     try:
         data = request.get_json(force=True)
         text_to_analyze = data.get("text", "")
@@ -42,9 +81,17 @@ def analyze_text():
         if not text_to_analyze:
              return jsonify({"error": "text field is missing or empty"}), 400
         
-        results = analyzer.analyze(text=text_to_analyze, language=language)
-        response_data = [res.to_dict() for res in results]
+        # Le seuil de confiance est appliqué ici, à la volée
+        score_threshold = data.get("score_threshold", config.get("ner_model_configuration", {}).get("confidence_threshold", {}).get("default", 0.35))
         
+        results = analyzer.analyze(
+            text=text_to_analyze,
+            language=language,
+            score_threshold=score_threshold,
+            allow_list=allow_list # On passe la allow list ici aussi
+        )
+        
+        response_data = [res.to_dict() for res in results]
         return make_response(jsonify(response_data), 200)
     except Exception as e:
         logger.exception("Error during analysis.")