new full

pipeline_manager.py

@@ -2,6 +2,7 @@ from typing import List
 from presidio_analyzer import RecognizerResult
 from entity_refiners import EntityRefinerManager
 from post_processors import DeduplicationProcessor, OverlapResolver
+from post_processors.cleanup_processor import CleanupProcessor
 import logging
 
 logger = logging.getLogger(__name__)
@@ -9,9 +10,10 @@ logger = logging.getLogger(__name__)
 class AnalysisPipeline:
     def __init__(self):
         self.refiner_manager = EntityRefinerManager()
+        self.cleanup_processor = CleanupProcessor()
         self.overlap_resolver = OverlapResolver()
         self.deduplicator = DeduplicationProcessor()
-        logger.info("🚀 Pipeline d'analyse initialisé")
+        logger.info("🚀 Pipeline d'analyse initialisé avec nettoyage avancé")
     
     def process(self, text: str, results: List[RecognizerResult], allow_list_terms: List[str]) -> List[RecognizerResult]:
         """Traite les résultats à travers le pipeline complet"""
@@ -38,10 +40,13 @@ class AnalysisPipeline:
                 )
                 refined_results.append(refined_result)
         
-        # 3. Résolution des chevauchements
-        resolved_results = self.overlap_resolver.process(refined_results, text)
+        # 3. Nettoyage avancé des résultats
+        cleaned_results = self.cleanup_processor.process(refined_results)
         
-        # 4. Déduplication
+        # 4. Résolution des chevauchements
+        resolved_results = self.overlap_resolver.process(cleaned_results, text)
+        
+        # 5. Déduplication
         final_results = self.deduplicator.process(resolved_results, text)
         
         logger.info(f"🎯 Pipeline complet: {len(results)} -> {len(final_results)} entités")