Update src/utils/search_docs_utils.py

src/utils/search_docs_utils.py

@@ -272,57 +272,34 @@ def select_parlement_transcript(query: str, embeddings_path="output/parlement_ti
 def search_relevant_documents(query: str, top_k: int = 3) -> dict:
     """
     Search for the most relevant documents following the logic of the provided graph:
-    - Detects the user's intention ("lois/règlements" or "parlement")
-    - If "lois/règlements": detects language, filters, then selects documents
-    - If "parlement": directly selects the most relevant parliamentary transcripts
     Returns a dictionary with intention, language (if applicable), and the relevant documents.
     """
-    # Detect user intention
-    intention = detect_intention(query)
+    
     # Detect language
     lang = detect_language(query)
 
-    if intention == "parlement":
-        if parlement_embeddings is None or parlement_metadatas is None:
-            return {
-                "error": "Parliamentary embeddings not available. Please run preprocessing first.",
-                "intention": intention,
-                "language": lang
-                }
-            
-        # Search parliamentary transcripts
-        results = select_parlement_transcript(query, top_k=top_k)
-            
+    if embeddings is None or metadatas is None:
         return {
-            "intention": intention,
-            "language": lang,
-            "results": results,
-            "count": len(results) if isinstance(results, list) else 1
+            "error": "Document embeddings not available. Please run preprocessing first.",
+            "language": lang
         }
-        
-    else:
-        if embeddings is None or metadatas is None:
-            return {
-                "error": "Document embeddings not available. Please run preprocessing first.",
-                "intention": intention,
-                "language": lang
-            }
             
-        # Filter by language and select documents
-        indices = filter_by_language(metadatas, lang)
-        filtered_embeddings = embeddings[indices]
-        filtered_metadatas = [metadatas[i] for i in indices]
-            
-        results = select_documents(
-            query,
-            filtered_embeddings,
-            filtered_metadatas,
-            lang=lang,
-            top_k=top_k
-        )
-        return {
-            "intention": intention,
-            "language": lang,
-            "results": results,
-            "count": len(results)
-        }
+    # Filter by language and select documents
+    indices = filter_by_language(metadatas, lang)
+    filtered_embeddings = embeddings[indices]
+    filtered_metadatas = [metadatas[i] for i in indices]
+    
+    results = select_documents(
+        query,
+        filtered_embeddings,
+        filtered_metadatas,
+        lang=lang,
+        top_k=top_k
+    )
+
+    return {
+        #"intention": 'intention',
+        "language": lang,
+        "results": results,
+        "count": len(results)
+    }