Update app/services/pinecone_service.py

app/services/pinecone_service.py

@@ -59,7 +59,7 @@ class PineconeService:
             raise
     
     async def store_embeddings(self, repository_id: int, embedded_chunks: List[Dict]):
-        """Store embeddings in Pinecone with repository namespace"""
+        """Store embeddings in Pinecone with minimal metadata (content stored in PostgreSQL)"""
         print(f"💾 [PINECONE] Storing {len(embedded_chunks)} embeddings for repository {repository_id}", flush=True)
         logger.info(f"💾 Storing {len(embedded_chunks)} embeddings for repository {repository_id}")
         
@@ -68,17 +68,18 @@ class PineconeService:
             for i, chunk in enumerate(embedded_chunks):
                 vector_id = f"repo_{repository_id}_chunk_{chunk['chunk_index']}_{i}"
                 
+                # Store ONLY identifiers - full content is in PostgreSQL
                 vector = {
                     "id": vector_id,
                     "values": chunk['embedding'],
                     "metadata": {
                         "repository_id": repository_id,
                         "file_path": chunk['file_path'],
+                        "chunk_index": chunk['chunk_index'],
                         "start_line": chunk['start_line'],
                         "end_line": chunk['end_line'],
-                        "chunk_type": chunk['chunk_type'],
-                        "content_length": chunk['content_length'],
-                        "content": chunk['content'][:1000]  # Pinecone metadata limit
+                        "chunk_type": chunk['chunk_type']
+                        # NO content field - saves Pinecone storage!
                     }
                 }
                 vectors.append(vector)
@@ -108,7 +109,7 @@ class PineconeService:
             raise
     
     async def search_similar_code(self, repository_id: int, query_embedding: List[float], top_k: int = 5) -> List[Dict]:
-        """Search for similar code using Pinecone"""
+        """Search for similar code using Pinecone - returns identifiers only"""
         try:
             print(f"🔍 [PINECONE] Searching for {top_k} similar chunks in repository {repository_id}", flush=True)
             
@@ -126,16 +127,18 @@ class PineconeService:
                 similarity = match.score  # Cosine similarity (0-1, higher is better)
                 metadata = match.metadata
                 
+                # Return identifiers to fetch full content from PostgreSQL
                 search_results.append({
-                    'content': metadata.get('content', ''),
-                    'metadata': metadata,
-                    'similarity': similarity,
+                    'repository_id': metadata.get('repository_id'),
                     'file_path': metadata.get('file_path', ''),
+                    'chunk_index': metadata.get('chunk_index', 0),
                     'start_line': metadata.get('start_line', 0),
-                    'end_line': metadata.get('end_line', 0)
+                    'end_line': metadata.get('end_line', 0),
+                    'chunk_type': metadata.get('chunk_type', ''),
+                    'similarity': similarity
                 })
             
-            print(f"✅ [PINECONE] Found {len(search_results)} similar code chunks", flush=True)
+            print(f"✅ [PINECONE] Found {len(search_results)} similar code chunks (identifiers only)", flush=True)
             logger.info(f"🔍 Found {len(search_results)} similar code chunks")
             return search_results