problem with fiding correct vector store

Author: MRGuziX

agents/community/langgraph_agentic_rag/data/load_documents.py

@@ -91,7 +91,7 @@ def load_and_index_documents(
     chunks = []
     for doc in all_chunks:
         content = doc.page_content.strip()
-        if content and not all(c in "=-_*#\n\r\t " for c in content):
+        if content and not all(c in "=-_*#|\n\r\t " for c in content):
             chunks.append(content)
     print(f"Created {len(chunks)} chunks (filtered out empty/separator chunks)")
 
@@ -131,8 +131,6 @@ def load_and_index_documents(
         vector_store_id=vector_store.id,
     )
 
-    print("\n =")
-
 
 if __name__ == "__main__":
     load_and_index_documents()

agents/community/langgraph_agentic_rag/deploy.sh

@@ -14,7 +14,7 @@
 set -e  # Exit on error
 
 source .env
-export CONTAINER_IMAGE BASE_URL MODEL_ID EMBEDDING_MODEL
+export CONTAINER_IMAGE BASE_URL MODEL_ID EMBEDDING_MODEL VECTOR_STORE_NAME
 
 ## ============================================
 # DOCKER BUILD

agents/community/langgraph_agentic_rag/src/langgraph_agentic_rag/tools.py

@@ -40,14 +40,22 @@ def get_retriever_components(
         api_key=getenv("API_KEY"),
     )
 
-    # Get the vector store ID
+    # Get the vector store ID by name
     vector_store_list = client.vector_stores.list()
+    vector_store_id = None
 
     for vs in vector_store_list.data:
         if vs.name == vector_store_name:
             print(f"Your Vector Store: {vs.id} ({vs.name})")
             vector_store_id = vs.id
 
+    if not vector_store_id:
+        available = [f"{vs.name} ({vs.id})" for vs in vector_store_list.data]
+        raise RuntimeError(
+            f"Vector store '{vector_store_name}' not found. "
+            f"Available: {available}. Run load_documents.py first."
+        )
+
     # Cache the components
     _client_cache = client
     _vector_store_id_cache = vector_store_id
@@ -105,7 +113,7 @@ def retriever_tool(query: str) -> str:
     for i, chunk in enumerate(response.chunks, 1):
         # Skip chunks that are empty or just separators/whitespace
         content = chunk.content.strip()
-        if not content or all(c in "=-_*#" for c in content):
+        if not content or all(c in "=-_*#|" for c in content):
             continue
 
         # Extract source from chunk metadata (Pydantic object)