fixed the RAg in test pipeline issue

app/routers/documents.py

@@ -94,7 +94,7 @@ async def upload_document(
         file_bytes=file_bytes,
         file_name=file.filename,
         doc_id=doc_id,
-        chatbot=chatbot,
+        chatbot_id=chatbot_id,
         supabase=supabase,
     )
 
@@ -105,16 +105,28 @@ async def _process_document_bg(
     file_bytes: bytes,
     file_name: str,
     doc_id: str,
-    chatbot: dict,
+    chatbot_id: str,
     supabase,
 ):
     """Background task to process and embed a document"""
     try:
+        # Re-fetch chatbot to guarantee we use the canonical collection and company_id,
+        # not a snapshot that could have been captured before an update.
+        chatbot_row = supabase.table("chatbots").select("company_id, qdrant_collection_name").eq("id", chatbot_id).execute()
+        if not chatbot_row.data:
+            logger.error(f"Chatbot {chatbot_id} not found during document processing")
+            supabase.table("documents").update({
+                "status": "failed",
+                "error_message": "Chatbot not found"
+            }).eq("id", doc_id).execute()
+            return
+
+        chatbot = chatbot_row.data[0]
         company_id = chatbot.get("company_id", "")
         collection_name = chatbot.get("qdrant_collection_name")
 
         if not collection_name:
-            logger.error(f"No Qdrant collection for chatbot {chatbot['id']}")
+            logger.error(f"No Qdrant collection for chatbot {chatbot_id}")
             supabase.table("documents").update({
                 "status": "failed",
                 "error_message": "Vector store not configured"
@@ -168,7 +180,7 @@ async def _process_document_bg(
         }).eq("id", doc_id).execute()
 
         response_cache.invalidate(collection_name)
-        logger.info(f"Document {doc_id} processed: {len(chunks)} chunks")
+        logger.info(f"Document {doc_id} processed: {len(chunks)} chunks → collection='{collection_name}' company='{company_id}'")
 
     except Exception as e:
         logger.error(f"Document processing error for {doc_id}: {e}")
@@ -274,7 +286,7 @@ async def retry_document_processing(
         file_bytes=file_bytes,
         file_name=document["file_name"],
         doc_id=document_id,
-        chatbot=chatbot,
+        chatbot_id=chatbot_id,
         supabase=supabase,
     )
 
@@ -333,7 +345,7 @@ async def add_url_source(
         _process_url_source,
         source_id=source_id,
         url=data.url,
-        chatbot=chatbot,
+        chatbot_id=chatbot_id,
         supabase=supabase,
     )
 
@@ -394,12 +406,12 @@ async def refresh_url_source(
         "chunk_count": 0,
     }).eq("id", source_id).returning("representation").execute()
 
-    background_tasks.add_task(_process_url_source, source_id, src["url"], chatbot, supabase)
+    background_tasks.add_task(_process_url_source, source_id, src["url"], chatbot_id, supabase)
 
     return UrlSourceResponse(**{**src, "status": "pending", "chunk_count": 0})
 
 
-async def _process_url_source(source_id: str, url: str, chatbot: dict, supabase):
+async def _process_url_source(source_id: str, url: str, chatbot_id: str, supabase):
     """Background task to scrape a URL and add its content to the vector store."""
     from app.services.web_scraper import scrape_url
     from app.services.document_processor import chunk_text
@@ -407,6 +419,18 @@ async def _process_url_source(source_id: str, url: str, chatbot: dict, supabase)
     from app.services.vector_store import vector_store
 
     try:
+        # Re-fetch chatbot to guarantee we use the canonical collection and company_id.
+        chatbot_row = supabase.table("chatbots").select("company_id, qdrant_collection_name").eq("id", chatbot_id).execute()
+        if not chatbot_row.data:
+            logger.error(f"Chatbot {chatbot_id} not found during URL source processing")
+            supabase.table("url_sources").update({
+                "status": "failed",
+                "error_message": "Chatbot not found",
+            }).eq("id", source_id).execute()
+            return
+
+        chatbot = chatbot_row.data[0]
+
         # Update status to processing
         supabase.table("url_sources").update({"status": "processing"}).eq("id", source_id).execute()
 
@@ -480,7 +504,8 @@ async def _process_url_source(source_id: str, url: str, chatbot: dict, supabase)
         }).eq("id", source_id).execute()
 
         response_cache.invalidate(collection_name)
-        logger.info(f"URL source {source_id} processed: {len(chunks)} chunks from {url}")
+        logger.info(f"URL source {source_id} processed: {len(chunks)} chunks from {url} → collection='{collection_name}' company='{chatbot.get('company_id', '')}'")
+
 
     except Exception as e:
         logger.error(f"URL source processing error {source_id}: {e}")