Fix compatibility issues with OpenAI embeddings and Qdrant client

streamlit_app.py

@@ -111,12 +111,45 @@ def get_agent_model():
 @st.cache_resource
 def get_embedding_model():
     """Get the embedding model."""
-    return OpenAIEmbeddings(model="text-embedding-3-small")
+    try:
+        # First try the standard approach
+        return OpenAIEmbeddings(model="text-embedding-3-small")
+    except Exception as e:
+        if "proxies" in str(e):
+            # If there's a proxies error, create with additional kwargs to override defaults
+            import os
+            from langchain_openai import OpenAIEmbeddings
+            
+            return OpenAIEmbeddings(
+                model="text-embedding-3-small",
+                openai_api_key=os.environ.get("OPENAI_API_KEY"),
+                show_progress_bar=False,
+                client_kwargs={"proxies": None, "timeout": 60}
+            )
+        else:
+            # Re-raise any other exceptions
+            raise
 
 @st.cache_resource
 def setup_qdrant_client():
     """Set up the Qdrant client."""
-    return QdrantClient(path=str(QDRANT_DIR))
+    try:
+        return QdrantClient(path=str(QDRANT_DIR))
+    except Exception as e:
+        # If there's an issue with the standard approach, print diagnostics and retry
+        print(f"QdrantClient initialization error: {str(e)}")
+        print(f"Checking if directory exists: {os.path.exists(str(QDRANT_DIR))}")
+        
+        # Try alternative approach with explicit collection params
+        if os.path.exists(str(QDRANT_DIR)):
+            try:
+                # Try with location parameter instead
+                return QdrantClient(location=str(QDRANT_DIR))
+            except Exception as e2:
+                print(f"Alternative initialization also failed: {str(e2)}")
+                raise
+        else:
+            raise ValueError(f"Qdrant directory does not exist: {str(QDRANT_DIR)}")
 
 def retrieve_documents(query, k=5):
     """Retrieve relevant documents for a query."""
@@ -132,11 +165,22 @@ def retrieve_documents(query, k=5):
     query_embedding = embedding_model.embed_query(query)
     
     # Search Qdrant
-    results = client.search(
-        collection_name="kohavi_ab_testing_pdf_collection",
-        query_vector=query_embedding,
-        limit=k
-    )
+    try:
+        # Try the new API method first
+        results = client.query_points(
+            collection_name="kohavi_ab_testing_pdf_collection",
+            query_vector=query_embedding,
+            limit=k
+        )
+    except Exception as e:
+        print(f"Error with query_points method: {str(e)}")
+        # Fall back to the deprecated method
+        results = client.search(
+            collection_name="kohavi_ab_testing_pdf_collection",
+            query_vector=query_embedding,
+            limit=k
+        )
+        print("Using deprecated search method")
     
     # Convert results to documents
     documents = []