Update app.py

app.py

@@ -42,9 +42,12 @@ pc = Pinecone(api_key=PINE_API_KEY)
 index = pc.Index(index_name)
 
 # ========== PART 3 ==========
+
+text_field = "child_content"
+
 splade_encoder = SpladeEncoder()
 retriever = PineconeHybridSearchRetriever(
-    embeddings=embed, sparse_encoder=splade_encoder, index=index
+    embeddings=embed, sparse_encoder=splade_encoder, index=index, text_field
 )
 
 
@@ -95,8 +98,8 @@ class PassParentContent(Runnable):
 
 rag_chain = (
     RunnablePassthrough()
-    # | SourceDedup()
-    # | PassParentContent()
+    | SourceDedup()
+    | PassParentContent()
     | prompt
     | model
     | StrOutputParser()
@@ -120,20 +123,20 @@ def generate_response(prompt):
             elif list(chunk.keys())[0] == 'context':
                 pass
                 # Sources DO NOT work the same with this code... removing for now.
-                # sources = chunk['context']
-                #for thing in chunk['context']:
-                    #print()
-                    #print(thing.metadata)
-                #sources = [doc.metadata['source'] for doc in chunk['context']]
-
-    #response = rag_chain_with_source.invoke(prompt)
-    #answer = response["answer"]
-    #sources_txt = "\n\nSources:\n" + "\n".join(sources)
-    #yield sources_txt
-
-#question = "How can I do hybrid search with a pinecone database?"
-#answer = generate_response(question)
-#print(answer)
+                sources = chunk['context']
+                for thing in chunk['context']:
+                    print()
+                    print(thing.metadata)
+                sources = [doc.metadata['source'] for doc in chunk['context']]
+
+    response = rag_chain_with_source.invoke(prompt)
+    answer = response["answer"]
+    sources_txt = "\n\nSources:\n" + "\n".join(sources)
+    yield sources_txt
+
+question = "How can I do hybrid search with a pinecone database?"
+answer = generate_response(question)
+print(answer)
 
 # ==================== THE REST OF THE STREAMLIT APP ====================