Update app.py

app.py

@@ -40,7 +40,7 @@ llm = ChatOpenAI(
 # List of product names
 products = []
 for id in db.index_to_docstore_id.values():
-  prod_name = db.docstore.search(id).metadata
+  prod_name = db.docstore.search(id).metadata['Product Name']
   products.append(prod_name)
 
     
@@ -153,11 +153,9 @@ def retrieve_products(query):
   if correct_prods:
     prods_mp = {}
     for prod in correct_prods:
-      print(prod)
     # Retrieve from db, each product 4-chunks
-      docs = db.similarity_search(query = "", filter = {'Product Name': prod['Product Name']}, k = 4, fetch_k = 49190)
-      print(docs)
-      prods_mp[prod['Product Name']] = "\n".join([doc.page_content for doc in docs])
+      docs = db.similarity_search(query = "", filter = {'Product Name': prod}, k = 4, fetch_k = 49190)
+      prods_mp[prod] = "\n".join([doc.page_content for doc in docs])
 
     return prods_mp
   return False
@@ -222,7 +220,7 @@ def qa_retrieve(chatlog,):
 
             Answer the following question: {query}
             Use the following documents for each product:
-            {"\n".join(products)}
+            {products}
 
             Only use the factual information from the documents to answer the question. You are very careful in the products or theory name and will not invent or imagine names.
             Make sure to always answer the question with the same language the question is in. If it's in chinese make sure to answer in chinese. If it's in english answer in english.
@@ -236,7 +234,7 @@ def qa_retrieve(chatlog,):
     # llm = BardLLM()
     chain = LLMChain(llm=llm, prompt = prompt)
     
-    response = chain.run(query=query, products=mp_products)
+    response = chain.run(query=query, products="\n".join(mp_products))
 
     
     chatlog[-1][1] = response