Update app.py

app.py

@@ -88,7 +88,7 @@ def index_urls_in_rag(urls=[]):
     # Create a vector store for storing embeddings of documents
     vector_store = Chroma(persist_directory="/home/user/.cache/chroma_db", embedding_function=embeddings)
     
-    print(urls)
+    print("Embedding " +urls)
 
 
     for url in urls:
@@ -99,7 +99,7 @@ def index_urls_in_rag(urls=[]):
         # Split the document into chunks
         text_splitter = RecursiveCharacterTextSplitter()
         document_chunks = text_splitter.split_documents(document)
-
+        print(document_chunks)
         # Index document chunks into the vector store
         vector_store.add_documents(document_chunks)
 
@@ -149,50 +149,25 @@ def get_response(message, history=[]):
     last_part = parts[-1]
     return last_part#[-1]['generation']['content']   
 
- 
-def get_conversational_rag_chain(retriever_chain): 
-    
-    llm = load_model()
-    
-    prompt = ChatPromptTemplate.from_messages([
-      ("system", "Du bist eine freundlicher Mitarbeiterin Namens Susie und arbeitest in einenm Call Center. Du beantwortest basierend auf dem Context. Benutze nur den Inhalt des Context. Füge wenn möglich die Quelle hinzu. Antworte mit: Ich bin mir nicht sicher. Wenn die Antwort nicht aus dem Context hervorgeht. Antworte auf Deutsch, bitte? CONTEXT:\n\n{context}"),
-      MessagesPlaceholder(variable_name="chat_history"),
-      ("user", "{input}"),
-    ])
-    
-    stuff_documents_chain = create_stuff_documents_chain(llm,prompt)
-    
-    return create_retrieval_chain(retriever_chain, stuff_documents_chain)
-
-    
-def get_response(message, history=[]): 
-
-    retriever_chain = index_urls_in_rag()
-    conversation_rag_chain = get_conversational_rag_chain(retriever_chain)
-    
-    response = conversation_rag_chain.invoke({
-        "chat_history": history,
-        "input": message  + " Assistant: ",
-        "chat_message": message + " Assistant: "
-    })
-    #print("get_response " +response)
-    res = response['answer']
-    parts = res.split(" Assistant: ")
-    last_part = parts[-1]
-    return last_part#[-1]['generation']['content']   
-
 
 # Index URLs on app startup
 @app.on_event("startup")
 async def startup():
-    domain_url = 'https://www.bofrost.de/faq/service-infos-fuer-neukunden.html'
-    links = get_all_links_from_domain(domain_url)
-    print(links)
-    retriever_chain = index_urls_in_rag(links)
-    retriever_chain.invoke("Was ist bofrost*")
-    get_response("Was kosten Schoko Osterhasen?")
+   # domain_url = 'https://www.bofrost.de/faq/service-infos-fuer-neukunden.html'
+   # links = get_all_links_from_domain(domain_url)
+   # print(links)
+   # retriever_chain = index_urls_in_rag(links)
+   # retriever_chain.invoke("Was ist bofrost*")
+   # get_response("Was kosten Schoko Osterhasen?")
 
 # Define API endpoint to receive queries and provide responses
 @app.post("/generate/")
 def generate(user_input): 
-    return get_response(user_input, [])
+    return get_response(user_input, [])
+
+# Define API endpoint to receive queries and provide responses
+@app.post("/update/")
+def generate(index_url): 
+    retriever_chain = index_urls_in_rag(domain_url)
+    retriever_chain.invoke("Was ist bofrost*")
+    get_response("Was kosten Schoko Osterhasen?")