Update app.py

app.py

@@ -46,7 +46,7 @@ hf_api = HfApi()
 
 
 def initialize():
-    global vectorstore, chunks, llm
+    global kadiAPY_Bot
 
     
     download_gitlab_repo_to_hfspace(GITLAB_API_URL, GITLAB_PROJECT_ID, GITLAB_PROJECT_VERSION, DATA_DIR, hf_api, HF_SPACE_NAME)
@@ -66,59 +66,13 @@ def initialize():
     vectorstore = setup_vectorstore(doc_chunks + code_chunks, EMBEDDING_MODEL_NAME, VECTORSTORE_DIRECTORY)
     llm = get_groq_llm(LLM_MODEL_NAME, LLM_MODEL_TEMPERATURE, GROQ_API_KEY)
 
+    kadiAPY_bot = KadiAPYBot(llm, vectorstore)
 
 initialize()
-
-def rag_workflow(query):
-    """
-    RAGChain class to perform the complete RAG workflow.
-    """
-    # Assume 'llm' and 'vectorstore' are already initialized instances
-    rag_chain = RAGChain(llm, vectorstore)
-
-
-    """
-    Pre-Retrieval-Stage  
-    """
-    # predict which python library to search in: (standard) kadiAPY-library or kadiAPY-cli-library 
-    code_library_usage_prediction = rag_chain.predict_library_usage(query)
-    print(f"Predicted library usage: {code_library_usage_prediction}")
-    
-    rewritten_query = rag_chain.rewrite_query(query)
-    print(f"\n\n Rewritten query: {rewritten_query}\n\n")
-
-    """
-    Retrieval-Stage
-    """  
-    kadiAPY_doc_documents = rag_chain.retrieve_contexts(query, k=5, filter={"usage": "doc"})
-    kadiAPY_code_documents = rag_chain.retrieve_contexts(str(rewritten_query.content), k=3, filter={"usage": code_library_usage_prediction})
-    
-    print("Retrieved Document Contexts:", kadiAPY_doc_documents)
-    print("Retrieved Code Contexts:", kadiAPY_code_documents)
-
-
-    
-    """
-    Pre-Generation-Stage
-    Adding each doc's metadata to the retrieved content (docs & code snippets) 
-    """
-    formatted_doc_snippets = rag_chain.format_documents(kadiAPY_doc_documents)
-    formatted_code_snippets = rag_chain.format_documents(kadiAPY_code_documents)
-    #print("FORMATTED Retrieved Document Contexts:", formatted_doc_snippets)
-    #print("FORMATTED Retrieved Code Contexts:" , formatted_code_snippets)
-
-    """
-    Generation-Stage
-    """
-    response = rag_chain.generate_response(query, formatted_doc_snippets, formatted_code_snippets)
-    print("Generated Response:", response)
-
-    return response
-
     
 def bot_kadi(history):
-    user_query = history[-1][0]
-    response = rag_workflow(user_query)
+    user_query = history[-1][0]   
+    response = kadiAPY_Bot.process_query(user_query)
     history[-1] = (user_query, response)
 
     yield history