Update app.py

app.py

@@ -2,26 +2,26 @@ import os
 import gradio as gr
 from langchain_huggingface import HuggingFaceEmbeddings, HuggingFaceEndpoint
 from langchain_community.vectorstores import FAISS
-from langchain.chains import create_retrieval_chain
-from langchain.chains.combine_documents import create_stuff_documents_chain
 from langchain_core.prompts import ChatPromptTemplate
+from langchain_core.runnables import RunnablePassthrough
+from langchain_core.output_parsers import StrOutputParser
 
-# 1. Load the pre-computed Vector Database
+# 1. Load the FULL pre-computed Vector Database
 embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
-# Allow dangerous deserialization is required for local FAISS loading
-vectorstore = FAISS.load_local("learncpp_faiss_index", embeddings, allow_dangerous_deserialization=True)
+vectorstore = FAISS.load_local("learncpp_faiss_index_full", embeddings, allow_dangerous_deserialization=True)
+retriever = vectorstore.as_retriever(search_kwargs={"k": 5})
 
-# 2. Set up the LLM (Using Mistral-7B for excellent coding logic)
+# 2. Set up the LLM
 hf_token = os.environ.get("HF_TOKEN")
 llm = HuggingFaceEndpoint(
-    repo_id="mistralai/Mistral-7B-Instruct-v0.2",
+    repo_id="HuggingFaceH4/zephyr-7b-beta",
     task="text-generation",
     max_new_tokens=512,
     temperature=0.1,
     huggingfacehub_api_token=hf_token
 )
 
-# 3. Create the RAG Chain
+# 3. Create the RAG Chain using modern LCEL (No langchain.chains needed!)
 system_prompt = (
     "You are an expert C++ programming assistant. You answer questions strictly based "
     "on the provided context from learncpp.com. If the answer is not in the context, "
@@ -34,22 +34,41 @@ prompt = ChatPromptTemplate.from_messages([
     ("human", "{input}"),
 ])
 
-qa_chain = create_stuff_documents_chain(llm, prompt)
-rag_chain = create_retrieval_chain(vectorstore.as_retriever(search_kwargs={"k": 3}), qa_chain)
+# Helper function to extract text from our vector chunks
+def format_docs(docs):
+    return "\n\n".join(doc.page_content for doc in docs)
 
-# 4. Define the Chat Interface
+# The modern LangChain pipeline
+rag_chain = (
+    {"context": retriever | format_docs, "input": RunnablePassthrough()}
+    | prompt
+    | llm
+    | StrOutputParser()
+)
+
+# 4. Define the Chat Interface WITH Memory
 def chat_function(message, history):
     try:
-        response = rag_chain.invoke({"input": message})
-        return response["answer"]
+        formatted_input = ""
+        if history:
+            formatted_input += "Previous Conversation Context:\n"
+            for user_msg, bot_msg in history:
+                formatted_input += f"User: {user_msg}\nBot: {bot_msg}\n"
+            formatted_input += "\n"
+        
+        formatted_input += f"Current Question: {message}"
+
+        # Invoke the LCEL chain
+        response = rag_chain.invoke(formatted_input)
+        return response
     except Exception as e:
         return f"Error: {str(e)}"
 
 demo = gr.ChatInterface(
     fn=chat_function,
     title="LearnCpp.com AI Assistant",
-    description="Ask me any C++ question! I retrieve my answers directly from the LearnCpp tutorials.",
-    examples=["What is a pointer?", "Explain dynamic memory allocation."]
+    description="Ask me any C++ question! I retrieve my answers directly from the complete LearnCpp tutorials.",
+    examples=["What is a pointer?", "Explain dynamic memory allocation.", "Give me an example of the previous concept."]
 )
 
 if __name__ == "__main__":