update app and requirements for API documentation chatbot

app.py

@@ -1,64 +1,129 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import os
+from langchain.chains import ConversationalRetrievalChain
+from langchain.memory import ConversationBufferMemory
+from langchain_openai import ChatOpenAI, OpenAIEmbeddings
+from langchain.prompts import PromptTemplate
+from langchain_community.vectorstores import Chroma
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+def create_qa_chain():
+    """
+    Create the QA chain with the loaded vectorstore
+    """
+    # Initialize embeddings and load vectorstore
+    embeddings = OpenAIEmbeddings()
+    vectorstore = Chroma(
+        persist_directory="./vectorstore",
+        embedding_function=embeddings
+    )
+    
+    # Set up retriever
+    retriever = vectorstore.as_retriever(
+        search_type="mmr",
+        search_kwargs={
+            "k": 6,
+            "fetch_k": 20,
+            "lambda_mult": 0.3,
+        }
+    )
+    
+    # Set up memory
+    memory = ConversationBufferMemory(
+        memory_key="chat_history",
+        return_messages=True,
+        output_key='answer'
+    )
+    
+    # Create prompt template
+    qa_prompt = PromptTemplate.from_template("""You are an expert technical writer specializing in API documentation. 
+When describing API endpoints, structure your response in this exact format:
 
+1. Start with the HTTP method and base URI structure
+2. List all key parameters with:
+   - Parameter name in bold (**parameter**)
+   - Type and requirement status
+   - Clear description
+   - Example values where applicable
+3. Show complete example requests with:
+   - Basic example
+   - Full example with all parameters
+   - Headers included
+4. Include any relevant response information
+   
+Use markdown formatting for:
+- Code blocks with syntax highlighting
+- Bold text for important terms
+- Clear section separation
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+Context: {context}
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+Question: {question}
 
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
+Technical answer (following the exact structure above):""")
+    
+    # Create the chain
+    qa_chain = ConversationalRetrievalChain.from_llm(
+        llm=ChatOpenAI(
+            temperature=0.1,
+            model_name="gpt-4-turbo-preview"
+        ),
+        retriever=retriever,
+        memory=memory,
+        return_source_documents=True,
+        combine_docs_chain_kwargs={"prompt": qa_prompt},
+        verbose=False
+    )
+    
+    return qa_chain
 
+def chat(message, history):
+    """
+    Process chat messages and return responses
+    """
+    # Get or create QA chain
+    if not hasattr(chat, 'qa_chain'):
+        chat.qa_chain = create_qa_chain()
+    
+    # Get response
+    result = chat.qa_chain({"question": message})
+    
+    # Format sources
+    sources = "\n\nSources:\n"
+    seen_components = set()
+    shown_sources = 0
+    
+    for doc in result["source_documents"]:
+        component = doc.metadata.get('component', '')
+        title = doc.metadata.get('title', '')
+        combo = (component, title)
+        
+        if combo not in seen_components and shown_sources < 3:
+            seen_components.add(combo)
+            shown_sources += 1
+            sources += f"\nSource {shown_sources}:\n"
+            sources += f"Title: {title}\n"
+            sources += f"Component: {component}\n"
+            sources += f"Content: {doc.page_content[:300]}...\n"
+    
+    # Combine response with sources
+    full_response = result["answer"] + sources
+    
+    return full_response
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# Create and launch the Gradio interface
 demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+    chat,
+    title="Apple Music API Documentation Assistant",
+    description="Ask questions about the Apple Music API documentation.",
+    examples=[
+        "How to search for songs on Apple Music API?",
+        "What are the required parameters for searching songs?",
+        "Show me an example request with all parameters"
     ],
+    retry_btn=None,
+    undo_btn=None,
+    clear_btn="Clear Chat"
 )
 
-
 if __name__ == "__main__":
     demo.launch()

requirements.txt

@@ -1 +1,7 @@
-huggingface_hub==0.25.2
+huggingface_hub==0.25.2
+gradio
+langchain
+langchain-openai
+langchain-community
+openai
+chromadb