.

app.py

@@ -153,7 +153,7 @@ class EnhancedRetriever:
         return vector_store
 
     @lru_cache(maxsize=500)
-    def retrieve(self, query: str) -> str:
+    def retrieve(self, query: str) -> Tuple[str, List[Document]]:
         try:
             processed_query = self._preprocess_query(query)
             expanded_query = self._hyde_expansion(processed_query)
@@ -163,10 +163,12 @@ class EnhancedRetriever:
             expanded_results = self.bm25.invoke(expanded_query)
             
             fused_results = self._fuse_results([bm25_results, vector_results, expanded_results])
-            return self._format_context(fused_results[:5])
+            top_docs = fused_results[:5]
+            formatted_context = self._format_context(top_docs)
+            return formatted_context, top_docs
         except Exception as e:
             logger.error(f"Retrieval Error: {str(e)}")
-            return ""
+            return "", []
 
     def _preprocess_query(self, query: str) -> str:
         return query.lower().strip()
@@ -228,7 +230,6 @@ SYSTEM_PROMPT = """
 Context: {context}
 """
 
-
 @retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=2, max=20))
 def get_ai_response(query: str, context: str, model: str) -> str:
     result = ""  # Initialize the result variable
@@ -262,15 +263,14 @@ def get_ai_response(query: str, context: str, model: str) -> str:
             if result is None:
                 result = "Failed to get response from llama3-70b-8192"
         # Append the model name to the response for clarity
-        # get the key name model from model mapping
         for key, value in model_mapping.items():
             if value == model:
                 model = key
-        result += f"\n\n**Model:** {model}"
+        # result += f"\n\n**Model:** {model}"
         return result
     except Exception as e:
         logger.error(f"Generation Error: {str(e)}")
-        return "I'm unable to generate a response right now. Please try again later."
+        return "I'm unable to generate a response right now. Please try again later or try another model."
 
 def _postprocess_response(response: str) -> str:
     response = re.sub(r"\[(.*?)\]", r"[\1](#)", response)
@@ -306,20 +306,38 @@ def get_groq_llama3_response(query: str) -> str:
     except requests.exceptions.RequestException as e:
         logger.error(f"Groq API Error: {str(e)}")
         return "An error occurred while contacting Groq's Llama 3 model."
+
 # --- Pipeline ---
 documents = load_and_chunk_data(data_file_name)
 retriever = EnhancedRetriever(documents)
 
 def generate_response(question: str, model: str) -> str:
     try:
-        context = retriever.retrieve(question)
-        return get_ai_response(question, context, model) if context else "No relevant information found."
+        formatted_context, retrieved_docs = retriever.retrieve(question)
+        if not formatted_context:
+            return "No relevant information found."
+        response = get_ai_response(question, formatted_context, model)
+        # Extract references from retrieved documents whose hyperlinks start with "https://asknature.org"
+        ref_links = []
+        for doc in retrieved_docs:
+            hyperlink = doc.metadata.get("hyperlink", "")
+            if hyperlink.startswith("https://asknature.org") and hyperlink not in ref_links:
+                ref_links.append(hyperlink)
+        if ref_links:
+            references_md = "\n\n**References:**\n"
+            for i, link in enumerate(ref_links, 1):
+                references_md += f"[{i}] {link}\n"
+            response += references_md
+        for key, value in model_mapping.items():
+            if value == model:
+                model = key
+        response += f"\n\n**Model:** {model}"
+        return response
     except Exception as e:
         logger.error(f"Pipeline Error: {str(e)}")
         return "An error occurred processing your request."
 
 # --- Gradio Interface ---
-# Define the mapping from display names to actual model identifiers
 model_mapping = {
     "Gemini-2.0-Flash": "gemini-2.0-flash",
     "Meta-llama-3-70b-instruct(GWDG)": "meta-llama-3-70b-instruct",
@@ -327,7 +345,7 @@ model_mapping = {
 }
 
 def chat_interface(question: str, history: List[Tuple[str, str]], display_model: str):
-    model = model_mapping.get(display_model, "gemini-2.0-flash")  # Default to Gemini if not found
+    model = model_mapping.get(display_model, "gemini-2.0-flash")
     response = generate_response(question, model)
     return "", history + [(question, response)]