Update app.py

app.py

@@ -328,52 +328,17 @@ textarea::placeholder {
 }
 """
 
-# def process_question(question: str) -> Iterator[str]:
-#     """
-#     Process the question and return a response generator for streaming.
-#     """
-#     if question in question_cache:
-#         yield question_cache[question][0]
-#         return
-
-#     relevant_docs = retriever(question)
-#     context = "\n".join([doc.page_content for doc in relevant_docs])
-
-#     prompt = prompt_template.format_messages(
-#         context=context,
-#         question=question
-#     )
-
-#     full_response = ""
-#     try:
-#         for chunk in llm.stream(prompt):
-#             if isinstance(chunk, str):
-#                 current_chunk = chunk
-#             else:
-#                 current_chunk = chunk.content
-
-#             full_response += current_chunk
-#             yield full_response  # Send the updated response in streaming
-
-#         question_cache[question] = (full_response, context)
-#     except Exception as e:
-#         yield f"Erreur lors du traitement : {str(e)}"
-
 def process_question(question: str) -> Iterator[str]:
     """
-    Process the question and return a response generator for streaming, including sources.
+    Process the question and return a response generator for streaming.
     """
     if question in question_cache:
         yield question_cache[question][0]
         return
 
-    # Récupérer les documents pertinents
     relevant_docs = retriever(question)
     context = "\n".join([doc.page_content for doc in relevant_docs])
-    sources = [doc.metadata.get("source", "Source inconnue") for doc in relevant_docs]
-    sources = os.path.splitext(sources[0])[0] if sources else "غير معروف"
 
-    # Générer le prompt
     prompt = prompt_template.format_messages(
         context=context,
         question=question
@@ -381,7 +346,6 @@ def process_question(question: str) -> Iterator[str]:
 
     full_response = ""
     try:
-        # Streaming de la réponse
         for chunk in llm.stream(prompt):
             if isinstance(chunk, str):
                 current_chunk = chunk
@@ -389,19 +353,55 @@ def process_question(question: str) -> Iterator[str]:
                 current_chunk = chunk.content
 
             full_response += current_chunk
-            yield full_response  # Envoyer la réponse mise à jour en streaming
-
-        # Ajouter les sources à la réponse finale
-        if sources:
-            sources_str = "\nSources :\n" + "\n".join(f"- {source}" for source in sources)
-            full_response += sources_str
-            yield sources_str  # Envoyer les sources
+            yield full_response  # Send the updated response in streaming
 
-        # Mettre en cache la réponse complète
         question_cache[question] = (full_response, context)
     except Exception as e:
         yield f"Erreur lors du traitement : {str(e)}"
 
+# def process_question(question: str) -> Iterator[str]:
+#     """
+#     Process the question and return a response generator for streaming, including sources.
+#     """
+#     if question in question_cache:
+#         yield question_cache[question][0]
+#         return
+
+#     # Récupérer les documents pertinents
+#     relevant_docs = retriever(question)
+#     context = "\n".join([doc.page_content for doc in relevant_docs])
+#     sources = [doc.metadata.get("source", "Source inconnue") for doc in relevant_docs]
+#     sources = os.path.splitext(sources[0])[0] if sources else "غير معروف"
+
+#     # Générer le prompt
+#     prompt = prompt_template.format_messages(
+#         context=context,
+#         question=question
+#     )
+
+#     full_response = ""
+#     try:
+#         # Streaming de la réponse
+#         for chunk in llm.stream(prompt):
+#             if isinstance(chunk, str):
+#                 current_chunk = chunk
+#             else:
+#                 current_chunk = chunk.content
+
+#             full_response += current_chunk
+#             yield full_response  # Envoyer la réponse mise à jour en streaming
+
+#         # Ajouter les sources à la réponse finale
+#         if sources:
+#             sources_str = "\nSources :\n" + "\n".join(f"- {source}" for source in sources)
+#             full_response += sources_str
+#             yield sources_str  # Envoyer les sources
+
+#         # Mettre en cache la réponse complète
+#         question_cache[question] = (full_response, context)
+#     except Exception as e:
+#         yield f"Erreur lors du traitement : {str(e)}"
+
 
 # def process_question(question: str) -> tuple[str, list[str]]:
 #     # Check cache first
@@ -440,44 +440,44 @@ def process_question(question: str) -> Iterator[str]:
 
 
 
-# def gradio_stream(question: str, chat_history: list) -> Iterator[list]:
-#     """
-#     Format the output for Gradio Chatbot component with streaming.
-#     """
-#     full_response = ""
-#     try:
-#         for partial_response in process_question(question):
-#             full_response = partial_response
-#             # Append the latest assistant response to chat history
-#             updated_chat = chat_history + [[question, partial_response]]
-#             yield updated_chat
-#     except Exception as e:
-#         # Handle errors during streaming
-#         updated_chat = chat_history + [[question, f"Erreur : {str(e)}"]]
-#         yield updated_chat
-
 def gradio_stream(question: str, chat_history: list) -> Iterator[list]:
     """
-    Format the output for Gradio Chatbot component with streaming, including sources.
+    Format the output for Gradio Chatbot component with streaming.
     """
     full_response = ""
-    sources_str = ""
     try:
         for partial_response in process_question(question):
-            if "Sources :" in partial_response:
-                # Les sources sont ajoutées à la réponse finale
-                sources_str = partial_response
-                updated_chat = chat_history + [[question, full_response + "\n" + sources_str]]
-            else:
-                # Construire progressivement la réponse
-                full_response = partial_response
-                updated_chat = chat_history + [[question, full_response]]
+            full_response = partial_response
+            # Append the latest assistant response to chat history
+            updated_chat = chat_history + [[question, partial_response]]
             yield updated_chat
     except Exception as e:
-        # Gestion des erreurs lors du streaming
+        # Handle errors during streaming
         updated_chat = chat_history + [[question, f"Erreur : {str(e)}"]]
         yield updated_chat
 
+# def gradio_stream(question: str, chat_history: list) -> Iterator[list]:
+#     """
+#     Format the output for Gradio Chatbot component with streaming, including sources.
+#     """
+#     full_response = ""
+#     sources_str = ""
+#     try:
+#         for partial_response in process_question(question):
+#             if "Sources :" in partial_response:
+#                 # Les sources sont ajoutées à la réponse finale
+#                 sources_str = partial_response
+#                 updated_chat = chat_history + [[question, full_response + "\n" + sources_str]]
+#             else:
+#                 # Construire progressivement la réponse
+#                 full_response = partial_response
+#                 updated_chat = chat_history + [[question, full_response]]
+#             yield updated_chat
+#     except Exception as e:
+#         # Gestion des erreurs lors du streaming
+#         updated_chat = chat_history + [[question, f"Erreur : {str(e)}"]]
+#         yield updated_chat
+
 
 # Gradio interface
 with gr.Blocks(css=css) as demo: