Update app.py

app.py

@@ -226,114 +226,6 @@ prompt_template = ChatPromptTemplate.from_messages([
 ])
 
 
-
-# def process_question(question: str):  
-#     """  
-#     Process the question and yield the answer progressively.  
-#     """  
-#     # Check cache first  
-#     if question in question_cache:  
-#         yield question_cache[question]  # Retourne directement depuis le cache si disponible  
-
-#     relevant_docs = retriever(question)  
-#     context = "\n".join([doc.page_content for doc in relevant_docs])  
-
-#     prompt = prompt_template.format_messages(  
-#         context=context,  
-#         question=question  
-#     )  
-
-#     response = ""  # Initialise la réponse  
-#     # Ici, nous supposons que 'llm.stream' est un générateur qui renvoie des chunks  
-#     for chunk in llm.stream(prompt):  # suppose que llm.stream renvoie des chunks de réponse  
-#         if isinstance(chunk, str):  
-#             response += chunk  # Accumulez la réponse si c'est déjà une chaîne  
-#         else:  
-#             response += chunk.content  # Sinon, prenez le contenu du chunk (si chunk est un type d'objet spécifique)  
-
-#         yield response, context  # Renvoie la réponse mise à jour et le contexte  
-
-#     # Mettez le résultat en cache à la fin  
-#     question_cache[question] = (response, context)   
-
-# # Custom CSS for right-aligned text in textboxes  
-# custom_css = """  
-# .rtl-text {  
-#     text-align: right !important;  
-#     direction: rtl !important;  
-# }  
-# .rtl-text textarea {  
-#     text-align: right !important;  
-#     direction: rtl !important;  
-# }  
-# """  
-
-# # Gradio interface with queue  
-# with gr.Blocks(css=custom_css) as iface:  
-#     with gr.Column():  
-#         input_text = gr.Textbox(  
-#             label="السؤال",  
-#             placeholder="اكتب سؤالك هنا...",  
-#             lines=2,  
-#             elem_classes="rtl-text"  
-#         )  
-        
-#         with gr.Row():  
-#             answer_box = gr.Textbox(  
-#                 label="الإجابة",  
-#                 lines=4,  
-#                 elem_classes="rtl-text"  
-#             )  
-#             context_box = gr.Textbox(  
-#                 label="السياق المستخدم",  
-#                 lines=8,  
-#                 elem_classes="rtl-text"  
-#             )  
-        
-#         submit_btn = gr.Button("إرسال")  
-        
-#         submit_btn.click(  
-#             fn=process_question,  
-#             inputs=input_text,  
-#             outputs=[answer_box, context_box],  
-#             api_name="predict",  
-#             queue=True  # Utiliser le système de queue pour un traitement asynchrone  
-#         )  
-
-# if __name__ == "__main__":  
-#     iface.launch(  
-#         share=True,  
-#         server_name="0.0.0.0",  
-#         server_port=7860,  
-#         max_threads=3,  # Controls concurrency  
-#         show_error=True  
-#     )
-
-
-
-# def process_question(question: str):  
-#     """  
-#     Process the question and return the answer and context
-#     """  
-#     # Check cache first  
-#     if question in question_cache:  
-#         return question_cache[question], ""  # Retourne la réponse cachée et un statut vide
-#     relevant_docs = retriever(question)  
-#     context = "\n".join([doc.page_content for doc in relevant_docs])  
-#     prompt = prompt_template.format_messages(  
-#         context=context,  
-#         question=question  
-#         )  
-#     response = ""
-#     for chunk in llm.stream(prompt):
-#         if isinstance(chunk, str):  
-#             response += chunk
-#         else:  
-#             response += chunk.content
-#     # Mettez le résultat en cache à la fin  
-#     question_cache[question] = (response, context)  
-#     return response, context
-
 def process_question(question: str):  
     """  
     Process the question and yield the answer progressively.  
@@ -488,18 +380,18 @@ with gr.Blocks(css=custom_css) as iface:
             )
         
         with gr.Row():
-            with gr.Column(scale=2):
+            with gr.Column():
                 answer_box = gr.Textbox(
                     label="الإجابة",
                     lines=4,
                     elem_classes="rtl-text textbox-container"
                 )
-            with gr.Column(scale=1):
-                context_box = gr.Textbox(
-                    label="السياق المستخدم",
-                    lines=4,
-                    elem_classes="rtl-text textbox-container"
-                )
+            # with gr.Column(scale=1):
+            #     context_box = gr.Textbox(
+            #         label="السياق المستخدم",
+            #         lines=4,
+            #         elem_classes="rtl-text textbox-container"
+            #     )
         
         submit_btn = gr.Button(
             "إرسال السؤال",
@@ -508,18 +400,6 @@ with gr.Blocks(css=custom_css) as iface:
         )
         
 
-        # def on_submit(question):
-        #     response = ""
-        #     context = ""
-        #     for chunk in process_question(question):
-        #         if isinstance(chunk, tuple):
-        #             chunk_content = chunk[0]  # Assurez-vous de prendre la bonne partie du tuple
-        #         else:
-        #             chunk_content = chunk
-                
-        #         response += chunk_content
-        #         yield [(question, response), context]  # Renvoie la réponse mise à jour et le contexte
-
         def on_submit(question):
             # Appeler process_question pour obtenir la réponse complète et le contexte
             response = ""
@@ -530,20 +410,16 @@ with gr.Blocks(css=custom_css) as iface:
             
             yield response, context  # Renvoie la réponse et le contexte final après la fin de la boucle
 
-
-
-
         
         submit_btn.click(
             fn=on_submit,
             inputs=input_text,
-            outputs=[answer_box, context_box],
+            outputs=answer_box
             api_name="predict",
             queue=True
 )
 
 
-
 if __name__ == "__main__":
     iface.launch(
         share=True,
@@ -553,3 +429,110 @@ if __name__ == "__main__":
         show_error=True
     )
 
+# def process_question(question: str):  
+#     """  
+#     Process the question and yield the answer progressively.  
+#     """  
+#     # Check cache first  
+#     if question in question_cache:  
+#         yield question_cache[question]  # Retourne directement depuis le cache si disponible  
+
+#     relevant_docs = retriever(question)  
+#     context = "\n".join([doc.page_content for doc in relevant_docs])  
+
+#     prompt = prompt_template.format_messages(  
+#         context=context,  
+#         question=question  
+#     )  
+
+#     response = ""  # Initialise la réponse  
+#     # Ici, nous supposons que 'llm.stream' est un générateur qui renvoie des chunks  
+#     for chunk in llm.stream(prompt):  # suppose que llm.stream renvoie des chunks de réponse  
+#         if isinstance(chunk, str):  
+#             response += chunk  # Accumulez la réponse si c'est déjà une chaîne  
+#         else:  
+#             response += chunk.content  # Sinon, prenez le contenu du chunk (si chunk est un type d'objet spécifique)  
+
+#         yield response, context  # Renvoie la réponse mise à jour et le contexte  
+
+#     # Mettez le résultat en cache à la fin  
+#     question_cache[question] = (response, context)   
+
+# # Custom CSS for right-aligned text in textboxes  
+# custom_css = """  
+# .rtl-text {  
+#     text-align: right !important;  
+#     direction: rtl !important;  
+# }  
+# .rtl-text textarea {  
+#     text-align: right !important;  
+#     direction: rtl !important;  
+# }  
+# """  
+
+# # Gradio interface with queue  
+# with gr.Blocks(css=custom_css) as iface:  
+#     with gr.Column():  
+#         input_text = gr.Textbox(  
+#             label="السؤال",  
+#             placeholder="اكتب سؤالك هنا...",  
+#             lines=2,  
+#             elem_classes="rtl-text"  
+#         )  
+        
+#         with gr.Row():  
+#             answer_box = gr.Textbox(  
+#                 label="الإجابة",  
+#                 lines=4,  
+#                 elem_classes="rtl-text"  
+#             )  
+#             context_box = gr.Textbox(  
+#                 label="السياق المستخدم",  
+#                 lines=8,  
+#                 elem_classes="rtl-text"  
+#             )  
+        
+#         submit_btn = gr.Button("إرسال")  
+        
+#         submit_btn.click(  
+#             fn=process_question,  
+#             inputs=input_text,  
+#             outputs=[answer_box, context_box],  
+#             api_name="predict",  
+#             queue=True  # Utiliser le système de queue pour un traitement asynchrone  
+#         )  
+
+# if __name__ == "__main__":  
+#     iface.launch(  
+#         share=True,  
+#         server_name="0.0.0.0",  
+#         server_port=7860,  
+#         max_threads=3,  # Controls concurrency  
+#         show_error=True  
+#     )
+
+
+
+# def process_question(question: str):  
+#     """  
+#     Process the question and return the answer and context
+#     """  
+#     # Check cache first  
+#     if question in question_cache:  
+#         return question_cache[question], ""  # Retourne la réponse cachée et un statut vide
+#     relevant_docs = retriever(question)  
+#     context = "\n".join([doc.page_content for doc in relevant_docs])  
+#     prompt = prompt_template.format_messages(  
+#         context=context,  
+#         question=question  
+#         )  
+#     response = ""
+#     for chunk in llm.stream(prompt):
+#         if isinstance(chunk, str):  
+#             response += chunk
+#         else:  
+#             response += chunk.content
+#     # Mettez le résultat en cache à la fin  
+#     question_cache[question] = (response, context)  
+#     return response, context
+