Update handle text and generate

app.py

@@ -91,6 +91,16 @@ def get_chat_response(chatbot, task_history):
 
 def handle_text_input(history, task_history, text):
     """Handle text input from the user."""
+    # Überprüfen, ob das Eingabefeld leer ist
+    if not text:
+        # Wenn das Eingabefeld leer ist, senden Sie eine vordefinierte Anfrage
+        text = "Describe the image for me..."
+        # Aktualisieren Sie das Chat- und Task-Verlauf mit der vordefinierten Anfrage
+        history = history + [(format_text(text), None)]
+        task_history = task_history + [(text, None)]
+        # Rufen Sie get_chat_response auf, um eine Antwort zu generieren
+        return get_chat_response(history, task_history)
+
     task_text = text
     if len(text) >= 2 and text[-1] in PUNCTUATION and text[-2] not in PUNCTUATION:
         task_text = text[:-1]
@@ -113,22 +123,33 @@ def clear_history(task_history):
     task_history.clear()
     return []
 
-def handle_regeneration(chatbot, task_history):
+def handle_regeneration(chatbot, task_history, input_field):
     """Handle the regeneration of the last response."""
     print("Regenerate clicked")
     print("Before:", task_history, chatbot)
     if not task_history:
         return chatbot
-    item = task_history[-1]
-    if item[1] is None:
-        return chatbot
-    task_history[-1] = (item[0], None)
-    chatbot_item = chatbot.pop(-1)
-    if chatbot_item[0] is None:
-        chatbot[-1] = (chatbot[-1][0], None)
+    
+    # Überprüfen, ob das Eingabefeld leer ist
+    if not input_field.value:
+        # Wenn das Eingabefeld leer ist, senden Sie eine vordefinierte Anfrage
+        predefined_query = "Describe this image for me..."
+        # Aktualisieren Sie das Eingabefeld mit der vordefinierten Anfrage
+        input_field.update(value=predefined_query)
+        # Führen Sie die normale Texteingabebehandlung durch
+        handle_text_input(chatbot, task_history, predefined_query)
     else:
-        chatbot.append((chatbot_item[0], None))
-    print("After:", task_history, chatbot)
+        item = task_history[-1]
+        if item[1] is None:
+            return chatbot
+        task_history[-1] = (item[0], None)
+        chatbot_item = chatbot.pop(-1)
+        if chatbot_item[0] is None:
+            chatbot[-1] = (chatbot[-1][0], None)
+        else:
+            chatbot.append((chatbot_item[0], None))
+        print("After:", task_history, chatbot)
+    
     return get_chat_response(chatbot, task_history)
 
 # Custom CSS
@@ -194,14 +215,14 @@ with gr.Blocks(css=css) as demo:
     
     gr.Markdown("### Key Features:\n- **Strong Performance**: Surpasses existing LVLMs on multiple English benchmarks including Zero-shot Captioning and VQA.\n- **Multi-lingual Support**: Supports English, Chinese, and multi-lingual conversation.\n- **High Resolution**: Utilizes 448*448 resolution for fine-grained recognition and understanding.")
     submit_btn.click(handle_text_input, [chatbot, task_history, query], [chatbot, task_history]).then(
-        get_chat_response, [chatbot, task_history], [chatbot], show_progress=True
-    )
+    get_chat_response, [chatbot, task_history], [chatbot], show_progress=True
+)
 
 
 
     submit_btn.click(clear_input, [], [query])
     clear_btn.click(clear_history, [task_history], [chatbot], show_progress=True)
-    regen_btn.click(handle_regeneration, [chatbot, task_history], [chatbot], show_progress=True)
+    regen_btn.click(handle_regeneration, [chatbot, task_history, query], [chatbot], show_progress=True)
     upload_btn.upload(handle_file_upload, [chatbot, task_history, upload_btn], [chatbot, task_history], show_progress=True)