Update app.py

app.py

@@ -943,7 +943,7 @@ async def chat_input_callback(*args):
         return state, state, None, audio,log_state,history
 
         
-async def upload_callback(image_input,state, log_state, visual_chatgpt=None, openai_api_key=None,language="English",narritive=None,history=None,autoplay=False,session="Session 1"):
+async def upload_callback(image_input,state, log_state, task_type, visual_chatgpt=None, openai_api_key=None,language="English",narritive=None,history=None,autoplay=False,session="Session 1"):
     print("narritive", narritive)
     if isinstance(image_input, dict):  # if upload from sketcher_input, input contains image and mask
         image_input = image_input['background']
@@ -993,11 +993,18 @@ async def upload_callback(image_input,state, log_state, visual_chatgpt=None, ope
         visual_chatgpt.current_image = new_image_path
         paragraph = get_gpt_response(openai_api_key, new_image_path,f"What's going on in this picture? in {language}")
         # visual_chatgpt.agent.memory.buffer = visual_chatgpt.agent.memory.buffer + visual_chatgpt.global_prompt 
-        parsed_data = get_gpt_response(openai_api_key, new_image_path,"Please provide the name, artist, year of creation (including the art historical period), and painting style used for this painting. Return the information in dictionary format without any newline characters. Format as follows: { \"name\": \"Name of the painting\", \"artist\": \"Name of the artist\", \"year\": \"Year of creation (Art historical period)\", \"style\": \"Painting style used in the painting\",\"gender\": \"The gender of the author\"}")
-        print(parsed_data)
-        parsed_data = json.loads(parsed_data.replace("'", "\""))
-        name, artist, year, material,gender= parsed_data["name"],parsed_data["artist"],parsed_data["year"], parsed_data["style"],parsed_data['gender']
-        gender=gender.lower()
+        if task_type=="task 3":
+            name="Along the River During the Qingming Festival"
+            artist="Zhang Zeduan"
+            year="12th century (Song Dynasty)"
+            material="Chinese painting"
+            gender="male"
+        else:
+            parsed_data = get_gpt_response(openai_api_key, new_image_path,"Please provide the name, artist, year of creation (including the art historical period), and painting style used for this painting. Return the information in dictionary format without any newline characters. Format as follows: { \"name\": \"Name of the painting\", \"artist\": \"Name of the artist\", \"year\": \"Year of creation (Art historical period)\", \"style\": \"Painting style used in the painting\",\"gender\": \"The gender of the author\"}")
+            print(parsed_data)
+            parsed_data = json.loads(parsed_data.replace("'", "\""))
+            name, artist, year, material,gender= parsed_data["name"],parsed_data["artist"],parsed_data["year"], parsed_data["style"],parsed_data['gender']
+            gender=gender.lower()
         print("gender",gender)
         
         
@@ -1668,7 +1675,7 @@ async def texttospeech(text, language,gender='female'):
         return None
 
 # give the reason of recommendation
-async def item_associate(new_crop,openai_api_key,language,autoplay,length,log_state,sort_score,narritive,state,evt: gr.SelectData):
+async def item_associate(new_crop,openai_api_key,language,autoplay,length,log_state,sort_score,narritive,state,history_log, evt: gr.SelectData):
     persona=naritive_mapping[narritive]
     rec_path=evt._data['value']['image']['path']
     index="Item Recommendation Picture "+ str(evt.index)
@@ -1677,9 +1684,10 @@ async def item_associate(new_crop,openai_api_key,language,autoplay,length,log_st
     
     image_paths=[new_crop,rec_path]
 
-    result=get_gpt_response(openai_api_key, image_paths, prompt)
+    result=get_gpt_response(openai_api_key, image_paths, prompt,history_log)
     print("recommend result",result)
     state += [(None, f"{result}")]
+    log_state += [("User wants to know object recomendation reason", None)]
     log_state = log_state + [(narritive, None)]
     log_state = log_state + [(f"image sort ranking {sort_score}", None)]
     log_state = log_state + [(None, f"{result}")]
@@ -1692,16 +1700,17 @@ async def item_associate(new_crop,openai_api_key,language,autoplay,length,log_st
     return state,state,audio_output,log_state,index,gr.update(value=[])
 
 
-async def style_associate(image_path,openai_api_key,language,autoplay,length,log_state,sort_score,narritive,state,evt: gr.SelectData):
+async def style_associate(image_path,openai_api_key,language,autoplay,length,log_state,sort_score,narritive,state,history_log,evt: gr.SelectData):
     persona=naritive_mapping[narritive]
     rec_path=evt._data['value']['image']['path']
     index="Style Recommendation Picture "+str(evt.index)
     print("rec_path",rec_path)
     prompt=recommendation_prompt[1][persona].format(language=language,length=length)
     image_paths=[image_path,rec_path]
-    result=get_gpt_response(openai_api_key, image_paths, prompt)
+    result=get_gpt_response(openai_api_key, image_paths, prompt,history_log)
     print("recommend result",result)
     state += [(None, f"{result}")]
+    log_state += [("User wants to know style recomendation reason", None)]
     log_state = log_state + [(narritive, None)]
     log_state = log_state + [(f"image sort ranking {sort_score}", None)]
     log_state = log_state + [(None, f"{result}")]
@@ -1866,7 +1875,7 @@ def create_ui():
             )
         with gr.Row():  
             with gr.Column(scale=1,min_width=50,visible=False) as instruct:
-                task_instuction=gr.Image(type="pil", interactive=False, elem_classes="task_instruct",height=650,label=None)                    
+                task_instuction=gr.Image(type="pil", interactive=False, elem_classes="task_instruct",height=650,label="Instruction")                    
             with gr.Column(scale=6):
                 with gr.Column(visible=False) as modules_not_need_gpt:
                     
@@ -2275,7 +2284,7 @@ def create_ui():
 
         gallery_result.select(
             item_associate,
-            inputs=[new_crop_save_path,openai_api_key,language,auto_play,length,log_state,sort_rec,naritive,recomended_state],
+            inputs=[new_crop_save_path,openai_api_key,language,auto_play,length,log_state,sort_rec,naritive,recomended_state,history_log],
             outputs=[recommend_bot,recomended_state,output_audio,log_state,pic_index,recommend_score],
             
             
@@ -2283,7 +2292,7 @@ def create_ui():
         
         style_gallery_result.select(
             style_associate,
-            inputs=[image_path,openai_api_key,language,auto_play,length,log_state,sort_rec,naritive,recomended_state],
+            inputs=[image_path,openai_api_key,language,auto_play,length,log_state,sort_rec,naritive,recomended_state,history_log],
             outputs=[recommend_bot,recomended_state,output_audio,log_state,pic_index,recommend_score],
             
             
@@ -2571,7 +2580,7 @@ def create_ui():
         #                     name_label_base2, artist_label_base2, year_label_base2, material_label_base2,name_label_traj, artist_label_traj, year_label_traj, material_label_traj, \
         #                         paragraph,artist,gender,image_path])
         
-        image_input.upload(upload_callback, [image_input, state, log_state,visual_chatgpt,openai_api_key,language,naritive,history_log,auto_play,session_type],
+        image_input.upload(upload_callback, [image_input, state, log_state,task_type, visual_chatgpt,openai_api_key,language,naritive,history_log,auto_play,session_type],
                            [chatbot, state, origin_image, click_state, image_input, image_input_base, sketcher_input,image_input_base_2,
                             image_embedding, original_size, input_size,name_label,artist_label,year_label,material_label,name_label_base, artist_label_base, year_label_base, material_label_base, \
                             name_label_base2, artist_label_base2, year_label_base2, material_label_base2,name_label_traj, artist_label_traj, year_label_traj, material_label_traj, \
@@ -2596,7 +2605,7 @@ def create_ui():
         # submit_button_text.click(chat_input_callback, [visual_chatgpt, chat_input, click_state, state, aux_state,language,auto_play],
         #                   [chatbot, state, aux_state,output_audio])
         # submit_button_text.click(lambda: "", None, chat_input)
-        example_image.change(upload_callback, [example_image, state, log_state, visual_chatgpt, openai_api_key,language,naritive,history_log,auto_play,session_type],
+        example_image.change(upload_callback, [example_image, state, log_state, task_type,  visual_chatgpt, openai_api_key,language,naritive,history_log,auto_play,session_type],
                              [chatbot, state, origin_image, click_state, image_input, image_input_base, sketcher_input,image_input_base_2,
                               image_embedding, original_size, input_size,name_label,artist_label,year_label,material_label,name_label_base, artist_label_base, year_label_base, material_label_base, \
                             name_label_base2, artist_label_base2, year_label_base2, material_label_base2,name_label_traj, artist_label_traj, year_label_traj, material_label_traj, \