app.py

# import gradio as gr
# from huggingface_hub import InferenceClient

# """
# For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
# """
# client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")


# def respond(
#     message,
#     history: list[tuple[str, str]],
#     system_message,
#     max_tokens,
#     temperature,
#     top_p,
# ):
#     messages = [{"role": "system", "content": system_message}]

#     for val in history:
#         if val[0]:
#             messages.append({"role": "user", "content": val[0]})
#         if val[1]:
#             messages.append({"role": "assistant", "content": val[1]})

#     messages.append({"role": "user", "content": message})

#     response = ""

#     for message in client.chat_completion(
#         messages,
#         max_tokens=max_tokens,
#         stream=True,
#         temperature=temperature,
#         top_p=top_p,
#     ):
#         token = message.choices[0].delta.content

#         response += token
#         yield response


# """
# For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
# """
# demo = gr.ChatInterface(
#     respond,
#     additional_inputs=[
#         gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
#         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
#         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
#         gr.Slider(
#             minimum=0.1,
#             maximum=1.0,
#             value=0.95,
#             step=0.05,
#             label="Top-p (nucleus sampling)",
#         ),
#     ],
# )


# if __name__ == "__main__":
#     demo.launch()


import gradio as gr
import requests
import re

# endpoint : 
deployment_name = "fimtrus-gpt-4o-ncus"
endpoint = "https://fimtrus-openai-ncus.openai.azure.com/openai/deployments/{}/chat/completions".format(deployment_name)
api_key = "FP7k76c0BGmz4MVot1LVqdtqcmQaKBeerYdFyXzW4NXN0940rRfEJQQJ99BCACHrzpqXJ3w3AAABACOG8189"

ai_search_endpoint = "https://fimtrus-ai-search3.search.windows.net"
ai_search_key = "lw8TYxNHT14hpM5oJQhCL3iOxR4PR494dV4RLDaiwzAzSeBecA5i"
ai_search_index_name = "travel-index-microsoft2"
ai_search_semantic = "travel-semantic"


def request_gpt(prompt):
  query_params = {
    "api-version": "2025-02-01-preview"
  }
  headers = {
      "Content-Type": "application/json",
      "api-key": api_key
  }
  
  body = {
    "data_sources": [
      {
        "type": "azure_search",
        "parameters": {
          "endpoint": ai_search_endpoint,
          "index_name": ai_search_index_name,
          "semantic_configuration": ai_search_semantic,
          "query_type": "semantic",
          "fields_mapping": {},
          "in_scope": True,
          "filter": None,
          "strictness": 3,
          "top_n_documents": 3,
          "authentication": {
            "type": "api_key",
            "key": ai_search_key
          },
          "key": ai_search_key
        }
      }
    ],
    "messages": [
      {
        "role": "system",
        "content": "사용자가 정보를 찾는 데 도움이 되는 AI 도우미입니다."
      },
      {
        "role": "user",
        "content": prompt
      }
    ],
    "temperature": 0.7,
    "top_p": 0.95,
    "max_tokens": 800,
    "stop": None,
    "stream": False,
    "frequency_penalty": 0,
    "presence_penalty": 0
  }

  response = requests.post(endpoint, params=query_params, headers=headers, json=body)
  
  if response.status_code == 200:
    response_json = response.json()
    message = response_json['choices'][0]['message']
    content = message['content']
    
    content = re.sub(r'\[doc(\d+)\]', r'[참조 \1]', content)
    
    if message['context']:
      citation_list = message['context']['citations']
    else:
      citation_list = list()
    return content, citation_list
  else:
    return "", list()

with gr.Blocks() as demo:
    
    citation_state = gr.State([])
        
    def click_send(prompt, histories):
        print(prompt)
        content, citation_list = request_gpt(prompt=prompt)
        message_list = [{"role": "user", "content": prompt}, {"role": "assistant", "content": content}]
        histories.extend(message_list)
        
        print(histories)
        print(len(citation_list), citation_list)
        return histories, citation_list, ""
      
    def select_tab(citation):
      print("tab!!!!!!")
      return gr.Textbox(label="", value=citation, lines=10, max_lines=10, visible=True)
    
    gr.Markdown("# 나의 GPT 월드")
    
    with gr.Row():
      chatbot = gr.Chatbot(label="나의 GPT", type="messages", scale=3)
    
      with gr.Column(scale=1):  
        @gr.render(inputs=[citation_state])
        def render_citations(citations):
          print(citations)
          for index in range(len(citations)):
            content = citations[index]['content']
            key = "참조{}".format(index + 1)
            if index == 0:
              gr.Tab("", id=-1, visible=False)
            with gr.Tab(key) as tab:
              citation_textbox = gr.Textbox(label="", value=content, lines=10, max_lines=10, visible=False)
            tab.select(fn=select_tab, inputs=[citation_textbox], outputs=[citation_textbox])
        
    with gr.Row():
        prompt_textbox = gr.Textbox(label="프롬프트", placeholder="프롬프트를 입력하세요.", scale=6)
        send_button = gr.Button("전송", scale=1)
        
    send_button.click(click_send, inputs=[prompt_textbox, chatbot], outputs=[chatbot, citation_state, prompt_textbox])
    prompt_textbox.submit(click_send, inputs=[prompt_textbox, chatbot], outputs=[chatbot, citation_state, prompt_textbox])

if __name__ == "__main__":
    demo.launch()