import gradio as gr import os import time import shutil import codecs # Chatimport shutilbot demo with multimodal input (text, markdown, LaTeX, code blocks, image, audio, & video). Plus shows support for streaming text. from src.template import rewrite_prompt_template from src.config import Config import os import platform from langchain.llms import OpenAI from src.requests_API import GPT4_requests with gr.Blocks() as demo: bot_name = gr.State(value="Rewriter") paper_path = gr.State(value="") file_dir = gr.State(value="files") public_file_dir = gr.State(value="") def build_prompt(history, name, mode="chat"): prompt = "" for query, response in history[:-1]: prompt += f"\n\nUser：{query}" prompt += f"\n\n{name}: \n{response}" if mode == "download": prompt += f"\n\nUser：{history[-1][0]}" prompt += f"\n\n{name}: \n{history[-1][1]}" else: prompt += f"\n\nUser：{history[-1][0]}" return prompt def print_like_dislike(x: gr.LikeData): print(x.index, x.value, x.liked) def read_file_contents(file_path): with open(file_path, 'r', encoding='utf8') as file: content = file.read() return content def add_text(history, text, name, cj, field, topic, language, paper_path): if text: if not history: if not paper_path: gr.Warning("Paper file not found.") return [], gr.Textbox(value="", interactive=True, placeholder="Enter text and press enter, or press submit button", ) bot_name = name # 读取文件内容 paper = read_file_contents(paper_path) # 格式化模板 rewrite_prompt = rewrite_prompt_template.format( name=name, cj=cj, field=field, topic=topic, language=language, paper=paper ) query = rewrite_prompt + text else: query = text history = history + [[query, None]] return history, gr.Textbox(value="", interactive=False), gr.Button(value="Submit", interactive=False), gr.Button(value="Reset", interactive=False) def bot(history, name, openai_key, temperature): if history: # print(history) prompt = build_prompt(history, name) os.environ["OPENAI_API_KEY"] = openai_key llm = OpenAI(temperature=temperature) response = llm(prompt) # response = GPT4_requests(prompt) history[-1][1] = response return history def add_file(history, file): history = history + [((file.name,), None)] return history def upload_file(file_obj, file_dir): print(file_obj.name) public_file_dir = os.path.dirname(file_obj.name) print(file_dir) # 将文件复制到目标目录 shutil.copy(file_obj.name, file_dir) # 获取上传Gradio的文件名称 FileName = os.path.basename(file_obj.name) # 获取拷贝在临时目录的新的文件地址 NewfilePath = os.path.join(file_dir, FileName) paper_path = NewfilePath print(paper_path) return paper_path, public_file_dir, gr.Button(value="下载", visible=True) def download_file(history, file_dir, paper_path, name): if not history: gr.Warning("Polished contents not found.") return print(paper_path) new_name = "rewrited_" + os.path.basename(paper_path) file_path = os.path.join(file_dir, new_name) with codecs.open(file_path, "w", encoding="utf-8") as file: file.write(build_prompt(history, name, mode="download")) # gr.Info("Click on the download link for the polished file.") return gr.Textbox( label="下载链接", info="复制文件路径拼接到服务器地址后得到完整下载地址", lines=1, visible=True, value="/file=" + file_path, interactive=False ) def reset_state(): return gr.Textbox(value="", interactive=True), gr.Button(value="Submit", interactive=True), gr.Button( value="Reset", interactive=True) def reset(): return [], None chatbot = gr.Chatbot( [], elem_id="Rewriter", bubble_full_width=False, height=666, avatar_images=(None, (os.path.join("images", "avatar.png"))), ) with gr.Column(): with gr.Row(): txt = gr.Textbox( scale=4, show_label=False, placeholder="Enter text and press enter, or press submit button", container=False, ) btn_submit = gr.Button(value="Submit") btn_reset = gr.Button(value="Reset") name_element = gr.Textbox( label="机器人名字", info="", lines=1, value="Academic Paper Rewriting Expert", ), openai_key_element = gr.Textbox( label="OpenAI Key", info="", lines=1, value="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", ), temperature_element = gr.Slider(0, 1, value=1, label="Temperature", info="Choose between 0 and 1"), cj_element = gr.Textbox( label="目标会议/期刊", info="", lines=1, value="ACL Conference", ), field_element = gr.Textbox( label="研究领域", info="", lines=1, value="LLM", ), topic_element = gr.Textbox( label="研究主题", info="", lines=1, value="Multi-Agent", ), language_element = gr.Textbox( label="语言", info="", lines=1, value="English", ), paper_element = gr.File(file_types=["txt"], label="论文") download_btn = gr.Button(value="下载", visible=False), download_element = gr.Textbox( label="下载链接", info="", lines=1, visible=False, value="", ) # txt_msg = txt.submit(add_text, [chatbot, txt, name_element[0], cj_element[0], field_element[0], topic_element[0]], # [chatbot, txt], queue=False).then( # bot, chatbot, chatbot, api_name="bot_response" # ) # btn_msg = btn.click(add_text, [chatbot, txt, name_element[0], cj_element[0], field_element[0], topic_element[0]], # [chatbot, txt, name_element[0]], queue=False).then( # bot, chatbot, chatbot, api_name="bot_response" # ) # btn_msg.then(lambda: gr.Textbox(interactive=True), None, [txt], queue=False) txt_msg = txt.submit(add_text, [ chatbot, txt, name_element[0], cj_element[0], field_element[0], topic_element[0], language_element[0], paper_path ], [chatbot, txt, btn_submit, btn_reset]).then(bot, [ chatbot, name_element[0], openai_key_element[0], temperature_element[0] ], [ chatbot ], api_name="bot_response") txt_msg.then(reset_state, None, [txt, btn_submit, btn_reset]) btn_msg = btn_submit.click(add_text, [ chatbot, txt, name_element[0], cj_element[0], field_element[0], topic_element[0], language_element[0], paper_path ], [chatbot, txt, btn_submit, btn_reset]).then(bot, [ chatbot, name_element[0], openai_key_element[0], temperature_element[0] ], [ chatbot ], api_name="bot_response") btn_msg.then(reset_state, None, [txt, btn_submit, btn_reset]) btn_reset.click(reset, None, [chatbot, paper_element]) paper_element.upload(upload_file, [paper_element, file_dir], [paper_path, public_file_dir, download_btn[0]]) download_btn[0].click(download_file, [chatbot, public_file_dir, paper_path, bot_name], [download_element]) chatbot.like(print_like_dislike, None, None) demo.queue() demo.launch()