Merge branch 'master' into huggingface

README.md

@@ -1,3 +1,15 @@
+---
+title: ChatImprovement
+emoji: 😻
+colorFrom: blue
+colorTo: blue
+sdk: gradio
+sdk_version: 3.32.0
+app_file: app.py
+pinned: false
+---
+
+# ChatGPT 学术优化
 > **Note**
 >
 > 2023.5.27 对Gradio依赖进行了调整，Fork并解决了官方Gradio的若干Bugs。请及时**更新代码**并重新更新pip依赖。安装依赖时，请严格选择`requirements.txt`中**指定的版本**： 

app.py

@@ -1,13 +1,15 @@
 import os; os.environ['no_proxy'] = '*' # 避免代理网络产生意外污染
 
 def main():
+    import subprocess, sys
+    subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'gradio-stable-fork'])
     import gradio as gr
-    if gr.__version__ not in ['3.28.3','3.32.2']: assert False, "需要特殊依赖，请务必用 pip install -r requirements.txt 指令安装依赖，详情信息见requirements.txt"
+    if gr.__version__ not in ['3.28.3','3.32.3']: assert False, "请用 pip install -r requirements.txt 安装依赖"
     from request_llm.bridge_all import predict
     from toolbox import format_io, find_free_port, on_file_uploaded, on_report_generated, get_conf, ArgsGeneralWrapper, DummyWith
     # 建议您复制一个config_private.py放自己的秘密, 如API和代理网址, 避免不小心传github被别人看到
-    proxies, WEB_PORT, LLM_MODEL, CONCURRENT_COUNT, AUTHENTICATION, CHATBOT_HEIGHT, LAYOUT, API_KEY, AVAIL_LLM_MODELS, AUTO_CLEAR_TXT = \
+    proxies, WEB_PORT, LLM_MODEL, CONCURRENT_COUNT, AUTHENTICATION, CHATBOT_HEIGHT, LAYOUT, API_KEY, AVAIL_LLM_MODELS = \
-        get_conf('proxies', 'WEB_PORT', 'LLM_MODEL', 'CONCURRENT_COUNT', 'AUTHENTICATION', 'CHATBOT_HEIGHT', 'LAYOUT', 'API_KEY', 'AVAIL_LLM_MODELS', 'AUTO_CLEAR_TXT')
+        get_conf('proxies', 'WEB_PORT', 'LLM_MODEL', 'CONCURRENT_COUNT', 'AUTHENTICATION', 'CHATBOT_HEIGHT', 'LAYOUT', 'API_KEY', 'AVAIL_LLM_MODELS')
 
     # 如果WEB_PORT是-1, 则随机选取WEB端口
     PORT = find_free_port() if WEB_PORT <= 0 else WEB_PORT
@@ -54,6 +56,7 @@ def main():
     cancel_handles = []
     with gr.Blocks(title="ChatGPT 学术优化", theme=set_theme, analytics_enabled=False, css=advanced_css) as demo:
         gr.HTML(title_html)
+        gr.HTML('''<center><a href="https://huggingface.co/spaces/qingxu98/gpt-academic?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>请您打开此页面后务必点击上方的“复制空间”（Duplicate Space）按钮！<font color="#FF00FF">使用时，先在输入框填入API-KEY然后回车。</font><br/>切忌在“复制空间”（Duplicate Space）之前填入API_KEY或进行提问，否则您的API_KEY将极可能被空间所有者攫取！<br/>支持任意数量的OpenAI的密钥和API2D的密钥共存，例如输入"OpenAI密钥1,API2D密钥2"，然后提交，即可同时使用两种模型接口。</center>''')
         cookies = gr.State({'api_key': API_KEY, 'llm_model': LLM_MODEL})
         with gr_L1():
             with gr_L2(scale=2):
@@ -63,7 +66,7 @@ def main():
             with gr_L2(scale=1):
                 with gr.Accordion("输入区", open=True) as area_input_primary:
                     with gr.Row():
-                        txt = gr.Textbox(show_label=False, placeholder="Input question here.").style(container=False)
+                        txt = gr.Textbox(show_label=False, lines=2, placeholder="输入问题或API密钥，输入多个密钥时，用英文逗号间隔。支持OpenAI密钥和API2D密钥共存。").style(container=False)
                     with gr.Row():
                         submitBtn = gr.Button("提交", variant="primary")
                     with gr.Row():
@@ -104,7 +107,7 @@ def main():
                     system_prompt = gr.Textbox(show_label=True, placeholder=f"System Prompt", label="System prompt", value=initial_prompt)
                     top_p = gr.Slider(minimum=-0, maximum=1.0, value=1.0, step=0.01,interactive=True, label="Top-p (nucleus sampling)",)
                     temperature = gr.Slider(minimum=-0, maximum=2.0, value=1.0, step=0.01, interactive=True, label="Temperature",)
-                    max_length_sl = gr.Slider(minimum=256, maximum=8192, value=4096, step=1, interactive=True, label="Local LLM MaxLength",)
+                    max_length_sl = gr.Slider(minimum=256, maximum=4096, value=512, step=1, interactive=True, label="Local LLM MaxLength",)
                     checkboxes = gr.CheckboxGroup(["基础功能区", "函数插件区", "底部输入区", "输入清除键", "插件参数区"], value=["基础功能区", "函数插件区"], label="显示/隐藏功能区")
                     md_dropdown = gr.Dropdown(AVAIL_LLM_MODELS, value=LLM_MODEL, label="更换LLM模型/请求源").style(container=False)
 
@@ -144,11 +147,6 @@ def main():
         resetBtn2.click(lambda: ([], [], "已重置"), None, [chatbot, history, status])
         clearBtn.click(lambda: ("",""), None, [txt, txt2])
         clearBtn2.click(lambda: ("",""), None, [txt, txt2])
-        if AUTO_CLEAR_TXT:
-            submitBtn.click(lambda: ("",""), None, [txt, txt2])
-            submitBtn2.click(lambda: ("",""), None, [txt, txt2])
-            txt.submit(lambda: ("",""), None, [txt, txt2])
-            txt2.submit(lambda: ("",""), None, [txt, txt2])
         # 基础功能区的回调函数注册
         for k in functional:
             if ("Visible" in functional[k]) and (not functional[k]["Visible"]): continue
@@ -202,10 +200,7 @@ def main():
         threading.Thread(target=warm_up_modules, name="warm-up", daemon=True).start()
 
     auto_opentab_delay()
-    demo.queue(concurrency_count=CONCURRENT_COUNT).launch(
+    demo.queue(concurrency_count=CONCURRENT_COUNT).launch(server_name="0.0.0.0", share=False, favicon_path="docs/logo.png", blocked_paths=["config.py","config_private.py","docker-compose.yml","Dockerfile"])
-        server_name="0.0.0.0", server_port=PORT,
-        favicon_path="docs/logo.png", auth=AUTHENTICATION,
-        blocked_paths=["config.py","config_private.py","docker-compose.yml","Dockerfile"])
 
     # 如果需要在二级路径下运行
     # CUSTOM_PATH, = get_conf('CUSTOM_PATH')

check_proxy.py

@@ -1,10 +1,10 @@
 
-def check_proxy(proxies: dict):
+def check_proxy(proxies):
     import requests
-    proxies_https = proxies.get('https') if proxies is not None else '无'
+    proxies_https = proxies['https'] if proxies is not None else '无'
     try:
         response = requests.get("https://ipapi.co/json/",
-                                proxies=proxies, timeout=30)
+                                proxies=proxies, timeout=4)
         data = response.json()
         print(f'查询代理的地理位置，返回的结果是{data}')
         if 'country_name' in data:
@@ -12,12 +12,10 @@ def check_proxy(proxies: dict):
             result = f"代理配置 {proxies_https}, 代理所在地：{country}"
         elif 'error' in data:
             result = f"代理配置 {proxies_https}, 代理所在地：未知，IP查询频率受限"
-        else:
-            result = f"代理配置 {proxies_https}, 代理数据解析失败：{data}"
         print(result)
         return result
-    except Exception as e:
+    except:
-        result = f"代理 {proxies_https} 查询出现异常: {e}，代理可能无效"
+        result = f"代理配置 {proxies_https}, 代理所在地查询超时，代理可能无效"
         print(result)
         return result
 

config.py

@@ -45,10 +45,9 @@ WEB_PORT = -1
 # 如果OpenAI不响应（网络卡顿、代理失败、KEY失效），重试的次数限制
 MAX_RETRY = 2
 
-# 模型选择是 (注意: LLM_MODEL是默认选中的模型, 同时它必须被包含在AVAIL_LLM_MODELS切换列表中 )
+# OpenAI模型选择是（gpt4现在只对申请成功的人开放）
-LLM_MODEL = "gpt-3.5-turbo" # 可选 ↓↓↓
+LLM_MODEL = "gpt-3.5-turbo" # 可选 "chatglm"
-AVAIL_LLM_MODELS = ["gpt-3.5-turbo-16k", "gpt-3.5-turbo", "azure-gpt35", "api2d-gpt-3.5-turbo", "gpt-4", "api2d-gpt-4", "chatglm", "moss", "newbing", "newbing-free", "stack-claude"]
+AVAIL_LLM_MODELS = ["newbing-free", "gpt-3.5-turbo", "gpt-4", "api2d-gpt-4", "api2d-gpt-3.5-turbo"]
-# P.S. 其他可用的模型还包括 ["gpt-3.5-turbo-0613", "gpt-3.5-turbo-16k-0613", "newbing-free", "jittorllms_rwkv", "jittorllms_pangualpha", "jittorllms_llama"]
 
 # 本地LLM模型如ChatGLM的执行方式 CPU/GPU
 LOCAL_MODEL_DEVICE = "cpu" # 可选 "cuda"
@@ -56,9 +55,6 @@ LOCAL_MODEL_DEVICE = "cpu" # 可选 "cuda"
 # 设置gradio的并行线程数（不需要修改）
 CONCURRENT_COUNT = 100
 
-# 是否在提交时自动清空输入框
-AUTO_CLEAR_TXT = False
-
 # 加一个live2d装饰
 ADD_WAIFU = False
 

core_functional.py

@@ -63,7 +63,6 @@ def get_core_functions():
             "Prefix":   r"我需要你找一张网络图片。使用Unsplash API(https://source.unsplash.com/960x640/?<英语关键词>)获取图片URL，" +
                         r"然后请使用Markdown格式封装，并且不要有反斜线，不要用代码块。现在，请按以下描述给我发送图片：" + "\n\n",
             "Suffix":   r"",
-            "Visible": False,
         },
         "解释代码": {
             "Prefix":   r"请解释以下代码：" + "\n```\n",
@@ -74,5 +73,6 @@ def get_core_functions():
                         r"Note that, reference styles maybe more than one kind, you should transform each item correctly." +
                         r"Items need to be transformed:",
             "Suffix":   r"",
+            "Visible": False,
         }
     }

crazy_functions/crazy_functions_test.py

@@ -193,9 +193,8 @@ def test_Latex():
     # txt = r"https://arxiv.org/abs/2212.10156"
     # txt = r"https://arxiv.org/abs/2211.11559"
     # txt = r"https://arxiv.org/abs/2303.08774"
-    # txt = r"https://arxiv.org/abs/2303.12712"
+    txt = r"https://arxiv.org/abs/2303.12712"
     # txt = r"C:\Users\fuqingxu\arxiv_cache\2303.12712\workfolder"
-    txt = r"2306.17157" # 这个paper有个input命令文件名大小写错误！
     
 
     for cookies, cb, hist, msg in (Latex翻译中文并重新编译PDF)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):

crazy_functions/crazy_utils.py

@@ -1,14 +1,11 @@
 from toolbox import update_ui, get_conf, trimmed_format_exc
 import threading
 
-
 def input_clipping(inputs, history, max_token_limit):
     import numpy as np
     from request_llm.bridge_all import model_info
     enc = model_info["gpt-3.5-turbo"]['tokenizer']
-
+    def get_token_num(txt): return len(enc.encode(txt, disallowed_special=()))
-    def get_token_num(txt):
-        return len(enc.encode(txt, disallowed_special=()))
 
     mode = 'input-and-history'
     # 当 输入部分的token占比 小于 全文的一半时，只裁剪历史
@@ -38,7 +35,6 @@ def input_clipping(inputs, history, max_token_limit):
     history = everything[1:]
     return inputs, history
 
-
 def request_gpt_model_in_new_thread_with_ui_alive(
         inputs, inputs_show_user, llm_kwargs, 
         chatbot, history, sys_prompt, refresh_interval=0.2,
@@ -71,7 +67,6 @@ def request_gpt_model_in_new_thread_with_ui_alive(
     yield from update_ui(chatbot=chatbot, history=[]) # 刷新界面
     executor = ThreadPoolExecutor(max_workers=16)
     mutable = ["", time.time(), ""]
-
     def _req_gpt(inputs, history, sys_prompt):
         retry_op = retry_times_at_unknown_error
         exceeded_cnt = 0
@@ -109,8 +104,7 @@ def request_gpt_model_in_new_thread_with_ui_alive(
                 mutable[0] += f"[Local Message] 警告，在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
                 if retry_op > 0:
                     retry_op -= 1
-                    mutable[
+                    mutable[0] += f"[Local Message] 重试中，请稍等 {retry_times_at_unknown_error-retry_op}/{retry_times_at_unknown_error}：\n\n"
-                        0] += f"[Local Message] 重试中，请稍等 {retry_times_at_unknown_error - retry_op}/{retry_times_at_unknown_error}：\n\n"
                     if ("Rate limit reached" in tb_str) or ("Too Many Requests" in tb_str):
                         time.sleep(30)
                     time.sleep(5)
@@ -177,10 +171,8 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
     assert len(inputs_array) == len(history_array)
     assert len(inputs_array) == len(sys_prompt_array)
     if max_workers == -1: # 读取配置文件
-        try:
+        try: max_workers, = get_conf('DEFAULT_WORKER_NUM')
-            max_workers, = get_conf('DEFAULT_WORKER_NUM')
+        except: max_workers = 8
-        except:
-            max_workers = 8
         if max_workers <= 0: max_workers = 3
     # 屏蔽掉 chatglm的多线程，可能会导致严重卡顿
     if not (llm_kwargs['llm_model'].startswith('gpt-') or llm_kwargs['llm_model'].startswith('api2d-')):
@@ -230,17 +222,15 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
                     # 【选择放弃】
                     tb_str = '```\n' + trimmed_format_exc() + '```'
                     gpt_say += f"[Local Message] 警告，线程{index}在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
-                    if len(mutable[index][0]) > 0:
+                    if len(mutable[index][0]) > 0: gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
-                        gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
                     mutable[index][2] = "输入过长已放弃"
                     return gpt_say # 放弃
-            except Exception as e:
+            except:
                 # 【第三种情况】：其他错误
                 tb_str = '```\n' + trimmed_format_exc() + '```'
-                print(f"发生异常：{e}, 调用栈信息：{tb_str}")
+                print(tb_str)
                 gpt_say += f"[Local Message] 警告，线程{index}在执行过程中遭遇问题, Traceback：\n\n{tb_str}\n\n"
-                if len(mutable[index][0]) > 0:
+                if len(mutable[index][0]) > 0: gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
-                    gpt_say += "此线程失败前收到的回答：\n\n" + mutable[index][0]
                 if retry_op > 0: 
                     retry_op -= 1
                     wait = random.randint(5, 20)
@@ -251,11 +241,9 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
                         fail_info = ""
                     # 也许等待十几秒后，情况会好转
                     for i in range(wait):
-                        mutable[index][2] = f"{fail_info}等待重试 {wait - i}";
+                        mutable[index][2] = f"{fail_info}等待重试 {wait-i}"; time.sleep(1)
-                        time.sleep(1)
                     # 开始重试
-                    mutable[index][
+                    mutable[index][2] = f"重试中 {retry_times_at_unknown_error-retry_op}/{retry_times_at_unknown_error}"
-                        2] = f"重试中 {retry_times_at_unknown_error - retry_op}/{retry_times_at_unknown_error}"
                     continue # 返回重试
                 else:
                     mutable[index][2] = "已失败"
@@ -264,8 +252,7 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
                     return gpt_say # 放弃
 
     # 异步任务开始
-    futures = [executor.submit(_req_gpt, index, inputs, history, sys_prompt) for index, inputs, history, sys_prompt in
+    futures = [executor.submit(_req_gpt, index, inputs, history, sys_prompt) for index, inputs, history, sys_prompt in zip(
-               zip(
         range(len(inputs_array)), inputs_array, history_array, sys_prompt_array)]
     cnt = 0
     while True:
@@ -319,7 +306,6 @@ def breakdown_txt_to_satisfy_token_limit(txt, get_token_fn, limit):
             lines = txt_tocut.split('\n')
             estimated_line_cut = limit / get_token_fn(txt_tocut) * len(lines)
             estimated_line_cut = int(estimated_line_cut)
-            cnt = 0
             for cnt in reversed(range(estimated_line_cut)):
                 if must_break_at_empty_line:
                     if lines[cnt] != "":
@@ -336,7 +322,6 @@ def breakdown_txt_to_satisfy_token_limit(txt, get_token_fn, limit):
             result = [prev]
             result.extend(cut(post, must_break_at_empty_line))
             return result
-
     try:
         return cut(txt, must_break_at_empty_line=True)
     except RuntimeError:
@@ -352,7 +337,6 @@ def force_breakdown(txt, limit, get_token_fn):
             return txt[:i], txt[i:]
     return "Tiktoken未知错误", "Tiktoken未知错误"
 
-
 def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
     # 递归
     def cut(txt_tocut, must_break_at_empty_line, break_anyway=False):  
@@ -381,7 +365,6 @@ def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
             result = [prev]
             result.extend(cut(post, must_break_at_empty_line, break_anyway=break_anyway))
             return result
-
     try:
         # 第1次尝试，将双空行（\n\n）作为切分点
         return cut(txt, must_break_at_empty_line=True)
@@ -404,6 +387,7 @@ def breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn, limit):
                     return cut(txt, must_break_at_empty_line=False, break_anyway=True)
 
 
+
 def read_and_clean_pdf_text(fp):
     """
     这个函数用于分割pdf，用了很多trick，逻辑较乱，效果奇好
@@ -433,7 +417,6 @@ def read_and_clean_pdf_text(fp):
     fb = 2  # Index 2 框框
     REMOVE_FOOT_NOTE = True # 是否丢弃掉 不是正文的内容 （比正文字体小，如参考文献、脚注、图注等）
     REMOVE_FOOT_FFSIZE_PERCENT = 0.95 # 小于正文的？时，判定为不是正文（有些文章的正文部分字体大小不是100%统一的，有肉眼不可见的小变化）
-
     def primary_ffsize(l):
         """
         提取文本块主字体
@@ -471,8 +454,7 @@ def read_and_clean_pdf_text(fp):
                         for wtf in l['spans']: # for l in t['lines']:
                             meta_span.append([wtf['text'], wtf['size'], len(wtf['text'])])
                     # meta_line.append(["NEW_BLOCK", pf])
-            # 块元提取                           for each word segment with in line                       for each line
+            # 块元提取                           for each word segment with in line                       for each line         cross-line words                          for each block
-            # cross-line words                          for each block
             meta_txt.extend([" ".join(["".join([wtf['text'] for wtf in l['spans']]) for l in t['lines']]).replace(
                 '- ', '') for t in text_areas['blocks'] if 'lines' in t])
             meta_font.extend([np.mean([np.mean([wtf['size'] for wtf in l['spans']])
@@ -504,8 +486,7 @@ def read_and_clean_pdf_text(fp):
                 # 尝试识别段落
                 if meta_line[index][fc].endswith('.') and\
                     (meta_line[index-1][fc] != 'NEW_BLOCK') and \
-                        (meta_line[index][fb][2] - meta_line[index][fb][0]) < (
+                    (meta_line[index][fb][2] - meta_line[index][fb][0]) < (meta_line[index-1][fb][2] - meta_line[index-1][fb][0]) * 0.7:
-                        meta_line[index - 1][fb][2] - meta_line[index - 1][fb][0]) * 0.7:
                     sec[-1] += line[fc]
                     sec[-1] += "\n\n"
                 else:
@@ -539,7 +520,6 @@ def read_and_clean_pdf_text(fp):
                 if len(block_txt) < 100:
                     meta_txt[index] = '\n'
             return meta_txt
-
         meta_txt = 把字符太少的块清除为回车(meta_txt)
 
         def 清理多余的空行(meta_txt):
@@ -547,7 +527,6 @@ def read_and_clean_pdf_text(fp):
                 if meta_txt[index] == '\n' and meta_txt[index-1] == '\n':
                     meta_txt.pop(index)
             return meta_txt
-
         meta_txt = 清理多余的空行(meta_txt)
 
         def 合并小写开头的段落块(meta_txt):
@@ -558,7 +537,6 @@ def read_and_clean_pdf_text(fp):
                     return True
                 else:
                     return False
-
             for _ in range(100):
                 for index, block_txt in enumerate(meta_txt):
                     if starts_with_lowercase_word(block_txt):
@@ -569,7 +547,6 @@ def read_and_clean_pdf_text(fp):
                         meta_txt[index-1] += meta_txt[index]
                         meta_txt[index] = '\n'
             return meta_txt
-
         meta_txt = 合并小写开头的段落块(meta_txt)
         meta_txt = 清理多余的空行(meta_txt)
 
@@ -611,8 +588,7 @@ def get_files_from_everything(txt, type):  # type='.md'
         from toolbox import get_conf
         proxies, = get_conf('proxies')
         r = requests.get(txt, proxies=proxies)
-        with open('./gpt_log/temp' + type, 'wb+') as f:
+        with open('./gpt_log/temp'+type, 'wb+') as f: f.write(r.content)
-            f.write(r.content)
         project_folder = './gpt_log/'
         file_manifest = ['./gpt_log/temp'+type]
     elif txt.endswith(type):
@@ -633,6 +609,8 @@ def get_files_from_everything(txt, type):  # type='.md'
     return success, file_manifest, project_folder
 
 
+
+
 def Singleton(cls):
     _instance = {}
  
@@ -664,6 +642,7 @@ class knowledge_archive_interface():
 
         return self.text2vec_large_chinese
 
+
     def feed_archive(self, file_manifest, id="default"):
         self.threadLock.acquire()
         # import uuid
@@ -715,7 +694,6 @@ class knowledge_archive_interface():
         self.threadLock.release()
         return resp, prompt
 
-
 def try_install_deps(deps):
     for dep in deps:
         import subprocess, sys

crazy_functions/latex_utils.py

@@ -203,7 +203,6 @@ def merge_tex_files_(project_foler, main_file, mode):
                 c = fx.read()
         else:  
             # e.g., \input{srcs/07_appendix}
-            assert os.path.exists(fp+'.tex'), f'即找不到{fp}，也找不到{fp}.tex，Tex源文件缺失！'
             with open(fp+'.tex', 'r', encoding='utf-8', errors='replace') as fx:
                 c = fx.read()
         c = merge_tex_files_(project_foler, c, mode)

crazy_functions/图片生成.py

@@ -27,10 +27,8 @@ def gen_image(llm_kwargs, prompt, resolution="256x256"):
     }
     response = requests.post(url, headers=headers, json=data, proxies=proxies)
     print(response.content)
-    try:
     image_url = json.loads(response.content.decode('utf8'))['data'][0]['url']
-    except:
+
-        raise RuntimeError(response.content.decode())
     # 文件保存到本地
     r = requests.get(image_url, proxies=proxies)
     file_path = 'gpt_log/image_gen/'

crazy_functions/批量翻译PDF文档_多线程.py

@@ -1,5 +1,5 @@
 from toolbox import CatchException, report_execption, write_results_to_file
-from toolbox import update_ui, promote_file_to_downloadzone
+from toolbox import update_ui
 from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
 from .crazy_utils import request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
 from .crazy_utils import read_and_clean_pdf_text
@@ -147,14 +147,23 @@ def 解析PDF(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot,
             print('writing html result failed:', trimmed_format_exc())
 
     # 准备文件的下载
+    import shutil
     for pdf_path in generated_conclusion_files:
         # 重命名文件
-        rename_file = f'翻译-{os.path.basename(pdf_path)}'
+        rename_file = f'./gpt_log/翻译-{os.path.basename(pdf_path)}'
-        promote_file_to_downloadzone(pdf_path, rename_file=rename_file, chatbot=chatbot)
+        if os.path.exists(rename_file):
+            os.remove(rename_file)
+        shutil.copyfile(pdf_path, rename_file)
+        if os.path.exists(pdf_path):
+            os.remove(pdf_path)
     for html_path in generated_html_files:
         # 重命名文件
-        rename_file = f'翻译-{os.path.basename(html_path)}'
+        rename_file = f'./gpt_log/翻译-{os.path.basename(html_path)}'
-        promote_file_to_downloadzone(html_path, rename_file=rename_file, chatbot=chatbot)
+        if os.path.exists(rename_file):
+            os.remove(rename_file)
+        shutil.copyfile(html_path, rename_file)
+        if os.path.exists(html_path):
+            os.remove(html_path)
     chatbot.append(("给出输出文件清单", str(generated_conclusion_files + generated_html_files)))
     yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
 

crazy_functions/询问多个大语言模型.py

@@ -13,11 +13,11 @@ def 同时问询(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt
     web_port        当前软件运行的端口号
     """
     history = []    # 清空历史，以免输入溢出
-    chatbot.append((txt, "正在同时咨询ChatGPT和ChatGLM……"))
+    chatbot.append((txt, "正在同时咨询gpt-3.5和gpt-4……"))
     yield from update_ui(chatbot=chatbot, history=history) # 刷新界面 # 由于请求gpt需要一段时间，我们先及时地做一次界面更新
 
     # llm_kwargs['llm_model'] = 'chatglm&gpt-3.5-turbo&api2d-gpt-3.5-turbo' # 支持任意数量的llm接口，用&符号分隔
-    llm_kwargs['llm_model'] = 'chatglm&gpt-3.5-turbo' # 支持任意数量的llm接口，用&符号分隔
+    llm_kwargs['llm_model'] = 'gpt-3.5-turbo&gpt-4' # 支持任意数量的llm接口，用&符号分隔
     gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
         inputs=txt, inputs_show_user=txt, 
         llm_kwargs=llm_kwargs, chatbot=chatbot, history=history, 

crazy_functions/读文章写摘要.py

@@ -1,13 +1,11 @@
-from toolbox import CatchException, report_execption, write_results_to_file
 from toolbox import update_ui
+from toolbox import CatchException, report_execption, write_results_to_file
 from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
-
 fast_debug = False
 
 
 def 解析Paper(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt):
-    import time
+    import time, glob, os
-    import os
     print('begin analysis on:', file_manifest)
     for index, fp in enumerate(file_manifest):
         with open(fp, 'r', encoding='utf-8', errors='replace') as f:
@@ -22,13 +20,10 @@ def 解析Paper(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbo
         if not fast_debug: 
             msg = '正常'
             # ** gpt request **
-            gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say_show_user, llm_kwargs,
+            gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say_show_user, llm_kwargs, chatbot, history=[], sys_prompt=system_prompt)   # 带超时倒计时
-                                                                               chatbot, history=[],
-                                                                               sys_prompt=system_prompt)  # 带超时倒计时
 
             chatbot[-1] = (i_say_show_user, gpt_say)
-            history.append(i_say_show_user);
+            history.append(i_say_show_user); history.append(gpt_say)
-            history.append(gpt_say)
             yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
             if not fast_debug: time.sleep(2)
 
@@ -40,31 +35,25 @@ def 解析Paper(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbo
     if not fast_debug: 
         msg = '正常'
         # ** gpt request **
-        gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say, llm_kwargs, chatbot,
+        gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say, llm_kwargs, chatbot, history=history, sys_prompt=system_prompt)   # 带超时倒计时
-                                                                           history=history,
-                                                                           sys_prompt=system_prompt)  # 带超时倒计时
 
         chatbot[-1] = (i_say, gpt_say)
-        history.append(i_say)
+        history.append(i_say); history.append(gpt_say)
-        history.append(gpt_say)
         yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
         res = write_results_to_file(history)
         chatbot.append(("完成了吗？", res))
         yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
 
 
+
 @CatchException
-def 读文章写摘要(txt, llm_kwargs, plugin_kwargs, chatbot, system_prompt, web_port, history=None):
+def 读文章写摘要(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
-    # history = []    # 清空历史，以免输入溢出
-    if history is None:
     history = []    # 清空历史，以免输入溢出
-    import glob
+    import glob, os
-    import os
     if os.path.exists(txt):
         project_folder = txt
     else:
-        if txt == "":
+        if txt == "": txt = '空空如也的输入栏'
-            txt = '空空如也的输入栏'
         report_execption(chatbot, history, a = f"解析项目: {txt}", b = f"找不到本地项目或无权访问: {txt}")
         yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
         return

crazy_functions/谷歌检索小助手.py

@@ -104,7 +104,7 @@ def 谷歌检索小助手(txt, llm_kwargs, plugin_kwargs, chatbot, history, syst
             meta_paper_info_list = meta_paper_info_list[batchsize:]
 
     chatbot.append(["状态？", 
-        "已经全部完成，您可以试试让AI写一个Related Works，例如您可以继续输入Write a \"Related Works\" section about \"你搜索的研究领域\" for me."])
+        "已经全部完成，您可以试试让AI写一个Related Works，例如您可以继续输入Write an academic \"Related Works\" section about \"你搜索的研究领域\" for me."])
     msg = '正常'
     yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
     res = write_results_to_file(history)

crazy_functions/辅助回答.py

@@ -1,28 +0,0 @@
-# encoding: utf-8
-# @Time   : 2023/4/19
-# @Author : Spike
-# @Descr   :
-from toolbox import update_ui
-from toolbox import CatchException, report_execption, write_results_to_file
-from crazy_functions.crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
-
-
-@CatchException
-def 猜你想问(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
-    if txt:
-        show_say = txt
-        prompt = txt+'\n回答完问题后，再列出用户可能提出的三个问题。'
-    else:
-        prompt = history[-1]+"\n分析上述回答，再列出用户可能提出的三个问题。"
-        show_say = '分析上述回答，再列出用户可能提出的三个问题。'
-    gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
-        inputs=prompt,
-        inputs_show_user=show_say,
-        llm_kwargs=llm_kwargs,
-        chatbot=chatbot,
-        history=history,
-        sys_prompt=system_prompt
-    )
-    chatbot[-1] = (show_say, gpt_say)
-    history.extend([show_say, gpt_say])
-    yield from update_ui(chatbot=chatbot, history=history)  # 刷新界面

request_llm/bridge_chatgpt.py

@@ -28,7 +28,6 @@ proxies, API_KEY, TIMEOUT_SECONDS, MAX_RETRY = \
 timeout_bot_msg = '[Local Message] Request timeout. Network error. Please check proxy settings in config.py.' + \
                   '网络错误，检查代理服务器是否可用，以及代理设置的格式是否正确，格式须是[协议]://[地址]:[端口]，缺一不可。'
 
-
 def get_full_error(chunk, stream_response):
     """
         获取完整的从Openai返回的报错
@@ -41,9 +40,7 @@ def get_full_error(chunk, stream_response):
     return chunk
 
 
-def predict_no_ui_long_connection(
+def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="", observe_window=None, console_slience=False):
-        inputs, llm_kwargs, history=None, sys_prompt="", observe_window=None, console_slience=False
-):
     """
     发送至chatGPT，等待回复，一次性完成，不显示中间过程。但内部用stream的方法避免中途网线被掐。
     inputs：
@@ -57,59 +54,45 @@ def predict_no_ui_long_connection(
     observe_window = None：
         用于负责跨越线程传递已经输出的部分，大部分时候仅仅为了fancy的视觉效果，留空即可。observe_window[0]：观测窗。observe_window[1]：看门狗
     """
-    if history is None:
-        history = []
     watch_dog_patience = 5 # 看门狗的耐心, 设置5秒即可
     headers, payload = generate_payload(inputs, llm_kwargs, history, system_prompt=sys_prompt, stream=True)
     retry = 0
-    from bridge_all import model_info
     while True:
         try:
             # make a POST request to the API endpoint, stream=False
+            from .bridge_all import model_info
             endpoint = model_info[llm_kwargs['llm_model']]['endpoint']
             response = requests.post(endpoint, headers=headers, proxies=proxies,
-                                     json=payload, stream=True, timeout=TIMEOUT_SECONDS)
+                                    json=payload, stream=True, timeout=TIMEOUT_SECONDS); break
-            stream_response = response.iter_lines()
+        except requests.exceptions.ReadTimeout as e:
-            break
-        except (requests.exceptions.ReadTimeout, requests.exceptions.ConnectionError):
             retry += 1
             traceback.print_exc()
-            if retry > MAX_RETRY:
+            if retry > MAX_RETRY: raise TimeoutError
-                raise TimeoutError
+            if MAX_RETRY!=0: print(f'请求超时，正在重试 ({retry}/{MAX_RETRY}) ……')
-            if MAX_RETRY != 0:
-                print(f'请求超时，正在重试 ({retry}/{MAX_RETRY}) ……')
-        except Exception as e:
-            print(f"出现异常：{e}")
-            raise e
 
+    stream_response =  response.iter_lines()
     result = ''
     while True:
-        try:
+        try: chunk = next(stream_response).decode()
-            chunk = next(stream_response).decode()
         except StopIteration: 
             break
-        # except requests.exceptions.ConnectionError:
+        except requests.exceptions.ConnectionError:
-        #     chunk = next(stream_response).decode()  # 失败了，重试一次？再失败就没办法了。
+            chunk = next(stream_response).decode() # 失败了，重试一次？再失败就没办法了。
-        if len(chunk) == 0:
+        if len(chunk)==0: continue
-            continue
         if not chunk.startswith('data:'): 
             error_msg = get_full_error(chunk.encode('utf8'), stream_response).decode()
             if "reduce the length" in error_msg:
                 raise ConnectionAbortedError("OpenAI拒绝了请求:" + error_msg)
             else:
                 raise RuntimeError("OpenAI拒绝了请求：" + error_msg)
-        if 'data: [DONE]' in chunk:
+        if ('data: [DONE]' in chunk): break # api2d 正常完成
-            break  # api2d 正常完成
         json_data = json.loads(chunk.lstrip('data:'))['choices'][0]
         delta = json_data["delta"]
-        if len(delta) == 0:
+        if len(delta) == 0: break
-            break
+        if "role" in delta: continue
-        if "role" in delta:
-            continue
         if "content" in delta: 
             result += delta["content"]
-            if not console_slience:
+            if not console_slience: print(delta["content"], end='')
-                print(delta["content"], end='')
             if observe_window is not None: 
                 # 观测窗，把已经获取的数据显示出去
                 if len(observe_window) >= 1: observe_window[0] += delta["content"]
@@ -117,8 +100,7 @@ def predict_no_ui_long_connection(
                 if len(observe_window) >= 2:  
                     if (time.time()-observe_window[1]) > watch_dog_patience:
                         raise RuntimeError("用户取消了程序。")
-        else:
+        else: raise RuntimeError("意外Json结构："+delta)
-            raise RuntimeError("意外Json结构："+delta)
     if json_data['finish_reason'] == 'length':
         raise ConnectionAbortedError("正常结束，但显示Token不足，导致输出不完整，请削减单次输入的文本量。")
     return result
@@ -246,7 +228,6 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
                     yield from update_ui(chatbot=chatbot, history=history, msg="Json异常" + error_msg) # 刷新界面
                     return
 
-
 def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
     """
     整合所有信息，选择LLM模型，生成http请求，为发送请求做准备
@@ -266,19 +247,23 @@ def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
     messages = [{"role": "system", "content": system_prompt}]
     if conversation_cnt:
         for index in range(0, 2*conversation_cnt, 2):
-            what_i_have_asked = {"role": "user", "content": history[index]}
+            what_i_have_asked = {}
-            what_gpt_answer = {"role": "assistant", "content": history[index + 1]}
+            what_i_have_asked["role"] = "user"
+            what_i_have_asked["content"] = history[index]
+            what_gpt_answer = {}
+            what_gpt_answer["role"] = "assistant"
+            what_gpt_answer["content"] = history[index+1]
             if what_i_have_asked["content"] != "":
-                if what_gpt_answer["content"] == "":
+                if what_gpt_answer["content"] == "": continue
-                    continue
+                if what_gpt_answer["content"] == timeout_bot_msg: continue
-                if what_gpt_answer["content"] == timeout_bot_msg:
-                    continue
                 messages.append(what_i_have_asked)
                 messages.append(what_gpt_answer)
             else:
                 messages[-1]['content'] = what_gpt_answer['content']
 
-    what_i_ask_now = {"role": "user", "content": inputs}
+    what_i_ask_now = {}
+    what_i_ask_now["role"] = "user"
+    what_i_ask_now["content"] = inputs
     messages.append(what_i_ask_now)
 
     payload = {
@@ -293,8 +278,8 @@ def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
     }
     try:
         print(f" {llm_kwargs['llm_model']} : {conversation_cnt} : {inputs[:100]} ..........")
-    except Exception as e:
+    except:
-        print(f'输入中可能存在乱码。抛出异常: {e}')
+        print('输入中可能存在乱码。')
     return headers,payload
 
 

requirements.txt

@@ -1,4 +1,3 @@
-gradio>=3.33.1
 tiktoken>=0.3.3
 requests[socks]
 transformers
@@ -16,5 +15,3 @@ openai
 numpy
 arxiv
 rich
-langchain
-zh_langchain

toolbox.py

@@ -21,7 +21,6 @@ pj = os.path.join
 ========================================================================
 """
 
-
 class ChatBotWithCookies(list):
     def __init__(self, cookie):
         self._cookies = cookie
@@ -72,13 +71,11 @@ def update_ui(chatbot, history, msg='正常', **kwargs):  # 刷新界面
     assert isinstance(chatbot, ChatBotWithCookies), "在传递chatbot的过程中不要将其丢弃。必要时，可用clear将其清空，然后用for+append循环重新赋值。"
     yield chatbot.get_cookies(), chatbot, history, msg
 
-
 def update_ui_lastest_msg(lastmsg, chatbot, history, delay=1):  # 刷新界面
     """
     刷新用户界面
     """
-    if len(chatbot) == 0:
+    if len(chatbot) == 0: chatbot.append(["update_ui_last_msg", lastmsg])
-        chatbot.append(["update_ui_last_msg", lastmsg])
     chatbot[-1] = list(chatbot[-1])
     chatbot[-1][-1] = lastmsg
     yield from update_ui(chatbot=chatbot, history=history)
@@ -86,25 +83,24 @@ def update_ui_lastest_msg(lastmsg, chatbot, history, delay=1):  # 刷新界面
 
 
 def trimmed_format_exc():
-    import os
+    import os, traceback
-    import traceback
+    str = traceback.format_exc()
-    _str = traceback.format_exc()
     current_path = os.getcwd()
     replace_path = "."
-    return _str.replace(current_path, replace_path)
+    return str.replace(current_path, replace_path)
-
 
 def CatchException(f):
     """
     装饰器函数，捕捉函数f中的异常并封装到一个生成器中返回，并显示到聊天当中。
     """
+
     @wraps(f)
     def decorated(txt, top_p, temperature, chatbot, history, systemPromptTxt, WEB_PORT=-1):
         try:
             yield from f(txt, top_p, temperature, chatbot, history, systemPromptTxt, WEB_PORT)
         except Exception as e:
             from check_proxy import check_proxy
-            # from toolbox import get_conf  # 不需要导入本文件内容
+            from toolbox import get_conf
             proxies, = get_conf('proxies')
             tb_str = '```\n' + trimmed_format_exc() + '```'
             if len(chatbot) == 0:
@@ -152,7 +148,6 @@ def HotReload(f):
 ========================================================================
 """
 
-
 def get_reduce_token_percent(text):
     """
         * 此函数未来将被弃用
@@ -212,6 +207,8 @@ def regular_txt_to_markdown(text):
     return text
 
 
+
+
 def report_execption(chatbot, history, a, b):
     """
     向chatbot中添加错误信息
@@ -241,7 +238,6 @@ def text_divide_paragraph(text):
         text = "</br>".join(lines)
         return pre + text + suf
 
-
 @lru_cache(maxsize=128) # 使用 lru缓存 加快转换速度
 def markdown_convertion(txt):
     """
@@ -444,7 +440,6 @@ def find_recent_files(directory):
 
     return recent_files
 
-
 def promote_file_to_downloadzone(file, rename_file=None, chatbot=None):
     # 将文件复制一份到下载区
     import shutil
@@ -457,7 +452,6 @@ def promote_file_to_downloadzone(file, rename_file=None, chatbot=None):
         else: current = []
         chatbot._cookies.update({'file_to_promote': [new_path] + current})
 
-
 def on_file_uploaded(files, chatbot, txt, txt2, checkboxes):
     """
     当文件被上传时的回调函数
@@ -511,20 +505,17 @@ def on_report_generated(cookies, files, chatbot):
     chatbot.append(['报告如何远程获取？', f'报告已经添加到右侧“文件上传区”（可能处于折叠状态），请查收。{file_links}'])
     return cookies, report_files, chatbot
 
-
 def is_openai_api_key(key):
     API_MATCH_ORIGINAL = re.match(r"sk-[a-zA-Z0-9]{48}$", key)
     API_MATCH_AZURE = re.match(r"[a-zA-Z0-9]{32}$", key)
     return bool(API_MATCH_ORIGINAL) or bool(API_MATCH_AZURE)
 
-
 def is_api2d_key(key):
     if key.startswith('fk') and len(key) == 41:
         return True
     else:
         return False
 
-
 def is_any_api_key(key):
     if ',' in key:
         keys = key.split(',')
@@ -534,7 +525,6 @@ def is_any_api_key(key):
     else:
         return is_openai_api_key(key) or is_api2d_key(key)
 
-
 def what_keys(keys):
     avail_key_list = {'OpenAI Key':0, "API2D Key":0}
     key_list = keys.split(',')
@@ -549,7 +539,6 @@ def what_keys(keys):
 
     return f"检测到： OpenAI Key {avail_key_list['OpenAI Key']} 个，API2D Key {avail_key_list['API2D Key']} 个"
 
-
 def select_api_key(keys, llm_model):
     import random
     avail_key_list = []
@@ -569,7 +558,6 @@ def select_api_key(keys, llm_model):
     api_key = random.choice(avail_key_list) # 随机负载均衡
     return api_key
 
-
 def read_env_variable(arg, default_value):
     """
     环境变量可以是 `GPT_ACADEMIC_CONFIG`(优先)，也可以直接是`CONFIG`
@@ -624,7 +612,6 @@ def read_env_variable(arg, default_value):
     print亮绿(f"[ENV_VAR] 成功读取环境变量{arg}")
     return r
 
-
 @lru_cache(maxsize=128)
 def read_single_conf_with_lru_cache(arg):
     from colorful import print亮红, print亮绿, print亮蓝
@@ -689,7 +676,6 @@ class DummyWith():
     def __exit__(self, exc_type, exc_value, traceback):
         return
 
-
 def run_gradio_in_subpath(demo, auth, port, custom_path):
     """
     把gradio的运行地址更改到指定的二次路径上
@@ -784,7 +770,6 @@ def clip_history(inputs, history, tokenizer, max_token_limit):
 ========================================================================
 """
 
-
 def zip_folder(source_folder, dest_folder, zip_name):
     import zipfile
     import os
@@ -816,7 +801,6 @@ def zip_folder(source_folder, dest_folder, zip_name):
 
     print(f"Zip file created at {zip_file}")
 
-
 def zip_result(folder):
     import time
     t = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
@@ -827,7 +811,6 @@ def gen_time_str():
     import time
     return time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
 
-
 class ProxyNetworkActivate():
     """
     这段代码定义了一个名为TempProxy的空上下文管理器, 用于给一小段代码上代理
@@ -847,14 +830,12 @@ class ProxyNetworkActivate():
         if 'HTTPS_PROXY' in os.environ: os.environ.pop('HTTPS_PROXY')
         return
 
-
 def objdump(obj, file='objdump.tmp'):
     import pickle
     with open(file, 'wb+') as f:
         pickle.dump(obj, f)
     return
 
-
 def objload(file='objdump.tmp'):
     import pickle, os
     if not os.path.exists(file):