Merge branch 'master' into improve_ui_master

crazy_functions/chatglm微调工具.py

@@ -0,0 +1,71 @@
+from toolbox import CatchException, update_ui, promote_file_to_downloadzone
+from .crazy_utils import request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
+import datetime, json
+
+def fetch_items(list_of_items, batch_size):
+    for i in range(0, len(list_of_items), batch_size):
+        yield list_of_items[i:i + batch_size]
+
+def string_to_options(arguments):
+    import argparse
+    import shlex
+
+    # Create an argparse.ArgumentParser instance
+    parser = argparse.ArgumentParser()
+
+    # Add command-line arguments
+    parser.add_argument("--llm_to_learn", type=str, help="LLM model to learn", default="gpt-3.5-turbo")
+    parser.add_argument("--prompt_prefix", type=str, help="Prompt prefix", default='')
+    parser.add_argument("--system_prompt", type=str, help="System prompt", default='')
+    parser.add_argument("--batch", type=int, help="System prompt", default=50)
+
+    # Parse the arguments
+    args = parser.parse_args(shlex.split(arguments))
+
+    return args
+
+@CatchException
+def 微调数据集生成(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+    """
+    txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
+    llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数
+    chatbot         聊天显示框的句柄，用于显示给用户
+    history         聊天历史，前情提要
+    system_prompt   给gpt的静默提醒
+    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(("这是什么功能？", "[Local Message] 微调数据集生成"))
+    if ("advanced_arg" in plugin_kwargs) and (plugin_kwargs["advanced_arg"] == ""): plugin_kwargs.pop("advanced_arg")
+    args = plugin_kwargs.get("advanced_arg", None)
+    if args is None: 
+        chatbot.append(("没给定指令", "退出"))
+        yield from update_ui(chatbot=chatbot, history=history); return
+    else:
+        arguments = string_to_options(arguments=args)
+
+    dat = []
+    with open(txt, 'r', encoding='utf8') as f:
+        for line in f.readlines():
+            json_dat = json.loads(line)
+            dat.append(json_dat["content"])
+
+    llm_kwargs['llm_model'] = arguments.llm_to_learn
+    for batch in fetch_items(dat, arguments.batch):
+        res = yield from request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
+            inputs_array=[f"{arguments.prompt_prefix}\n\n{b}" for b in (batch)],
+            inputs_show_user_array=[f"Show Nothing" for _ in (batch)],
+            llm_kwargs=llm_kwargs,
+            chatbot=chatbot,
+            history_array=[[] for _ in (batch)],
+            sys_prompt_array=[arguments.system_prompt for _ in (batch)],
+            max_workers=10  # OpenAI所允许的最大并行过载
+        )
+    
+        with open(txt+'.generated.json', 'a+', encoding='utf8') as f:
+            for b, r in zip(batch, res[1::2]):
+                f.write(json.dumps({"content":b, "summary":r}, ensure_ascii=False)+'\n')
+
+    promote_file_to_downloadzone(txt+'.generated.json', rename_file='generated.json', chatbot=chatbot)
+    return

crazy_functions/crazy_functions_test.py

@@ -211,22 +211,30 @@ def test_Latex():
     # # for cookies, cb, hist, msg in silence_stdout(编译Latex)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
     #     cli_printer.print(cb)   #  print(cb)
 
+def test_chatglm_finetune():
+    from crazy_functions.chatglm微调工具 import 微调数据集生成
+    txt = 'build/dev.json'
+    plugin_kwargs = {"advanced_arg":"--llm_to_learn=gpt-3.5-turbo --prompt_prefix='根据下面的服装类型提示，想象一个穿着者，对这个人外貌、身处的环境、内心世界、人设进行描写。要求：100字以内，用第二人称。' --system_prompt=''" }
+
+    for cookies, cb, hist, msg in (微调数据集生成)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+        cli_printer.print(cb)
 
 
-# test_解析一个Python项目()
-# test_Latex英文润色()
-# test_Markdown中译英()
-# test_批量翻译PDF文档()
-# test_谷歌检索小助手()
-# test_总结word文档()
-# test_下载arxiv论文并翻译摘要()
-# test_解析一个Cpp项目()
-# test_联网回答问题()
-# test_解析ipynb文件()
-# test_数学动画生成manim()
-# test_Langchain知识库()
-# test_Langchain知识库读取()
 if __name__ == "__main__":
-    test_Latex()
+    # test_解析一个Python项目()
+    # test_Latex英文润色()
+    # test_Markdown中译英()
+    # test_批量翻译PDF文档()
+    # test_谷歌检索小助手()
+    # test_总结word文档()
+    # test_下载arxiv论文并翻译摘要()
+    # test_解析一个Cpp项目()
+    # test_联网回答问题()
+    # test_解析ipynb文件()
+    # test_数学动画生成manim()
+    # test_Langchain知识库()
+    # test_Langchain知识库读取()
+    # test_Latex()
+    test_chatglm_finetune()
     input("程序完成，回车退出。")
     print("退出。")

crazy_functions/crazy_utils.py

@@ -130,6 +130,11 @@ def request_gpt_model_in_new_thread_with_ui_alive(
     yield from update_ui(chatbot=chatbot, history=[]) # 如果最后成功了，则删除报错信息
     return final_result
 
+def can_multi_process(llm):
+    if llm.startswith('gpt-'): return True
+    if llm.startswith('api2d-'): return True
+    if llm.startswith('azure-'): return True
+    return False
 
 def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
         inputs_array, inputs_show_user_array, llm_kwargs, 
@@ -175,7 +180,7 @@ def request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
         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-')):
+    if not can_multi_process(llm_kwargs['llm_model']):
         max_workers = 1
         
     executor = ThreadPoolExecutor(max_workers=max_workers)

crazy_functions/latex_utils.py

@@ -189,6 +189,18 @@ def rm_comments(main_file):
     main_file = re.sub(r'(?<!\\)%.*', '', main_file)  # 使用正则表达式查找半行注释, 并替换为空字符串
     return main_file
 
+def find_tex_file_ignore_case(fp):
+    dir_name = os.path.dirname(fp)
+    base_name = os.path.basename(fp)
+    if not base_name.endswith('.tex'): base_name+='.tex'
+    if os.path.exists(pj(dir_name, base_name)): return pj(dir_name, base_name)
+    # go case in-sensitive
+    import glob
+    for f in glob.glob(dir_name+'/*.tex'):
+        base_name_s = os.path.basename(fp)
+        if base_name_s.lower() == base_name.lower(): return f
+    return None
+
 def merge_tex_files_(project_foler, main_file, mode):
     """
     Merge Tex project recrusively
@@ -197,15 +209,11 @@ def merge_tex_files_(project_foler, main_file, mode):
     for s in reversed([q for q in re.finditer(r"\\input\{(.*?)\}", main_file, re.M)]):
         f = s.group(1)
         fp = os.path.join(project_foler, f)
-        if os.path.exists(fp):  
-            # e.g., \input{srcs/07_appendix.tex}
-            with open(fp, 'r', encoding='utf-8', errors='replace') as fx:
-                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()
+        fp = find_tex_file_ignore_case(fp)
+        if fp:
+            with open(fp, 'r', encoding='utf-8', errors='replace') as fx: c = fx.read()
+        else:
+            raise RuntimeError(f'找不到{fp}，Tex源文件缺失！')
         c = merge_tex_files_(project_foler, c, mode)
         main_file = main_file[:s.span()[0]] + c + main_file[s.span()[1]:]
     return main_file
@@ -324,7 +332,7 @@ def split_subprocess(txt, project_folder, return_dict, opts):
     # 吸收在42行以内的begin-end组合
     text, mask = set_forbidden_text_begin_end(text, mask, r"\\begin\{([a-z\*]*)\}(.*?)\\end\{\1\}", re.DOTALL, limit_n_lines=42)
     # 吸收匿名公式
-    text, mask = set_forbidden_text(text, mask, [ r"\$\$(.*?)\$\$",  r"\\\[.*?\\\]" ], re.DOTALL)
+    text, mask = set_forbidden_text(text, mask, [ r"\$\$([^$]+)\$\$",  r"\\\[.*?\\\]" ], re.DOTALL)
     # 吸收其他杂项
     text, mask = set_forbidden_text(text, mask, [ r"\\section\{(.*?)\}", r"\\section\*\{(.*?)\}", r"\\subsection\{(.*?)\}", r"\\subsubsection\{(.*?)\}" ])
     text, mask = set_forbidden_text(text, mask, [ r"\\bibliography\{(.*?)\}", r"\\bibliographystyle\{(.*?)\}" ])

crazy_functions/交互功能函数模板.py

@@ -0,0 +1,63 @@
+from toolbox import CatchException, update_ui
+from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
+
+
+@CatchException
+def 交互功能模板函数(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+    """
+    txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
+    llm_kwargs      gpt模型参数, 如温度和top_p等, 一般原样传递下去就行
+    plugin_kwargs   插件模型的参数, 如温度和top_p等, 一般原样传递下去就行
+    chatbot         聊天显示框的句柄，用于显示给用户
+    history         聊天历史，前情提要
+    system_prompt   给gpt的静默提醒
+    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(("这是什么功能？", "交互功能函数模板。在执行完成之后, 可以将自身的状态存储到cookie中, 等待用户的再次调用。"))
+    yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+
+    state = chatbot._cookies.get('plugin_state_0001', None) # 初始化插件状态
+
+    if state is None:
+        chatbot._cookies['lock_plugin'] = 'crazy_functions.交互功能函数模板->交互功能模板函数'      # 赋予插件锁定 锁定插件回调路径，当下一次用户提交时，会直接转到该函数
+        chatbot._cookies['plugin_state_0001'] = 'wait_user_keyword'                              # 赋予插件状态
+
+        chatbot.append(("第一次调用：", "请输入关键词, 我将为您查找相关壁纸, 建议使用英文单词, 插件锁定中，请直接提交即可。"))
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        return
+
+    if state == 'wait_user_keyword':
+        chatbot._cookies['lock_plugin'] = None          # 解除插件锁定，避免遗忘导致死锁
+        chatbot._cookies['plugin_state_0001'] = None    # 解除插件状态，避免遗忘导致死锁
+
+        # 解除插件锁定
+        chatbot.append((f"获取关键词：{txt}", ""))
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        page_return = get_image_page_by_keyword(txt)
+        inputs=inputs_show_user=f"Extract all image urls in this html page, pick the first 5 images and show them with markdown format: \n\n {page_return}"
+        gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
+            inputs=inputs, inputs_show_user=inputs_show_user,
+            llm_kwargs=llm_kwargs, chatbot=chatbot, history=[], 
+            sys_prompt="When you want to show an image, use markdown format. e.g. ![image_description](image_url). If there are no image url provided, answer 'no image url provided'"
+        )
+        chatbot[-1] = [chatbot[-1][0], gpt_say]
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        return
+
+
+
+# ---------------------------------------------------------------------------------
+
+def get_image_page_by_keyword(keyword):
+    import requests
+    from bs4 import BeautifulSoup
+    response = requests.get(f'https://wallhaven.cc/search?q={keyword}', timeout=2)
+    res = "image urls: \n"
+    for image_element in BeautifulSoup(response.content, 'html.parser').findAll("img"):
+        try:
+            res += image_element["data-src"]
+            res += "\n"
+        except:
+            pass
+    return res

crazy_functions/总结word文档.py

@@ -14,17 +14,19 @@ def 解析docx(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot
             doc = Document(fp)
             file_content = "\n".join([para.text for para in doc.paragraphs])
         else:
-            import win32com.client
-            word = win32com.client.Dispatch("Word.Application")
-            word.visible = False
-            # 打开文件
-            print('fp', os.getcwd())
-            doc = word.Documents.Open(os.getcwd() + '/' + fp)
-            # file_content = doc.Content.Text
-            doc = word.ActiveDocument
-            file_content = doc.Range().Text
-            doc.Close()
-            word.Quit()
+            try:
+                import win32com.client
+                word = win32com.client.Dispatch("Word.Application")
+                word.visible = False
+                # 打开文件
+                doc = word.Documents.Open(os.getcwd() + '/' + fp)
+                # file_content = doc.Content.Text
+                doc = word.ActiveDocument
+                file_content = doc.Range().Text
+                doc.Close()
+                word.Quit()
+            except:
+                raise RuntimeError('请先将.doc文档转换为.docx文档。')
 
         print(file_content)
         # private_upload里面的文件名在解压zip后容易出现乱码（rar和7z格式正常），故可以只分析文章内容，不输入文件名

crazy_functions/批量总结PDF文档.py

@@ -1,121 +1,107 @@
-from toolbox import update_ui
+from toolbox import update_ui, promote_file_to_downloadzone, gen_time_str
 from toolbox import CatchException, report_execption, write_results_to_file
-import re
-import unicodedata
-fast_debug = False
 from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
+from .crazy_utils import read_and_clean_pdf_text
+from .crazy_utils import input_clipping
 
-def is_paragraph_break(match):
-    """
-    根据给定的匹配结果来判断换行符是否表示段落分隔。
-    如果换行符前为句子结束标志（句号，感叹号，问号），且下一个字符为大写字母，则换行符更有可能表示段落分隔。
-    也可以根据之前的内容长度来判断段落是否已经足够长。
-    """
-    prev_char, next_char = match.groups()
 
-    # 句子结束标志
-    sentence_endings = ".!?"
-
-    # 设定一个最小段落长度阈值
-    min_paragraph_length = 140
-
-    if prev_char in sentence_endings and next_char.isupper() and len(match.string[:match.start(1)]) > min_paragraph_length:
-        return "\n\n" 
-    else:
-        return " "
-
-def normalize_text(text):
-    """
-    通过把连字（ligatures）等文本特殊符号转换为其基本形式来对文本进行归一化处理。
-    例如，将连字 "fi" 转换为 "f" 和 "i"。
-    """
-    # 对文本进行归一化处理，分解连字
-    normalized_text = unicodedata.normalize("NFKD", text)
-
-    # 替换其他特殊字符
-    cleaned_text = re.sub(r'[^\x00-\x7F]+', '', normalized_text)
-
-    return cleaned_text
-
-def clean_text(raw_text):
-    """
-    对从 PDF 提取出的原始文本进行清洗和格式化处理。
-    1. 对原始文本进行归一化处理。
-    2. 替换跨行的连词
-    3. 根据 heuristic 规则判断换行符是否是段落分隔，并相应地进行替换
-    """
-    # 对文本进行归一化处理
-    normalized_text = normalize_text(raw_text)
-
-    # 替换跨行的连词
-    text = re.sub(r'(\w+-\n\w+)', lambda m: m.group(1).replace('-\n', ''), normalized_text)
-
-    # 根据前后相邻字符的特点，找到原文本中的换行符
-    newlines = re.compile(r'(\S)\n(\S)')
-
-    # 根据 heuristic 规则，用空格或段落分隔符替换原换行符
-    final_text = re.sub(newlines, lambda m: m.group(1) + is_paragraph_break(m) + m.group(2), text)
-
-    return final_text.strip()
 
 def 解析PDF(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt):
-    import time, glob, os, fitz
-    print('begin analysis on:', file_manifest)
-    for index, fp in enumerate(file_manifest):
-        with fitz.open(fp) as doc:
-            file_content = ""
-            for page in doc:
-                file_content += page.get_text()
-            file_content = clean_text(file_content)
-            print(file_content)
+    file_write_buffer = []
+    for file_name in file_manifest:
+        print('begin analysis on:', file_name)
+        ############################## <第 0 步，切割PDF> ##################################
+        # 递归地切割PDF文件，每一块（尽量是完整的一个section，比如introduction，experiment等，必要时再进行切割）
+        # 的长度必须小于 2500 个 Token
+        file_content, page_one = read_and_clean_pdf_text(file_name) # （尝试）按照章节切割PDF
+        file_content = file_content.encode('utf-8', 'ignore').decode()   # avoid reading non-utf8 chars
+        page_one = str(page_one).encode('utf-8', 'ignore').decode()  # avoid reading non-utf8 chars
+        
+        TOKEN_LIMIT_PER_FRAGMENT = 2500
 
-        prefix = "接下来请你逐文件分析下面的论文文件，概括其内容" if index==0 else ""
-        i_say = prefix + f'请对下面的文章片段用中文做一个概述，文件名是{os.path.relpath(fp, project_folder)}，文章内容是 ```{file_content}```'
-        i_say_show_user = prefix + f'[{index}/{len(file_manifest)}] 请对下面的文章片段做一个概述: {os.path.abspath(fp)}'
-        chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
-        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        from .crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
+        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=()))
+        paper_fragments = breakdown_txt_to_satisfy_token_limit_for_pdf(
+            txt=file_content,  get_token_fn=get_token_num, limit=TOKEN_LIMIT_PER_FRAGMENT)
+        page_one_fragments = breakdown_txt_to_satisfy_token_limit_for_pdf(
+            txt=str(page_one), get_token_fn=get_token_num, limit=TOKEN_LIMIT_PER_FRAGMENT//4)
+        # 为了更好的效果，我们剥离Introduction之后的部分（如果有）
+        paper_meta = page_one_fragments[0].split('introduction')[0].split('Introduction')[0].split('INTRODUCTION')[0]
+        
+        ############################## <第 1 步，从摘要中提取高价值信息，放到history中> ##################################
+        final_results = []
+        final_results.append(paper_meta)
 
-        if not fast_debug: 
-            msg = '正常'
-            # ** gpt request **
-            gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
-                inputs=i_say, 
-                inputs_show_user=i_say_show_user, 
-                llm_kwargs=llm_kwargs,
-                chatbot=chatbot, 
-                history=[],
-                sys_prompt="总结文章。"
-            )  # 带超时倒计时
-                
+        ############################## <第 2 步，迭代地历遍整个文章，提取精炼信息> ##################################
+        i_say_show_user = f'首先你在中文语境下通读整篇论文。'; gpt_say = "[Local Message] 收到。"           # 用户提示
+        chatbot.append([i_say_show_user, gpt_say]); yield from update_ui(chatbot=chatbot, history=[])    # 更新UI
 
-            chatbot[-1] = (i_say_show_user, gpt_say)
-            history.append(i_say_show_user); 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)
+        iteration_results = []
+        last_iteration_result = paper_meta  # 初始值是摘要
+        MAX_WORD_TOTAL = 4096 * 0.7
+        n_fragment = len(paper_fragments)
+        if n_fragment >= 20: print('文章极长，不能达到预期效果')
+        for i in range(n_fragment):
+            NUM_OF_WORD = MAX_WORD_TOTAL // n_fragment
+            i_say = f"Read this section, recapitulate the content of this section with less than {NUM_OF_WORD} Chinese characters: {paper_fragments[i]}"
+            i_say_show_user = f"[{i+1}/{n_fragment}] Read this section, recapitulate the content of this section with less than {NUM_OF_WORD} Chinese characters: {paper_fragments[i][:200]}"
+            gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say_show_user,  # i_say=真正给chatgpt的提问， i_say_show_user=给用户看的提问
+                                                                                llm_kwargs, chatbot, 
+                                                                                history=["The main idea of the previous section is?", last_iteration_result], # 迭代上一次的结果
+                                                                                sys_prompt="Extract the main idea of this section with Chinese."  # 提示
+                                                                                ) 
+            iteration_results.append(gpt_say)
+            last_iteration_result = gpt_say
 
-    all_file = ', '.join([os.path.relpath(fp, project_folder) for index, fp in enumerate(file_manifest)])
-    i_say = f'根据以上你自己的分析，对全文进行概括，用学术性语言写一段中文摘要，然后再写一段英文摘要（包括{all_file}）。'
-    chatbot.append((i_say, "[Local Message] waiting gpt response."))
-    yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
-
-    if not fast_debug: 
-        msg = '正常'
-        # ** gpt request **
+        ############################## <第 3 步，整理history，提取总结> ##################################
+        final_results.extend(iteration_results)
+        final_results.append(f'Please conclude this paper discussed above。')
+        # This prompt is from https://github.com/kaixindelele/ChatPaper/blob/main/chat_paper.py
+        NUM_OF_WORD = 1000
+        i_say = """
+1. Mark the title of the paper (with Chinese translation)
+2. list all the authors' names (use English)
+3. mark the first author's affiliation (output Chinese translation only)
+4. mark the keywords of this article (use English)
+5. link to the paper, Github code link (if available, fill in Github:None if not)
+6. summarize according to the following four points.Be sure to use Chinese answers (proper nouns need to be marked in English)
+    - (1):What is the research background of this article?
+    - (2):What are the past methods? What are the problems with them? Is the approach well motivated?
+    - (3):What is the research methodology proposed in this paper?
+    - (4):On what task and what performance is achieved by the methods in this paper? Can the performance support their goals?
+Follow the format of the output that follows:                  
+1. Title: xxx\n\n
+2. Authors: xxx\n\n
+3. Affiliation: xxx\n\n
+4. Keywords: xxx\n\n
+5. Urls: xxx or xxx , xxx \n\n
+6. Summary: \n\n
+    - (1):xxx;\n 
+    - (2):xxx;\n 
+    - (3):xxx;\n
+    - (4):xxx.\n\n
+Be sure to use Chinese answers (proper nouns need to be marked in English), statements as concise and academic as possible,
+do not have too much repetitive information, numerical values using the original numbers.
+        """
+        # This prompt is from https://github.com/kaixindelele/ChatPaper/blob/main/chat_paper.py
+        file_write_buffer.extend(final_results)
+        i_say, final_results = input_clipping(i_say, final_results, max_token_limit=2000)
         gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
-            inputs=i_say, 
-            inputs_show_user=i_say, 
-            llm_kwargs=llm_kwargs,
-            chatbot=chatbot, 
-            history=history,
-            sys_prompt="总结文章。"
-        )  # 带超时倒计时
+            inputs=i_say, inputs_show_user='开始最终总结', 
+            llm_kwargs=llm_kwargs, chatbot=chatbot, history=final_results, 
+            sys_prompt= f"Extract the main idea of this paper with less than {NUM_OF_WORD} Chinese characters"
+        )
+        final_results.append(gpt_say)
+        file_write_buffer.extend([i_say, gpt_say])
+        ############################## <第 4 步，设置一个token上限> ##################################
+        _, final_results = input_clipping("", final_results, max_token_limit=3200)
+        yield from update_ui(chatbot=chatbot, history=final_results) # 注意这里的历史记录被替代了
 
-        chatbot[-1] = (i_say, gpt_say)
-        history.append(i_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
+    res = write_results_to_file(file_write_buffer, file_name=gen_time_str())
+    promote_file_to_downloadzone(res.split('\t')[-1], chatbot=chatbot)
+    yield from update_ui(chatbot=chatbot, history=final_results) # 刷新界面
 
 
 @CatchException
@@ -151,10 +137,7 @@ def 批量总结PDF文档(txt, llm_kwargs, plugin_kwargs, chatbot, history, syst
         return
 
     # 搜索需要处理的文件清单
-    file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.pdf', recursive=True)] # + \
-                    # [f for f in glob.glob(f'{project_folder}/**/*.tex', recursive=True)] + \
-                    # [f for f in glob.glob(f'{project_folder}/**/*.cpp', recursive=True)] + \
-                    # [f for f in glob.glob(f'{project_folder}/**/*.c', recursive=True)]
+    file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.pdf', recursive=True)]
     
     # 如果没找到任何文件
     if len(file_manifest) == 0:

crazy_functions/询问多个大语言模型.py

@@ -6,7 +6,7 @@ def 同时问询(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt
     """
     txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
     llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
-    plugin_kwargs   插件模型的参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数，用于灵活调整复杂功能的各种参数
     chatbot         聊天显示框的句柄，用于显示给用户
     history         聊天历史，前情提要
     system_prompt   给gpt的静默提醒
@@ -35,7 +35,7 @@ def 同时问询_指定模型(txt, llm_kwargs, plugin_kwargs, chatbot, history,
     """
     txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
     llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
-    plugin_kwargs   插件模型的参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数，用于灵活调整复杂功能的各种参数
     chatbot         聊天显示框的句柄，用于显示给用户
     history         聊天历史，前情提要
     system_prompt   给gpt的静默提醒

crazy_functions/高级功能函数模板.py

@@ -1,13 +1,12 @@
 from toolbox import CatchException, update_ui
 from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
-import datetime, re
-
+import datetime
 @CatchException
 def 高阶功能模板函数(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
     """
     txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
     llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
-    plugin_kwargs   插件模型的参数，暂时没有用武之地
+    plugin_kwargs   插件模型的参数，用于灵活调整复杂功能的各种参数
     chatbot         聊天显示框的句柄，用于显示给用户
     history         聊天历史，前情提要
     system_prompt   给gpt的静默提醒
@@ -19,34 +18,12 @@ def 高阶功能模板函数(txt, llm_kwargs, plugin_kwargs, chatbot, history, s
     for i in range(5):
         currentMonth = (datetime.date.today() + datetime.timedelta(days=i)).month
         currentDay = (datetime.date.today() + datetime.timedelta(days=i)).day
-        i_say = f'历史中哪些事件发生在{currentMonth}月{currentDay}日？用中文列举两条，然后分别给出描述事件的两个英文单词。' + '当你给出关键词时，使用以下json格式：{"KeyWords":[EnglishKeyWord1,EnglishKeyWord2]}。'
+        i_say = f'历史中哪些事件发生在{currentMonth}月{currentDay}日？列举两条并发送相关图片。发送图片时，请使用Markdown，将Unsplash API中的PUT_YOUR_QUERY_HERE替换成描述该事件的一个最重要的单词。'
         gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
             inputs=i_say, inputs_show_user=i_say, 
             llm_kwargs=llm_kwargs, chatbot=chatbot, history=[], 
-            sys_prompt='输出格式示例：1908年，美国消防救援事业发展的“美国消防协会”成立。关键词：{"KeyWords":["Fire","American"]}。'
+            sys_prompt="当你想发送一张照片时，请使用Markdown, 并且不要有反斜线, 不要用代码块。使用 Unsplash API (https://source.unsplash.com/1280x720/? < PUT_YOUR_QUERY_HERE >)。"
         )
-        gpt_say = get_images(gpt_say)
         chatbot[-1] = (i_say, gpt_say)
         history.append(i_say);history.append(gpt_say)
-        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面 # 界面更新
-
-
-def get_images(gpt_say):
-    def get_image_by_keyword(keyword):
-        import requests
-        from bs4 import BeautifulSoup
-        response = requests.get(f'https://wallhaven.cc/search?q={keyword}', timeout=2)
-        for image_element in BeautifulSoup(response.content, 'html.parser').findAll("img"):
-            if "data-src" in image_element: break
-        return image_element["data-src"]
-
-    for keywords in re.findall('{"KeyWords":\[(.*?)\]}', gpt_say):
-        keywords = [n.strip('"') for n in keywords.split(',')]
-        try:
-            description = keywords[0]
-            url = get_image_by_keyword(keywords[0])
-            img_tag = f"\n\n![{description}]({url})"
-            gpt_say += img_tag
-        except:
-            continue
-    return gpt_say
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面 # 界面更新