chunzhi/chatgpt_academic
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'master' into frontier

Browse Source
This commit is contained in:
binary-husky 2023-09-19 17:21:02 +08:00
commit 5e647ff149
8 changed files with 235 additions and 210 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
README.md
View File

@ -101,9 +101,11 @@ cd gpt_academic
2. 配置API_KEY 2. 配置API_KEY
在`config.py`中，配置API KEY等设置，[点击查看特殊网络环境设置方法](https://github.com/binary-husky/gpt_academic/issues/1) 。 在`config.py`中，配置API KEY等设置，[点击查看特殊网络环境设置方法](https://github.com/binary-husky/gpt_academic/issues/1) 。[Wiki页面](https://github.com/binary-husky/gpt_academic/wiki/%E9%A1%B9%E7%9B%AE%E9%85%8D%E7%BD%AE%E8%AF%B4%E6%98%8E)。
(P.S. 程序运行时会优先检查是否存在名为`config_private.py`的私密配置文件，并用其中的配置覆盖`config.py`的同名配置。因此，如果您能理解我们的配置读取逻辑，我们强烈建议您在`config.py`旁边创建一个名为`config_private.py`的新配置文件，并把`config.py`中的配置转移（复制）到`config_private.py`中（仅复制您修改过的配置条目即可）。`config_private.py`不受git管控，可以让您的隐私信息更加安全。P.S.项目同样支持通过`环境变量`配置大多数选项，环境变量的书写格式参考`docker-compose`文件。读取优先级: `环境变量` > `config_private.py` > `config.py`) 「 程序会优先检查是否存在名为`config_private.py`的私密配置文件，并用其中的配置覆盖`config.py`的同名配置。如您能理解该读取逻辑，我们强烈建议您在`config.py`旁边创建一个名为`config_private.py`的新配置文件，并把`config.py`中的配置转移（复制）到`config_private.py`中（仅复制您修改过的配置条目即可）。 」
「 支持通过`环境变量`配置项目，环境变量的书写格式参考`docker-compose.yml`文件或者我们的[Wiki页面](https://github.com/binary-husky/gpt_academic/wiki/%E9%A1%B9%E7%9B%AE%E9%85%8D%E7%BD%AE%E8%AF%B4%E6%98%8E)。配置读取优先级: `环境变量` > `config_private.py` > `config.py`。 」
3. 安装依赖 3. 安装依赖
@ -111,7 +113,7 @@ cd gpt_academic
# （选择I: 如熟悉python）（python版本3.9以上，越新越好），备注：使用官方pip源或者阿里pip源,临时换源方法：python -m pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/ # （选择I: 如熟悉python）（python版本3.9以上，越新越好），备注：使用官方pip源或者阿里pip源,临时换源方法：python -m pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/
python -m pip install -r requirements.txt python -m pip install -r requirements.txt
# （选择II: 如不熟悉python）使用anaconda，步骤也是类似的 (https://www.bilibili.com/video/BV1rc411W7Dr)： # （选择II: 使用Anaconda）步骤也是类似的 (https://www.bilibili.com/video/BV1rc411W7Dr)：
conda create -n gptac_venv python=3.11 # 创建anaconda环境 conda create -n gptac_venv python=3.11 # 创建anaconda环境
conda activate gptac_venv # 激活anaconda环境 conda activate gptac_venv # 激活anaconda环境
python -m pip install -r requirements.txt # 这个步骤和pip安装一样的步骤 python -m pip install -r requirements.txt # 这个步骤和pip安装一样的步骤
@ -149,26 +151,25 @@ python main.py
### 安装方法II：使用Docker ### 安装方法II：使用Docker
0. 部署项目的全部能力（这个是包含cuda和latex的大型镜像。如果您网速慢、硬盘小或没有显卡，则不推荐使用这个，建议使用方案1）（需要熟悉[Nvidia Docker](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#installing-on-ubuntu-and-debian)运行时）
[![fullcapacity](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-all-capacity.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml) [![fullcapacity](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-all-capacity.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml)
1. 仅ChatGPT（推荐大多数人选择，等价于docker-compose方案1） ``` sh
# 修改docker-compose.yml，保留方案0并删除其他方案。修改docker-compose.yml中方案0的配置，参考其中注释即可
docker-compose up
```
1. 仅ChatGPT+文心一言+spark等在线模型（推荐大多数人选择）
[![basic](https://github.com/binary-husky/gpt_academic/actions/workflows/build-without-local-llms.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-without-local-llms.yml) [![basic](https://github.com/binary-husky/gpt_academic/actions/workflows/build-without-local-llms.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-without-local-llms.yml)
[![basiclatex](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-latex.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-latex.yml) [![basiclatex](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-latex.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-latex.yml)
[![basicaudio](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml) [![basicaudio](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml)
``` sh ``` sh
git clone --depth=1 https://github.com/binary-husky/gpt_academic.git # 下载项目 # 修改docker-compose.yml，保留方案1并删除其他方案。修改docker-compose.yml中方案1的配置，参考其中注释即可
cd gpt_academic # 进入路径 docker-compose up
nano config.py # 用任意文本编辑器编辑config.py, 配置 “Proxy”， “API_KEY” 以及 “WEB_PORT” (例如50923) 等
docker build -t gpt-academic . # 安装
#（最后一步-Linux操作系统）用`--net=host`更方便快捷
docker run --rm -it --net=host gpt-academic
#（最后一步-MacOS/Windows操作系统）只能用-p选项将容器上的端口(例如50923)暴露给主机上的端口
docker run --rm -it -e WEB_PORT=50923 -p 50923:50923 gpt-academic
``` ```
P.S. 如果需要依赖Latex的插件功能，请见Wiki。另外，您也可以直接使用docker-compose获取Latex功能（修改docker-compose.yml，保留方案4并删除其他方案）。
P.S. 如果需要依赖Latex的插件功能，请见Wiki。另外，您也可以直接使用方案4或者方案0获取Latex功能。
2. ChatGPT + ChatGLM2 + MOSS + LLAMA2 + 通义千问（需要熟悉[Nvidia Docker](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#installing-on-ubuntu-and-debian)运行时） 2. ChatGPT + ChatGLM2 + MOSS + LLAMA2 + 通义千问（需要熟悉[Nvidia Docker](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#installing-on-ubuntu-and-debian)运行时）
[![chatglm](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-chatglm.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-chatglm.yml) [![chatglm](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-chatglm.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-chatglm.yml)
@ -309,6 +310,7 @@ Tip：不指定文件直接点击 `载入对话历史存档` 可以查看历史h
### II：版本: ### II：版本:
- version 3.60（todo）: 优化虚空终端，引入code interpreter和更多插件 - version 3.60（todo）: 优化虚空终端，引入code interpreter和更多插件
- version 3.53: 支持动态选择不同界面主题，提高稳定性&解决多用户冲突问题
- version 3.50: 使用自然语言调用本项目的所有函数插件（虚空终端），支持插件分类，改进UI，设计新主题 - version 3.50: 使用自然语言调用本项目的所有函数插件（虚空终端），支持插件分类，改进UI，设计新主题
- version 3.49: 支持百度千帆平台和文心一言 - version 3.49: 支持百度千帆平台和文心一言
- version 3.48: 支持阿里达摩院通义千问，上海AI-Lab书生，讯飞星火 - version 3.48: 支持阿里达摩院通义千问，上海AI-Lab书生，讯飞星火

140
crazy_functions/pdf_fns/parse_pdf.py
View File

@ -1,6 +1,14 @@
from functools import lru_cache
from toolbox import gen_time_str
from toolbox import promote_file_to_downloadzone
from toolbox import write_history_to_file, promote_file_to_downloadzone
from colorful import *
import requests import requests
import random import random
from functools import lru_cache import copy
import os
import math
class GROBID_OFFLINE_EXCEPTION(Exception): pass class GROBID_OFFLINE_EXCEPTION(Exception): pass
def get_avail_grobid_url(): def get_avail_grobid_url():
@ -28,3 +36,133 @@ def parse_pdf(pdf_path, grobid_url):
raise RuntimeError("解析PDF失败，请检查PDF是否损坏。") raise RuntimeError("解析PDF失败，请检查PDF是否损坏。")
return article_dict return article_dict
def produce_report_markdown(gpt_response_collection, meta, paper_meta_info, chatbot, fp, generated_conclusion_files):
# -=-=-=-=-=-=-=-= 写出第1个文件：翻译前后混合 -=-=-=-=-=-=-=-=
res_path = write_history_to_file(meta + ["# Meta Translation" , paper_meta_info] + gpt_response_collection, file_basename=f"{gen_time_str()}translated_and_original.md", file_fullname=None)
promote_file_to_downloadzone(res_path, rename_file=os.path.basename(res_path)+'.md', chatbot=chatbot)
generated_conclusion_files.append(res_path)
# -=-=-=-=-=-=-=-= 写出第2个文件：仅翻译后的文本 -=-=-=-=-=-=-=-=
translated_res_array = []
# 记录当前的大章节标题：
last_section_name = ""
for index, value in enumerate(gpt_response_collection):
# 先挑选偶数序列号：
if index % 2 != 0:
# 先提取当前英文标题：
cur_section_name = gpt_response_collection[index-1].split('\n')[0].split(" Part")[0]
# 如果index是1的话，则直接使用first section name：
if cur_section_name != last_section_name:
cur_value = cur_section_name + '\n'
last_section_name = copy.deepcopy(cur_section_name)
else:
cur_value = ""
# 再做一个小修改：重新修改当前part的标题，默认用英文的
cur_value += value
translated_res_array.append(cur_value)
res_path = write_history_to_file(meta + ["# Meta Translation" , paper_meta_info] + translated_res_array,
file_basename = f"{gen_time_str()}-translated_only.md",
file_fullname = None,
auto_caption = False)
promote_file_to_downloadzone(res_path, rename_file=os.path.basename(res_path)+'.md', chatbot=chatbot)
generated_conclusion_files.append(res_path)
return res_path
def translate_pdf(article_dict, llm_kwargs, chatbot, fp, generated_conclusion_files, TOKEN_LIMIT_PER_FRAGMENT, DST_LANG):
from crazy_functions.crazy_utils import construct_html
from crazy_functions.crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
from crazy_functions.crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
from crazy_functions.crazy_utils import request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
prompt = "以下是一篇学术论文的基本信息:\n"
# title
title = article_dict.get('title', '无法获取 title'); prompt += f'title:{title}\n\n'
# authors
authors = article_dict.get('authors', '无法获取 authors'); prompt += f'authors:{authors}\n\n'
# abstract
abstract = article_dict.get('abstract', '无法获取 abstract'); prompt += f'abstract:{abstract}\n\n'
# command
prompt += f"请将题目和摘要翻译为{DST_LANG}。"
meta = [f'# Title:\n\n', title, f'# Abstract:\n\n', abstract ]
# 单线，获取文章meta信息
paper_meta_info = yield from request_gpt_model_in_new_thread_with_ui_alive(
inputs=prompt,
inputs_show_user=prompt,
llm_kwargs=llm_kwargs,
chatbot=chatbot, history=[],
sys_prompt="You are an academic paper reader。",
)
# 多线，翻译
inputs_array = []
inputs_show_user_array = []
# get_token_num
from request_llm.bridge_all import model_info
enc = model_info[llm_kwargs['llm_model']]['tokenizer']
def get_token_num(txt): return len(enc.encode(txt, disallowed_special=()))
def break_down(txt):
raw_token_num = get_token_num(txt)
if raw_token_num <= TOKEN_LIMIT_PER_FRAGMENT:
return [txt]
else:
# raw_token_num > TOKEN_LIMIT_PER_FRAGMENT
# find a smooth token limit to achieve even seperation
count = int(math.ceil(raw_token_num / TOKEN_LIMIT_PER_FRAGMENT))
token_limit_smooth = raw_token_num // count + count
return breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn=get_token_num, limit=token_limit_smooth)
for section in article_dict.get('sections'):
if len(section['text']) == 0: continue
section_frags = break_down(section['text'])
for i, fragment in enumerate(section_frags):
heading = section['heading']
if len(section_frags) > 1: heading += f' Part-{i+1}'
inputs_array.append(
f"你需要翻译{heading}章节，内容如下: \n\n{fragment}"
)
inputs_show_user_array.append(
f"# {heading}\n\n{fragment}"
)
gpt_response_collection = yield from request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
inputs_array=inputs_array,
inputs_show_user_array=inputs_show_user_array,
llm_kwargs=llm_kwargs,
chatbot=chatbot,
history_array=[meta for _ in inputs_array],
sys_prompt_array=[
"请你作为一个学术翻译，负责把学术论文准确翻译成中文。注意文章中的每一句话都要翻译。" for _ in inputs_array],
)
# -=-=-=-=-=-=-=-= 写出Markdown文件 -=-=-=-=-=-=-=-=
produce_report_markdown(gpt_response_collection, meta, paper_meta_info, chatbot, fp, generated_conclusion_files)
# -=-=-=-=-=-=-=-= 写出HTML文件 -=-=-=-=-=-=-=-=
ch = construct_html()
orig = ""
trans = ""
gpt_response_collection_html = copy.deepcopy(gpt_response_collection)
for i,k in enumerate(gpt_response_collection_html):
if i%2==0:
gpt_response_collection_html[i] = inputs_show_user_array[i//2]
else:
# 先提取当前英文标题：
cur_section_name = gpt_response_collection[i-1].split('\n')[0].split(" Part")[0]
cur_value = cur_section_name + "\n" + gpt_response_collection_html[i]
gpt_response_collection_html[i] = cur_value
final = ["", "", "一、论文概况", "", "Abstract", paper_meta_info, "二、论文翻译", ""]
final.extend(gpt_response_collection_html)
for i, k in enumerate(final):
if i%2==0:
orig = k
if i%2==1:
trans = k
ch.add_row(a=orig, b=trans)
create_report_file_name = f"{os.path.basename(fp)}.trans.html"
html_file = ch.save_file(create_report_file_name)
generated_conclusion_files.append(html_file)
promote_file_to_downloadzone(html_file, rename_file=os.path.basename(html_file), chatbot=chatbot)

100
crazy_functions/批量翻译PDF文档_NOUGAT.py
View File

@ -1,11 +1,12 @@
from toolbox import CatchException, report_execption, gen_time_str from toolbox import CatchException, report_execption, get_log_folder, gen_time_str
from toolbox import update_ui, promote_file_to_downloadzone, update_ui_lastest_msg, disable_auto_promotion from toolbox import update_ui, promote_file_to_downloadzone, update_ui_lastest_msg, disable_auto_promotion
from toolbox import write_history_to_file, get_log_folder from toolbox import write_history_to_file, promote_file_to_downloadzone
from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive 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 request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
from .crazy_utils import read_and_clean_pdf_text from .crazy_utils import read_and_clean_pdf_text
from .pdf_fns.parse_pdf import parse_pdf, get_avail_grobid_url from .pdf_fns.parse_pdf import parse_pdf, get_avail_grobid_url, translate_pdf
from colorful import * from colorful import *
import copy
import os import os
import math import math
import logging import logging
@ -92,7 +93,7 @@ def 批量翻译PDF文档(txt, llm_kwargs, plugin_kwargs, chatbot, history, syst
def 解析PDF_基于NOUGAT(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt): def 解析PDF_基于NOUGAT(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt):
import copy import copy
import tiktoken import tiktoken
TOKEN_LIMIT_PER_FRAGMENT = 1280 TOKEN_LIMIT_PER_FRAGMENT = 1024
generated_conclusion_files = [] generated_conclusion_files = []
generated_html_files = [] generated_html_files = []
DST_LANG = "中文" DST_LANG = "中文"
@ -106,96 +107,7 @@ def 解析PDF_基于NOUGAT(file_manifest, project_folder, llm_kwargs, plugin_kwa
article_content = f.readlines() article_content = f.readlines()
article_dict = markdown_to_dict(article_content) article_dict = markdown_to_dict(article_content)
logging.info(article_dict) logging.info(article_dict)
yield from translate_pdf(article_dict, llm_kwargs, chatbot, fp, generated_conclusion_files, TOKEN_LIMIT_PER_FRAGMENT, DST_LANG)
prompt = "以下是一篇学术论文的基本信息:\n"
# title
title = article_dict.get('title', '无法获取 title'); prompt += f'title:{title}\n\n'
# authors
authors = article_dict.get('authors', '无法获取 authors'); prompt += f'authors:{authors}\n\n'
# abstract
abstract = article_dict.get('abstract', '无法获取 abstract'); prompt += f'abstract:{abstract}\n\n'
# command
prompt += f"请将题目和摘要翻译为{DST_LANG}。"
meta = [f'# Title:\n\n', title, f'# Abstract:\n\n', abstract ]
# 单线，获取文章meta信息
paper_meta_info = yield from request_gpt_model_in_new_thread_with_ui_alive(
inputs=prompt,
inputs_show_user=prompt,
llm_kwargs=llm_kwargs,
chatbot=chatbot, history=[],
sys_prompt="You are an academic paper reader。",
)
# 多线，翻译
inputs_array = []
inputs_show_user_array = []
# get_token_num
from request_llm.bridge_all import model_info
enc = model_info[llm_kwargs['llm_model']]['tokenizer']
def get_token_num(txt): return len(enc.encode(txt, disallowed_special=()))
from .crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
def break_down(txt):
raw_token_num = get_token_num(txt)
if raw_token_num <= TOKEN_LIMIT_PER_FRAGMENT:
return [txt]
else:
# raw_token_num > TOKEN_LIMIT_PER_FRAGMENT
# find a smooth token limit to achieve even seperation
count = int(math.ceil(raw_token_num / TOKEN_LIMIT_PER_FRAGMENT))
token_limit_smooth = raw_token_num // count + count
return breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn=get_token_num, limit=token_limit_smooth)
for section in article_dict.get('sections'):
if len(section['text']) == 0: continue
section_frags = break_down(section['text'])
for i, fragment in enumerate(section_frags):
heading = section['heading']
if len(section_frags) > 1: heading += f' Part-{i+1}'
inputs_array.append(
f"你需要翻译{heading}章节，内容如下: \n\n{fragment}"
)
inputs_show_user_array.append(
f"# {heading}\n\n{fragment}"
)
gpt_response_collection = yield from request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
inputs_array=inputs_array,
inputs_show_user_array=inputs_show_user_array,
llm_kwargs=llm_kwargs,
chatbot=chatbot,
history_array=[meta for _ in inputs_array],
sys_prompt_array=[
"请你作为一个学术翻译，负责把学术论文准确翻译成中文。注意文章中的每一句话都要翻译。" for _ in inputs_array],
)
res_path = write_history_to_file(meta + ["# Meta Translation" , paper_meta_info] + gpt_response_collection, file_basename=None, file_fullname=None)
promote_file_to_downloadzone(res_path, rename_file=os.path.basename(fp)+'.md', chatbot=chatbot)
generated_conclusion_files.append(res_path)
ch = construct_html()
orig = ""
trans = ""
gpt_response_collection_html = copy.deepcopy(gpt_response_collection)
for i,k in enumerate(gpt_response_collection_html):
if i%2==0:
gpt_response_collection_html[i] = inputs_show_user_array[i//2]
else:
gpt_response_collection_html[i] = gpt_response_collection_html[i]
final = ["", "", "一、论文概况", "", "Abstract", paper_meta_info, "二、论文翻译", ""]
final.extend(gpt_response_collection_html)
for i, k in enumerate(final):
if i%2==0:
orig = k
if i%2==1:
trans = k
ch.add_row(a=orig, b=trans)
create_report_file_name = f"{os.path.basename(fp)}.trans.html"
html_file = ch.save_file(create_report_file_name)
generated_html_files.append(html_file)
promote_file_to_downloadzone(html_file, rename_file=os.path.basename(html_file), chatbot=chatbot)
chatbot.append(("给出输出文件清单", str(generated_conclusion_files + generated_html_files))) 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 yield from update_ui(chatbot=chatbot, history=history) # 刷新界面

111
crazy_functions/批量翻译PDF文档_多线程.py
View File

@ -1,12 +1,12 @@
from toolbox import CatchException, report_execption, get_log_folder from toolbox import CatchException, report_execption, get_log_folder, gen_time_str
from toolbox import update_ui, promote_file_to_downloadzone, update_ui_lastest_msg, disable_auto_promotion from toolbox import update_ui, promote_file_to_downloadzone, update_ui_lastest_msg, disable_auto_promotion
from toolbox import write_history_to_file, promote_file_to_downloadzone from toolbox import write_history_to_file, promote_file_to_downloadzone
from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive 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 request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
from .crazy_utils import read_and_clean_pdf_text from .crazy_utils import read_and_clean_pdf_text
from .pdf_fns.parse_pdf import parse_pdf, get_avail_grobid_url from .pdf_fns.parse_pdf import parse_pdf, get_avail_grobid_url, translate_pdf
from colorful import * from colorful import *
import glob import copy
import os import os
import math import math
@ -58,8 +58,8 @@ def 批量翻译PDF文档(txt, llm_kwargs, plugin_kwargs, chatbot, history, syst
def 解析PDF_基于GROBID(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, grobid_url): def 解析PDF_基于GROBID(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, grobid_url):
import copy import copy, json
TOKEN_LIMIT_PER_FRAGMENT = 1280 TOKEN_LIMIT_PER_FRAGMENT = 1024
generated_conclusion_files = [] generated_conclusion_files = []
generated_html_files = [] generated_html_files = []
DST_LANG = "中文" DST_LANG = "中文"
@ -67,104 +67,23 @@ def 解析PDF_基于GROBID(file_manifest, project_folder, llm_kwargs, plugin_kwa
for index, fp in enumerate(file_manifest): for index, fp in enumerate(file_manifest):
chatbot.append(["当前进度：", f"正在连接GROBID服务，请稍候: {grobid_url}\n如果等待时间过长，请修改config中的GROBID_URL，可修改成本地GROBID服务。"]); yield from update_ui(chatbot=chatbot, history=history) # 刷新界面 chatbot.append(["当前进度：", f"正在连接GROBID服务，请稍候: {grobid_url}\n如果等待时间过长，请修改config中的GROBID_URL，可修改成本地GROBID服务。"]); yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
article_dict = parse_pdf(fp, grobid_url) article_dict = parse_pdf(fp, grobid_url)
grobid_json_res = os.path.join(get_log_folder(), gen_time_str() + "grobid.json")
with open(grobid_json_res, 'w+', encoding='utf8') as f:
f.write(json.dumps(article_dict, indent=4, ensure_ascii=False))
promote_file_to_downloadzone(grobid_json_res, chatbot=chatbot)
if article_dict is None: raise RuntimeError("解析PDF失败，请检查PDF是否损坏。") if article_dict is None: raise RuntimeError("解析PDF失败，请检查PDF是否损坏。")
prompt = "以下是一篇学术论文的基本信息:\n" yield from translate_pdf(article_dict, llm_kwargs, chatbot, fp, generated_conclusion_files, TOKEN_LIMIT_PER_FRAGMENT, DST_LANG)
# title
title = article_dict.get('title', '无法获取 title'); prompt += f'title:{title}\n\n'
# authors
authors = article_dict.get('authors', '无法获取 authors'); prompt += f'authors:{authors}\n\n'
# abstract
abstract = article_dict.get('abstract', '无法获取 abstract'); prompt += f'abstract:{abstract}\n\n'
# command
prompt += f"请将题目和摘要翻译为{DST_LANG}。"
meta = [f'# Title:\n\n', title, f'# Abstract:\n\n', abstract ]
# 单线，获取文章meta信息
paper_meta_info = yield from request_gpt_model_in_new_thread_with_ui_alive(
inputs=prompt,
inputs_show_user=prompt,
llm_kwargs=llm_kwargs,
chatbot=chatbot, history=[],
sys_prompt="You are an academic paper reader。",
)
# 多线，翻译
inputs_array = []
inputs_show_user_array = []
# get_token_num
from request_llm.bridge_all import model_info
enc = model_info[llm_kwargs['llm_model']]['tokenizer']
def get_token_num(txt): return len(enc.encode(txt, disallowed_special=()))
from .crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
def break_down(txt):
raw_token_num = get_token_num(txt)
if raw_token_num <= TOKEN_LIMIT_PER_FRAGMENT:
return [txt]
else:
# raw_token_num > TOKEN_LIMIT_PER_FRAGMENT
# find a smooth token limit to achieve even seperation
count = int(math.ceil(raw_token_num / TOKEN_LIMIT_PER_FRAGMENT))
token_limit_smooth = raw_token_num // count + count
return breakdown_txt_to_satisfy_token_limit_for_pdf(txt, get_token_fn=get_token_num, limit=token_limit_smooth)
for section in article_dict.get('sections'):
if len(section['text']) == 0: continue
section_frags = break_down(section['text'])
for i, fragment in enumerate(section_frags):
heading = section['heading']
if len(section_frags) > 1: heading += f' Part-{i+1}'
inputs_array.append(
f"你需要翻译{heading}章节，内容如下: \n\n{fragment}"
)
inputs_show_user_array.append(
f"# {heading}\n\n{fragment}"
)
gpt_response_collection = yield from request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
inputs_array=inputs_array,
inputs_show_user_array=inputs_show_user_array,
llm_kwargs=llm_kwargs,
chatbot=chatbot,
history_array=[meta for _ in inputs_array],
sys_prompt_array=[
"请你作为一个学术翻译，负责把学术论文准确翻译成中文。注意文章中的每一句话都要翻译。" for _ in inputs_array],
)
res_path = write_history_to_file(meta + ["# Meta Translation" , paper_meta_info] + gpt_response_collection, file_basename=None, file_fullname=None)
promote_file_to_downloadzone(res_path, rename_file=os.path.basename(fp)+'.md', chatbot=chatbot)
generated_conclusion_files.append(res_path)
ch = construct_html()
orig = ""
trans = ""
gpt_response_collection_html = copy.deepcopy(gpt_response_collection)
for i,k in enumerate(gpt_response_collection_html):
if i%2==0:
gpt_response_collection_html[i] = inputs_show_user_array[i//2]
else:
gpt_response_collection_html[i] = gpt_response_collection_html[i]
final = ["", "", "一、论文概况", "", "Abstract", paper_meta_info, "二、论文翻译", ""]
final.extend(gpt_response_collection_html)
for i, k in enumerate(final):
if i%2==0:
orig = k
if i%2==1:
trans = k
ch.add_row(a=orig, b=trans)
create_report_file_name = f"{os.path.basename(fp)}.trans.html"
html_file = ch.save_file(create_report_file_name)
generated_html_files.append(html_file)
promote_file_to_downloadzone(html_file, rename_file=os.path.basename(html_file), chatbot=chatbot)
chatbot.append(("给出输出文件清单", str(generated_conclusion_files + generated_html_files))) 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 yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
def 解析PDF(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt): def 解析PDF(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt):
"""
此函数已经弃用
"""
import copy import copy
TOKEN_LIMIT_PER_FRAGMENT = 1280 TOKEN_LIMIT_PER_FRAGMENT = 1024
generated_conclusion_files = [] generated_conclusion_files = []
generated_html_files = [] generated_html_files = []
from crazy_functions.crazy_utils import construct_html from crazy_functions.crazy_utils import construct_html

49
docker-compose.yml
View File

@ -1,5 +1,54 @@
#【请修改完参数后，删除此行】请在以下方案中选择一种，然后删除其他的方案，最后docker-compose up运行 | Please choose from one of these options below, delete other options as well as This Line #【请修改完参数后，删除此行】请在以下方案中选择一种，然后删除其他的方案，最后docker-compose up运行 | Please choose from one of these options below, delete other options as well as This Line
## ===================================================
## 【方案零】 部署项目的全部能力（这个是包含cuda和latex的大型镜像。如果您网速慢、硬盘小或没有显卡，则不推荐使用这个）
## ===================================================
version: '3'
services:
gpt_academic_full_capability:
image: ghcr.io/binary-husky/gpt_academic_with_all_capacity:master
environment:
# 请查阅 `config.py`或者 github wiki 以查看所有的配置信息
API_KEY: ' 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 '
# USE_PROXY: ' True '
# proxies: ' { "http": "http://localhost:10881", "https": "http://localhost:10881", } '
LLM_MODEL: ' gpt-3.5-turbo '
AVAIL_LLM_MODELS: ' ["gpt-3.5-turbo", "gpt-4", "qianfan", "sparkv2", "spark", "chatglm"] '
BAIDU_CLOUD_API_KEY : ' bTUtwEAveBrQipEowUvDwYWq '
BAIDU_CLOUD_SECRET_KEY : ' jqXtLvXiVw6UNdjliATTS61rllG8Iuni '
XFYUN_APPID: ' 53a8d816 '
XFYUN_API_SECRET: ' MjMxNDQ4NDE4MzM0OSNlNjQ2NTlhMTkx '
XFYUN_API_KEY: ' 95ccdec285364869d17b33e75ee96447 '
ENABLE_AUDIO: ' False '
DEFAULT_WORKER_NUM: ' 20 '
WEB_PORT: ' 12345 '
ADD_WAIFU: ' False '
ALIYUN_APPKEY: ' RxPlZrM88DnAFkZK '
THEME: ' Chuanhu-Small-and-Beautiful '
ALIYUN_ACCESSKEY: ' LTAI5t6BrFUzxRXVGUWnekh1 '
ALIYUN_SECRET: ' eHmI20SVWIwQZxCiTD2bGQVspP9i68 '
# LOCAL_MODEL_DEVICE: ' cuda '
# 加载英伟达显卡运行时
# runtime: nvidia
# deploy:
# resources:
# reservations:
# devices:
# - driver: nvidia
# count: 1
# capabilities: [gpu]
# 与宿主的网络融合
network_mode: "host"
# 不使用代理网络拉取最新代码
command: >
bash -c "python3 -u main.py"
## =================================================== ## ===================================================
## 【方案一】 如果不需要运行本地模型（仅 chatgpt, azure, 星火, 千帆, claude 等在线大模型服务） ## 【方案一】 如果不需要运行本地模型（仅 chatgpt, azure, 星火, 千帆, claude 等在线大模型服务）
## =================================================== ## ===================================================

5
themes/common.css
View File

@ -19,3 +19,8 @@
.wrap.svelte-xwlu1w { .wrap.svelte-xwlu1w {
min-height: var(--size-32); min-height: var(--size-32);
} }
/* status bar height */
.min.svelte-1yrv54 {
min-height: var(--size-12);
}

4
toolbox.py
View File

@ -216,7 +216,7 @@ def get_reduce_token_percent(text):
return 0.5, '不详' return 0.5, '不详'
def write_history_to_file(history, file_basename=None, file_fullname=None): def write_history_to_file(history, file_basename=None, file_fullname=None, auto_caption=True):
""" """
将对话记录history以Markdown格式写入文件中。如果没有指定文件名，则使用当前时间生成文件名。 将对话记录history以Markdown格式写入文件中。如果没有指定文件名，则使用当前时间生成文件名。
""" """
@ -235,7 +235,7 @@ def write_history_to_file(history, file_basename=None, file_fullname=None):
if type(content) != str: content = str(content) if type(content) != str: content = str(content)
except: except:
continue continue
if i % 2 == 0: if i % 2 == 0 and auto_caption:
f.write('## ') f.write('## ')
try: try:
f.write(content) f.write(content)

4
version
View File

@ -1,5 +1,5 @@
{ {
"version": 3.52, "version": 3.53,
"show_feature": true, "show_feature": true,
"new_feature": "提高稳定性&解决多用户冲突问题 <-> 支持插件分类和更多UI皮肤外观 <-> 支持用户使用自然语言调度各个插件(虚空终端) ！ <-> 改进UI，设计新主题 <-> 支持借助GROBID实现PDF高精度翻译 <-> 接入百度千帆平台和文心一言 <-> 接入阿里通义千问、讯飞星火、上海AI-Lab书生 <-> 优化一键升级 <-> 提高arxiv翻译速度和成功率" "new_feature": "支持动态选择不同界面主题 <-> 提高稳定性&解决多用户冲突问题 <-> 支持插件分类和更多UI皮肤外观 <-> 支持用户使用自然语言调度各个插件(虚空终端) ！ <-> 改进UI，设计新主题 <-> 支持借助GROBID实现PDF高精度翻译 <-> 接入百度千帆平台和文心一言 <-> 接入阿里通义千问、讯飞星火、上海AI-Lab书生 <-> 优化一键升级 <-> 提高arxiv翻译速度和成功率"
} }