中文 | English
ChatPaper2XMind is a tool for generating concise XMind notes from PDF papers using ChatGPT, including images and formulas, to improve reading efficiency.
Note: Due to the limitations of the ChatGPT generation model's accuracy, the generated XMind notes are more suitable as note drafts, which can be used as a basis for creating reading notes, rather than directly using them for paper reading.
Contents
For the meaning of configuration options, please refer to the source code in config.py.
Drag and drop a PDF file or a folder containing PDFs into the input box to generate with a single click.
git clone --recursive https://github.com/MasterYip/ChatPaper2Xmind.git
cd <work-dir>
pip install -r requirements.txt
pip install -r ./XmindCopilot/requirements.txt
OpenAI API Settings
"""OpenAI API"""
APIBASE = "" # OpenAI API base, default is "https://api.openai.com/v1" for now (Leave it as empty if you are not sure)
APIKEYS = [""] # Your OpenAI API keys
MODEL = "gpt-3.5-turbo" # GPT model name
LANGUAGE = "English" # Only partially support Chinese
KEYWORD = "Science&Engineering" # Keyword for GPT model (What field you want the model to focus on)
PROXY = None # Your proxy address
# Note: If you are in China, you may need to use a proxy to access the OpenAI API
# (If your system's global proxy is set, you can leave it as None)
# PROXY = "http://127.0.0.1:7890"
- APIBASE: URL of the OpenAI model request server
- You can replace it with any model that supports OpenAI request format (ChatGLM/LLaMA, etc.)
- APIKEYS (Must be configured): API keys for OpenAI model requests
- Multiple API keys can be added to support multi-threaded requests
- If using a different model, the length of the APIKEYS list determines the number of request threads; the content can be any value
- For those without API keys, you can refer to ChatGPT_API_NoKey to configure a fake API server and change openai.api_base to achieve fake API access; in this case, APIKEY should be set to any value (cannot be empty)
- MODEL: Select OpenAI model
- LANGUAGE: Generated language
- KEYWORD: Field to which the paper belongs
- PROXY: Proxy server
- If you are in the China region, you may need to use a proxy to access the OpenAI website
- When left as None, it will follow the system's global proxy
Generation Format Settings
"""Generation"""
GEN_IMGS = True
GEN_EQUATIONS = True
# PDFFigure2
USE_PDFFIGURE2 = True # Use PDFFigure2 to generate images & tables (This requires you to install JVM)
SNAP_WITH_CAPTION = True # Generate images & tables with captions (Only valid when USE_PDFFIGURE2 is True)
# Max generation item number
TEXT2LIST_MAX_NUM = 4 # Max number of items for each list
TEXT2TREE_MAX_NUM = 4 # Max number of subtopics for each topic
FAKE_GPT_RESPONSE = "Fake" # Fake GPT response when GPT_ENABLE is False
if True: # Use the true GPT model
GPT_ENABLE = True
THREAD_RATE_LIMIT = 100 # Each APIKEY can send 3 requests per minute (limited by OpenAI)
else: # Use the fake GPT model
GPT_ENABLE = False
THREAD_RATE_LIMIT = 6000
- GEN_IMGS: Capture and generate paper images
- GEN_EQUATIONS: Capture and generate paper equations
- USE_PDFFIGURE2: Use PDFFIGURE2 to capture paper images (Requires Java environment; set to False if Java environment is not installed)
- SNAP_WITH_CAPTION: When USE_PDFFIGURE2 is True, capture images and tables along with captions
- TEXT2LIST_MAX_NUM (Currently ineffective)
- TEXT2TREE_MAX_NUM (Currently ineffective)
- FAKE_GPT_RESPONSE: Pseudo GPT response when not using ChatGPT
- GPT_ENABLE: Whether to use GPT
- If not using ChatGPT to generate text summaries and only generating the table of contents and images/equations, set it to False
- THREAD_RATE_LIMIT: Number of requests per minute for a single thread (single API key)
- OpenAI has frequency limits on requests; typically 3/min for standard packages
PDF Parser: Regular Expression Settings
"""PDF Parser - Regular Expression"""
# Special title
ABS_MATCHSTR = "ABSTRACT|Abstract|abstract"
INTRO_MATCHSTR = "I.[\s]{1,3}(INTRODUCTION|Introduction|introduction)"
REF_MATCHSTR = "Reference|REFERENCE|Bibliography"
APD_MATCHSTR = "APPENDIX|Appendix" # Not used for now
# General title
# FIXME: Misidentification exists
SECTION_TITLE_MATCHSTR = ["[IVX1-9]{1,4}[\.\s][\sA-Za-z]{1,}|[1-9]{1,2}[\s\.\n][\sA-Za-z]{1,}", # Level 1
"[A-M]{1}\.[\sA-Za-z]{1,}|[1-9]\.[1-9]\.[\sA-Za-z]{1,}"] # Level 2
# Equation & Image
EQUATION_MATCHSTR = '[\s]{0,}\([\d]{1,}[a-zA-Z]{0,1}\)'
IMG_MATCHSTR = 'Fig.[\s]{1,3}[\d]{1,2}|Figure[\s]{1,3}[\d]{1,2}|Tab.[\s]{1,3}[\dIVX]{1,3}|Table[\s]{1,3}[\dIVX]{1,3}' # Figure & Table
Xmind Sytle Template
"""Xmind Sytle Template"""
TEMPLATE_XMIND_PATH = 'template.xmind'
Debug Info
"""Debuging"""
DEBUG_MODE = False
Convert PDF papers to XMind
cd <root-dir>
python paper2xmind.py --path <pdf_path_or_folder_path>
Run the demo
python paper2xmind.py
FileNotFoundError: [WinError 2] The system cannot find the file specified.
Exception ignored in: <function PDFPaperParser.del at 0x000001F4388C2C00>
Traceback (most recent call last):
File "D:\git\ChatPaper2Xmind\pdf_parser.py", line 31, in __del__
self.pdf.close()
^^^^^^^^
AttributeError: 'PDFFigure2PaperParser' object has no attribute 'pdf'
- Incorrect PDF input path
- PDF file doesn't exist
- Input path contains spaces and is not enclosed in double quotes
- Java environment not installed and using PDFFIGURE2 to generate images
- Set USE_PDFFIGURE2=False or install Java environment and add it to the system PATH environment variable
Traceback (most recent call last):
File "D:\academic chatgpt series\ChatPaper2Xmind-main\paper2xmind.py", line 3, in <module>
from XmindCopilot import xmind, fileshrink
ImportError: cannot import name 'xmind' from 'XmindCopilot' (unknown location)
- Git repository XmindCopilot not cloned properly
- Need to properly execute environment setup
git clone --recursive https://github.com/MasterYip/ChatPaper2Xmind.git
cd <work-dir>
pip install -r requirements.txt
pip install -r ./XmindCopilot/requirements.txt
- Reduce GPT requests to speed up XMind generation.
- Add metadata and resource parsing capabilities.
- Add functionality for generating Markdown notes.
- Optimize formula detection (boundary detection).
Thanks to the following projects for their valuable contributions to this project:
And all other projects that might have been inadvertently overlooked :)
Special thanks to the open-source community and all contributors who have contributed to this project.
This project is released under the MIT License. For more information, see the LICENSE file.
Master Yip
Email: 2205929492@qq.com
GitHub: Master Yip
QQ Group：
