https://github.com/opendatalab/magic-html
https://github.com/opendatalab/magic-html
Last synced: 8 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/opendatalab/magic-html
- Owner: opendatalab
- License: apache-2.0
- Created: 2024-04-25T07:59:53.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-03-13T06:27:20.000Z (10 months ago)
- Last Synced: 2025-04-19T05:07:10.998Z (9 months ago)
- Language: Python
- Size: 14.6 MB
- Stars: 439
- Watchers: 7
- Forks: 38
- Open Issues: 8
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# magic-html - 通用HTML数据提取器
欢迎使用magic-html，这是一个旨在简化从HTML中提取主体区域内容的Python库。
## 项目描述
magic-html提供了一套工具，能够轻松地从HTML中提取主体区域内容。无论您处理的是复杂的HTML结构还是简单的网页，这个库都旨在为您的HTML抽取需求提供一个便捷高效的接口。
## 特点
- 返回主体区域html结构，可自定义输出纯文本/markdown
- 支持多模态抽取
- 支持多种版面extractor，文章/论坛
- 支持latex公式提取转换
## 安装
```shell
pip install https://github.com/opendatalab/magic-html/releases/download/magic_html-0.1.2-released/magic_html-0.1.2-py3-none-any.whl
```
## 使用
```python
from magic_html import GeneralExtractor
# 初始化提取器
extractor = GeneralExtractor()
url = "http://example.com/"
html = """
Example Domain
Example Domain
This domain is for use in illustrative examples in documents. You may use this
domain in literature without prior coordination or asking for permission.
"""
# 文章类型HTML提取数据
data = extractor.extract(html, base_url=url)
# 论坛类型HTML提取数据
# data = extractor.extract(html, base_url=url, html_type="forum")
# 微信文章HTML提取数据
# data = extractor.extract(html, base_url=url, html_type="weixin")
print(data)
```
## benchmark report
根据html页面类型，文章/论坛，对比不同开源通用抽取框架抽取准确性
文章类型：选取头部新闻、博客站点共标注158个html页面
```Python
╒══════════════════════╤═════════════╤════════════╤═══════════╕
│ func │ prec_mean │ rec_mean │ f1_mean │
╞══════════════════════╪═════════════╪════════════╪═══════════╡
│ magic_html │ 0.908865 │ 0.95032 │ 0.92913 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ trafilatura │ 0.833434 │ 0.912384 │ 0.871124 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ trafilatura_fallback │ 0.831229 │ 0.933713 │ 0.879496 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ readability-lxml │ 0.86587 │ 0.861391 │ 0.863625 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ newspaper3k │ 0.409585 │ 0.372083 │ 0.389935 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ goose3 │ 0.525717 │ 0.457669 │ 0.489339 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ justext │ 0.224945 │ 0.117092 │ 0.154014 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ gne │ 0.828849 │ 0.629112 │ 0.715299 │
╘══════════════════════╧═════════════╧════════════╧═══════════╛
```
论坛类型：选取头部论坛、问答站点与开源建站框架搭建站点共103个html页面
```Python
╒══════════════════════╤═════════════╤════════════╤═══════════╕
│ func │ prec_mean │ rec_mean │ f1_mean │
╞══════════════════════╪═════════════╪════════════╪═══════════╡
│ magic_html │ 0.796252 │ 0.826819 │ 0.811248 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ trafilatura │ 0.716009 │ 0.695947 │ 0.705835 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ trafilatura_fallback │ 0.730304 │ 0.691328 │ 0.710282 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ readability-lxml │ 0.788018 │ 0.445087 │ 0.568867 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ newspaper3k │ 0.596976 │ 0.298322 │ 0.397837 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ goose3 │ 0.675835 │ 0.312969 │ 0.427821 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ justext │ 0.175889 │ 0.0517628 │ 0.0799863 │
├──────────────────────┼─────────────┼────────────┼───────────┤
│ gne │ 0.81003 │ 0.389709 │ 0.526241 │
╘══════════════════════╧═════════════╧════════════╧═══════════╛
```
## 许可
本项目代码采用[Apache 2.0 license](https://www.apache.org/licenses/LICENSE-2.0.html)授权。
## 鸣谢
- [trafilatura](https://github.com/adbar/trafilatura)
- [readability-lxml](https://github.com/buriy/python-readability)