Python 3.x tool converting Confluence html to Markdown. Primarily, it uses Github flavored Markdown, but also adds some stuff that is used on stackoverflow. Focusses on code-blocks, tables, lists, images. Might miss some other advanced html tags. See details below. Designed for Confluence 5.8.2.
You need
- python3
- bs4 - BeautifulSoup4
Python3 should be available on most Linux distros. If on Mac, install via brew or download.
bs4 is a HtmlParser. Can be installed via package manager on several Linux distros (e.g. apt-get install python3-bs4). Otherwise install via pip (e.g. pip3 install beautifulsoup4)
This project might not support ALL tags. It worked for me for the most part to convert my confluence wiki to github markdown. So clone it, use it as-is, see the results and maybe make modifications for you own needs.
- Open Confluence, go to space tools (need to be space admin), export, select html format, download the html zip-file, extract it somewhere
- c2m.py is the only file you need. Either clone the repo or download the single file. No installation required. chmod +x c2m.py
- Run
python3 c2m.py [sourcefolder] [destinationfolder]
Sample:
python3 c2m.py /path/to/extracted/confluence/html/ /tmp/my-markdown
Confluence adds a lot of garbage to any html-page. Not only confluence-specific-css, but also "generated by Confluence"-notes, link-hierarchy, "last-modified"-notes, etc. All this data is ignored, i.e. will not be contained in generated .md files.
Code blocks are rendered Gibhub-flavored, using three backticks. stackoverflow uses 4-space-indention instead, but code on SO can be easily indented (CTRL-K). SO also uses optional HTML-comments to specify language. Github allows to specify language appended to backticks.
```python
some code
```
Tables are rendered as-is, i.e. html is kept. Even though tables are often written with pipes in markup, pipe-tables have limitations.