I decided to put some work into a project with a MediaWiki instance. In the past, one of their websites went off the web, so I wanted to make sure my (and others) work doesn’t face a similar fate-

Luckily, MediaWiki has an API, and can be backed up without direct database or shell access to the wiki.

WikiTeam has created a script to backup wikis including media using that API: dumpgenerator.py (via MediaWiki docs). It requires python2, and can be setup in a virtual env as follows:

virtualenv env
. env/bin/activate
pip install lxml mwclient kitchen
wget https://raw.githubusercontent.com/WikiTeam/wikiteam/master/dumpgenerator.py

I’ve created a little wrapper, which includes backup rotation:



DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

cd $DIR
./env/bin/python dumpgenerator.py --xml --images $URL >/dev/null
if [ $? -eq 0 ]; then
	ls -1dt *-wikidump | sed 1,7d | xargs -r rm -r
	echo "error while backing up $URL"

You can call this with ./backup.sh <Wiki URL>, e.g. ./backup.sh https://altpwr.net/.