A simple shell script (based on information found at https://help.github.com/articles/remove-sensitive-data) to remove files based on their extension. The idea was for me to permanently remove some binary files I had in a git repository history.

git-remove-files-from-history

#!/bin/sh

function updateFiles()
{
	export files=`git rev-list --all --objects | awk '{ print $2; }' | sort | uniq | grep -e "\.dll$" -e "\.dll\.a$" -e "\.dylib$" -e "\.lib$" -e "\.pdf$" -e "\.pptx$" -e "\.so$" -e "\.so\..$" -e "\.zip$" | sort | uniq`
}

updateFiles

needCleaningUp=0

while [ "$files" != "" ]
do
	# Note: from personal experience, the loop is needed to make sure that all
	#       files are actually removed. YMMV, but it certainly doesn't harm
	#       having it...

	needCleaningUp=1

	git filter-branch --force --index-filter 'git rm -f --cached --ignore-unmatch $files' --prune-empty --tag-name-filter cat -- --all

	rm -rf .git/refs/original
	git reflog expire --expire=now --all

	updateFiles
done

if [ $needCleaningUp -eq 1 ]
then
	git gc --prune=now
	git gc --aggressive --prune=now
fi

# Note: following the execution of this script, I force pushed everything back
#       to my git repository using
#
#           $ git push --force --all
#
#       and then my tags using
#
#           $ git push --tags