Compares all commits in a git repo against a directory, and outputs the commit message prefixed by the line count of the diff between the two

git-compare.sh

#!/bin/bash
#
# Usage: ./git-compare.sh PATH_TO_REPO DIR_TO_COMPARE_AGAINST
#
# Compares all commits in a git repo against a directory, and outputs the commit
# message prefixed by the line count of the diff between the two.
#
# If somobody copies the content of your repository (i.e. downloads a tarball
# instead of cloning it) and possibly does some changes, you can later use this
# to figure out on which commit it was based on.
#
# `diff -r` is used to do the comparison. ".git" is excluded from the diff.
# This probably means that changes in files like .gitignore will not count.

REPO=`realpath $1`
DIFF_TO=`realpath $2`

cd $REPO
COMMIT_HASHES=`git log --pretty=format:"%h"`
ORIGINAL_COMMIT=`git log --pretty=format:"%h" | head -1`

do_work() {
    for h in $COMMIT_HASHES ; do
        git checkout $h 2>/dev/null
        printf "%8s %s\n"\
            `diff -r -x '.git' --suppress-common-lines --new-file\
                $REPO $DIFF_TO | wc -l`\
            "`git log --oneline | head -1`"
    done
}

# Trims the output to fit into a terminal
do_work | cut -c-`tput cols`

# FIXME This returns to the right commit, but you are detached
git checkout $ORIGINAL_COMMIT 2>/dev/null