CODEWITHSUNDEEP
git
Patrick
Patrick

Reputation: 13974

How to determine if a file is identical to one in git's history

I have arbitrary file foo, not committed to git.

I also have a file whose entire history has been kept in git, bar.

How can I determine if foo is identical to any version of bar that has ever existed?

Upvotes: 3

Views: 490

Answers (2)

torek
torek

Reputation: 488183

It's easy to tell if the contents of file foo appear somewhere in the repo:

file=foo  # or argument to script, etc
sha1=$(git hash-object -t blob $file)
repotype=$(git cat-file -t $sha1 2>/dev/null) || {
    echo "file \"$file\" is not in the repo"
    exit 1
}
[ $repotype = blob ] || {
    echo "uh oh: file \"$file\" matches non-file ($repotype) object"
    exit 1
}

However, just because foo appears in the repo as a blob, does not mean it appears under the name bar (or maybe not even at all, it could have been git added but never checked in under a commit). So now look at every (reasonable?) commit, extracting the blob-ID for your target path, skipping the commit if it's not in there:

target_path=bar

git rev-list --branches |     # or --all, or HEAD, or (etc)
while read id; do
    file_id=$(git rev-parse -q --verify $id:$target_path) || continue
    [ $file_id = $sha1 ] || continue
    echo "found \"$file\" as \"$target_path\" in $id"
    # do more here if you like, e.g., git show $id
done

If you want to find it under any name, rather than some particular explicit name, you can git ls-tree -r each commit to find all blobs and check their IDs.

(note: untested except piecemeal, and an occasional bit may have been retyped or altered along the way, beware of typos or silly errors)

Upvotes: 4

interskh
interskh

Reputation: 2591

Using a combination of md5sum, git log and grep would work:

for SHA in `git log --pretty=format:"%h" bar`; do 
  git show $SHA:bar | md5sum
done | grep `md5sum foo| cut -d' ' -f1`

The above command git log --pretty=format:"%h" bar get the list of all commits for bar file then we do md5sum on each one of them (git show to show the file in that commit). At the end we grep the md5sum of foo file

Upvotes: 2

Related Questions