<< Back to list

Recovering a file from the git index

Ever lost a file that you knew for sure you added to git?

Well, you'll be happy to hear that git stores every file that's added to the index and we can use that to retrieve any file git has seen. The problem is they're all addressed by SHA1 hashes of their content.

The .git/objects directory contains all of the objects git has recorded. The directory contains lots of subdirectories, which are just the two first letter of the object hash. Inside each subdirectory you will find the actual blobs. You can't read them directly, but you can instruct git to print the file to stdout for you using the git cat-file command.

I wrote the following script to concatenate all of the git objects and print their contents. Pipe everything through less and search for something that's in the file you're looking for. There might be multiple versions of the file so look carefully. Once you found the file, you can either copy the content to a new file on disk or you can take the object hash that was printed above the content and give that to git cat-file -p <hash> to print only that blob.

#!/usr/bin/env bash

for dir in ./.git/objects/*; do
  hashPrefix=$(basename "$dir")

  if [ "$hashPrefix" == "pack" ] || [ "$hashPrefix" == "info" ]; then

  for file in $dir/*; do
    hashSuffix=$(basename "$file")

    echo "========================================"
    echo "object: $object"
    git cat-file -p "$object"