repohardlink

#!/bin/bash
#
# repohardlink
# version 0.1
# 
# Optimizes storage by hardlinking common git pack objects between two repos.
#
# This script runs "git gc" to repack the pack objects before using the hardlink
# tool to unify the storage between identical pack objects in two repos.
# Normally it only considers git repos that were changed.  However if you 
# use the optional --all parameter it will repack and unify all git objects
# between the two repos.  This is useful to conserve storage when you have
# preexisting independent repos on your local filesystem.
#
# Copyright 2012 Warren Togami <wtogami@gmail.com>
# Licensed under the GNU GPL v3+.

errorexit() {
  echo "ERROR: You must invoke this command with two parameters."
  echo
  echo "    $0 <fromrepo> <torepo> [--all]"
  echo
  echo "      * <fromrepo> must be in the current directory."
  echo "      * <torepo> must be in the current directory, but you can move it later."
  echo "      * Neither names should have a trailing slash."
  echo
  echo "      [--all] parameter runs git gc on all git repositories.  Useful if you"
  echo "              are merging storage of two repos for the first time."
  echo
  exit 255
}

# Sanity Checks
# TODO: Make this more flexible in dealing with relative or full pathnames
[ -z "$1" ] || [ -z "$2" ] && errorexit
[ "$1" != $(basename $1) ] || [ "$2" != $(basename $2) ] && errorexit

# Abort immediately if any error occurs
set -e

GITLIST1=$(find $1/ -type d -name .git)
GITLIST2=$(find $2/ -type d -name .git)
unset REPACKLIST
for gitdir in $GITLIST1 $GITLIST2; do
  # Skip if objects directory does not exist
  [ ! -d $gitdir/objects/ ] && continue
  # Skip if two directories in .git/objects/, no repacking necessary
  if ls -l $gitdir/objects/ |grep -q "^total 8"; then
    # Unless --all was specified
    [ "$3" != "--all" ] && continue
  fi
  cd $gitdir/..
  pwd
  git gc
  cd - > /dev/null
  REPACKLIST="$REPACKLIST $gitdir/objects/pack/"
done

set -x
hardlink -cvv $REPACKLIST