bin/what-made-this-repo-so-large.sh

   1 #!/bin/sh
   2
   3 # This script tries to find out what made your repository so large.
   4 # It sorts the loose and packed objects by size (for packed objects, we look
   5 # at the size in the pack) and shows the largest ten.
   6
   7 die () {
   8         echo "$*" >&2
   9         exit 1
  10 }
  11
  12 identify_object () {
  13         LINES="$(git log --root --raw --abbrev=40 --all --format="commit %H" |
  14                 grep -e '^commit' -e "^:[^ ]* [^ ]* [^ ]* $1" |
  15                 grep -m1 -B1 "^:[^ ]* [^ ]* [^ ]* $1")"
  16         if test -n "$LINES"
  17         then
  18                 COMMIT="$(git name-rev $(echo "$LINES" |
  19                                 sed -n 's/commit //p') |
  20                         sed 's/^[^ ]* //')"
  21                 PATH="$(echo "$LINES" | sed -n 's/^.*   //p')"
  22                 printf "'%s:%s'" $COMMIT "$PATH"
  23         fi
  24 }
  25
  26 N=10
  27 test $# -gt 0 && N=$1
  28
  29 GIT_DIR="$(git rev-parse --git-dir)" || die "No repository found"
  30 OBJECT_LIST="$( (cd "$GIT_DIR" &&
  31         ls -l objects/??/* |
  32         sed 's/^\([^ ]* *\)\{4\}\([^ ]*\).* objects\/\(..\)\/\(.\{38\}\)$/\2 \3\4/' &&
  33         for pack in "$(ls objects/pack/*.pack 2> /dev/null)"
  34         do
  35                 git verify-pack -v "$pack" 2>&1 |
  36                 sed -n 's/^\([^ ]\{40\}\) *[^ ]* *[^ ]* *\([1-9][0-9]*\).*$/\2 \1/p'
  37         done) |
  38         sort -n -r |
  39         uniq |
  40         head -n "$N")"
  41
  42 echo "$OBJECT_LIST" |
  43 while read size sha1
  44 do
  45         echo $size $sha1 "$(identify_object $sha1)"
  46 done