bin/recodetree

   1 #!/bin/sh
   2 # reencode file names in git tree objects from Windows system encoding to UTF-8
   3
   4 usage()
   5 {
   6         echo "WARNING this script is not intended as a full-fledged and fool-proof
   7 migration utility, but rather as a proof of concept or collection of examples.
   8 Use at your own risk.
   9
  10 In particular, the script doesn't check or fix any non-ascii names in config
  11 files (such as .git/config, .gitignore or .gitmodules). For specific migration
  12 tasks, some of the commands in this script may require tweaking.
  13
  14 Usage: recodetree <command>
  15 recodetree check    Checks git repository for non-ascii file names.
  16 recodetree preview  Same as 'check', but prints file names as if converted
  17                     from cp$(getcp) to UTF-8.
  18 recodetree head     Converts the current HEAD from cp$(getcp) to UTF-8.
  19                     Does not commit, use git status to check results first.
  20 recodetree history  Converts entire repository history from cp$(getcp) to UTF-8.
  21                     WARNING: this rewrites the history of the repository.
  22                     - Make a BACKUP copy of the repository before using this!
  23                     - Read 'git help filter-branch' for implications of
  24                       rewriting history."
  25 }
  26
  27 check()
  28 {
  29         # use fast-export to dump all file names in the history
  30         git fast-export --no-data --signed-tags=strip --tag-of-filtered-object=drop --all |
  31         # use awk to filter for non-ascii names
  32         awk --posix '/^M [0-9]{6} [0-9a-f]{40} .*[\200-\377]/{print substr($0,51);}'
  33 }
  34
  35 recode()
  36 {
  37         # convert from system encoding to UTF-8
  38         iconv -c -f cp$(getcp) -t utf-8
  39 }
  40
  41 recode_tree()
  42 {
  43         # clear the index
  44         git rm -f -r -q --cached --ignore-unmatch \*
  45         # list specified commit, reencode and add to index
  46         git ls-tree -z -r $1 | recode | git update-index -z --index-info
  47 }
  48
  49 case $1 in
  50         check)
  51                 check
  52                 ;;
  53
  54         preview)
  55                 check | recode
  56                 ;;
  57
  58         head)
  59                 recode_tree HEAD
  60                 ;;
  61
  62         history)
  63                 git filter-branch --index-filter 'recodetree filter' -- --all
  64                 ;;
  65
  66         filter)
  67                 # used internally by recodetree history
  68                 recode_tree $GIT_COMMIT
  69                 ;;
  70
  71         *)
  72                 usage
  73                 ;;
  74 esac