git-filter-branch.sh

   1 #!/bin/sh
   2 #
   3 # Rewrite revision history
   4 # Copyright (c) Petr Baudis, 2006
   5 # Minimal changes to "port" it to core-git (c) Johannes Schindelin, 2007
   6 #
   7 # Lets you rewrite the revision history of the current branch, creating
   8 # a new branch. You can specify a number of filters to modify the commits,
   9 # files and trees.
  10
  11 warn () {
  12         echo "$*" >&2
  13 }
  14
  15 map()
  16 {
  17         # if it was not rewritten, take the original
  18         if test -r "$workdir/../map/$1"
  19         then
  20                 cat "$workdir/../map/$1"
  21         else
  22                 echo "$1"
  23         fi
  24 }
  25
  26 # if you run 'skip_commit "$@"' in a commit filter, it will print
  27 # the (mapped) parents, effectively skipping the commit.
  28
  29 skip_commit()
  30 {
  31         shift;
  32         while [ -n "$1" ];
  33         do
  34                 shift;
  35                 map "$1";
  36                 shift;
  37         done;
  38 }
  39
  40 # override die(): this version puts in an extra line break, so that
  41 # the progress is still visible
  42
  43 die()
  44 {
  45         echo >&2
  46         echo "$*" >&2
  47         exit 1
  48 }
  49
  50 # When piped a commit, output a script to set the ident of either
  51 # "author" or "committer
  52
  53 set_ident () {
  54         lid="$(echo "$1" | tr "A-Z" "a-z")"
  55         uid="$(echo "$1" | tr "a-z" "A-Z")"
  56         pick_id_script='
  57                 /^'$lid' /{
  58                         s/'\''/'\''\\'\'\''/g
  59                         h
  60                         s/^'$lid' \([^<]*\) <[^>]*> .*$/\1/
  61                         s/'\''/'\''\'\'\''/g
  62                         s/.*/export GIT_'$uid'_NAME='\''&'\''/p
  63
  64                         g
  65                         s/^'$lid' [^<]* <\([^>]*\)> .*$/\1/
  66                         s/'\''/'\''\'\'\''/g
  67                         s/.*/export GIT_'$uid'_EMAIL='\''&'\''/p
  68
  69                         g
  70                         s/^'$lid' [^<]* <[^>]*> \(.*\)$/\1/
  71                         s/'\''/'\''\'\'\''/g
  72                         s/.*/export GIT_'$uid'_DATE='\''&'\''/p
  73
  74                         q
  75                 }
  76         '
  77
  78         LANG=C LC_ALL=C sed -ne "$pick_id_script"
  79         # Ensure non-empty id name.
  80         echo "[ -n \"\$GIT_${uid}_NAME\" ] || export GIT_${uid}_NAME=\"\${GIT_${uid}_EMAIL%%@*}\""
  81 }
  82
  83 # This script can be sourced by the commit filter to get the functions
  84 test "a$SOURCE_FUNCTIONS" = a1 && return
  85 this_script="$(cd "$(dirname "$0")"; pwd)"/$(basename "$0")
  86 export this_script
  87
  88 USAGE="[--env-filter <command>] [--tree-filter <command>] \
  89 [--index-filter <command>] [--parent-filter <command>] \
  90 [--msg-filter <command>] [--commit-filter <command>] \
  91 [--tag-name-filter <command>] [--subdirectory-filter <directory>] \
  92 [--original <namespace>] [-d <directory>] [-f | --force] \
  93 [<rev-list options>...]"
  94
  95 OPTIONS_SPEC=
  96 . git-sh-setup
  97
  98 git diff-files --quiet &&
  99         git diff-index --cached --quiet HEAD ||
 100         die "Cannot rewrite branch(es) with a dirty working directory."
 101
 102 tempdir=.git-rewrite
 103 filter_env=
 104 filter_tree=
 105 filter_index=
 106 filter_parent=
 107 filter_msg=cat
 108 filter_commit='git commit-tree "$@"'
 109 filter_tag_name=
 110 filter_subdir=
 111 orig_namespace=refs/original/
 112 force=
 113 while :
 114 do
 115         test $# = 0 && usage
 116         case "$1" in
 117         --)
 118                 shift
 119                 break
 120                 ;;
 121         --force|-f)
 122                 shift
 123                 force=t
 124                 continue
 125                 ;;
 126         -*)
 127                 ;;
 128         *)
 129                 break;
 130         esac
 131
 132         # all switches take one argument
 133         ARG="$1"
 134         case "$#" in 1) usage ;; esac
 135         shift
 136         OPTARG="$1"
 137         shift
 138
 139         case "$ARG" in
 140         -d)
 141                 tempdir="$OPTARG"
 142                 ;;
 143         --env-filter)
 144                 filter_env="$OPTARG"
 145                 ;;
 146         --tree-filter)
 147                 filter_tree="$OPTARG"
 148                 ;;
 149         --index-filter)
 150                 filter_index="$OPTARG"
 151                 ;;
 152         --parent-filter)
 153                 filter_parent="$OPTARG"
 154                 ;;
 155         --msg-filter)
 156                 filter_msg="$OPTARG"
 157                 ;;
 158         --commit-filter)
 159                 filter_commit='SOURCE_FUNCTIONS=1 . "$this_script";'" $OPTARG"
 160                 ;;
 161         --tag-name-filter)
 162                 filter_tag_name="$OPTARG"
 163                 ;;
 164         --subdirectory-filter)
 165                 filter_subdir="$OPTARG"
 166                 ;;
 167         --original)
 168                 orig_namespace=$(expr "$OPTARG/" : '\(.*[^/]\)/*$')/
 169                 ;;
 170         *)
 171                 usage
 172                 ;;
 173         esac
 174 done
 175
 176 case "$force" in
 177 t)
 178         rm -rf "$tempdir"
 179 ;;
 180 '')
 181         test -d "$tempdir" &&
 182                 die "$tempdir already exists, please remove it"
 183 esac
 184 mkdir -p "$tempdir/t" &&
 185 tempdir="$(cd "$tempdir"; pwd)" &&
 186 cd "$tempdir/t" &&
 187 workdir="$(pwd)" ||
 188 die ""
 189
 190 # Make sure refs/original is empty
 191 git for-each-ref > "$tempdir"/backup-refs
 192 while read sha1 type name
 193 do
 194         case "$force,$name" in
 195         ,$orig_namespace*)
 196                 die "Namespace $orig_namespace not empty"
 197         ;;
 198         t,$orig_namespace*)
 199                 git update-ref -d "$name" $sha1
 200         ;;
 201         esac
 202 done < "$tempdir"/backup-refs
 203
 204 ORIG_GIT_DIR="$GIT_DIR"
 205 ORIG_GIT_WORK_TREE="$GIT_WORK_TREE"
 206 ORIG_GIT_INDEX_FILE="$GIT_INDEX_FILE"
 207 export GIT_DIR GIT_WORK_TREE=.
 208
 209 # These refs should be updated if their heads were rewritten
 210
 211 git rev-parse --revs-only --symbolic "$@" |
 212 while read ref
 213 do
 214         # normalize ref
 215         case "$ref" in
 216         HEAD)
 217                 ref="$(git symbolic-ref "$ref")"
 218         ;;
 219         refs/*)
 220         ;;
 221         *)
 222                 ref="$(git for-each-ref --format='%(refname)' |
 223                         grep /"$ref")"
 224         esac
 225
 226         git check-ref-format "$ref" && echo "$ref"
 227 done > "$tempdir"/heads
 228
 229 test -s "$tempdir"/heads ||
 230         die "Which ref do you want to rewrite?"
 231
 232 export GIT_INDEX_FILE="$(pwd)/../index"
 233 git read-tree || die "Could not seed the index"
 234
 235 ret=0
 236
 237 # map old->new commit ids for rewriting parents
 238 mkdir ../map || die "Could not create map/ directory"
 239
 240 case "$filter_subdir" in
 241 "")
 242         git rev-list --reverse --topo-order --default HEAD \
 243                 --parents "$@"
 244         ;;
 245 *)
 246         git rev-list --reverse --topo-order --default HEAD \
 247                 --parents --full-history "$@" -- "$filter_subdir"
 248 esac > ../revs || die "Could not get the commits"
 249 commits=$(wc -l <../revs | tr -d " ")
 250
 251 test $commits -eq 0 && die "Found nothing to rewrite"
 252
 253 # Rewrite the commits
 254
 255 i=0
 256 while read commit parents; do
 257         i=$(($i+1))
 258         printf "\rRewrite $commit ($i/$commits)"
 259
 260         case "$filter_subdir" in
 261         "")
 262                 git read-tree -i -m $commit
 263                 ;;
 264         *)
 265                 git read-tree -i -m $commit:"$filter_subdir"
 266         esac || die "Could not initialize the index"
 267
 268         export GIT_COMMIT=$commit
 269         git cat-file commit "$commit" >../commit ||
 270                 die "Cannot read commit $commit"
 271
 272         eval "$(set_ident AUTHOR <../commit)" ||
 273                 die "setting author failed for commit $commit"
 274         eval "$(set_ident COMMITTER <../commit)" ||
 275                 die "setting committer failed for commit $commit"
 276         eval "$filter_env" < /dev/null ||
 277                 die "env filter failed: $filter_env"
 278
 279         if [ "$filter_tree" ]; then
 280                 git checkout-index -f -u -a ||
 281                         die "Could not checkout the index"
 282                 # files that $commit removed are now still in the working tree;
 283                 # remove them, else they would be added again
 284                 git ls-files -z --others | xargs -0 rm -f
 285                 eval "$filter_tree" < /dev/null ||
 286                         die "tree filter failed: $filter_tree"
 287
 288                 git diff-index -r $commit | cut -f 2- | tr '\n' '\0' | \
 289                         xargs -0 git update-index --add --replace --remove
 290                 git ls-files -z --others | \
 291                         xargs -0 git update-index --add --replace --remove
 292         fi
 293
 294         eval "$filter_index" < /dev/null ||
 295                 die "index filter failed: $filter_index"
 296
 297         parentstr=
 298         for parent in $parents; do
 299                 for reparent in $(map "$parent"); do
 300                         parentstr="$parentstr -p $reparent"
 301                 done
 302         done
 303         if [ "$filter_parent" ]; then
 304                 parentstr="$(echo "$parentstr" | eval "$filter_parent")" ||
 305                                 die "parent filter failed: $filter_parent"
 306         fi
 307
 308         sed -e '1,/^$/d' <../commit | \
 309                 eval "$filter_msg" > ../message ||
 310                         die "msg filter failed: $filter_msg"
 311         sh -c "$filter_commit" "git commit-tree" \
 312                 $(git write-tree) $parentstr < ../message > ../map/$commit
 313 done <../revs
 314
 315 # In case of a subdirectory filter, it is possible that a specified head
 316 # is not in the set of rewritten commits, because it was pruned by the
 317 # revision walker.  Fix it by mapping these heads to the next rewritten
 318 # ancestor(s), i.e. the boundaries in the set of rewritten commits.
 319
 320 # NEEDSWORK: we should sort the unmapped refs topologically first
 321 while read ref
 322 do
 323         sha1=$(git rev-parse "$ref"^0)
 324         test -f "$workdir"/../map/$sha1 && continue
 325         # Assign the boundarie(s) in the set of rewritten commits
 326         # as the replacement commit(s).
 327         # (This would look a bit nicer if --not --stdin worked.)
 328         for p in $( (cd "$workdir"/../map; ls | sed "s/^/^/") |
 329                 git rev-list $ref --boundary --stdin |
 330                 sed -n "s/^-//p")
 331         do
 332                 map $p >> "$workdir"/../map/$sha1
 333         done
 334 done < "$tempdir"/heads
 335
 336 # Finally update the refs
 337
 338 _x40='[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]'
 339 _x40="$_x40$_x40$_x40$_x40$_x40$_x40$_x40$_x40"
 340 count=0
 341 echo
 342 while read ref
 343 do
 344         # avoid rewriting a ref twice
 345         test -f "$orig_namespace$ref" && continue
 346
 347         sha1=$(git rev-parse "$ref"^0)
 348         rewritten=$(map $sha1)
 349
 350         test $sha1 = "$rewritten" &&
 351                 warn "WARNING: Ref '$ref' is unchanged" &&
 352                 continue
 353
 354         case "$rewritten" in
 355         '')
 356                 echo "Ref '$ref' was deleted"
 357                 git update-ref -m "filter-branch: delete" -d "$ref" $sha1 ||
 358                         die "Could not delete $ref"
 359         ;;
 360         $_x40)
 361                 echo "Ref '$ref' was rewritten"
 362                 git update-ref -m "filter-branch: rewrite" \
 363                                 "$ref" $rewritten $sha1 ||
 364                         die "Could not rewrite $ref"
 365         ;;
 366         *)
 367                 # NEEDSWORK: possibly add -Werror, making this an error
 368                 warn "WARNING: '$ref' was rewritten into multiple commits:"
 369                 warn "$rewritten"
 370                 warn "WARNING: Ref '$ref' points to the first one now."
 371                 rewritten=$(echo "$rewritten" | head -n 1)
 372                 git update-ref -m "filter-branch: rewrite to first" \
 373                                 "$ref" $rewritten $sha1 ||
 374                         die "Could not rewrite $ref"
 375         ;;
 376         esac
 377         git update-ref -m "filter-branch: backup" "$orig_namespace$ref" $sha1
 378         count=$(($count+1))
 379 done < "$tempdir"/heads
 380
 381 # TODO: This should possibly go, with the semantics that all positive given
 382 #       refs are updated, and their original heads stored in refs/original/
 383 # Filter tags
 384
 385 if [ "$filter_tag_name" ]; then
 386         git for-each-ref --format='%(objectname) %(objecttype) %(refname)' refs/tags |
 387         while read sha1 type ref; do
 388                 ref="${ref#refs/tags/}"
 389                 # XXX: Rewrite tagged trees as well?
 390                 if [ "$type" != "commit" -a "$type" != "tag" ]; then
 391                         continue;
 392                 fi
 393
 394                 if [ "$type" = "tag" ]; then
 395                         # Dereference to a commit
 396                         sha1t="$sha1"
 397                         sha1="$(git rev-parse "$sha1"^{commit} 2>/dev/null)" || continue
 398                 fi
 399
 400                 [ -f "../map/$sha1" ] || continue
 401                 new_sha1="$(cat "../map/$sha1")"
 402                 export GIT_COMMIT="$sha1"
 403                 new_ref="$(echo "$ref" | eval "$filter_tag_name")" ||
 404                         die "tag name filter failed: $filter_tag_name"
 405
 406                 echo "$ref -> $new_ref ($sha1 -> $new_sha1)"
 407
 408                 if [ "$type" = "tag" ]; then
 409                         # Warn that we are not rewriting the tag object itself.
 410                         warn "unreferencing tag object $sha1t"
 411                 fi
 412
 413                 git update-ref "refs/tags/$new_ref" "$new_sha1" ||
 414                         die "Could not write tag $new_ref"
 415         done
 416 fi
 417
 418 cd ../..
 419 rm -rf "$tempdir"
 420 echo
 421 test $count -gt 0 && echo "These refs were rewritten:"
 422 git show-ref | grep ^"$orig_namespace"
 423
 424 unset GIT_DIR GIT_WORK_TREE GIT_INDEX_FILE
 425 test -z "$ORIG_GIT_DIR" || GIT_DIR="$ORIG_GIT_DIR" && export GIT_DIR
 426 test -z "$ORIG_GIT_WORK_TREE" || GIT_WORK_TREE="$ORIG_GIT_WORK_TREE" &&
 427         export GIT_WORK_TREE
 428 test -z "$ORIG_GIT_INDEX_FILE" || GIT_INDEX_FILE="$ORIG_GIT_INDEX_FILE" &&
 429         export GIT_INDEX_FILE
 430 git read-tree -u -m HEAD
 431
 432 exit $ret