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 # The following functions will also be available in the commit filter:
  12
  13 functions=$(cat << \EOF
  14 warn () {
  15         echo "$*" >&2
  16 }
  17
  18 map()
  19 {
  20         # if it was not rewritten, take the original
  21         if test -r "$workdir/../map/$1"
  22         then
  23                 cat "$workdir/../map/$1"
  24         else
  25                 echo "$1"
  26         fi
  27 }
  28
  29 # if you run 'skip_commit "$@"' in a commit filter, it will print
  30 # the (mapped) parents, effectively skipping the commit.
  31
  32 skip_commit()
  33 {
  34         shift;
  35         while [ -n "$1" ];
  36         do
  37                 shift;
  38                 map "$1";
  39                 shift;
  40         done;
  41 }
  42
  43 # override die(): this version puts in an extra line break, so that
  44 # the progress is still visible
  45
  46 die()
  47 {
  48         echo >&2
  49         echo "$*" >&2
  50         exit 1
  51 }
  52 EOF
  53 )
  54
  55 eval "$functions"
  56
  57 # When piped a commit, output a script to set the ident of either
  58 # "author" or "committer
  59
  60 set_ident () {
  61         lid="$(echo "$1" | tr "A-Z" "a-z")"
  62         uid="$(echo "$1" | tr "a-z" "A-Z")"
  63         pick_id_script='
  64                 /^'$lid' /{
  65                         s/'\''/'\''\\'\'\''/g
  66                         h
  67                         s/^'$lid' \([^<]*\) <[^>]*> .*$/\1/
  68                         s/'\''/'\''\'\'\''/g
  69                         s/.*/GIT_'$uid'_NAME='\''&'\''; export GIT_'$uid'_NAME/p
  70
  71                         g
  72                         s/^'$lid' [^<]* <\([^>]*\)> .*$/\1/
  73                         s/'\''/'\''\'\'\''/g
  74                         s/.*/GIT_'$uid'_EMAIL='\''&'\''; export GIT_'$uid'_EMAIL/p
  75
  76                         g
  77                         s/^'$lid' [^<]* <[^>]*> \(.*\)$/\1/
  78                         s/'\''/'\''\'\'\''/g
  79                         s/.*/GIT_'$uid'_DATE='\''&'\''; export GIT_'$uid'_DATE/p
  80
  81                         q
  82                 }
  83         '
  84
  85         LANG=C LC_ALL=C sed -ne "$pick_id_script"
  86         # Ensure non-empty id name.
  87         echo "case \"\$GIT_${uid}_NAME\" in \"\") GIT_${uid}_NAME=\"\${GIT_${uid}_EMAIL%%@*}\" && export GIT_${uid}_NAME;; esac"
  88 }
  89
  90 USAGE="[--env-filter <command>] [--tree-filter <command>] \
  91 [--index-filter <command>] [--parent-filter <command>] \
  92 [--msg-filter <command>] [--commit-filter <command>] \
  93 [--tag-name-filter <command>] [--subdirectory-filter <directory>] \
  94 [--original <namespace>] [-d <directory>] [-f | --force] \
  95 [<rev-list options>...]"
  96
  97 OPTIONS_SPEC=
  98 . git-sh-setup
  99
 100 git diff-files --quiet &&
 101         git diff-index --cached --quiet HEAD -- ||
 102         die "Cannot rewrite branch(es) with a dirty working directory."
 103
 104 tempdir=.git-rewrite
 105 filter_env=
 106 filter_tree=
 107 filter_index=
 108 filter_parent=
 109 filter_msg=cat
 110 filter_commit='git commit-tree "$@"'
 111 filter_tag_name=
 112 filter_subdir=
 113 orig_namespace=refs/original/
 114 force=
 115 while :
 116 do
 117         test $# = 0 && usage
 118         case "$1" in
 119         --)
 120                 shift
 121                 break
 122                 ;;
 123         --force|-f)
 124                 shift
 125                 force=t
 126                 continue
 127                 ;;
 128         -*)
 129                 ;;
 130         *)
 131                 break;
 132         esac
 133
 134         # all switches take one argument
 135         ARG="$1"
 136         case "$#" in 1) usage ;; esac
 137         shift
 138         OPTARG="$1"
 139         shift
 140
 141         case "$ARG" in
 142         -d)
 143                 tempdir="$OPTARG"
 144                 ;;
 145         --env-filter)
 146                 filter_env="$OPTARG"
 147                 ;;
 148         --tree-filter)
 149                 filter_tree="$OPTARG"
 150                 ;;
 151         --index-filter)
 152                 filter_index="$OPTARG"
 153                 ;;
 154         --parent-filter)
 155                 filter_parent="$OPTARG"
 156                 ;;
 157         --msg-filter)
 158                 filter_msg="$OPTARG"
 159                 ;;
 160         --commit-filter)
 161                 filter_commit="$functions; $OPTARG"
 162                 ;;
 163         --tag-name-filter)
 164                 filter_tag_name="$OPTARG"
 165                 ;;
 166         --subdirectory-filter)
 167                 filter_subdir="$OPTARG"
 168                 ;;
 169         --original)
 170                 orig_namespace=$(expr "$OPTARG/" : '\(.*[^/]\)/*$')/
 171                 ;;
 172         *)
 173                 usage
 174                 ;;
 175         esac
 176 done
 177
 178 case "$force" in
 179 t)
 180         rm -rf "$tempdir"
 181 ;;
 182 '')
 183         test -d "$tempdir" &&
 184                 die "$tempdir already exists, please remove it"
 185 esac
 186 mkdir -p "$tempdir/t" &&
 187 tempdir="$(cd "$tempdir"; pwd)" &&
 188 cd "$tempdir/t" &&
 189 workdir="$(pwd)" ||
 190 die ""
 191
 192 # Make sure refs/original is empty
 193 git for-each-ref > "$tempdir"/backup-refs
 194 while read sha1 type name
 195 do
 196         case "$force,$name" in
 197         ,$orig_namespace*)
 198                 die "Namespace $orig_namespace not empty"
 199         ;;
 200         t,$orig_namespace*)
 201                 git update-ref -d "$name" $sha1
 202         ;;
 203         esac
 204 done < "$tempdir"/backup-refs
 205
 206 ORIG_GIT_DIR="$GIT_DIR"
 207 ORIG_GIT_WORK_TREE="$GIT_WORK_TREE"
 208 ORIG_GIT_INDEX_FILE="$GIT_INDEX_FILE"
 209 GIT_WORK_TREE=.
 210 export GIT_DIR GIT_WORK_TREE
 211
 212 # These refs should be updated if their heads were rewritten
 213
 214 git rev-parse --revs-only --symbolic "$@" |
 215 while read ref
 216 do
 217         # normalize ref
 218         case "$ref" in
 219         HEAD)
 220                 ref="$(git symbolic-ref "$ref")"
 221         ;;
 222         refs/*)
 223         ;;
 224         *)
 225                 ref="$(git for-each-ref --format='%(refname)' |
 226                         grep /"$ref")"
 227         esac
 228
 229         git check-ref-format "$ref" && echo "$ref"
 230 done > "$tempdir"/heads
 231
 232 test -s "$tempdir"/heads ||
 233         die "Which ref do you want to rewrite?"
 234
 235 GIT_INDEX_FILE="$(pwd)/../index"
 236 export GIT_INDEX_FILE
 237 git read-tree || die "Could not seed the index"
 238
 239 ret=0
 240
 241 # map old->new commit ids for rewriting parents
 242 mkdir ../map || die "Could not create map/ directory"
 243
 244 case "$filter_subdir" in
 245 "")
 246         git rev-list --reverse --topo-order --default HEAD \
 247                 --parents "$@"
 248         ;;
 249 *)
 250         git rev-list --reverse --topo-order --default HEAD \
 251                 --parents --full-history "$@" -- "$filter_subdir"
 252 esac > ../revs || die "Could not get the commits"
 253 commits=$(wc -l <../revs | tr -d " ")
 254
 255 test $commits -eq 0 && die "Found nothing to rewrite"
 256
 257 # Rewrite the commits
 258
 259 i=0
 260 while read commit parents; do
 261         i=$(($i+1))
 262         printf "\rRewrite $commit ($i/$commits)"
 263
 264         case "$filter_subdir" in
 265         "")
 266                 git read-tree -i -m $commit
 267                 ;;
 268         *)
 269                 git read-tree -i -m $commit:"$filter_subdir"
 270         esac || die "Could not initialize the index"
 271
 272         GIT_COMMIT=$commit
 273         export GIT_COMMIT
 274         git cat-file commit "$commit" >../commit ||
 275                 die "Cannot read commit $commit"
 276
 277         eval "$(set_ident AUTHOR <../commit)" ||
 278                 die "setting author failed for commit $commit"
 279         eval "$(set_ident COMMITTER <../commit)" ||
 280                 die "setting committer failed for commit $commit"
 281         eval "$filter_env" < /dev/null ||
 282                 die "env filter failed: $filter_env"
 283
 284         if [ "$filter_tree" ]; then
 285                 git checkout-index -f -u -a ||
 286                         die "Could not checkout the index"
 287                 # files that $commit removed are now still in the working tree;
 288                 # remove them, else they would be added again
 289                 git ls-files -z --others | xargs -0 rm -f
 290                 eval "$filter_tree" < /dev/null ||
 291                         die "tree filter failed: $filter_tree"
 292
 293                 git diff-index -r $commit | cut -f 2- | tr '\n' '\0' | \
 294                         xargs -0 git update-index --add --replace --remove
 295                 git ls-files -z --others | \
 296                         xargs -0 git update-index --add --replace --remove
 297         fi
 298
 299         eval "$filter_index" < /dev/null ||
 300                 die "index filter failed: $filter_index"
 301
 302         parentstr=
 303         for parent in $parents; do
 304                 for reparent in $(map "$parent"); do
 305                         parentstr="$parentstr -p $reparent"
 306                 done
 307         done
 308         if [ "$filter_parent" ]; then
 309                 parentstr="$(echo "$parentstr" | eval "$filter_parent")" ||
 310                                 die "parent filter failed: $filter_parent"
 311         fi
 312
 313         sed -e '1,/^$/d' <../commit | \
 314                 eval "$filter_msg" > ../message ||
 315                         die "msg filter failed: $filter_msg"
 316         sh -c "$filter_commit" "git commit-tree" \
 317                 $(git write-tree) $parentstr < ../message > ../map/$commit
 318 done <../revs
 319
 320 # In case of a subdirectory filter, it is possible that a specified head
 321 # is not in the set of rewritten commits, because it was pruned by the
 322 # revision walker.  Fix it by mapping these heads to the next rewritten
 323 # ancestor(s), i.e. the boundaries in the set of rewritten commits.
 324
 325 # NEEDSWORK: we should sort the unmapped refs topologically first
 326 while read ref
 327 do
 328         sha1=$(git rev-parse "$ref"^0)
 329         test -f "$workdir"/../map/$sha1 && continue
 330         # Assign the boundarie(s) in the set of rewritten commits
 331         # as the replacement commit(s).
 332         # (This would look a bit nicer if --not --stdin worked.)
 333         for p in $( (cd "$workdir"/../map; ls | sed "s/^/^/") |
 334                 git rev-list $ref --boundary --stdin |
 335                 sed -n "s/^-//p")
 336         do
 337                 map $p >> "$workdir"/../map/$sha1
 338         done
 339 done < "$tempdir"/heads
 340
 341 # Finally update the refs
 342
 343 _x40='[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]'
 344 _x40="$_x40$_x40$_x40$_x40$_x40$_x40$_x40$_x40"
 345 count=0
 346 echo
 347 while read ref
 348 do
 349         # avoid rewriting a ref twice
 350         test -f "$orig_namespace$ref" && continue
 351
 352         sha1=$(git rev-parse "$ref"^0)
 353         rewritten=$(map $sha1)
 354
 355         test $sha1 = "$rewritten" &&
 356                 warn "WARNING: Ref '$ref' is unchanged" &&
 357                 continue
 358
 359         case "$rewritten" in
 360         '')
 361                 echo "Ref '$ref' was deleted"
 362                 git update-ref -m "filter-branch: delete" -d "$ref" $sha1 ||
 363                         die "Could not delete $ref"
 364         ;;
 365         $_x40)
 366                 echo "Ref '$ref' was rewritten"
 367                 git update-ref -m "filter-branch: rewrite" \
 368                                 "$ref" $rewritten $sha1 ||
 369                         die "Could not rewrite $ref"
 370         ;;
 371         *)
 372                 # NEEDSWORK: possibly add -Werror, making this an error
 373                 warn "WARNING: '$ref' was rewritten into multiple commits:"
 374                 warn "$rewritten"
 375                 warn "WARNING: Ref '$ref' points to the first one now."
 376                 rewritten=$(echo "$rewritten" | head -n 1)
 377                 git update-ref -m "filter-branch: rewrite to first" \
 378                                 "$ref" $rewritten $sha1 ||
 379                         die "Could not rewrite $ref"
 380         ;;
 381         esac
 382         git update-ref -m "filter-branch: backup" "$orig_namespace$ref" $sha1
 383         count=$(($count+1))
 384 done < "$tempdir"/heads
 385
 386 # TODO: This should possibly go, with the semantics that all positive given
 387 #       refs are updated, and their original heads stored in refs/original/
 388 # Filter tags
 389
 390 if [ "$filter_tag_name" ]; then
 391         git for-each-ref --format='%(objectname) %(objecttype) %(refname)' refs/tags |
 392         while read sha1 type ref; do
 393                 ref="${ref#refs/tags/}"
 394                 # XXX: Rewrite tagged trees as well?
 395                 if [ "$type" != "commit" -a "$type" != "tag" ]; then
 396                         continue;
 397                 fi
 398
 399                 if [ "$type" = "tag" ]; then
 400                         # Dereference to a commit
 401                         sha1t="$sha1"
 402                         sha1="$(git rev-parse "$sha1"^{commit} 2>/dev/null)" || continue
 403                 fi
 404
 405                 [ -f "../map/$sha1" ] || continue
 406                 new_sha1="$(cat "../map/$sha1")"
 407                 GIT_COMMIT="$sha1"
 408                 export GIT_COMMIT
 409                 new_ref="$(echo "$ref" | eval "$filter_tag_name")" ||
 410                         die "tag name filter failed: $filter_tag_name"
 411
 412                 echo "$ref -> $new_ref ($sha1 -> $new_sha1)"
 413
 414                 if [ "$type" = "tag" ]; then
 415                         # Warn that we are not rewriting the tag object itself.
 416                         warn "unreferencing tag object $sha1t"
 417                 fi
 418
 419                 git update-ref "refs/tags/$new_ref" "$new_sha1" ||
 420                         die "Could not write tag $new_ref"
 421         done
 422 fi
 423
 424 cd ../..
 425 rm -rf "$tempdir"
 426 echo
 427 test $count -gt 0 && echo "These refs were rewritten:"
 428 git show-ref | grep ^"$orig_namespace"
 429
 430 unset GIT_DIR GIT_WORK_TREE GIT_INDEX_FILE
 431 test -z "$ORIG_GIT_DIR" || GIT_DIR="$ORIG_GIT_DIR" && export GIT_DIR
 432 test -z "$ORIG_GIT_WORK_TREE" || GIT_WORK_TREE="$ORIG_GIT_WORK_TREE" &&
 433         export GIT_WORK_TREE
 434 test -z "$ORIG_GIT_INDEX_FILE" || GIT_INDEX_FILE="$ORIG_GIT_INDEX_FILE" &&
 435         export GIT_INDEX_FILE
 436 git read-tree -u -m HEAD
 437
 438 exit $ret