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=$(cat << \EOF
  14 EMPTY_TREE=$(git hash-object -t tree /dev/null)
  15
  16 warn () {
  17         echo "$*" >&2
  18 }
  19
  20 map()
  21 {
  22         # if it was not rewritten, take the original
  23         if test -r "$workdir/../map/$1"
  24         then
  25                 cat "$workdir/../map/$1"
  26         else
  27                 echo "$1"
  28         fi
  29 }
  30
  31 # if you run 'skip_commit "$@"' in a commit filter, it will print
  32 # the (mapped) parents, effectively skipping the commit.
  33
  34 skip_commit()
  35 {
  36         shift;
  37         while [ -n "$1" ];
  38         do
  39                 shift;
  40                 map "$1";
  41                 shift;
  42         done;
  43 }
  44
  45 # if you run 'git_commit_non_empty_tree "$@"' in a commit filter,
  46 # it will skip commits that leave the tree untouched, commit the other.
  47 git_commit_non_empty_tree()
  48 {
  49         if test $# = 3 && test "$1" = $(git rev-parse "$3^{tree}"); then
  50                 map "$3"
  51         elif test $# = 1 && test "$1" = $EMPTY_TREE; then
  52                 :
  53         else
  54                 git commit-tree "$@"
  55         fi
  56 }
  57 # override die(): this version puts in an extra line break, so that
  58 # the progress is still visible
  59
  60 die()
  61 {
  62         echo >&2
  63         echo "$*" >&2
  64         exit 1
  65 }
  66 EOF
  67 )
  68
  69 eval "$functions"
  70
  71 finish_ident() {
  72         # Ensure non-empty id name.
  73         echo "case \"\$GIT_$1_NAME\" in \"\") GIT_$1_NAME=\"\${GIT_$1_EMAIL%%@*}\" && export GIT_$1_NAME;; esac"
  74         # And make sure everything is exported.
  75         echo "export GIT_$1_NAME"
  76         echo "export GIT_$1_EMAIL"
  77         echo "export GIT_$1_DATE"
  78 }
  79
  80 set_ident () {
  81         parse_ident_from_commit author AUTHOR committer COMMITTER
  82         finish_ident AUTHOR
  83         finish_ident COMMITTER
  84 }
  85
  86 USAGE="[--setup <command>] [--subdirectory-filter <directory>] [--env-filter <command>]
  87         [--tree-filter <command>] [--index-filter <command>]
  88         [--parent-filter <command>] [--msg-filter <command>]
  89         [--commit-filter <command>] [--tag-name-filter <command>]
  90         [--original <namespace>]
  91         [-d <directory>] [-f | --force] [--state-branch <branch>]
  92         [--] [<rev-list options>...]"
  93
  94 OPTIONS_SPEC=
  95 . git-sh-setup
  96
  97 if [ "$(is_bare_repository)" = false ]; then
  98         require_clean_work_tree 'rewrite branches'
  99 fi
 100
 101 tempdir=.git-rewrite
 102 filter_setup=
 103 filter_env=
 104 filter_tree=
 105 filter_index=
 106 filter_parent=
 107 filter_msg=cat
 108 filter_commit=
 109 filter_tag_name=
 110 filter_subdir=
 111 state_branch=
 112 orig_namespace=refs/original/
 113 force=
 114 prune_empty=
 115 remap_to_ancestor=
 116 while :
 117 do
 118         case "$1" in
 119         --)
 120                 shift
 121                 break
 122                 ;;
 123         --force|-f)
 124                 shift
 125                 force=t
 126                 continue
 127                 ;;
 128         --remap-to-ancestor)
 129                 # deprecated ($remap_to_ancestor is set now automatically)
 130                 shift
 131                 remap_to_ancestor=t
 132                 continue
 133                 ;;
 134         --prune-empty)
 135                 shift
 136                 prune_empty=t
 137                 continue
 138                 ;;
 139         -*)
 140                 ;;
 141         *)
 142                 break;
 143         esac
 144
 145         # all switches take one argument
 146         ARG="$1"
 147         case "$#" in 1) usage ;; esac
 148         shift
 149         OPTARG="$1"
 150         shift
 151
 152         case "$ARG" in
 153         -d)
 154                 tempdir="$OPTARG"
 155                 ;;
 156         --setup)
 157                 filter_setup="$OPTARG"
 158                 ;;
 159         --subdirectory-filter)
 160                 filter_subdir="$OPTARG"
 161                 remap_to_ancestor=t
 162                 ;;
 163         --env-filter)
 164                 filter_env="$OPTARG"
 165                 ;;
 166         --tree-filter)
 167                 filter_tree="$OPTARG"
 168                 ;;
 169         --index-filter)
 170                 filter_index="$OPTARG"
 171                 ;;
 172         --parent-filter)
 173                 filter_parent="$OPTARG"
 174                 ;;
 175         --msg-filter)
 176                 filter_msg="$OPTARG"
 177                 ;;
 178         --commit-filter)
 179                 filter_commit="$functions; $OPTARG"
 180                 ;;
 181         --tag-name-filter)
 182                 filter_tag_name="$OPTARG"
 183                 ;;
 184         --original)
 185                 orig_namespace=$(expr "$OPTARG/" : '\(.*[^/]\)/*$')/
 186                 ;;
 187         --state-branch)
 188                 state_branch="$OPTARG"
 189                 ;;
 190         *)
 191                 usage
 192                 ;;
 193         esac
 194 done
 195
 196 case "$prune_empty,$filter_commit" in
 197 ,)
 198         filter_commit='git commit-tree "$@"';;
 199 t,)
 200         filter_commit="$functions;"' git_commit_non_empty_tree "$@"';;
 201 ,*)
 202         ;;
 203 *)
 204         die "Cannot set --prune-empty and --commit-filter at the same time"
 205 esac
 206
 207 case "$force" in
 208 t)
 209         rm -rf "$tempdir"
 210 ;;
 211 '')
 212         test -d "$tempdir" &&
 213                 die "$tempdir already exists, please remove it"
 214 esac
 215 orig_dir=$(pwd)
 216 mkdir -p "$tempdir/t" &&
 217 tempdir="$(cd "$tempdir"; pwd)" &&
 218 cd "$tempdir/t" &&
 219 workdir="$(pwd)" ||
 220 die ""
 221
 222 # Remove tempdir on exit
 223 trap 'cd "$orig_dir"; rm -rf "$tempdir"' 0
 224
 225 ORIG_GIT_DIR="$GIT_DIR"
 226 ORIG_GIT_WORK_TREE="$GIT_WORK_TREE"
 227 ORIG_GIT_INDEX_FILE="$GIT_INDEX_FILE"
 228 ORIG_GIT_AUTHOR_NAME="$GIT_AUTHOR_NAME"
 229 ORIG_GIT_AUTHOR_EMAIL="$GIT_AUTHOR_EMAIL"
 230 ORIG_GIT_AUTHOR_DATE="$GIT_AUTHOR_DATE"
 231 ORIG_GIT_COMMITTER_NAME="$GIT_COMMITTER_NAME"
 232 ORIG_GIT_COMMITTER_EMAIL="$GIT_COMMITTER_EMAIL"
 233 ORIG_GIT_COMMITTER_DATE="$GIT_COMMITTER_DATE"
 234
 235 GIT_WORK_TREE=.
 236 export GIT_DIR GIT_WORK_TREE
 237
 238 # Make sure refs/original is empty
 239 git for-each-ref > "$tempdir"/backup-refs || exit
 240 while read sha1 type name
 241 do
 242         case "$force,$name" in
 243         ,$orig_namespace*)
 244                 die "Cannot create a new backup.
 245 A previous backup already exists in $orig_namespace
 246 Force overwriting the backup with -f"
 247         ;;
 248         t,$orig_namespace*)
 249                 git update-ref -d "$name" $sha1
 250         ;;
 251         esac
 252 done < "$tempdir"/backup-refs
 253
 254 # The refs should be updated if their heads were rewritten
 255 git rev-parse --no-flags --revs-only --symbolic-full-name \
 256         --default HEAD "$@" > "$tempdir"/raw-refs || exit
 257 while read ref
 258 do
 259         case "$ref" in ^?*) continue ;; esac
 260
 261         if git rev-parse --verify "$ref"^0 >/dev/null 2>&1
 262         then
 263                 echo "$ref"
 264         else
 265                 warn "WARNING: not rewriting '$ref' (not a committish)"
 266         fi
 267 done >"$tempdir"/heads <"$tempdir"/raw-refs
 268
 269 test -s "$tempdir"/heads ||
 270         die "You must specify a ref to rewrite."
 271
 272 GIT_INDEX_FILE="$(pwd)/../index"
 273 export GIT_INDEX_FILE
 274
 275 # map old->new commit ids for rewriting parents
 276 mkdir ../map || die "Could not create map/ directory"
 277
 278 if test -n "$state_branch"
 279 then
 280         state_commit=$(git rev-parse --no-flags --revs-only "$state_branch")
 281         if test -n "$state_commit"
 282         then
 283                 echo "Populating map from $state_branch ($state_commit)" 1>&2
 284                 perl -e'open(MAP, "-|", "git show $ARGV[0]:filter.map") or die;
 285                         while (<MAP>) {
 286                                 m/(.*):(.*)/ or die;
 287                                 open F, ">../map/$1" or die;
 288                                 print F "$2" or die;
 289                                 close(F) or die;
 290                         }
 291                         close(MAP) or die;' "$state_commit" \
 292                                 || die "Unable to load state from $state_branch:filter.map"
 293         else
 294                 echo "Branch $state_branch does not exist. Will create" 1>&2
 295         fi
 296 fi
 297
 298 # we need "--" only if there are no path arguments in $@
 299 nonrevs=$(git rev-parse --no-revs "$@") || exit
 300 if test -z "$nonrevs"
 301 then
 302         dashdash=--
 303 else
 304         dashdash=
 305         remap_to_ancestor=t
 306 fi
 307
 308 git rev-parse --revs-only "$@" >../parse
 309
 310 case "$filter_subdir" in
 311 "")
 312         eval set -- "$(git rev-parse --sq --no-revs "$@")"
 313         ;;
 314 *)
 315         eval set -- "$(git rev-parse --sq --no-revs "$@" $dashdash \
 316                 "$filter_subdir")"
 317         ;;
 318 esac
 319
 320 git rev-list --reverse --topo-order --default HEAD \
 321         --parents --simplify-merges --stdin "$@" <../parse >../revs ||
 322         die "Could not get the commits"
 323 commits=$(wc -l <../revs | tr -d " ")
 324
 325 test $commits -eq 0 && die_with_status 2 "Found nothing to rewrite"
 326
 327 # Rewrite the commits
 328 report_progress ()
 329 {
 330         if test -n "$progress" &&
 331                 test $git_filter_branch__commit_count -gt $next_sample_at
 332         then
 333                 count=$git_filter_branch__commit_count
 334
 335                 now=$(date +%s)
 336                 elapsed=$(($now - $start_timestamp))
 337                 remaining=$(( ($commits - $count) * $elapsed / $count ))
 338                 if test $elapsed -gt 0
 339                 then
 340                         next_sample_at=$(( ($elapsed + 1) * $count / $elapsed ))
 341                 else
 342                         next_sample_at=$(($next_sample_at + 1))
 343                 fi
 344                 progress=" ($elapsed seconds passed, remaining $remaining predicted)"
 345         fi
 346         printf "\rRewrite $commit ($count/$commits)$progress    "
 347 }
 348
 349 git_filter_branch__commit_count=0
 350
 351 progress= start_timestamp=
 352 if date '+%s' 2>/dev/null | grep -q '^[0-9][0-9]*$'
 353 then
 354         next_sample_at=0
 355         progress="dummy to ensure this is not empty"
 356         start_timestamp=$(date '+%s')
 357 fi
 358
 359 if test -n "$filter_index" ||
 360    test -n "$filter_tree" ||
 361    test -n "$filter_subdir"
 362 then
 363         need_index=t
 364 else
 365         need_index=
 366 fi
 367
 368 eval "$filter_setup" < /dev/null ||
 369         die "filter setup failed: $filter_setup"
 370
 371 while read commit parents; do
 372         git_filter_branch__commit_count=$(($git_filter_branch__commit_count+1))
 373
 374         report_progress
 375
 376         case "$filter_subdir" in
 377         "")
 378                 if test -n "$need_index"
 379                 then
 380                         GIT_ALLOW_NULL_SHA1=1 git read-tree -i -m $commit
 381                 fi
 382                 ;;
 383         *)
 384                 # The commit may not have the subdirectory at all
 385                 err=$(GIT_ALLOW_NULL_SHA1=1 \
 386                       git read-tree -i -m $commit:"$filter_subdir" 2>&1) || {
 387                         if ! git rev-parse -q --verify $commit:"$filter_subdir"
 388                         then
 389                                 rm -f "$GIT_INDEX_FILE"
 390                         else
 391                                 echo >&2 "$err"
 392                                 false
 393                         fi
 394                 }
 395         esac || die "Could not initialize the index"
 396
 397         GIT_COMMIT=$commit
 398         export GIT_COMMIT
 399         git cat-file commit "$commit" >../commit ||
 400                 die "Cannot read commit $commit"
 401
 402         eval "$(set_ident <../commit)" ||
 403                 die "setting author/committer failed for commit $commit"
 404         eval "$filter_env" < /dev/null ||
 405                 die "env filter failed: $filter_env"
 406
 407         if [ "$filter_tree" ]; then
 408                 git checkout-index -f -u -a ||
 409                         die "Could not checkout the index"
 410                 # files that $commit removed are now still in the working tree;
 411                 # remove them, else they would be added again
 412                 git clean -d -q -f -x
 413                 eval "$filter_tree" < /dev/null ||
 414                         die "tree filter failed: $filter_tree"
 415
 416                 (
 417                         git diff-index -r --name-only --ignore-submodules $commit -- &&
 418                         git ls-files --others
 419                 ) > "$tempdir"/tree-state || exit
 420                 git update-index --add --replace --remove --stdin \
 421                         < "$tempdir"/tree-state || exit
 422         fi
 423
 424         eval "$filter_index" < /dev/null ||
 425                 die "index filter failed: $filter_index"
 426
 427         parentstr=
 428         for parent in $parents; do
 429                 for reparent in $(map "$parent"); do
 430                         case "$parentstr " in
 431                         *" -p $reparent "*)
 432                                 ;;
 433                         *)
 434                                 parentstr="$parentstr -p $reparent"
 435                                 ;;
 436                         esac
 437                 done
 438         done
 439         if [ "$filter_parent" ]; then
 440                 parentstr="$(echo "$parentstr" | eval "$filter_parent")" ||
 441                                 die "parent filter failed: $filter_parent"
 442         fi
 443
 444         {
 445                 while IFS='' read -r header_line && test -n "$header_line"
 446                 do
 447                         # skip header lines...
 448                         :;
 449                 done
 450                 # and output the actual commit message
 451                 cat
 452         } <../commit |
 453                 eval "$filter_msg" > ../message ||
 454                         die "msg filter failed: $filter_msg"
 455
 456         if test -n "$need_index"
 457         then
 458                 tree=$(git write-tree)
 459         else
 460                 tree=$(git rev-parse "$commit^{tree}")
 461         fi
 462         workdir=$workdir @SHELL_PATH@ -c "$filter_commit" "git commit-tree" \
 463                 "$tree" $parentstr < ../message > ../map/$commit ||
 464                         die "could not write rewritten commit"
 465 done <../revs
 466
 467 # If we are filtering for paths, as in the case of a subdirectory
 468 # filter, it is possible that a specified head is not in the set of
 469 # rewritten commits, because it was pruned by the revision walker.
 470 # Ancestor remapping fixes this by mapping these heads to the unique
 471 # nearest ancestor that survived the pruning.
 472
 473 if test "$remap_to_ancestor" = t
 474 then
 475         while read ref
 476         do
 477                 sha1=$(git rev-parse "$ref"^0)
 478                 test -f "$workdir"/../map/$sha1 && continue
 479                 ancestor=$(git rev-list --simplify-merges -1 "$ref" "$@")
 480                 test "$ancestor" && echo $(map $ancestor) >> "$workdir"/../map/$sha1
 481         done < "$tempdir"/heads
 482 fi
 483
 484 # Finally update the refs
 485
 486 _x40='[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]'
 487 _x40="$_x40$_x40$_x40$_x40$_x40$_x40$_x40$_x40"
 488 echo
 489 while read ref
 490 do
 491         # avoid rewriting a ref twice
 492         test -f "$orig_namespace$ref" && continue
 493
 494         sha1=$(git rev-parse "$ref"^0)
 495         rewritten=$(map $sha1)
 496
 497         test $sha1 = "$rewritten" &&
 498                 warn "WARNING: Ref '$ref' is unchanged" &&
 499                 continue
 500
 501         case "$rewritten" in
 502         '')
 503                 echo "Ref '$ref' was deleted"
 504                 git update-ref -m "filter-branch: delete" -d "$ref" $sha1 ||
 505                         die "Could not delete $ref"
 506         ;;
 507         $_x40)
 508                 echo "Ref '$ref' was rewritten"
 509                 if ! git update-ref -m "filter-branch: rewrite" \
 510                                         "$ref" $rewritten $sha1 2>/dev/null; then
 511                         if test $(git cat-file -t "$ref") = tag; then
 512                                 if test -z "$filter_tag_name"; then
 513                                         warn "WARNING: You said to rewrite tagged commits, but not the corresponding tag."
 514                                         warn "WARNING: Perhaps use '--tag-name-filter cat' to rewrite the tag."
 515                                 fi
 516                         else
 517                                 die "Could not rewrite $ref"
 518                         fi
 519                 fi
 520         ;;
 521         *)
 522                 # NEEDSWORK: possibly add -Werror, making this an error
 523                 warn "WARNING: '$ref' was rewritten into multiple commits:"
 524                 warn "$rewritten"
 525                 warn "WARNING: Ref '$ref' points to the first one now."
 526                 rewritten=$(echo "$rewritten" | head -n 1)
 527                 git update-ref -m "filter-branch: rewrite to first" \
 528                                 "$ref" $rewritten $sha1 ||
 529                         die "Could not rewrite $ref"
 530         ;;
 531         esac
 532         git update-ref -m "filter-branch: backup" "$orig_namespace$ref" $sha1 ||
 533                  exit
 534 done < "$tempdir"/heads
 535
 536 # TODO: This should possibly go, with the semantics that all positive given
 537 #       refs are updated, and their original heads stored in refs/original/
 538 # Filter tags
 539
 540 if [ "$filter_tag_name" ]; then
 541         git for-each-ref --format='%(objectname) %(objecttype) %(refname)' refs/tags |
 542         while read sha1 type ref; do
 543                 ref="${ref#refs/tags/}"
 544                 # XXX: Rewrite tagged trees as well?
 545                 if [ "$type" != "commit" -a "$type" != "tag" ]; then
 546                         continue;
 547                 fi
 548
 549                 if [ "$type" = "tag" ]; then
 550                         # Dereference to a commit
 551                         sha1t="$sha1"
 552                         sha1="$(git rev-parse -q "$sha1"^{commit})" || continue
 553                 fi
 554
 555                 [ -f "../map/$sha1" ] || continue
 556                 new_sha1="$(cat "../map/$sha1")"
 557                 GIT_COMMIT="$sha1"
 558                 export GIT_COMMIT
 559                 new_ref="$(echo "$ref" | eval "$filter_tag_name")" ||
 560                         die "tag name filter failed: $filter_tag_name"
 561
 562                 echo "$ref -> $new_ref ($sha1 -> $new_sha1)"
 563
 564                 if [ "$type" = "tag" ]; then
 565                         new_sha1=$( ( printf 'object %s\ntype commit\ntag %s\n' \
 566                                                 "$new_sha1" "$new_ref"
 567                                 git cat-file tag "$ref" |
 568                                 sed -n \
 569                                     -e '1,/^$/{
 570                                           /^object /d
 571                                           /^type /d
 572                                           /^tag /d
 573                                         }' \
 574                                     -e '/^-----BEGIN PGP SIGNATURE-----/q' \
 575                                     -e 'p' ) |
 576                                 git hash-object -t tag -w --stdin) ||
 577                                 die "Could not create new tag object for $ref"
 578                         if git cat-file tag "$ref" | \
 579                            sane_grep '^-----BEGIN PGP SIGNATURE-----' >/dev/null 2>&1
 580                         then
 581                                 warn "gpg signature stripped from tag object $sha1t"
 582                         fi
 583                 fi
 584
 585                 git update-ref "refs/tags/$new_ref" "$new_sha1" ||
 586                         die "Could not write tag $new_ref"
 587         done
 588 fi
 589
 590 unset GIT_DIR GIT_WORK_TREE GIT_INDEX_FILE
 591 unset GIT_AUTHOR_NAME GIT_AUTHOR_EMAIL GIT_AUTHOR_DATE
 592 unset GIT_COMMITTER_NAME GIT_COMMITTER_EMAIL GIT_COMMITTER_DATE
 593 test -z "$ORIG_GIT_DIR" || {
 594         GIT_DIR="$ORIG_GIT_DIR" && export GIT_DIR
 595 }
 596 test -z "$ORIG_GIT_WORK_TREE" || {
 597         GIT_WORK_TREE="$ORIG_GIT_WORK_TREE" &&
 598         export GIT_WORK_TREE
 599 }
 600 test -z "$ORIG_GIT_INDEX_FILE" || {
 601         GIT_INDEX_FILE="$ORIG_GIT_INDEX_FILE" &&
 602         export GIT_INDEX_FILE
 603 }
 604 test -z "$ORIG_GIT_AUTHOR_NAME" || {
 605         GIT_AUTHOR_NAME="$ORIG_GIT_AUTHOR_NAME" &&
 606         export GIT_AUTHOR_NAME
 607 }
 608 test -z "$ORIG_GIT_AUTHOR_EMAIL" || {
 609         GIT_AUTHOR_EMAIL="$ORIG_GIT_AUTHOR_EMAIL" &&
 610         export GIT_AUTHOR_EMAIL
 611 }
 612 test -z "$ORIG_GIT_AUTHOR_DATE" || {
 613         GIT_AUTHOR_DATE="$ORIG_GIT_AUTHOR_DATE" &&
 614         export GIT_AUTHOR_DATE
 615 }
 616 test -z "$ORIG_GIT_COMMITTER_NAME" || {
 617         GIT_COMMITTER_NAME="$ORIG_GIT_COMMITTER_NAME" &&
 618         export GIT_COMMITTER_NAME
 619 }
 620 test -z "$ORIG_GIT_COMMITTER_EMAIL" || {
 621         GIT_COMMITTER_EMAIL="$ORIG_GIT_COMMITTER_EMAIL" &&
 622         export GIT_COMMITTER_EMAIL
 623 }
 624 test -z "$ORIG_GIT_COMMITTER_DATE" || {
 625         GIT_COMMITTER_DATE="$ORIG_GIT_COMMITTER_DATE" &&
 626         export GIT_COMMITTER_DATE
 627 }
 628
 629 if test -n "$state_branch"
 630 then
 631         echo "Saving rewrite state to $state_branch" 1>&2
 632         state_blob=$(
 633                 perl -e'opendir D, "../map" or die;
 634                         open H, "|-", "git hash-object -w --stdin" or die;
 635                         foreach (sort readdir(D)) {
 636                                 next if m/^\.\.?$/;
 637                                 open F, "<../map/$_" or die;
 638                                 chomp($f = <F>);
 639                                 print H "$_:$f\n" or die;
 640                         }
 641                         close(H) or die;' || die "Unable to save state")
 642         state_tree=$(printf '100644 blob %s\tfilter.map\n' "$state_blob" | git mktree)
 643         if test -n "$state_commit"
 644         then
 645                 state_commit=$(echo "Sync" | git commit-tree "$state_tree" -p "$state_commit")
 646         else
 647                 state_commit=$(echo "Sync" | git commit-tree "$state_tree" )
 648         fi
 649         git update-ref "$state_branch" "$state_commit"
 650 fi
 651
 652 cd "$orig_dir"
 653 rm -rf "$tempdir"
 654
 655 trap - 0
 656
 657 if [ "$(is_bare_repository)" = false ]; then
 658         git read-tree -u -m HEAD || exit
 659 fi
 660
 661 exit 0