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