Merge branch 'dt/cat-file-follow-symlinks'
[git] / 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 "$1";
39                 shift;
40         done;
41 }
42
43 # if you run 'git_commit_non_empty_tree "$@"' in a commit filter,
44 # it will skip commits that leave the tree untouched, commit the other.
45 git_commit_non_empty_tree()
46 {
47         if test $# = 3 && test "$1" = $(git rev-parse "$3^{tree}"); then
48                 map "$3"
49         else
50                 git commit-tree "$@"
51         fi
52 }
53 # override die(): this version puts in an extra line break, so that
54 # the progress is still visible
55
56 die()
57 {
58         echo >&2
59         echo "$*" >&2
60         exit 1
61 }
62 EOF
63 )
64
65 eval "$functions"
66
67 finish_ident() {
68         # Ensure non-empty id name.
69         echo "case \"\$GIT_$1_NAME\" in \"\") GIT_$1_NAME=\"\${GIT_$1_EMAIL%%@*}\" && export GIT_$1_NAME;; esac"
70         # And make sure everything is exported.
71         echo "export GIT_$1_NAME"
72         echo "export GIT_$1_EMAIL"
73         echo "export GIT_$1_DATE"
74 }
75
76 set_ident () {
77         parse_ident_from_commit author AUTHOR committer COMMITTER
78         finish_ident AUTHOR
79         finish_ident COMMITTER
80 }
81
82 USAGE="[--env-filter <command>] [--tree-filter <command>]
83         [--index-filter <command>] [--parent-filter <command>]
84         [--msg-filter <command>] [--commit-filter <command>]
85         [--tag-name-filter <command>] [--subdirectory-filter <directory>]
86         [--original <namespace>] [-d <directory>] [-f | --force]
87         [<rev-list options>...]"
88
89 OPTIONS_SPEC=
90 . git-sh-setup
91
92 if [ "$(is_bare_repository)" = false ]; then
93         require_clean_work_tree 'rewrite branches'
94 fi
95
96 tempdir=.git-rewrite
97 filter_env=
98 filter_tree=
99 filter_index=
100 filter_parent=
101 filter_msg=cat
102 filter_commit=
103 filter_tag_name=
104 filter_subdir=
105 orig_namespace=refs/original/
106 force=
107 prune_empty=
108 remap_to_ancestor=
109 while :
110 do
111         case "$1" in
112         --)
113                 shift
114                 break
115                 ;;
116         --force|-f)
117                 shift
118                 force=t
119                 continue
120                 ;;
121         --remap-to-ancestor)
122                 # deprecated ($remap_to_ancestor is set now automatically)
123                 shift
124                 remap_to_ancestor=t
125                 continue
126                 ;;
127         --prune-empty)
128                 shift
129                 prune_empty=t
130                 continue
131                 ;;
132         -*)
133                 ;;
134         *)
135                 break;
136         esac
137
138         # all switches take one argument
139         ARG="$1"
140         case "$#" in 1) usage ;; esac
141         shift
142         OPTARG="$1"
143         shift
144
145         case "$ARG" in
146         -d)
147                 tempdir="$OPTARG"
148                 ;;
149         --env-filter)
150                 filter_env="$OPTARG"
151                 ;;
152         --tree-filter)
153                 filter_tree="$OPTARG"
154                 ;;
155         --index-filter)
156                 filter_index="$OPTARG"
157                 ;;
158         --parent-filter)
159                 filter_parent="$OPTARG"
160                 ;;
161         --msg-filter)
162                 filter_msg="$OPTARG"
163                 ;;
164         --commit-filter)
165                 filter_commit="$functions; $OPTARG"
166                 ;;
167         --tag-name-filter)
168                 filter_tag_name="$OPTARG"
169                 ;;
170         --subdirectory-filter)
171                 filter_subdir="$OPTARG"
172                 remap_to_ancestor=t
173                 ;;
174         --original)
175                 orig_namespace=$(expr "$OPTARG/" : '\(.*[^/]\)/*$')/
176                 ;;
177         *)
178                 usage
179                 ;;
180         esac
181 done
182
183 case "$prune_empty,$filter_commit" in
184 ,)
185         filter_commit='git commit-tree "$@"';;
186 t,)
187         filter_commit="$functions;"' git_commit_non_empty_tree "$@"';;
188 ,*)
189         ;;
190 *)
191         die "Cannot set --prune-empty and --commit-filter at the same time"
192 esac
193
194 case "$force" in
195 t)
196         rm -rf "$tempdir"
197 ;;
198 '')
199         test -d "$tempdir" &&
200                 die "$tempdir already exists, please remove it"
201 esac
202 orig_dir=$(pwd)
203 mkdir -p "$tempdir/t" &&
204 tempdir="$(cd "$tempdir"; pwd)" &&
205 cd "$tempdir/t" &&
206 workdir="$(pwd)" ||
207 die ""
208
209 # Remove tempdir on exit
210 trap 'cd "$orig_dir"; rm -rf "$tempdir"' 0
211
212 ORIG_GIT_DIR="$GIT_DIR"
213 ORIG_GIT_WORK_TREE="$GIT_WORK_TREE"
214 ORIG_GIT_INDEX_FILE="$GIT_INDEX_FILE"
215 GIT_WORK_TREE=.
216 export GIT_DIR GIT_WORK_TREE
217
218 # Make sure refs/original is empty
219 git for-each-ref > "$tempdir"/backup-refs || exit
220 while read sha1 type name
221 do
222         case "$force,$name" in
223         ,$orig_namespace*)
224                 die "Cannot create a new backup.
225 A previous backup already exists in $orig_namespace
226 Force overwriting the backup with -f"
227         ;;
228         t,$orig_namespace*)
229                 git update-ref -d "$name" $sha1
230         ;;
231         esac
232 done < "$tempdir"/backup-refs
233
234 # The refs should be updated if their heads were rewritten
235 git rev-parse --no-flags --revs-only --symbolic-full-name \
236         --default HEAD "$@" > "$tempdir"/raw-heads || exit
237 sed -e '/^^/d' "$tempdir"/raw-heads >"$tempdir"/heads
238
239 test -s "$tempdir"/heads ||
240         die "Which ref do you want to rewrite?"
241
242 GIT_INDEX_FILE="$(pwd)/../index"
243 export GIT_INDEX_FILE
244
245 # map old->new commit ids for rewriting parents
246 mkdir ../map || die "Could not create map/ directory"
247
248 # we need "--" only if there are no path arguments in $@
249 nonrevs=$(git rev-parse --no-revs "$@") || exit
250 if test -z "$nonrevs"
251 then
252         dashdash=--
253 else
254         dashdash=
255         remap_to_ancestor=t
256 fi
257
258 git rev-parse --revs-only "$@" >../parse
259
260 case "$filter_subdir" in
261 "")
262         eval set -- "$(git rev-parse --sq --no-revs "$@")"
263         ;;
264 *)
265         eval set -- "$(git rev-parse --sq --no-revs "$@" $dashdash \
266                 "$filter_subdir")"
267         ;;
268 esac
269
270 git rev-list --reverse --topo-order --default HEAD \
271         --parents --simplify-merges --stdin "$@" <../parse >../revs ||
272         die "Could not get the commits"
273 commits=$(wc -l <../revs | tr -d " ")
274
275 test $commits -eq 0 && die "Found nothing to rewrite"
276
277 # Rewrite the commits
278
279 git_filter_branch__commit_count=0
280 while read commit parents; do
281         git_filter_branch__commit_count=$(($git_filter_branch__commit_count+1))
282         printf "\rRewrite $commit ($git_filter_branch__commit_count/$commits)"
283
284         case "$filter_subdir" in
285         "")
286                 GIT_ALLOW_NULL_SHA1=1 git read-tree -i -m $commit
287                 ;;
288         *)
289                 # The commit may not have the subdirectory at all
290                 err=$(GIT_ALLOW_NULL_SHA1=1 \
291                       git read-tree -i -m $commit:"$filter_subdir" 2>&1) || {
292                         if ! git rev-parse -q --verify $commit:"$filter_subdir"
293                         then
294                                 rm -f "$GIT_INDEX_FILE"
295                         else
296                                 echo >&2 "$err"
297                                 false
298                         fi
299                 }
300         esac || die "Could not initialize the index"
301
302         GIT_COMMIT=$commit
303         export GIT_COMMIT
304         git cat-file commit "$commit" >../commit ||
305                 die "Cannot read commit $commit"
306
307         eval "$(set_ident <../commit)" ||
308                 die "setting author/committer failed for commit $commit"
309         eval "$filter_env" < /dev/null ||
310                 die "env filter failed: $filter_env"
311
312         if [ "$filter_tree" ]; then
313                 git checkout-index -f -u -a ||
314                         die "Could not checkout the index"
315                 # files that $commit removed are now still in the working tree;
316                 # remove them, else they would be added again
317                 git clean -d -q -f -x
318                 eval "$filter_tree" < /dev/null ||
319                         die "tree filter failed: $filter_tree"
320
321                 (
322                         git diff-index -r --name-only --ignore-submodules $commit &&
323                         git ls-files --others
324                 ) > "$tempdir"/tree-state || exit
325                 git update-index --add --replace --remove --stdin \
326                         < "$tempdir"/tree-state || exit
327         fi
328
329         eval "$filter_index" < /dev/null ||
330                 die "index filter failed: $filter_index"
331
332         parentstr=
333         for parent in $parents; do
334                 for reparent in $(map "$parent"); do
335                         case "$parentstr " in
336                         *" -p $reparent "*)
337                                 ;;
338                         *)
339                                 parentstr="$parentstr -p $reparent"
340                                 ;;
341                         esac
342                 done
343         done
344         if [ "$filter_parent" ]; then
345                 parentstr="$(echo "$parentstr" | eval "$filter_parent")" ||
346                                 die "parent filter failed: $filter_parent"
347         fi
348
349         {
350                 while read -r header_line && test -n "$header_line"
351                 do
352                         # skip header lines...
353                         :;
354                 done
355                 # and output the actual commit message
356                 cat
357         } <../commit |
358                 eval "$filter_msg" > ../message ||
359                         die "msg filter failed: $filter_msg"
360         workdir=$workdir @SHELL_PATH@ -c "$filter_commit" "git commit-tree" \
361                 $(git write-tree) $parentstr < ../message > ../map/$commit ||
362                         die "could not write rewritten commit"
363 done <../revs
364
365 # If we are filtering for paths, as in the case of a subdirectory
366 # filter, it is possible that a specified head is not in the set of
367 # rewritten commits, because it was pruned by the revision walker.
368 # Ancestor remapping fixes this by mapping these heads to the unique
369 # nearest ancestor that survived the pruning.
370
371 if test "$remap_to_ancestor" = t
372 then
373         while read ref
374         do
375                 sha1=$(git rev-parse "$ref"^0)
376                 test -f "$workdir"/../map/$sha1 && continue
377                 ancestor=$(git rev-list --simplify-merges -1 "$ref" "$@")
378                 test "$ancestor" && echo $(map $ancestor) >> "$workdir"/../map/$sha1
379         done < "$tempdir"/heads
380 fi
381
382 # Finally update the refs
383
384 _x40='[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]'
385 _x40="$_x40$_x40$_x40$_x40$_x40$_x40$_x40$_x40"
386 echo
387 while read ref
388 do
389         # avoid rewriting a ref twice
390         test -f "$orig_namespace$ref" && continue
391
392         sha1=$(git rev-parse "$ref"^0)
393         rewritten=$(map $sha1)
394
395         test $sha1 = "$rewritten" &&
396                 warn "WARNING: Ref '$ref' is unchanged" &&
397                 continue
398
399         case "$rewritten" in
400         '')
401                 echo "Ref '$ref' was deleted"
402                 git update-ref -m "filter-branch: delete" -d "$ref" $sha1 ||
403                         die "Could not delete $ref"
404         ;;
405         $_x40)
406                 echo "Ref '$ref' was rewritten"
407                 if ! git update-ref -m "filter-branch: rewrite" \
408                                         "$ref" $rewritten $sha1 2>/dev/null; then
409                         if test $(git cat-file -t "$ref") = tag; then
410                                 if test -z "$filter_tag_name"; then
411                                         warn "WARNING: You said to rewrite tagged commits, but not the corresponding tag."
412                                         warn "WARNING: Perhaps use '--tag-name-filter cat' to rewrite the tag."
413                                 fi
414                         else
415                                 die "Could not rewrite $ref"
416                         fi
417                 fi
418         ;;
419         *)
420                 # NEEDSWORK: possibly add -Werror, making this an error
421                 warn "WARNING: '$ref' was rewritten into multiple commits:"
422                 warn "$rewritten"
423                 warn "WARNING: Ref '$ref' points to the first one now."
424                 rewritten=$(echo "$rewritten" | head -n 1)
425                 git update-ref -m "filter-branch: rewrite to first" \
426                                 "$ref" $rewritten $sha1 ||
427                         die "Could not rewrite $ref"
428         ;;
429         esac
430         git update-ref -m "filter-branch: backup" "$orig_namespace$ref" $sha1 ||
431                  exit
432 done < "$tempdir"/heads
433
434 # TODO: This should possibly go, with the semantics that all positive given
435 #       refs are updated, and their original heads stored in refs/original/
436 # Filter tags
437
438 if [ "$filter_tag_name" ]; then
439         git for-each-ref --format='%(objectname) %(objecttype) %(refname)' refs/tags |
440         while read sha1 type ref; do
441                 ref="${ref#refs/tags/}"
442                 # XXX: Rewrite tagged trees as well?
443                 if [ "$type" != "commit" -a "$type" != "tag" ]; then
444                         continue;
445                 fi
446
447                 if [ "$type" = "tag" ]; then
448                         # Dereference to a commit
449                         sha1t="$sha1"
450                         sha1="$(git rev-parse -q "$sha1"^{commit})" || continue
451                 fi
452
453                 [ -f "../map/$sha1" ] || continue
454                 new_sha1="$(cat "../map/$sha1")"
455                 GIT_COMMIT="$sha1"
456                 export GIT_COMMIT
457                 new_ref="$(echo "$ref" | eval "$filter_tag_name")" ||
458                         die "tag name filter failed: $filter_tag_name"
459
460                 echo "$ref -> $new_ref ($sha1 -> $new_sha1)"
461
462                 if [ "$type" = "tag" ]; then
463                         new_sha1=$( ( printf 'object %s\ntype commit\ntag %s\n' \
464                                                 "$new_sha1" "$new_ref"
465                                 git cat-file tag "$ref" |
466                                 sed -n \
467                                     -e '1,/^$/{
468                                           /^object /d
469                                           /^type /d
470                                           /^tag /d
471                                         }' \
472                                     -e '/^-----BEGIN PGP SIGNATURE-----/q' \
473                                     -e 'p' ) |
474                                 git mktag) ||
475                                 die "Could not create new tag object for $ref"
476                         if git cat-file tag "$ref" | \
477                            sane_grep '^-----BEGIN PGP SIGNATURE-----' >/dev/null 2>&1
478                         then
479                                 warn "gpg signature stripped from tag object $sha1t"
480                         fi
481                 fi
482
483                 git update-ref "refs/tags/$new_ref" "$new_sha1" ||
484                         die "Could not write tag $new_ref"
485         done
486 fi
487
488 cd "$orig_dir"
489 rm -rf "$tempdir"
490
491 trap - 0
492
493 unset GIT_DIR GIT_WORK_TREE GIT_INDEX_FILE
494 test -z "$ORIG_GIT_DIR" || {
495         GIT_DIR="$ORIG_GIT_DIR" && export GIT_DIR
496 }
497 test -z "$ORIG_GIT_WORK_TREE" || {
498         GIT_WORK_TREE="$ORIG_GIT_WORK_TREE" &&
499         export GIT_WORK_TREE
500 }
501 test -z "$ORIG_GIT_INDEX_FILE" || {
502         GIT_INDEX_FILE="$ORIG_GIT_INDEX_FILE" &&
503         export GIT_INDEX_FILE
504 }
505
506 if [ "$(is_bare_repository)" = false ]; then
507         git read-tree -u -m HEAD || exit
508 fi
509
510 exit 0