Merge branch 'er/doc-fast-import-done' into maint
[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 # When piped a commit, output a script to set the ident of either
68 # "author" or "committer
69
70 set_ident () {
71         lid="$(echo "$1" | tr "[A-Z]" "[a-z]")"
72         uid="$(echo "$1" | tr "[a-z]" "[A-Z]")"
73         pick_id_script='
74                 /^'$lid' /{
75                         s/'\''/'\''\\'\'\''/g
76                         h
77                         s/^'$lid' \([^<]*\) <[^>]*> .*$/\1/
78                         s/'\''/'\''\'\'\''/g
79                         s/.*/GIT_'$uid'_NAME='\''&'\''; export GIT_'$uid'_NAME/p
80
81                         g
82                         s/^'$lid' [^<]* <\([^>]*\)> .*$/\1/
83                         s/'\''/'\''\'\'\''/g
84                         s/.*/GIT_'$uid'_EMAIL='\''&'\''; export GIT_'$uid'_EMAIL/p
85
86                         g
87                         s/^'$lid' [^<]* <[^>]*> \(.*\)$/@\1/
88                         s/'\''/'\''\'\'\''/g
89                         s/.*/GIT_'$uid'_DATE='\''&'\''; export GIT_'$uid'_DATE/p
90
91                         q
92                 }
93         '
94
95         LANG=C LC_ALL=C sed -ne "$pick_id_script"
96         # Ensure non-empty id name.
97         echo "case \"\$GIT_${uid}_NAME\" in \"\") GIT_${uid}_NAME=\"\${GIT_${uid}_EMAIL%%@*}\" && export GIT_${uid}_NAME;; esac"
98 }
99
100 USAGE="[--env-filter <command>] [--tree-filter <command>]
101         [--index-filter <command>] [--parent-filter <command>]
102         [--msg-filter <command>] [--commit-filter <command>]
103         [--tag-name-filter <command>] [--subdirectory-filter <directory>]
104         [--original <namespace>] [-d <directory>] [-f | --force]
105         [<rev-list options>...]"
106
107 OPTIONS_SPEC=
108 . git-sh-setup
109
110 if [ "$(is_bare_repository)" = false ]; then
111         require_clean_work_tree 'rewrite branches'
112 fi
113
114 tempdir=.git-rewrite
115 filter_env=
116 filter_tree=
117 filter_index=
118 filter_parent=
119 filter_msg=cat
120 filter_commit=
121 filter_tag_name=
122 filter_subdir=
123 orig_namespace=refs/original/
124 force=
125 prune_empty=
126 remap_to_ancestor=
127 while :
128 do
129         case "$1" in
130         --)
131                 shift
132                 break
133                 ;;
134         --force|-f)
135                 shift
136                 force=t
137                 continue
138                 ;;
139         --remap-to-ancestor)
140                 # deprecated ($remap_to_ancestor is set now automatically)
141                 shift
142                 remap_to_ancestor=t
143                 continue
144                 ;;
145         --prune-empty)
146                 shift
147                 prune_empty=t
148                 continue
149                 ;;
150         -*)
151                 ;;
152         *)
153                 break;
154         esac
155
156         # all switches take one argument
157         ARG="$1"
158         case "$#" in 1) usage ;; esac
159         shift
160         OPTARG="$1"
161         shift
162
163         case "$ARG" in
164         -d)
165                 tempdir="$OPTARG"
166                 ;;
167         --env-filter)
168                 filter_env="$OPTARG"
169                 ;;
170         --tree-filter)
171                 filter_tree="$OPTARG"
172                 ;;
173         --index-filter)
174                 filter_index="$OPTARG"
175                 ;;
176         --parent-filter)
177                 filter_parent="$OPTARG"
178                 ;;
179         --msg-filter)
180                 filter_msg="$OPTARG"
181                 ;;
182         --commit-filter)
183                 filter_commit="$functions; $OPTARG"
184                 ;;
185         --tag-name-filter)
186                 filter_tag_name="$OPTARG"
187                 ;;
188         --subdirectory-filter)
189                 filter_subdir="$OPTARG"
190                 remap_to_ancestor=t
191                 ;;
192         --original)
193                 orig_namespace=$(expr "$OPTARG/" : '\(.*[^/]\)/*$')/
194                 ;;
195         *)
196                 usage
197                 ;;
198         esac
199 done
200
201 case "$prune_empty,$filter_commit" in
202 ,)
203         filter_commit='git commit-tree "$@"';;
204 t,)
205         filter_commit="$functions;"' git_commit_non_empty_tree "$@"';;
206 ,*)
207         ;;
208 *)
209         die "Cannot set --prune-empty and --commit-filter at the same time"
210 esac
211
212 case "$force" in
213 t)
214         rm -rf "$tempdir"
215 ;;
216 '')
217         test -d "$tempdir" &&
218                 die "$tempdir already exists, please remove it"
219 esac
220 mkdir -p "$tempdir/t" &&
221 tempdir="$(cd "$tempdir"; pwd)" &&
222 cd "$tempdir/t" &&
223 workdir="$(pwd)" ||
224 die ""
225
226 # Remove tempdir on exit
227 trap 'cd ../..; rm -rf "$tempdir"' 0
228
229 ORIG_GIT_DIR="$GIT_DIR"
230 ORIG_GIT_WORK_TREE="$GIT_WORK_TREE"
231 ORIG_GIT_INDEX_FILE="$GIT_INDEX_FILE"
232 GIT_WORK_TREE=.
233 export GIT_DIR GIT_WORK_TREE
234
235 # Make sure refs/original is empty
236 git for-each-ref > "$tempdir"/backup-refs || exit
237 while read sha1 type name
238 do
239         case "$force,$name" in
240         ,$orig_namespace*)
241                 die "Cannot create a new backup.
242 A previous backup already exists in $orig_namespace
243 Force overwriting the backup with -f"
244         ;;
245         t,$orig_namespace*)
246                 git update-ref -d "$name" $sha1
247         ;;
248         esac
249 done < "$tempdir"/backup-refs
250
251 # The refs should be updated if their heads were rewritten
252 git rev-parse --no-flags --revs-only --symbolic-full-name \
253         --default HEAD "$@" > "$tempdir"/raw-heads || exit
254 sed -e '/^^/d' "$tempdir"/raw-heads >"$tempdir"/heads
255
256 test -s "$tempdir"/heads ||
257         die "Which ref do you want to rewrite?"
258
259 GIT_INDEX_FILE="$(pwd)/../index"
260 export GIT_INDEX_FILE
261
262 # map old->new commit ids for rewriting parents
263 mkdir ../map || die "Could not create map/ directory"
264
265 # we need "--" only if there are no path arguments in $@
266 nonrevs=$(git rev-parse --no-revs "$@") || exit
267 if test -z "$nonrevs"
268 then
269         dashdash=--
270 else
271         dashdash=
272         remap_to_ancestor=t
273 fi
274
275 rev_args=$(git rev-parse --revs-only "$@")
276
277 case "$filter_subdir" in
278 "")
279         eval set -- "$(git rev-parse --sq --no-revs "$@")"
280         ;;
281 *)
282         eval set -- "$(git rev-parse --sq --no-revs "$@" $dashdash \
283                 "$filter_subdir")"
284         ;;
285 esac
286
287 git rev-list --reverse --topo-order --default HEAD \
288         --parents --simplify-merges $rev_args "$@" > ../revs ||
289         die "Could not get the commits"
290 commits=$(wc -l <../revs | tr -d " ")
291
292 test $commits -eq 0 && die "Found nothing to rewrite"
293
294 # Rewrite the commits
295
296 git_filter_branch__commit_count=0
297 while read commit parents; do
298         git_filter_branch__commit_count=$(($git_filter_branch__commit_count+1))
299         printf "\rRewrite $commit ($git_filter_branch__commit_count/$commits)"
300
301         case "$filter_subdir" in
302         "")
303                 git read-tree -i -m $commit
304                 ;;
305         *)
306                 # The commit may not have the subdirectory at all
307                 err=$(git read-tree -i -m $commit:"$filter_subdir" 2>&1) || {
308                         if ! git rev-parse -q --verify $commit:"$filter_subdir"
309                         then
310                                 rm -f "$GIT_INDEX_FILE"
311                         else
312                                 echo >&2 "$err"
313                                 false
314                         fi
315                 }
316         esac || die "Could not initialize the index"
317
318         GIT_COMMIT=$commit
319         export GIT_COMMIT
320         git cat-file commit "$commit" >../commit ||
321                 die "Cannot read commit $commit"
322
323         eval "$(set_ident AUTHOR <../commit)" ||
324                 die "setting author failed for commit $commit"
325         eval "$(set_ident COMMITTER <../commit)" ||
326                 die "setting committer failed for commit $commit"
327         eval "$filter_env" < /dev/null ||
328                 die "env filter failed: $filter_env"
329
330         if [ "$filter_tree" ]; then
331                 git checkout-index -f -u -a ||
332                         die "Could not checkout the index"
333                 # files that $commit removed are now still in the working tree;
334                 # remove them, else they would be added again
335                 git clean -d -q -f -x
336                 eval "$filter_tree" < /dev/null ||
337                         die "tree filter failed: $filter_tree"
338
339                 (
340                         git diff-index -r --name-only --ignore-submodules $commit &&
341                         git ls-files --others
342                 ) > "$tempdir"/tree-state || exit
343                 git update-index --add --replace --remove --stdin \
344                         < "$tempdir"/tree-state || exit
345         fi
346
347         eval "$filter_index" < /dev/null ||
348                 die "index filter failed: $filter_index"
349
350         parentstr=
351         for parent in $parents; do
352                 for reparent in $(map "$parent"); do
353                         parentstr="$parentstr -p $reparent"
354                 done
355         done
356         if [ "$filter_parent" ]; then
357                 parentstr="$(echo "$parentstr" | eval "$filter_parent")" ||
358                                 die "parent filter failed: $filter_parent"
359         fi
360
361         sed -e '1,/^$/d' <../commit | \
362                 eval "$filter_msg" > ../message ||
363                         die "msg filter failed: $filter_msg"
364         workdir=$workdir @SHELL_PATH@ -c "$filter_commit" "git commit-tree" \
365                 $(git write-tree) $parentstr < ../message > ../map/$commit ||
366                         die "could not write rewritten commit"
367 done <../revs
368
369 # If we are filtering for paths, as in the case of a subdirectory
370 # filter, it is possible that a specified head is not in the set of
371 # rewritten commits, because it was pruned by the revision walker.
372 # Ancestor remapping fixes this by mapping these heads to the unique
373 # nearest ancestor that survived the pruning.
374
375 if test "$remap_to_ancestor" = t
376 then
377         while read ref
378         do
379                 sha1=$(git rev-parse "$ref"^0)
380                 test -f "$workdir"/../map/$sha1 && continue
381                 ancestor=$(git rev-list --simplify-merges -1 "$ref" "$@")
382                 test "$ancestor" && echo $(map $ancestor) >> "$workdir"/../map/$sha1
383         done < "$tempdir"/heads
384 fi
385
386 # Finally update the refs
387
388 _x40='[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]'
389 _x40="$_x40$_x40$_x40$_x40$_x40$_x40$_x40$_x40"
390 echo
391 while read ref
392 do
393         # avoid rewriting a ref twice
394         test -f "$orig_namespace$ref" && continue
395
396         sha1=$(git rev-parse "$ref"^0)
397         rewritten=$(map $sha1)
398
399         test $sha1 = "$rewritten" &&
400                 warn "WARNING: Ref '$ref' is unchanged" &&
401                 continue
402
403         case "$rewritten" in
404         '')
405                 echo "Ref '$ref' was deleted"
406                 git update-ref -m "filter-branch: delete" -d "$ref" $sha1 ||
407                         die "Could not delete $ref"
408         ;;
409         $_x40)
410                 echo "Ref '$ref' was rewritten"
411                 if ! git update-ref -m "filter-branch: rewrite" \
412                                         "$ref" $rewritten $sha1 2>/dev/null; then
413                         if test $(git cat-file -t "$ref") = tag; then
414                                 if test -z "$filter_tag_name"; then
415                                         warn "WARNING: You said to rewrite tagged commits, but not the corresponding tag."
416                                         warn "WARNING: Perhaps use '--tag-name-filter cat' to rewrite the tag."
417                                 fi
418                         else
419                                 die "Could not rewrite $ref"
420                         fi
421                 fi
422         ;;
423         *)
424                 # NEEDSWORK: possibly add -Werror, making this an error
425                 warn "WARNING: '$ref' was rewritten into multiple commits:"
426                 warn "$rewritten"
427                 warn "WARNING: Ref '$ref' points to the first one now."
428                 rewritten=$(echo "$rewritten" | head -n 1)
429                 git update-ref -m "filter-branch: rewrite to first" \
430                                 "$ref" $rewritten $sha1 ||
431                         die "Could not rewrite $ref"
432         ;;
433         esac
434         git update-ref -m "filter-branch: backup" "$orig_namespace$ref" $sha1 ||
435                  exit
436 done < "$tempdir"/heads
437
438 # TODO: This should possibly go, with the semantics that all positive given
439 #       refs are updated, and their original heads stored in refs/original/
440 # Filter tags
441
442 if [ "$filter_tag_name" ]; then
443         git for-each-ref --format='%(objectname) %(objecttype) %(refname)' refs/tags |
444         while read sha1 type ref; do
445                 ref="${ref#refs/tags/}"
446                 # XXX: Rewrite tagged trees as well?
447                 if [ "$type" != "commit" -a "$type" != "tag" ]; then
448                         continue;
449                 fi
450
451                 if [ "$type" = "tag" ]; then
452                         # Dereference to a commit
453                         sha1t="$sha1"
454                         sha1="$(git rev-parse -q "$sha1"^{commit})" || continue
455                 fi
456
457                 [ -f "../map/$sha1" ] || continue
458                 new_sha1="$(cat "../map/$sha1")"
459                 GIT_COMMIT="$sha1"
460                 export GIT_COMMIT
461                 new_ref="$(echo "$ref" | eval "$filter_tag_name")" ||
462                         die "tag name filter failed: $filter_tag_name"
463
464                 echo "$ref -> $new_ref ($sha1 -> $new_sha1)"
465
466                 if [ "$type" = "tag" ]; then
467                         new_sha1=$( ( printf 'object %s\ntype commit\ntag %s\n' \
468                                                 "$new_sha1" "$new_ref"
469                                 git cat-file tag "$ref" |
470                                 sed -n \
471                                     -e '1,/^$/{
472                                           /^object /d
473                                           /^type /d
474                                           /^tag /d
475                                         }' \
476                                     -e '/^-----BEGIN PGP SIGNATURE-----/q' \
477                                     -e 'p' ) |
478                                 git mktag) ||
479                                 die "Could not create new tag object for $ref"
480                         if git cat-file tag "$ref" | \
481                            sane_grep '^-----BEGIN PGP SIGNATURE-----' >/dev/null 2>&1
482                         then
483                                 warn "gpg signature stripped from tag object $sha1t"
484                         fi
485                 fi
486
487                 git update-ref "refs/tags/$new_ref" "$new_sha1" ||
488                         die "Could not write tag $new_ref"
489         done
490 fi
491
492 cd ../..
493 rm -rf "$tempdir"
494
495 trap - 0
496
497 unset GIT_DIR GIT_WORK_TREE GIT_INDEX_FILE
498 test -z "$ORIG_GIT_DIR" || {
499         GIT_DIR="$ORIG_GIT_DIR" && export GIT_DIR
500 }
501 test -z "$ORIG_GIT_WORK_TREE" || {
502         GIT_WORK_TREE="$ORIG_GIT_WORK_TREE" &&
503         export GIT_WORK_TREE
504 }
505 test -z "$ORIG_GIT_INDEX_FILE" || {
506         GIT_INDEX_FILE="$ORIG_GIT_INDEX_FILE" &&
507         export GIT_INDEX_FILE
508 }
509
510 if [ "$(is_bare_repository)" = false ]; then
511         git read-tree -u -m HEAD || exit
512 fi
513
514 exit 0