]>
Commit | Line | Data |
---|---|---|
1 | git-filter-branch(1) | |
2 | ==================== | |
3 | ||
4 | NAME | |
5 | ---- | |
6 | git-filter-branch - Rewrite branches | |
7 | ||
8 | SYNOPSIS | |
9 | -------- | |
10 | [verse] | |
11 | 'git filter-branch' [--env-filter <command>] [--tree-filter <command>] | |
12 | [--index-filter <command>] [--parent-filter <command>] | |
13 | [--msg-filter <command>] [--commit-filter <command>] | |
14 | [--tag-name-filter <command>] [--subdirectory-filter <directory>] | |
15 | [--prune-empty] | |
16 | [--original <namespace>] [-d <directory>] [-f | --force] | |
17 | [--] [<rev-list options>...] | |
18 | ||
19 | DESCRIPTION | |
20 | ----------- | |
21 | Lets you rewrite git revision history by rewriting the branches mentioned | |
22 | in the <rev-list options>, applying custom filters on each revision. | |
23 | Those filters can modify each tree (e.g. removing a file or running | |
24 | a perl rewrite on all files) or information about each commit. | |
25 | Otherwise, all information (including original commit times or merge | |
26 | information) will be preserved. | |
27 | ||
28 | The command will only rewrite the _positive_ refs mentioned in the | |
29 | command line (e.g. if you pass 'a..b', only 'b' will be rewritten). | |
30 | If you specify no filters, the commits will be recommitted without any | |
31 | changes, which would normally have no effect. Nevertheless, this may be | |
32 | useful in the future for compensating for some git bugs or such, | |
33 | therefore such a usage is permitted. | |
34 | ||
35 | *NOTE*: This command honors `.git/info/grafts` file and refs in | |
36 | the `refs/replace/` namespace. | |
37 | If you have any grafts or replacement refs defined, running this command | |
38 | will make them permanent. | |
39 | ||
40 | *WARNING*! The rewritten history will have different object names for all | |
41 | the objects and will not converge with the original branch. You will not | |
42 | be able to easily push and distribute the rewritten branch on top of the | |
43 | original branch. Please do not use this command if you do not know the | |
44 | full implications, and avoid using it anyway, if a simple single commit | |
45 | would suffice to fix your problem. (See the "RECOVERING FROM UPSTREAM | |
46 | REBASE" section in linkgit:git-rebase[1] for further information about | |
47 | rewriting published history.) | |
48 | ||
49 | Always verify that the rewritten version is correct: The original refs, | |
50 | if different from the rewritten ones, will be stored in the namespace | |
51 | 'refs/original/'. | |
52 | ||
53 | Note that since this operation is very I/O expensive, it might | |
54 | be a good idea to redirect the temporary directory off-disk with the | |
55 | '-d' option, e.g. on tmpfs. Reportedly the speedup is very noticeable. | |
56 | ||
57 | ||
58 | Filters | |
59 | ~~~~~~~ | |
60 | ||
61 | The filters are applied in the order as listed below. The <command> | |
62 | argument is always evaluated in the shell context using the 'eval' command | |
63 | (with the notable exception of the commit filter, for technical reasons). | |
64 | Prior to that, the $GIT_COMMIT environment variable will be set to contain | |
65 | the id of the commit being rewritten. Also, GIT_AUTHOR_NAME, | |
66 | GIT_AUTHOR_EMAIL, GIT_AUTHOR_DATE, GIT_COMMITTER_NAME, GIT_COMMITTER_EMAIL, | |
67 | and GIT_COMMITTER_DATE are set according to the current commit. The values | |
68 | of these variables after the filters have run, are used for the new commit. | |
69 | If any evaluation of <command> returns a non-zero exit status, the whole | |
70 | operation will be aborted. | |
71 | ||
72 | A 'map' function is available that takes an "original sha1 id" argument | |
73 | and outputs a "rewritten sha1 id" if the commit has been already | |
74 | rewritten, and "original sha1 id" otherwise; the 'map' function can | |
75 | return several ids on separate lines if your commit filter emitted | |
76 | multiple commits. | |
77 | ||
78 | ||
79 | OPTIONS | |
80 | ------- | |
81 | ||
82 | --env-filter <command>:: | |
83 | This filter may be used if you only need to modify the environment | |
84 | in which the commit will be performed. Specifically, you might | |
85 | want to rewrite the author/committer name/email/time environment | |
86 | variables (see linkgit:git-commit-tree[1] for details). Do not forget | |
87 | to re-export the variables. | |
88 | ||
89 | --tree-filter <command>:: | |
90 | This is the filter for rewriting the tree and its contents. | |
91 | The argument is evaluated in shell with the working | |
92 | directory set to the root of the checked out tree. The new tree | |
93 | is then used as-is (new files are auto-added, disappeared files | |
94 | are auto-removed - neither .gitignore files nor any other ignore | |
95 | rules *HAVE ANY EFFECT*!). | |
96 | ||
97 | --index-filter <command>:: | |
98 | This is the filter for rewriting the index. It is similar to the | |
99 | tree filter but does not check out the tree, which makes it much | |
100 | faster. Frequently used with `git rm --cached | |
101 | --ignore-unmatch ...`, see EXAMPLES below. For hairy | |
102 | cases, see linkgit:git-update-index[1]. | |
103 | ||
104 | --parent-filter <command>:: | |
105 | This is the filter for rewriting the commit's parent list. | |
106 | It will receive the parent string on stdin and shall output | |
107 | the new parent string on stdout. The parent string is in | |
108 | the format described in linkgit:git-commit-tree[1]: empty for | |
109 | the initial commit, "-p parent" for a normal commit and | |
110 | "-p parent1 -p parent2 -p parent3 ..." for a merge commit. | |
111 | ||
112 | --msg-filter <command>:: | |
113 | This is the filter for rewriting the commit messages. | |
114 | The argument is evaluated in the shell with the original | |
115 | commit message on standard input; its standard output is | |
116 | used as the new commit message. | |
117 | ||
118 | --commit-filter <command>:: | |
119 | This is the filter for performing the commit. | |
120 | If this filter is specified, it will be called instead of the | |
121 | 'git commit-tree' command, with arguments of the form | |
122 | "<TREE_ID> [(-p <PARENT_COMMIT_ID>)...]" and the log message on | |
123 | stdin. The commit id is expected on stdout. | |
124 | + | |
125 | As a special extension, the commit filter may emit multiple | |
126 | commit ids; in that case, the rewritten children of the original commit will | |
127 | have all of them as parents. | |
128 | + | |
129 | You can use the 'map' convenience function in this filter, and other | |
130 | convenience functions, too. For example, calling 'skip_commit "$@"' | |
131 | will leave out the current commit (but not its changes! If you want | |
132 | that, use 'git rebase' instead). | |
133 | + | |
134 | You can also use the `git_commit_non_empty_tree "$@"` instead of | |
135 | `git commit-tree "$@"` if you don't wish to keep commits with a single parent | |
136 | and that makes no change to the tree. | |
137 | ||
138 | --tag-name-filter <command>:: | |
139 | This is the filter for rewriting tag names. When passed, | |
140 | it will be called for every tag ref that points to a rewritten | |
141 | object (or to a tag object which points to a rewritten object). | |
142 | The original tag name is passed via standard input, and the new | |
143 | tag name is expected on standard output. | |
144 | + | |
145 | The original tags are not deleted, but can be overwritten; | |
146 | use "--tag-name-filter cat" to simply update the tags. In this | |
147 | case, be very careful and make sure you have the old tags | |
148 | backed up in case the conversion has run afoul. | |
149 | + | |
150 | Nearly proper rewriting of tag objects is supported. If the tag has | |
151 | a message attached, a new tag object will be created with the same message, | |
152 | author, and timestamp. If the tag has a signature attached, the | |
153 | signature will be stripped. It is by definition impossible to preserve | |
154 | signatures. The reason this is "nearly" proper, is because ideally if | |
155 | the tag did not change (points to the same object, has the same name, etc.) | |
156 | it should retain any signature. That is not the case, signatures will always | |
157 | be removed, buyer beware. There is also no support for changing the | |
158 | author or timestamp (or the tag message for that matter). Tags which point | |
159 | to other tags will be rewritten to point to the underlying commit. | |
160 | ||
161 | --subdirectory-filter <directory>:: | |
162 | Only look at the history which touches the given subdirectory. | |
163 | The result will contain that directory (and only that) as its | |
164 | project root. Implies <<Remap_to_ancestor>>. | |
165 | ||
166 | --prune-empty:: | |
167 | Some kind of filters will generate empty commits, that left the tree | |
168 | untouched. This switch allow git-filter-branch to ignore such | |
169 | commits. Though, this switch only applies for commits that have one | |
170 | and only one parent, it will hence keep merges points. Also, this | |
171 | option is not compatible with the use of '--commit-filter'. Though you | |
172 | just need to use the function 'git_commit_non_empty_tree "$@"' instead | |
173 | of the `git commit-tree "$@"` idiom in your commit filter to make that | |
174 | happen. | |
175 | ||
176 | --original <namespace>:: | |
177 | Use this option to set the namespace where the original commits | |
178 | will be stored. The default value is 'refs/original'. | |
179 | ||
180 | -d <directory>:: | |
181 | Use this option to set the path to the temporary directory used for | |
182 | rewriting. When applying a tree filter, the command needs to | |
183 | temporarily check out the tree to some directory, which may consume | |
184 | considerable space in case of large projects. By default it | |
185 | does this in the '.git-rewrite/' directory but you can override | |
186 | that choice by this parameter. | |
187 | ||
188 | -f:: | |
189 | --force:: | |
190 | 'git filter-branch' refuses to start with an existing temporary | |
191 | directory or when there are already refs starting with | |
192 | 'refs/original/', unless forced. | |
193 | ||
194 | <rev-list options>...:: | |
195 | Arguments for 'git rev-list'. All positive refs included by | |
196 | these options are rewritten. You may also specify options | |
197 | such as '--all', but you must use '--' to separate them from | |
198 | the 'git filter-branch' options. Implies <<Remap_to_ancestor>>. | |
199 | ||
200 | ||
201 | [[Remap_to_ancestor]] | |
202 | Remap to ancestor | |
203 | ~~~~~~~~~~~~~~~~~ | |
204 | ||
205 | By using linkgit:rev-list[1] arguments, e.g., path limiters, you can limit the | |
206 | set of revisions which get rewritten. However, positive refs on the command | |
207 | line are distinguished: we don't let them be excluded by such limiters. For | |
208 | this purpose, they are instead rewritten to point at the nearest ancestor that | |
209 | was not excluded. | |
210 | ||
211 | ||
212 | Examples | |
213 | -------- | |
214 | ||
215 | Suppose you want to remove a file (containing confidential information | |
216 | or copyright violation) from all commits: | |
217 | ||
218 | ------------------------------------------------------- | |
219 | git filter-branch --tree-filter 'rm filename' HEAD | |
220 | ------------------------------------------------------- | |
221 | ||
222 | However, if the file is absent from the tree of some commit, | |
223 | a simple `rm filename` will fail for that tree and commit. | |
224 | Thus you may instead want to use `rm -f filename` as the script. | |
225 | ||
226 | Using `--index-filter` with 'git rm' yields a significantly faster | |
227 | version. Like with using `rm filename`, `git rm --cached filename` | |
228 | will fail if the file is absent from the tree of a commit. If you | |
229 | want to "completely forget" a file, it does not matter when it entered | |
230 | history, so we also add `--ignore-unmatch`: | |
231 | ||
232 | -------------------------------------------------------------------------- | |
233 | git filter-branch --index-filter 'git rm --cached --ignore-unmatch filename' HEAD | |
234 | -------------------------------------------------------------------------- | |
235 | ||
236 | Now, you will get the rewritten history saved in HEAD. | |
237 | ||
238 | To rewrite the repository to look as if `foodir/` had been its project | |
239 | root, and discard all other history: | |
240 | ||
241 | ------------------------------------------------------- | |
242 | git filter-branch --subdirectory-filter foodir -- --all | |
243 | ------------------------------------------------------- | |
244 | ||
245 | Thus you can, e.g., turn a library subdirectory into a repository of | |
246 | its own. Note the `--` that separates 'filter-branch' options from | |
247 | revision options, and the `--all` to rewrite all branches and tags. | |
248 | ||
249 | To set a commit (which typically is at the tip of another | |
250 | history) to be the parent of the current initial commit, in | |
251 | order to paste the other history behind the current history: | |
252 | ||
253 | ------------------------------------------------------------------- | |
254 | git filter-branch --parent-filter 'sed "s/^\$/-p <graft-id>/"' HEAD | |
255 | ------------------------------------------------------------------- | |
256 | ||
257 | (if the parent string is empty - which happens when we are dealing with | |
258 | the initial commit - add graftcommit as a parent). Note that this assumes | |
259 | history with a single root (that is, no merge without common ancestors | |
260 | happened). If this is not the case, use: | |
261 | ||
262 | -------------------------------------------------------------------------- | |
263 | git filter-branch --parent-filter \ | |
264 | 'test $GIT_COMMIT = <commit-id> && echo "-p <graft-id>" || cat' HEAD | |
265 | -------------------------------------------------------------------------- | |
266 | ||
267 | or even simpler: | |
268 | ||
269 | ----------------------------------------------- | |
270 | echo "$commit-id $graft-id" >> .git/info/grafts | |
271 | git filter-branch $graft-id..HEAD | |
272 | ----------------------------------------------- | |
273 | ||
274 | To remove commits authored by "Darl McBribe" from the history: | |
275 | ||
276 | ------------------------------------------------------------------------------ | |
277 | git filter-branch --commit-filter ' | |
278 | if [ "$GIT_AUTHOR_NAME" = "Darl McBribe" ]; | |
279 | then | |
280 | skip_commit "$@"; | |
281 | else | |
282 | git commit-tree "$@"; | |
283 | fi' HEAD | |
284 | ------------------------------------------------------------------------------ | |
285 | ||
286 | The function 'skip_commit' is defined as follows: | |
287 | ||
288 | -------------------------- | |
289 | skip_commit() | |
290 | { | |
291 | shift; | |
292 | while [ -n "$1" ]; | |
293 | do | |
294 | shift; | |
295 | map "$1"; | |
296 | shift; | |
297 | done; | |
298 | } | |
299 | -------------------------- | |
300 | ||
301 | The shift magic first throws away the tree id and then the -p | |
302 | parameters. Note that this handles merges properly! In case Darl | |
303 | committed a merge between P1 and P2, it will be propagated properly | |
304 | and all children of the merge will become merge commits with P1,P2 | |
305 | as their parents instead of the merge commit. | |
306 | ||
307 | *NOTE* the changes introduced by the commits, and which are not reverted | |
308 | by subsequent commits, will still be in the rewritten branch. If you want | |
309 | to throw out _changes_ together with the commits, you should use the | |
310 | interactive mode of 'git rebase'. | |
311 | ||
312 | You can rewrite the commit log messages using `--msg-filter`. For | |
313 | example, 'git svn-id' strings in a repository created by 'git svn' can | |
314 | be removed this way: | |
315 | ||
316 | ------------------------------------------------------- | |
317 | git filter-branch --msg-filter ' | |
318 | sed -e "/^git-svn-id:/d" | |
319 | ' | |
320 | ------------------------------------------------------- | |
321 | ||
322 | If you need to add 'Acked-by' lines to, say, the last 10 commits (none | |
323 | of which is a merge), use this command: | |
324 | ||
325 | -------------------------------------------------------- | |
326 | git filter-branch --msg-filter ' | |
327 | cat && | |
328 | echo "Acked-by: Bugs Bunny <bunny@bugzilla.org>" | |
329 | ' HEAD~10..HEAD | |
330 | -------------------------------------------------------- | |
331 | ||
332 | To restrict rewriting to only part of the history, specify a revision | |
333 | range in addition to the new branch name. The new branch name will | |
334 | point to the top-most revision that a 'git rev-list' of this range | |
335 | will print. | |
336 | ||
337 | Consider this history: | |
338 | ||
339 | ------------------ | |
340 | D--E--F--G--H | |
341 | / / | |
342 | A--B-----C | |
343 | ------------------ | |
344 | ||
345 | To rewrite only commits D,E,F,G,H, but leave A, B and C alone, use: | |
346 | ||
347 | -------------------------------- | |
348 | git filter-branch ... C..H | |
349 | -------------------------------- | |
350 | ||
351 | To rewrite commits E,F,G,H, use one of these: | |
352 | ||
353 | ---------------------------------------- | |
354 | git filter-branch ... C..H --not D | |
355 | git filter-branch ... D..H --not C | |
356 | ---------------------------------------- | |
357 | ||
358 | To move the whole tree into a subdirectory, or remove it from there: | |
359 | ||
360 | --------------------------------------------------------------- | |
361 | git filter-branch --index-filter \ | |
362 | 'git ls-files -s | sed "s-\t\"*-&newsubdir/-" | | |
363 | GIT_INDEX_FILE=$GIT_INDEX_FILE.new \ | |
364 | git update-index --index-info && | |
365 | mv "$GIT_INDEX_FILE.new" "$GIT_INDEX_FILE"' HEAD | |
366 | --------------------------------------------------------------- | |
367 | ||
368 | ||
369 | ||
370 | Checklist for Shrinking a Repository | |
371 | ------------------------------------ | |
372 | ||
373 | git-filter-branch is often used to get rid of a subset of files, | |
374 | usually with some combination of `--index-filter` and | |
375 | `--subdirectory-filter`. People expect the resulting repository to | |
376 | be smaller than the original, but you need a few more steps to | |
377 | actually make it smaller, because git tries hard not to lose your | |
378 | objects until you tell it to. First make sure that: | |
379 | ||
380 | * You really removed all variants of a filename, if a blob was moved | |
381 | over its lifetime. `git log --name-only --follow --all -- filename` | |
382 | can help you find renames. | |
383 | ||
384 | * You really filtered all refs: use `--tag-name-filter cat -- --all` | |
385 | when calling git-filter-branch. | |
386 | ||
387 | Then there are two ways to get a smaller repository. A safer way is | |
388 | to clone, that keeps your original intact. | |
389 | ||
390 | * Clone it with `git clone file:///path/to/repo`. The clone | |
391 | will not have the removed objects. See linkgit:git-clone[1]. (Note | |
392 | that cloning with a plain path just hardlinks everything!) | |
393 | ||
394 | If you really don't want to clone it, for whatever reasons, check the | |
395 | following points instead (in this order). This is a very destructive | |
396 | approach, so *make a backup* or go back to cloning it. You have been | |
397 | warned. | |
398 | ||
399 | * Remove the original refs backed up by git-filter-branch: say `git | |
400 | for-each-ref --format="%(refname)" refs/original/ | xargs -n 1 git | |
401 | update-ref -d`. | |
402 | ||
403 | * Expire all reflogs with `git reflog expire --expire=now --all`. | |
404 | ||
405 | * Garbage collect all unreferenced objects with `git gc --prune=now` | |
406 | (or if your git-gc is not new enough to support arguments to | |
407 | `--prune`, use `git repack -ad; git prune` instead). | |
408 | ||
409 | GIT | |
410 | --- | |
411 | Part of the linkgit:git[1] suite |