Merge branch 'ab/gc-auto-in-commit'
[git/git.git] / Documentation / gitattributes.txt
1 gitattributes(5)
2 ================
5 ----
6 gitattributes - defining attributes per path
9 --------
10 $GIT_DIR/info/attributes, .gitattributes
14 -----------
16 A `gitattributes` file is a simple text file that gives
17 `attributes` to pathnames.
19 Each line in `gitattributes` file is of form:
21 pattern attr1 attr2 ...
23 That is, a pattern followed by an attributes list,
24 separated by whitespaces. Leading and trailing whitespaces are
25 ignored. Lines that begin with '#' are ignored. Patterns
26 that begin with a double quote are quoted in C style.
27 When the pattern matches the path in question, the attributes
28 listed on the line are given to the path.
30 Each attribute can be in one of these states for a given path:
32 Set::
34 The path has the attribute with special value "true";
35 this is specified by listing only the name of the
36 attribute in the attribute list.
38 Unset::
40 The path has the attribute with special value "false";
41 this is specified by listing the name of the attribute
42 prefixed with a dash `-` in the attribute list.
44 Set to a value::
46 The path has the attribute with specified string value;
47 this is specified by listing the name of the attribute
48 followed by an equal sign `=` and its value in the
49 attribute list.
51 Unspecified::
53 No pattern matches the path, and nothing says if
54 the path has or does not have the attribute, the
55 attribute for the path is said to be Unspecified.
57 When more than one pattern matches the path, a later line
58 overrides an earlier line. This overriding is done per
59 attribute. The rules how the pattern matches paths are the
60 same as in `.gitignore` files; see linkgit:gitignore[5].
61 Unlike `.gitignore`, negative patterns are forbidden.
63 When deciding what attributes are assigned to a path, Git
64 consults `$GIT_DIR/info/attributes` file (which has the highest
65 precedence), `.gitattributes` file in the same directory as the
66 path in question, and its parent directories up to the toplevel of the
67 work tree (the further the directory that contains `.gitattributes`
68 is from the path in question, the lower its precedence). Finally
69 global and system-wide files are considered (they have the lowest
70 precedence).
72 When the `.gitattributes` file is missing from the work tree, the
73 path in the index is used as a fall-back. During checkout process,
74 `.gitattributes` in the index is used and then the file in the
75 working tree is used as a fall-back.
77 If you wish to affect only a single repository (i.e., to assign
78 attributes to files that are particular to
79 one user's workflow for that repository), then
80 attributes should be placed in the `$GIT_DIR/info/attributes` file.
81 Attributes which should be version-controlled and distributed to other
82 repositories (i.e., attributes of interest to all users) should go into
83 `.gitattributes` files. Attributes that should affect all repositories
84 for a single user should be placed in a file specified by the
85 `core.attributesFile` configuration option (see linkgit:git-config[1]).
86 Its default value is $XDG_CONFIG_HOME/git/attributes. If $XDG_CONFIG_HOME
87 is either not set or empty, $HOME/.config/git/attributes is used instead.
88 Attributes for all users on a system should be placed in the
89 `$(prefix)/etc/gitattributes` file.
91 Sometimes you would need to override a setting of an attribute
92 for a path to `Unspecified` state. This can be done by listing
93 the name of the attribute prefixed with an exclamation point `!`.
97 -------
99 Certain operations by Git can be influenced by assigning
100 particular attributes to a path. Currently, the following
101 operations are attributes-aware.
103 Checking-out and checking-in
104 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
106 These attributes affect how the contents stored in the
107 repository are copied to the working tree files when commands
108 such as 'git checkout' and 'git merge' run. They also affect how
109 Git stores the contents you prepare in the working tree in the
110 repository upon 'git add' and 'git commit'.
112 `text`
113 ^^^^^^
115 This attribute enables and controls end-of-line normalization. When a
116 text file is normalized, its line endings are converted to LF in the
117 repository. To control what line ending style is used in the working
118 directory, use the `eol` attribute for a single file and the
119 `core.eol` configuration variable for all text files.
120 Note that `core.autocrlf` overrides `core.eol`
122 Set::
124 Setting the `text` attribute on a path enables end-of-line
125 normalization and marks the path as a text file. End-of-line
126 conversion takes place without guessing the content type.
128 Unset::
130 Unsetting the `text` attribute on a path tells Git not to
131 attempt any end-of-line conversion upon checkin or checkout.
133 Set to string value "auto"::
135 When `text` is set to "auto", the path is marked for automatic
136 end-of-line conversion. If Git decides that the content is
137 text, its line endings are converted to LF on checkin.
138 When the file has been committed with CRLF, no conversion is done.
140 Unspecified::
142 If the `text` attribute is unspecified, Git uses the
143 `core.autocrlf` configuration variable to determine if the
144 file should be converted.
146 Any other value causes Git to act as if `text` has been left
147 unspecified.
149 `eol`
150 ^^^^^
152 This attribute sets a specific line-ending style to be used in the
153 working directory. It enables end-of-line conversion without any
154 content checks, effectively setting the `text` attribute. Note that
155 setting this attribute on paths which are in the index with CRLF line
156 endings may make the paths to be considered dirty. Adding the path to
157 the index again will normalize the line endings in the index.
159 Set to string value "crlf"::
161 This setting forces Git to normalize line endings for this
162 file on checkin and convert them to CRLF when the file is
163 checked out.
165 Set to string value "lf"::
167 This setting forces Git to normalize line endings to LF on
168 checkin and prevents conversion to CRLF when the file is
169 checked out.
171 Backwards compatibility with `crlf` attribute
172 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
174 For backwards compatibility, the `crlf` attribute is interpreted as
175 follows:
177 ------------------------
178 crlf text
179 -crlf -text
180 crlf=input eol=lf
181 ------------------------
183 End-of-line conversion
184 ^^^^^^^^^^^^^^^^^^^^^^
186 While Git normally leaves file contents alone, it can be configured to
187 normalize line endings to LF in the repository and, optionally, to
188 convert them to CRLF when files are checked out.
190 If you simply want to have CRLF line endings in your working directory
191 regardless of the repository you are working with, you can set the
192 config variable "core.autocrlf" without using any attributes.
194 ------------------------
195 [core]
196 autocrlf = true
197 ------------------------
199 This does not force normalization of text files, but does ensure
200 that text files that you introduce to the repository have their line
201 endings normalized to LF when they are added, and that files that are
202 already normalized in the repository stay normalized.
204 If you want to ensure that text files that any contributor introduces to
205 the repository have their line endings normalized, you can set the
206 `text` attribute to "auto" for _all_ files.
208 ------------------------
209 * text=auto
210 ------------------------
212 The attributes allow a fine-grained control, how the line endings
213 are converted.
214 Here is an example that will make Git normalize .txt, .vcproj and .sh
215 files, ensure that .vcproj files have CRLF and .sh files have LF in
216 the working directory, and prevent .jpg files from being normalized
217 regardless of their content.
219 ------------------------
220 * text=auto
221 *.txt text
222 *.vcproj text eol=crlf
223 *.sh text eol=lf
224 *.jpg -text
225 ------------------------
227 NOTE: When `text=auto` conversion is enabled in a cross-platform
228 project using push and pull to a central repository the text files
229 containing CRLFs should be normalized.
231 From a clean working directory:
233 -------------------------------------------------
234 $ echo "* text=auto" >.gitattributes
235 $ git add --renormalize .
236 $ git status # Show files that will be normalized
237 $ git commit -m "Introduce end-of-line normalization"
238 -------------------------------------------------
240 If any files that should not be normalized show up in 'git status',
241 unset their `text` attribute before running 'git add -u'.
243 ------------------------
244 manual.pdf -text
245 ------------------------
247 Conversely, text files that Git does not detect can have normalization
248 enabled manually.
250 ------------------------
251 weirdchars.txt text
252 ------------------------
254 If `core.safecrlf` is set to "true" or "warn", Git verifies if
255 the conversion is reversible for the current setting of
256 `core.autocrlf`. For "true", Git rejects irreversible
257 conversions; for "warn", Git only prints a warning but accepts
258 an irreversible conversion. The safety triggers to prevent such
259 a conversion done to the files in the work tree, but there are a
260 few exceptions. Even though...
262 - 'git add' itself does not touch the files in the work tree, the
263 next checkout would, so the safety triggers;
265 - 'git apply' to update a text file with a patch does touch the files
266 in the work tree, but the operation is about text files and CRLF
267 conversion is about fixing the line ending inconsistencies, so the
268 safety does not trigger;
270 - 'git diff' itself does not touch the files in the work tree, it is
271 often run to inspect the changes you intend to next 'git add'. To
272 catch potential problems early, safety triggers.
275 `ident`
276 ^^^^^^^
278 When the attribute `ident` is set for a path, Git replaces
279 `$Id$` in the blob object with `$Id:`, followed by the
280 40-character hexadecimal blob object name, followed by a dollar
281 sign `$` upon checkout. Any byte sequence that begins with
282 `$Id:` and ends with `$` in the worktree file is replaced
283 with `$Id$` upon check-in.
286 `filter`
287 ^^^^^^^^
289 A `filter` attribute can be set to a string value that names a
290 filter driver specified in the configuration.
292 A filter driver consists of a `clean` command and a `smudge`
293 command, either of which can be left unspecified. Upon
294 checkout, when the `smudge` command is specified, the command is
295 fed the blob object from its standard input, and its standard
296 output is used to update the worktree file. Similarly, the
297 `clean` command is used to convert the contents of worktree file
298 upon checkin. By default these commands process only a single
299 blob and terminate. If a long running `process` filter is used
300 in place of `clean` and/or `smudge` filters, then Git can process
301 all blobs with a single filter command invocation for the entire
302 life of a single Git command, for example `git add --all`. If a
303 long running `process` filter is configured then it always takes
304 precedence over a configured single blob filter. See section
305 below for the description of the protocol used to communicate with
306 a `process` filter.
308 One use of the content filtering is to massage the content into a shape
309 that is more convenient for the platform, filesystem, and the user to use.
310 For this mode of operation, the key phrase here is "more convenient" and
311 not "turning something unusable into usable". In other words, the intent
312 is that if someone unsets the filter driver definition, or does not have
313 the appropriate filter program, the project should still be usable.
315 Another use of the content filtering is to store the content that cannot
316 be directly used in the repository (e.g. a UUID that refers to the true
317 content stored outside Git, or an encrypted content) and turn it into a
318 usable form upon checkout (e.g. download the external content, or decrypt
319 the encrypted content).
321 These two filters behave differently, and by default, a filter is taken as
322 the former, massaging the contents into more convenient shape. A missing
323 filter driver definition in the config, or a filter driver that exits with
324 a non-zero status, is not an error but makes the filter a no-op passthru.
326 You can declare that a filter turns a content that by itself is unusable
327 into a usable content by setting the filter.<driver>.required configuration
328 variable to `true`.
330 Note: Whenever the clean filter is changed, the repo should be renormalized:
331 $ git add --renormalize .
333 For example, in .gitattributes, you would assign the `filter`
334 attribute for paths.
336 ------------------------
337 *.c filter=indent
338 ------------------------
340 Then you would define a "filter.indent.clean" and "filter.indent.smudge"
341 configuration in your .git/config to specify a pair of commands to
342 modify the contents of C programs when the source files are checked
343 in ("clean" is run) and checked out (no change is made because the
344 command is "cat").
346 ------------------------
347 [filter "indent"]
348 clean = indent
349 smudge = cat
350 ------------------------
352 For best results, `clean` should not alter its output further if it is
353 run twice ("clean->clean" should be equivalent to "clean"), and
354 multiple `smudge` commands should not alter `clean`'s output
355 ("smudge->smudge->clean" should be equivalent to "clean"). See the
356 section on merging below.
358 The "indent" filter is well-behaved in this regard: it will not modify
359 input that is already correctly indented. In this case, the lack of a
360 smudge filter means that the clean filter _must_ accept its own output
361 without modifying it.
363 If a filter _must_ succeed in order to make the stored contents usable,
364 you can declare that the filter is `required`, in the configuration:
366 ------------------------
367 [filter "crypt"]
368 clean = openssl enc ...
369 smudge = openssl enc -d ...
370 required
371 ------------------------
373 Sequence "%f" on the filter command line is replaced with the name of
374 the file the filter is working on. A filter might use this in keyword
375 substitution. For example:
377 ------------------------
378 [filter "p4"]
379 clean = git-p4-filter --clean %f
380 smudge = git-p4-filter --smudge %f
381 ------------------------
383 Note that "%f" is the name of the path that is being worked on. Depending
384 on the version that is being filtered, the corresponding file on disk may
385 not exist, or may have different contents. So, smudge and clean commands
386 should not try to access the file on disk, but only act as filters on the
387 content provided to them on standard input.
389 Long Running Filter Process
390 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
392 If the filter command (a string value) is defined via
393 `filter.<driver>.process` then Git can process all blobs with a
394 single filter invocation for the entire life of a single Git
395 command. This is achieved by using the long-running process protocol
396 (described in technical/long-running-process-protocol.txt).
398 When Git encounters the first file that needs to be cleaned or smudged,
399 it starts the filter and performs the handshake. In the handshake, the
400 welcome message sent by Git is "git-filter-client", only version 2 is
401 suppported, and the supported capabilities are "clean", "smudge", and
402 "delay".
404 Afterwards Git sends a list of "key=value" pairs terminated with
405 a flush packet. The list will contain at least the filter command
406 (based on the supported capabilities) and the pathname of the file
407 to filter relative to the repository root. Right after the flush packet
408 Git sends the content split in zero or more pkt-line packets and a
409 flush packet to terminate content. Please note, that the filter
410 must not send any response before it received the content and the
411 final flush packet. Also note that the "value" of a "key=value" pair
412 can contain the "=" character whereas the key would never contain
413 that character.
414 ------------------------
415 packet: git> command=smudge
416 packet: git> pathname=path/testfile.dat
417 packet: git> 0000
418 packet: git> CONTENT
419 packet: git> 0000
420 ------------------------
422 The filter is expected to respond with a list of "key=value" pairs
423 terminated with a flush packet. If the filter does not experience
424 problems then the list must contain a "success" status. Right after
425 these packets the filter is expected to send the content in zero
426 or more pkt-line packets and a flush packet at the end. Finally, a
427 second list of "key=value" pairs terminated with a flush packet
428 is expected. The filter can change the status in the second list
429 or keep the status as is with an empty list. Please note that the
430 empty list must be terminated with a flush packet regardless.
432 ------------------------
433 packet: git< status=success
434 packet: git< 0000
435 packet: git< SMUDGED_CONTENT
436 packet: git< 0000
437 packet: git< 0000 # empty list, keep "status=success" unchanged!
438 ------------------------
440 If the result content is empty then the filter is expected to respond
441 with a "success" status and a flush packet to signal the empty content.
442 ------------------------
443 packet: git< status=success
444 packet: git< 0000
445 packet: git< 0000 # empty content!
446 packet: git< 0000 # empty list, keep "status=success" unchanged!
447 ------------------------
449 In case the filter cannot or does not want to process the content,
450 it is expected to respond with an "error" status.
451 ------------------------
452 packet: git< status=error
453 packet: git< 0000
454 ------------------------
456 If the filter experiences an error during processing, then it can
457 send the status "error" after the content was (partially or
458 completely) sent.
459 ------------------------
460 packet: git< status=success
461 packet: git< 0000
463 packet: git< 0000
464 packet: git< status=error
465 packet: git< 0000
466 ------------------------
468 In case the filter cannot or does not want to process the content
469 as well as any future content for the lifetime of the Git process,
470 then it is expected to respond with an "abort" status at any point
471 in the protocol.
472 ------------------------
473 packet: git< status=abort
474 packet: git< 0000
475 ------------------------
477 Git neither stops nor restarts the filter process in case the
478 "error"/"abort" status is set. However, Git sets its exit code
479 according to the `filter.<driver>.required` flag, mimicking the
480 behavior of the `filter.<driver>.clean` / `filter.<driver>.smudge`
481 mechanism.
483 If the filter dies during the communication or does not adhere to
484 the protocol then Git will stop the filter process and restart it
485 with the next file that needs to be processed. Depending on the
486 `filter.<driver>.required` flag Git will interpret that as error.
488 Delay
489 ^^^^^
491 If the filter supports the "delay" capability, then Git can send the
492 flag "can-delay" after the filter command and pathname. This flag
493 denotes that the filter can delay filtering the current blob (e.g. to
494 compensate network latencies) by responding with no content but with
495 the status "delayed" and a flush packet.
496 ------------------------
497 packet: git> command=smudge
498 packet: git> pathname=path/testfile.dat
499 packet: git> can-delay=1
500 packet: git> 0000
501 packet: git> CONTENT
502 packet: git> 0000
503 packet: git< status=delayed
504 packet: git< 0000
505 ------------------------
507 If the filter supports the "delay" capability then it must support the
508 "list_available_blobs" command. If Git sends this command, then the
509 filter is expected to return a list of pathnames representing blobs
510 that have been delayed earlier and are now available.
511 The list must be terminated with a flush packet followed
512 by a "success" status that is also terminated with a flush packet. If
513 no blobs for the delayed paths are available, yet, then the filter is
514 expected to block the response until at least one blob becomes
515 available. The filter can tell Git that it has no more delayed blobs
516 by sending an empty list. As soon as the filter responds with an empty
517 list, Git stops asking. All blobs that Git has not received at this
518 point are considered missing and will result in an error.
520 ------------------------
521 packet: git> command=list_available_blobs
522 packet: git> 0000
523 packet: git< pathname=path/testfile.dat
524 packet: git< pathname=path/otherfile.dat
525 packet: git< 0000
526 packet: git< status=success
527 packet: git< 0000
528 ------------------------
530 After Git received the pathnames, it will request the corresponding
531 blobs again. These requests contain a pathname and an empty content
532 section. The filter is expected to respond with the smudged content
533 in the usual way as explained above.
534 ------------------------
535 packet: git> command=smudge
536 packet: git> pathname=path/testfile.dat
537 packet: git> 0000
538 packet: git> 0000 # empty content!
539 packet: git< status=success
540 packet: git< 0000
541 packet: git< SMUDGED_CONTENT
542 packet: git< 0000
543 packet: git< 0000 # empty list, keep "status=success" unchanged!
544 ------------------------
546 Example
547 ^^^^^^^
549 A long running filter demo implementation can be found in
550 `contrib/long-running-filter/` located in the Git
551 core repository. If you develop your own long running filter
552 process then the `GIT_TRACE_PACKET` environment variables can be
553 very helpful for debugging (see linkgit:git[1]).
555 Please note that you cannot use an existing `filter.<driver>.clean`
556 or `filter.<driver>.smudge` command with `filter.<driver>.process`
557 because the former two use a different inter process communication
558 protocol than the latter one.
561 Interaction between checkin/checkout attributes
562 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
564 In the check-in codepath, the worktree file is first converted
565 with `filter` driver (if specified and corresponding driver
566 defined), then the result is processed with `ident` (if
567 specified), and then finally with `text` (again, if specified
568 and applicable).
570 In the check-out codepath, the blob content is first converted
571 with `text`, and then `ident` and fed to `filter`.
574 Merging branches with differing checkin/checkout attributes
575 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
577 If you have added attributes to a file that cause the canonical
578 repository format for that file to change, such as adding a
579 clean/smudge filter or text/eol/ident attributes, merging anything
580 where the attribute is not in place would normally cause merge
581 conflicts.
583 To prevent these unnecessary merge conflicts, Git can be told to run a
584 virtual check-out and check-in of all three stages of a file when
585 resolving a three-way merge by setting the `merge.renormalize`
586 configuration variable. This prevents changes caused by check-in
587 conversion from causing spurious merge conflicts when a converted file
588 is merged with an unconverted file.
590 As long as a "smudge->clean" results in the same output as a "clean"
591 even on files that are already smudged, this strategy will
592 automatically resolve all filter-related conflicts. Filters that do
593 not act in this way may cause additional merge conflicts that must be
594 resolved manually.
597 Generating diff text
598 ~~~~~~~~~~~~~~~~~~~~
600 `diff`
601 ^^^^^^
603 The attribute `diff` affects how Git generates diffs for particular
604 files. It can tell Git whether to generate a textual patch for the path
605 or to treat the path as a binary file. It can also affect what line is
606 shown on the hunk header `@@ -k,l +n,m @@` line, tell Git to use an
607 external command to generate the diff, or ask Git to convert binary
608 files to a text format before generating the diff.
610 Set::
612 A path to which the `diff` attribute is set is treated
613 as text, even when they contain byte values that
614 normally never appear in text files, such as NUL.
616 Unset::
618 A path to which the `diff` attribute is unset will
619 generate `Binary files differ` (or a binary patch, if
620 binary patches are enabled).
622 Unspecified::
624 A path to which the `diff` attribute is unspecified
625 first gets its contents inspected, and if it looks like
626 text and is smaller than core.bigFileThreshold, it is treated
627 as text. Otherwise it would generate `Binary files differ`.
629 String::
631 Diff is shown using the specified diff driver. Each driver may
632 specify one or more options, as described in the following
633 section. The options for the diff driver "foo" are defined
634 by the configuration variables in the "" section of the
635 Git config file.
638 Defining an external diff driver
639 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
641 The definition of a diff driver is done in `gitconfig`, not
642 `gitattributes` file, so strictly speaking this manual page is a
643 wrong place to talk about it. However...
645 To define an external diff driver `jcdiff`, add a section to your
646 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
648 ----------------------------------------------------------------
649 [diff "jcdiff"]
650 command = j-c-diff
651 ----------------------------------------------------------------
653 When Git needs to show you a diff for the path with `diff`
654 attribute set to `jcdiff`, it calls the command you specified
655 with the above configuration, i.e. `j-c-diff`, with 7
656 parameters, just like `GIT_EXTERNAL_DIFF` program is called.
657 See linkgit:git[1] for details.
660 Defining a custom hunk-header
661 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
663 Each group of changes (called a "hunk") in the textual diff output
664 is prefixed with a line of the form:
666 @@ -k,l +n,m @@ TEXT
668 This is called a 'hunk header'. The "TEXT" portion is by default a line
669 that begins with an alphabet, an underscore or a dollar sign; this
670 matches what GNU 'diff -p' output uses. This default selection however
671 is not suited for some contents, and you can use a customized pattern
672 to make a selection.
674 First, in .gitattributes, you would assign the `diff` attribute
675 for paths.
677 ------------------------
678 *.tex diff=tex
679 ------------------------
681 Then, you would define a "diff.tex.xfuncname" configuration to
682 specify a regular expression that matches a line that you would
683 want to appear as the hunk header "TEXT". Add a section to your
684 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
686 ------------------------
687 [diff "tex"]
688 xfuncname = "^(\\\\(sub)*section\\{.*)$"
689 ------------------------
691 Note. A single level of backslashes are eaten by the
692 configuration file parser, so you would need to double the
693 backslashes; the pattern above picks a line that begins with a
694 backslash, and zero or more occurrences of `sub` followed by
695 `section` followed by open brace, to the end of line.
697 There are a few built-in patterns to make this easier, and `tex`
698 is one of them, so you do not have to write the above in your
699 configuration file (you still need to enable this with the
700 attribute mechanism, via `.gitattributes`). The following built in
701 patterns are available:
703 - `ada` suitable for source code in the Ada language.
705 - `bibtex` suitable for files with BibTeX coded references.
707 - `cpp` suitable for source code in the C and C++ languages.
709 - `csharp` suitable for source code in the C# language.
711 - `css` suitable for cascading style sheets.
713 - `fortran` suitable for source code in the Fortran language.
715 - `fountain` suitable for Fountain documents.
717 - `html` suitable for HTML/XHTML documents.
719 - `java` suitable for source code in the Java language.
721 - `matlab` suitable for source code in the MATLAB language.
723 - `objc` suitable for source code in the Objective-C language.
725 - `pascal` suitable for source code in the Pascal/Delphi language.
727 - `perl` suitable for source code in the Perl language.
729 - `php` suitable for source code in the PHP language.
731 - `python` suitable for source code in the Python language.
733 - `ruby` suitable for source code in the Ruby language.
735 - `tex` suitable for source code for LaTeX documents.
738 Customizing word diff
739 ^^^^^^^^^^^^^^^^^^^^^
741 You can customize the rules that `git diff --word-diff` uses to
742 split words in a line, by specifying an appropriate regular expression
743 in the "diff.*.wordRegex" configuration variable. For example, in TeX
744 a backslash followed by a sequence of letters forms a command, but
745 several such commands can be run together without intervening
746 whitespace. To separate them, use a regular expression in your
747 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
749 ------------------------
750 [diff "tex"]
751 wordRegex = "\\\\[a-zA-Z]+|[{}]|\\\\.|[^\\{}[:space:]]+"
752 ------------------------
754 A built-in pattern is provided for all languages listed in the
755 previous section.
758 Performing text diffs of binary files
759 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
761 Sometimes it is desirable to see the diff of a text-converted
762 version of some binary files. For example, a word processor
763 document can be converted to an ASCII text representation, and
764 the diff of the text shown. Even though this conversion loses
765 some information, the resulting diff is useful for human
766 viewing (but cannot be applied directly).
768 The `textconv` config option is used to define a program for
769 performing such a conversion. The program should take a single
770 argument, the name of a file to convert, and produce the
771 resulting text on stdout.
773 For example, to show the diff of the exif information of a
774 file instead of the binary information (assuming you have the
775 exif tool installed), add the following section to your
776 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file):
778 ------------------------
779 [diff "jpg"]
780 textconv = exif
781 ------------------------
783 NOTE: The text conversion is generally a one-way conversion;
784 in this example, we lose the actual image contents and focus
785 just on the text data. This means that diffs generated by
786 textconv are _not_ suitable for applying. For this reason,
787 only `git diff` and the `git log` family of commands (i.e.,
788 log, whatchanged, show) will perform text conversion. `git
789 format-patch` will never generate this output. If you want to
790 send somebody a text-converted diff of a binary file (e.g.,
791 because it quickly conveys the changes you have made), you
792 should generate it separately and send it as a comment _in
793 addition to_ the usual binary diff that you might send.
795 Because text conversion can be slow, especially when doing a
796 large number of them with `git log -p`, Git provides a mechanism
797 to cache the output and use it in future diffs. To enable
798 caching, set the "cachetextconv" variable in your diff driver's
799 config. For example:
801 ------------------------
802 [diff "jpg"]
803 textconv = exif
804 cachetextconv = true
805 ------------------------
807 This will cache the result of running "exif" on each blob
808 indefinitely. If you change the textconv config variable for a
809 diff driver, Git will automatically invalidate the cache entries
810 and re-run the textconv filter. If you want to invalidate the
811 cache manually (e.g., because your version of "exif" was updated
812 and now produces better output), you can remove the cache
813 manually with `git update-ref -d refs/notes/textconv/jpg` (where
814 "jpg" is the name of the diff driver, as in the example above).
816 Choosing textconv versus external diff
817 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
819 If you want to show differences between binary or specially-formatted
820 blobs in your repository, you can choose to use either an external diff
821 command, or to use textconv to convert them to a diff-able text format.
822 Which method you choose depends on your exact situation.
824 The advantage of using an external diff command is flexibility. You are
825 not bound to find line-oriented changes, nor is it necessary for the
826 output to resemble unified diff. You are free to locate and report
827 changes in the most appropriate way for your data format.
829 A textconv, by comparison, is much more limiting. You provide a
830 transformation of the data into a line-oriented text format, and Git
831 uses its regular diff tools to generate the output. There are several
832 advantages to choosing this method:
834 1. Ease of use. It is often much simpler to write a binary to text
835 transformation than it is to perform your own diff. In many cases,
836 existing programs can be used as textconv filters (e.g., exif,
837 odt2txt).
839 2. Git diff features. By performing only the transformation step
840 yourself, you can still utilize many of Git's diff features,
841 including colorization, word-diff, and combined diffs for merges.
843 3. Caching. Textconv caching can speed up repeated diffs, such as those
844 you might trigger by running `git log -p`.
847 Marking files as binary
848 ^^^^^^^^^^^^^^^^^^^^^^^
850 Git usually guesses correctly whether a blob contains text or binary
851 data by examining the beginning of the contents. However, sometimes you
852 may want to override its decision, either because a blob contains binary
853 data later in the file, or because the content, while technically
854 composed of text characters, is opaque to a human reader. For example,
855 many postscript files contain only ASCII characters, but produce noisy
856 and meaningless diffs.
858 The simplest way to mark a file as binary is to unset the diff
859 attribute in the `.gitattributes` file:
861 ------------------------
862 *.ps -diff
863 ------------------------
865 This will cause Git to generate `Binary files differ` (or a binary
866 patch, if binary patches are enabled) instead of a regular diff.
868 However, one may also want to specify other diff driver attributes. For
869 example, you might want to use `textconv` to convert postscript files to
870 an ASCII representation for human viewing, but otherwise treat them as
871 binary files. You cannot specify both `-diff` and `diff=ps` attributes.
872 The solution is to use the `diff.*.binary` config option:
874 ------------------------
875 [diff "ps"]
876 textconv = ps2ascii
877 binary = true
878 ------------------------
880 Performing a three-way merge
881 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
883 `merge`
884 ^^^^^^^
886 The attribute `merge` affects how three versions of a file are
887 merged when a file-level merge is necessary during `git merge`,
888 and other commands such as `git revert` and `git cherry-pick`.
890 Set::
892 Built-in 3-way merge driver is used to merge the
893 contents in a way similar to 'merge' command of `RCS`
894 suite. This is suitable for ordinary text files.
896 Unset::
898 Take the version from the current branch as the
899 tentative merge result, and declare that the merge has
900 conflicts. This is suitable for binary files that do
901 not have a well-defined merge semantics.
903 Unspecified::
905 By default, this uses the same built-in 3-way merge
906 driver as is the case when the `merge` attribute is set.
907 However, the `merge.default` configuration variable can name
908 different merge driver to be used with paths for which the
909 `merge` attribute is unspecified.
911 String::
913 3-way merge is performed using the specified custom
914 merge driver. The built-in 3-way merge driver can be
915 explicitly specified by asking for "text" driver; the
916 built-in "take the current branch" driver can be
917 requested with "binary".
920 Built-in merge drivers
921 ^^^^^^^^^^^^^^^^^^^^^^
923 There are a few built-in low-level merge drivers defined that
924 can be asked for via the `merge` attribute.
926 text::
928 Usual 3-way file level merge for text files. Conflicted
929 regions are marked with conflict markers `<<<<<<<`,
930 `=======` and `>>>>>>>`. The version from your branch
931 appears before the `=======` marker, and the version
932 from the merged branch appears after the `=======`
933 marker.
935 binary::
937 Keep the version from your branch in the work tree, but
938 leave the path in the conflicted state for the user to
939 sort out.
941 union::
943 Run 3-way file level merge for text files, but take
944 lines from both versions, instead of leaving conflict
945 markers. This tends to leave the added lines in the
946 resulting file in random order and the user should
947 verify the result. Do not use this if you do not
948 understand the implications.
951 Defining a custom merge driver
952 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
954 The definition of a merge driver is done in the `.git/config`
955 file, not in the `gitattributes` file, so strictly speaking this
956 manual page is a wrong place to talk about it. However...
958 To define a custom merge driver `filfre`, add a section to your
959 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
961 ----------------------------------------------------------------
962 [merge "filfre"]
963 name = feel-free merge driver
964 driver = filfre %O %A %B %L %P
965 recursive = binary
966 ----------------------------------------------------------------
968 The `merge.*.name` variable gives the driver a human-readable
969 name.
971 The `merge.*.driver` variable's value is used to construct a
972 command to run to merge ancestor's version (`%O`), current
973 version (`%A`) and the other branches' version (`%B`). These
974 three tokens are replaced with the names of temporary files that
975 hold the contents of these versions when the command line is
976 built. Additionally, %L will be replaced with the conflict marker
977 size (see below).
979 The merge driver is expected to leave the result of the merge in
980 the file named with `%A` by overwriting it, and exit with zero
981 status if it managed to merge them cleanly, or non-zero if there
982 were conflicts.
984 The `merge.*.recursive` variable specifies what other merge
985 driver to use when the merge driver is called for an internal
986 merge between common ancestors, when there are more than one.
987 When left unspecified, the driver itself is used for both
988 internal merge and the final merge.
990 The merge driver can learn the pathname in which the merged result
991 will be stored via placeholder `%P`.
994 `conflict-marker-size`
995 ^^^^^^^^^^^^^^^^^^^^^^
997 This attribute controls the length of conflict markers left in
998 the work tree file during a conflicted merge. Only setting to
999 the value to a positive integer has any meaningful effect.
1001 For example, this line in `.gitattributes` can be used to tell the merge
1002 machinery to leave much longer (instead of the usual 7-character-long)
1003 conflict markers when merging the file `Documentation/git-merge.txt`
1004 results in a conflict.
1006 ------------------------
1007 Documentation/git-merge.txt conflict-marker-size=32
1008 ------------------------
1011 Checking whitespace errors
1012 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1014 `whitespace`
1015 ^^^^^^^^^^^^
1017 The `core.whitespace` configuration variable allows you to define what
1018 'diff' and 'apply' should consider whitespace errors for all paths in
1019 the project (See linkgit:git-config[1]). This attribute gives you finer
1020 control per path.
1022 Set::
1024 Notice all types of potential whitespace errors known to Git.
1025 The tab width is taken from the value of the `core.whitespace`
1026 configuration variable.
1028 Unset::
1030 Do not notice anything as error.
1032 Unspecified::
1034 Use the value of the `core.whitespace` configuration variable to
1035 decide what to notice as error.
1037 String::
1039 Specify a comma separate list of common whitespace problems to
1040 notice in the same format as the `core.whitespace` configuration
1041 variable.
1044 Creating an archive
1045 ~~~~~~~~~~~~~~~~~~~
1047 `export-ignore`
1048 ^^^^^^^^^^^^^^^
1050 Files and directories with the attribute `export-ignore` won't be added to
1051 archive files.
1053 `export-subst`
1054 ^^^^^^^^^^^^^^
1056 If the attribute `export-subst` is set for a file then Git will expand
1057 several placeholders when adding this file to an archive. The
1058 expansion depends on the availability of a commit ID, i.e., if
1059 linkgit:git-archive[1] has been given a tree instead of a commit or a
1060 tag then no replacement will be done. The placeholders are the same
1061 as those for the option `--pretty=format:` of linkgit:git-log[1],
1062 except that they need to be wrapped like this: `$Format:PLACEHOLDERS$`
1063 in the file. E.g. the string `$Format:%H$` will be replaced by the
1064 commit hash.
1067 Packing objects
1068 ~~~~~~~~~~~~~~~
1070 `delta`
1071 ^^^^^^^
1073 Delta compression will not be attempted for blobs for paths with the
1074 attribute `delta` set to false.
1077 Viewing files in GUI tools
1078 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1080 `encoding`
1081 ^^^^^^^^^^
1083 The value of this attribute specifies the character encoding that should
1084 be used by GUI tools (e.g. linkgit:gitk[1] and linkgit:git-gui[1]) to
1085 display the contents of the relevant file. Note that due to performance
1086 considerations linkgit:gitk[1] does not use this attribute unless you
1087 manually enable per-file encodings in its options.
1089 If this attribute is not set or has an invalid value, the value of the
1090 `gui.encoding` configuration variable is used instead
1091 (See linkgit:git-config[1]).
1095 ----------------------
1097 You do not want any end-of-line conversions applied to, nor textual diffs
1098 produced for, any binary file you track. You would need to specify e.g.
1100 ------------
1101 *.jpg -text -diff
1102 ------------
1104 but that may become cumbersome, when you have many attributes. Using
1105 macro attributes, you can define an attribute that, when set, also
1106 sets or unsets a number of other attributes at the same time. The
1107 system knows a built-in macro attribute, `binary`:
1109 ------------
1110 *.jpg binary
1111 ------------
1113 Setting the "binary" attribute also unsets the "text" and "diff"
1114 attributes as above. Note that macro attributes can only be "Set",
1115 though setting one might have the effect of setting or unsetting other
1116 attributes or even returning other attributes to the "Unspecified"
1117 state.
1121 -------------------------
1123 Custom macro attributes can be defined only in top-level gitattributes
1124 files (`$GIT_DIR/info/attributes`, the `.gitattributes` file at the
1125 top level of the working tree, or the global or system-wide
1126 gitattributes files), not in `.gitattributes` files in working tree
1127 subdirectories. The built-in macro attribute "binary" is equivalent
1128 to:
1130 ------------
1131 [attr]binary -diff -merge -text
1132 ------------
1136 -------
1138 If you have these three `gitattributes` file:
1140 ----------------------------------------------------------------
1141 (in $GIT_DIR/info/attributes)
1143 a* foo !bar -baz
1145 (in .gitattributes)
1146 abc foo bar baz
1148 (in t/.gitattributes)
1149 ab* merge=filfre
1150 abc -foo -bar
1151 *.c frotz
1152 ----------------------------------------------------------------
1154 the attributes given to path `t/abc` are computed as follows:
1156 1. By examining `t/.gitattributes` (which is in the same
1157 directory as the path in question), Git finds that the first
1158 line matches. `merge` attribute is set. It also finds that
1159 the second line matches, and attributes `foo` and `bar`
1160 are unset.
1162 2. Then it examines `.gitattributes` (which is in the parent
1163 directory), and finds that the first line matches, but
1164 `t/.gitattributes` file already decided how `merge`, `foo`
1165 and `bar` attributes should be given to this path, so it
1166 leaves `foo` and `bar` unset. Attribute `baz` is set.
1168 3. Finally it examines `$GIT_DIR/info/attributes`. This file
1169 is used to override the in-tree settings. The first line is
1170 a match, and `foo` is set, `bar` is reverted to unspecified
1171 state, and `baz` is unset.
1173 As the result, the attributes assignment to `t/abc` becomes:
1175 ----------------------------------------------------------------
1176 foo set to true
1177 bar unspecified
1178 baz set to false
1179 merge set to string value "filfre"
1180 frotz unspecified
1181 ----------------------------------------------------------------
1185 --------
1186 linkgit:git-check-attr[1].
1188 GIT
1189 ---
1190 Part of the linkgit:git[1] suite