Merge branch 'tb/doc-eol-normalization'
[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.
156 Set to string value "crlf"::
158 This setting forces Git to normalize line endings for this
159 file on checkin and convert them to CRLF when the file is
160 checked out.
162 Set to string value "lf"::
164 This setting forces Git to normalize line endings to LF on
165 checkin and prevents conversion to CRLF when the file is
166 checked out.
168 Backwards compatibility with `crlf` attribute
169 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
171 For backwards compatibility, the `crlf` attribute is interpreted as
172 follows:
174 ------------------------
175 crlf text
176 -crlf -text
177 crlf=input eol=lf
178 ------------------------
180 End-of-line conversion
181 ^^^^^^^^^^^^^^^^^^^^^^
183 While Git normally leaves file contents alone, it can be configured to
184 normalize line endings to LF in the repository and, optionally, to
185 convert them to CRLF when files are checked out.
187 If you simply want to have CRLF line endings in your working directory
188 regardless of the repository you are working with, you can set the
189 config variable "core.autocrlf" without using any attributes.
191 ------------------------
192 [core]
193 autocrlf = true
194 ------------------------
196 This does not force normalization of text files, but does ensure
197 that text files that you introduce to the repository have their line
198 endings normalized to LF when they are added, and that files that are
199 already normalized in the repository stay normalized.
201 If you want to ensure that text files that any contributor introduces to
202 the repository have their line endings normalized, you can set the
203 `text` attribute to "auto" for _all_ files.
205 ------------------------
206 * text=auto
207 ------------------------
209 The attributes allow a fine-grained control, how the line endings
210 are converted.
211 Here is an example that will make Git normalize .txt, .vcproj and .sh
212 files, ensure that .vcproj files have CRLF and .sh files have LF in
213 the working directory, and prevent .jpg files from being normalized
214 regardless of their content.
216 ------------------------
217 * text=auto
218 *.txt text
219 *.vcproj text eol=crlf
220 *.sh text eol=lf
221 *.jpg -text
222 ------------------------
224 NOTE: When `text=auto` conversion is enabled in a cross-platform
225 project using push and pull to a central repository the text files
226 containing CRLFs should be normalized.
228 From a clean working directory:
230 -------------------------------------------------
231 $ echo "* text=auto" >.gitattributes
232 $ rm .git/index # Remove the index to re-scan the working directory
233 $ git add .
234 $ git status # Show files that will be normalized
235 $ git commit -m "Introduce end-of-line normalization"
236 -------------------------------------------------
238 If any files that should not be normalized show up in 'git status',
239 unset their `text` attribute before running 'git add -u'.
241 ------------------------
242 manual.pdf -text
243 ------------------------
245 Conversely, text files that Git does not detect can have normalization
246 enabled manually.
248 ------------------------
249 weirdchars.txt text
250 ------------------------
252 If `core.safecrlf` is set to "true" or "warn", Git verifies if
253 the conversion is reversible for the current setting of
254 `core.autocrlf`. For "true", Git rejects irreversible
255 conversions; for "warn", Git only prints a warning but accepts
256 an irreversible conversion. The safety triggers to prevent such
257 a conversion done to the files in the work tree, but there are a
258 few exceptions. Even though...
260 - 'git add' itself does not touch the files in the work tree, the
261 next checkout would, so the safety triggers;
263 - 'git apply' to update a text file with a patch does touch the files
264 in the work tree, but the operation is about text files and CRLF
265 conversion is about fixing the line ending inconsistencies, so the
266 safety does not trigger;
268 - 'git diff' itself does not touch the files in the work tree, it is
269 often run to inspect the changes you intend to next 'git add'. To
270 catch potential problems early, safety triggers.
273 `ident`
274 ^^^^^^^
276 When the attribute `ident` is set for a path, Git replaces
277 `$Id$` in the blob object with `$Id:`, followed by the
278 40-character hexadecimal blob object name, followed by a dollar
279 sign `$` upon checkout. Any byte sequence that begins with
280 `$Id:` and ends with `$` in the worktree file is replaced
281 with `$Id$` upon check-in.
284 `filter`
285 ^^^^^^^^
287 A `filter` attribute can be set to a string value that names a
288 filter driver specified in the configuration.
290 A filter driver consists of a `clean` command and a `smudge`
291 command, either of which can be left unspecified. Upon
292 checkout, when the `smudge` command is specified, the command is
293 fed the blob object from its standard input, and its standard
294 output is used to update the worktree file. Similarly, the
295 `clean` command is used to convert the contents of worktree file
296 upon checkin. By default these commands process only a single
297 blob and terminate. If a long running `process` filter is used
298 in place of `clean` and/or `smudge` filters, then Git can process
299 all blobs with a single filter command invocation for the entire
300 life of a single Git command, for example `git add --all`. If a
301 long running `process` filter is configured then it always takes
302 precedence over a configured single blob filter. See section
303 below for the description of the protocol used to communicate with
304 a `process` filter.
306 One use of the content filtering is to massage the content into a shape
307 that is more convenient for the platform, filesystem, and the user to use.
308 For this mode of operation, the key phrase here is "more convenient" and
309 not "turning something unusable into usable". In other words, the intent
310 is that if someone unsets the filter driver definition, or does not have
311 the appropriate filter program, the project should still be usable.
313 Another use of the content filtering is to store the content that cannot
314 be directly used in the repository (e.g. a UUID that refers to the true
315 content stored outside Git, or an encrypted content) and turn it into a
316 usable form upon checkout (e.g. download the external content, or decrypt
317 the encrypted content).
319 These two filters behave differently, and by default, a filter is taken as
320 the former, massaging the contents into more convenient shape. A missing
321 filter driver definition in the config, or a filter driver that exits with
322 a non-zero status, is not an error but makes the filter a no-op passthru.
324 You can declare that a filter turns a content that by itself is unusable
325 into a usable content by setting the filter.<driver>.required configuration
326 variable to `true`.
328 For example, in .gitattributes, you would assign the `filter`
329 attribute for paths.
331 ------------------------
332 *.c filter=indent
333 ------------------------
335 Then you would define a "filter.indent.clean" and "filter.indent.smudge"
336 configuration in your .git/config to specify a pair of commands to
337 modify the contents of C programs when the source files are checked
338 in ("clean" is run) and checked out (no change is made because the
339 command is "cat").
341 ------------------------
342 [filter "indent"]
343 clean = indent
344 smudge = cat
345 ------------------------
347 For best results, `clean` should not alter its output further if it is
348 run twice ("clean->clean" should be equivalent to "clean"), and
349 multiple `smudge` commands should not alter `clean`'s output
350 ("smudge->smudge->clean" should be equivalent to "clean"). See the
351 section on merging below.
353 The "indent" filter is well-behaved in this regard: it will not modify
354 input that is already correctly indented. In this case, the lack of a
355 smudge filter means that the clean filter _must_ accept its own output
356 without modifying it.
358 If a filter _must_ succeed in order to make the stored contents usable,
359 you can declare that the filter is `required`, in the configuration:
361 ------------------------
362 [filter "crypt"]
363 clean = openssl enc ...
364 smudge = openssl enc -d ...
365 required
366 ------------------------
368 Sequence "%f" on the filter command line is replaced with the name of
369 the file the filter is working on. A filter might use this in keyword
370 substitution. For example:
372 ------------------------
373 [filter "p4"]
374 clean = git-p4-filter --clean %f
375 smudge = git-p4-filter --smudge %f
376 ------------------------
378 Note that "%f" is the name of the path that is being worked on. Depending
379 on the version that is being filtered, the corresponding file on disk may
380 not exist, or may have different contents. So, smudge and clean commands
381 should not try to access the file on disk, but only act as filters on the
382 content provided to them on standard input.
384 Long Running Filter Process
385 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
387 If the filter command (a string value) is defined via
388 `filter.<driver>.process` then Git can process all blobs with a
389 single filter invocation for the entire life of a single Git
390 command. This is achieved by using a packet format (pkt-line,
391 see technical/protocol-common.txt) based protocol over standard
392 input and standard output as follows. All packets, except for the
393 "*CONTENT" packets and the "0000" flush packet, are considered
394 text and therefore are terminated by a LF.
396 Git starts the filter when it encounters the first file
397 that needs to be cleaned or smudged. After the filter started
398 Git sends a welcome message ("git-filter-client"), a list of supported
399 protocol version numbers, and a flush packet. Git expects to read a welcome
400 response message ("git-filter-server"), exactly one protocol version number
401 from the previously sent list, and a flush packet. All further
402 communication will be based on the selected version. The remaining
403 protocol description below documents "version=2". Please note that
404 "version=42" in the example below does not exist and is only there
405 to illustrate how the protocol would look like with more than one
406 version.
408 After the version negotiation Git sends a list of all capabilities that
409 it supports and a flush packet. Git expects to read a list of desired
410 capabilities, which must be a subset of the supported capabilities list,
411 and a flush packet as response:
412 ------------------------
413 packet: git> git-filter-client
414 packet: git> version=2
415 packet: git> version=42
416 packet: git> 0000
417 packet: git< git-filter-server
418 packet: git< version=2
419 packet: git< 0000
420 packet: git> capability=clean
421 packet: git> capability=smudge
422 packet: git> capability=not-yet-invented
423 packet: git> 0000
424 packet: git< capability=clean
425 packet: git< capability=smudge
426 packet: git< 0000
427 ------------------------
428 Supported filter capabilities in version 2 are "clean" and
429 "smudge".
431 Afterwards Git sends a list of "key=value" pairs terminated with
432 a flush packet. The list will contain at least the filter command
433 (based on the supported capabilities) and the pathname of the file
434 to filter relative to the repository root. Right after the flush packet
435 Git sends the content split in zero or more pkt-line packets and a
436 flush packet to terminate content. Please note, that the filter
437 must not send any response before it received the content and the
438 final flush packet. Also note that the "value" of a "key=value" pair
439 can contain the "=" character whereas the key would never contain
440 that character.
441 ------------------------
442 packet: git> command=smudge
443 packet: git> pathname=path/testfile.dat
444 packet: git> 0000
445 packet: git> CONTENT
446 packet: git> 0000
447 ------------------------
449 The filter is expected to respond with a list of "key=value" pairs
450 terminated with a flush packet. If the filter does not experience
451 problems then the list must contain a "success" status. Right after
452 these packets the filter is expected to send the content in zero
453 or more pkt-line packets and a flush packet at the end. Finally, a
454 second list of "key=value" pairs terminated with a flush packet
455 is expected. The filter can change the status in the second list
456 or keep the status as is with an empty list. Please note that the
457 empty list must be terminated with a flush packet regardless.
459 ------------------------
460 packet: git< status=success
461 packet: git< 0000
462 packet: git< SMUDGED_CONTENT
463 packet: git< 0000
464 packet: git< 0000 # empty list, keep "status=success" unchanged!
465 ------------------------
467 If the result content is empty then the filter is expected to respond
468 with a "success" status and a flush packet to signal the empty content.
469 ------------------------
470 packet: git< status=success
471 packet: git< 0000
472 packet: git< 0000 # empty content!
473 packet: git< 0000 # empty list, keep "status=success" unchanged!
474 ------------------------
476 In case the filter cannot or does not want to process the content,
477 it is expected to respond with an "error" status.
478 ------------------------
479 packet: git< status=error
480 packet: git< 0000
481 ------------------------
483 If the filter experiences an error during processing, then it can
484 send the status "error" after the content was (partially or
485 completely) sent.
486 ------------------------
487 packet: git< status=success
488 packet: git< 0000
490 packet: git< 0000
491 packet: git< status=error
492 packet: git< 0000
493 ------------------------
495 In case the filter cannot or does not want to process the content
496 as well as any future content for the lifetime of the Git process,
497 then it is expected to respond with an "abort" status at any point
498 in the protocol.
499 ------------------------
500 packet: git< status=abort
501 packet: git< 0000
502 ------------------------
504 Git neither stops nor restarts the filter process in case the
505 "error"/"abort" status is set. However, Git sets its exit code
506 according to the `filter.<driver>.required` flag, mimicking the
507 behavior of the `filter.<driver>.clean` / `filter.<driver>.smudge`
508 mechanism.
510 If the filter dies during the communication or does not adhere to
511 the protocol then Git will stop the filter process and restart it
512 with the next file that needs to be processed. Depending on the
513 `filter.<driver>.required` flag Git will interpret that as error.
515 After the filter has processed a blob it is expected to wait for
516 the next "key=value" list containing a command. Git will close
517 the command pipe on exit. The filter is expected to detect EOF
518 and exit gracefully on its own. Git will wait until the filter
519 process has stopped.
521 A long running filter demo implementation can be found in
522 `contrib/long-running-filter/` located in the Git
523 core repository. If you develop your own long running filter
524 process then the `GIT_TRACE_PACKET` environment variables can be
525 very helpful for debugging (see linkgit:git[1]).
527 Please note that you cannot use an existing `filter.<driver>.clean`
528 or `filter.<driver>.smudge` command with `filter.<driver>.process`
529 because the former two use a different inter process communication
530 protocol than the latter one.
533 Interaction between checkin/checkout attributes
534 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
536 In the check-in codepath, the worktree file is first converted
537 with `filter` driver (if specified and corresponding driver
538 defined), then the result is processed with `ident` (if
539 specified), and then finally with `text` (again, if specified
540 and applicable).
542 In the check-out codepath, the blob content is first converted
543 with `text`, and then `ident` and fed to `filter`.
546 Merging branches with differing checkin/checkout attributes
547 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
549 If you have added attributes to a file that cause the canonical
550 repository format for that file to change, such as adding a
551 clean/smudge filter or text/eol/ident attributes, merging anything
552 where the attribute is not in place would normally cause merge
553 conflicts.
555 To prevent these unnecessary merge conflicts, Git can be told to run a
556 virtual check-out and check-in of all three stages of a file when
557 resolving a three-way merge by setting the `merge.renormalize`
558 configuration variable. This prevents changes caused by check-in
559 conversion from causing spurious merge conflicts when a converted file
560 is merged with an unconverted file.
562 As long as a "smudge->clean" results in the same output as a "clean"
563 even on files that are already smudged, this strategy will
564 automatically resolve all filter-related conflicts. Filters that do
565 not act in this way may cause additional merge conflicts that must be
566 resolved manually.
569 Generating diff text
570 ~~~~~~~~~~~~~~~~~~~~
572 `diff`
573 ^^^^^^
575 The attribute `diff` affects how Git generates diffs for particular
576 files. It can tell Git whether to generate a textual patch for the path
577 or to treat the path as a binary file. It can also affect what line is
578 shown on the hunk header `@@ -k,l +n,m @@` line, tell Git to use an
579 external command to generate the diff, or ask Git to convert binary
580 files to a text format before generating the diff.
582 Set::
584 A path to which the `diff` attribute is set is treated
585 as text, even when they contain byte values that
586 normally never appear in text files, such as NUL.
588 Unset::
590 A path to which the `diff` attribute is unset will
591 generate `Binary files differ` (or a binary patch, if
592 binary patches are enabled).
594 Unspecified::
596 A path to which the `diff` attribute is unspecified
597 first gets its contents inspected, and if it looks like
598 text and is smaller than core.bigFileThreshold, it is treated
599 as text. Otherwise it would generate `Binary files differ`.
601 String::
603 Diff is shown using the specified diff driver. Each driver may
604 specify one or more options, as described in the following
605 section. The options for the diff driver "foo" are defined
606 by the configuration variables in the "" section of the
607 Git config file.
610 Defining an external diff driver
611 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
613 The definition of a diff driver is done in `gitconfig`, not
614 `gitattributes` file, so strictly speaking this manual page is a
615 wrong place to talk about it. However...
617 To define an external diff driver `jcdiff`, add a section to your
618 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
620 ----------------------------------------------------------------
621 [diff "jcdiff"]
622 command = j-c-diff
623 ----------------------------------------------------------------
625 When Git needs to show you a diff for the path with `diff`
626 attribute set to `jcdiff`, it calls the command you specified
627 with the above configuration, i.e. `j-c-diff`, with 7
628 parameters, just like `GIT_EXTERNAL_DIFF` program is called.
629 See linkgit:git[1] for details.
632 Defining a custom hunk-header
633 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
635 Each group of changes (called a "hunk") in the textual diff output
636 is prefixed with a line of the form:
638 @@ -k,l +n,m @@ TEXT
640 This is called a 'hunk header'. The "TEXT" portion is by default a line
641 that begins with an alphabet, an underscore or a dollar sign; this
642 matches what GNU 'diff -p' output uses. This default selection however
643 is not suited for some contents, and you can use a customized pattern
644 to make a selection.
646 First, in .gitattributes, you would assign the `diff` attribute
647 for paths.
649 ------------------------
650 *.tex diff=tex
651 ------------------------
653 Then, you would define a "diff.tex.xfuncname" configuration to
654 specify a regular expression that matches a line that you would
655 want to appear as the hunk header "TEXT". Add a section to your
656 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
658 ------------------------
659 [diff "tex"]
660 xfuncname = "^(\\\\(sub)*section\\{.*)$"
661 ------------------------
663 Note. A single level of backslashes are eaten by the
664 configuration file parser, so you would need to double the
665 backslashes; the pattern above picks a line that begins with a
666 backslash, and zero or more occurrences of `sub` followed by
667 `section` followed by open brace, to the end of line.
669 There are a few built-in patterns to make this easier, and `tex`
670 is one of them, so you do not have to write the above in your
671 configuration file (you still need to enable this with the
672 attribute mechanism, via `.gitattributes`). The following built in
673 patterns are available:
675 - `ada` suitable for source code in the Ada language.
677 - `bibtex` suitable for files with BibTeX coded references.
679 - `cpp` suitable for source code in the C and C++ languages.
681 - `csharp` suitable for source code in the C# language.
683 - `css` suitable for cascading style sheets.
685 - `fortran` suitable for source code in the Fortran language.
687 - `fountain` suitable for Fountain documents.
689 - `html` suitable for HTML/XHTML documents.
691 - `java` suitable for source code in the Java language.
693 - `matlab` suitable for source code in the MATLAB language.
695 - `objc` suitable for source code in the Objective-C language.
697 - `pascal` suitable for source code in the Pascal/Delphi language.
699 - `perl` suitable for source code in the Perl language.
701 - `php` suitable for source code in the PHP language.
703 - `python` suitable for source code in the Python language.
705 - `ruby` suitable for source code in the Ruby language.
707 - `tex` suitable for source code for LaTeX documents.
710 Customizing word diff
711 ^^^^^^^^^^^^^^^^^^^^^
713 You can customize the rules that `git diff --word-diff` uses to
714 split words in a line, by specifying an appropriate regular expression
715 in the "diff.*.wordRegex" configuration variable. For example, in TeX
716 a backslash followed by a sequence of letters forms a command, but
717 several such commands can be run together without intervening
718 whitespace. To separate them, use a regular expression in your
719 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
721 ------------------------
722 [diff "tex"]
723 wordRegex = "\\\\[a-zA-Z]+|[{}]|\\\\.|[^\\{}[:space:]]+"
724 ------------------------
726 A built-in pattern is provided for all languages listed in the
727 previous section.
730 Performing text diffs of binary files
731 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
733 Sometimes it is desirable to see the diff of a text-converted
734 version of some binary files. For example, a word processor
735 document can be converted to an ASCII text representation, and
736 the diff of the text shown. Even though this conversion loses
737 some information, the resulting diff is useful for human
738 viewing (but cannot be applied directly).
740 The `textconv` config option is used to define a program for
741 performing such a conversion. The program should take a single
742 argument, the name of a file to convert, and produce the
743 resulting text on stdout.
745 For example, to show the diff of the exif information of a
746 file instead of the binary information (assuming you have the
747 exif tool installed), add the following section to your
748 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file):
750 ------------------------
751 [diff "jpg"]
752 textconv = exif
753 ------------------------
755 NOTE: The text conversion is generally a one-way conversion;
756 in this example, we lose the actual image contents and focus
757 just on the text data. This means that diffs generated by
758 textconv are _not_ suitable for applying. For this reason,
759 only `git diff` and the `git log` family of commands (i.e.,
760 log, whatchanged, show) will perform text conversion. `git
761 format-patch` will never generate this output. If you want to
762 send somebody a text-converted diff of a binary file (e.g.,
763 because it quickly conveys the changes you have made), you
764 should generate it separately and send it as a comment _in
765 addition to_ the usual binary diff that you might send.
767 Because text conversion can be slow, especially when doing a
768 large number of them with `git log -p`, Git provides a mechanism
769 to cache the output and use it in future diffs. To enable
770 caching, set the "cachetextconv" variable in your diff driver's
771 config. For example:
773 ------------------------
774 [diff "jpg"]
775 textconv = exif
776 cachetextconv = true
777 ------------------------
779 This will cache the result of running "exif" on each blob
780 indefinitely. If you change the textconv config variable for a
781 diff driver, Git will automatically invalidate the cache entries
782 and re-run the textconv filter. If you want to invalidate the
783 cache manually (e.g., because your version of "exif" was updated
784 and now produces better output), you can remove the cache
785 manually with `git update-ref -d refs/notes/textconv/jpg` (where
786 "jpg" is the name of the diff driver, as in the example above).
788 Choosing textconv versus external diff
789 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
791 If you want to show differences between binary or specially-formatted
792 blobs in your repository, you can choose to use either an external diff
793 command, or to use textconv to convert them to a diff-able text format.
794 Which method you choose depends on your exact situation.
796 The advantage of using an external diff command is flexibility. You are
797 not bound to find line-oriented changes, nor is it necessary for the
798 output to resemble unified diff. You are free to locate and report
799 changes in the most appropriate way for your data format.
801 A textconv, by comparison, is much more limiting. You provide a
802 transformation of the data into a line-oriented text format, and Git
803 uses its regular diff tools to generate the output. There are several
804 advantages to choosing this method:
806 1. Ease of use. It is often much simpler to write a binary to text
807 transformation than it is to perform your own diff. In many cases,
808 existing programs can be used as textconv filters (e.g., exif,
809 odt2txt).
811 2. Git diff features. By performing only the transformation step
812 yourself, you can still utilize many of Git's diff features,
813 including colorization, word-diff, and combined diffs for merges.
815 3. Caching. Textconv caching can speed up repeated diffs, such as those
816 you might trigger by running `git log -p`.
819 Marking files as binary
820 ^^^^^^^^^^^^^^^^^^^^^^^
822 Git usually guesses correctly whether a blob contains text or binary
823 data by examining the beginning of the contents. However, sometimes you
824 may want to override its decision, either because a blob contains binary
825 data later in the file, or because the content, while technically
826 composed of text characters, is opaque to a human reader. For example,
827 many postscript files contain only ASCII characters, but produce noisy
828 and meaningless diffs.
830 The simplest way to mark a file as binary is to unset the diff
831 attribute in the `.gitattributes` file:
833 ------------------------
834 *.ps -diff
835 ------------------------
837 This will cause Git to generate `Binary files differ` (or a binary
838 patch, if binary patches are enabled) instead of a regular diff.
840 However, one may also want to specify other diff driver attributes. For
841 example, you might want to use `textconv` to convert postscript files to
842 an ASCII representation for human viewing, but otherwise treat them as
843 binary files. You cannot specify both `-diff` and `diff=ps` attributes.
844 The solution is to use the `diff.*.binary` config option:
846 ------------------------
847 [diff "ps"]
848 textconv = ps2ascii
849 binary = true
850 ------------------------
852 Performing a three-way merge
853 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
855 `merge`
856 ^^^^^^^
858 The attribute `merge` affects how three versions of a file are
859 merged when a file-level merge is necessary during `git merge`,
860 and other commands such as `git revert` and `git cherry-pick`.
862 Set::
864 Built-in 3-way merge driver is used to merge the
865 contents in a way similar to 'merge' command of `RCS`
866 suite. This is suitable for ordinary text files.
868 Unset::
870 Take the version from the current branch as the
871 tentative merge result, and declare that the merge has
872 conflicts. This is suitable for binary files that do
873 not have a well-defined merge semantics.
875 Unspecified::
877 By default, this uses the same built-in 3-way merge
878 driver as is the case when the `merge` attribute is set.
879 However, the `merge.default` configuration variable can name
880 different merge driver to be used with paths for which the
881 `merge` attribute is unspecified.
883 String::
885 3-way merge is performed using the specified custom
886 merge driver. The built-in 3-way merge driver can be
887 explicitly specified by asking for "text" driver; the
888 built-in "take the current branch" driver can be
889 requested with "binary".
892 Built-in merge drivers
893 ^^^^^^^^^^^^^^^^^^^^^^
895 There are a few built-in low-level merge drivers defined that
896 can be asked for via the `merge` attribute.
898 text::
900 Usual 3-way file level merge for text files. Conflicted
901 regions are marked with conflict markers `<<<<<<<`,
902 `=======` and `>>>>>>>`. The version from your branch
903 appears before the `=======` marker, and the version
904 from the merged branch appears after the `=======`
905 marker.
907 binary::
909 Keep the version from your branch in the work tree, but
910 leave the path in the conflicted state for the user to
911 sort out.
913 union::
915 Run 3-way file level merge for text files, but take
916 lines from both versions, instead of leaving conflict
917 markers. This tends to leave the added lines in the
918 resulting file in random order and the user should
919 verify the result. Do not use this if you do not
920 understand the implications.
923 Defining a custom merge driver
924 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
926 The definition of a merge driver is done in the `.git/config`
927 file, not in the `gitattributes` file, so strictly speaking this
928 manual page is a wrong place to talk about it. However...
930 To define a custom merge driver `filfre`, add a section to your
931 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
933 ----------------------------------------------------------------
934 [merge "filfre"]
935 name = feel-free merge driver
936 driver = filfre %O %A %B %L %P
937 recursive = binary
938 ----------------------------------------------------------------
940 The `merge.*.name` variable gives the driver a human-readable
941 name.
943 The `merge.*.driver` variable's value is used to construct a
944 command to run to merge ancestor's version (`%O`), current
945 version (`%A`) and the other branches' version (`%B`). These
946 three tokens are replaced with the names of temporary files that
947 hold the contents of these versions when the command line is
948 built. Additionally, %L will be replaced with the conflict marker
949 size (see below).
951 The merge driver is expected to leave the result of the merge in
952 the file named with `%A` by overwriting it, and exit with zero
953 status if it managed to merge them cleanly, or non-zero if there
954 were conflicts.
956 The `merge.*.recursive` variable specifies what other merge
957 driver to use when the merge driver is called for an internal
958 merge between common ancestors, when there are more than one.
959 When left unspecified, the driver itself is used for both
960 internal merge and the final merge.
962 The merge driver can learn the pathname in which the merged result
963 will be stored via placeholder `%P`.
966 `conflict-marker-size`
967 ^^^^^^^^^^^^^^^^^^^^^^
969 This attribute controls the length of conflict markers left in
970 the work tree file during a conflicted merge. Only setting to
971 the value to a positive integer has any meaningful effect.
973 For example, this line in `.gitattributes` can be used to tell the merge
974 machinery to leave much longer (instead of the usual 7-character-long)
975 conflict markers when merging the file `Documentation/git-merge.txt`
976 results in a conflict.
978 ------------------------
979 Documentation/git-merge.txt conflict-marker-size=32
980 ------------------------
983 Checking whitespace errors
984 ~~~~~~~~~~~~~~~~~~~~~~~~~~
986 `whitespace`
987 ^^^^^^^^^^^^
989 The `core.whitespace` configuration variable allows you to define what
990 'diff' and 'apply' should consider whitespace errors for all paths in
991 the project (See linkgit:git-config[1]). This attribute gives you finer
992 control per path.
994 Set::
996 Notice all types of potential whitespace errors known to Git.
997 The tab width is taken from the value of the `core.whitespace`
998 configuration variable.
1000 Unset::
1002 Do not notice anything as error.
1004 Unspecified::
1006 Use the value of the `core.whitespace` configuration variable to
1007 decide what to notice as error.
1009 String::
1011 Specify a comma separate list of common whitespace problems to
1012 notice in the same format as the `core.whitespace` configuration
1013 variable.
1016 Creating an archive
1017 ~~~~~~~~~~~~~~~~~~~
1019 `export-ignore`
1020 ^^^^^^^^^^^^^^^
1022 Files and directories with the attribute `export-ignore` won't be added to
1023 archive files.
1025 `export-subst`
1026 ^^^^^^^^^^^^^^
1028 If the attribute `export-subst` is set for a file then Git will expand
1029 several placeholders when adding this file to an archive. The
1030 expansion depends on the availability of a commit ID, i.e., if
1031 linkgit:git-archive[1] has been given a tree instead of a commit or a
1032 tag then no replacement will be done. The placeholders are the same
1033 as those for the option `--pretty=format:` of linkgit:git-log[1],
1034 except that they need to be wrapped like this: `$Format:PLACEHOLDERS$`
1035 in the file. E.g. the string `$Format:%H$` will be replaced by the
1036 commit hash.
1039 Packing objects
1040 ~~~~~~~~~~~~~~~
1042 `delta`
1043 ^^^^^^^
1045 Delta compression will not be attempted for blobs for paths with the
1046 attribute `delta` set to false.
1049 Viewing files in GUI tools
1050 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1052 `encoding`
1053 ^^^^^^^^^^
1055 The value of this attribute specifies the character encoding that should
1056 be used by GUI tools (e.g. linkgit:gitk[1] and linkgit:git-gui[1]) to
1057 display the contents of the relevant file. Note that due to performance
1058 considerations linkgit:gitk[1] does not use this attribute unless you
1059 manually enable per-file encodings in its options.
1061 If this attribute is not set or has an invalid value, the value of the
1062 `gui.encoding` configuration variable is used instead
1063 (See linkgit:git-config[1]).
1067 ----------------------
1069 You do not want any end-of-line conversions applied to, nor textual diffs
1070 produced for, any binary file you track. You would need to specify e.g.
1072 ------------
1073 *.jpg -text -diff
1074 ------------
1076 but that may become cumbersome, when you have many attributes. Using
1077 macro attributes, you can define an attribute that, when set, also
1078 sets or unsets a number of other attributes at the same time. The
1079 system knows a built-in macro attribute, `binary`:
1081 ------------
1082 *.jpg binary
1083 ------------
1085 Setting the "binary" attribute also unsets the "text" and "diff"
1086 attributes as above. Note that macro attributes can only be "Set",
1087 though setting one might have the effect of setting or unsetting other
1088 attributes or even returning other attributes to the "Unspecified"
1089 state.
1093 -------------------------
1095 Custom macro attributes can be defined only in top-level gitattributes
1096 files (`$GIT_DIR/info/attributes`, the `.gitattributes` file at the
1097 top level of the working tree, or the global or system-wide
1098 gitattributes files), not in `.gitattributes` files in working tree
1099 subdirectories. The built-in macro attribute "binary" is equivalent
1100 to:
1102 ------------
1103 [attr]binary -diff -merge -text
1104 ------------
1108 -------
1110 If you have these three `gitattributes` file:
1112 ----------------------------------------------------------------
1113 (in $GIT_DIR/info/attributes)
1115 a* foo !bar -baz
1117 (in .gitattributes)
1118 abc foo bar baz
1120 (in t/.gitattributes)
1121 ab* merge=filfre
1122 abc -foo -bar
1123 *.c frotz
1124 ----------------------------------------------------------------
1126 the attributes given to path `t/abc` are computed as follows:
1128 1. By examining `t/.gitattributes` (which is in the same
1129 directory as the path in question), Git finds that the first
1130 line matches. `merge` attribute is set. It also finds that
1131 the second line matches, and attributes `foo` and `bar`
1132 are unset.
1134 2. Then it examines `.gitattributes` (which is in the parent
1135 directory), and finds that the first line matches, but
1136 `t/.gitattributes` file already decided how `merge`, `foo`
1137 and `bar` attributes should be given to this path, so it
1138 leaves `foo` and `bar` unset. Attribute `baz` is set.
1140 3. Finally it examines `$GIT_DIR/info/attributes`. This file
1141 is used to override the in-tree settings. The first line is
1142 a match, and `foo` is set, `bar` is reverted to unspecified
1143 state, and `baz` is unset.
1145 As the result, the attributes assignment to `t/abc` becomes:
1147 ----------------------------------------------------------------
1148 foo set to true
1149 bar unspecified
1150 baz set to false
1151 merge set to string value "filfre"
1152 frotz unspecified
1153 ----------------------------------------------------------------
1157 --------
1158 linkgit:git-check-attr[1].
1160 GIT
1161 ---
1162 Part of the linkgit:git[1] suite