1GITATTRIBUTES(5) Git Manual GITATTRIBUTES(5)
2
3
4
6 gitattributes - Defining attributes per path
7
9 $GIT_DIR/info/attributes, .gitattributes
10
12 A gitattributes file is a simple text file that gives attributes to
13 pathnames.
14
15 Each line in gitattributes file is of form:
16
17 pattern attr1 attr2 ...
18
19 That is, a pattern followed by an attributes list, separated by
20 whitespaces. Leading and trailing whitespaces are ignored. Lines that
21 begin with # are ignored. Patterns that begin with a double quote are
22 quoted in C style. When the pattern matches the path in question, the
23 attributes listed on the line are given to the path.
24
25 Each attribute can be in one of these states for a given path:
26
27 Set
28 The path has the attribute with special value "true"; this is
29 specified by listing only the name of the attribute in the
30 attribute list.
31
32 Unset
33 The path has the attribute with special value "false"; this is
34 specified by listing the name of the attribute prefixed with a dash
35 - in the attribute list.
36
37 Set to a value
38 The path has the attribute with specified string value; this is
39 specified by listing the name of the attribute followed by an equal
40 sign = and its value in the attribute list.
41
42 Unspecified
43 No pattern matches the path, and nothing says if the path has or
44 does not have the attribute, the attribute for the path is said to
45 be Unspecified.
46
47 When more than one pattern matches the path, a later line overrides an
48 earlier line. This overriding is done per attribute.
49
50 The rules by which the pattern matches paths are the same as in
51 .gitignore files (see gitignore(5)), with a few exceptions:
52
53 • negative patterns are forbidden
54
55 • patterns that match a directory do not recursively match paths
56 inside that directory (so using the trailing-slash path/ syntax is
57 pointless in an attributes file; use path/** instead)
58
59 When deciding what attributes are assigned to a path, Git consults
60 $GIT_DIR/info/attributes file (which has the highest precedence),
61 .gitattributes file in the same directory as the path in question, and
62 its parent directories up to the toplevel of the work tree (the further
63 the directory that contains .gitattributes is from the path in
64 question, the lower its precedence). Finally global and system-wide
65 files are considered (they have the lowest precedence).
66
67 When the .gitattributes file is missing from the work tree, the path in
68 the index is used as a fall-back. During checkout process,
69 .gitattributes in the index is used and then the file in the working
70 tree is used as a fall-back.
71
72 If you wish to affect only a single repository (i.e., to assign
73 attributes to files that are particular to one user’s workflow for that
74 repository), then attributes should be placed in the
75 $GIT_DIR/info/attributes file. Attributes which should be
76 version-controlled and distributed to other repositories (i.e.,
77 attributes of interest to all users) should go into .gitattributes
78 files. Attributes that should affect all repositories for a single user
79 should be placed in a file specified by the core.attributesFile
80 configuration option (see git-config(1)). Its default value is
81 $XDG_CONFIG_HOME/git/attributes. If $XDG_CONFIG_HOME is either not set
82 or empty, $HOME/.config/git/attributes is used instead. Attributes for
83 all users on a system should be placed in the
84 $(prefix)/etc/gitattributes file.
85
86 Sometimes you would need to override a setting of an attribute for a
87 path to Unspecified state. This can be done by listing the name of the
88 attribute prefixed with an exclamation point !.
89
91 Certain operations by Git can be influenced by assigning particular
92 attributes to a path. Currently, the following operations are
93 attributes-aware.
94
95 Checking-out and checking-in
96 These attributes affect how the contents stored in the repository are
97 copied to the working tree files when commands such as git switch, git
98 checkout and git merge run. They also affect how Git stores the
99 contents you prepare in the working tree in the repository upon git add
100 and git commit.
101
102 text
103 This attribute enables and controls end-of-line normalization. When
104 a text file is normalized, its line endings are converted to LF in
105 the repository. To control what line ending style is used in the
106 working directory, use the eol attribute for a single file and the
107 core.eol configuration variable for all text files. Note that
108 setting core.autocrlf to true or input overrides core.eol (see the
109 definitions of those options in git-config(1)).
110
111 Set
112 Setting the text attribute on a path enables end-of-line
113 normalization and marks the path as a text file. End-of-line
114 conversion takes place without guessing the content type.
115
116 Unset
117 Unsetting the text attribute on a path tells Git not to attempt
118 any end-of-line conversion upon checkin or checkout.
119
120 Set to string value "auto"
121 When text is set to "auto", the path is marked for automatic
122 end-of-line conversion. If Git decides that the content is
123 text, its line endings are converted to LF on checkin. When the
124 file has been committed with CRLF, no conversion is done.
125
126 Unspecified
127 If the text attribute is unspecified, Git uses the
128 core.autocrlf configuration variable to determine if the file
129 should be converted.
130
131 Any other value causes Git to act as if text has been left
132 unspecified.
133
134 eol
135 This attribute sets a specific line-ending style to be used in the
136 working directory. It enables end-of-line conversion without any
137 content checks, effectively setting the text attribute. Note that
138 setting this attribute on paths which are in the index with CRLF
139 line endings may make the paths to be considered dirty. Adding the
140 path to the index again will normalize the line endings in the
141 index.
142
143 Set to string value "crlf"
144 This setting forces Git to normalize line endings for this file
145 on checkin and convert them to CRLF when the file is checked
146 out.
147
148 Set to string value "lf"
149 This setting forces Git to normalize line endings to LF on
150 checkin and prevents conversion to CRLF when the file is
151 checked out.
152
153 Backwards compatibility with crlf attribute
154 For backwards compatibility, the crlf attribute is interpreted as
155 follows:
156
157 crlf text
158 -crlf -text
159 crlf=input eol=lf
160
161 End-of-line conversion
162 While Git normally leaves file contents alone, it can be configured
163 to normalize line endings to LF in the repository and, optionally,
164 to convert them to CRLF when files are checked out.
165
166 If you simply want to have CRLF line endings in your working
167 directory regardless of the repository you are working with, you
168 can set the config variable "core.autocrlf" without using any
169 attributes.
170
171 [core]
172 autocrlf = true
173
174 This does not force normalization of text files, but does ensure
175 that text files that you introduce to the repository have their
176 line endings normalized to LF when they are added, and that files
177 that are already normalized in the repository stay normalized.
178
179 If you want to ensure that text files that any contributor
180 introduces to the repository have their line endings normalized,
181 you can set the text attribute to "auto" for all files.
182
183 * text=auto
184
185 The attributes allow a fine-grained control, how the line endings
186 are converted. Here is an example that will make Git normalize
187 .txt, .vcproj and .sh files, ensure that .vcproj files have CRLF
188 and .sh files have LF in the working directory, and prevent .jpg
189 files from being normalized regardless of their content.
190
191 * text=auto
192 *.txt text
193 *.vcproj text eol=crlf
194 *.sh text eol=lf
195 *.jpg -text
196
197
198 Note
199 When text=auto conversion is enabled in a cross-platform
200 project using push and pull to a central repository the text
201 files containing CRLFs should be normalized.
202
203 From a clean working directory:
204
205 $ echo "* text=auto" >.gitattributes
206 $ git add --renormalize .
207 $ git status # Show files that will be normalized
208 $ git commit -m "Introduce end-of-line normalization"
209
210 If any files that should not be normalized show up in git status,
211 unset their text attribute before running git add -u.
212
213 manual.pdf -text
214
215 Conversely, text files that Git does not detect can have
216 normalization enabled manually.
217
218 weirdchars.txt text
219
220 If core.safecrlf is set to "true" or "warn", Git verifies if the
221 conversion is reversible for the current setting of core.autocrlf.
222 For "true", Git rejects irreversible conversions; for "warn", Git
223 only prints a warning but accepts an irreversible conversion. The
224 safety triggers to prevent such a conversion done to the files in
225 the work tree, but there are a few exceptions. Even though...
226
227 • git add itself does not touch the files in the work tree, the
228 next checkout would, so the safety triggers;
229
230 • git apply to update a text file with a patch does touch the
231 files in the work tree, but the operation is about text files
232 and CRLF conversion is about fixing the line ending
233 inconsistencies, so the safety does not trigger;
234
235 • git diff itself does not touch the files in the work tree, it
236 is often run to inspect the changes you intend to next git add.
237 To catch potential problems early, safety triggers.
238
239 working-tree-encoding
240 Git recognizes files encoded in ASCII or one of its supersets (e.g.
241 UTF-8, ISO-8859-1, ...) as text files. Files encoded in certain
242 other encodings (e.g. UTF-16) are interpreted as binary and
243 consequently built-in Git text processing tools (e.g. git diff) as
244 well as most Git web front ends do not visualize the contents of
245 these files by default.
246
247 In these cases you can tell Git the encoding of a file in the
248 working directory with the working-tree-encoding attribute. If a
249 file with this attribute is added to Git, then Git re-encodes the
250 content from the specified encoding to UTF-8. Finally, Git stores
251 the UTF-8 encoded content in its internal data structure (called
252 "the index"). On checkout the content is re-encoded back to the
253 specified encoding.
254
255 Please note that using the working-tree-encoding attribute may have
256 a number of pitfalls:
257
258 • Alternative Git implementations (e.g. JGit or libgit2) and
259 older Git versions (as of March 2018) do not support the
260 working-tree-encoding attribute. If you decide to use the
261 working-tree-encoding attribute in your repository, then it is
262 strongly recommended to ensure that all clients working with
263 the repository support it.
264
265 For example, Microsoft Visual Studio resources files (*.rc) or
266 PowerShell script files (*.ps1) are sometimes encoded in
267 UTF-16. If you declare *.ps1 as files as UTF-16 and you add
268 foo.ps1 with a working-tree-encoding enabled Git client, then
269 foo.ps1 will be stored as UTF-8 internally. A client without
270 working-tree-encoding support will checkout foo.ps1 as UTF-8
271 encoded file. This will typically cause trouble for the users
272 of this file.
273
274 If a Git client that does not support the working-tree-encoding
275 attribute adds a new file bar.ps1, then bar.ps1 will be stored
276 "as-is" internally (in this example probably as UTF-16). A
277 client with working-tree-encoding support will interpret the
278 internal contents as UTF-8 and try to convert it to UTF-16 on
279 checkout. That operation will fail and cause an error.
280
281 • Reencoding content to non-UTF encodings can cause errors as the
282 conversion might not be UTF-8 round trip safe. If you suspect
283 your encoding to not be round trip safe, then add it to
284 core.checkRoundtripEncoding to make Git check the round trip
285 encoding (see git-config(1)). SHIFT-JIS (Japanese character
286 set) is known to have round trip issues with UTF-8 and is
287 checked by default.
288
289 • Reencoding content requires resources that might slow down
290 certain Git operations (e.g git checkout or git add).
291
292 Use the working-tree-encoding attribute only if you cannot store a
293 file in UTF-8 encoding and if you want Git to be able to process
294 the content as text.
295
296 As an example, use the following attributes if your *.ps1 files are
297 UTF-16 encoded with byte order mark (BOM) and you want Git to
298 perform automatic line ending conversion based on your platform.
299
300 *.ps1 text working-tree-encoding=UTF-16
301
302 Use the following attributes if your *.ps1 files are UTF-16 little
303 endian encoded without BOM and you want Git to use Windows line
304 endings in the working directory (use UTF-16LE-BOM instead of
305 UTF-16LE if you want UTF-16 little endian with BOM). Please note,
306 it is highly recommended to explicitly define the line endings with
307 eol if the working-tree-encoding attribute is used to avoid
308 ambiguity.
309
310 *.ps1 text working-tree-encoding=UTF-16LE eol=CRLF
311
312 You can get a list of all available encodings on your platform with
313 the following command:
314
315 iconv --list
316
317 If you do not know the encoding of a file, then you can use the
318 file command to guess the encoding:
319
320 file foo.ps1
321
322 ident
323 When the attribute ident is set for a path, Git replaces $Id$ in
324 the blob object with $Id:, followed by the 40-character hexadecimal
325 blob object name, followed by a dollar sign $ upon checkout. Any
326 byte sequence that begins with $Id: and ends with $ in the worktree
327 file is replaced with $Id$ upon check-in.
328
329 filter
330 A filter attribute can be set to a string value that names a filter
331 driver specified in the configuration.
332
333 A filter driver consists of a clean command and a smudge command,
334 either of which can be left unspecified. Upon checkout, when the
335 smudge command is specified, the command is fed the blob object
336 from its standard input, and its standard output is used to update
337 the worktree file. Similarly, the clean command is used to convert
338 the contents of worktree file upon checkin. By default these
339 commands process only a single blob and terminate. If a long
340 running process filter is used in place of clean and/or smudge
341 filters, then Git can process all blobs with a single filter
342 command invocation for the entire life of a single Git command, for
343 example git add --all. If a long running process filter is
344 configured then it always takes precedence over a configured single
345 blob filter. See section below for the description of the protocol
346 used to communicate with a process filter.
347
348 One use of the content filtering is to massage the content into a
349 shape that is more convenient for the platform, filesystem, and the
350 user to use. For this mode of operation, the key phrase here is
351 "more convenient" and not "turning something unusable into usable".
352 In other words, the intent is that if someone unsets the filter
353 driver definition, or does not have the appropriate filter program,
354 the project should still be usable.
355
356 Another use of the content filtering is to store the content that
357 cannot be directly used in the repository (e.g. a UUID that refers
358 to the true content stored outside Git, or an encrypted content)
359 and turn it into a usable form upon checkout (e.g. download the
360 external content, or decrypt the encrypted content).
361
362 These two filters behave differently, and by default, a filter is
363 taken as the former, massaging the contents into more convenient
364 shape. A missing filter driver definition in the config, or a
365 filter driver that exits with a non-zero status, is not an error
366 but makes the filter a no-op passthru.
367
368 You can declare that a filter turns a content that by itself is
369 unusable into a usable content by setting the
370 filter.<driver>.required configuration variable to true.
371
372 Note: Whenever the clean filter is changed, the repo should be
373 renormalized: $ git add --renormalize .
374
375 For example, in .gitattributes, you would assign the filter
376 attribute for paths.
377
378 *.c filter=indent
379
380 Then you would define a "filter.indent.clean" and
381 "filter.indent.smudge" configuration in your .git/config to specify
382 a pair of commands to modify the contents of C programs when the
383 source files are checked in ("clean" is run) and checked out (no
384 change is made because the command is "cat").
385
386 [filter "indent"]
387 clean = indent
388 smudge = cat
389
390 For best results, clean should not alter its output further if it
391 is run twice ("clean→clean" should be equivalent to "clean"), and
392 multiple smudge commands should not alter clean's output
393 ("smudge→smudge→clean" should be equivalent to "clean"). See the
394 section on merging below.
395
396 The "indent" filter is well-behaved in this regard: it will not
397 modify input that is already correctly indented. In this case, the
398 lack of a smudge filter means that the clean filter must accept its
399 own output without modifying it.
400
401 If a filter must succeed in order to make the stored contents
402 usable, you can declare that the filter is required, in the
403 configuration:
404
405 [filter "crypt"]
406 clean = openssl enc ...
407 smudge = openssl enc -d ...
408 required
409
410 Sequence "%f" on the filter command line is replaced with the name
411 of the file the filter is working on. A filter might use this in
412 keyword substitution. For example:
413
414 [filter "p4"]
415 clean = git-p4-filter --clean %f
416 smudge = git-p4-filter --smudge %f
417
418 Note that "%f" is the name of the path that is being worked on.
419 Depending on the version that is being filtered, the corresponding
420 file on disk may not exist, or may have different contents. So,
421 smudge and clean commands should not try to access the file on
422 disk, but only act as filters on the content provided to them on
423 standard input.
424
425 Long Running Filter Process
426 If the filter command (a string value) is defined via
427 filter.<driver>.process then Git can process all blobs with a
428 single filter invocation for the entire life of a single Git
429 command. This is achieved by using the long-running process
430 protocol (described in
431 technical/long-running-process-protocol.txt).
432
433 When Git encounters the first file that needs to be cleaned or
434 smudged, it starts the filter and performs the handshake. In the
435 handshake, the welcome message sent by Git is "git-filter-client",
436 only version 2 is supported, and the supported capabilities are
437 "clean", "smudge", and "delay".
438
439 Afterwards Git sends a list of "key=value" pairs terminated with a
440 flush packet. The list will contain at least the filter command
441 (based on the supported capabilities) and the pathname of the file
442 to filter relative to the repository root. Right after the flush
443 packet Git sends the content split in zero or more pkt-line packets
444 and a flush packet to terminate content. Please note, that the
445 filter must not send any response before it received the content
446 and the final flush packet. Also note that the "value" of a
447 "key=value" pair can contain the "=" character whereas the key
448 would never contain that character.
449
450 packet: git> command=smudge
451 packet: git> pathname=path/testfile.dat
452 packet: git> 0000
453 packet: git> CONTENT
454 packet: git> 0000
455
456 The filter is expected to respond with a list of "key=value" pairs
457 terminated with a flush packet. If the filter does not experience
458 problems then the list must contain a "success" status. Right after
459 these packets the filter is expected to send the content in zero or
460 more pkt-line packets and a flush packet at the end. Finally, a
461 second list of "key=value" pairs terminated with a flush packet is
462 expected. The filter can change the status in the second list or
463 keep the status as is with an empty list. Please note that the
464 empty list must be terminated with a flush packet regardless.
465
466 packet: git< status=success
467 packet: git< 0000
468 packet: git< SMUDGED_CONTENT
469 packet: git< 0000
470 packet: git< 0000 # empty list, keep "status=success" unchanged!
471
472 If the result content is empty then the filter is expected to
473 respond with a "success" status and a flush packet to signal the
474 empty content.
475
476 packet: git< status=success
477 packet: git< 0000
478 packet: git< 0000 # empty content!
479 packet: git< 0000 # empty list, keep "status=success" unchanged!
480
481 In case the filter cannot or does not want to process the content,
482 it is expected to respond with an "error" status.
483
484 packet: git< status=error
485 packet: git< 0000
486
487 If the filter experiences an error during processing, then it can
488 send the status "error" after the content was (partially or
489 completely) sent.
490
491 packet: git< status=success
492 packet: git< 0000
493 packet: git< HALF_WRITTEN_ERRONEOUS_CONTENT
494 packet: git< 0000
495 packet: git< status=error
496 packet: git< 0000
497
498 In case the filter cannot or does not want to process the content
499 as well as any future content for the lifetime of the Git process,
500 then it is expected to respond with an "abort" status at any point
501 in the protocol.
502
503 packet: git< status=abort
504 packet: git< 0000
505
506 Git neither stops nor restarts the filter process in case the
507 "error"/"abort" status is set. However, Git sets its exit code
508 according to the filter.<driver>.required flag, mimicking the
509 behavior of the filter.<driver>.clean / filter.<driver>.smudge
510 mechanism.
511
512 If the filter dies during the communication or does not adhere to
513 the protocol then Git will stop the filter process and restart it
514 with the next file that needs to be processed. Depending on the
515 filter.<driver>.required flag Git will interpret that as error.
516
517 Delay
518 If the filter supports the "delay" capability, then Git can send
519 the flag "can-delay" after the filter command and pathname. This
520 flag denotes that the filter can delay filtering the current blob
521 (e.g. to compensate network latencies) by responding with no
522 content but with the status "delayed" and a flush packet.
523
524 packet: git> command=smudge
525 packet: git> pathname=path/testfile.dat
526 packet: git> can-delay=1
527 packet: git> 0000
528 packet: git> CONTENT
529 packet: git> 0000
530 packet: git< status=delayed
531 packet: git< 0000
532
533 If the filter supports the "delay" capability then it must support
534 the "list_available_blobs" command. If Git sends this command, then
535 the filter is expected to return a list of pathnames representing
536 blobs that have been delayed earlier and are now available. The
537 list must be terminated with a flush packet followed by a "success"
538 status that is also terminated with a flush packet. If no blobs for
539 the delayed paths are available, yet, then the filter is expected
540 to block the response until at least one blob becomes available.
541 The filter can tell Git that it has no more delayed blobs by
542 sending an empty list. As soon as the filter responds with an empty
543 list, Git stops asking. All blobs that Git has not received at this
544 point are considered missing and will result in an error.
545
546 packet: git> command=list_available_blobs
547 packet: git> 0000
548 packet: git< pathname=path/testfile.dat
549 packet: git< pathname=path/otherfile.dat
550 packet: git< 0000
551 packet: git< status=success
552 packet: git< 0000
553
554 After Git received the pathnames, it will request the corresponding
555 blobs again. These requests contain a pathname and an empty content
556 section. The filter is expected to respond with the smudged content
557 in the usual way as explained above.
558
559 packet: git> command=smudge
560 packet: git> pathname=path/testfile.dat
561 packet: git> 0000
562 packet: git> 0000 # empty content!
563 packet: git< status=success
564 packet: git< 0000
565 packet: git< SMUDGED_CONTENT
566 packet: git< 0000
567 packet: git< 0000 # empty list, keep "status=success" unchanged!
568
569 Example
570 A long running filter demo implementation can be found in
571 contrib/long-running-filter/example.pl located in the Git core
572 repository. If you develop your own long running filter process
573 then the GIT_TRACE_PACKET environment variables can be very helpful
574 for debugging (see git(1)).
575
576 Please note that you cannot use an existing filter.<driver>.clean
577 or filter.<driver>.smudge command with filter.<driver>.process
578 because the former two use a different inter process communication
579 protocol than the latter one.
580
581 Interaction between checkin/checkout attributes
582 In the check-in codepath, the worktree file is first converted with
583 filter driver (if specified and corresponding driver defined), then
584 the result is processed with ident (if specified), and then finally
585 with text (again, if specified and applicable).
586
587 In the check-out codepath, the blob content is first converted with
588 text, and then ident and fed to filter.
589
590 Merging branches with differing checkin/checkout attributes
591 If you have added attributes to a file that cause the canonical
592 repository format for that file to change, such as adding a
593 clean/smudge filter or text/eol/ident attributes, merging anything
594 where the attribute is not in place would normally cause merge
595 conflicts.
596
597 To prevent these unnecessary merge conflicts, Git can be told to
598 run a virtual check-out and check-in of all three stages of a file
599 when resolving a three-way merge by setting the merge.renormalize
600 configuration variable. This prevents changes caused by check-in
601 conversion from causing spurious merge conflicts when a converted
602 file is merged with an unconverted file.
603
604 As long as a "smudge→clean" results in the same output as a "clean"
605 even on files that are already smudged, this strategy will
606 automatically resolve all filter-related conflicts. Filters that do
607 not act in this way may cause additional merge conflicts that must
608 be resolved manually.
609
610 Generating diff text
611 diff
612 The attribute diff affects how Git generates diffs for particular
613 files. It can tell Git whether to generate a textual patch for the
614 path or to treat the path as a binary file. It can also affect what
615 line is shown on the hunk header @@ -k,l +n,m @@ line, tell Git to
616 use an external command to generate the diff, or ask Git to convert
617 binary files to a text format before generating the diff.
618
619 Set
620 A path to which the diff attribute is set is treated as text,
621 even when they contain byte values that normally never appear
622 in text files, such as NUL.
623
624 Unset
625 A path to which the diff attribute is unset will generate
626 Binary files differ (or a binary patch, if binary patches are
627 enabled).
628
629 Unspecified
630 A path to which the diff attribute is unspecified first gets
631 its contents inspected, and if it looks like text and is
632 smaller than core.bigFileThreshold, it is treated as text.
633 Otherwise it would generate Binary files differ.
634
635 String
636 Diff is shown using the specified diff driver. Each driver may
637 specify one or more options, as described in the following
638 section. The options for the diff driver "foo" are defined by
639 the configuration variables in the "diff.foo" section of the
640 Git config file.
641
642 Defining an external diff driver
643 The definition of a diff driver is done in gitconfig, not
644 gitattributes file, so strictly speaking this manual page is a
645 wrong place to talk about it. However...
646
647 To define an external diff driver jcdiff, add a section to your
648 $GIT_DIR/config file (or $HOME/.gitconfig file) like this:
649
650 [diff "jcdiff"]
651 command = j-c-diff
652
653 When Git needs to show you a diff for the path with diff attribute
654 set to jcdiff, it calls the command you specified with the above
655 configuration, i.e. j-c-diff, with 7 parameters, just like
656 GIT_EXTERNAL_DIFF program is called. See git(1) for details.
657
658 Defining a custom hunk-header
659 Each group of changes (called a "hunk") in the textual diff output
660 is prefixed with a line of the form:
661
662 @@ -k,l +n,m @@ TEXT
663
664 This is called a hunk header. The "TEXT" portion is by default a
665 line that begins with an alphabet, an underscore or a dollar sign;
666 this matches what GNU diff -p output uses. This default selection
667 however is not suited for some contents, and you can use a
668 customized pattern to make a selection.
669
670 First, in .gitattributes, you would assign the diff attribute for
671 paths.
672
673 *.tex diff=tex
674
675 Then, you would define a "diff.tex.xfuncname" configuration to
676 specify a regular expression that matches a line that you would
677 want to appear as the hunk header "TEXT". Add a section to your
678 $GIT_DIR/config file (or $HOME/.gitconfig file) like this:
679
680 [diff "tex"]
681 xfuncname = "^(\\\\(sub)*section\\{.*)$"
682
683 Note. A single level of backslashes are eaten by the configuration
684 file parser, so you would need to double the backslashes; the
685 pattern above picks a line that begins with a backslash, and zero
686 or more occurrences of sub followed by section followed by open
687 brace, to the end of line.
688
689 There are a few built-in patterns to make this easier, and tex is
690 one of them, so you do not have to write the above in your
691 configuration file (you still need to enable this with the
692 attribute mechanism, via .gitattributes). The following built in
693 patterns are available:
694
695 • ada suitable for source code in the Ada language.
696
697 • bash suitable for source code in the Bourne-Again SHell
698 language. Covers a superset of POSIX shell function
699 definitions.
700
701 • bibtex suitable for files with BibTeX coded references.
702
703 • cpp suitable for source code in the C and C++ languages.
704
705 • csharp suitable for source code in the C# language.
706
707 • css suitable for cascading style sheets.
708
709 • dts suitable for devicetree (DTS) files.
710
711 • elixir suitable for source code in the Elixir language.
712
713 • fortran suitable for source code in the Fortran language.
714
715 • fountain suitable for Fountain documents.
716
717 • golang suitable for source code in the Go language.
718
719 • html suitable for HTML/XHTML documents.
720
721 • java suitable for source code in the Java language.
722
723 • markdown suitable for Markdown documents.
724
725 • matlab suitable for source code in the MATLAB and Octave
726 languages.
727
728 • objc suitable for source code in the Objective-C language.
729
730 • pascal suitable for source code in the Pascal/Delphi language.
731
732 • perl suitable for source code in the Perl language.
733
734 • php suitable for source code in the PHP language.
735
736 • python suitable for source code in the Python language.
737
738 • ruby suitable for source code in the Ruby language.
739
740 • rust suitable for source code in the Rust language.
741
742 • scheme suitable for source code in the Scheme language.
743
744 • tex suitable for source code for LaTeX documents.
745
746 Customizing word diff
747 You can customize the rules that git diff --word-diff uses to split
748 words in a line, by specifying an appropriate regular expression in
749 the "diff.*.wordRegex" configuration variable. For example, in TeX
750 a backslash followed by a sequence of letters forms a command, but
751 several such commands can be run together without intervening
752 whitespace. To separate them, use a regular expression in your
753 $GIT_DIR/config file (or $HOME/.gitconfig file) like this:
754
755 [diff "tex"]
756 wordRegex = "\\\\[a-zA-Z]+|[{}]|\\\\.|[^\\{}[:space:]]+"
757
758 A built-in pattern is provided for all languages listed in the
759 previous section.
760
761 Performing text diffs of binary files
762 Sometimes it is desirable to see the diff of a text-converted
763 version of some binary files. For example, a word processor
764 document can be converted to an ASCII text representation, and the
765 diff of the text shown. Even though this conversion loses some
766 information, the resulting diff is useful for human viewing (but
767 cannot be applied directly).
768
769 The textconv config option is used to define a program for
770 performing such a conversion. The program should take a single
771 argument, the name of a file to convert, and produce the resulting
772 text on stdout.
773
774 For example, to show the diff of the exif information of a file
775 instead of the binary information (assuming you have the exif tool
776 installed), add the following section to your $GIT_DIR/config file
777 (or $HOME/.gitconfig file):
778
779 [diff "jpg"]
780 textconv = exif
781
782
783 Note
784 The text conversion is generally a one-way conversion; in this
785 example, we lose the actual image contents and focus just on
786 the text data. This means that diffs generated by textconv are
787 not suitable for applying. For this reason, only git diff and
788 the git log family of commands (i.e., log, whatchanged, show)
789 will perform text conversion. git format-patch will never
790 generate this output. If you want to send somebody a
791 text-converted diff of a binary file (e.g., because it quickly
792 conveys the changes you have made), you should generate it
793 separately and send it as a comment in addition to the usual
794 binary diff that you might send.
795
796 Because text conversion can be slow, especially when doing a large
797 number of them with git log -p, Git provides a mechanism to cache
798 the output and use it in future diffs. To enable caching, set the
799 "cachetextconv" variable in your diff driver’s config. For example:
800
801 [diff "jpg"]
802 textconv = exif
803 cachetextconv = true
804
805 This will cache the result of running "exif" on each blob
806 indefinitely. If you change the textconv config variable for a diff
807 driver, Git will automatically invalidate the cache entries and
808 re-run the textconv filter. If you want to invalidate the cache
809 manually (e.g., because your version of "exif" was updated and now
810 produces better output), you can remove the cache manually with git
811 update-ref -d refs/notes/textconv/jpg (where "jpg" is the name of
812 the diff driver, as in the example above).
813
814 Choosing textconv versus external diff
815 If you want to show differences between binary or
816 specially-formatted blobs in your repository, you can choose to use
817 either an external diff command, or to use textconv to convert them
818 to a diff-able text format. Which method you choose depends on your
819 exact situation.
820
821 The advantage of using an external diff command is flexibility. You
822 are not bound to find line-oriented changes, nor is it necessary
823 for the output to resemble unified diff. You are free to locate and
824 report changes in the most appropriate way for your data format.
825
826 A textconv, by comparison, is much more limiting. You provide a
827 transformation of the data into a line-oriented text format, and
828 Git uses its regular diff tools to generate the output. There are
829 several advantages to choosing this method:
830
831 1. Ease of use. It is often much simpler to write a binary to text
832 transformation than it is to perform your own diff. In many
833 cases, existing programs can be used as textconv filters (e.g.,
834 exif, odt2txt).
835
836 2. Git diff features. By performing only the transformation step
837 yourself, you can still utilize many of Git’s diff features,
838 including colorization, word-diff, and combined diffs for
839 merges.
840
841 3. Caching. Textconv caching can speed up repeated diffs, such as
842 those you might trigger by running git log -p.
843
844 Marking files as binary
845 Git usually guesses correctly whether a blob contains text or
846 binary data by examining the beginning of the contents. However,
847 sometimes you may want to override its decision, either because a
848 blob contains binary data later in the file, or because the
849 content, while technically composed of text characters, is opaque
850 to a human reader. For example, many postscript files contain only
851 ASCII characters, but produce noisy and meaningless diffs.
852
853 The simplest way to mark a file as binary is to unset the diff
854 attribute in the .gitattributes file:
855
856 *.ps -diff
857
858 This will cause Git to generate Binary files differ (or a binary
859 patch, if binary patches are enabled) instead of a regular diff.
860
861 However, one may also want to specify other diff driver attributes.
862 For example, you might want to use textconv to convert postscript
863 files to an ASCII representation for human viewing, but otherwise
864 treat them as binary files. You cannot specify both -diff and
865 diff=ps attributes. The solution is to use the diff.*.binary config
866 option:
867
868 [diff "ps"]
869 textconv = ps2ascii
870 binary = true
871
872 Performing a three-way merge
873 merge
874 The attribute merge affects how three versions of a file are merged
875 when a file-level merge is necessary during git merge, and other
876 commands such as git revert and git cherry-pick.
877
878 Set
879 Built-in 3-way merge driver is used to merge the contents in a
880 way similar to merge command of RCS suite. This is suitable for
881 ordinary text files.
882
883 Unset
884 Take the version from the current branch as the tentative merge
885 result, and declare that the merge has conflicts. This is
886 suitable for binary files that do not have a well-defined merge
887 semantics.
888
889 Unspecified
890 By default, this uses the same built-in 3-way merge driver as
891 is the case when the merge attribute is set. However, the
892 merge.default configuration variable can name different merge
893 driver to be used with paths for which the merge attribute is
894 unspecified.
895
896 String
897 3-way merge is performed using the specified custom merge
898 driver. The built-in 3-way merge driver can be explicitly
899 specified by asking for "text" driver; the built-in "take the
900 current branch" driver can be requested with "binary".
901
902 Built-in merge drivers
903 There are a few built-in low-level merge drivers defined that can
904 be asked for via the merge attribute.
905
906 text
907 Usual 3-way file level merge for text files. Conflicted regions
908 are marked with conflict markers <<<<<<<, ======= and >>>>>>>.
909 The version from your branch appears before the ======= marker,
910 and the version from the merged branch appears after the
911 ======= marker.
912
913 binary
914 Keep the version from your branch in the work tree, but leave
915 the path in the conflicted state for the user to sort out.
916
917 union
918 Run 3-way file level merge for text files, but take lines from
919 both versions, instead of leaving conflict markers. This tends
920 to leave the added lines in the resulting file in random order
921 and the user should verify the result. Do not use this if you
922 do not understand the implications.
923
924 Defining a custom merge driver
925 The definition of a merge driver is done in the .git/config file,
926 not in the gitattributes file, so strictly speaking this manual
927 page is a wrong place to talk about it. However...
928
929 To define a custom merge driver filfre, add a section to your
930 $GIT_DIR/config file (or $HOME/.gitconfig file) like this:
931
932 [merge "filfre"]
933 name = feel-free merge driver
934 driver = filfre %O %A %B %L %P
935 recursive = binary
936
937 The merge.*.name variable gives the driver a human-readable name.
938
939 The merge.*.driver variable’s value is used to construct a command
940 to run to merge ancestor’s version (%O), current version (%A) and
941 the other branches' version (%B). These three tokens are replaced
942 with the names of temporary files that hold the contents of these
943 versions when the command line is built. Additionally, %L will be
944 replaced with the conflict marker size (see below).
945
946 The merge driver is expected to leave the result of the merge in
947 the file named with %A by overwriting it, and exit with zero status
948 if it managed to merge them cleanly, or non-zero if there were
949 conflicts.
950
951 The merge.*.recursive variable specifies what other merge driver to
952 use when the merge driver is called for an internal merge between
953 common ancestors, when there are more than one. When left
954 unspecified, the driver itself is used for both internal merge and
955 the final merge.
956
957 The merge driver can learn the pathname in which the merged result
958 will be stored via placeholder %P.
959
960 conflict-marker-size
961 This attribute controls the length of conflict markers left in the
962 work tree file during a conflicted merge. Only setting to the value
963 to a positive integer has any meaningful effect.
964
965 For example, this line in .gitattributes can be used to tell the
966 merge machinery to leave much longer (instead of the usual
967 7-character-long) conflict markers when merging the file
968 Documentation/git-merge.txt results in a conflict.
969
970 Documentation/git-merge.txt conflict-marker-size=32
971
972 Checking whitespace errors
973 whitespace
974 The core.whitespace configuration variable allows you to define
975 what diff and apply should consider whitespace errors for all paths
976 in the project (See git-config(1)). This attribute gives you finer
977 control per path.
978
979 Set
980 Notice all types of potential whitespace errors known to Git.
981 The tab width is taken from the value of the core.whitespace
982 configuration variable.
983
984 Unset
985 Do not notice anything as error.
986
987 Unspecified
988 Use the value of the core.whitespace configuration variable to
989 decide what to notice as error.
990
991 String
992 Specify a comma separate list of common whitespace problems to
993 notice in the same format as the core.whitespace configuration
994 variable.
995
996 Creating an archive
997 export-ignore
998 Files and directories with the attribute export-ignore won’t be
999 added to archive files.
1000
1001 export-subst
1002 If the attribute export-subst is set for a file then Git will
1003 expand several placeholders when adding this file to an archive.
1004 The expansion depends on the availability of a commit ID, i.e., if
1005 git-archive(1) has been given a tree instead of a commit or a tag
1006 then no replacement will be done. The placeholders are the same as
1007 those for the option --pretty=format: of git-log(1), except that
1008 they need to be wrapped like this: $Format:PLACEHOLDERS$ in the
1009 file. E.g. the string $Format:%H$ will be replaced by the commit
1010 hash. However, only one %(describe) placeholder is expanded per
1011 archive to avoid denial-of-service attacks.
1012
1013 Packing objects
1014 delta
1015 Delta compression will not be attempted for blobs for paths with
1016 the attribute delta set to false.
1017
1018 Viewing files in GUI tools
1019 encoding
1020 The value of this attribute specifies the character encoding that
1021 should be used by GUI tools (e.g. gitk(1) and git-gui(1)) to
1022 display the contents of the relevant file. Note that due to
1023 performance considerations gitk(1) does not use this attribute
1024 unless you manually enable per-file encodings in its options.
1025
1026 If this attribute is not set or has an invalid value, the value of
1027 the gui.encoding configuration variable is used instead (See git-
1028 config(1)).
1029
1031 You do not want any end-of-line conversions applied to, nor textual
1032 diffs produced for, any binary file you track. You would need to
1033 specify e.g.
1034
1035 *.jpg -text -diff
1036
1037 but that may become cumbersome, when you have many attributes. Using
1038 macro attributes, you can define an attribute that, when set, also sets
1039 or unsets a number of other attributes at the same time. The system
1040 knows a built-in macro attribute, binary:
1041
1042 *.jpg binary
1043
1044 Setting the "binary" attribute also unsets the "text" and "diff"
1045 attributes as above. Note that macro attributes can only be "Set",
1046 though setting one might have the effect of setting or unsetting other
1047 attributes or even returning other attributes to the "Unspecified"
1048 state.
1049
1051 Custom macro attributes can be defined only in top-level gitattributes
1052 files ($GIT_DIR/info/attributes, the .gitattributes file at the top
1053 level of the working tree, or the global or system-wide gitattributes
1054 files), not in .gitattributes files in working tree subdirectories. The
1055 built-in macro attribute "binary" is equivalent to:
1056
1057 [attr]binary -diff -merge -text
1058
1060 Git does not follow symbolic links when accessing a .gitattributes file
1061 in the working tree. This keeps behavior consistent when the file is
1062 accessed from the index or a tree versus from the filesystem.
1063
1065 If you have these three gitattributes file:
1066
1067 (in $GIT_DIR/info/attributes)
1068
1069 a* foo !bar -baz
1070
1071 (in .gitattributes)
1072 abc foo bar baz
1073
1074 (in t/.gitattributes)
1075 ab* merge=filfre
1076 abc -foo -bar
1077 *.c frotz
1078
1079 the attributes given to path t/abc are computed as follows:
1080
1081 1. By examining t/.gitattributes (which is in the same directory as
1082 the path in question), Git finds that the first line matches.
1083 merge attribute is set. It also finds that the second line matches,
1084 and attributes foo and bar are unset.
1085
1086 2. Then it examines .gitattributes (which is in the parent directory),
1087 and finds that the first line matches, but t/.gitattributes file
1088 already decided how merge, foo and bar attributes should be given
1089 to this path, so it leaves foo and bar unset. Attribute baz is set.
1090
1091 3. Finally it examines $GIT_DIR/info/attributes. This file is used to
1092 override the in-tree settings. The first line is a match, and foo
1093 is set, bar is reverted to unspecified state, and baz is unset.
1094
1095 As the result, the attributes assignment to t/abc becomes:
1096
1097 foo set to true
1098 bar unspecified
1099 baz set to false
1100 merge set to string value "filfre"
1101 frotz unspecified
1102
1104 git-check-attr(1).
1105
1107 Part of the git(1) suite
1108
1109
1110
1111Git 2.33.1 2021-10-12 GITATTRIBUTES(5)