6 gitattributes - Defining attributes per path
10 $GIT_DIR/info/attributes, .gitattributes
16 A `gitattributes` file is a simple text file that gives
17 `attributes` to pathnames.
19 Each line in `gitattributes` file is of form:
21 pattern attr1 attr2 ...
23 That is, a pattern followed by an attributes list,
24 separated by whitespaces. Leading and trailing whitespaces are
25 ignored. Lines that begin with '#' are ignored. Patterns
26 that begin with a double quote are quoted in C style.
27 When the pattern matches the path in question, the attributes
28 listed on the line are given to the path.
30 Each attribute can be in one of these states for a given path:
34 The path has the attribute with special value "true";
35 this is specified by listing only the name of the
36 attribute in the attribute list.
40 The path has the attribute with special value "false";
41 this is specified by listing the name of the attribute
42 prefixed with a dash `-` in the attribute list.
46 The path has the attribute with specified string value;
47 this is specified by listing the name of the attribute
48 followed by an equal sign `=` and its value in the
53 No pattern matches the path, and nothing says if
54 the path has or does not have the attribute, the
55 attribute for the path is said to be Unspecified.
57 When more than one pattern matches the path, a later line
58 overrides an earlier line. This overriding is done per
61 The rules by which the pattern matches paths are the same as in
62 `.gitignore` files (see linkgit:gitignore[5]), with a few exceptions:
64 - negative patterns are forbidden
66 - patterns that match a directory do not recursively match paths
67 inside that directory (so using the trailing-slash `path/` syntax is
68 pointless in an attributes file; use `path/**` instead)
70 When deciding what attributes are assigned to a path, Git
71 consults `$GIT_DIR/info/attributes` file (which has the highest
72 precedence), `.gitattributes` file in the same directory as the
73 path in question, and its parent directories up to the toplevel of the
74 work tree (the further the directory that contains `.gitattributes`
75 is from the path in question, the lower its precedence). Finally
76 global and system-wide files are considered (they have the lowest
79 When the `.gitattributes` file is missing from the work tree, the
80 path in the index is used as a fall-back. During checkout process,
81 `.gitattributes` in the index is used and then the file in the
82 working tree is used as a fall-back.
84 If you wish to affect only a single repository (i.e., to assign
85 attributes to files that are particular to
86 one user's workflow for that repository), then
87 attributes should be placed in the `$GIT_DIR/info/attributes` file.
88 Attributes which should be version-controlled and distributed to other
89 repositories (i.e., attributes of interest to all users) should go into
90 `.gitattributes` files. Attributes that should affect all repositories
91 for a single user should be placed in a file specified by the
92 `core.attributesFile` configuration option (see linkgit:git-config[1]).
93 Its default value is $XDG_CONFIG_HOME/git/attributes. If $XDG_CONFIG_HOME
94 is either not set or empty, $HOME/.config/git/attributes is used instead.
95 Attributes for all users on a system should be placed in the
96 `$(prefix)/etc/gitattributes` file.
98 Sometimes you would need to override a setting of an attribute
99 for a path to `Unspecified` state. This can be done by listing
100 the name of the attribute prefixed with an exclamation point `!`.
106 Certain operations by Git can be influenced by assigning
107 particular attributes to a path. Currently, the following
108 operations are attributes-aware.
110 Checking-out and checking-in
111 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
113 These attributes affect how the contents stored in the
114 repository are copied to the working tree files when commands
115 such as 'git switch', 'git checkout' and 'git merge' run.
117 Git stores the contents you prepare in the working tree in the
118 repository upon 'git add' and 'git commit'.
123 This attribute enables and controls end-of-line normalization. When a
124 text file is normalized, its line endings are converted to LF in the
125 repository. To control what line ending style is used in the working
126 directory, use the `eol` attribute for a single file and the
127 `core.eol` configuration variable for all text files.
128 Note that setting `core.autocrlf` to `true` or `input` overrides
129 `core.eol` (see the definitions of those options in
130 linkgit:git-config[1]).
134 Setting the `text` attribute on a path enables end-of-line
135 normalization and marks the path as a text file. End-of-line
136 conversion takes place without guessing the content type.
140 Unsetting the `text` attribute on a path tells Git not to
141 attempt any end-of-line conversion upon checkin or checkout.
143 Set to string value "auto"::
145 When `text` is set to "auto", the path is marked for automatic
146 end-of-line conversion. If Git decides that the content is
147 text, its line endings are converted to LF on checkin.
148 When the file has been committed with CRLF, no conversion is done.
152 If the `text` attribute is unspecified, Git uses the
153 `core.autocrlf` configuration variable to determine if the
154 file should be converted.
156 Any other value causes Git to act as if `text` has been left
162 This attribute sets a specific line-ending style to be used in the
163 working directory. This attribute has effect only if the `text`
164 attribute is set or unspecified, or if it is set to `auto` and the file
165 is detected as text. Note that setting this attribute on paths which
166 are in the index with CRLF line endings may make the paths to be
167 considered dirty. Adding the path to the index again will normalize the
168 line endings in the index.
170 Set to string value "crlf"::
172 This setting forces Git to normalize line endings for this
173 file on checkin and convert them to CRLF when the file is
176 Set to string value "lf"::
178 This setting forces Git to normalize line endings to LF on
179 checkin and prevents conversion to CRLF when the file is
182 Backwards compatibility with `crlf` attribute
183 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
185 For backwards compatibility, the `crlf` attribute is interpreted as
188 ------------------------
192 ------------------------
194 End-of-line conversion
195 ^^^^^^^^^^^^^^^^^^^^^^
197 While Git normally leaves file contents alone, it can be configured to
198 normalize line endings to LF in the repository and, optionally, to
199 convert them to CRLF when files are checked out.
201 If you simply want to have CRLF line endings in your working directory
202 regardless of the repository you are working with, you can set the
203 config variable "core.autocrlf" without using any attributes.
205 ------------------------
208 ------------------------
210 This does not force normalization of text files, but does ensure
211 that text files that you introduce to the repository have their line
212 endings normalized to LF when they are added, and that files that are
213 already normalized in the repository stay normalized.
215 If you want to ensure that text files that any contributor introduces to
216 the repository have their line endings normalized, you can set the
217 `text` attribute to "auto" for _all_ files.
219 ------------------------
221 ------------------------
223 The attributes allow a fine-grained control, how the line endings
225 Here is an example that will make Git normalize .txt, .vcproj and .sh
226 files, ensure that .vcproj files have CRLF and .sh files have LF in
227 the working directory, and prevent .jpg files from being normalized
228 regardless of their content.
230 ------------------------
233 *.vcproj text eol=crlf
236 ------------------------
238 NOTE: When `text=auto` conversion is enabled in a cross-platform
239 project using push and pull to a central repository the text files
240 containing CRLFs should be normalized.
242 From a clean working directory:
244 -------------------------------------------------
245 $ echo "* text=auto" >.gitattributes
246 $ git add --renormalize .
247 $ git status # Show files that will be normalized
248 $ git commit -m "Introduce end-of-line normalization"
249 -------------------------------------------------
251 If any files that should not be normalized show up in 'git status',
252 unset their `text` attribute before running 'git add -u'.
254 ------------------------
256 ------------------------
258 Conversely, text files that Git does not detect can have normalization
261 ------------------------
263 ------------------------
265 If `core.safecrlf` is set to "true" or "warn", Git verifies if
266 the conversion is reversible for the current setting of
267 `core.autocrlf`. For "true", Git rejects irreversible
268 conversions; for "warn", Git only prints a warning but accepts
269 an irreversible conversion. The safety triggers to prevent such
270 a conversion done to the files in the work tree, but there are a
271 few exceptions. Even though...
273 - 'git add' itself does not touch the files in the work tree, the
274 next checkout would, so the safety triggers;
276 - 'git apply' to update a text file with a patch does touch the files
277 in the work tree, but the operation is about text files and CRLF
278 conversion is about fixing the line ending inconsistencies, so the
279 safety does not trigger;
281 - 'git diff' itself does not touch the files in the work tree, it is
282 often run to inspect the changes you intend to next 'git add'. To
283 catch potential problems early, safety triggers.
286 `working-tree-encoding`
287 ^^^^^^^^^^^^^^^^^^^^^^^
289 Git recognizes files encoded in ASCII or one of its supersets (e.g.
290 UTF-8, ISO-8859-1, ...) as text files. Files encoded in certain other
291 encodings (e.g. UTF-16) are interpreted as binary and consequently
292 built-in Git text processing tools (e.g. 'git diff') as well as most Git
293 web front ends do not visualize the contents of these files by default.
295 In these cases you can tell Git the encoding of a file in the working
296 directory with the `working-tree-encoding` attribute. If a file with this
297 attribute is added to Git, then Git re-encodes the content from the
298 specified encoding to UTF-8. Finally, Git stores the UTF-8 encoded
299 content in its internal data structure (called "the index"). On checkout
300 the content is re-encoded back to the specified encoding.
302 Please note that using the `working-tree-encoding` attribute may have a
305 - Alternative Git implementations (e.g. JGit or libgit2) and older Git
306 versions (as of March 2018) do not support the `working-tree-encoding`
307 attribute. If you decide to use the `working-tree-encoding` attribute
308 in your repository, then it is strongly recommended to ensure that all
309 clients working with the repository support it.
311 For example, Microsoft Visual Studio resources files (`*.rc`) or
312 PowerShell script files (`*.ps1`) are sometimes encoded in UTF-16.
313 If you declare `*.ps1` as files as UTF-16 and you add `foo.ps1` with
314 a `working-tree-encoding` enabled Git client, then `foo.ps1` will be
315 stored as UTF-8 internally. A client without `working-tree-encoding`
316 support will checkout `foo.ps1` as UTF-8 encoded file. This will
317 typically cause trouble for the users of this file.
319 If a Git client that does not support the `working-tree-encoding`
320 attribute adds a new file `bar.ps1`, then `bar.ps1` will be
321 stored "as-is" internally (in this example probably as UTF-16).
322 A client with `working-tree-encoding` support will interpret the
323 internal contents as UTF-8 and try to convert it to UTF-16 on checkout.
324 That operation will fail and cause an error.
326 - Reencoding content to non-UTF encodings can cause errors as the
327 conversion might not be UTF-8 round trip safe. If you suspect your
328 encoding to not be round trip safe, then add it to
329 `core.checkRoundtripEncoding` to make Git check the round trip
330 encoding (see linkgit:git-config[1]). SHIFT-JIS (Japanese character
331 set) is known to have round trip issues with UTF-8 and is checked by
334 - Reencoding content requires resources that might slow down certain
335 Git operations (e.g 'git checkout' or 'git add').
337 Use the `working-tree-encoding` attribute only if you cannot store a file
338 in UTF-8 encoding and if you want Git to be able to process the content
341 As an example, use the following attributes if your '*.ps1' files are
342 UTF-16 encoded with byte order mark (BOM) and you want Git to perform
343 automatic line ending conversion based on your platform.
345 ------------------------
346 *.ps1 text working-tree-encoding=UTF-16
347 ------------------------
349 Use the following attributes if your '*.ps1' files are UTF-16 little
350 endian encoded without BOM and you want Git to use Windows line endings
351 in the working directory (use `UTF-16LE-BOM` instead of `UTF-16LE` if
352 you want UTF-16 little endian with BOM).
353 Please note, it is highly recommended to
354 explicitly define the line endings with `eol` if the `working-tree-encoding`
355 attribute is used to avoid ambiguity.
357 ------------------------
358 *.ps1 text working-tree-encoding=UTF-16LE eol=CRLF
359 ------------------------
361 You can get a list of all available encodings on your platform with the
364 ------------------------
366 ------------------------
368 If you do not know the encoding of a file, then you can use the `file`
369 command to guess the encoding:
371 ------------------------
373 ------------------------
379 When the attribute `ident` is set for a path, Git replaces
380 `$Id$` in the blob object with `$Id:`, followed by the
381 40-character hexadecimal blob object name, followed by a dollar
382 sign `$` upon checkout. Any byte sequence that begins with
383 `$Id:` and ends with `$` in the worktree file is replaced
384 with `$Id$` upon check-in.
390 A `filter` attribute can be set to a string value that names a
391 filter driver specified in the configuration.
393 A filter driver consists of a `clean` command and a `smudge`
394 command, either of which can be left unspecified. Upon
395 checkout, when the `smudge` command is specified, the command is
396 fed the blob object from its standard input, and its standard
397 output is used to update the worktree file. Similarly, the
398 `clean` command is used to convert the contents of worktree file
399 upon checkin. By default these commands process only a single
400 blob and terminate. If a long running `process` filter is used
401 in place of `clean` and/or `smudge` filters, then Git can process
402 all blobs with a single filter command invocation for the entire
403 life of a single Git command, for example `git add --all`. If a
404 long running `process` filter is configured then it always takes
405 precedence over a configured single blob filter. See section
406 below for the description of the protocol used to communicate with
409 One use of the content filtering is to massage the content into a shape
410 that is more convenient for the platform, filesystem, and the user to use.
411 For this mode of operation, the key phrase here is "more convenient" and
412 not "turning something unusable into usable". In other words, the intent
413 is that if someone unsets the filter driver definition, or does not have
414 the appropriate filter program, the project should still be usable.
416 Another use of the content filtering is to store the content that cannot
417 be directly used in the repository (e.g. a UUID that refers to the true
418 content stored outside Git, or an encrypted content) and turn it into a
419 usable form upon checkout (e.g. download the external content, or decrypt
420 the encrypted content).
422 These two filters behave differently, and by default, a filter is taken as
423 the former, massaging the contents into more convenient shape. A missing
424 filter driver definition in the config, or a filter driver that exits with
425 a non-zero status, is not an error but makes the filter a no-op passthru.
427 You can declare that a filter turns a content that by itself is unusable
428 into a usable content by setting the filter.<driver>.required configuration
431 Note: Whenever the clean filter is changed, the repo should be renormalized:
432 $ git add --renormalize .
434 For example, in .gitattributes, you would assign the `filter`
437 ------------------------
439 ------------------------
441 Then you would define a "filter.indent.clean" and "filter.indent.smudge"
442 configuration in your .git/config to specify a pair of commands to
443 modify the contents of C programs when the source files are checked
444 in ("clean" is run) and checked out (no change is made because the
447 ------------------------
451 ------------------------
453 For best results, `clean` should not alter its output further if it is
454 run twice ("clean->clean" should be equivalent to "clean"), and
455 multiple `smudge` commands should not alter `clean`'s output
456 ("smudge->smudge->clean" should be equivalent to "clean"). See the
457 section on merging below.
459 The "indent" filter is well-behaved in this regard: it will not modify
460 input that is already correctly indented. In this case, the lack of a
461 smudge filter means that the clean filter _must_ accept its own output
462 without modifying it.
464 If a filter _must_ succeed in order to make the stored contents usable,
465 you can declare that the filter is `required`, in the configuration:
467 ------------------------
469 clean = openssl enc ...
470 smudge = openssl enc -d ...
472 ------------------------
474 Sequence "%f" on the filter command line is replaced with the name of
475 the file the filter is working on. A filter might use this in keyword
476 substitution. For example:
478 ------------------------
480 clean = git-p4-filter --clean %f
481 smudge = git-p4-filter --smudge %f
482 ------------------------
484 Note that "%f" is the name of the path that is being worked on. Depending
485 on the version that is being filtered, the corresponding file on disk may
486 not exist, or may have different contents. So, smudge and clean commands
487 should not try to access the file on disk, but only act as filters on the
488 content provided to them on standard input.
490 Long Running Filter Process
491 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
493 If the filter command (a string value) is defined via
494 `filter.<driver>.process` then Git can process all blobs with a
495 single filter invocation for the entire life of a single Git
496 command. This is achieved by using the long-running process protocol
497 (described in technical/long-running-process-protocol.txt).
499 When Git encounters the first file that needs to be cleaned or smudged,
500 it starts the filter and performs the handshake. In the handshake, the
501 welcome message sent by Git is "git-filter-client", only version 2 is
502 supported, and the supported capabilities are "clean", "smudge", and
505 Afterwards Git sends a list of "key=value" pairs terminated with
506 a flush packet. The list will contain at least the filter command
507 (based on the supported capabilities) and the pathname of the file
508 to filter relative to the repository root. Right after the flush packet
509 Git sends the content split in zero or more pkt-line packets and a
510 flush packet to terminate content. Please note, that the filter
511 must not send any response before it received the content and the
512 final flush packet. Also note that the "value" of a "key=value" pair
513 can contain the "=" character whereas the key would never contain
515 ------------------------
516 packet: git> command=smudge
517 packet: git> pathname=path/testfile.dat
521 ------------------------
523 The filter is expected to respond with a list of "key=value" pairs
524 terminated with a flush packet. If the filter does not experience
525 problems then the list must contain a "success" status. Right after
526 these packets the filter is expected to send the content in zero
527 or more pkt-line packets and a flush packet at the end. Finally, a
528 second list of "key=value" pairs terminated with a flush packet
529 is expected. The filter can change the status in the second list
530 or keep the status as is with an empty list. Please note that the
531 empty list must be terminated with a flush packet regardless.
533 ------------------------
534 packet: git< status=success
536 packet: git< SMUDGED_CONTENT
538 packet: git< 0000 # empty list, keep "status=success" unchanged!
539 ------------------------
541 If the result content is empty then the filter is expected to respond
542 with a "success" status and a flush packet to signal the empty content.
543 ------------------------
544 packet: git< status=success
546 packet: git< 0000 # empty content!
547 packet: git< 0000 # empty list, keep "status=success" unchanged!
548 ------------------------
550 In case the filter cannot or does not want to process the content,
551 it is expected to respond with an "error" status.
552 ------------------------
553 packet: git< status=error
555 ------------------------
557 If the filter experiences an error during processing, then it can
558 send the status "error" after the content was (partially or
560 ------------------------
561 packet: git< status=success
563 packet: git< HALF_WRITTEN_ERRONEOUS_CONTENT
565 packet: git< status=error
567 ------------------------
569 In case the filter cannot or does not want to process the content
570 as well as any future content for the lifetime of the Git process,
571 then it is expected to respond with an "abort" status at any point
573 ------------------------
574 packet: git< status=abort
576 ------------------------
578 Git neither stops nor restarts the filter process in case the
579 "error"/"abort" status is set. However, Git sets its exit code
580 according to the `filter.<driver>.required` flag, mimicking the
581 behavior of the `filter.<driver>.clean` / `filter.<driver>.smudge`
584 If the filter dies during the communication or does not adhere to
585 the protocol then Git will stop the filter process and restart it
586 with the next file that needs to be processed. Depending on the
587 `filter.<driver>.required` flag Git will interpret that as error.
592 If the filter supports the "delay" capability, then Git can send the
593 flag "can-delay" after the filter command and pathname. This flag
594 denotes that the filter can delay filtering the current blob (e.g. to
595 compensate network latencies) by responding with no content but with
596 the status "delayed" and a flush packet.
597 ------------------------
598 packet: git> command=smudge
599 packet: git> pathname=path/testfile.dat
600 packet: git> can-delay=1
604 packet: git< status=delayed
606 ------------------------
608 If the filter supports the "delay" capability then it must support the
609 "list_available_blobs" command. If Git sends this command, then the
610 filter is expected to return a list of pathnames representing blobs
611 that have been delayed earlier and are now available.
612 The list must be terminated with a flush packet followed
613 by a "success" status that is also terminated with a flush packet. If
614 no blobs for the delayed paths are available, yet, then the filter is
615 expected to block the response until at least one blob becomes
616 available. The filter can tell Git that it has no more delayed blobs
617 by sending an empty list. As soon as the filter responds with an empty
618 list, Git stops asking. All blobs that Git has not received at this
619 point are considered missing and will result in an error.
621 ------------------------
622 packet: git> command=list_available_blobs
624 packet: git< pathname=path/testfile.dat
625 packet: git< pathname=path/otherfile.dat
627 packet: git< status=success
629 ------------------------
631 After Git received the pathnames, it will request the corresponding
632 blobs again. These requests contain a pathname and an empty content
633 section. The filter is expected to respond with the smudged content
634 in the usual way as explained above.
635 ------------------------
636 packet: git> command=smudge
637 packet: git> pathname=path/testfile.dat
639 packet: git> 0000 # empty content!
640 packet: git< status=success
642 packet: git< SMUDGED_CONTENT
644 packet: git< 0000 # empty list, keep "status=success" unchanged!
645 ------------------------
650 A long running filter demo implementation can be found in
651 `contrib/long-running-filter/example.pl` located in the Git
652 core repository. If you develop your own long running filter
653 process then the `GIT_TRACE_PACKET` environment variables can be
654 very helpful for debugging (see linkgit:git[1]).
656 Please note that you cannot use an existing `filter.<driver>.clean`
657 or `filter.<driver>.smudge` command with `filter.<driver>.process`
658 because the former two use a different inter process communication
659 protocol than the latter one.
662 Interaction between checkin/checkout attributes
663 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
665 In the check-in codepath, the worktree file is first converted
666 with `filter` driver (if specified and corresponding driver
667 defined), then the result is processed with `ident` (if
668 specified), and then finally with `text` (again, if specified
671 In the check-out codepath, the blob content is first converted
672 with `text`, and then `ident` and fed to `filter`.
675 Merging branches with differing checkin/checkout attributes
676 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
678 If you have added attributes to a file that cause the canonical
679 repository format for that file to change, such as adding a
680 clean/smudge filter or text/eol/ident attributes, merging anything
681 where the attribute is not in place would normally cause merge
684 To prevent these unnecessary merge conflicts, Git can be told to run a
685 virtual check-out and check-in of all three stages of a file when
686 resolving a three-way merge by setting the `merge.renormalize`
687 configuration variable. This prevents changes caused by check-in
688 conversion from causing spurious merge conflicts when a converted file
689 is merged with an unconverted file.
691 As long as a "smudge->clean" results in the same output as a "clean"
692 even on files that are already smudged, this strategy will
693 automatically resolve all filter-related conflicts. Filters that do
694 not act in this way may cause additional merge conflicts that must be
704 The attribute `diff` affects how Git generates diffs for particular
705 files. It can tell Git whether to generate a textual patch for the path
706 or to treat the path as a binary file. It can also affect what line is
707 shown on the hunk header `@@ -k,l +n,m @@` line, tell Git to use an
708 external command to generate the diff, or ask Git to convert binary
709 files to a text format before generating the diff.
713 A path to which the `diff` attribute is set is treated
714 as text, even when they contain byte values that
715 normally never appear in text files, such as NUL.
719 A path to which the `diff` attribute is unset will
720 generate `Binary files differ` (or a binary patch, if
721 binary patches are enabled).
725 A path to which the `diff` attribute is unspecified
726 first gets its contents inspected, and if it looks like
727 text and is smaller than core.bigFileThreshold, it is treated
728 as text. Otherwise it would generate `Binary files differ`.
732 Diff is shown using the specified diff driver. Each driver may
733 specify one or more options, as described in the following
734 section. The options for the diff driver "foo" are defined
735 by the configuration variables in the "diff.foo" section of the
739 Defining an external diff driver
740 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
742 The definition of a diff driver is done in `gitconfig`, not
743 `gitattributes` file, so strictly speaking this manual page is a
744 wrong place to talk about it. However...
746 To define an external diff driver `jcdiff`, add a section to your
747 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
749 ----------------------------------------------------------------
752 ----------------------------------------------------------------
754 When Git needs to show you a diff for the path with `diff`
755 attribute set to `jcdiff`, it calls the command you specified
756 with the above configuration, i.e. `j-c-diff`, with 7
757 parameters, just like `GIT_EXTERNAL_DIFF` program is called.
758 See linkgit:git[1] for details.
761 Defining a custom hunk-header
762 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
764 Each group of changes (called a "hunk") in the textual diff output
765 is prefixed with a line of the form:
769 This is called a 'hunk header'. The "TEXT" portion is by default a line
770 that begins with an alphabet, an underscore or a dollar sign; this
771 matches what GNU 'diff -p' output uses. This default selection however
772 is not suited for some contents, and you can use a customized pattern
775 First, in .gitattributes, you would assign the `diff` attribute
778 ------------------------
780 ------------------------
782 Then, you would define a "diff.tex.xfuncname" configuration to
783 specify a regular expression that matches a line that you would
784 want to appear as the hunk header "TEXT". Add a section to your
785 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
787 ------------------------
789 xfuncname = "^(\\\\(sub)*section\\{.*)$"
790 ------------------------
792 Note. A single level of backslashes are eaten by the
793 configuration file parser, so you would need to double the
794 backslashes; the pattern above picks a line that begins with a
795 backslash, and zero or more occurrences of `sub` followed by
796 `section` followed by open brace, to the end of line.
798 There are a few built-in patterns to make this easier, and `tex`
799 is one of them, so you do not have to write the above in your
800 configuration file (you still need to enable this with the
801 attribute mechanism, via `.gitattributes`). The following built in
802 patterns are available:
804 - `ada` suitable for source code in the Ada language.
806 - `bash` suitable for source code in the Bourne-Again SHell language.
807 Covers a superset of POSIX shell function definitions.
809 - `bibtex` suitable for files with BibTeX coded references.
811 - `cpp` suitable for source code in the C and C++ languages.
813 - `csharp` suitable for source code in the C# language.
815 - `css` suitable for cascading style sheets.
817 - `dts` suitable for devicetree (DTS) files.
819 - `elixir` suitable for source code in the Elixir language.
821 - `fortran` suitable for source code in the Fortran language.
823 - `fountain` suitable for Fountain documents.
825 - `golang` suitable for source code in the Go language.
827 - `html` suitable for HTML/XHTML documents.
829 - `java` suitable for source code in the Java language.
831 - `markdown` suitable for Markdown documents.
833 - `matlab` suitable for source code in the MATLAB and Octave languages.
835 - `objc` suitable for source code in the Objective-C language.
837 - `pascal` suitable for source code in the Pascal/Delphi language.
839 - `perl` suitable for source code in the Perl language.
841 - `php` suitable for source code in the PHP language.
843 - `python` suitable for source code in the Python language.
845 - `ruby` suitable for source code in the Ruby language.
847 - `rust` suitable for source code in the Rust language.
849 - `scheme` suitable for source code in the Scheme language.
851 - `tex` suitable for source code for LaTeX documents.
854 Customizing word diff
855 ^^^^^^^^^^^^^^^^^^^^^
857 You can customize the rules that `git diff --word-diff` uses to
858 split words in a line, by specifying an appropriate regular expression
859 in the "diff.*.wordRegex" configuration variable. For example, in TeX
860 a backslash followed by a sequence of letters forms a command, but
861 several such commands can be run together without intervening
862 whitespace. To separate them, use a regular expression in your
863 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
865 ------------------------
867 wordRegex = "\\\\[a-zA-Z]+|[{}]|\\\\.|[^\\{}[:space:]]+"
868 ------------------------
870 A built-in pattern is provided for all languages listed in the
874 Performing text diffs of binary files
875 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
877 Sometimes it is desirable to see the diff of a text-converted
878 version of some binary files. For example, a word processor
879 document can be converted to an ASCII text representation, and
880 the diff of the text shown. Even though this conversion loses
881 some information, the resulting diff is useful for human
882 viewing (but cannot be applied directly).
884 The `textconv` config option is used to define a program for
885 performing such a conversion. The program should take a single
886 argument, the name of a file to convert, and produce the
887 resulting text on stdout.
889 For example, to show the diff of the exif information of a
890 file instead of the binary information (assuming you have the
891 exif tool installed), add the following section to your
892 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file):
894 ------------------------
897 ------------------------
899 NOTE: The text conversion is generally a one-way conversion;
900 in this example, we lose the actual image contents and focus
901 just on the text data. This means that diffs generated by
902 textconv are _not_ suitable for applying. For this reason,
903 only `git diff` and the `git log` family of commands (i.e.,
904 log, whatchanged, show) will perform text conversion. `git
905 format-patch` will never generate this output. If you want to
906 send somebody a text-converted diff of a binary file (e.g.,
907 because it quickly conveys the changes you have made), you
908 should generate it separately and send it as a comment _in
909 addition to_ the usual binary diff that you might send.
911 Because text conversion can be slow, especially when doing a
912 large number of them with `git log -p`, Git provides a mechanism
913 to cache the output and use it in future diffs. To enable
914 caching, set the "cachetextconv" variable in your diff driver's
917 ------------------------
921 ------------------------
923 This will cache the result of running "exif" on each blob
924 indefinitely. If you change the textconv config variable for a
925 diff driver, Git will automatically invalidate the cache entries
926 and re-run the textconv filter. If you want to invalidate the
927 cache manually (e.g., because your version of "exif" was updated
928 and now produces better output), you can remove the cache
929 manually with `git update-ref -d refs/notes/textconv/jpg` (where
930 "jpg" is the name of the diff driver, as in the example above).
932 Choosing textconv versus external diff
933 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
935 If you want to show differences between binary or specially-formatted
936 blobs in your repository, you can choose to use either an external diff
937 command, or to use textconv to convert them to a diff-able text format.
938 Which method you choose depends on your exact situation.
940 The advantage of using an external diff command is flexibility. You are
941 not bound to find line-oriented changes, nor is it necessary for the
942 output to resemble unified diff. You are free to locate and report
943 changes in the most appropriate way for your data format.
945 A textconv, by comparison, is much more limiting. You provide a
946 transformation of the data into a line-oriented text format, and Git
947 uses its regular diff tools to generate the output. There are several
948 advantages to choosing this method:
950 1. Ease of use. It is often much simpler to write a binary to text
951 transformation than it is to perform your own diff. In many cases,
952 existing programs can be used as textconv filters (e.g., exif,
955 2. Git diff features. By performing only the transformation step
956 yourself, you can still utilize many of Git's diff features,
957 including colorization, word-diff, and combined diffs for merges.
959 3. Caching. Textconv caching can speed up repeated diffs, such as those
960 you might trigger by running `git log -p`.
963 Marking files as binary
964 ^^^^^^^^^^^^^^^^^^^^^^^
966 Git usually guesses correctly whether a blob contains text or binary
967 data by examining the beginning of the contents. However, sometimes you
968 may want to override its decision, either because a blob contains binary
969 data later in the file, or because the content, while technically
970 composed of text characters, is opaque to a human reader. For example,
971 many postscript files contain only ASCII characters, but produce noisy
972 and meaningless diffs.
974 The simplest way to mark a file as binary is to unset the diff
975 attribute in the `.gitattributes` file:
977 ------------------------
979 ------------------------
981 This will cause Git to generate `Binary files differ` (or a binary
982 patch, if binary patches are enabled) instead of a regular diff.
984 However, one may also want to specify other diff driver attributes. For
985 example, you might want to use `textconv` to convert postscript files to
986 an ASCII representation for human viewing, but otherwise treat them as
987 binary files. You cannot specify both `-diff` and `diff=ps` attributes.
988 The solution is to use the `diff.*.binary` config option:
990 ------------------------
994 ------------------------
996 Performing a three-way merge
997 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1002 The attribute `merge` affects how three versions of a file are
1003 merged when a file-level merge is necessary during `git merge`,
1004 and other commands such as `git revert` and `git cherry-pick`.
1008 Built-in 3-way merge driver is used to merge the
1009 contents in a way similar to 'merge' command of `RCS`
1010 suite. This is suitable for ordinary text files.
1014 Take the version from the current branch as the
1015 tentative merge result, and declare that the merge has
1016 conflicts. This is suitable for binary files that do
1017 not have a well-defined merge semantics.
1021 By default, this uses the same built-in 3-way merge
1022 driver as is the case when the `merge` attribute is set.
1023 However, the `merge.default` configuration variable can name
1024 different merge driver to be used with paths for which the
1025 `merge` attribute is unspecified.
1029 3-way merge is performed using the specified custom
1030 merge driver. The built-in 3-way merge driver can be
1031 explicitly specified by asking for "text" driver; the
1032 built-in "take the current branch" driver can be
1033 requested with "binary".
1036 Built-in merge drivers
1037 ^^^^^^^^^^^^^^^^^^^^^^
1039 There are a few built-in low-level merge drivers defined that
1040 can be asked for via the `merge` attribute.
1044 Usual 3-way file level merge for text files. Conflicted
1045 regions are marked with conflict markers `<<<<<<<`,
1046 `=======` and `>>>>>>>`. The version from your branch
1047 appears before the `=======` marker, and the version
1048 from the merged branch appears after the `=======`
1053 Keep the version from your branch in the work tree, but
1054 leave the path in the conflicted state for the user to
1059 Run 3-way file level merge for text files, but take
1060 lines from both versions, instead of leaving conflict
1061 markers. This tends to leave the added lines in the
1062 resulting file in random order and the user should
1063 verify the result. Do not use this if you do not
1064 understand the implications.
1067 Defining a custom merge driver
1068 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1070 The definition of a merge driver is done in the `.git/config`
1071 file, not in the `gitattributes` file, so strictly speaking this
1072 manual page is a wrong place to talk about it. However...
1074 To define a custom merge driver `filfre`, add a section to your
1075 `$GIT_DIR/config` file (or `$HOME/.gitconfig` file) like this:
1077 ----------------------------------------------------------------
1079 name = feel-free merge driver
1080 driver = filfre %O %A %B %L %P
1082 ----------------------------------------------------------------
1084 The `merge.*.name` variable gives the driver a human-readable
1087 The `merge.*.driver` variable's value is used to construct a
1088 command to run to merge ancestor's version (`%O`), current
1089 version (`%A`) and the other branches' version (`%B`). These
1090 three tokens are replaced with the names of temporary files that
1091 hold the contents of these versions when the command line is
1092 built. Additionally, %L will be replaced with the conflict marker
1095 The merge driver is expected to leave the result of the merge in
1096 the file named with `%A` by overwriting it, and exit with zero
1097 status if it managed to merge them cleanly, or non-zero if there
1100 The `merge.*.recursive` variable specifies what other merge
1101 driver to use when the merge driver is called for an internal
1102 merge between common ancestors, when there are more than one.
1103 When left unspecified, the driver itself is used for both
1104 internal merge and the final merge.
1106 The merge driver can learn the pathname in which the merged result
1107 will be stored via placeholder `%P`.
1110 `conflict-marker-size`
1111 ^^^^^^^^^^^^^^^^^^^^^^
1113 This attribute controls the length of conflict markers left in
1114 the work tree file during a conflicted merge. Only setting to
1115 the value to a positive integer has any meaningful effect.
1117 For example, this line in `.gitattributes` can be used to tell the merge
1118 machinery to leave much longer (instead of the usual 7-character-long)
1119 conflict markers when merging the file `Documentation/git-merge.txt`
1120 results in a conflict.
1122 ------------------------
1123 Documentation/git-merge.txt conflict-marker-size=32
1124 ------------------------
1127 Checking whitespace errors
1128 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1133 The `core.whitespace` configuration variable allows you to define what
1134 'diff' and 'apply' should consider whitespace errors for all paths in
1135 the project (See linkgit:git-config[1]). This attribute gives you finer
1140 Notice all types of potential whitespace errors known to Git.
1141 The tab width is taken from the value of the `core.whitespace`
1142 configuration variable.
1146 Do not notice anything as error.
1150 Use the value of the `core.whitespace` configuration variable to
1151 decide what to notice as error.
1155 Specify a comma separate list of common whitespace problems to
1156 notice in the same format as the `core.whitespace` configuration
1166 Files and directories with the attribute `export-ignore` won't be added to
1172 If the attribute `export-subst` is set for a file then Git will expand
1173 several placeholders when adding this file to an archive. The
1174 expansion depends on the availability of a commit ID, i.e., if
1175 linkgit:git-archive[1] has been given a tree instead of a commit or a
1176 tag then no replacement will be done. The placeholders are the same
1177 as those for the option `--pretty=format:` of linkgit:git-log[1],
1178 except that they need to be wrapped like this: `$Format:PLACEHOLDERS$`
1179 in the file. E.g. the string `$Format:%H$` will be replaced by the
1180 commit hash. However, only one `%(describe)` placeholder is expanded
1181 per archive to avoid denial-of-service attacks.
1190 Delta compression will not be attempted for blobs for paths with the
1191 attribute `delta` set to false.
1194 Viewing files in GUI tools
1195 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1200 The value of this attribute specifies the character encoding that should
1201 be used by GUI tools (e.g. linkgit:gitk[1] and linkgit:git-gui[1]) to
1202 display the contents of the relevant file. Note that due to performance
1203 considerations linkgit:gitk[1] does not use this attribute unless you
1204 manually enable per-file encodings in its options.
1206 If this attribute is not set or has an invalid value, the value of the
1207 `gui.encoding` configuration variable is used instead
1208 (See linkgit:git-config[1]).
1211 USING MACRO ATTRIBUTES
1212 ----------------------
1214 You do not want any end-of-line conversions applied to, nor textual diffs
1215 produced for, any binary file you track. You would need to specify e.g.
1221 but that may become cumbersome, when you have many attributes. Using
1222 macro attributes, you can define an attribute that, when set, also
1223 sets or unsets a number of other attributes at the same time. The
1224 system knows a built-in macro attribute, `binary`:
1230 Setting the "binary" attribute also unsets the "text" and "diff"
1231 attributes as above. Note that macro attributes can only be "Set",
1232 though setting one might have the effect of setting or unsetting other
1233 attributes or even returning other attributes to the "Unspecified"
1237 DEFINING MACRO ATTRIBUTES
1238 -------------------------
1240 Custom macro attributes can be defined only in top-level gitattributes
1241 files (`$GIT_DIR/info/attributes`, the `.gitattributes` file at the
1242 top level of the working tree, or the global or system-wide
1243 gitattributes files), not in `.gitattributes` files in working tree
1244 subdirectories. The built-in macro attribute "binary" is equivalent
1248 [attr]binary -diff -merge -text
1254 Git does not follow symbolic links when accessing a `.gitattributes`
1255 file in the working tree. This keeps behavior consistent when the file
1256 is accessed from the index or a tree versus from the filesystem.
1261 If you have these three `gitattributes` file:
1263 ----------------------------------------------------------------
1264 (in $GIT_DIR/info/attributes)
1271 (in t/.gitattributes)
1275 ----------------------------------------------------------------
1277 the attributes given to path `t/abc` are computed as follows:
1279 1. By examining `t/.gitattributes` (which is in the same
1280 directory as the path in question), Git finds that the first
1281 line matches. `merge` attribute is set. It also finds that
1282 the second line matches, and attributes `foo` and `bar`
1285 2. Then it examines `.gitattributes` (which is in the parent
1286 directory), and finds that the first line matches, but
1287 `t/.gitattributes` file already decided how `merge`, `foo`
1288 and `bar` attributes should be given to this path, so it
1289 leaves `foo` and `bar` unset. Attribute `baz` is set.
1291 3. Finally it examines `$GIT_DIR/info/attributes`. This file
1292 is used to override the in-tree settings. The first line is
1293 a match, and `foo` is set, `bar` is reverted to unspecified
1294 state, and `baz` is unset.
1296 As the result, the attributes assignment to `t/abc` becomes:
1298 ----------------------------------------------------------------
1302 merge set to string value "filfre"
1304 ----------------------------------------------------------------
1309 linkgit:git-check-attr[1].
1313 Part of the linkgit:git[1] suite