1GIT-DIFF-FILES(1) Git Manual GIT-DIFF-FILES(1)
2
6 git-diff-files - Compares files in the working tree and the index
7
9 git diff-files [-q] [-0|-1|-2|-3|-c|--cc] [<common-diff-options>] [<path>...]
10
12 Compares the files in the working tree and the index. When paths are
13 specified, compares only those named paths. Otherwise all entries in
14 the index are compared. The output format is the same as for git
15 diff-index and git diff-tree.
16
18 -p, -u, --patch
19 Generate patch (see section on generating patches).
20 -s, --no-patch
22 Suppress diff output. Useful for commands like git show that show
23 the patch by default, or to cancel the effect of --patch.
24 -U<n>, --unified=<n>
26 Generate diffs with <n> lines of context instead of the usual
27 three. Implies --patch.
28 --output=<file>
30 Output to a specific file instead of stdout.
31 --output-indicator-new=<char>, --output-indicator-old=<char>,
33 --output-indicator-context=<char>
34 Specify the character used to indicate new, old or context lines in
35 the generated patch. Normally they are +, - and ' ' respectively.
36 --raw
38 Generate the diff in raw format. This is the default.
39 --patch-with-raw
41 Synonym for -p --raw.
42 --indent-heuristic
44 Enable the heuristic that shifts diff hunk boundaries to make
45 patches easier to read. This is the default.
46 --no-indent-heuristic
48 Disable the indent heuristic.
49 --minimal
51 Spend extra time to make sure the smallest possible diff is
52 produced.
53 --patience
55 Generate a diff using the "patience diff" algorithm.
56 --histogram
58 Generate a diff using the "histogram diff" algorithm.
59 --anchored=<text>
61 Generate a diff using the "anchored diff" algorithm.
62 This option may be specified more than once.
64 If a line exists in both the source and destination, exists only
66 once, and starts with this text, this algorithm attempts to prevent
67 it from appearing as a deletion or addition in the output. It uses
68 the "patience diff" algorithm internally.
69 --diff-algorithm={patience|minimal|histogram|myers}
71 Choose a diff algorithm. The variants are as follows:
72 default, myers
74 The basic greedy diff algorithm. Currently, this is the
75 default.
76 minimal
78 Spend extra time to make sure the smallest possible diff is
79 produced.
80 patience
82 Use "patience diff" algorithm when generating patches.
83 histogram
85 This algorithm extends the patience algorithm to "support
86 low-occurrence common elements".
87 For instance, if you configured the diff.algorithm variable to a
89 non-default value and want to use the default one, then you have to
90 use --diff-algorithm=default option.
91 --stat[=<width>[,<name-width>[,<count>]]]
93 Generate a diffstat. By default, as much space as necessary will be
94 used for the filename part, and the rest for the graph part.
95 Maximum width defaults to terminal width, or 80 columns if not
96 connected to a terminal, and can be overridden by <width>. The
97 width of the filename part can be limited by giving another width
98 <name-width> after a comma. The width of the graph part can be
99 limited by using --stat-graph-width=<width> (affects all commands
100 generating a stat graph) or by setting diff.statGraphWidth=<width>
101 (does not affect git format-patch). By giving a third parameter
102 <count>, you can limit the output to the first <count> lines,
103 followed by ... if there are more.
104 These parameters can also be set individually with
106 --stat-width=<width>, --stat-name-width=<name-width> and
107 --stat-count=<count>.
108 --compact-summary
110 Output a condensed summary of extended header information such as
111 file creations or deletions ("new" or "gone", optionally "+l" if
112 it’s a symlink) and mode changes ("+x" or "-x" for adding or
113 removing executable bit respectively) in diffstat. The information
114 is put between the filename part and the graph part. Implies
115 --stat.
116 --numstat
118 Similar to --stat, but shows number of added and deleted lines in
119 decimal notation and pathname without abbreviation, to make it more
120 machine friendly. For binary files, outputs two - instead of saying
121 0 0.
122 --shortstat
124 Output only the last line of the --stat format containing total
125 number of modified files, as well as number of added and deleted
126 lines.
127 -X[<param1,param2,...>], --dirstat[=<param1,param2,...>]
129 Output the distribution of relative amount of changes for each
130 sub-directory. The behavior of --dirstat can be customized by
131 passing it a comma separated list of parameters. The defaults are
132 controlled by the diff.dirstat configuration variable (see git-
133 config(1)). The following parameters are available:
134 changes
136 Compute the dirstat numbers by counting the lines that have
137 been removed from the source, or added to the destination. This
138 ignores the amount of pure code movements within a file. In
139 other words, rearranging lines in a file is not counted as much
140 as other changes. This is the default behavior when no
141 parameter is given.
142 lines
144 Compute the dirstat numbers by doing the regular line-based
145 diff analysis, and summing the removed/added line counts. (For
146 binary files, count 64-byte chunks instead, since binary files
147 have no natural concept of lines). This is a more expensive
148 --dirstat behavior than the changes behavior, but it does count
149 rearranged lines within a file as much as other changes. The
150 resulting output is consistent with what you get from the other
151 --*stat options.
152 files
154 Compute the dirstat numbers by counting the number of files
155 changed. Each changed file counts equally in the dirstat
156 analysis. This is the computationally cheapest --dirstat
157 behavior, since it does not have to look at the file contents
158 at all.
159 cumulative
161 Count changes in a child directory for the parent directory as
162 well. Note that when using cumulative, the sum of the
163 percentages reported may exceed 100%. The default
164 (non-cumulative) behavior can be specified with the
165 noncumulative parameter.
166 <limit>
168 An integer parameter specifies a cut-off percent (3% by
169 default). Directories contributing less than this percentage of
170 the changes are not shown in the output.
171 Example: The following will count changed files, while ignoring
173 directories with less than 10% of the total amount of changed
174 files, and accumulating child directory counts in the parent
175 directories: --dirstat=files,10,cumulative.
176 --cumulative
178 Synonym for --dirstat=cumulative
179 --dirstat-by-file[=<param1,param2>...]
181 Synonym for --dirstat=files,param1,param2...
182 --summary
184 Output a condensed summary of extended header information such as
185 creations, renames and mode changes.
186 --patch-with-stat
188 Synonym for -p --stat.
189 -z
191 When --raw, --numstat, --name-only or --name-status has been given,
192 do not munge pathnames and use NULs as output field terminators.
193 Without this option, pathnames with "unusual" characters are quoted
195 as explained for the configuration variable core.quotePath (see
196 git-config(1)).
197 --name-only
199 Show only names of changed files. The file names are often encoded
200 in UTF-8. For more information see the discussion about encoding in
201 the git-log(1) manual page.
202 --name-status
204 Show only names and status of changed files. See the description of
205 the --diff-filter option on what the status letters mean. Just like
206 --name-only the file names are often encoded in UTF-8.
207 --submodule[=<format>]
209 Specify how differences in submodules are shown. When specifying
210 --submodule=short the short format is used. This format just shows
211 the names of the commits at the beginning and end of the range.
212 When --submodule or --submodule=log is specified, the log format is
213 used. This format lists the commits in the range like git-
214 submodule(1) summary does. When --submodule=diff is specified, the
215 diff format is used. This format shows an inline diff of the
216 changes in the submodule contents between the commit range.
217 Defaults to diff.submodule or the short format if the config option
218 is unset.
219 --color[=<when>]
221 Show colored diff. --color (i.e. without =<when>) is the same as
222 --color=always. <when> can be one of always, never, or auto.
223 --no-color
225 Turn off colored diff. It is the same as --color=never.
226 --color-moved[=<mode>]
228 Moved lines of code are colored differently. The <mode> defaults to
229 no if the option is not given and to zebra if the option with no
230 mode is given. The mode must be one of:
231 no
233 Moved lines are not highlighted.
234 default
236 Is a synonym for zebra. This may change to a more sensible mode
237 in the future.
238 plain
240 Any line that is added in one location and was removed in
241 another location will be colored with color.diff.newMoved.
242 Similarly color.diff.oldMoved will be used for removed lines
243 that are added somewhere else in the diff. This mode picks up
244 any moved line, but it is not very useful in a review to
245 determine if a block of code was moved without permutation.
246 blocks
248 Blocks of moved text of at least 20 alphanumeric characters are
249 detected greedily. The detected blocks are painted using either
250 the color.diff.{old,new}Moved color. Adjacent blocks cannot be
251 told apart.
252 zebra
254 Blocks of moved text are detected as in blocks mode. The blocks
255 are painted using either the color.diff.{old,new}Moved color or
256 color.diff.{old,new}MovedAlternative. The change between the
257 two colors indicates that a new block was detected.
258 dimmed-zebra
260 Similar to zebra, but additional dimming of uninteresting parts
261 of moved code is performed. The bordering lines of two adjacent
262 blocks are considered interesting, the rest is uninteresting.
263 dimmed_zebra is a deprecated synonym.
264 --no-color-moved
266 Turn off move detection. This can be used to override configuration
267 settings. It is the same as --color-moved=no.
268 --color-moved-ws=<modes>
270 This configures how whitespace is ignored when performing the move
271 detection for --color-moved. These modes can be given as a comma
272 separated list:
273 no
275 Do not ignore whitespace when performing move detection.
276 ignore-space-at-eol
278 Ignore changes in whitespace at EOL.
279 ignore-space-change
281 Ignore changes in amount of whitespace. This ignores whitespace
282 at line end, and considers all other sequences of one or more
283 whitespace characters to be equivalent.
284 ignore-all-space
286 Ignore whitespace when comparing lines. This ignores
287 differences even if one line has whitespace where the other
288 line has none.
289 allow-indentation-change
291 Initially ignore any whitespace in the move detection, then
292 group the moved code blocks only into a block if the change in
293 whitespace is the same per line. This is incompatible with the
294 other modes.
295 --no-color-moved-ws
297 Do not ignore whitespace when performing move detection. This can
298 be used to override configuration settings. It is the same as
299 --color-moved-ws=no.
300 --word-diff[=<mode>]
302 Show a word diff, using the <mode> to delimit changed words. By
303 default, words are delimited by whitespace; see --word-diff-regex
304 below. The <mode> defaults to plain, and must be one of:
305 color
307 Highlight changed words using only colors. Implies --color.
308 plain
310 Show words as [-removed-] and {+added+}. Makes no attempts to
311 escape the delimiters if they appear in the input, so the
312 output may be ambiguous.
313 porcelain
315 Use a special line-based format intended for script
316 consumption. Added/removed/unchanged runs are printed in the
317 usual unified diff format, starting with a +/-/` ` character at
318 the beginning of the line and extending to the end of the line.
319 Newlines in the input are represented by a tilde ~ on a line of
320 its own.
321 none
323 Disable word diff again.
324 Note that despite the name of the first mode, color is used to
326 highlight the changed parts in all modes if enabled.
327 --word-diff-regex=<regex>
329 Use <regex> to decide what a word is, instead of considering runs
330 of non-whitespace to be a word. Also implies --word-diff unless it
331 was already enabled.
332 Every non-overlapping match of the <regex> is considered a word.
334 Anything between these matches is considered whitespace and
335 ignored(!) for the purposes of finding differences. You may want to
336 append |[^[:space:]] to your regular expression to make sure that
337 it matches all non-whitespace characters. A match that contains a
338 newline is silently truncated(!) at the newline.
339 For example, --word-diff-regex=. will treat each character as a
341 word and, correspondingly, show differences character by character.
342 The regex can also be set via a diff driver or configuration
344 option, see gitattributes(5) or git-config(1). Giving it explicitly
345 overrides any diff driver or configuration setting. Diff drivers
346 override configuration settings.
347 --color-words[=<regex>]
349 Equivalent to --word-diff=color plus (if a regex was specified)
350 --word-diff-regex=<regex>.
351 --no-renames
353 Turn off rename detection, even when the configuration file gives
354 the default to do so.
355 --[no-]rename-empty
357 Whether to use empty blobs as rename source.
358 --check
360 Warn if changes introduce conflict markers or whitespace errors.
361 What are considered whitespace errors is controlled by
362 core.whitespace configuration. By default, trailing whitespaces
363 (including lines that consist solely of whitespaces) and a space
364 character that is immediately followed by a tab character inside
365 the initial indent of the line are considered whitespace errors.
366 Exits with non-zero status if problems are found. Not compatible
367 with --exit-code.
368 --ws-error-highlight=<kind>
370 Highlight whitespace errors in the context, old or new lines of the
371 diff. Multiple values are separated by comma, none resets previous
372 values, default reset the list to new and all is a shorthand for
373 old,new,context. When this option is not given, and the
374 configuration variable diff.wsErrorHighlight is not set, only
375 whitespace errors in new lines are highlighted. The whitespace
376 errors are colored with color.diff.whitespace.
377 --full-index
379 Instead of the first handful of characters, show the full pre- and
380 post-image blob object names on the "index" line when generating
381 patch format output.
382 --binary
384 In addition to --full-index, output a binary diff that can be
385 applied with git-apply. Implies --patch.
386 --abbrev[=<n>]
388 Instead of showing the full 40-byte hexadecimal object name in
389 diff-raw format output and diff-tree header lines, show the
390 shortest prefix that is at least <n> hexdigits long that uniquely
391 refers the object. In diff-patch output format, --full-index takes
392 higher precedence, i.e. if --full-index is specified, full blob
393 names will be shown regardless of --abbrev. Non default number of
394 digits can be specified with --abbrev=<n>.
395 -B[<n>][/<m>], --break-rewrites[=[<n>][/<m>]]
397 Break complete rewrite changes into pairs of delete and create.
398 This serves two purposes:
399 It affects the way a change that amounts to a total rewrite of a
401 file not as a series of deletion and insertion mixed together with
402 a very few lines that happen to match textually as the context, but
403 as a single deletion of everything old followed by a single
404 insertion of everything new, and the number m controls this aspect
405 of the -B option (defaults to 60%). -B/70% specifies that less
406 than 30% of the original should remain in the result for Git to
407 consider it a total rewrite (i.e. otherwise the resulting patch
408 will be a series of deletion and insertion mixed together with
409 context lines).
410 When used with -M, a totally-rewritten file is also considered as
412 the source of a rename (usually -M only considers a file that
413 disappeared as the source of a rename), and the number n controls
414 this aspect of the -B option (defaults to 50%). -B20% specifies
415 that a change with addition and deletion compared to 20% or more of
416 the file’s size are eligible for being picked up as a possible
417 source of a rename to another file.
418 -M[<n>], --find-renames[=<n>]
420 Detect renames. If n is specified, it is a threshold on the
421 similarity index (i.e. amount of addition/deletions compared to the
422 file’s size). For example, -M90% means Git should consider a
423 delete/add pair to be a rename if more than 90% of the file hasn’t
424 changed. Without a % sign, the number is to be read as a fraction,
425 with a decimal point before it. I.e., -M5 becomes 0.5, and is thus
426 the same as -M50%. Similarly, -M05 is the same as -M5%. To limit
427 detection to exact renames, use -M100%. The default similarity
428 index is 50%.
429 -C[<n>], --find-copies[=<n>]
431 Detect copies as well as renames. See also --find-copies-harder. If
432 n is specified, it has the same meaning as for -M<n>.
433 --find-copies-harder
435 For performance reasons, by default, -C option finds copies only if
436 the original file of the copy was modified in the same changeset.
437 This flag makes the command inspect unmodified files as candidates
438 for the source of copy. This is a very expensive operation for
439 large projects, so use it with caution. Giving more than one -C
440 option has the same effect.
441 -D, --irreversible-delete
443 Omit the preimage for deletes, i.e. print only the header but not
444 the diff between the preimage and /dev/null. The resulting patch is
445 not meant to be applied with patch or git apply; this is solely for
446 people who want to just concentrate on reviewing the text after the
447 change. In addition, the output obviously lacks enough information
448 to apply such a patch in reverse, even manually, hence the name of
449 the option.
450 When used together with -B, omit also the preimage in the deletion
452 part of a delete/create pair.
453 -l<num>
455 The -M and -C options involve some preliminary steps that can
456 detect subsets of renames/copies cheaply, followed by an exhaustive
457 fallback portion that compares all remaining unpaired destinations
458 to all relevant sources. (For renames, only remaining unpaired
459 sources are relevant; for copies, all original sources are
460 relevant.) For N sources and destinations, this exhaustive check is
461 O(N^2). This option prevents the exhaustive portion of rename/copy
462 detection from running if the number of source/destination files
463 involved exceeds the specified number. Defaults to
464 diff.renameLimit. Note that a value of 0 is treated as unlimited.
465 --diff-filter=[(A|C|D|M|R|T|U|X|B)...[*]]
467 Select only files that are Added (A), Copied (C), Deleted (D),
468 Modified (M), Renamed (R), have their type (i.e. regular file,
469 symlink, submodule, ...) changed (T), are Unmerged (U), are Unknown
470 (X), or have had their pairing Broken (B). Any combination of the
471 filter characters (including none) can be used. When *
472 (All-or-none) is added to the combination, all paths are selected
473 if there is any file that matches other criteria in the comparison;
474 if there is no file that matches other criteria, nothing is
475 selected.
476 Also, these upper-case letters can be downcased to exclude. E.g.
478 --diff-filter=ad excludes added and deleted paths.
479 Note that not all diffs can feature all types. For instance, copied
481 and renamed entries cannot appear if detection for those types is
482 disabled.
483 -S<string>
485 Look for differences that change the number of occurrences of the
486 specified string (i.e. addition/deletion) in a file. Intended for
487 the scripter’s use.
488 It is useful when you’re looking for an exact block of code (like a
490 struct), and want to know the history of that block since it first
491 came into being: use the feature iteratively to feed the
492 interesting block in the preimage back into -S, and keep going
493 until you get the very first version of the block.
494 Binary files are searched as well.
496 -G<regex>
498 Look for differences whose patch text contains added/removed lines
499 that match <regex>.
500 To illustrate the difference between -S<regex> --pickaxe-regex and
502 -G<regex>, consider a commit with the following diff in the same
503 file:
504 + return frotz(nitfol, two->ptr, 1, 0);
506 ...
507 - hit = frotz(nitfol, mf2.ptr, 1, 0);
508 While git log -G"frotz\(nitfol" will show this commit, git log
510 -S"frotz\(nitfol" --pickaxe-regex will not (because the number of
511 occurrences of that string did not change).
512 Unless --text is supplied patches of binary files without a
514 textconv filter will be ignored.
515 See the pickaxe entry in gitdiffcore(7) for more information.
517 --find-object=<object-id>
519 Look for differences that change the number of occurrences of the
520 specified object. Similar to -S, just the argument is different in
521 that it doesn’t search for a specific string but for a specific
522 object id.
523 The object can be a blob or a submodule commit. It implies the -t
525 option in git-log to also find trees.
526 --pickaxe-all
528 When -S or -G finds a change, show all the changes in that
529 changeset, not just the files that contain the change in <string>.
530 --pickaxe-regex
532 Treat the <string> given to -S as an extended POSIX regular
533 expression to match.
534 -O<orderfile>
536 Control the order in which files appear in the output. This
537 overrides the diff.orderFile configuration variable (see git-
538 config(1)). To cancel diff.orderFile, use -O/dev/null.
539 The output order is determined by the order of glob patterns in
541 <orderfile>. All files with pathnames that match the first pattern
542 are output first, all files with pathnames that match the second
543 pattern (but not the first) are output next, and so on. All files
544 with pathnames that do not match any pattern are output last, as if
545 there was an implicit match-all pattern at the end of the file. If
546 multiple pathnames have the same rank (they match the same pattern
547 but no earlier patterns), their output order relative to each other
548 is the normal order.
549 <orderfile> is parsed as follows:
551 • Blank lines are ignored, so they can be used as separators for
553 readability.
554 • Lines starting with a hash ("#") are ignored, so they can be
556 used for comments. Add a backslash ("\") to the beginning of
557 the pattern if it starts with a hash.
558 • Each other line contains a single pattern.
560 Patterns have the same syntax and semantics as patterns used for
562 fnmatch(3) without the FNM_PATHNAME flag, except a pathname also
563 matches a pattern if removing any number of the final pathname
564 components matches the pattern. For example, the pattern "foo*bar"
565 matches "fooasdfbar" and "foo/bar/baz/asdf" but not "foobarx".
566 --skip-to=<file>, --rotate-to=<file>
568 Discard the files before the named <file> from the output (i.e.
569 skip to), or move them to the end of the output (i.e. rotate to).
570 These were invented primarily for use of the git difftool command,
571 and may not be very useful otherwise.
572 -R
574 Swap two inputs; that is, show differences from index or on-disk
575 file to tree contents.
576 --relative[=<path>], --no-relative
578 When run from a subdirectory of the project, it can be told to
579 exclude changes outside the directory and show pathnames relative
580 to it with this option. When you are not in a subdirectory (e.g. in
581 a bare repository), you can name which subdirectory to make the
582 output relative to by giving a <path> as an argument.
583 --no-relative can be used to countermand both diff.relative config
584 option and previous --relative.
585 -a, --text
587 Treat all files as text.
588 --ignore-cr-at-eol
590 Ignore carriage-return at the end of line when doing a comparison.
591 --ignore-space-at-eol
593 Ignore changes in whitespace at EOL.
594 -b, --ignore-space-change
596 Ignore changes in amount of whitespace. This ignores whitespace at
597 line end, and considers all other sequences of one or more
598 whitespace characters to be equivalent.
599 -w, --ignore-all-space
601 Ignore whitespace when comparing lines. This ignores differences
602 even if one line has whitespace where the other line has none.
603 --ignore-blank-lines
605 Ignore changes whose lines are all blank.
606 -I<regex>, --ignore-matching-lines=<regex>
608 Ignore changes whose all lines match <regex>. This option may be
609 specified more than once.
610 --inter-hunk-context=<lines>
612 Show the context between diff hunks, up to the specified number of
613 lines, thereby fusing hunks that are close to each other. Defaults
614 to diff.interHunkContext or 0 if the config option is unset.
615 -W, --function-context
617 Show whole function as context lines for each change. The function
618 names are determined in the same way as git diff works out patch
619 hunk headers (see Defining a custom hunk-header in
620 gitattributes(5)).
621 --exit-code
623 Make the program exit with codes similar to diff(1). That is, it
624 exits with 1 if there were differences and 0 means no differences.
625 --quiet
627 Disable all output of the program. Implies --exit-code.
628 --ext-diff
630 Allow an external diff helper to be executed. If you set an
631 external diff driver with gitattributes(5), you need to use this
632 option with git-log(1) and friends.
633 --no-ext-diff
635 Disallow external diff drivers.
636 --textconv, --no-textconv
638 Allow (or disallow) external text conversion filters to be run when
639 comparing binary files. See gitattributes(5) for details. Because
640 textconv filters are typically a one-way conversion, the resulting
641 diff is suitable for human consumption, but cannot be applied. For
642 this reason, textconv filters are enabled by default only for git-
643 diff(1) and git-log(1), but not for git-format-patch(1) or diff
644 plumbing commands.
645 --ignore-submodules[=<when>]
647 Ignore changes to submodules in the diff generation. <when> can be
648 either "none", "untracked", "dirty" or "all", which is the default.
649 Using "none" will consider the submodule modified when it either
650 contains untracked or modified files or its HEAD differs from the
651 commit recorded in the superproject and can be used to override any
652 settings of the ignore option in git-config(1) or gitmodules(5).
653 When "untracked" is used submodules are not considered dirty when
654 they only contain untracked content (but they are still scanned for
655 modified content). Using "dirty" ignores all changes to the work
656 tree of submodules, only changes to the commits stored in the
657 superproject are shown (this was the behavior until 1.7.0). Using
658 "all" hides all changes to submodules.
659 --src-prefix=<prefix>
661 Show the given source prefix instead of "a/".
662 --dst-prefix=<prefix>
664 Show the given destination prefix instead of "b/".
665 --no-prefix
667 Do not show any source or destination prefix.
668 --line-prefix=<prefix>
670 Prepend an additional prefix to every line of output.
671 --ita-invisible-in-index
673 By default entries added by "git add -N" appear as an existing
674 empty file in "git diff" and a new file in "git diff --cached".
675 This option makes the entry appear as a new file in "git diff" and
676 non-existent in "git diff --cached". This option could be reverted
677 with --ita-visible-in-index. Both options are experimental and
678 could be removed in future.
679 For more detailed explanation on these common options, see also
681 gitdiffcore(7).
682 -1 --base, -2 --ours, -3 --theirs, -0
684 Diff against the "base" version, "our branch" or "their branch"
685 respectively. With these options, diffs for merged entries are not
686 shown.
687 The default is to diff against our branch (-2) and the cleanly
689 resolved paths. The option -0 can be given to omit diff output for
690 unmerged entries and just show "Unmerged".
691 -c, --cc
693 This compares stage 2 (our branch), stage 3 (their branch) and the
694 working tree file and outputs a combined diff, similar to the way
695 diff-tree shows a merge commit with these flags.
696 -q
698 Remain silent even on nonexistent files
699
701 The raw output format from "git-diff-index", "git-diff-tree",
702 "git-diff-files" and "git diff --raw" are very similar.
703 These commands all compare two sets of things; what is compared
705 differs:
706 git-diff-index <tree-ish>
708 compares the <tree-ish> and the files on the filesystem.
709 git-diff-index --cached <tree-ish>
711 compares the <tree-ish> and the index.
712 git-diff-tree [-r] <tree-ish-1> <tree-ish-2> [<pattern>...]
714 compares the trees named by the two arguments.
715 git-diff-files [<pattern>...]
717 compares the index and the files on the filesystem.
718 The "git-diff-tree" command begins its output by printing the hash of
720 what is being compared. After that, all the commands print one output
721 line per changed file.
722 An output line is formatted this way:
724 in-place edit :100644 100644 bcd1234 0123456 M file0
726 copy-edit :100644 100644 abcd123 1234567 C68 file1 file2
727 rename-edit :100644 100644 abcd123 1234567 R86 file1 file3
728 create :000000 100644 0000000 1234567 A file4
729 delete :100644 000000 1234567 0000000 D file5
730 unmerged :000000 000000 0000000 0000000 U file6
731 That is, from the left to the right:
733 1. a colon.
735 2. mode for "src"; 000000 if creation or unmerged.
737 3. a space.
739 4. mode for "dst"; 000000 if deletion or unmerged.
741 5. a space.
743 6. sha1 for "src"; 0{40} if creation or unmerged.
745 7. a space.
747 8. sha1 for "dst"; 0{40} if creation, unmerged or "look at work tree".
749 9. a space.
751 10. status, followed by optional "score" number.
753 11. a tab or a NUL when -z option is used.
755 12. path for "src"
757 13. a tab or a NUL when -z option is used; only exists for C or R.
759 14. path for "dst"; only exists for C or R.
761 15. an LF or a NUL when -z option is used, to terminate the record.
763 Possible status letters are:
765 • A: addition of a file
767 • C: copy of a file into a new one
769 • D: deletion of a file
771 • M: modification of the contents or mode of a file
773 • R: renaming of a file
775 • T: change in the type of the file (regular file, symbolic link or
777 submodule)
778 • U: file is unmerged (you must complete the merge before it can be
780 committed)
781 • X: "unknown" change type (most probably a bug, please report it)
783 Status letters C and R are always followed by a score (denoting the
785 percentage of similarity between the source and target of the move or
786 copy). Status letter M may be followed by a score (denoting the
787 percentage of dissimilarity) for file rewrites.
788 <sha1> is shown as all 0’s if a file is new on the filesystem and it is
790 out of sync with the index.
791 Example:
793 :100644 100644 5be4a4a 0000000 M file.c
795 Without the -z option, pathnames with "unusual" characters are quoted
797 as explained for the configuration variable core.quotePath (see git-
798 config(1)). Using -z the filename is output verbatim and the line is
799 terminated by a NUL byte.
800
802 "git-diff-tree", "git-diff-files" and "git-diff --raw" can take -c or
803 --cc option to generate diff output also for merge commits. The output
804 differs from the format described above in the following way:
805 1. there is a colon for each parent
807 2. there are more "src" modes and "src" sha1
809 3. status is concatenated status characters for each parent
811 4. no optional "score" number
813 5. tab-separated pathname(s) of the file
815 For -c and --cc, only the destination or final path is shown even if
817 the file was renamed on any side of history. With --combined-all-paths,
818 the name of the path in each parent is shown followed by the name of
819 the path in the merge commit.
820 Examples for -c and --cc without --combined-all-paths:
822 ::100644 100644 100644 fabadb8 cc95eb0 4866510 MM desc.c
824 ::100755 100755 100755 52b7a2d 6d1ac04 d2ac7d7 RM bar.sh
825 ::100644 100644 100644 e07d6c5 9042e82 ee91881 RR phooey.c
826 Examples when --combined-all-paths added to either -c or --cc:
828 ::100644 100644 100644 fabadb8 cc95eb0 4866510 MM desc.c desc.c desc.c
830 ::100755 100755 100755 52b7a2d 6d1ac04 d2ac7d7 RM foo.sh bar.sh bar.sh
831 ::100644 100644 100644 e07d6c5 9042e82 ee91881 RR fooey.c fuey.c phooey.c
832 Note that combined diff lists only files which were modified from all
834 parents.
835
837 Running git-diff(1), git-log(1), git-show(1), git-diff-index(1), git-
838 diff-tree(1), or git-diff-files(1) with the -p option produces patch
839 text. You can customize the creation of patch text via the
840 GIT_EXTERNAL_DIFF and the GIT_DIFF_OPTS environment variables (see
841 git(1)), and the diff attribute (see gitattributes(5)).
842 What the -p option produces is slightly different from the traditional
844 diff format:
845 1. It is preceded with a "git diff" header that looks like this:
847 diff --git a/file1 b/file2
849 The a/ and b/ filenames are the same unless rename/copy is
851 involved. Especially, even for a creation or a deletion, /dev/null
852 is not used in place of the a/ or b/ filenames.
853 When rename/copy is involved, file1 and file2 show the name of the
855 source file of the rename/copy and the name of the file that
856 rename/copy produces, respectively.
857 2. It is followed by one or more extended header lines:
859 old mode <mode>
861 new mode <mode>
862 deleted file mode <mode>
863 new file mode <mode>
864 copy from <path>
865 copy to <path>
866 rename from <path>
867 rename to <path>
868 similarity index <number>
869 dissimilarity index <number>
870 index <hash>..<hash> <mode>
871 File modes are printed as 6-digit octal numbers including the file
873 type and file permission bits.
874 Path names in extended headers do not include the a/ and b/
876 prefixes.
877 The similarity index is the percentage of unchanged lines, and the
879 dissimilarity index is the percentage of changed lines. It is a
880 rounded down integer, followed by a percent sign. The similarity
881 index value of 100% is thus reserved for two equal files, while
882 100% dissimilarity means that no line from the old file made it
883 into the new one.
884 The index line includes the blob object names before and after the
886 change. The <mode> is included if the file mode does not change;
887 otherwise, separate lines indicate the old and the new mode.
888 3. Pathnames with "unusual" characters are quoted as explained for the
890 configuration variable core.quotePath (see git-config(1)).
891 4. All the file1 files in the output refer to files before the commit,
893 and all the file2 files refer to files after the commit. It is
894 incorrect to apply each change to each file sequentially. For
895 example, this patch will swap a and b:
896 diff --git a/a b/b
898 rename from a
899 rename to b
900 diff --git a/b b/a
901 rename from b
902 rename to a
903 5. Hunk headers mention the name of the function to which the hunk
905 applies. See "Defining a custom hunk-header" in gitattributes(5)
906 for details of how to tailor to this to specific languages.
907
909 Any diff-generating command can take the -c or --cc option to produce a
910 combined diff when showing a merge. This is the default format when
911 showing merges with git-diff(1) or git-show(1). Note also that you can
912 give suitable --diff-merges option to any of these commands to force
913 generation of diffs in specific format.
914 A "combined diff" format looks like this:
916 diff --combined describe.c
918 index fabadb8,cc95eb0..4866510
919 --- a/describe.c
920 +++ b/describe.c
921 @@@ -98,20 -98,12 +98,20 @@@
922 return (a_date > b_date) ? -1 : (a_date == b_date) ? 0 : 1;
923 }
924 - static void describe(char *arg)
926 -static void describe(struct commit *cmit, int last_one)
927 ++static void describe(char *arg, int last_one)
928 {
929 + unsigned char sha1[20];
930 + struct commit *cmit;
931 struct commit_list *list;
932 static int initialized = 0;
933 struct commit_name *n;
934 + if (get_sha1(arg, sha1) < 0)
936 + usage(describe_usage);
937 + cmit = lookup_commit_reference(sha1);
938 + if (!cmit)
939 + usage(describe_usage);
940 +
941 if (!initialized) {
942 initialized = 1;
943 for_each_ref(get_name);
944 1. It is preceded with a "git diff" header, that looks like this (when
946 the -c option is used):
947 diff --combined file
949 or like this (when the --cc option is used):
951 diff --cc file
953 2. It is followed by one or more extended header lines (this example
955 shows a merge with two parents):
956 index <hash>,<hash>..<hash>
958 mode <mode>,<mode>..<mode>
959 new file mode <mode>
960 deleted file mode <mode>,<mode>
961 The mode <mode>,<mode>..<mode> line appears only if at least one of
963 the <mode> is different from the rest. Extended headers with
964 information about detected contents movement (renames and copying
965 detection) are designed to work with diff of two <tree-ish> and are
966 not used by combined diff format.
967 3. It is followed by two-line from-file/to-file header
969 --- a/file
971 +++ b/file
972 Similar to two-line header for traditional unified diff format,
974 /dev/null is used to signal created or deleted files.
975 However, if the --combined-all-paths option is provided, instead of
977 a two-line from-file/to-file you get a N+1 line from-file/to-file
978 header, where N is the number of parents in the merge commit
979 --- a/file
981 --- a/file
982 --- a/file
983 +++ b/file
984 This extended format can be useful if rename or copy detection is
986 active, to allow you to see the original name of the file in
987 different parents.
988 4. Chunk header format is modified to prevent people from accidentally
990 feeding it to patch -p1. Combined diff format was created for
991 review of merge commit changes, and was not meant to be applied.
992 The change is similar to the change in the extended index header:
993 @@@ <from-file-range> <from-file-range> <to-file-range> @@@
995 There are (number of parents + 1) @ characters in the chunk header
997 for combined diff format.
998 Unlike the traditional unified diff format, which shows two files A and
1000 B with a single column that has - (minus — appears in A but removed in
1001 B), + (plus — missing in A but added to B), or " " (space — unchanged)
1002 prefix, this format compares two or more files file1, file2,... with
1003 one file X, and shows how X differs from each of fileN. One column for
1004 each of fileN is prepended to the output line to note how X’s line is
1005 different from it.
1006 A - character in the column N means that the line appears in fileN but
1008 it does not appear in the result. A + character in the column N means
1009 that the line appears in the result, and fileN does not have that line
1010 (in other words, the line was added, from the point of view of that
1011 parent).
1012 In the above example output, the function signature was changed from
1014 both files (hence two - removals from both file1 and file2, plus ++ to
1015 mean one line that was added does not appear in either file1 or file2).
1016 Also eight other lines are the same from file1 but do not appear in
1017 file2 (hence prefixed with +).
1018 When shown by git diff-tree -c, it compares the parents of a merge
1020 commit with the merge result (i.e. file1..fileN are the parents). When
1021 shown by git diff-files -c, it compares the two unresolved merge
1022 parents with the working tree file (i.e. file1 is stage 2 aka "our
1023 version", file2 is stage 3 aka "their version").
1024
1026 The --summary option describes newly added, deleted, renamed and copied
1027 files. The --stat option adds diffstat(1) graph to the output. These
1028 options can be combined with other options, such as -p, and are meant
1029 for human consumption.
1030 When showing a change that involves a rename or a copy, --stat output
1032 formats the pathnames compactly by combining common prefix and suffix
1033 of the pathnames. For example, a change that moves arch/i386/Makefile
1034 to arch/x86/Makefile while modifying 4 lines will be shown like this:
1035 arch/{i386 => x86}/Makefile | 4 +--
1037 The --numstat option gives the diffstat(1) information but is designed
1039 for easier machine consumption. An entry in --numstat output looks like
1040 this:
1041 1 2 README
1043 3 1 arch/{i386 => x86}/Makefile
1044 That is, from left to right:
1046 1. the number of added lines;
1048 2. a tab;
1050 3. the number of deleted lines;
1052 4. a tab;
1054 5. pathname (possibly with rename/copy information);
1056 6. a newline.
1058 When -z output option is in effect, the output is formatted this way:
1060 1 2 README NUL
1062 3 1 NUL arch/i386/Makefile NUL arch/x86/Makefile NUL
1063 That is:
1065 1. the number of added lines;
1067 2. a tab;
1069 3. the number of deleted lines;
1071 4. a tab;
1073 5. a NUL (only exists if renamed/copied);
1075 6. pathname in preimage;
1077 7. a NUL (only exists if renamed/copied);
1079 8. pathname in postimage (only exists if renamed/copied);
1081 9. a NUL.
1083 The extra NUL before the preimage path in renamed case is to allow
1085 scripts that read the output to tell if the current record being read
1086 is a single-path record or a rename/copy record without reading ahead.
1087 After reading added and deleted lines, reading up to NUL would yield
1088 the pathname, but if that is NUL, the record will show two paths.
1089
1091 Part of the git(1) suite
1092Git 2.36.1 2022-05-05 GIT-DIFF-FILES(1)