1GIT-DIFF-INDEX(1) Git Manual GIT-DIFF-INDEX(1)
2
6 git-diff-index - Compares content and mode of blobs between the index
7 and repository
8
10 git diff-index [-m] [--cached] [<common diff options>] <tree-ish>
11 [<path>...]
12
14 Compares the content and mode of the blobs found via a tree object with
15 the content of the current index and, optionally ignoring the stat
16 state of the file on disk. When paths are specified, compares only
17 those named paths. Otherwise all entries in the index are compared.
18
20 -p, -u
21 Generate patch (see section on generating patches).
22 -U<n>, --unified=<n>
24 Generate diffs with <n> lines of context instead of the usual
25 three. Implies -p.
26 --raw
28 Generate the raw format. This is the default.
29 --patch-with-raw
31 Synonym for -p --raw.
32 --patience
34 Generate a diff using the "patience diff" algorithm.
35 --stat[=width[,name-width]]
37 Generate a diffstat. You can override the default output width for
38 80-column terminal by --stat=width. The width of the filename part
39 can be controlled by giving another width to it separated by a
40 comma.
41 --numstat
43 Similar to --stat, but shows number of added and deleted lines in
44 decimal notation and pathname without abbreviation, to make it more
45 machine friendly. For binary files, outputs two - instead of saying
46 0 0.
47 --shortstat
49 Output only the last line of the --stat format containing total
50 number of modified files, as well as number of added and deleted
51 lines.
52 --dirstat[=limit]
54 Output the distribution of relative amount of changes (number of
55 lines added or removed) for each sub-directory. Directories with
56 changes below a cut-off percent (3% by default) are not shown. The
57 cut-off percent can be set with --dirstat=limit. Changes in a child
58 directory is not counted for the parent directory, unless
59 --cumulative is used.
60 --dirstat-by-file[=limit]
62 Same as --dirstat, but counts changed files instead of lines.
63 --summary
65 Output a condensed summary of extended header information such as
66 creations, renames and mode changes.
67 --patch-with-stat
69 Synonym for -p --stat.
70 -z
72 When --raw, --numstat, --name-only or --name-status has been given,
73 do not munge pathnames and use NULs as output field terminators.
74 Without this option, each pathname output will have TAB, LF, double
76 quotes, and backslash characters replaced with \t, \n, \", and \\,
77 respectively, and the pathname will be enclosed in double quotes if
78 any of those replacements occurred.
79 --name-only
81 Show only names of changed files.
82 --name-status
84 Show only names and status of changed files. See the description of
85 the --diff-filter option on what the status letters mean.
86 --submodule[=<format>]
88 Chose the output format for submodule differences. <format> can be
89 one of short and log. short just shows pairs of commit names, this
90 format is used when this option is not given. log is the default
91 value for this option and lists the commits in that commit range
92 like the summary option of git-submodule(1) does.
93 --color[=<when>]
95 Show colored diff. The value must be always (the default), never,
96 or auto.
97 --no-color
99 Turn off colored diff, even when the configuration file gives the
100 default to color output. Same as --color=never.
101 --color-words[=<regex>]
103 Show colored word diff, i.e., color words which have changed. By
104 default, words are separated by whitespace.
105 When a <regex> is specified, every non-overlapping match of the
107 <regex> is considered a word. Anything between these matches is
108 considered whitespace and ignored(!) for the purposes of finding
109 differences. You may want to append |[^[:space:]] to your regular
110 expression to make sure that it matches all non-whitespace
111 characters. A match that contains a newline is silently
112 truncated(!) at the newline.
113 The regex can also be set via a diff driver or configuration
115 option, see gitattributes(1) or git-config(1). Giving it explicitly
116 overrides any diff driver or configuration setting. Diff drivers
117 override configuration settings.
118 --no-renames
120 Turn off rename detection, even when the configuration file gives
121 the default to do so.
122 --check
124 Warn if changes introduce trailing whitespace or an indent that
125 uses a space before a tab. Exits with non-zero status if problems
126 are found. Not compatible with --exit-code.
127 --full-index
129 Instead of the first handful of characters, show the full pre- and
130 post-image blob object names on the "index" line when generating
131 patch format output.
132 --binary
134 In addition to --full-index, output a binary diff that can be
135 applied with git-apply.
136 --abbrev[=<n>]
138 Instead of showing the full 40-byte hexadecimal object name in
139 diff-raw format output and diff-tree header lines, show only a
140 partial prefix. This is independent of the --full-index option
141 above, which controls the diff-patch output format. Non default
142 number of digits can be specified with --abbrev=<n>.
143 -B
145 Break complete rewrite changes into pairs of delete and create.
146 -M
148 Detect renames.
149 -C
151 Detect copies as well as renames. See also --find-copies-harder.
152 --diff-filter=[ACDMRTUXB*]
154 Select only files that are Added (A), Copied (C), Deleted (D),
155 Modified (M), Renamed (R), have their type (i.e. regular file,
156 symlink, submodule, ...) changed (T), are Unmerged (U), are Unknown
157 (X), or have had their pairing Broken (B). Any combination of the
158 filter characters may be used. When * (All-or-none) is added to the
159 combination, all paths are selected if there is any file that
160 matches other criteria in the comparison; if there is no file that
161 matches other criteria, nothing is selected.
162 --find-copies-harder
164 For performance reasons, by default, -C option finds copies only if
165 the original file of the copy was modified in the same changeset.
166 This flag makes the command inspect unmodified files as candidates
167 for the source of copy. This is a very expensive operation for
168 large projects, so use it with caution. Giving more than one -C
169 option has the same effect.
170 -l<num>
172 The -M and -C options require O(n^2) processing time where n is the
173 number of potential rename/copy targets. This option prevents
174 rename/copy detection from running if the number of rename/copy
175 targets exceeds the specified number.
176 -S<string>
178 Look for differences that introduce or remove an instance of
179 <string>. Note that this is different than the string simply
180 appearing in diff output; see the pickaxe entry in gitdiffcore(7)
181 for more details.
182 --pickaxe-all
184 When -S finds a change, show all the changes in that changeset, not
185 just the files that contain the change in <string>.
186 --pickaxe-regex
188 Make the <string> not a plain string but an extended POSIX regex to
189 match.
190 -O<orderfile>
192 Output the patch in the order specified in the <orderfile>, which
193 has one shell glob pattern per line.
194 -R
196 Swap two inputs; that is, show differences from index or on-disk
197 file to tree contents.
198 --relative[=<path>]
200 When run from a subdirectory of the project, it can be told to
201 exclude changes outside the directory and show pathnames relative
202 to it with this option. When you are not in a subdirectory (e.g. in
203 a bare repository), you can name which subdirectory to make the
204 output relative to by giving a <path> as an argument.
205 -a, --text
207 Treat all files as text.
208 --ignore-space-at-eol
210 Ignore changes in whitespace at EOL.
211 -b, --ignore-space-change
213 Ignore changes in amount of whitespace. This ignores whitespace at
214 line end, and considers all other sequences of one or more
215 whitespace characters to be equivalent.
216 -w, --ignore-all-space
218 Ignore whitespace when comparing lines. This ignores differences
219 even if one line has whitespace where the other line has none.
220 --inter-hunk-context=<lines>
222 Show the context between diff hunks, up to the specified number of
223 lines, thereby fusing hunks that are close to each other.
224 --exit-code
226 Make the program exit with codes similar to diff(1). That is, it
227 exits with 1 if there were differences and 0 means no differences.
228 --quiet
230 Disable all output of the program. Implies --exit-code.
231 --ext-diff
233 Allow an external diff helper to be executed. If you set an
234 external diff driver with gitattributes(5), you need to use this
235 option with git-log(1) and friends.
236 --no-ext-diff
238 Disallow external diff drivers.
239 --ignore-submodules
241 Ignore changes to submodules in the diff generation.
242 --src-prefix=<prefix>
244 Show the given source prefix instead of "a/".
245 --dst-prefix=<prefix>
247 Show the given destination prefix instead of "b/".
248 --no-prefix
250 Do not show any source or destination prefix.
251 For more detailed explanation on these common options, see also
253 gitdiffcore(7).
254 <tree-ish>
256 The id of a tree object to diff against.
257 --cached
259 do not consider the on-disk file at all
260 -m
262 By default, files recorded in the index but not checked out are
263 reported as deleted. This flag makes git diff-index say that all
264 non-checked-out files are up to date.
265
267 The raw output format from "git-diff-index", "git-diff-tree",
268 "git-diff-files" and "git diff --raw" are very similar.
269 These commands all compare two sets of things; what is compared
271 differs:
272 git-diff-index <tree-ish>
274 compares the <tree-ish> and the files on the filesystem.
275 git-diff-index --cached <tree-ish>
277 compares the <tree-ish> and the index.
278 git-diff-tree [-r] <tree-ish-1> <tree-ish-2> [<pattern>...]
280 compares the trees named by the two arguments.
281 git-diff-files [<pattern>...]
283 compares the index and the files on the filesystem.
284 The "git-diff-tree" command begins its output by printing the hash of
286 what is being compared. After that, all the commands print one output
287 line per changed file.
288 An output line is formatted this way:
290 in-place edit :100644 100644 bcd1234... 0123456... M file0
292 copy-edit :100644 100644 abcd123... 1234567... C68 file1 file2
293 rename-edit :100644 100644 abcd123... 1234567... R86 file1 file3
294 create :000000 100644 0000000... 1234567... A file4
295 delete :100644 000000 1234567... 0000000... D file5
296 unmerged :000000 000000 0000000... 0000000... U file6
297 That is, from the left to the right:
300 1. a colon.
302 2. mode for "src"; 000000 if creation or unmerged.
304 3. a space.
306 4. mode for "dst"; 000000 if deletion or unmerged.
308 5. a space.
310 6. sha1 for "src"; 0{40} if creation or unmerged.
312 7. a space.
314 8. sha1 for "dst"; 0{40} if creation, unmerged or "look at work tree".
316 9. a space.
318 10. status, followed by optional "score" number.
320 11. a tab or a NUL when -z option is used.
322 12. path for "src"
324 13. a tab or a NUL when -z option is used; only exists for C or R.
326 14. path for "dst"; only exists for C or R.
328 15. an LF or a NUL when -z option is used, to terminate the record.
330 Possible status letters are:
332 · A: addition of a file
334 · C: copy of a file into a new one
336 · D: deletion of a file
338 · M: modification of the contents or mode of a file
340 · R: renaming of a file
342 · T: change in the type of the file
344 · U: file is unmerged (you must complete the merge before it can be
346 committed)
347 · X: "unknown" change type (most probably a bug, please report it)
349 Status letters C and R are always followed by a score (denoting the
351 percentage of similarity between the source and target of the move or
352 copy), and are the only ones to be so.
353 <sha1> is shown as all 0’s if a file is new on the filesystem and it is
355 out of sync with the index.
356 Example:
358 :100644 100644 5be4a4...... 000000...... M file.c
360 When -z option is not used, TAB, LF, and backslash characters in
363 pathnames are represented as \t, \n, and \\, respectively.
364
366 "git-diff-tree", "git-diff-files" and "git-diff --raw" can take -c or
367 --cc option to generate diff output also for merge commits. The output
368 differs from the format described above in the following way:
369 1. there is a colon for each parent
371 2. there are more "src" modes and "src" sha1
373 3. status is concatenated status characters for each parent
375 4. no optional "score" number
377 5. single path, only for "dst"
379 Example:
381 ::100644 100644 100644 fabadb8... cc95eb0... 4866510... MM describe.c
383 Note that combined diff lists only files which were modified from all
386 parents.
387
389 When "git-diff-index", "git-diff-tree", or "git-diff-files" are run
390 with a -p option, "git diff" without the --raw option, or "git log"
391 with the "-p" option, they do not produce the output described above;
392 instead they produce a patch file. You can customize the creation of
393 such patches via the GIT_EXTERNAL_DIFF and the GIT_DIFF_OPTS
394 environment variables.
395 What the -p option produces is slightly different from the traditional
397 diff format.
398 1. It is preceded with a "git diff" header, that looks like this:
400 diff --git a/file1 b/file2
402 The a/ and b/ filenames are the same unless rename/copy is
404 involved. Especially, even for a creation or a deletion, /dev/null
405 is not used in place of a/ or b/ filenames.
406 When rename/copy is involved, file1 and file2 show the name of the
408 source file of the rename/copy and the name of the file that
409 rename/copy produces, respectively.
410 2. It is followed by one or more extended header lines:
412 old mode <mode>
414 new mode <mode>
415 deleted file mode <mode>
416 new file mode <mode>
417 copy from <path>
418 copy to <path>
419 rename from <path>
420 rename to <path>
421 similarity index <number>
422 dissimilarity index <number>
423 index <hash>..<hash> <mode>
424 3. TAB, LF, double quote and backslash characters in pathnames are
426 represented as \t, \n, \" and \\, respectively. If there is need
427 for such substitution then the whole pathname is put in double
428 quotes.
429 The similarity index is the percentage of unchanged lines, and the
431 dissimilarity index is the percentage of changed lines. It is a rounded
432 down integer, followed by a percent sign. The similarity index value of
433 100% is thus reserved for two equal files, while 100% dissimilarity
434 means that no line from the old file made it into the new one.
435
437 "git-diff-tree", "git-diff-files" and "git-diff" can take -c or --cc
438 option to produce combined diff. For showing a merge commit with "git
439 log -p", this is the default format; you can force showing full diff
440 with the -m option. A combined diff format looks like this:
441 diff --combined describe.c
443 index fabadb8,cc95eb0..4866510
444 --- a/describe.c
445 +++ b/describe.c
446 @@@ -98,20 -98,12 +98,20 @@@
447 return (a_date > b_date) ? -1 : (a_date == b_date) ? 0 : 1;
448 }
449 - static void describe(char *arg)
451 -static void describe(struct commit *cmit, int last_one)
452 ++static void describe(char *arg, int last_one)
453 {
454 + unsigned char sha1[20];
455 + struct commit *cmit;
456 struct commit_list *list;
457 static int initialized = 0;
458 struct commit_name *n;
459 + if (get_sha1(arg, sha1) < 0)
461 + usage(describe_usage);
462 + cmit = lookup_commit_reference(sha1);
463 + if (!cmit)
464 + usage(describe_usage);
465 +
466 if (!initialized) {
467 initialized = 1;
468 for_each_ref(get_name);
469 1. It is preceded with a "git diff" header, that looks like this (when
473 -c option is used):
474 diff --combined file
476 or like this (when --cc option is used):
478 diff --cc file
480 2. It is followed by one or more extended header lines (this example
482 shows a merge with two parents):
483 index <hash>,<hash>..<hash>
485 mode <mode>,<mode>..<mode>
486 new file mode <mode>
487 deleted file mode <mode>,<mode>
488 The mode <mode>,<mode>..<mode> line appears only if at least one of
490 the <mode> is different from the rest. Extended headers with
491 information about detected contents movement (renames and copying
492 detection) are designed to work with diff of two <tree-ish> and are
493 not used by combined diff format.
494 3. It is followed by two-line from-file/to-file header
496 --- a/file
498 +++ b/file
499 Similar to two-line header for traditional unified diff format,
501 /dev/null is used to signal created or deleted files.
502 4. Chunk header format is modified to prevent people from accidentally
504 feeding it to patch -p1. Combined diff format was created for
505 review of merge commit changes, and was not meant for apply. The
506 change is similar to the change in the extended index header:
507 @@@ <from-file-range> <from-file-range> <to-file-range> @@@
509 There are (number of parents + 1) @ characters in the chunk header
511 for combined diff format.
512 Unlike the traditional unified diff format, which shows two files A and
514 B with a single column that has - (minus — appears in A but removed in
515 B), + (plus — missing in A but added to B), or " " (space — unchanged)
516 prefix, this format compares two or more files file1, file2,... with
517 one file X, and shows how X differs from each of fileN. One column for
518 each of fileN is prepended to the output line to note how X’s line is
519 different from it.
520 A - character in the column N means that the line appears in fileN but
522 it does not appear in the result. A + character in the column N means
523 that the line appears in the result, and fileN does not have that line
524 (in other words, the line was added, from the point of view of that
525 parent).
526 In the above example output, the function signature was changed from
528 both files (hence two - removals from both file1 and file2, plus ++ to
529 mean one line that was added does not appear in either file1 nor
530 file2). Also eight other lines are the same from file1 but do not
531 appear in file2 (hence prefixed with +).
532 When shown by git diff-tree -c, it compares the parents of a merge
534 commit with the merge result (i.e. file1..fileN are the parents). When
535 shown by git diff-files -c, it compares the two unresolved merge
536 parents with the working tree file (i.e. file1 is stage 2 aka "our
537 version", file2 is stage 3 aka "their version").
538
540 The --summary option describes newly added, deleted, renamed and copied
541 files. The --stat option adds diffstat(1) graph to the output. These
542 options can be combined with other options, such as -p, and are meant
543 for human consumption.
544 When showing a change that involves a rename or a copy, --stat output
546 formats the pathnames compactly by combining common prefix and suffix
547 of the pathnames. For example, a change that moves arch/i386/Makefile
548 to arch/x86/Makefile while modifying 4 lines will be shown like this:
549 arch/{i386 => x86}/Makefile | 4 +--
551 The --numstat option gives the diffstat(1) information but is designed
554 for easier machine consumption. An entry in --numstat output looks like
555 this:
556 1 2 README
558 3 1 arch/{i386 => x86}/Makefile
559 That is, from left to right:
562 1. the number of added lines;
564 2. a tab;
566 3. the number of deleted lines;
568 4. a tab;
570 5. pathname (possibly with rename/copy information);
572 6. a newline.
574 When -z output option is in effect, the output is formatted this way:
576 1 2 README NUL
578 3 1 NUL arch/i386/Makefile NUL arch/x86/Makefile NUL
579 That is:
582 1. the number of added lines;
584 2. a tab;
586 3. the number of deleted lines;
588 4. a tab;
590 5. a NUL (only exists if renamed/copied);
592 6. pathname in preimage;
594 7. a NUL (only exists if renamed/copied);
596 8. pathname in postimage (only exists if renamed/copied);
598 9. a NUL.
600 The extra NUL before the preimage path in renamed case is to allow
602 scripts that read the output to tell if the current record being read
603 is a single-path record or a rename/copy record without reading ahead.
604 After reading added and deleted lines, reading up to NUL would yield
605 the pathname, but if that is NUL, the record will show two paths.
606
608 You can choose whether you want to trust the index file entirely (using
609 the --cached flag) or ask the diff logic to show any files that don’t
610 match the stat state as being "tentatively changed". Both of these
611 operations are very useful indeed.
612
614 If --cached is specified, it allows you to ask:
615 show me the differences between HEAD and the current index
617 contents (the ones I´d write using ´git write-tree´)
618 For example, let’s say that you have worked on your working directory,
620 updated some files in the index and are ready to commit. You want to
621 see exactly what you are going to commit, without having to write a new
622 tree object and compare it that way, and to do that, you just do
623 git diff-index --cached HEAD
625 Example: let’s say I had renamed commit.c to git-commit.c, and I had
627 done an update-index to make that effective in the index file. git
628 diff-files wouldn’t show anything at all, since the index file matches
629 my working directory. But doing a git diff-index does:
630 torvalds@ppc970:~/git> git diff-index --cached HEAD
632 -100644 blob 4161aecc6700a2eb579e842af0b7f22b98443f74 commit.c
633 +100644 blob 4161aecc6700a2eb579e842af0b7f22b98443f74 git-commit.c
634 You can see easily that the above is a rename.
636 In fact, git diff-index --cached should always be entirely equivalent
638 to actually doing a git write-tree and comparing that. Except this one
639 is much nicer for the case where you just want to check where you are.
640 So doing a git diff-index --cached is basically very useful when you
642 are asking yourself "what have I already marked for being committed,
643 and what’s the difference to a previous tree".
644
646 The "non-cached" mode takes a different approach, and is potentially
647 the more useful of the two in that what it does can’t be emulated with
648 a git write-tree + git diff-tree. Thus that’s the default mode. The
649 non-cached version asks the question:
650 show me the differences between HEAD and the currently checked out
652 tree - index contents _and_ files that aren´t up-to-date
653 which is obviously a very useful question too, since that tells you
655 what you could commit. Again, the output matches the git diff-tree -r
656 output to a tee, but with a twist.
657 The twist is that if some file doesn’t match the index, we don’t have a
659 backing store thing for it, and we use the magic "all-zero" sha1 to
660 show that. So let’s say that you have edited kernel/sched.c, but have
661 not actually done a git update-index on it yet - there is no "object"
662 associated with the new state, and you get:
663 torvalds@ppc970:~/v2.6/linux> git diff-index HEAD
665 *100644->100664 blob 7476bb......->000000...... kernel/sched.c
666 i.e., it shows that the tree has changed, and that kernel/sched.c has
668 is not up-to-date and may contain new stuff. The all-zero sha1 means
669 that to get the real diff, you need to look at the object in the
670 working directory directly rather than do an object-to-object diff.
671 Note
673 As with other commands of this type, git diff-index does not
674 actually look at the contents of the file at all. So maybe
675 kernel/sched.c hasn’t actually changed, and it’s just that you
676 touched it. In either case, it’s a note that you need to git
677 update-index it to make the index be in sync.
678 Note
680 You can have a mixture of files show up as "has been updated" and
681 "is still dirty in the working directory" together. You can always
682 tell which file is in which state, since the "has been updated"
683 ones show a valid sha1, and the "not in sync with the index" ones
684 will always have the special all-zero sha1.
685
687 Written by Linus Torvalds <torvalds@osdl.org[1]>
688
690 Documentation by David Greaves, Junio C Hamano and the git-list
691 <git@vger.kernel.org[2]>.
692
694 Part of the git(1) suite
695
697 1. torvalds@osdl.org
698 mailto:torvalds@osdl.org
699 2. git@vger.kernel.org
701 mailto:git@vger.kernel.org
702Git 1.7.1 08/16/2017 GIT-DIFF-INDEX(1)