Merge branch 'mh/write-refs-sooner-2.4' into maint
[git.git] / diff.c
blob7500c5509550ccd9a86d620e06c51a71d61f8b6c
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15 #include "sigchain.h"
16 #include "submodule.h"
17 #include "ll-merge.h"
18 #include "string-list.h"
19 #include "argv-array.h"
21 #ifdef NO_FAST_WORKING_DIRECTORY
22 #define FAST_WORKING_DIRECTORY 0
23 #else
24 #define FAST_WORKING_DIRECTORY 1
25 #endif
27 static int diff_detect_rename_default;
28 static int diff_rename_limit_default = 400;
29 static int diff_suppress_blank_empty;
30 static int diff_use_color_default = -1;
31 static int diff_context_default = 3;
32 static const char *diff_word_regex_cfg;
33 static const char *external_diff_cmd_cfg;
34 static const char *diff_order_file_cfg;
35 int diff_auto_refresh_index = 1;
36 static int diff_mnemonic_prefix;
37 static int diff_no_prefix;
38 static int diff_stat_graph_width;
39 static int diff_dirstat_permille_default = 30;
40 static struct diff_options default_diff_options;
41 static long diff_algorithm;
43 static char diff_colors[][COLOR_MAXLEN] = {
44 GIT_COLOR_RESET,
45 GIT_COLOR_NORMAL, /* PLAIN */
46 GIT_COLOR_BOLD, /* METAINFO */
47 GIT_COLOR_CYAN, /* FRAGINFO */
48 GIT_COLOR_RED, /* OLD */
49 GIT_COLOR_GREEN, /* NEW */
50 GIT_COLOR_YELLOW, /* COMMIT */
51 GIT_COLOR_BG_RED, /* WHITESPACE */
52 GIT_COLOR_NORMAL, /* FUNCINFO */
55 static int parse_diff_color_slot(const char *var)
57 if (!strcasecmp(var, "plain"))
58 return DIFF_PLAIN;
59 if (!strcasecmp(var, "meta"))
60 return DIFF_METAINFO;
61 if (!strcasecmp(var, "frag"))
62 return DIFF_FRAGINFO;
63 if (!strcasecmp(var, "old"))
64 return DIFF_FILE_OLD;
65 if (!strcasecmp(var, "new"))
66 return DIFF_FILE_NEW;
67 if (!strcasecmp(var, "commit"))
68 return DIFF_COMMIT;
69 if (!strcasecmp(var, "whitespace"))
70 return DIFF_WHITESPACE;
71 if (!strcasecmp(var, "func"))
72 return DIFF_FUNCINFO;
73 return -1;
76 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
77 struct strbuf *errmsg)
79 char *params_copy = xstrdup(params_string);
80 struct string_list params = STRING_LIST_INIT_NODUP;
81 int ret = 0;
82 int i;
84 if (*params_copy)
85 string_list_split_in_place(&params, params_copy, ',', -1);
86 for (i = 0; i < params.nr; i++) {
87 const char *p = params.items[i].string;
88 if (!strcmp(p, "changes")) {
89 DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
90 DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
91 } else if (!strcmp(p, "lines")) {
92 DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
93 DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
94 } else if (!strcmp(p, "files")) {
95 DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
96 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
97 } else if (!strcmp(p, "noncumulative")) {
98 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
99 } else if (!strcmp(p, "cumulative")) {
100 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
101 } else if (isdigit(*p)) {
102 char *end;
103 int permille = strtoul(p, &end, 10) * 10;
104 if (*end == '.' && isdigit(*++end)) {
105 /* only use first digit */
106 permille += *end - '0';
107 /* .. and ignore any further digits */
108 while (isdigit(*++end))
109 ; /* nothing */
111 if (!*end)
112 options->dirstat_permille = permille;
113 else {
114 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
116 ret++;
118 } else {
119 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
120 ret++;
124 string_list_clear(&params, 0);
125 free(params_copy);
126 return ret;
129 static int parse_submodule_params(struct diff_options *options, const char *value)
131 if (!strcmp(value, "log"))
132 DIFF_OPT_SET(options, SUBMODULE_LOG);
133 else if (!strcmp(value, "short"))
134 DIFF_OPT_CLR(options, SUBMODULE_LOG);
135 else
136 return -1;
137 return 0;
140 static int git_config_rename(const char *var, const char *value)
142 if (!value)
143 return DIFF_DETECT_RENAME;
144 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
145 return DIFF_DETECT_COPY;
146 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
149 long parse_algorithm_value(const char *value)
151 if (!value)
152 return -1;
153 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
154 return 0;
155 else if (!strcasecmp(value, "minimal"))
156 return XDF_NEED_MINIMAL;
157 else if (!strcasecmp(value, "patience"))
158 return XDF_PATIENCE_DIFF;
159 else if (!strcasecmp(value, "histogram"))
160 return XDF_HISTOGRAM_DIFF;
161 return -1;
165 * These are to give UI layer defaults.
166 * The core-level commands such as git-diff-files should
167 * never be affected by the setting of diff.renames
168 * the user happens to have in the configuration file.
170 int git_diff_ui_config(const char *var, const char *value, void *cb)
172 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
173 diff_use_color_default = git_config_colorbool(var, value);
174 return 0;
176 if (!strcmp(var, "diff.context")) {
177 diff_context_default = git_config_int(var, value);
178 if (diff_context_default < 0)
179 return -1;
180 return 0;
182 if (!strcmp(var, "diff.renames")) {
183 diff_detect_rename_default = git_config_rename(var, value);
184 return 0;
186 if (!strcmp(var, "diff.autorefreshindex")) {
187 diff_auto_refresh_index = git_config_bool(var, value);
188 return 0;
190 if (!strcmp(var, "diff.mnemonicprefix")) {
191 diff_mnemonic_prefix = git_config_bool(var, value);
192 return 0;
194 if (!strcmp(var, "diff.noprefix")) {
195 diff_no_prefix = git_config_bool(var, value);
196 return 0;
198 if (!strcmp(var, "diff.statgraphwidth")) {
199 diff_stat_graph_width = git_config_int(var, value);
200 return 0;
202 if (!strcmp(var, "diff.external"))
203 return git_config_string(&external_diff_cmd_cfg, var, value);
204 if (!strcmp(var, "diff.wordregex"))
205 return git_config_string(&diff_word_regex_cfg, var, value);
206 if (!strcmp(var, "diff.orderfile"))
207 return git_config_pathname(&diff_order_file_cfg, var, value);
209 if (!strcmp(var, "diff.ignoresubmodules"))
210 handle_ignore_submodules_arg(&default_diff_options, value);
212 if (!strcmp(var, "diff.submodule")) {
213 if (parse_submodule_params(&default_diff_options, value))
214 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
215 value);
216 return 0;
219 if (!strcmp(var, "diff.algorithm")) {
220 diff_algorithm = parse_algorithm_value(value);
221 if (diff_algorithm < 0)
222 return -1;
223 return 0;
226 if (git_color_config(var, value, cb) < 0)
227 return -1;
229 return git_diff_basic_config(var, value, cb);
232 int git_diff_basic_config(const char *var, const char *value, void *cb)
234 const char *name;
236 if (!strcmp(var, "diff.renamelimit")) {
237 diff_rename_limit_default = git_config_int(var, value);
238 return 0;
241 if (userdiff_config(var, value) < 0)
242 return -1;
244 if (skip_prefix(var, "diff.color.", &name) ||
245 skip_prefix(var, "color.diff.", &name)) {
246 int slot = parse_diff_color_slot(name);
247 if (slot < 0)
248 return 0;
249 if (!value)
250 return config_error_nonbool(var);
251 return color_parse(value, diff_colors[slot]);
254 /* like GNU diff's --suppress-blank-empty option */
255 if (!strcmp(var, "diff.suppressblankempty") ||
256 /* for backwards compatibility */
257 !strcmp(var, "diff.suppress-blank-empty")) {
258 diff_suppress_blank_empty = git_config_bool(var, value);
259 return 0;
262 if (!strcmp(var, "diff.dirstat")) {
263 struct strbuf errmsg = STRBUF_INIT;
264 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
265 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
266 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
267 errmsg.buf);
268 strbuf_release(&errmsg);
269 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
270 return 0;
273 if (starts_with(var, "submodule."))
274 return parse_submodule_config_option(var, value);
276 return git_default_config(var, value, cb);
279 static char *quote_two(const char *one, const char *two)
281 int need_one = quote_c_style(one, NULL, NULL, 1);
282 int need_two = quote_c_style(two, NULL, NULL, 1);
283 struct strbuf res = STRBUF_INIT;
285 if (need_one + need_two) {
286 strbuf_addch(&res, '"');
287 quote_c_style(one, &res, NULL, 1);
288 quote_c_style(two, &res, NULL, 1);
289 strbuf_addch(&res, '"');
290 } else {
291 strbuf_addstr(&res, one);
292 strbuf_addstr(&res, two);
294 return strbuf_detach(&res, NULL);
297 static const char *external_diff(void)
299 static const char *external_diff_cmd = NULL;
300 static int done_preparing = 0;
302 if (done_preparing)
303 return external_diff_cmd;
304 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
305 if (!external_diff_cmd)
306 external_diff_cmd = external_diff_cmd_cfg;
307 done_preparing = 1;
308 return external_diff_cmd;
311 static struct diff_tempfile {
312 const char *name; /* filename external diff should read from */
313 char hex[41];
314 char mode[10];
315 char tmp_path[PATH_MAX];
316 } diff_temp[2];
318 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
320 struct emit_callback {
321 int color_diff;
322 unsigned ws_rule;
323 int blank_at_eof_in_preimage;
324 int blank_at_eof_in_postimage;
325 int lno_in_preimage;
326 int lno_in_postimage;
327 sane_truncate_fn truncate;
328 const char **label_path;
329 struct diff_words_data *diff_words;
330 struct diff_options *opt;
331 int *found_changesp;
332 struct strbuf *header;
335 static int count_lines(const char *data, int size)
337 int count, ch, completely_empty = 1, nl_just_seen = 0;
338 count = 0;
339 while (0 < size--) {
340 ch = *data++;
341 if (ch == '\n') {
342 count++;
343 nl_just_seen = 1;
344 completely_empty = 0;
346 else {
347 nl_just_seen = 0;
348 completely_empty = 0;
351 if (completely_empty)
352 return 0;
353 if (!nl_just_seen)
354 count++; /* no trailing newline */
355 return count;
358 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
360 if (!DIFF_FILE_VALID(one)) {
361 mf->ptr = (char *)""; /* does not matter */
362 mf->size = 0;
363 return 0;
365 else if (diff_populate_filespec(one, 0))
366 return -1;
368 mf->ptr = one->data;
369 mf->size = one->size;
370 return 0;
373 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
374 static unsigned long diff_filespec_size(struct diff_filespec *one)
376 if (!DIFF_FILE_VALID(one))
377 return 0;
378 diff_populate_filespec(one, CHECK_SIZE_ONLY);
379 return one->size;
382 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
384 char *ptr = mf->ptr;
385 long size = mf->size;
386 int cnt = 0;
388 if (!size)
389 return cnt;
390 ptr += size - 1; /* pointing at the very end */
391 if (*ptr != '\n')
392 ; /* incomplete line */
393 else
394 ptr--; /* skip the last LF */
395 while (mf->ptr < ptr) {
396 char *prev_eol;
397 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
398 if (*prev_eol == '\n')
399 break;
400 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
401 break;
402 cnt++;
403 ptr = prev_eol - 1;
405 return cnt;
408 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
409 struct emit_callback *ecbdata)
411 int l1, l2, at;
412 unsigned ws_rule = ecbdata->ws_rule;
413 l1 = count_trailing_blank(mf1, ws_rule);
414 l2 = count_trailing_blank(mf2, ws_rule);
415 if (l2 <= l1) {
416 ecbdata->blank_at_eof_in_preimage = 0;
417 ecbdata->blank_at_eof_in_postimage = 0;
418 return;
420 at = count_lines(mf1->ptr, mf1->size);
421 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
423 at = count_lines(mf2->ptr, mf2->size);
424 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
427 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
428 int first, const char *line, int len)
430 int has_trailing_newline, has_trailing_carriage_return;
431 int nofirst;
432 FILE *file = o->file;
434 fputs(diff_line_prefix(o), file);
436 if (len == 0) {
437 has_trailing_newline = (first == '\n');
438 has_trailing_carriage_return = (!has_trailing_newline &&
439 (first == '\r'));
440 nofirst = has_trailing_newline || has_trailing_carriage_return;
441 } else {
442 has_trailing_newline = (len > 0 && line[len-1] == '\n');
443 if (has_trailing_newline)
444 len--;
445 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
446 if (has_trailing_carriage_return)
447 len--;
448 nofirst = 0;
451 if (len || !nofirst) {
452 fputs(set, file);
453 if (!nofirst)
454 fputc(first, file);
455 fwrite(line, len, 1, file);
456 fputs(reset, file);
458 if (has_trailing_carriage_return)
459 fputc('\r', file);
460 if (has_trailing_newline)
461 fputc('\n', file);
464 static void emit_line(struct diff_options *o, const char *set, const char *reset,
465 const char *line, int len)
467 emit_line_0(o, set, reset, line[0], line+1, len-1);
470 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
472 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
473 ecbdata->blank_at_eof_in_preimage &&
474 ecbdata->blank_at_eof_in_postimage &&
475 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
476 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
477 return 0;
478 return ws_blank_line(line, len, ecbdata->ws_rule);
481 static void emit_add_line(const char *reset,
482 struct emit_callback *ecbdata,
483 const char *line, int len)
485 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
486 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
488 if (!*ws)
489 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
490 else if (new_blank_line_at_eof(ecbdata, line, len))
491 /* Blank line at EOF - paint '+' as well */
492 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
493 else {
494 /* Emit just the prefix, then the rest. */
495 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
496 ws_check_emit(line, len, ecbdata->ws_rule,
497 ecbdata->opt->file, set, reset, ws);
501 static void emit_hunk_header(struct emit_callback *ecbdata,
502 const char *line, int len)
504 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
505 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
506 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
507 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
508 static const char atat[2] = { '@', '@' };
509 const char *cp, *ep;
510 struct strbuf msgbuf = STRBUF_INIT;
511 int org_len = len;
512 int i = 1;
515 * As a hunk header must begin with "@@ -<old>, +<new> @@",
516 * it always is at least 10 bytes long.
518 if (len < 10 ||
519 memcmp(line, atat, 2) ||
520 !(ep = memmem(line + 2, len - 2, atat, 2))) {
521 emit_line(ecbdata->opt, plain, reset, line, len);
522 return;
524 ep += 2; /* skip over @@ */
526 /* The hunk header in fraginfo color */
527 strbuf_addstr(&msgbuf, frag);
528 strbuf_add(&msgbuf, line, ep - line);
529 strbuf_addstr(&msgbuf, reset);
532 * trailing "\r\n"
534 for ( ; i < 3; i++)
535 if (line[len - i] == '\r' || line[len - i] == '\n')
536 len--;
538 /* blank before the func header */
539 for (cp = ep; ep - line < len; ep++)
540 if (*ep != ' ' && *ep != '\t')
541 break;
542 if (ep != cp) {
543 strbuf_addstr(&msgbuf, plain);
544 strbuf_add(&msgbuf, cp, ep - cp);
545 strbuf_addstr(&msgbuf, reset);
548 if (ep < line + len) {
549 strbuf_addstr(&msgbuf, func);
550 strbuf_add(&msgbuf, ep, line + len - ep);
551 strbuf_addstr(&msgbuf, reset);
554 strbuf_add(&msgbuf, line + len, org_len - len);
555 emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
556 strbuf_release(&msgbuf);
559 static struct diff_tempfile *claim_diff_tempfile(void) {
560 int i;
561 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
562 if (!diff_temp[i].name)
563 return diff_temp + i;
564 die("BUG: diff is failing to clean up its tempfiles");
567 static int remove_tempfile_installed;
569 static void remove_tempfile(void)
571 int i;
572 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
573 if (diff_temp[i].name == diff_temp[i].tmp_path)
574 unlink_or_warn(diff_temp[i].name);
575 diff_temp[i].name = NULL;
579 static void remove_tempfile_on_signal(int signo)
581 remove_tempfile();
582 sigchain_pop(signo);
583 raise(signo);
586 static void print_line_count(FILE *file, int count)
588 switch (count) {
589 case 0:
590 fprintf(file, "0,0");
591 break;
592 case 1:
593 fprintf(file, "1");
594 break;
595 default:
596 fprintf(file, "1,%d", count);
597 break;
601 static void emit_rewrite_lines(struct emit_callback *ecb,
602 int prefix, const char *data, int size)
604 const char *endp = NULL;
605 static const char *nneof = " No newline at end of file\n";
606 const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
607 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
609 while (0 < size) {
610 int len;
612 endp = memchr(data, '\n', size);
613 len = endp ? (endp - data + 1) : size;
614 if (prefix != '+') {
615 ecb->lno_in_preimage++;
616 emit_line_0(ecb->opt, old, reset, '-',
617 data, len);
618 } else {
619 ecb->lno_in_postimage++;
620 emit_add_line(reset, ecb, data, len);
622 size -= len;
623 data += len;
625 if (!endp) {
626 const char *plain = diff_get_color(ecb->color_diff,
627 DIFF_PLAIN);
628 putc('\n', ecb->opt->file);
629 emit_line_0(ecb->opt, plain, reset, '\\',
630 nneof, strlen(nneof));
634 static void emit_rewrite_diff(const char *name_a,
635 const char *name_b,
636 struct diff_filespec *one,
637 struct diff_filespec *two,
638 struct userdiff_driver *textconv_one,
639 struct userdiff_driver *textconv_two,
640 struct diff_options *o)
642 int lc_a, lc_b;
643 const char *name_a_tab, *name_b_tab;
644 const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
645 const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
646 const char *reset = diff_get_color(o->use_color, DIFF_RESET);
647 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
648 const char *a_prefix, *b_prefix;
649 char *data_one, *data_two;
650 size_t size_one, size_two;
651 struct emit_callback ecbdata;
652 const char *line_prefix = diff_line_prefix(o);
654 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
655 a_prefix = o->b_prefix;
656 b_prefix = o->a_prefix;
657 } else {
658 a_prefix = o->a_prefix;
659 b_prefix = o->b_prefix;
662 name_a += (*name_a == '/');
663 name_b += (*name_b == '/');
664 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
665 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
667 strbuf_reset(&a_name);
668 strbuf_reset(&b_name);
669 quote_two_c_style(&a_name, a_prefix, name_a, 0);
670 quote_two_c_style(&b_name, b_prefix, name_b, 0);
672 size_one = fill_textconv(textconv_one, one, &data_one);
673 size_two = fill_textconv(textconv_two, two, &data_two);
675 memset(&ecbdata, 0, sizeof(ecbdata));
676 ecbdata.color_diff = want_color(o->use_color);
677 ecbdata.found_changesp = &o->found_changes;
678 ecbdata.ws_rule = whitespace_rule(name_b);
679 ecbdata.opt = o;
680 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
681 mmfile_t mf1, mf2;
682 mf1.ptr = (char *)data_one;
683 mf2.ptr = (char *)data_two;
684 mf1.size = size_one;
685 mf2.size = size_two;
686 check_blank_at_eof(&mf1, &mf2, &ecbdata);
688 ecbdata.lno_in_preimage = 1;
689 ecbdata.lno_in_postimage = 1;
691 lc_a = count_lines(data_one, size_one);
692 lc_b = count_lines(data_two, size_two);
693 fprintf(o->file,
694 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
695 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
696 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
697 line_prefix, fraginfo);
698 if (!o->irreversible_delete)
699 print_line_count(o->file, lc_a);
700 else
701 fprintf(o->file, "?,?");
702 fprintf(o->file, " +");
703 print_line_count(o->file, lc_b);
704 fprintf(o->file, " @@%s\n", reset);
705 if (lc_a && !o->irreversible_delete)
706 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
707 if (lc_b)
708 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
709 if (textconv_one)
710 free((char *)data_one);
711 if (textconv_two)
712 free((char *)data_two);
715 struct diff_words_buffer {
716 mmfile_t text;
717 long alloc;
718 struct diff_words_orig {
719 const char *begin, *end;
720 } *orig;
721 int orig_nr, orig_alloc;
724 static void diff_words_append(char *line, unsigned long len,
725 struct diff_words_buffer *buffer)
727 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
728 line++;
729 len--;
730 memcpy(buffer->text.ptr + buffer->text.size, line, len);
731 buffer->text.size += len;
732 buffer->text.ptr[buffer->text.size] = '\0';
735 struct diff_words_style_elem {
736 const char *prefix;
737 const char *suffix;
738 const char *color; /* NULL; filled in by the setup code if
739 * color is enabled */
742 struct diff_words_style {
743 enum diff_words_type type;
744 struct diff_words_style_elem new, old, ctx;
745 const char *newline;
748 static struct diff_words_style diff_words_styles[] = {
749 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
750 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
751 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
754 struct diff_words_data {
755 struct diff_words_buffer minus, plus;
756 const char *current_plus;
757 int last_minus;
758 struct diff_options *opt;
759 regex_t *word_regex;
760 enum diff_words_type type;
761 struct diff_words_style *style;
764 static int fn_out_diff_words_write_helper(FILE *fp,
765 struct diff_words_style_elem *st_el,
766 const char *newline,
767 size_t count, const char *buf,
768 const char *line_prefix)
770 int print = 0;
772 while (count) {
773 char *p = memchr(buf, '\n', count);
774 if (print)
775 fputs(line_prefix, fp);
776 if (p != buf) {
777 if (st_el->color && fputs(st_el->color, fp) < 0)
778 return -1;
779 if (fputs(st_el->prefix, fp) < 0 ||
780 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
781 fputs(st_el->suffix, fp) < 0)
782 return -1;
783 if (st_el->color && *st_el->color
784 && fputs(GIT_COLOR_RESET, fp) < 0)
785 return -1;
787 if (!p)
788 return 0;
789 if (fputs(newline, fp) < 0)
790 return -1;
791 count -= p + 1 - buf;
792 buf = p + 1;
793 print = 1;
795 return 0;
799 * '--color-words' algorithm can be described as:
801 * 1. collect a the minus/plus lines of a diff hunk, divided into
802 * minus-lines and plus-lines;
804 * 2. break both minus-lines and plus-lines into words and
805 * place them into two mmfile_t with one word for each line;
807 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
809 * And for the common parts of the both file, we output the plus side text.
810 * diff_words->current_plus is used to trace the current position of the plus file
811 * which printed. diff_words->last_minus is used to trace the last minus word
812 * printed.
814 * For '--graph' to work with '--color-words', we need to output the graph prefix
815 * on each line of color words output. Generally, there are two conditions on
816 * which we should output the prefix.
818 * 1. diff_words->last_minus == 0 &&
819 * diff_words->current_plus == diff_words->plus.text.ptr
821 * that is: the plus text must start as a new line, and if there is no minus
822 * word printed, a graph prefix must be printed.
824 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
825 * *(diff_words->current_plus - 1) == '\n'
827 * that is: a graph prefix must be printed following a '\n'
829 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
831 if ((diff_words->last_minus == 0 &&
832 diff_words->current_plus == diff_words->plus.text.ptr) ||
833 (diff_words->current_plus > diff_words->plus.text.ptr &&
834 *(diff_words->current_plus - 1) == '\n')) {
835 return 1;
836 } else {
837 return 0;
841 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
843 struct diff_words_data *diff_words = priv;
844 struct diff_words_style *style = diff_words->style;
845 int minus_first, minus_len, plus_first, plus_len;
846 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
847 struct diff_options *opt = diff_words->opt;
848 const char *line_prefix;
850 if (line[0] != '@' || parse_hunk_header(line, len,
851 &minus_first, &minus_len, &plus_first, &plus_len))
852 return;
854 assert(opt);
855 line_prefix = diff_line_prefix(opt);
857 /* POSIX requires that first be decremented by one if len == 0... */
858 if (minus_len) {
859 minus_begin = diff_words->minus.orig[minus_first].begin;
860 minus_end =
861 diff_words->minus.orig[minus_first + minus_len - 1].end;
862 } else
863 minus_begin = minus_end =
864 diff_words->minus.orig[minus_first].end;
866 if (plus_len) {
867 plus_begin = diff_words->plus.orig[plus_first].begin;
868 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
869 } else
870 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
872 if (color_words_output_graph_prefix(diff_words)) {
873 fputs(line_prefix, diff_words->opt->file);
875 if (diff_words->current_plus != plus_begin) {
876 fn_out_diff_words_write_helper(diff_words->opt->file,
877 &style->ctx, style->newline,
878 plus_begin - diff_words->current_plus,
879 diff_words->current_plus, line_prefix);
880 if (*(plus_begin - 1) == '\n')
881 fputs(line_prefix, diff_words->opt->file);
883 if (minus_begin != minus_end) {
884 fn_out_diff_words_write_helper(diff_words->opt->file,
885 &style->old, style->newline,
886 minus_end - minus_begin, minus_begin,
887 line_prefix);
889 if (plus_begin != plus_end) {
890 fn_out_diff_words_write_helper(diff_words->opt->file,
891 &style->new, style->newline,
892 plus_end - plus_begin, plus_begin,
893 line_prefix);
896 diff_words->current_plus = plus_end;
897 diff_words->last_minus = minus_first;
900 /* This function starts looking at *begin, and returns 0 iff a word was found. */
901 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
902 int *begin, int *end)
904 if (word_regex && *begin < buffer->size) {
905 regmatch_t match[1];
906 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
907 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
908 '\n', match[0].rm_eo - match[0].rm_so);
909 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
910 *begin += match[0].rm_so;
911 return *begin >= *end;
913 return -1;
916 /* find the next word */
917 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
918 (*begin)++;
919 if (*begin >= buffer->size)
920 return -1;
922 /* find the end of the word */
923 *end = *begin + 1;
924 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
925 (*end)++;
927 return 0;
931 * This function splits the words in buffer->text, stores the list with
932 * newline separator into out, and saves the offsets of the original words
933 * in buffer->orig.
935 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
936 regex_t *word_regex)
938 int i, j;
939 long alloc = 0;
941 out->size = 0;
942 out->ptr = NULL;
944 /* fake an empty "0th" word */
945 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
946 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
947 buffer->orig_nr = 1;
949 for (i = 0; i < buffer->text.size; i++) {
950 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
951 return;
953 /* store original boundaries */
954 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
955 buffer->orig_alloc);
956 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
957 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
958 buffer->orig_nr++;
960 /* store one word */
961 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
962 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
963 out->ptr[out->size + j - i] = '\n';
964 out->size += j - i + 1;
966 i = j - 1;
970 /* this executes the word diff on the accumulated buffers */
971 static void diff_words_show(struct diff_words_data *diff_words)
973 xpparam_t xpp;
974 xdemitconf_t xecfg;
975 mmfile_t minus, plus;
976 struct diff_words_style *style = diff_words->style;
978 struct diff_options *opt = diff_words->opt;
979 const char *line_prefix;
981 assert(opt);
982 line_prefix = diff_line_prefix(opt);
984 /* special case: only removal */
985 if (!diff_words->plus.text.size) {
986 fputs(line_prefix, diff_words->opt->file);
987 fn_out_diff_words_write_helper(diff_words->opt->file,
988 &style->old, style->newline,
989 diff_words->minus.text.size,
990 diff_words->minus.text.ptr, line_prefix);
991 diff_words->minus.text.size = 0;
992 return;
995 diff_words->current_plus = diff_words->plus.text.ptr;
996 diff_words->last_minus = 0;
998 memset(&xpp, 0, sizeof(xpp));
999 memset(&xecfg, 0, sizeof(xecfg));
1000 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1001 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1002 xpp.flags = 0;
1003 /* as only the hunk header will be parsed, we need a 0-context */
1004 xecfg.ctxlen = 0;
1005 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1006 &xpp, &xecfg);
1007 free(minus.ptr);
1008 free(plus.ptr);
1009 if (diff_words->current_plus != diff_words->plus.text.ptr +
1010 diff_words->plus.text.size) {
1011 if (color_words_output_graph_prefix(diff_words))
1012 fputs(line_prefix, diff_words->opt->file);
1013 fn_out_diff_words_write_helper(diff_words->opt->file,
1014 &style->ctx, style->newline,
1015 diff_words->plus.text.ptr + diff_words->plus.text.size
1016 - diff_words->current_plus, diff_words->current_plus,
1017 line_prefix);
1019 diff_words->minus.text.size = diff_words->plus.text.size = 0;
1022 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1023 static void diff_words_flush(struct emit_callback *ecbdata)
1025 if (ecbdata->diff_words->minus.text.size ||
1026 ecbdata->diff_words->plus.text.size)
1027 diff_words_show(ecbdata->diff_words);
1030 static void diff_filespec_load_driver(struct diff_filespec *one)
1032 /* Use already-loaded driver */
1033 if (one->driver)
1034 return;
1036 if (S_ISREG(one->mode))
1037 one->driver = userdiff_find_by_path(one->path);
1039 /* Fallback to default settings */
1040 if (!one->driver)
1041 one->driver = userdiff_find_by_name("default");
1044 static const char *userdiff_word_regex(struct diff_filespec *one)
1046 diff_filespec_load_driver(one);
1047 return one->driver->word_regex;
1050 static void init_diff_words_data(struct emit_callback *ecbdata,
1051 struct diff_options *orig_opts,
1052 struct diff_filespec *one,
1053 struct diff_filespec *two)
1055 int i;
1056 struct diff_options *o = xmalloc(sizeof(struct diff_options));
1057 memcpy(o, orig_opts, sizeof(struct diff_options));
1059 ecbdata->diff_words =
1060 xcalloc(1, sizeof(struct diff_words_data));
1061 ecbdata->diff_words->type = o->word_diff;
1062 ecbdata->diff_words->opt = o;
1063 if (!o->word_regex)
1064 o->word_regex = userdiff_word_regex(one);
1065 if (!o->word_regex)
1066 o->word_regex = userdiff_word_regex(two);
1067 if (!o->word_regex)
1068 o->word_regex = diff_word_regex_cfg;
1069 if (o->word_regex) {
1070 ecbdata->diff_words->word_regex = (regex_t *)
1071 xmalloc(sizeof(regex_t));
1072 if (regcomp(ecbdata->diff_words->word_regex,
1073 o->word_regex,
1074 REG_EXTENDED | REG_NEWLINE))
1075 die ("Invalid regular expression: %s",
1076 o->word_regex);
1078 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1079 if (o->word_diff == diff_words_styles[i].type) {
1080 ecbdata->diff_words->style =
1081 &diff_words_styles[i];
1082 break;
1085 if (want_color(o->use_color)) {
1086 struct diff_words_style *st = ecbdata->diff_words->style;
1087 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1088 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1089 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1093 static void free_diff_words_data(struct emit_callback *ecbdata)
1095 if (ecbdata->diff_words) {
1096 diff_words_flush(ecbdata);
1097 free (ecbdata->diff_words->opt);
1098 free (ecbdata->diff_words->minus.text.ptr);
1099 free (ecbdata->diff_words->minus.orig);
1100 free (ecbdata->diff_words->plus.text.ptr);
1101 free (ecbdata->diff_words->plus.orig);
1102 if (ecbdata->diff_words->word_regex) {
1103 regfree(ecbdata->diff_words->word_regex);
1104 free(ecbdata->diff_words->word_regex);
1106 free(ecbdata->diff_words);
1107 ecbdata->diff_words = NULL;
1111 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1113 if (want_color(diff_use_color))
1114 return diff_colors[ix];
1115 return "";
1118 const char *diff_line_prefix(struct diff_options *opt)
1120 struct strbuf *msgbuf;
1121 if (!opt->output_prefix)
1122 return "";
1124 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1125 return msgbuf->buf;
1128 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1130 const char *cp;
1131 unsigned long allot;
1132 size_t l = len;
1134 if (ecb->truncate)
1135 return ecb->truncate(line, len);
1136 cp = line;
1137 allot = l;
1138 while (0 < l) {
1139 (void) utf8_width(&cp, &l);
1140 if (!cp)
1141 break; /* truncated in the middle? */
1143 return allot - l;
1146 static void find_lno(const char *line, struct emit_callback *ecbdata)
1148 const char *p;
1149 ecbdata->lno_in_preimage = 0;
1150 ecbdata->lno_in_postimage = 0;
1151 p = strchr(line, '-');
1152 if (!p)
1153 return; /* cannot happen */
1154 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1155 p = strchr(p, '+');
1156 if (!p)
1157 return; /* cannot happen */
1158 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1161 static void fn_out_consume(void *priv, char *line, unsigned long len)
1163 struct emit_callback *ecbdata = priv;
1164 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1165 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1166 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1167 struct diff_options *o = ecbdata->opt;
1168 const char *line_prefix = diff_line_prefix(o);
1170 if (ecbdata->header) {
1171 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1172 strbuf_reset(ecbdata->header);
1173 ecbdata->header = NULL;
1175 *(ecbdata->found_changesp) = 1;
1177 if (ecbdata->label_path[0]) {
1178 const char *name_a_tab, *name_b_tab;
1180 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1181 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1183 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1184 line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1185 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1186 line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1187 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1190 if (diff_suppress_blank_empty
1191 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1192 line[0] = '\n';
1193 len = 1;
1196 if (line[0] == '@') {
1197 if (ecbdata->diff_words)
1198 diff_words_flush(ecbdata);
1199 len = sane_truncate_line(ecbdata, line, len);
1200 find_lno(line, ecbdata);
1201 emit_hunk_header(ecbdata, line, len);
1202 if (line[len-1] != '\n')
1203 putc('\n', ecbdata->opt->file);
1204 return;
1207 if (len < 1) {
1208 emit_line(ecbdata->opt, reset, reset, line, len);
1209 if (ecbdata->diff_words
1210 && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1211 fputs("~\n", ecbdata->opt->file);
1212 return;
1215 if (ecbdata->diff_words) {
1216 if (line[0] == '-') {
1217 diff_words_append(line, len,
1218 &ecbdata->diff_words->minus);
1219 return;
1220 } else if (line[0] == '+') {
1221 diff_words_append(line, len,
1222 &ecbdata->diff_words->plus);
1223 return;
1224 } else if (starts_with(line, "\\ ")) {
1226 * Eat the "no newline at eof" marker as if we
1227 * saw a "+" or "-" line with nothing on it,
1228 * and return without diff_words_flush() to
1229 * defer processing. If this is the end of
1230 * preimage, more "+" lines may come after it.
1232 return;
1234 diff_words_flush(ecbdata);
1235 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1236 emit_line(ecbdata->opt, plain, reset, line, len);
1237 fputs("~\n", ecbdata->opt->file);
1238 } else {
1240 * Skip the prefix character, if any. With
1241 * diff_suppress_blank_empty, there may be
1242 * none.
1244 if (line[0] != '\n') {
1245 line++;
1246 len--;
1248 emit_line(ecbdata->opt, plain, reset, line, len);
1250 return;
1253 if (line[0] != '+') {
1254 const char *color =
1255 diff_get_color(ecbdata->color_diff,
1256 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1257 ecbdata->lno_in_preimage++;
1258 if (line[0] == ' ')
1259 ecbdata->lno_in_postimage++;
1260 emit_line(ecbdata->opt, color, reset, line, len);
1261 } else {
1262 ecbdata->lno_in_postimage++;
1263 emit_add_line(reset, ecbdata, line + 1, len - 1);
1267 static char *pprint_rename(const char *a, const char *b)
1269 const char *old = a;
1270 const char *new = b;
1271 struct strbuf name = STRBUF_INIT;
1272 int pfx_length, sfx_length;
1273 int pfx_adjust_for_slash;
1274 int len_a = strlen(a);
1275 int len_b = strlen(b);
1276 int a_midlen, b_midlen;
1277 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1278 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1280 if (qlen_a || qlen_b) {
1281 quote_c_style(a, &name, NULL, 0);
1282 strbuf_addstr(&name, " => ");
1283 quote_c_style(b, &name, NULL, 0);
1284 return strbuf_detach(&name, NULL);
1287 /* Find common prefix */
1288 pfx_length = 0;
1289 while (*old && *new && *old == *new) {
1290 if (*old == '/')
1291 pfx_length = old - a + 1;
1292 old++;
1293 new++;
1296 /* Find common suffix */
1297 old = a + len_a;
1298 new = b + len_b;
1299 sfx_length = 0;
1301 * If there is a common prefix, it must end in a slash. In
1302 * that case we let this loop run 1 into the prefix to see the
1303 * same slash.
1305 * If there is no common prefix, we cannot do this as it would
1306 * underrun the input strings.
1308 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1309 while (a + pfx_length - pfx_adjust_for_slash <= old &&
1310 b + pfx_length - pfx_adjust_for_slash <= new &&
1311 *old == *new) {
1312 if (*old == '/')
1313 sfx_length = len_a - (old - a);
1314 old--;
1315 new--;
1319 * pfx{mid-a => mid-b}sfx
1320 * {pfx-a => pfx-b}sfx
1321 * pfx{sfx-a => sfx-b}
1322 * name-a => name-b
1324 a_midlen = len_a - pfx_length - sfx_length;
1325 b_midlen = len_b - pfx_length - sfx_length;
1326 if (a_midlen < 0)
1327 a_midlen = 0;
1328 if (b_midlen < 0)
1329 b_midlen = 0;
1331 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1332 if (pfx_length + sfx_length) {
1333 strbuf_add(&name, a, pfx_length);
1334 strbuf_addch(&name, '{');
1336 strbuf_add(&name, a + pfx_length, a_midlen);
1337 strbuf_addstr(&name, " => ");
1338 strbuf_add(&name, b + pfx_length, b_midlen);
1339 if (pfx_length + sfx_length) {
1340 strbuf_addch(&name, '}');
1341 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1343 return strbuf_detach(&name, NULL);
1346 struct diffstat_t {
1347 int nr;
1348 int alloc;
1349 struct diffstat_file {
1350 char *from_name;
1351 char *name;
1352 char *print_name;
1353 unsigned is_unmerged:1;
1354 unsigned is_binary:1;
1355 unsigned is_renamed:1;
1356 unsigned is_interesting:1;
1357 uintmax_t added, deleted;
1358 } **files;
1361 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1362 const char *name_a,
1363 const char *name_b)
1365 struct diffstat_file *x;
1366 x = xcalloc(1, sizeof(*x));
1367 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1368 diffstat->files[diffstat->nr++] = x;
1369 if (name_b) {
1370 x->from_name = xstrdup(name_a);
1371 x->name = xstrdup(name_b);
1372 x->is_renamed = 1;
1374 else {
1375 x->from_name = NULL;
1376 x->name = xstrdup(name_a);
1378 return x;
1381 static void diffstat_consume(void *priv, char *line, unsigned long len)
1383 struct diffstat_t *diffstat = priv;
1384 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1386 if (line[0] == '+')
1387 x->added++;
1388 else if (line[0] == '-')
1389 x->deleted++;
1392 const char mime_boundary_leader[] = "------------";
1394 static int scale_linear(int it, int width, int max_change)
1396 if (!it)
1397 return 0;
1399 * make sure that at least one '-' or '+' is printed if
1400 * there is any change to this path. The easiest way is to
1401 * scale linearly as if the alloted width is one column shorter
1402 * than it is, and then add 1 to the result.
1404 return 1 + (it * (width - 1) / max_change);
1407 static void show_name(FILE *file,
1408 const char *prefix, const char *name, int len)
1410 fprintf(file, " %s%-*s |", prefix, len, name);
1413 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1415 if (cnt <= 0)
1416 return;
1417 fprintf(file, "%s", set);
1418 while (cnt--)
1419 putc(ch, file);
1420 fprintf(file, "%s", reset);
1423 static void fill_print_name(struct diffstat_file *file)
1425 char *pname;
1427 if (file->print_name)
1428 return;
1430 if (!file->is_renamed) {
1431 struct strbuf buf = STRBUF_INIT;
1432 if (quote_c_style(file->name, &buf, NULL, 0)) {
1433 pname = strbuf_detach(&buf, NULL);
1434 } else {
1435 pname = file->name;
1436 strbuf_release(&buf);
1438 } else {
1439 pname = pprint_rename(file->from_name, file->name);
1441 file->print_name = pname;
1444 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1446 struct strbuf sb = STRBUF_INIT;
1447 int ret;
1449 if (!files) {
1450 assert(insertions == 0 && deletions == 0);
1451 return fprintf(fp, "%s\n", " 0 files changed");
1454 strbuf_addf(&sb,
1455 (files == 1) ? " %d file changed" : " %d files changed",
1456 files);
1459 * For binary diff, the caller may want to print "x files
1460 * changed" with insertions == 0 && deletions == 0.
1462 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1463 * is probably less confusing (i.e skip over "2 files changed
1464 * but nothing about added/removed lines? Is this a bug in Git?").
1466 if (insertions || deletions == 0) {
1467 strbuf_addf(&sb,
1468 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1469 insertions);
1472 if (deletions || insertions == 0) {
1473 strbuf_addf(&sb,
1474 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1475 deletions);
1477 strbuf_addch(&sb, '\n');
1478 ret = fputs(sb.buf, fp);
1479 strbuf_release(&sb);
1480 return ret;
1483 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1485 int i, len, add, del, adds = 0, dels = 0;
1486 uintmax_t max_change = 0, max_len = 0;
1487 int total_files = data->nr, count;
1488 int width, name_width, graph_width, number_width = 0, bin_width = 0;
1489 const char *reset, *add_c, *del_c;
1490 const char *line_prefix = "";
1491 int extra_shown = 0;
1493 if (data->nr == 0)
1494 return;
1496 line_prefix = diff_line_prefix(options);
1497 count = options->stat_count ? options->stat_count : data->nr;
1499 reset = diff_get_color_opt(options, DIFF_RESET);
1500 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1501 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1504 * Find the longest filename and max number of changes
1506 for (i = 0; (i < count) && (i < data->nr); i++) {
1507 struct diffstat_file *file = data->files[i];
1508 uintmax_t change = file->added + file->deleted;
1510 if (!file->is_interesting && (change == 0)) {
1511 count++; /* not shown == room for one more */
1512 continue;
1514 fill_print_name(file);
1515 len = strlen(file->print_name);
1516 if (max_len < len)
1517 max_len = len;
1519 if (file->is_unmerged) {
1520 /* "Unmerged" is 8 characters */
1521 bin_width = bin_width < 8 ? 8 : bin_width;
1522 continue;
1524 if (file->is_binary) {
1525 /* "Bin XXX -> YYY bytes" */
1526 int w = 14 + decimal_width(file->added)
1527 + decimal_width(file->deleted);
1528 bin_width = bin_width < w ? w : bin_width;
1529 /* Display change counts aligned with "Bin" */
1530 number_width = 3;
1531 continue;
1534 if (max_change < change)
1535 max_change = change;
1537 count = i; /* where we can stop scanning in data->files[] */
1540 * We have width = stat_width or term_columns() columns total.
1541 * We want a maximum of min(max_len, stat_name_width) for the name part.
1542 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1543 * We also need 1 for " " and 4 + decimal_width(max_change)
1544 * for " | NNNN " and one the empty column at the end, altogether
1545 * 6 + decimal_width(max_change).
1547 * If there's not enough space, we will use the smaller of
1548 * stat_name_width (if set) and 5/8*width for the filename,
1549 * and the rest for constant elements + graph part, but no more
1550 * than stat_graph_width for the graph part.
1551 * (5/8 gives 50 for filename and 30 for the constant parts + graph
1552 * for the standard terminal size).
1554 * In other words: stat_width limits the maximum width, and
1555 * stat_name_width fixes the maximum width of the filename,
1556 * and is also used to divide available columns if there
1557 * aren't enough.
1559 * Binary files are displayed with "Bin XXX -> YYY bytes"
1560 * instead of the change count and graph. This part is treated
1561 * similarly to the graph part, except that it is not
1562 * "scaled". If total width is too small to accommodate the
1563 * guaranteed minimum width of the filename part and the
1564 * separators and this message, this message will "overflow"
1565 * making the line longer than the maximum width.
1568 if (options->stat_width == -1)
1569 width = term_columns() - options->output_prefix_length;
1570 else
1571 width = options->stat_width ? options->stat_width : 80;
1572 number_width = decimal_width(max_change) > number_width ?
1573 decimal_width(max_change) : number_width;
1575 if (options->stat_graph_width == -1)
1576 options->stat_graph_width = diff_stat_graph_width;
1579 * Guarantee 3/8*16==6 for the graph part
1580 * and 5/8*16==10 for the filename part
1582 if (width < 16 + 6 + number_width)
1583 width = 16 + 6 + number_width;
1586 * First assign sizes that are wanted, ignoring available width.
1587 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1588 * starting from "XXX" should fit in graph_width.
1590 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1591 if (options->stat_graph_width &&
1592 options->stat_graph_width < graph_width)
1593 graph_width = options->stat_graph_width;
1595 name_width = (options->stat_name_width > 0 &&
1596 options->stat_name_width < max_len) ?
1597 options->stat_name_width : max_len;
1600 * Adjust adjustable widths not to exceed maximum width
1602 if (name_width + number_width + 6 + graph_width > width) {
1603 if (graph_width > width * 3/8 - number_width - 6) {
1604 graph_width = width * 3/8 - number_width - 6;
1605 if (graph_width < 6)
1606 graph_width = 6;
1609 if (options->stat_graph_width &&
1610 graph_width > options->stat_graph_width)
1611 graph_width = options->stat_graph_width;
1612 if (name_width > width - number_width - 6 - graph_width)
1613 name_width = width - number_width - 6 - graph_width;
1614 else
1615 graph_width = width - number_width - 6 - name_width;
1619 * From here name_width is the width of the name area,
1620 * and graph_width is the width of the graph area.
1621 * max_change is used to scale graph properly.
1623 for (i = 0; i < count; i++) {
1624 const char *prefix = "";
1625 struct diffstat_file *file = data->files[i];
1626 char *name = file->print_name;
1627 uintmax_t added = file->added;
1628 uintmax_t deleted = file->deleted;
1629 int name_len;
1631 if (!file->is_interesting && (added + deleted == 0))
1632 continue;
1635 * "scale" the filename
1637 len = name_width;
1638 name_len = strlen(name);
1639 if (name_width < name_len) {
1640 char *slash;
1641 prefix = "...";
1642 len -= 3;
1643 name += name_len - len;
1644 slash = strchr(name, '/');
1645 if (slash)
1646 name = slash;
1649 if (file->is_binary) {
1650 fprintf(options->file, "%s", line_prefix);
1651 show_name(options->file, prefix, name, len);
1652 fprintf(options->file, " %*s", number_width, "Bin");
1653 if (!added && !deleted) {
1654 putc('\n', options->file);
1655 continue;
1657 fprintf(options->file, " %s%"PRIuMAX"%s",
1658 del_c, deleted, reset);
1659 fprintf(options->file, " -> ");
1660 fprintf(options->file, "%s%"PRIuMAX"%s",
1661 add_c, added, reset);
1662 fprintf(options->file, " bytes");
1663 fprintf(options->file, "\n");
1664 continue;
1666 else if (file->is_unmerged) {
1667 fprintf(options->file, "%s", line_prefix);
1668 show_name(options->file, prefix, name, len);
1669 fprintf(options->file, " Unmerged\n");
1670 continue;
1674 * scale the add/delete
1676 add = added;
1677 del = deleted;
1679 if (graph_width <= max_change) {
1680 int total = scale_linear(add + del, graph_width, max_change);
1681 if (total < 2 && add && del)
1682 /* width >= 2 due to the sanity check */
1683 total = 2;
1684 if (add < del) {
1685 add = scale_linear(add, graph_width, max_change);
1686 del = total - add;
1687 } else {
1688 del = scale_linear(del, graph_width, max_change);
1689 add = total - del;
1692 fprintf(options->file, "%s", line_prefix);
1693 show_name(options->file, prefix, name, len);
1694 fprintf(options->file, " %*"PRIuMAX"%s",
1695 number_width, added + deleted,
1696 added + deleted ? " " : "");
1697 show_graph(options->file, '+', add, add_c, reset);
1698 show_graph(options->file, '-', del, del_c, reset);
1699 fprintf(options->file, "\n");
1702 for (i = 0; i < data->nr; i++) {
1703 struct diffstat_file *file = data->files[i];
1704 uintmax_t added = file->added;
1705 uintmax_t deleted = file->deleted;
1707 if (file->is_unmerged ||
1708 (!file->is_interesting && (added + deleted == 0))) {
1709 total_files--;
1710 continue;
1713 if (!file->is_binary) {
1714 adds += added;
1715 dels += deleted;
1717 if (i < count)
1718 continue;
1719 if (!extra_shown)
1720 fprintf(options->file, "%s ...\n", line_prefix);
1721 extra_shown = 1;
1723 fprintf(options->file, "%s", line_prefix);
1724 print_stat_summary(options->file, total_files, adds, dels);
1727 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1729 int i, adds = 0, dels = 0, total_files = data->nr;
1731 if (data->nr == 0)
1732 return;
1734 for (i = 0; i < data->nr; i++) {
1735 int added = data->files[i]->added;
1736 int deleted= data->files[i]->deleted;
1738 if (data->files[i]->is_unmerged ||
1739 (!data->files[i]->is_interesting && (added + deleted == 0))) {
1740 total_files--;
1741 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1742 adds += added;
1743 dels += deleted;
1746 fprintf(options->file, "%s", diff_line_prefix(options));
1747 print_stat_summary(options->file, total_files, adds, dels);
1750 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1752 int i;
1754 if (data->nr == 0)
1755 return;
1757 for (i = 0; i < data->nr; i++) {
1758 struct diffstat_file *file = data->files[i];
1760 fprintf(options->file, "%s", diff_line_prefix(options));
1762 if (file->is_binary)
1763 fprintf(options->file, "-\t-\t");
1764 else
1765 fprintf(options->file,
1766 "%"PRIuMAX"\t%"PRIuMAX"\t",
1767 file->added, file->deleted);
1768 if (options->line_termination) {
1769 fill_print_name(file);
1770 if (!file->is_renamed)
1771 write_name_quoted(file->name, options->file,
1772 options->line_termination);
1773 else {
1774 fputs(file->print_name, options->file);
1775 putc(options->line_termination, options->file);
1777 } else {
1778 if (file->is_renamed) {
1779 putc('\0', options->file);
1780 write_name_quoted(file->from_name, options->file, '\0');
1782 write_name_quoted(file->name, options->file, '\0');
1787 struct dirstat_file {
1788 const char *name;
1789 unsigned long changed;
1792 struct dirstat_dir {
1793 struct dirstat_file *files;
1794 int alloc, nr, permille, cumulative;
1797 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1798 unsigned long changed, const char *base, int baselen)
1800 unsigned long this_dir = 0;
1801 unsigned int sources = 0;
1802 const char *line_prefix = diff_line_prefix(opt);
1804 while (dir->nr) {
1805 struct dirstat_file *f = dir->files;
1806 int namelen = strlen(f->name);
1807 unsigned long this;
1808 char *slash;
1810 if (namelen < baselen)
1811 break;
1812 if (memcmp(f->name, base, baselen))
1813 break;
1814 slash = strchr(f->name + baselen, '/');
1815 if (slash) {
1816 int newbaselen = slash + 1 - f->name;
1817 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1818 sources++;
1819 } else {
1820 this = f->changed;
1821 dir->files++;
1822 dir->nr--;
1823 sources += 2;
1825 this_dir += this;
1829 * We don't report dirstat's for
1830 * - the top level
1831 * - or cases where everything came from a single directory
1832 * under this directory (sources == 1).
1834 if (baselen && sources != 1) {
1835 if (this_dir) {
1836 int permille = this_dir * 1000 / changed;
1837 if (permille >= dir->permille) {
1838 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1839 permille / 10, permille % 10, baselen, base);
1840 if (!dir->cumulative)
1841 return 0;
1845 return this_dir;
1848 static int dirstat_compare(const void *_a, const void *_b)
1850 const struct dirstat_file *a = _a;
1851 const struct dirstat_file *b = _b;
1852 return strcmp(a->name, b->name);
1855 static void show_dirstat(struct diff_options *options)
1857 int i;
1858 unsigned long changed;
1859 struct dirstat_dir dir;
1860 struct diff_queue_struct *q = &diff_queued_diff;
1862 dir.files = NULL;
1863 dir.alloc = 0;
1864 dir.nr = 0;
1865 dir.permille = options->dirstat_permille;
1866 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1868 changed = 0;
1869 for (i = 0; i < q->nr; i++) {
1870 struct diff_filepair *p = q->queue[i];
1871 const char *name;
1872 unsigned long copied, added, damage;
1873 int content_changed;
1875 name = p->two->path ? p->two->path : p->one->path;
1877 if (p->one->sha1_valid && p->two->sha1_valid)
1878 content_changed = hashcmp(p->one->sha1, p->two->sha1);
1879 else
1880 content_changed = 1;
1882 if (!content_changed) {
1884 * The SHA1 has not changed, so pre-/post-content is
1885 * identical. We can therefore skip looking at the
1886 * file contents altogether.
1888 damage = 0;
1889 goto found_damage;
1892 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1894 * In --dirstat-by-file mode, we don't really need to
1895 * look at the actual file contents at all.
1896 * The fact that the SHA1 changed is enough for us to
1897 * add this file to the list of results
1898 * (with each file contributing equal damage).
1900 damage = 1;
1901 goto found_damage;
1904 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1905 diff_populate_filespec(p->one, 0);
1906 diff_populate_filespec(p->two, 0);
1907 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1908 &copied, &added);
1909 diff_free_filespec_data(p->one);
1910 diff_free_filespec_data(p->two);
1911 } else if (DIFF_FILE_VALID(p->one)) {
1912 diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1913 copied = added = 0;
1914 diff_free_filespec_data(p->one);
1915 } else if (DIFF_FILE_VALID(p->two)) {
1916 diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1917 copied = 0;
1918 added = p->two->size;
1919 diff_free_filespec_data(p->two);
1920 } else
1921 continue;
1924 * Original minus copied is the removed material,
1925 * added is the new material. They are both damages
1926 * made to the preimage.
1927 * If the resulting damage is zero, we know that
1928 * diffcore_count_changes() considers the two entries to
1929 * be identical, but since content_changed is true, we
1930 * know that there must have been _some_ kind of change,
1931 * so we force all entries to have damage > 0.
1933 damage = (p->one->size - copied) + added;
1934 if (!damage)
1935 damage = 1;
1937 found_damage:
1938 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1939 dir.files[dir.nr].name = name;
1940 dir.files[dir.nr].changed = damage;
1941 changed += damage;
1942 dir.nr++;
1945 /* This can happen even with many files, if everything was renames */
1946 if (!changed)
1947 return;
1949 /* Show all directories with more than x% of the changes */
1950 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1951 gather_dirstat(options, &dir, changed, "", 0);
1954 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1956 int i;
1957 unsigned long changed;
1958 struct dirstat_dir dir;
1960 if (data->nr == 0)
1961 return;
1963 dir.files = NULL;
1964 dir.alloc = 0;
1965 dir.nr = 0;
1966 dir.permille = options->dirstat_permille;
1967 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1969 changed = 0;
1970 for (i = 0; i < data->nr; i++) {
1971 struct diffstat_file *file = data->files[i];
1972 unsigned long damage = file->added + file->deleted;
1973 if (file->is_binary)
1975 * binary files counts bytes, not lines. Must find some
1976 * way to normalize binary bytes vs. textual lines.
1977 * The following heuristic assumes that there are 64
1978 * bytes per "line".
1979 * This is stupid and ugly, but very cheap...
1981 damage = (damage + 63) / 64;
1982 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1983 dir.files[dir.nr].name = file->name;
1984 dir.files[dir.nr].changed = damage;
1985 changed += damage;
1986 dir.nr++;
1989 /* This can happen even with many files, if everything was renames */
1990 if (!changed)
1991 return;
1993 /* Show all directories with more than x% of the changes */
1994 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1995 gather_dirstat(options, &dir, changed, "", 0);
1998 static void free_diffstat_info(struct diffstat_t *diffstat)
2000 int i;
2001 for (i = 0; i < diffstat->nr; i++) {
2002 struct diffstat_file *f = diffstat->files[i];
2003 if (f->name != f->print_name)
2004 free(f->print_name);
2005 free(f->name);
2006 free(f->from_name);
2007 free(f);
2009 free(diffstat->files);
2012 struct checkdiff_t {
2013 const char *filename;
2014 int lineno;
2015 int conflict_marker_size;
2016 struct diff_options *o;
2017 unsigned ws_rule;
2018 unsigned status;
2021 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2023 char firstchar;
2024 int cnt;
2026 if (len < marker_size + 1)
2027 return 0;
2028 firstchar = line[0];
2029 switch (firstchar) {
2030 case '=': case '>': case '<': case '|':
2031 break;
2032 default:
2033 return 0;
2035 for (cnt = 1; cnt < marker_size; cnt++)
2036 if (line[cnt] != firstchar)
2037 return 0;
2038 /* line[1] thru line[marker_size-1] are same as firstchar */
2039 if (len < marker_size + 1 || !isspace(line[marker_size]))
2040 return 0;
2041 return 1;
2044 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2046 struct checkdiff_t *data = priv;
2047 int marker_size = data->conflict_marker_size;
2048 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2049 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2050 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2051 char *err;
2052 const char *line_prefix;
2054 assert(data->o);
2055 line_prefix = diff_line_prefix(data->o);
2057 if (line[0] == '+') {
2058 unsigned bad;
2059 data->lineno++;
2060 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2061 data->status |= 1;
2062 fprintf(data->o->file,
2063 "%s%s:%d: leftover conflict marker\n",
2064 line_prefix, data->filename, data->lineno);
2066 bad = ws_check(line + 1, len - 1, data->ws_rule);
2067 if (!bad)
2068 return;
2069 data->status |= bad;
2070 err = whitespace_error_string(bad);
2071 fprintf(data->o->file, "%s%s:%d: %s.\n",
2072 line_prefix, data->filename, data->lineno, err);
2073 free(err);
2074 emit_line(data->o, set, reset, line, 1);
2075 ws_check_emit(line + 1, len - 1, data->ws_rule,
2076 data->o->file, set, reset, ws);
2077 } else if (line[0] == ' ') {
2078 data->lineno++;
2079 } else if (line[0] == '@') {
2080 char *plus = strchr(line, '+');
2081 if (plus)
2082 data->lineno = strtol(plus, NULL, 10) - 1;
2083 else
2084 die("invalid diff");
2088 static unsigned char *deflate_it(char *data,
2089 unsigned long size,
2090 unsigned long *result_size)
2092 int bound;
2093 unsigned char *deflated;
2094 git_zstream stream;
2096 git_deflate_init(&stream, zlib_compression_level);
2097 bound = git_deflate_bound(&stream, size);
2098 deflated = xmalloc(bound);
2099 stream.next_out = deflated;
2100 stream.avail_out = bound;
2102 stream.next_in = (unsigned char *)data;
2103 stream.avail_in = size;
2104 while (git_deflate(&stream, Z_FINISH) == Z_OK)
2105 ; /* nothing */
2106 git_deflate_end(&stream);
2107 *result_size = stream.total_out;
2108 return deflated;
2111 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2112 const char *prefix)
2114 void *cp;
2115 void *delta;
2116 void *deflated;
2117 void *data;
2118 unsigned long orig_size;
2119 unsigned long delta_size;
2120 unsigned long deflate_size;
2121 unsigned long data_size;
2123 /* We could do deflated delta, or we could do just deflated two,
2124 * whichever is smaller.
2126 delta = NULL;
2127 deflated = deflate_it(two->ptr, two->size, &deflate_size);
2128 if (one->size && two->size) {
2129 delta = diff_delta(one->ptr, one->size,
2130 two->ptr, two->size,
2131 &delta_size, deflate_size);
2132 if (delta) {
2133 void *to_free = delta;
2134 orig_size = delta_size;
2135 delta = deflate_it(delta, delta_size, &delta_size);
2136 free(to_free);
2140 if (delta && delta_size < deflate_size) {
2141 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2142 free(deflated);
2143 data = delta;
2144 data_size = delta_size;
2146 else {
2147 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2148 free(delta);
2149 data = deflated;
2150 data_size = deflate_size;
2153 /* emit data encoded in base85 */
2154 cp = data;
2155 while (data_size) {
2156 int bytes = (52 < data_size) ? 52 : data_size;
2157 char line[70];
2158 data_size -= bytes;
2159 if (bytes <= 26)
2160 line[0] = bytes + 'A' - 1;
2161 else
2162 line[0] = bytes - 26 + 'a' - 1;
2163 encode_85(line + 1, cp, bytes);
2164 cp = (char *) cp + bytes;
2165 fprintf(file, "%s", prefix);
2166 fputs(line, file);
2167 fputc('\n', file);
2169 fprintf(file, "%s\n", prefix);
2170 free(data);
2173 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2174 const char *prefix)
2176 fprintf(file, "%sGIT binary patch\n", prefix);
2177 emit_binary_diff_body(file, one, two, prefix);
2178 emit_binary_diff_body(file, two, one, prefix);
2181 int diff_filespec_is_binary(struct diff_filespec *one)
2183 if (one->is_binary == -1) {
2184 diff_filespec_load_driver(one);
2185 if (one->driver->binary != -1)
2186 one->is_binary = one->driver->binary;
2187 else {
2188 if (!one->data && DIFF_FILE_VALID(one))
2189 diff_populate_filespec(one, CHECK_BINARY);
2190 if (one->is_binary == -1 && one->data)
2191 one->is_binary = buffer_is_binary(one->data,
2192 one->size);
2193 if (one->is_binary == -1)
2194 one->is_binary = 0;
2197 return one->is_binary;
2200 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2202 diff_filespec_load_driver(one);
2203 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2206 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2208 if (!options->a_prefix)
2209 options->a_prefix = a;
2210 if (!options->b_prefix)
2211 options->b_prefix = b;
2214 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2216 if (!DIFF_FILE_VALID(one))
2217 return NULL;
2219 diff_filespec_load_driver(one);
2220 return userdiff_get_textconv(one->driver);
2223 static void builtin_diff(const char *name_a,
2224 const char *name_b,
2225 struct diff_filespec *one,
2226 struct diff_filespec *two,
2227 const char *xfrm_msg,
2228 int must_show_header,
2229 struct diff_options *o,
2230 int complete_rewrite)
2232 mmfile_t mf1, mf2;
2233 const char *lbl[2];
2234 char *a_one, *b_two;
2235 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2236 const char *reset = diff_get_color_opt(o, DIFF_RESET);
2237 const char *a_prefix, *b_prefix;
2238 struct userdiff_driver *textconv_one = NULL;
2239 struct userdiff_driver *textconv_two = NULL;
2240 struct strbuf header = STRBUF_INIT;
2241 const char *line_prefix = diff_line_prefix(o);
2243 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2244 (!one->mode || S_ISGITLINK(one->mode)) &&
2245 (!two->mode || S_ISGITLINK(two->mode))) {
2246 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2247 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2248 show_submodule_summary(o->file, one->path ? one->path : two->path,
2249 line_prefix,
2250 one->sha1, two->sha1, two->dirty_submodule,
2251 meta, del, add, reset);
2252 return;
2255 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2256 textconv_one = get_textconv(one);
2257 textconv_two = get_textconv(two);
2260 diff_set_mnemonic_prefix(o, "a/", "b/");
2261 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2262 a_prefix = o->b_prefix;
2263 b_prefix = o->a_prefix;
2264 } else {
2265 a_prefix = o->a_prefix;
2266 b_prefix = o->b_prefix;
2269 /* Never use a non-valid filename anywhere if at all possible */
2270 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2271 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2273 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2274 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2275 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2276 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2277 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2278 if (lbl[0][0] == '/') {
2279 /* /dev/null */
2280 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2281 if (xfrm_msg)
2282 strbuf_addstr(&header, xfrm_msg);
2283 must_show_header = 1;
2285 else if (lbl[1][0] == '/') {
2286 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2287 if (xfrm_msg)
2288 strbuf_addstr(&header, xfrm_msg);
2289 must_show_header = 1;
2291 else {
2292 if (one->mode != two->mode) {
2293 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2294 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2295 must_show_header = 1;
2297 if (xfrm_msg)
2298 strbuf_addstr(&header, xfrm_msg);
2301 * we do not run diff between different kind
2302 * of objects.
2304 if ((one->mode ^ two->mode) & S_IFMT)
2305 goto free_ab_and_return;
2306 if (complete_rewrite &&
2307 (textconv_one || !diff_filespec_is_binary(one)) &&
2308 (textconv_two || !diff_filespec_is_binary(two))) {
2309 fprintf(o->file, "%s", header.buf);
2310 strbuf_reset(&header);
2311 emit_rewrite_diff(name_a, name_b, one, two,
2312 textconv_one, textconv_two, o);
2313 o->found_changes = 1;
2314 goto free_ab_and_return;
2318 if (o->irreversible_delete && lbl[1][0] == '/') {
2319 fprintf(o->file, "%s", header.buf);
2320 strbuf_reset(&header);
2321 goto free_ab_and_return;
2322 } else if (!DIFF_OPT_TST(o, TEXT) &&
2323 ( (!textconv_one && diff_filespec_is_binary(one)) ||
2324 (!textconv_two && diff_filespec_is_binary(two)) )) {
2325 if (!one->data && !two->data &&
2326 S_ISREG(one->mode) && S_ISREG(two->mode) &&
2327 !DIFF_OPT_TST(o, BINARY)) {
2328 if (!hashcmp(one->sha1, two->sha1)) {
2329 if (must_show_header)
2330 fprintf(o->file, "%s", header.buf);
2331 goto free_ab_and_return;
2333 fprintf(o->file, "%s", header.buf);
2334 fprintf(o->file, "%sBinary files %s and %s differ\n",
2335 line_prefix, lbl[0], lbl[1]);
2336 goto free_ab_and_return;
2338 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2339 die("unable to read files to diff");
2340 /* Quite common confusing case */
2341 if (mf1.size == mf2.size &&
2342 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2343 if (must_show_header)
2344 fprintf(o->file, "%s", header.buf);
2345 goto free_ab_and_return;
2347 fprintf(o->file, "%s", header.buf);
2348 strbuf_reset(&header);
2349 if (DIFF_OPT_TST(o, BINARY))
2350 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2351 else
2352 fprintf(o->file, "%sBinary files %s and %s differ\n",
2353 line_prefix, lbl[0], lbl[1]);
2354 o->found_changes = 1;
2355 } else {
2356 /* Crazy xdl interfaces.. */
2357 const char *diffopts = getenv("GIT_DIFF_OPTS");
2358 const char *v;
2359 xpparam_t xpp;
2360 xdemitconf_t xecfg;
2361 struct emit_callback ecbdata;
2362 const struct userdiff_funcname *pe;
2364 if (must_show_header) {
2365 fprintf(o->file, "%s", header.buf);
2366 strbuf_reset(&header);
2369 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2370 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2372 pe = diff_funcname_pattern(one);
2373 if (!pe)
2374 pe = diff_funcname_pattern(two);
2376 memset(&xpp, 0, sizeof(xpp));
2377 memset(&xecfg, 0, sizeof(xecfg));
2378 memset(&ecbdata, 0, sizeof(ecbdata));
2379 ecbdata.label_path = lbl;
2380 ecbdata.color_diff = want_color(o->use_color);
2381 ecbdata.found_changesp = &o->found_changes;
2382 ecbdata.ws_rule = whitespace_rule(name_b);
2383 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2384 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2385 ecbdata.opt = o;
2386 ecbdata.header = header.len ? &header : NULL;
2387 xpp.flags = o->xdl_opts;
2388 xecfg.ctxlen = o->context;
2389 xecfg.interhunkctxlen = o->interhunkcontext;
2390 xecfg.flags = XDL_EMIT_FUNCNAMES;
2391 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2392 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2393 if (pe)
2394 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2395 if (!diffopts)
2397 else if (skip_prefix(diffopts, "--unified=", &v))
2398 xecfg.ctxlen = strtoul(v, NULL, 10);
2399 else if (skip_prefix(diffopts, "-u", &v))
2400 xecfg.ctxlen = strtoul(v, NULL, 10);
2401 if (o->word_diff)
2402 init_diff_words_data(&ecbdata, o, one, two);
2403 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2404 &xpp, &xecfg);
2405 if (o->word_diff)
2406 free_diff_words_data(&ecbdata);
2407 if (textconv_one)
2408 free(mf1.ptr);
2409 if (textconv_two)
2410 free(mf2.ptr);
2411 xdiff_clear_find_func(&xecfg);
2414 free_ab_and_return:
2415 strbuf_release(&header);
2416 diff_free_filespec_data(one);
2417 diff_free_filespec_data(two);
2418 free(a_one);
2419 free(b_two);
2420 return;
2423 static void builtin_diffstat(const char *name_a, const char *name_b,
2424 struct diff_filespec *one,
2425 struct diff_filespec *two,
2426 struct diffstat_t *diffstat,
2427 struct diff_options *o,
2428 struct diff_filepair *p)
2430 mmfile_t mf1, mf2;
2431 struct diffstat_file *data;
2432 int same_contents;
2433 int complete_rewrite = 0;
2435 if (!DIFF_PAIR_UNMERGED(p)) {
2436 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2437 complete_rewrite = 1;
2440 data = diffstat_add(diffstat, name_a, name_b);
2441 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2443 if (!one || !two) {
2444 data->is_unmerged = 1;
2445 return;
2448 same_contents = !hashcmp(one->sha1, two->sha1);
2450 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2451 data->is_binary = 1;
2452 if (same_contents) {
2453 data->added = 0;
2454 data->deleted = 0;
2455 } else {
2456 data->added = diff_filespec_size(two);
2457 data->deleted = diff_filespec_size(one);
2461 else if (complete_rewrite) {
2462 diff_populate_filespec(one, 0);
2463 diff_populate_filespec(two, 0);
2464 data->deleted = count_lines(one->data, one->size);
2465 data->added = count_lines(two->data, two->size);
2468 else if (!same_contents) {
2469 /* Crazy xdl interfaces.. */
2470 xpparam_t xpp;
2471 xdemitconf_t xecfg;
2473 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2474 die("unable to read files to diff");
2476 memset(&xpp, 0, sizeof(xpp));
2477 memset(&xecfg, 0, sizeof(xecfg));
2478 xpp.flags = o->xdl_opts;
2479 xecfg.ctxlen = o->context;
2480 xecfg.interhunkctxlen = o->interhunkcontext;
2481 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2482 &xpp, &xecfg);
2485 diff_free_filespec_data(one);
2486 diff_free_filespec_data(two);
2489 static void builtin_checkdiff(const char *name_a, const char *name_b,
2490 const char *attr_path,
2491 struct diff_filespec *one,
2492 struct diff_filespec *two,
2493 struct diff_options *o)
2495 mmfile_t mf1, mf2;
2496 struct checkdiff_t data;
2498 if (!two)
2499 return;
2501 memset(&data, 0, sizeof(data));
2502 data.filename = name_b ? name_b : name_a;
2503 data.lineno = 0;
2504 data.o = o;
2505 data.ws_rule = whitespace_rule(attr_path);
2506 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2508 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2509 die("unable to read files to diff");
2512 * All the other codepaths check both sides, but not checking
2513 * the "old" side here is deliberate. We are checking the newly
2514 * introduced changes, and as long as the "new" side is text, we
2515 * can and should check what it introduces.
2517 if (diff_filespec_is_binary(two))
2518 goto free_and_return;
2519 else {
2520 /* Crazy xdl interfaces.. */
2521 xpparam_t xpp;
2522 xdemitconf_t xecfg;
2524 memset(&xpp, 0, sizeof(xpp));
2525 memset(&xecfg, 0, sizeof(xecfg));
2526 xecfg.ctxlen = 1; /* at least one context line */
2527 xpp.flags = 0;
2528 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2529 &xpp, &xecfg);
2531 if (data.ws_rule & WS_BLANK_AT_EOF) {
2532 struct emit_callback ecbdata;
2533 int blank_at_eof;
2535 ecbdata.ws_rule = data.ws_rule;
2536 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2537 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2539 if (blank_at_eof) {
2540 static char *err;
2541 if (!err)
2542 err = whitespace_error_string(WS_BLANK_AT_EOF);
2543 fprintf(o->file, "%s:%d: %s.\n",
2544 data.filename, blank_at_eof, err);
2545 data.status = 1; /* report errors */
2549 free_and_return:
2550 diff_free_filespec_data(one);
2551 diff_free_filespec_data(two);
2552 if (data.status)
2553 DIFF_OPT_SET(o, CHECK_FAILED);
2556 struct diff_filespec *alloc_filespec(const char *path)
2558 int namelen = strlen(path);
2559 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2561 memset(spec, 0, sizeof(*spec));
2562 spec->path = (char *)(spec + 1);
2563 memcpy(spec->path, path, namelen+1);
2564 spec->count = 1;
2565 spec->is_binary = -1;
2566 return spec;
2569 void free_filespec(struct diff_filespec *spec)
2571 if (!--spec->count) {
2572 diff_free_filespec_data(spec);
2573 free(spec);
2577 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2578 int sha1_valid, unsigned short mode)
2580 if (mode) {
2581 spec->mode = canon_mode(mode);
2582 hashcpy(spec->sha1, sha1);
2583 spec->sha1_valid = sha1_valid;
2588 * Given a name and sha1 pair, if the index tells us the file in
2589 * the work tree has that object contents, return true, so that
2590 * prepare_temp_file() does not have to inflate and extract.
2592 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2594 const struct cache_entry *ce;
2595 struct stat st;
2596 int pos, len;
2599 * We do not read the cache ourselves here, because the
2600 * benchmark with my previous version that always reads cache
2601 * shows that it makes things worse for diff-tree comparing
2602 * two linux-2.6 kernel trees in an already checked out work
2603 * tree. This is because most diff-tree comparisons deal with
2604 * only a small number of files, while reading the cache is
2605 * expensive for a large project, and its cost outweighs the
2606 * savings we get by not inflating the object to a temporary
2607 * file. Practically, this code only helps when we are used
2608 * by diff-cache --cached, which does read the cache before
2609 * calling us.
2611 if (!active_cache)
2612 return 0;
2614 /* We want to avoid the working directory if our caller
2615 * doesn't need the data in a normal file, this system
2616 * is rather slow with its stat/open/mmap/close syscalls,
2617 * and the object is contained in a pack file. The pack
2618 * is probably already open and will be faster to obtain
2619 * the data through than the working directory. Loose
2620 * objects however would tend to be slower as they need
2621 * to be individually opened and inflated.
2623 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2624 return 0;
2626 len = strlen(name);
2627 pos = cache_name_pos(name, len);
2628 if (pos < 0)
2629 return 0;
2630 ce = active_cache[pos];
2633 * This is not the sha1 we are looking for, or
2634 * unreusable because it is not a regular file.
2636 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2637 return 0;
2640 * If ce is marked as "assume unchanged", there is no
2641 * guarantee that work tree matches what we are looking for.
2643 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2644 return 0;
2647 * If ce matches the file in the work tree, we can reuse it.
2649 if (ce_uptodate(ce) ||
2650 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2651 return 1;
2653 return 0;
2656 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2658 int len;
2659 char *data = xmalloc(100), *dirty = "";
2661 /* Are we looking at the work tree? */
2662 if (s->dirty_submodule)
2663 dirty = "-dirty";
2665 len = snprintf(data, 100,
2666 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2667 s->data = data;
2668 s->size = len;
2669 s->should_free = 1;
2670 if (size_only) {
2671 s->data = NULL;
2672 free(data);
2674 return 0;
2678 * While doing rename detection and pickaxe operation, we may need to
2679 * grab the data for the blob (or file) for our own in-core comparison.
2680 * diff_filespec has data and size fields for this purpose.
2682 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2684 int size_only = flags & CHECK_SIZE_ONLY;
2685 int err = 0;
2687 * demote FAIL to WARN to allow inspecting the situation
2688 * instead of refusing.
2690 enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2691 ? SAFE_CRLF_WARN
2692 : safe_crlf);
2694 if (!DIFF_FILE_VALID(s))
2695 die("internal error: asking to populate invalid file.");
2696 if (S_ISDIR(s->mode))
2697 return -1;
2699 if (s->data)
2700 return 0;
2702 if (size_only && 0 < s->size)
2703 return 0;
2705 if (S_ISGITLINK(s->mode))
2706 return diff_populate_gitlink(s, size_only);
2708 if (!s->sha1_valid ||
2709 reuse_worktree_file(s->path, s->sha1, 0)) {
2710 struct strbuf buf = STRBUF_INIT;
2711 struct stat st;
2712 int fd;
2714 if (lstat(s->path, &st) < 0) {
2715 if (errno == ENOENT) {
2716 err_empty:
2717 err = -1;
2718 empty:
2719 s->data = (char *)"";
2720 s->size = 0;
2721 return err;
2724 s->size = xsize_t(st.st_size);
2725 if (!s->size)
2726 goto empty;
2727 if (S_ISLNK(st.st_mode)) {
2728 struct strbuf sb = STRBUF_INIT;
2730 if (strbuf_readlink(&sb, s->path, s->size))
2731 goto err_empty;
2732 s->size = sb.len;
2733 s->data = strbuf_detach(&sb, NULL);
2734 s->should_free = 1;
2735 return 0;
2737 if (size_only)
2738 return 0;
2739 if ((flags & CHECK_BINARY) &&
2740 s->size > big_file_threshold && s->is_binary == -1) {
2741 s->is_binary = 1;
2742 return 0;
2744 fd = open(s->path, O_RDONLY);
2745 if (fd < 0)
2746 goto err_empty;
2747 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2748 close(fd);
2749 s->should_munmap = 1;
2752 * Convert from working tree format to canonical git format
2754 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2755 size_t size = 0;
2756 munmap(s->data, s->size);
2757 s->should_munmap = 0;
2758 s->data = strbuf_detach(&buf, &size);
2759 s->size = size;
2760 s->should_free = 1;
2763 else {
2764 enum object_type type;
2765 if (size_only || (flags & CHECK_BINARY)) {
2766 type = sha1_object_info(s->sha1, &s->size);
2767 if (type < 0)
2768 die("unable to read %s", sha1_to_hex(s->sha1));
2769 if (size_only)
2770 return 0;
2771 if (s->size > big_file_threshold && s->is_binary == -1) {
2772 s->is_binary = 1;
2773 return 0;
2776 s->data = read_sha1_file(s->sha1, &type, &s->size);
2777 if (!s->data)
2778 die("unable to read %s", sha1_to_hex(s->sha1));
2779 s->should_free = 1;
2781 return 0;
2784 void diff_free_filespec_blob(struct diff_filespec *s)
2786 if (s->should_free)
2787 free(s->data);
2788 else if (s->should_munmap)
2789 munmap(s->data, s->size);
2791 if (s->should_free || s->should_munmap) {
2792 s->should_free = s->should_munmap = 0;
2793 s->data = NULL;
2797 void diff_free_filespec_data(struct diff_filespec *s)
2799 diff_free_filespec_blob(s);
2800 free(s->cnt_data);
2801 s->cnt_data = NULL;
2804 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2805 void *blob,
2806 unsigned long size,
2807 const unsigned char *sha1,
2808 int mode)
2810 int fd;
2811 struct strbuf buf = STRBUF_INIT;
2812 struct strbuf template = STRBUF_INIT;
2813 char *path_dup = xstrdup(path);
2814 const char *base = basename(path_dup);
2816 /* Generate "XXXXXX_basename.ext" */
2817 strbuf_addstr(&template, "XXXXXX_");
2818 strbuf_addstr(&template, base);
2820 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2821 strlen(base) + 1);
2822 if (fd < 0)
2823 die_errno("unable to create temp-file");
2824 if (convert_to_working_tree(path,
2825 (const char *)blob, (size_t)size, &buf)) {
2826 blob = buf.buf;
2827 size = buf.len;
2829 if (write_in_full(fd, blob, size) != size)
2830 die_errno("unable to write temp-file");
2831 close(fd);
2832 temp->name = temp->tmp_path;
2833 strcpy(temp->hex, sha1_to_hex(sha1));
2834 temp->hex[40] = 0;
2835 sprintf(temp->mode, "%06o", mode);
2836 strbuf_release(&buf);
2837 strbuf_release(&template);
2838 free(path_dup);
2841 static struct diff_tempfile *prepare_temp_file(const char *name,
2842 struct diff_filespec *one)
2844 struct diff_tempfile *temp = claim_diff_tempfile();
2846 if (!DIFF_FILE_VALID(one)) {
2847 not_a_valid_file:
2848 /* A '-' entry produces this for file-2, and
2849 * a '+' entry produces this for file-1.
2851 temp->name = "/dev/null";
2852 strcpy(temp->hex, ".");
2853 strcpy(temp->mode, ".");
2854 return temp;
2857 if (!remove_tempfile_installed) {
2858 atexit(remove_tempfile);
2859 sigchain_push_common(remove_tempfile_on_signal);
2860 remove_tempfile_installed = 1;
2863 if (!S_ISGITLINK(one->mode) &&
2864 (!one->sha1_valid ||
2865 reuse_worktree_file(name, one->sha1, 1))) {
2866 struct stat st;
2867 if (lstat(name, &st) < 0) {
2868 if (errno == ENOENT)
2869 goto not_a_valid_file;
2870 die_errno("stat(%s)", name);
2872 if (S_ISLNK(st.st_mode)) {
2873 struct strbuf sb = STRBUF_INIT;
2874 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2875 die_errno("readlink(%s)", name);
2876 prep_temp_blob(name, temp, sb.buf, sb.len,
2877 (one->sha1_valid ?
2878 one->sha1 : null_sha1),
2879 (one->sha1_valid ?
2880 one->mode : S_IFLNK));
2881 strbuf_release(&sb);
2883 else {
2884 /* we can borrow from the file in the work tree */
2885 temp->name = name;
2886 if (!one->sha1_valid)
2887 strcpy(temp->hex, sha1_to_hex(null_sha1));
2888 else
2889 strcpy(temp->hex, sha1_to_hex(one->sha1));
2890 /* Even though we may sometimes borrow the
2891 * contents from the work tree, we always want
2892 * one->mode. mode is trustworthy even when
2893 * !(one->sha1_valid), as long as
2894 * DIFF_FILE_VALID(one).
2896 sprintf(temp->mode, "%06o", one->mode);
2898 return temp;
2900 else {
2901 if (diff_populate_filespec(one, 0))
2902 die("cannot read data blob for %s", one->path);
2903 prep_temp_blob(name, temp, one->data, one->size,
2904 one->sha1, one->mode);
2906 return temp;
2909 static void add_external_diff_name(struct argv_array *argv,
2910 const char *name,
2911 struct diff_filespec *df)
2913 struct diff_tempfile *temp = prepare_temp_file(name, df);
2914 argv_array_push(argv, temp->name);
2915 argv_array_push(argv, temp->hex);
2916 argv_array_push(argv, temp->mode);
2919 /* An external diff command takes:
2921 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2922 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2925 static void run_external_diff(const char *pgm,
2926 const char *name,
2927 const char *other,
2928 struct diff_filespec *one,
2929 struct diff_filespec *two,
2930 const char *xfrm_msg,
2931 int complete_rewrite,
2932 struct diff_options *o)
2934 struct argv_array argv = ARGV_ARRAY_INIT;
2935 struct argv_array env = ARGV_ARRAY_INIT;
2936 struct diff_queue_struct *q = &diff_queued_diff;
2938 argv_array_push(&argv, pgm);
2939 argv_array_push(&argv, name);
2941 if (one && two) {
2942 add_external_diff_name(&argv, name, one);
2943 if (!other)
2944 add_external_diff_name(&argv, name, two);
2945 else {
2946 add_external_diff_name(&argv, other, two);
2947 argv_array_push(&argv, other);
2948 argv_array_push(&argv, xfrm_msg);
2952 argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2953 argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2955 if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2956 die(_("external diff died, stopping at %s"), name);
2958 remove_tempfile();
2959 argv_array_clear(&argv);
2960 argv_array_clear(&env);
2963 static int similarity_index(struct diff_filepair *p)
2965 return p->score * 100 / MAX_SCORE;
2968 static void fill_metainfo(struct strbuf *msg,
2969 const char *name,
2970 const char *other,
2971 struct diff_filespec *one,
2972 struct diff_filespec *two,
2973 struct diff_options *o,
2974 struct diff_filepair *p,
2975 int *must_show_header,
2976 int use_color)
2978 const char *set = diff_get_color(use_color, DIFF_METAINFO);
2979 const char *reset = diff_get_color(use_color, DIFF_RESET);
2980 const char *line_prefix = diff_line_prefix(o);
2982 *must_show_header = 1;
2983 strbuf_init(msg, PATH_MAX * 2 + 300);
2984 switch (p->status) {
2985 case DIFF_STATUS_COPIED:
2986 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2987 line_prefix, set, similarity_index(p));
2988 strbuf_addf(msg, "%s\n%s%scopy from ",
2989 reset, line_prefix, set);
2990 quote_c_style(name, msg, NULL, 0);
2991 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2992 quote_c_style(other, msg, NULL, 0);
2993 strbuf_addf(msg, "%s\n", reset);
2994 break;
2995 case DIFF_STATUS_RENAMED:
2996 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2997 line_prefix, set, similarity_index(p));
2998 strbuf_addf(msg, "%s\n%s%srename from ",
2999 reset, line_prefix, set);
3000 quote_c_style(name, msg, NULL, 0);
3001 strbuf_addf(msg, "%s\n%s%srename to ",
3002 reset, line_prefix, set);
3003 quote_c_style(other, msg, NULL, 0);
3004 strbuf_addf(msg, "%s\n", reset);
3005 break;
3006 case DIFF_STATUS_MODIFIED:
3007 if (p->score) {
3008 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3009 line_prefix,
3010 set, similarity_index(p), reset);
3011 break;
3013 /* fallthru */
3014 default:
3015 *must_show_header = 0;
3017 if (one && two && hashcmp(one->sha1, two->sha1)) {
3018 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3020 if (DIFF_OPT_TST(o, BINARY)) {
3021 mmfile_t mf;
3022 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3023 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3024 abbrev = 40;
3026 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3027 find_unique_abbrev(one->sha1, abbrev));
3028 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3029 if (one->mode == two->mode)
3030 strbuf_addf(msg, " %06o", one->mode);
3031 strbuf_addf(msg, "%s\n", reset);
3035 static void run_diff_cmd(const char *pgm,
3036 const char *name,
3037 const char *other,
3038 const char *attr_path,
3039 struct diff_filespec *one,
3040 struct diff_filespec *two,
3041 struct strbuf *msg,
3042 struct diff_options *o,
3043 struct diff_filepair *p)
3045 const char *xfrm_msg = NULL;
3046 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3047 int must_show_header = 0;
3050 if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3051 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3052 if (drv && drv->external)
3053 pgm = drv->external;
3056 if (msg) {
3058 * don't use colors when the header is intended for an
3059 * external diff driver
3061 fill_metainfo(msg, name, other, one, two, o, p,
3062 &must_show_header,
3063 want_color(o->use_color) && !pgm);
3064 xfrm_msg = msg->len ? msg->buf : NULL;
3067 if (pgm) {
3068 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3069 complete_rewrite, o);
3070 return;
3072 if (one && two)
3073 builtin_diff(name, other ? other : name,
3074 one, two, xfrm_msg, must_show_header,
3075 o, complete_rewrite);
3076 else
3077 fprintf(o->file, "* Unmerged path %s\n", name);
3080 static void diff_fill_sha1_info(struct diff_filespec *one)
3082 if (DIFF_FILE_VALID(one)) {
3083 if (!one->sha1_valid) {
3084 struct stat st;
3085 if (one->is_stdin) {
3086 hashcpy(one->sha1, null_sha1);
3087 return;
3089 if (lstat(one->path, &st) < 0)
3090 die_errno("stat '%s'", one->path);
3091 if (index_path(one->sha1, one->path, &st, 0))
3092 die("cannot hash %s", one->path);
3095 else
3096 hashclr(one->sha1);
3099 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3101 /* Strip the prefix but do not molest /dev/null and absolute paths */
3102 if (*namep && **namep != '/') {
3103 *namep += prefix_length;
3104 if (**namep == '/')
3105 ++*namep;
3107 if (*otherp && **otherp != '/') {
3108 *otherp += prefix_length;
3109 if (**otherp == '/')
3110 ++*otherp;
3114 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3116 const char *pgm = external_diff();
3117 struct strbuf msg;
3118 struct diff_filespec *one = p->one;
3119 struct diff_filespec *two = p->two;
3120 const char *name;
3121 const char *other;
3122 const char *attr_path;
3124 name = p->one->path;
3125 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3126 attr_path = name;
3127 if (o->prefix_length)
3128 strip_prefix(o->prefix_length, &name, &other);
3130 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3131 pgm = NULL;
3133 if (DIFF_PAIR_UNMERGED(p)) {
3134 run_diff_cmd(pgm, name, NULL, attr_path,
3135 NULL, NULL, NULL, o, p);
3136 return;
3139 diff_fill_sha1_info(one);
3140 diff_fill_sha1_info(two);
3142 if (!pgm &&
3143 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3144 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3146 * a filepair that changes between file and symlink
3147 * needs to be split into deletion and creation.
3149 struct diff_filespec *null = alloc_filespec(two->path);
3150 run_diff_cmd(NULL, name, other, attr_path,
3151 one, null, &msg, o, p);
3152 free(null);
3153 strbuf_release(&msg);
3155 null = alloc_filespec(one->path);
3156 run_diff_cmd(NULL, name, other, attr_path,
3157 null, two, &msg, o, p);
3158 free(null);
3160 else
3161 run_diff_cmd(pgm, name, other, attr_path,
3162 one, two, &msg, o, p);
3164 strbuf_release(&msg);
3167 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3168 struct diffstat_t *diffstat)
3170 const char *name;
3171 const char *other;
3173 if (DIFF_PAIR_UNMERGED(p)) {
3174 /* unmerged */
3175 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3176 return;
3179 name = p->one->path;
3180 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3182 if (o->prefix_length)
3183 strip_prefix(o->prefix_length, &name, &other);
3185 diff_fill_sha1_info(p->one);
3186 diff_fill_sha1_info(p->two);
3188 builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3191 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3193 const char *name;
3194 const char *other;
3195 const char *attr_path;
3197 if (DIFF_PAIR_UNMERGED(p)) {
3198 /* unmerged */
3199 return;
3202 name = p->one->path;
3203 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3204 attr_path = other ? other : name;
3206 if (o->prefix_length)
3207 strip_prefix(o->prefix_length, &name, &other);
3209 diff_fill_sha1_info(p->one);
3210 diff_fill_sha1_info(p->two);
3212 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3215 void diff_setup(struct diff_options *options)
3217 memcpy(options, &default_diff_options, sizeof(*options));
3219 options->file = stdout;
3221 options->line_termination = '\n';
3222 options->break_opt = -1;
3223 options->rename_limit = -1;
3224 options->dirstat_permille = diff_dirstat_permille_default;
3225 options->context = diff_context_default;
3226 DIFF_OPT_SET(options, RENAME_EMPTY);
3228 /* pathchange left =NULL by default */
3229 options->change = diff_change;
3230 options->add_remove = diff_addremove;
3231 options->use_color = diff_use_color_default;
3232 options->detect_rename = diff_detect_rename_default;
3233 options->xdl_opts |= diff_algorithm;
3235 options->orderfile = diff_order_file_cfg;
3237 if (diff_no_prefix) {
3238 options->a_prefix = options->b_prefix = "";
3239 } else if (!diff_mnemonic_prefix) {
3240 options->a_prefix = "a/";
3241 options->b_prefix = "b/";
3245 void diff_setup_done(struct diff_options *options)
3247 int count = 0;
3249 if (options->set_default)
3250 options->set_default(options);
3252 if (options->output_format & DIFF_FORMAT_NAME)
3253 count++;
3254 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3255 count++;
3256 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3257 count++;
3258 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3259 count++;
3260 if (count > 1)
3261 die("--name-only, --name-status, --check and -s are mutually exclusive");
3264 * Most of the time we can say "there are changes"
3265 * only by checking if there are changed paths, but
3266 * --ignore-whitespace* options force us to look
3267 * inside contents.
3270 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3271 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3272 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3273 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3274 else
3275 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3277 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3278 options->detect_rename = DIFF_DETECT_COPY;
3280 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3281 options->prefix = NULL;
3282 if (options->prefix)
3283 options->prefix_length = strlen(options->prefix);
3284 else
3285 options->prefix_length = 0;
3287 if (options->output_format & (DIFF_FORMAT_NAME |
3288 DIFF_FORMAT_NAME_STATUS |
3289 DIFF_FORMAT_CHECKDIFF |
3290 DIFF_FORMAT_NO_OUTPUT))
3291 options->output_format &= ~(DIFF_FORMAT_RAW |
3292 DIFF_FORMAT_NUMSTAT |
3293 DIFF_FORMAT_DIFFSTAT |
3294 DIFF_FORMAT_SHORTSTAT |
3295 DIFF_FORMAT_DIRSTAT |
3296 DIFF_FORMAT_SUMMARY |
3297 DIFF_FORMAT_PATCH);
3300 * These cases always need recursive; we do not drop caller-supplied
3301 * recursive bits for other formats here.
3303 if (options->output_format & (DIFF_FORMAT_PATCH |
3304 DIFF_FORMAT_NUMSTAT |
3305 DIFF_FORMAT_DIFFSTAT |
3306 DIFF_FORMAT_SHORTSTAT |
3307 DIFF_FORMAT_DIRSTAT |
3308 DIFF_FORMAT_SUMMARY |
3309 DIFF_FORMAT_CHECKDIFF))
3310 DIFF_OPT_SET(options, RECURSIVE);
3312 * Also pickaxe would not work very well if you do not say recursive
3314 if (options->pickaxe)
3315 DIFF_OPT_SET(options, RECURSIVE);
3317 * When patches are generated, submodules diffed against the work tree
3318 * must be checked for dirtiness too so it can be shown in the output
3320 if (options->output_format & DIFF_FORMAT_PATCH)
3321 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3323 if (options->detect_rename && options->rename_limit < 0)
3324 options->rename_limit = diff_rename_limit_default;
3325 if (options->setup & DIFF_SETUP_USE_CACHE) {
3326 if (!active_cache)
3327 /* read-cache does not die even when it fails
3328 * so it is safe for us to do this here. Also
3329 * it does not smudge active_cache or active_nr
3330 * when it fails, so we do not have to worry about
3331 * cleaning it up ourselves either.
3333 read_cache();
3335 if (options->abbrev <= 0 || 40 < options->abbrev)
3336 options->abbrev = 40; /* full */
3339 * It does not make sense to show the first hit we happened
3340 * to have found. It does not make sense not to return with
3341 * exit code in such a case either.
3343 if (DIFF_OPT_TST(options, QUICK)) {
3344 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3345 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3348 options->diff_path_counter = 0;
3350 if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3351 die(_("--follow requires exactly one pathspec"));
3354 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3356 char c, *eq;
3357 int len;
3359 if (*arg != '-')
3360 return 0;
3361 c = *++arg;
3362 if (!c)
3363 return 0;
3364 if (c == arg_short) {
3365 c = *++arg;
3366 if (!c)
3367 return 1;
3368 if (val && isdigit(c)) {
3369 char *end;
3370 int n = strtoul(arg, &end, 10);
3371 if (*end)
3372 return 0;
3373 *val = n;
3374 return 1;
3376 return 0;
3378 if (c != '-')
3379 return 0;
3380 arg++;
3381 eq = strchrnul(arg, '=');
3382 len = eq - arg;
3383 if (!len || strncmp(arg, arg_long, len))
3384 return 0;
3385 if (*eq) {
3386 int n;
3387 char *end;
3388 if (!isdigit(*++eq))
3389 return 0;
3390 n = strtoul(eq, &end, 10);
3391 if (*end)
3392 return 0;
3393 *val = n;
3395 return 1;
3398 static int diff_scoreopt_parse(const char *opt);
3400 static inline int short_opt(char opt, const char **argv,
3401 const char **optarg)
3403 const char *arg = argv[0];
3404 if (arg[0] != '-' || arg[1] != opt)
3405 return 0;
3406 if (arg[2] != '\0') {
3407 *optarg = arg + 2;
3408 return 1;
3410 if (!argv[1])
3411 die("Option '%c' requires a value", opt);
3412 *optarg = argv[1];
3413 return 2;
3416 int parse_long_opt(const char *opt, const char **argv,
3417 const char **optarg)
3419 const char *arg = argv[0];
3420 if (!skip_prefix(arg, "--", &arg))
3421 return 0;
3422 if (!skip_prefix(arg, opt, &arg))
3423 return 0;
3424 if (*arg == '=') { /* stuck form: --option=value */
3425 *optarg = arg + 1;
3426 return 1;
3428 if (*arg != '\0')
3429 return 0;
3430 /* separate form: --option value */
3431 if (!argv[1])
3432 die("Option '--%s' requires a value", opt);
3433 *optarg = argv[1];
3434 return 2;
3437 static int stat_opt(struct diff_options *options, const char **av)
3439 const char *arg = av[0];
3440 char *end;
3441 int width = options->stat_width;
3442 int name_width = options->stat_name_width;
3443 int graph_width = options->stat_graph_width;
3444 int count = options->stat_count;
3445 int argcount = 1;
3447 if (!skip_prefix(arg, "--stat", &arg))
3448 die("BUG: stat option does not begin with --stat: %s", arg);
3449 end = (char *)arg;
3451 switch (*arg) {
3452 case '-':
3453 if (skip_prefix(arg, "-width", &arg)) {
3454 if (*arg == '=')
3455 width = strtoul(arg + 1, &end, 10);
3456 else if (!*arg && !av[1])
3457 die("Option '--stat-width' requires a value");
3458 else if (!*arg) {
3459 width = strtoul(av[1], &end, 10);
3460 argcount = 2;
3462 } else if (skip_prefix(arg, "-name-width", &arg)) {
3463 if (*arg == '=')
3464 name_width = strtoul(arg + 1, &end, 10);
3465 else if (!*arg && !av[1])
3466 die("Option '--stat-name-width' requires a value");
3467 else if (!*arg) {
3468 name_width = strtoul(av[1], &end, 10);
3469 argcount = 2;
3471 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3472 if (*arg == '=')
3473 graph_width = strtoul(arg + 1, &end, 10);
3474 else if (!*arg && !av[1])
3475 die("Option '--stat-graph-width' requires a value");
3476 else if (!*arg) {
3477 graph_width = strtoul(av[1], &end, 10);
3478 argcount = 2;
3480 } else if (skip_prefix(arg, "-count", &arg)) {
3481 if (*arg == '=')
3482 count = strtoul(arg + 1, &end, 10);
3483 else if (!*arg && !av[1])
3484 die("Option '--stat-count' requires a value");
3485 else if (!*arg) {
3486 count = strtoul(av[1], &end, 10);
3487 argcount = 2;
3490 break;
3491 case '=':
3492 width = strtoul(arg+1, &end, 10);
3493 if (*end == ',')
3494 name_width = strtoul(end+1, &end, 10);
3495 if (*end == ',')
3496 count = strtoul(end+1, &end, 10);
3499 /* Important! This checks all the error cases! */
3500 if (*end)
3501 return 0;
3502 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3503 options->stat_name_width = name_width;
3504 options->stat_graph_width = graph_width;
3505 options->stat_width = width;
3506 options->stat_count = count;
3507 return argcount;
3510 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3512 struct strbuf errmsg = STRBUF_INIT;
3513 if (parse_dirstat_params(options, params, &errmsg))
3514 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3515 errmsg.buf);
3516 strbuf_release(&errmsg);
3518 * The caller knows a dirstat-related option is given from the command
3519 * line; allow it to say "return this_function();"
3521 options->output_format |= DIFF_FORMAT_DIRSTAT;
3522 return 1;
3525 static int parse_submodule_opt(struct diff_options *options, const char *value)
3527 if (parse_submodule_params(options, value))
3528 die(_("Failed to parse --submodule option parameter: '%s'"),
3529 value);
3530 return 1;
3533 static const char diff_status_letters[] = {
3534 DIFF_STATUS_ADDED,
3535 DIFF_STATUS_COPIED,
3536 DIFF_STATUS_DELETED,
3537 DIFF_STATUS_MODIFIED,
3538 DIFF_STATUS_RENAMED,
3539 DIFF_STATUS_TYPE_CHANGED,
3540 DIFF_STATUS_UNKNOWN,
3541 DIFF_STATUS_UNMERGED,
3542 DIFF_STATUS_FILTER_AON,
3543 DIFF_STATUS_FILTER_BROKEN,
3544 '\0',
3547 static unsigned int filter_bit['Z' + 1];
3549 static void prepare_filter_bits(void)
3551 int i;
3553 if (!filter_bit[DIFF_STATUS_ADDED]) {
3554 for (i = 0; diff_status_letters[i]; i++)
3555 filter_bit[(int) diff_status_letters[i]] = (1 << i);
3559 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3561 return opt->filter & filter_bit[(int) status];
3564 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3566 int i, optch;
3568 prepare_filter_bits();
3571 * If there is a negation e.g. 'd' in the input, and we haven't
3572 * initialized the filter field with another --diff-filter, start
3573 * from full set of bits, except for AON.
3575 if (!opt->filter) {
3576 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3577 if (optch < 'a' || 'z' < optch)
3578 continue;
3579 opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3580 opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3581 break;
3585 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3586 unsigned int bit;
3587 int negate;
3589 if ('a' <= optch && optch <= 'z') {
3590 negate = 1;
3591 optch = toupper(optch);
3592 } else {
3593 negate = 0;
3596 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3597 if (!bit)
3598 return optarg[i];
3599 if (negate)
3600 opt->filter &= ~bit;
3601 else
3602 opt->filter |= bit;
3604 return 0;
3607 static void enable_patch_output(int *fmt) {
3608 *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3609 *fmt |= DIFF_FORMAT_PATCH;
3612 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3614 const char *arg = av[0];
3615 const char *optarg;
3616 int argcount;
3618 /* Output format options */
3619 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3620 || opt_arg(arg, 'U', "unified", &options->context))
3621 enable_patch_output(&options->output_format);
3622 else if (!strcmp(arg, "--raw"))
3623 options->output_format |= DIFF_FORMAT_RAW;
3624 else if (!strcmp(arg, "--patch-with-raw")) {
3625 enable_patch_output(&options->output_format);
3626 options->output_format |= DIFF_FORMAT_RAW;
3627 } else if (!strcmp(arg, "--numstat"))
3628 options->output_format |= DIFF_FORMAT_NUMSTAT;
3629 else if (!strcmp(arg, "--shortstat"))
3630 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3631 else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3632 return parse_dirstat_opt(options, "");
3633 else if (skip_prefix(arg, "-X", &arg))
3634 return parse_dirstat_opt(options, arg);
3635 else if (skip_prefix(arg, "--dirstat=", &arg))
3636 return parse_dirstat_opt(options, arg);
3637 else if (!strcmp(arg, "--cumulative"))
3638 return parse_dirstat_opt(options, "cumulative");
3639 else if (!strcmp(arg, "--dirstat-by-file"))
3640 return parse_dirstat_opt(options, "files");
3641 else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3642 parse_dirstat_opt(options, "files");
3643 return parse_dirstat_opt(options, arg);
3645 else if (!strcmp(arg, "--check"))
3646 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3647 else if (!strcmp(arg, "--summary"))
3648 options->output_format |= DIFF_FORMAT_SUMMARY;
3649 else if (!strcmp(arg, "--patch-with-stat")) {
3650 enable_patch_output(&options->output_format);
3651 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3652 } else if (!strcmp(arg, "--name-only"))
3653 options->output_format |= DIFF_FORMAT_NAME;
3654 else if (!strcmp(arg, "--name-status"))
3655 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3656 else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3657 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3658 else if (starts_with(arg, "--stat"))
3659 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3660 return stat_opt(options, av);
3662 /* renames options */
3663 else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3664 !strcmp(arg, "--break-rewrites")) {
3665 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3666 return error("invalid argument to -B: %s", arg+2);
3668 else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3669 !strcmp(arg, "--find-renames")) {
3670 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3671 return error("invalid argument to -M: %s", arg+2);
3672 options->detect_rename = DIFF_DETECT_RENAME;
3674 else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3675 options->irreversible_delete = 1;
3677 else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3678 !strcmp(arg, "--find-copies")) {
3679 if (options->detect_rename == DIFF_DETECT_COPY)
3680 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3681 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3682 return error("invalid argument to -C: %s", arg+2);
3683 options->detect_rename = DIFF_DETECT_COPY;
3685 else if (!strcmp(arg, "--no-renames"))
3686 options->detect_rename = 0;
3687 else if (!strcmp(arg, "--rename-empty"))
3688 DIFF_OPT_SET(options, RENAME_EMPTY);
3689 else if (!strcmp(arg, "--no-rename-empty"))
3690 DIFF_OPT_CLR(options, RENAME_EMPTY);
3691 else if (!strcmp(arg, "--relative"))
3692 DIFF_OPT_SET(options, RELATIVE_NAME);
3693 else if (skip_prefix(arg, "--relative=", &arg)) {
3694 DIFF_OPT_SET(options, RELATIVE_NAME);
3695 options->prefix = arg;
3698 /* xdiff options */
3699 else if (!strcmp(arg, "--minimal"))
3700 DIFF_XDL_SET(options, NEED_MINIMAL);
3701 else if (!strcmp(arg, "--no-minimal"))
3702 DIFF_XDL_CLR(options, NEED_MINIMAL);
3703 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3704 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3705 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3706 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3707 else if (!strcmp(arg, "--ignore-space-at-eol"))
3708 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3709 else if (!strcmp(arg, "--ignore-blank-lines"))
3710 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3711 else if (!strcmp(arg, "--patience"))
3712 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3713 else if (!strcmp(arg, "--histogram"))
3714 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3715 else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3716 long value = parse_algorithm_value(optarg);
3717 if (value < 0)
3718 return error("option diff-algorithm accepts \"myers\", "
3719 "\"minimal\", \"patience\" and \"histogram\"");
3720 /* clear out previous settings */
3721 DIFF_XDL_CLR(options, NEED_MINIMAL);
3722 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3723 options->xdl_opts |= value;
3724 return argcount;
3727 /* flags options */
3728 else if (!strcmp(arg, "--binary")) {
3729 enable_patch_output(&options->output_format);
3730 DIFF_OPT_SET(options, BINARY);
3732 else if (!strcmp(arg, "--full-index"))
3733 DIFF_OPT_SET(options, FULL_INDEX);
3734 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3735 DIFF_OPT_SET(options, TEXT);
3736 else if (!strcmp(arg, "-R"))
3737 DIFF_OPT_SET(options, REVERSE_DIFF);
3738 else if (!strcmp(arg, "--find-copies-harder"))
3739 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3740 else if (!strcmp(arg, "--follow"))
3741 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3742 else if (!strcmp(arg, "--no-follow"))
3743 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3744 else if (!strcmp(arg, "--color"))
3745 options->use_color = 1;
3746 else if (skip_prefix(arg, "--color=", &arg)) {
3747 int value = git_config_colorbool(NULL, arg);
3748 if (value < 0)
3749 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3750 options->use_color = value;
3752 else if (!strcmp(arg, "--no-color"))
3753 options->use_color = 0;
3754 else if (!strcmp(arg, "--color-words")) {
3755 options->use_color = 1;
3756 options->word_diff = DIFF_WORDS_COLOR;
3758 else if (skip_prefix(arg, "--color-words=", &arg)) {
3759 options->use_color = 1;
3760 options->word_diff = DIFF_WORDS_COLOR;
3761 options->word_regex = arg;
3763 else if (!strcmp(arg, "--word-diff")) {
3764 if (options->word_diff == DIFF_WORDS_NONE)
3765 options->word_diff = DIFF_WORDS_PLAIN;
3767 else if (skip_prefix(arg, "--word-diff=", &arg)) {
3768 if (!strcmp(arg, "plain"))
3769 options->word_diff = DIFF_WORDS_PLAIN;
3770 else if (!strcmp(arg, "color")) {
3771 options->use_color = 1;
3772 options->word_diff = DIFF_WORDS_COLOR;
3774 else if (!strcmp(arg, "porcelain"))
3775 options->word_diff = DIFF_WORDS_PORCELAIN;
3776 else if (!strcmp(arg, "none"))
3777 options->word_diff = DIFF_WORDS_NONE;
3778 else
3779 die("bad --word-diff argument: %s", arg);
3781 else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {