t4200: use cut instead of sed
[git/peff.git] / diff.c
blob00e1590c6eec291c2a6b3f831b62f402768c7042
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"
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 100;
23 int diff_use_color_default = -1;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
27 static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
38 static int parse_diff_color_slot(const char *var, int ofs)
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
57 static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 const char *cmd;
61 } *user_diff, **user_diff_tail;
64 * Currently there is only "diff.<drivername>.command" variable;
65 * because there are "diff.color.<slot>" variables, we are parsing
66 * this in a bit convoluted way to allow low level diff driver
67 * called "color".
69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
71 const char *name;
72 int namelen;
73 struct ll_diff_driver *drv;
75 name = var + 5;
76 namelen = ep - name;
77 for (drv = user_diff; drv; drv = drv->next)
78 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79 break;
80 if (!drv) {
81 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82 drv->name = xmemdupz(name, namelen);
83 if (!user_diff_tail)
84 user_diff_tail = &user_diff;
85 *user_diff_tail = drv;
86 user_diff_tail = &(drv->next);
89 return git_config_string(&(drv->cmd), var, value);
93 * 'diff.<what>.funcname' attribute can be specified in the configuration
94 * to define a customized regexp to find the beginning of a function to
95 * be used for hunk header lines of "diff -p" style output.
97 static struct funcname_pattern {
98 char *name;
99 char *pattern;
100 struct funcname_pattern *next;
101 } *funcname_pattern_list;
103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
105 const char *name;
106 int namelen;
107 struct funcname_pattern *pp;
109 name = var + 5; /* "diff." */
110 namelen = ep - name;
112 for (pp = funcname_pattern_list; pp; pp = pp->next)
113 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114 break;
115 if (!pp) {
116 pp = xcalloc(1, sizeof(*pp));
117 pp->name = xmemdupz(name, namelen);
118 pp->next = funcname_pattern_list;
119 funcname_pattern_list = pp;
121 free(pp->pattern);
122 pp->pattern = xstrdup(value);
123 return 0;
127 * These are to give UI layer defaults.
128 * The core-level commands such as git-diff-files should
129 * never be affected by the setting of diff.renames
130 * the user happens to have in the configuration file.
132 int git_diff_ui_config(const char *var, const char *value)
134 if (!strcmp(var, "diff.renamelimit")) {
135 diff_rename_limit_default = git_config_int(var, value);
136 return 0;
138 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
139 diff_use_color_default = git_config_colorbool(var, value, -1);
140 return 0;
142 if (!strcmp(var, "diff.renames")) {
143 if (!value)
144 diff_detect_rename_default = DIFF_DETECT_RENAME;
145 else if (!strcasecmp(value, "copies") ||
146 !strcasecmp(value, "copy"))
147 diff_detect_rename_default = DIFF_DETECT_COPY;
148 else if (git_config_bool(var,value))
149 diff_detect_rename_default = DIFF_DETECT_RENAME;
150 return 0;
152 if (!strcmp(var, "diff.autorefreshindex")) {
153 diff_auto_refresh_index = git_config_bool(var, value);
154 return 0;
156 if (!strcmp(var, "diff.external")) {
157 if (!value)
158 return config_error_nonbool(var);
159 external_diff_cmd_cfg = xstrdup(value);
160 return 0;
162 if (!prefixcmp(var, "diff.")) {
163 const char *ep = strrchr(var, '.');
165 if (ep != var + 4 && !strcmp(ep, ".command"))
166 return parse_lldiff_command(var, ep, value);
169 return git_diff_basic_config(var, value);
172 int git_diff_basic_config(const char *var, const char *value)
174 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
175 int slot = parse_diff_color_slot(var, 11);
176 if (!value)
177 return config_error_nonbool(var);
178 color_parse(value, var, diff_colors[slot]);
179 return 0;
182 if (!prefixcmp(var, "diff.")) {
183 const char *ep = strrchr(var, '.');
184 if (ep != var + 4) {
185 if (!strcmp(ep, ".funcname")) {
186 if (!value)
187 return config_error_nonbool(var);
188 return parse_funcname_pattern(var, ep, value);
193 return git_color_default_config(var, value);
196 static char *quote_two(const char *one, const char *two)
198 int need_one = quote_c_style(one, NULL, NULL, 1);
199 int need_two = quote_c_style(two, NULL, NULL, 1);
200 struct strbuf res;
202 strbuf_init(&res, 0);
203 if (need_one + need_two) {
204 strbuf_addch(&res, '"');
205 quote_c_style(one, &res, NULL, 1);
206 quote_c_style(two, &res, NULL, 1);
207 strbuf_addch(&res, '"');
208 } else {
209 strbuf_addstr(&res, one);
210 strbuf_addstr(&res, two);
212 return strbuf_detach(&res, NULL);
215 static const char *external_diff(void)
217 static const char *external_diff_cmd = NULL;
218 static int done_preparing = 0;
220 if (done_preparing)
221 return external_diff_cmd;
222 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
223 if (!external_diff_cmd)
224 external_diff_cmd = external_diff_cmd_cfg;
225 done_preparing = 1;
226 return external_diff_cmd;
229 static struct diff_tempfile {
230 const char *name; /* filename external diff should read from */
231 char hex[41];
232 char mode[10];
233 char tmp_path[PATH_MAX];
234 } diff_temp[2];
236 static int count_lines(const char *data, int size)
238 int count, ch, completely_empty = 1, nl_just_seen = 0;
239 count = 0;
240 while (0 < size--) {
241 ch = *data++;
242 if (ch == '\n') {
243 count++;
244 nl_just_seen = 1;
245 completely_empty = 0;
247 else {
248 nl_just_seen = 0;
249 completely_empty = 0;
252 if (completely_empty)
253 return 0;
254 if (!nl_just_seen)
255 count++; /* no trailing newline */
256 return count;
259 static void print_line_count(int count)
261 switch (count) {
262 case 0:
263 printf("0,0");
264 break;
265 case 1:
266 printf("1");
267 break;
268 default:
269 printf("1,%d", count);
270 break;
274 static void copy_file_with_prefix(int prefix, const char *data, int size,
275 const char *set, const char *reset)
277 int ch, nl_just_seen = 1;
278 while (0 < size--) {
279 ch = *data++;
280 if (nl_just_seen) {
281 fputs(set, stdout);
282 putchar(prefix);
284 if (ch == '\n') {
285 nl_just_seen = 1;
286 fputs(reset, stdout);
287 } else
288 nl_just_seen = 0;
289 putchar(ch);
291 if (!nl_just_seen)
292 printf("%s\n\\ No newline at end of file\n", reset);
295 static void emit_rewrite_diff(const char *name_a,
296 const char *name_b,
297 struct diff_filespec *one,
298 struct diff_filespec *two,
299 struct diff_options *o)
301 int lc_a, lc_b;
302 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
303 const char *name_a_tab, *name_b_tab;
304 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
305 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
306 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
307 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
308 const char *reset = diff_get_color(color_diff, DIFF_RESET);
309 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
311 name_a += (*name_a == '/');
312 name_b += (*name_b == '/');
313 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
314 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
316 strbuf_reset(&a_name);
317 strbuf_reset(&b_name);
318 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
319 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
321 diff_populate_filespec(one, 0);
322 diff_populate_filespec(two, 0);
323 lc_a = count_lines(one->data, one->size);
324 lc_b = count_lines(two->data, two->size);
325 printf("%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
326 metainfo, a_name.buf, name_a_tab, reset,
327 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
328 print_line_count(lc_a);
329 printf(" +");
330 print_line_count(lc_b);
331 printf(" @@%s\n", reset);
332 if (lc_a)
333 copy_file_with_prefix('-', one->data, one->size, old, reset);
334 if (lc_b)
335 copy_file_with_prefix('+', two->data, two->size, new, reset);
338 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
340 if (!DIFF_FILE_VALID(one)) {
341 mf->ptr = (char *)""; /* does not matter */
342 mf->size = 0;
343 return 0;
345 else if (diff_populate_filespec(one, 0))
346 return -1;
347 mf->ptr = one->data;
348 mf->size = one->size;
349 return 0;
352 struct diff_words_buffer {
353 mmfile_t text;
354 long alloc;
355 long current; /* output pointer */
356 int suppressed_newline;
359 static void diff_words_append(char *line, unsigned long len,
360 struct diff_words_buffer *buffer)
362 if (buffer->text.size + len > buffer->alloc) {
363 buffer->alloc = (buffer->text.size + len) * 3 / 2;
364 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
366 line++;
367 len--;
368 memcpy(buffer->text.ptr + buffer->text.size, line, len);
369 buffer->text.size += len;
372 struct diff_words_data {
373 struct xdiff_emit_state xm;
374 struct diff_words_buffer minus, plus;
377 static void print_word(struct diff_words_buffer *buffer, int len, int color,
378 int suppress_newline)
380 const char *ptr;
381 int eol = 0;
383 if (len == 0)
384 return;
386 ptr = buffer->text.ptr + buffer->current;
387 buffer->current += len;
389 if (ptr[len - 1] == '\n') {
390 eol = 1;
391 len--;
394 fputs(diff_get_color(1, color), stdout);
395 fwrite(ptr, len, 1, stdout);
396 fputs(diff_get_color(1, DIFF_RESET), stdout);
398 if (eol) {
399 if (suppress_newline)
400 buffer->suppressed_newline = 1;
401 else
402 putchar('\n');
406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
408 struct diff_words_data *diff_words = priv;
410 if (diff_words->minus.suppressed_newline) {
411 if (line[0] != '+')
412 putchar('\n');
413 diff_words->minus.suppressed_newline = 0;
416 len--;
417 switch (line[0]) {
418 case '-':
419 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
420 break;
421 case '+':
422 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
423 break;
424 case ' ':
425 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
426 diff_words->minus.current += len;
427 break;
431 /* this executes the word diff on the accumulated buffers */
432 static void diff_words_show(struct diff_words_data *diff_words)
434 xpparam_t xpp;
435 xdemitconf_t xecfg;
436 xdemitcb_t ecb;
437 mmfile_t minus, plus;
438 int i;
440 memset(&xecfg, 0, sizeof(xecfg));
441 minus.size = diff_words->minus.text.size;
442 minus.ptr = xmalloc(minus.size);
443 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
444 for (i = 0; i < minus.size; i++)
445 if (isspace(minus.ptr[i]))
446 minus.ptr[i] = '\n';
447 diff_words->minus.current = 0;
449 plus.size = diff_words->plus.text.size;
450 plus.ptr = xmalloc(plus.size);
451 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
452 for (i = 0; i < plus.size; i++)
453 if (isspace(plus.ptr[i]))
454 plus.ptr[i] = '\n';
455 diff_words->plus.current = 0;
457 xpp.flags = XDF_NEED_MINIMAL;
458 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
459 ecb.outf = xdiff_outf;
460 ecb.priv = diff_words;
461 diff_words->xm.consume = fn_out_diff_words_aux;
462 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
464 free(minus.ptr);
465 free(plus.ptr);
466 diff_words->minus.text.size = diff_words->plus.text.size = 0;
468 if (diff_words->minus.suppressed_newline) {
469 putchar('\n');
470 diff_words->minus.suppressed_newline = 0;
474 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
476 struct emit_callback {
477 struct xdiff_emit_state xm;
478 int nparents, color_diff;
479 unsigned ws_rule;
480 sane_truncate_fn truncate;
481 const char **label_path;
482 struct diff_words_data *diff_words;
483 int *found_changesp;
486 static void free_diff_words_data(struct emit_callback *ecbdata)
488 if (ecbdata->diff_words) {
489 /* flush buffers */
490 if (ecbdata->diff_words->minus.text.size ||
491 ecbdata->diff_words->plus.text.size)
492 diff_words_show(ecbdata->diff_words);
494 free (ecbdata->diff_words->minus.text.ptr);
495 free (ecbdata->diff_words->plus.text.ptr);
496 free(ecbdata->diff_words);
497 ecbdata->diff_words = NULL;
501 const char *diff_get_color(int diff_use_color, enum color_diff ix)
503 if (diff_use_color)
504 return diff_colors[ix];
505 return "";
508 static void emit_line(const char *set, const char *reset, const char *line, int len)
510 fputs(set, stdout);
511 fwrite(line, len, 1, stdout);
512 fputs(reset, stdout);
515 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
517 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
518 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
520 if (!*ws)
521 emit_line(set, reset, line, len);
522 else {
523 /* Emit just the prefix, then the rest. */
524 emit_line(set, reset, line, ecbdata->nparents);
525 (void)check_and_emit_line(line + ecbdata->nparents,
526 len - ecbdata->nparents, ecbdata->ws_rule,
527 stdout, set, reset, ws);
531 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
533 const char *cp;
534 unsigned long allot;
535 size_t l = len;
537 if (ecb->truncate)
538 return ecb->truncate(line, len);
539 cp = line;
540 allot = l;
541 while (0 < l) {
542 (void) utf8_width(&cp, &l);
543 if (!cp)
544 break; /* truncated in the middle? */
546 return allot - l;
549 static void fn_out_consume(void *priv, char *line, unsigned long len)
551 int i;
552 int color;
553 struct emit_callback *ecbdata = priv;
554 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
555 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
556 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
558 *(ecbdata->found_changesp) = 1;
560 if (ecbdata->label_path[0]) {
561 const char *name_a_tab, *name_b_tab;
563 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
564 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
566 printf("%s--- %s%s%s\n",
567 meta, ecbdata->label_path[0], reset, name_a_tab);
568 printf("%s+++ %s%s%s\n",
569 meta, ecbdata->label_path[1], reset, name_b_tab);
570 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
573 /* This is not really necessary for now because
574 * this codepath only deals with two-way diffs.
576 for (i = 0; i < len && line[i] == '@'; i++)
578 if (2 <= i && i < len && line[i] == ' ') {
579 ecbdata->nparents = i - 1;
580 len = sane_truncate_line(ecbdata, line, len);
581 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
582 reset, line, len);
583 if (line[len-1] != '\n')
584 putchar('\n');
585 return;
588 if (len < ecbdata->nparents) {
589 emit_line(reset, reset, line, len);
590 return;
593 color = DIFF_PLAIN;
594 if (ecbdata->diff_words && ecbdata->nparents != 1)
595 /* fall back to normal diff */
596 free_diff_words_data(ecbdata);
597 if (ecbdata->diff_words) {
598 if (line[0] == '-') {
599 diff_words_append(line, len,
600 &ecbdata->diff_words->minus);
601 return;
602 } else if (line[0] == '+') {
603 diff_words_append(line, len,
604 &ecbdata->diff_words->plus);
605 return;
607 if (ecbdata->diff_words->minus.text.size ||
608 ecbdata->diff_words->plus.text.size)
609 diff_words_show(ecbdata->diff_words);
610 line++;
611 len--;
612 emit_line(plain, reset, line, len);
613 return;
615 for (i = 0; i < ecbdata->nparents && len; i++) {
616 if (line[i] == '-')
617 color = DIFF_FILE_OLD;
618 else if (line[i] == '+')
619 color = DIFF_FILE_NEW;
622 if (color != DIFF_FILE_NEW) {
623 emit_line(diff_get_color(ecbdata->color_diff, color),
624 reset, line, len);
625 return;
627 emit_add_line(reset, ecbdata, line, len);
630 static char *pprint_rename(const char *a, const char *b)
632 const char *old = a;
633 const char *new = b;
634 struct strbuf name;
635 int pfx_length, sfx_length;
636 int len_a = strlen(a);
637 int len_b = strlen(b);
638 int a_midlen, b_midlen;
639 int qlen_a = quote_c_style(a, NULL, NULL, 0);
640 int qlen_b = quote_c_style(b, NULL, NULL, 0);
642 strbuf_init(&name, 0);
643 if (qlen_a || qlen_b) {
644 quote_c_style(a, &name, NULL, 0);
645 strbuf_addstr(&name, " => ");
646 quote_c_style(b, &name, NULL, 0);
647 return strbuf_detach(&name, NULL);
650 /* Find common prefix */
651 pfx_length = 0;
652 while (*old && *new && *old == *new) {
653 if (*old == '/')
654 pfx_length = old - a + 1;
655 old++;
656 new++;
659 /* Find common suffix */
660 old = a + len_a;
661 new = b + len_b;
662 sfx_length = 0;
663 while (a <= old && b <= new && *old == *new) {
664 if (*old == '/')
665 sfx_length = len_a - (old - a);
666 old--;
667 new--;
671 * pfx{mid-a => mid-b}sfx
672 * {pfx-a => pfx-b}sfx
673 * pfx{sfx-a => sfx-b}
674 * name-a => name-b
676 a_midlen = len_a - pfx_length - sfx_length;
677 b_midlen = len_b - pfx_length - sfx_length;
678 if (a_midlen < 0)
679 a_midlen = 0;
680 if (b_midlen < 0)
681 b_midlen = 0;
683 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
684 if (pfx_length + sfx_length) {
685 strbuf_add(&name, a, pfx_length);
686 strbuf_addch(&name, '{');
688 strbuf_add(&name, a + pfx_length, a_midlen);
689 strbuf_addstr(&name, " => ");
690 strbuf_add(&name, b + pfx_length, b_midlen);
691 if (pfx_length + sfx_length) {
692 strbuf_addch(&name, '}');
693 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
695 return strbuf_detach(&name, NULL);
698 struct diffstat_t {
699 struct xdiff_emit_state xm;
701 int nr;
702 int alloc;
703 struct diffstat_file {
704 char *from_name;
705 char *name;
706 char *print_name;
707 unsigned is_unmerged:1;
708 unsigned is_binary:1;
709 unsigned is_renamed:1;
710 unsigned int added, deleted;
711 } **files;
714 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
715 const char *name_a,
716 const char *name_b)
718 struct diffstat_file *x;
719 x = xcalloc(sizeof (*x), 1);
720 if (diffstat->nr == diffstat->alloc) {
721 diffstat->alloc = alloc_nr(diffstat->alloc);
722 diffstat->files = xrealloc(diffstat->files,
723 diffstat->alloc * sizeof(x));
725 diffstat->files[diffstat->nr++] = x;
726 if (name_b) {
727 x->from_name = xstrdup(name_a);
728 x->name = xstrdup(name_b);
729 x->is_renamed = 1;
731 else {
732 x->from_name = NULL;
733 x->name = xstrdup(name_a);
735 return x;
738 static void diffstat_consume(void *priv, char *line, unsigned long len)
740 struct diffstat_t *diffstat = priv;
741 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
743 if (line[0] == '+')
744 x->added++;
745 else if (line[0] == '-')
746 x->deleted++;
749 const char mime_boundary_leader[] = "------------";
751 static int scale_linear(int it, int width, int max_change)
754 * make sure that at least one '-' is printed if there were deletions,
755 * and likewise for '+'.
757 if (max_change < 2)
758 return it;
759 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
762 static void show_name(const char *prefix, const char *name, int len,
763 const char *reset, const char *set)
765 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
768 static void show_graph(char ch, int cnt, const char *set, const char *reset)
770 if (cnt <= 0)
771 return;
772 printf("%s", set);
773 while (cnt--)
774 putchar(ch);
775 printf("%s", reset);
778 static void fill_print_name(struct diffstat_file *file)
780 char *pname;
782 if (file->print_name)
783 return;
785 if (!file->is_renamed) {
786 struct strbuf buf;
787 strbuf_init(&buf, 0);
788 if (quote_c_style(file->name, &buf, NULL, 0)) {
789 pname = strbuf_detach(&buf, NULL);
790 } else {
791 pname = file->name;
792 strbuf_release(&buf);
794 } else {
795 pname = pprint_rename(file->from_name, file->name);
797 file->print_name = pname;
800 static void show_stats(struct diffstat_t* data, struct diff_options *options)
802 int i, len, add, del, total, adds = 0, dels = 0;
803 int max_change = 0, max_len = 0;
804 int total_files = data->nr;
805 int width, name_width;
806 const char *reset, *set, *add_c, *del_c;
808 if (data->nr == 0)
809 return;
811 width = options->stat_width ? options->stat_width : 80;
812 name_width = options->stat_name_width ? options->stat_name_width : 50;
814 /* Sanity: give at least 5 columns to the graph,
815 * but leave at least 10 columns for the name.
817 if (width < name_width + 15) {
818 if (name_width <= 25)
819 width = name_width + 15;
820 else
821 name_width = width - 15;
824 /* Find the longest filename and max number of changes */
825 reset = diff_get_color_opt(options, DIFF_RESET);
826 set = diff_get_color_opt(options, DIFF_PLAIN);
827 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
828 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
830 for (i = 0; i < data->nr; i++) {
831 struct diffstat_file *file = data->files[i];
832 int change = file->added + file->deleted;
833 fill_print_name(file);
834 len = strlen(file->print_name);
835 if (max_len < len)
836 max_len = len;
838 if (file->is_binary || file->is_unmerged)
839 continue;
840 if (max_change < change)
841 max_change = change;
844 /* Compute the width of the graph part;
845 * 10 is for one blank at the beginning of the line plus
846 * " | count " between the name and the graph.
848 * From here on, name_width is the width of the name area,
849 * and width is the width of the graph area.
851 name_width = (name_width < max_len) ? name_width : max_len;
852 if (width < (name_width + 10) + max_change)
853 width = width - (name_width + 10);
854 else
855 width = max_change;
857 for (i = 0; i < data->nr; i++) {
858 const char *prefix = "";
859 char *name = data->files[i]->print_name;
860 int added = data->files[i]->added;
861 int deleted = data->files[i]->deleted;
862 int name_len;
865 * "scale" the filename
867 len = name_width;
868 name_len = strlen(name);
869 if (name_width < name_len) {
870 char *slash;
871 prefix = "...";
872 len -= 3;
873 name += name_len - len;
874 slash = strchr(name, '/');
875 if (slash)
876 name = slash;
879 if (data->files[i]->is_binary) {
880 show_name(prefix, name, len, reset, set);
881 printf(" Bin ");
882 printf("%s%d%s", del_c, deleted, reset);
883 printf(" -> ");
884 printf("%s%d%s", add_c, added, reset);
885 printf(" bytes");
886 printf("\n");
887 continue;
889 else if (data->files[i]->is_unmerged) {
890 show_name(prefix, name, len, reset, set);
891 printf(" Unmerged\n");
892 continue;
894 else if (!data->files[i]->is_renamed &&
895 (added + deleted == 0)) {
896 total_files--;
897 continue;
901 * scale the add/delete
903 add = added;
904 del = deleted;
905 total = add + del;
906 adds += add;
907 dels += del;
909 if (width <= max_change) {
910 add = scale_linear(add, width, max_change);
911 del = scale_linear(del, width, max_change);
912 total = add + del;
914 show_name(prefix, name, len, reset, set);
915 printf("%5d ", added + deleted);
916 show_graph('+', add, add_c, reset);
917 show_graph('-', del, del_c, reset);
918 putchar('\n');
920 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
921 set, total_files, adds, dels, reset);
924 static void show_shortstats(struct diffstat_t* data)
926 int i, adds = 0, dels = 0, total_files = data->nr;
928 if (data->nr == 0)
929 return;
931 for (i = 0; i < data->nr; i++) {
932 if (!data->files[i]->is_binary &&
933 !data->files[i]->is_unmerged) {
934 int added = data->files[i]->added;
935 int deleted= data->files[i]->deleted;
936 if (!data->files[i]->is_renamed &&
937 (added + deleted == 0)) {
938 total_files--;
939 } else {
940 adds += added;
941 dels += deleted;
945 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
946 total_files, adds, dels);
949 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
951 int i;
953 if (data->nr == 0)
954 return;
956 for (i = 0; i < data->nr; i++) {
957 struct diffstat_file *file = data->files[i];
959 if (file->is_binary)
960 printf("-\t-\t");
961 else
962 printf("%d\t%d\t", file->added, file->deleted);
963 if (options->line_termination) {
964 fill_print_name(file);
965 if (!file->is_renamed)
966 write_name_quoted(file->name, stdout,
967 options->line_termination);
968 else {
969 fputs(file->print_name, stdout);
970 putchar(options->line_termination);
972 } else {
973 if (file->is_renamed) {
974 putchar('\0');
975 write_name_quoted(file->from_name, stdout, '\0');
977 write_name_quoted(file->name, stdout, '\0');
982 struct diffstat_dir {
983 struct diffstat_file **files;
984 int nr, percent, cumulative;
987 static long gather_dirstat(struct diffstat_dir *dir, unsigned long changed, const char *base, int baselen)
989 unsigned long this_dir = 0;
990 unsigned int sources = 0;
992 while (dir->nr) {
993 struct diffstat_file *f = *dir->files;
994 int namelen = strlen(f->name);
995 unsigned long this;
996 char *slash;
998 if (namelen < baselen)
999 break;
1000 if (memcmp(f->name, base, baselen))
1001 break;
1002 slash = strchr(f->name + baselen, '/');
1003 if (slash) {
1004 int newbaselen = slash + 1 - f->name;
1005 this = gather_dirstat(dir, changed, f->name, newbaselen);
1006 sources++;
1007 } else {
1008 if (f->is_unmerged || f->is_binary)
1009 this = 0;
1010 else
1011 this = f->added + f->deleted;
1012 dir->files++;
1013 dir->nr--;
1014 sources += 2;
1016 this_dir += this;
1020 * We don't report dirstat's for
1021 * - the top level
1022 * - or cases where everything came from a single directory
1023 * under this directory (sources == 1).
1025 if (baselen && sources != 1) {
1026 int permille = this_dir * 1000 / changed;
1027 if (permille) {
1028 int percent = permille / 10;
1029 if (percent >= dir->percent) {
1030 printf("%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1031 if (!dir->cumulative)
1032 return 0;
1036 return this_dir;
1039 static void show_dirstat(struct diffstat_t *data, struct diff_options *options)
1041 int i;
1042 unsigned long changed;
1043 struct diffstat_dir dir;
1045 /* Calculate total changes */
1046 changed = 0;
1047 for (i = 0; i < data->nr; i++) {
1048 if (data->files[i]->is_binary || data->files[i]->is_unmerged)
1049 continue;
1050 changed += data->files[i]->added;
1051 changed += data->files[i]->deleted;
1054 /* This can happen even with many files, if everything was renames */
1055 if (!changed)
1056 return;
1058 /* Show all directories with more than x% of the changes */
1059 dir.files = data->files;
1060 dir.nr = data->nr;
1061 dir.percent = options->dirstat_percent;
1062 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1063 gather_dirstat(&dir, changed, "", 0);
1066 static void free_diffstat_info(struct diffstat_t *diffstat)
1068 int i;
1069 for (i = 0; i < diffstat->nr; i++) {
1070 struct diffstat_file *f = diffstat->files[i];
1071 if (f->name != f->print_name)
1072 free(f->print_name);
1073 free(f->name);
1074 free(f->from_name);
1075 free(f);
1077 free(diffstat->files);
1080 struct checkdiff_t {
1081 struct xdiff_emit_state xm;
1082 const char *filename;
1083 int lineno, color_diff;
1084 unsigned ws_rule;
1085 unsigned status;
1088 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1090 struct checkdiff_t *data = priv;
1091 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1092 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1093 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1094 char *err;
1096 if (line[0] == '+') {
1097 data->lineno++;
1098 data->status = check_and_emit_line(line + 1, len - 1,
1099 data->ws_rule, NULL, NULL, NULL, NULL);
1100 if (!data->status)
1101 return;
1102 err = whitespace_error_string(data->status);
1103 printf("%s:%d: %s.\n", data->filename, data->lineno, err);
1104 free(err);
1105 emit_line(set, reset, line, 1);
1106 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1107 stdout, set, reset, ws);
1108 } else if (line[0] == ' ')
1109 data->lineno++;
1110 else if (line[0] == '@') {
1111 char *plus = strchr(line, '+');
1112 if (plus)
1113 data->lineno = strtol(plus, NULL, 10) - 1;
1114 else
1115 die("invalid diff");
1119 static unsigned char *deflate_it(char *data,
1120 unsigned long size,
1121 unsigned long *result_size)
1123 int bound;
1124 unsigned char *deflated;
1125 z_stream stream;
1127 memset(&stream, 0, sizeof(stream));
1128 deflateInit(&stream, zlib_compression_level);
1129 bound = deflateBound(&stream, size);
1130 deflated = xmalloc(bound);
1131 stream.next_out = deflated;
1132 stream.avail_out = bound;
1134 stream.next_in = (unsigned char *)data;
1135 stream.avail_in = size;
1136 while (deflate(&stream, Z_FINISH) == Z_OK)
1137 ; /* nothing */
1138 deflateEnd(&stream);
1139 *result_size = stream.total_out;
1140 return deflated;
1143 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1145 void *cp;
1146 void *delta;
1147 void *deflated;
1148 void *data;
1149 unsigned long orig_size;
1150 unsigned long delta_size;
1151 unsigned long deflate_size;
1152 unsigned long data_size;
1154 /* We could do deflated delta, or we could do just deflated two,
1155 * whichever is smaller.
1157 delta = NULL;
1158 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1159 if (one->size && two->size) {
1160 delta = diff_delta(one->ptr, one->size,
1161 two->ptr, two->size,
1162 &delta_size, deflate_size);
1163 if (delta) {
1164 void *to_free = delta;
1165 orig_size = delta_size;
1166 delta = deflate_it(delta, delta_size, &delta_size);
1167 free(to_free);
1171 if (delta && delta_size < deflate_size) {
1172 printf("delta %lu\n", orig_size);
1173 free(deflated);
1174 data = delta;
1175 data_size = delta_size;
1177 else {
1178 printf("literal %lu\n", two->size);
1179 free(delta);
1180 data = deflated;
1181 data_size = deflate_size;
1184 /* emit data encoded in base85 */
1185 cp = data;
1186 while (data_size) {
1187 int bytes = (52 < data_size) ? 52 : data_size;
1188 char line[70];
1189 data_size -= bytes;
1190 if (bytes <= 26)
1191 line[0] = bytes + 'A' - 1;
1192 else
1193 line[0] = bytes - 26 + 'a' - 1;
1194 encode_85(line + 1, cp, bytes);
1195 cp = (char *) cp + bytes;
1196 puts(line);
1198 printf("\n");
1199 free(data);
1202 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1204 printf("GIT binary patch\n");
1205 emit_binary_diff_body(one, two);
1206 emit_binary_diff_body(two, one);
1209 static void setup_diff_attr_check(struct git_attr_check *check)
1211 static struct git_attr *attr_diff;
1213 if (!attr_diff) {
1214 attr_diff = git_attr("diff", 4);
1216 check[0].attr = attr_diff;
1219 static void diff_filespec_check_attr(struct diff_filespec *one)
1221 struct git_attr_check attr_diff_check;
1222 int check_from_data = 0;
1224 if (one->checked_attr)
1225 return;
1227 setup_diff_attr_check(&attr_diff_check);
1228 one->is_binary = 0;
1229 one->funcname_pattern_ident = NULL;
1231 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1232 const char *value;
1234 /* binaryness */
1235 value = attr_diff_check.value;
1236 if (ATTR_TRUE(value))
1238 else if (ATTR_FALSE(value))
1239 one->is_binary = 1;
1240 else
1241 check_from_data = 1;
1243 /* funcname pattern ident */
1244 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1246 else
1247 one->funcname_pattern_ident = value;
1250 if (check_from_data) {
1251 if (!one->data && DIFF_FILE_VALID(one))
1252 diff_populate_filespec(one, 0);
1254 if (one->data)
1255 one->is_binary = buffer_is_binary(one->data, one->size);
1259 int diff_filespec_is_binary(struct diff_filespec *one)
1261 diff_filespec_check_attr(one);
1262 return one->is_binary;
1265 static const char *funcname_pattern(const char *ident)
1267 struct funcname_pattern *pp;
1269 for (pp = funcname_pattern_list; pp; pp = pp->next)
1270 if (!strcmp(ident, pp->name))
1271 return pp->pattern;
1272 return NULL;
1275 static struct builtin_funcname_pattern {
1276 const char *name;
1277 const char *pattern;
1278 } builtin_funcname_pattern[] = {
1279 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1280 "new\\|return\\|switch\\|throw\\|while\\)\n"
1281 "^[ ]*\\(\\([ ]*"
1282 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1283 "[ ]*([^;]*\\)$" },
1284 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1287 static const char *diff_funcname_pattern(struct diff_filespec *one)
1289 const char *ident, *pattern;
1290 int i;
1292 diff_filespec_check_attr(one);
1293 ident = one->funcname_pattern_ident;
1295 if (!ident)
1297 * If the config file has "funcname.default" defined, that
1298 * regexp is used; otherwise NULL is returned and xemit uses
1299 * the built-in default.
1301 return funcname_pattern("default");
1303 /* Look up custom "funcname.$ident" regexp from config. */
1304 pattern = funcname_pattern(ident);
1305 if (pattern)
1306 return pattern;
1309 * And define built-in fallback patterns here. Note that
1310 * these can be overridden by the user's config settings.
1312 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1313 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1314 return builtin_funcname_pattern[i].pattern;
1316 return NULL;
1319 static void builtin_diff(const char *name_a,
1320 const char *name_b,
1321 struct diff_filespec *one,
1322 struct diff_filespec *two,
1323 const char *xfrm_msg,
1324 struct diff_options *o,
1325 int complete_rewrite)
1327 mmfile_t mf1, mf2;
1328 const char *lbl[2];
1329 char *a_one, *b_two;
1330 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1331 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1333 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1334 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1335 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1336 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1337 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1338 if (lbl[0][0] == '/') {
1339 /* /dev/null */
1340 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1341 if (xfrm_msg && xfrm_msg[0])
1342 printf("%s%s%s\n", set, xfrm_msg, reset);
1344 else if (lbl[1][0] == '/') {
1345 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1346 if (xfrm_msg && xfrm_msg[0])
1347 printf("%s%s%s\n", set, xfrm_msg, reset);
1349 else {
1350 if (one->mode != two->mode) {
1351 printf("%sold mode %06o%s\n", set, one->mode, reset);
1352 printf("%snew mode %06o%s\n", set, two->mode, reset);
1354 if (xfrm_msg && xfrm_msg[0])
1355 printf("%s%s%s\n", set, xfrm_msg, reset);
1357 * we do not run diff between different kind
1358 * of objects.
1360 if ((one->mode ^ two->mode) & S_IFMT)
1361 goto free_ab_and_return;
1362 if (complete_rewrite) {
1363 emit_rewrite_diff(name_a, name_b, one, two, o);
1364 o->found_changes = 1;
1365 goto free_ab_and_return;
1369 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1370 die("unable to read files to diff");
1372 if (!DIFF_OPT_TST(o, TEXT) &&
1373 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1374 /* Quite common confusing case */
1375 if (mf1.size == mf2.size &&
1376 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1377 goto free_ab_and_return;
1378 if (DIFF_OPT_TST(o, BINARY))
1379 emit_binary_diff(&mf1, &mf2);
1380 else
1381 printf("Binary files %s and %s differ\n",
1382 lbl[0], lbl[1]);
1383 o->found_changes = 1;
1385 else {
1386 /* Crazy xdl interfaces.. */
1387 const char *diffopts = getenv("GIT_DIFF_OPTS");
1388 xpparam_t xpp;
1389 xdemitconf_t xecfg;
1390 xdemitcb_t ecb;
1391 struct emit_callback ecbdata;
1392 const char *funcname_pattern;
1394 funcname_pattern = diff_funcname_pattern(one);
1395 if (!funcname_pattern)
1396 funcname_pattern = diff_funcname_pattern(two);
1398 memset(&xecfg, 0, sizeof(xecfg));
1399 memset(&ecbdata, 0, sizeof(ecbdata));
1400 ecbdata.label_path = lbl;
1401 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1402 ecbdata.found_changesp = &o->found_changes;
1403 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1404 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1405 xecfg.ctxlen = o->context;
1406 xecfg.flags = XDL_EMIT_FUNCNAMES;
1407 if (funcname_pattern)
1408 xdiff_set_find_func(&xecfg, funcname_pattern);
1409 if (!diffopts)
1411 else if (!prefixcmp(diffopts, "--unified="))
1412 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1413 else if (!prefixcmp(diffopts, "-u"))
1414 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1415 ecb.outf = xdiff_outf;
1416 ecb.priv = &ecbdata;
1417 ecbdata.xm.consume = fn_out_consume;
1418 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1419 ecbdata.diff_words =
1420 xcalloc(1, sizeof(struct diff_words_data));
1421 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1422 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1423 free_diff_words_data(&ecbdata);
1426 free_ab_and_return:
1427 diff_free_filespec_data(one);
1428 diff_free_filespec_data(two);
1429 free(a_one);
1430 free(b_two);
1431 return;
1434 static void builtin_diffstat(const char *name_a, const char *name_b,
1435 struct diff_filespec *one,
1436 struct diff_filespec *two,
1437 struct diffstat_t *diffstat,
1438 struct diff_options *o,
1439 int complete_rewrite)
1441 mmfile_t mf1, mf2;
1442 struct diffstat_file *data;
1444 data = diffstat_add(diffstat, name_a, name_b);
1446 if (!one || !two) {
1447 data->is_unmerged = 1;
1448 return;
1450 if (complete_rewrite) {
1451 diff_populate_filespec(one, 0);
1452 diff_populate_filespec(two, 0);
1453 data->deleted = count_lines(one->data, one->size);
1454 data->added = count_lines(two->data, two->size);
1455 goto free_and_return;
1457 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1458 die("unable to read files to diff");
1460 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1461 data->is_binary = 1;
1462 data->added = mf2.size;
1463 data->deleted = mf1.size;
1464 } else {
1465 /* Crazy xdl interfaces.. */
1466 xpparam_t xpp;
1467 xdemitconf_t xecfg;
1468 xdemitcb_t ecb;
1470 memset(&xecfg, 0, sizeof(xecfg));
1471 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1472 ecb.outf = xdiff_outf;
1473 ecb.priv = diffstat;
1474 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1477 free_and_return:
1478 diff_free_filespec_data(one);
1479 diff_free_filespec_data(two);
1482 static void builtin_checkdiff(const char *name_a, const char *name_b,
1483 const char *attr_path,
1484 struct diff_filespec *one,
1485 struct diff_filespec *two, struct diff_options *o)
1487 mmfile_t mf1, mf2;
1488 struct checkdiff_t data;
1490 if (!two)
1491 return;
1493 memset(&data, 0, sizeof(data));
1494 data.xm.consume = checkdiff_consume;
1495 data.filename = name_b ? name_b : name_a;
1496 data.lineno = 0;
1497 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1498 data.ws_rule = whitespace_rule(attr_path);
1500 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1501 die("unable to read files to diff");
1503 if (diff_filespec_is_binary(two))
1504 goto free_and_return;
1505 else {
1506 /* Crazy xdl interfaces.. */
1507 xpparam_t xpp;
1508 xdemitconf_t xecfg;
1509 xdemitcb_t ecb;
1511 memset(&xecfg, 0, sizeof(xecfg));
1512 xpp.flags = XDF_NEED_MINIMAL;
1513 ecb.outf = xdiff_outf;
1514 ecb.priv = &data;
1515 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1517 free_and_return:
1518 diff_free_filespec_data(one);
1519 diff_free_filespec_data(two);
1520 if (data.status)
1521 DIFF_OPT_SET(o, CHECK_FAILED);
1524 struct diff_filespec *alloc_filespec(const char *path)
1526 int namelen = strlen(path);
1527 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1529 memset(spec, 0, sizeof(*spec));
1530 spec->path = (char *)(spec + 1);
1531 memcpy(spec->path, path, namelen+1);
1532 spec->count = 1;
1533 return spec;
1536 void free_filespec(struct diff_filespec *spec)
1538 if (!--spec->count) {
1539 diff_free_filespec_data(spec);
1540 free(spec);
1544 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1545 unsigned short mode)
1547 if (mode) {
1548 spec->mode = canon_mode(mode);
1549 hashcpy(spec->sha1, sha1);
1550 spec->sha1_valid = !is_null_sha1(sha1);
1555 * Given a name and sha1 pair, if the index tells us the file in
1556 * the work tree has that object contents, return true, so that
1557 * prepare_temp_file() does not have to inflate and extract.
1559 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1561 struct cache_entry *ce;
1562 struct stat st;
1563 int pos, len;
1565 /* We do not read the cache ourselves here, because the
1566 * benchmark with my previous version that always reads cache
1567 * shows that it makes things worse for diff-tree comparing
1568 * two linux-2.6 kernel trees in an already checked out work
1569 * tree. This is because most diff-tree comparisons deal with
1570 * only a small number of files, while reading the cache is
1571 * expensive for a large project, and its cost outweighs the
1572 * savings we get by not inflating the object to a temporary
1573 * file. Practically, this code only helps when we are used
1574 * by diff-cache --cached, which does read the cache before
1575 * calling us.
1577 if (!active_cache)
1578 return 0;
1580 /* We want to avoid the working directory if our caller
1581 * doesn't need the data in a normal file, this system
1582 * is rather slow with its stat/open/mmap/close syscalls,
1583 * and the object is contained in a pack file. The pack
1584 * is probably already open and will be faster to obtain
1585 * the data through than the working directory. Loose
1586 * objects however would tend to be slower as they need
1587 * to be individually opened and inflated.
1589 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1590 return 0;
1592 len = strlen(name);
1593 pos = cache_name_pos(name, len);
1594 if (pos < 0)
1595 return 0;
1596 ce = active_cache[pos];
1599 * This is not the sha1 we are looking for, or
1600 * unreusable because it is not a regular file.
1602 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1603 return 0;
1606 * If ce matches the file in the work tree, we can reuse it.
1608 if (ce_uptodate(ce) ||
1609 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1610 return 1;
1612 return 0;
1615 static int populate_from_stdin(struct diff_filespec *s)
1617 struct strbuf buf;
1618 size_t size = 0;
1620 strbuf_init(&buf, 0);
1621 if (strbuf_read(&buf, 0, 0) < 0)
1622 return error("error while reading from stdin %s",
1623 strerror(errno));
1625 s->should_munmap = 0;
1626 s->data = strbuf_detach(&buf, &size);
1627 s->size = size;
1628 s->should_free = 1;
1629 return 0;
1632 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1634 int len;
1635 char *data = xmalloc(100);
1636 len = snprintf(data, 100,
1637 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1638 s->data = data;
1639 s->size = len;
1640 s->should_free = 1;
1641 if (size_only) {
1642 s->data = NULL;
1643 free(data);
1645 return 0;
1649 * While doing rename detection and pickaxe operation, we may need to
1650 * grab the data for the blob (or file) for our own in-core comparison.
1651 * diff_filespec has data and size fields for this purpose.
1653 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1655 int err = 0;
1656 if (!DIFF_FILE_VALID(s))
1657 die("internal error: asking to populate invalid file.");
1658 if (S_ISDIR(s->mode))
1659 return -1;
1661 if (s->data)
1662 return 0;
1664 if (size_only && 0 < s->size)
1665 return 0;
1667 if (S_ISGITLINK(s->mode))
1668 return diff_populate_gitlink(s, size_only);
1670 if (!s->sha1_valid ||
1671 reuse_worktree_file(s->path, s->sha1, 0)) {
1672 struct strbuf buf;
1673 struct stat st;
1674 int fd;
1676 if (!strcmp(s->path, "-"))
1677 return populate_from_stdin(s);
1679 if (lstat(s->path, &st) < 0) {
1680 if (errno == ENOENT) {
1681 err_empty:
1682 err = -1;
1683 empty:
1684 s->data = (char *)"";
1685 s->size = 0;
1686 return err;
1689 s->size = xsize_t(st.st_size);
1690 if (!s->size)
1691 goto empty;
1692 if (size_only)
1693 return 0;
1694 if (S_ISLNK(st.st_mode)) {
1695 int ret;
1696 s->data = xmalloc(s->size);
1697 s->should_free = 1;
1698 ret = readlink(s->path, s->data, s->size);
1699 if (ret < 0) {
1700 free(s->data);
1701 goto err_empty;
1703 return 0;
1705 fd = open(s->path, O_RDONLY);
1706 if (fd < 0)
1707 goto err_empty;
1708 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1709 close(fd);
1710 s->should_munmap = 1;
1713 * Convert from working tree format to canonical git format
1715 strbuf_init(&buf, 0);
1716 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1717 size_t size = 0;
1718 munmap(s->data, s->size);
1719 s->should_munmap = 0;
1720 s->data = strbuf_detach(&buf, &size);
1721 s->size = size;
1722 s->should_free = 1;
1725 else {
1726 enum object_type type;
1727 if (size_only)
1728 type = sha1_object_info(s->sha1, &s->size);
1729 else {
1730 s->data = read_sha1_file(s->sha1, &type, &s->size);
1731 s->should_free = 1;
1734 return 0;
1737 void diff_free_filespec_blob(struct diff_filespec *s)
1739 if (s->should_free)
1740 free(s->data);
1741 else if (s->should_munmap)
1742 munmap(s->data, s->size);
1744 if (s->should_free || s->should_munmap) {
1745 s->should_free = s->should_munmap = 0;
1746 s->data = NULL;
1750 void diff_free_filespec_data(struct diff_filespec *s)
1752 diff_free_filespec_blob(s);
1753 free(s->cnt_data);
1754 s->cnt_data = NULL;
1757 static void prep_temp_blob(struct diff_tempfile *temp,
1758 void *blob,
1759 unsigned long size,
1760 const unsigned char *sha1,
1761 int mode)
1763 int fd;
1765 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1766 if (fd < 0)
1767 die("unable to create temp-file: %s", strerror(errno));
1768 if (write_in_full(fd, blob, size) != size)
1769 die("unable to write temp-file");
1770 close(fd);
1771 temp->name = temp->tmp_path;
1772 strcpy(temp->hex, sha1_to_hex(sha1));
1773 temp->hex[40] = 0;
1774 sprintf(temp->mode, "%06o", mode);
1777 static void prepare_temp_file(const char *name,
1778 struct diff_tempfile *temp,
1779 struct diff_filespec *one)
1781 if (!DIFF_FILE_VALID(one)) {
1782 not_a_valid_file:
1783 /* A '-' entry produces this for file-2, and
1784 * a '+' entry produces this for file-1.
1786 temp->name = "/dev/null";
1787 strcpy(temp->hex, ".");
1788 strcpy(temp->mode, ".");
1789 return;
1792 if (!one->sha1_valid ||
1793 reuse_worktree_file(name, one->sha1, 1)) {
1794 struct stat st;
1795 if (lstat(name, &st) < 0) {
1796 if (errno == ENOENT)
1797 goto not_a_valid_file;
1798 die("stat(%s): %s", name, strerror(errno));
1800 if (S_ISLNK(st.st_mode)) {
1801 int ret;
1802 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1803 size_t sz = xsize_t(st.st_size);
1804 if (sizeof(buf) <= st.st_size)
1805 die("symlink too long: %s", name);
1806 ret = readlink(name, buf, sz);
1807 if (ret < 0)
1808 die("readlink(%s)", name);
1809 prep_temp_blob(temp, buf, sz,
1810 (one->sha1_valid ?
1811 one->sha1 : null_sha1),
1812 (one->sha1_valid ?
1813 one->mode : S_IFLNK));
1815 else {
1816 /* we can borrow from the file in the work tree */
1817 temp->name = name;
1818 if (!one->sha1_valid)
1819 strcpy(temp->hex, sha1_to_hex(null_sha1));
1820 else
1821 strcpy(temp->hex, sha1_to_hex(one->sha1));
1822 /* Even though we may sometimes borrow the
1823 * contents from the work tree, we always want
1824 * one->mode. mode is trustworthy even when
1825 * !(one->sha1_valid), as long as
1826 * DIFF_FILE_VALID(one).
1828 sprintf(temp->mode, "%06o", one->mode);
1830 return;
1832 else {
1833 if (diff_populate_filespec(one, 0))
1834 die("cannot read data blob for %s", one->path);
1835 prep_temp_blob(temp, one->data, one->size,
1836 one->sha1, one->mode);
1840 static void remove_tempfile(void)
1842 int i;
1844 for (i = 0; i < 2; i++)
1845 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1846 unlink(diff_temp[i].name);
1847 diff_temp[i].name = NULL;
1851 static void remove_tempfile_on_signal(int signo)
1853 remove_tempfile();
1854 signal(SIGINT, SIG_DFL);
1855 raise(signo);
1858 /* An external diff command takes:
1860 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1861 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1864 static void run_external_diff(const char *pgm,
1865 const char *name,
1866 const char *other,
1867 struct diff_filespec *one,
1868 struct diff_filespec *two,
1869 const char *xfrm_msg,
1870 int complete_rewrite)
1872 const char *spawn_arg[10];
1873 struct diff_tempfile *temp = diff_temp;
1874 int retval;
1875 static int atexit_asked = 0;
1876 const char *othername;
1877 const char **arg = &spawn_arg[0];
1879 othername = (other? other : name);
1880 if (one && two) {
1881 prepare_temp_file(name, &temp[0], one);
1882 prepare_temp_file(othername, &temp[1], two);
1883 if (! atexit_asked &&
1884 (temp[0].name == temp[0].tmp_path ||
1885 temp[1].name == temp[1].tmp_path)) {
1886 atexit_asked = 1;
1887 atexit(remove_tempfile);
1889 signal(SIGINT, remove_tempfile_on_signal);
1892 if (one && two) {
1893 *arg++ = pgm;
1894 *arg++ = name;
1895 *arg++ = temp[0].name;
1896 *arg++ = temp[0].hex;
1897 *arg++ = temp[0].mode;
1898 *arg++ = temp[1].name;
1899 *arg++ = temp[1].hex;
1900 *arg++ = temp[1].mode;
1901 if (other) {
1902 *arg++ = other;
1903 *arg++ = xfrm_msg;
1905 } else {
1906 *arg++ = pgm;
1907 *arg++ = name;
1909 *arg = NULL;
1910 fflush(NULL);
1911 retval = run_command_v_opt(spawn_arg, 0);
1912 remove_tempfile();
1913 if (retval) {
1914 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1915 exit(1);
1919 static const char *external_diff_attr(const char *name)
1921 struct git_attr_check attr_diff_check;
1923 if (!name)
1924 return NULL;
1926 setup_diff_attr_check(&attr_diff_check);
1927 if (!git_checkattr(name, 1, &attr_diff_check)) {
1928 const char *value = attr_diff_check.value;
1929 if (!ATTR_TRUE(value) &&
1930 !ATTR_FALSE(value) &&
1931 !ATTR_UNSET(value)) {
1932 struct ll_diff_driver *drv;
1934 for (drv = user_diff; drv; drv = drv->next)
1935 if (!strcmp(drv->name, value))
1936 return drv->cmd;
1939 return NULL;
1942 static void run_diff_cmd(const char *pgm,
1943 const char *name,
1944 const char *other,
1945 const char *attr_path,
1946 struct diff_filespec *one,
1947 struct diff_filespec *two,
1948 const char *xfrm_msg,
1949 struct diff_options *o,
1950 int complete_rewrite)
1952 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1953 pgm = NULL;
1954 else {
1955 const char *cmd = external_diff_attr(attr_path);
1956 if (cmd)
1957 pgm = cmd;
1960 if (pgm) {
1961 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1962 complete_rewrite);
1963 return;
1965 if (one && two)
1966 builtin_diff(name, other ? other : name,
1967 one, two, xfrm_msg, o, complete_rewrite);
1968 else
1969 printf("* Unmerged path %s\n", name);
1972 static void diff_fill_sha1_info(struct diff_filespec *one)
1974 if (DIFF_FILE_VALID(one)) {
1975 if (!one->sha1_valid) {
1976 struct stat st;
1977 if (!strcmp(one->path, "-")) {
1978 hashcpy(one->sha1, null_sha1);
1979 return;
1981 if (lstat(one->path, &st) < 0)
1982 die("stat %s", one->path);
1983 if (index_path(one->sha1, one->path, &st, 0))
1984 die("cannot hash %s\n", one->path);
1987 else
1988 hashclr(one->sha1);
1991 static int similarity_index(struct diff_filepair *p)
1993 return p->score * 100 / MAX_SCORE;
1996 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
1998 /* Strip the prefix but do not molest /dev/null and absolute paths */
1999 if (*namep && **namep != '/')
2000 *namep += prefix_length;
2001 if (*otherp && **otherp != '/')
2002 *otherp += prefix_length;
2005 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2007 const char *pgm = external_diff();
2008 struct strbuf msg;
2009 char *xfrm_msg;
2010 struct diff_filespec *one = p->one;
2011 struct diff_filespec *two = p->two;
2012 const char *name;
2013 const char *other;
2014 const char *attr_path;
2015 int complete_rewrite = 0;
2017 name = p->one->path;
2018 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2019 attr_path = name;
2020 if (o->prefix_length)
2021 strip_prefix(o->prefix_length, &name, &other);
2023 if (DIFF_PAIR_UNMERGED(p)) {
2024 run_diff_cmd(pgm, name, NULL, attr_path,
2025 NULL, NULL, NULL, o, 0);
2026 return;
2029 diff_fill_sha1_info(one);
2030 diff_fill_sha1_info(two);
2032 strbuf_init(&msg, PATH_MAX * 2 + 300);
2033 switch (p->status) {
2034 case DIFF_STATUS_COPIED:
2035 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2036 strbuf_addstr(&msg, "\ncopy from ");
2037 quote_c_style(name, &msg, NULL, 0);
2038 strbuf_addstr(&msg, "\ncopy to ");
2039 quote_c_style(other, &msg, NULL, 0);
2040 strbuf_addch(&msg, '\n');
2041 break;
2042 case DIFF_STATUS_RENAMED:
2043 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2044 strbuf_addstr(&msg, "\nrename from ");
2045 quote_c_style(name, &msg, NULL, 0);
2046 strbuf_addstr(&msg, "\nrename to ");
2047 quote_c_style(other, &msg, NULL, 0);
2048 strbuf_addch(&msg, '\n');
2049 break;
2050 case DIFF_STATUS_MODIFIED:
2051 if (p->score) {
2052 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2053 similarity_index(p));
2054 complete_rewrite = 1;
2055 break;
2057 /* fallthru */
2058 default:
2059 /* nothing */
2063 if (hashcmp(one->sha1, two->sha1)) {
2064 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2066 if (DIFF_OPT_TST(o, BINARY)) {
2067 mmfile_t mf;
2068 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2069 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2070 abbrev = 40;
2072 strbuf_addf(&msg, "index %.*s..%.*s",
2073 abbrev, sha1_to_hex(one->sha1),
2074 abbrev, sha1_to_hex(two->sha1));
2075 if (one->mode == two->mode)
2076 strbuf_addf(&msg, " %06o", one->mode);
2077 strbuf_addch(&msg, '\n');
2080 if (msg.len)
2081 strbuf_setlen(&msg, msg.len - 1);
2082 xfrm_msg = msg.len ? msg.buf : NULL;
2084 if (!pgm &&
2085 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2086 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2087 /* a filepair that changes between file and symlink
2088 * needs to be split into deletion and creation.
2090 struct diff_filespec *null = alloc_filespec(two->path);
2091 run_diff_cmd(NULL, name, other, attr_path,
2092 one, null, xfrm_msg, o, 0);
2093 free(null);
2094 null = alloc_filespec(one->path);
2095 run_diff_cmd(NULL, name, other, attr_path,
2096 null, two, xfrm_msg, o, 0);
2097 free(null);
2099 else
2100 run_diff_cmd(pgm, name, other, attr_path,
2101 one, two, xfrm_msg, o, complete_rewrite);
2103 strbuf_release(&msg);
2106 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2107 struct diffstat_t *diffstat)
2109 const char *name;
2110 const char *other;
2111 int complete_rewrite = 0;
2113 if (DIFF_PAIR_UNMERGED(p)) {
2114 /* unmerged */
2115 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2116 return;
2119 name = p->one->path;
2120 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2122 if (o->prefix_length)
2123 strip_prefix(o->prefix_length, &name, &other);
2125 diff_fill_sha1_info(p->one);
2126 diff_fill_sha1_info(p->two);
2128 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2129 complete_rewrite = 1;
2130 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2133 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2135 const char *name;
2136 const char *other;
2137 const char *attr_path;
2139 if (DIFF_PAIR_UNMERGED(p)) {
2140 /* unmerged */
2141 return;
2144 name = p->one->path;
2145 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2146 attr_path = other ? other : name;
2148 if (o->prefix_length)
2149 strip_prefix(o->prefix_length, &name, &other);
2151 diff_fill_sha1_info(p->one);
2152 diff_fill_sha1_info(p->two);
2154 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2157 void diff_setup(struct diff_options *options)
2159 memset(options, 0, sizeof(*options));
2160 options->line_termination = '\n';
2161 options->break_opt = -1;
2162 options->rename_limit = -1;
2163 options->dirstat_percent = 3;
2164 options->context = 3;
2165 options->msg_sep = "";
2167 options->change = diff_change;
2168 options->add_remove = diff_addremove;
2169 if (diff_use_color_default > 0)
2170 DIFF_OPT_SET(options, COLOR_DIFF);
2171 else
2172 DIFF_OPT_CLR(options, COLOR_DIFF);
2173 options->detect_rename = diff_detect_rename_default;
2175 options->a_prefix = "a/";
2176 options->b_prefix = "b/";
2179 int diff_setup_done(struct diff_options *options)
2181 int count = 0;
2183 if (options->output_format & DIFF_FORMAT_NAME)
2184 count++;
2185 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2186 count++;
2187 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2188 count++;
2189 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2190 count++;
2191 if (count > 1)
2192 die("--name-only, --name-status, --check and -s are mutually exclusive");
2194 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2195 options->detect_rename = DIFF_DETECT_COPY;
2197 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2198 options->prefix = NULL;
2199 if (options->prefix)
2200 options->prefix_length = strlen(options->prefix);
2201 else
2202 options->prefix_length = 0;
2204 if (options->output_format & (DIFF_FORMAT_NAME |
2205 DIFF_FORMAT_NAME_STATUS |
2206 DIFF_FORMAT_CHECKDIFF |
2207 DIFF_FORMAT_NO_OUTPUT))
2208 options->output_format &= ~(DIFF_FORMAT_RAW |
2209 DIFF_FORMAT_NUMSTAT |
2210 DIFF_FORMAT_DIFFSTAT |
2211 DIFF_FORMAT_SHORTSTAT |
2212 DIFF_FORMAT_DIRSTAT |
2213 DIFF_FORMAT_SUMMARY |
2214 DIFF_FORMAT_PATCH);
2217 * These cases always need recursive; we do not drop caller-supplied
2218 * recursive bits for other formats here.
2220 if (options->output_format & (DIFF_FORMAT_PATCH |
2221 DIFF_FORMAT_NUMSTAT |
2222 DIFF_FORMAT_DIFFSTAT |
2223 DIFF_FORMAT_SHORTSTAT |
2224 DIFF_FORMAT_DIRSTAT |
2225 DIFF_FORMAT_SUMMARY |
2226 DIFF_FORMAT_CHECKDIFF))
2227 DIFF_OPT_SET(options, RECURSIVE);
2229 * Also pickaxe would not work very well if you do not say recursive
2231 if (options->pickaxe)
2232 DIFF_OPT_SET(options, RECURSIVE);
2234 if (options->detect_rename && options->rename_limit < 0)
2235 options->rename_limit = diff_rename_limit_default;
2236 if (options->setup & DIFF_SETUP_USE_CACHE) {
2237 if (!active_cache)
2238 /* read-cache does not die even when it fails
2239 * so it is safe for us to do this here. Also
2240 * it does not smudge active_cache or active_nr
2241 * when it fails, so we do not have to worry about
2242 * cleaning it up ourselves either.
2244 read_cache();
2246 if (options->abbrev <= 0 || 40 < options->abbrev)
2247 options->abbrev = 40; /* full */
2250 * It does not make sense to show the first hit we happened
2251 * to have found. It does not make sense not to return with
2252 * exit code in such a case either.
2254 if (DIFF_OPT_TST(options, QUIET)) {
2255 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2256 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2260 * If we postprocess in diffcore, we cannot simply return
2261 * upon the first hit. We need to run diff as usual.
2263 if (options->pickaxe || options->filter)
2264 DIFF_OPT_CLR(options, QUIET);
2266 return 0;
2269 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2271 char c, *eq;
2272 int len;
2274 if (*arg != '-')
2275 return 0;
2276 c = *++arg;
2277 if (!c)
2278 return 0;
2279 if (c == arg_short) {
2280 c = *++arg;
2281 if (!c)
2282 return 1;
2283 if (val && isdigit(c)) {
2284 char *end;
2285 int n = strtoul(arg, &end, 10);
2286 if (*end)
2287 return 0;
2288 *val = n;
2289 return 1;
2291 return 0;
2293 if (c != '-')
2294 return 0;
2295 arg++;
2296 eq = strchr(arg, '=');
2297 if (eq)
2298 len = eq - arg;
2299 else
2300 len = strlen(arg);
2301 if (!len || strncmp(arg, arg_long, len))
2302 return 0;
2303 if (eq) {
2304 int n;
2305 char *end;
2306 if (!isdigit(*++eq))
2307 return 0;
2308 n = strtoul(eq, &end, 10);
2309 if (*end)
2310 return 0;
2311 *val = n;
2313 return 1;
2316 static int diff_scoreopt_parse(const char *opt);
2318 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2320 const char *arg = av[0];
2322 /* Output format options */
2323 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2324 options->output_format |= DIFF_FORMAT_PATCH;
2325 else if (opt_arg(arg, 'U', "unified", &options->context))
2326 options->output_format |= DIFF_FORMAT_PATCH;
2327 else if (!strcmp(arg, "--raw"))
2328 options->output_format |= DIFF_FORMAT_RAW;
2329 else if (!strcmp(arg, "--patch-with-raw"))
2330 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2331 else if (!strcmp(arg, "--numstat"))
2332 options->output_format |= DIFF_FORMAT_NUMSTAT;
2333 else if (!strcmp(arg, "--shortstat"))
2334 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2335 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2336 options->output_format |= DIFF_FORMAT_DIRSTAT;
2337 else if (!strcmp(arg, "--cumulative"))
2338 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2339 else if (!strcmp(arg, "--check"))
2340 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2341 else if (!strcmp(arg, "--summary"))
2342 options->output_format |= DIFF_FORMAT_SUMMARY;
2343 else if (!strcmp(arg, "--patch-with-stat"))
2344 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2345 else if (!strcmp(arg, "--name-only"))
2346 options->output_format |= DIFF_FORMAT_NAME;
2347 else if (!strcmp(arg, "--name-status"))
2348 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2349 else if (!strcmp(arg, "-s"))
2350 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2351 else if (!prefixcmp(arg, "--stat")) {
2352 char *end;
2353 int width = options->stat_width;
2354 int name_width = options->stat_name_width;
2355 arg += 6;
2356 end = (char *)arg;
2358 switch (*arg) {
2359 case '-':
2360 if (!prefixcmp(arg, "-width="))
2361 width = strtoul(arg + 7, &end, 10);
2362 else if (!prefixcmp(arg, "-name-width="))
2363 name_width = strtoul(arg + 12, &end, 10);
2364 break;
2365 case '=':
2366 width = strtoul(arg+1, &end, 10);
2367 if (*end == ',')
2368 name_width = strtoul(end+1, &end, 10);
2371 /* Important! This checks all the error cases! */
2372 if (*end)
2373 return 0;
2374 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2375 options->stat_name_width = name_width;
2376 options->stat_width = width;
2379 /* renames options */
2380 else if (!prefixcmp(arg, "-B")) {
2381 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2382 return -1;
2384 else if (!prefixcmp(arg, "-M")) {
2385 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2386 return -1;
2387 options->detect_rename = DIFF_DETECT_RENAME;
2389 else if (!prefixcmp(arg, "-C")) {
2390 if (options->detect_rename == DIFF_DETECT_COPY)
2391 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2392 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2393 return -1;
2394 options->detect_rename = DIFF_DETECT_COPY;
2396 else if (!strcmp(arg, "--no-renames"))
2397 options->detect_rename = 0;
2398 else if (!strcmp(arg, "--relative"))
2399 DIFF_OPT_SET(options, RELATIVE_NAME);
2400 else if (!prefixcmp(arg, "--relative=")) {
2401 DIFF_OPT_SET(options, RELATIVE_NAME);
2402 options->prefix = arg + 11;
2405 /* xdiff options */
2406 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2407 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2408 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2409 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2410 else if (!strcmp(arg, "--ignore-space-at-eol"))
2411 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2413 /* flags options */
2414 else if (!strcmp(arg, "--binary")) {
2415 options->output_format |= DIFF_FORMAT_PATCH;
2416 DIFF_OPT_SET(options, BINARY);
2418 else if (!strcmp(arg, "--full-index"))
2419 DIFF_OPT_SET(options, FULL_INDEX);
2420 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2421 DIFF_OPT_SET(options, TEXT);
2422 else if (!strcmp(arg, "-R"))
2423 DIFF_OPT_SET(options, REVERSE_DIFF);
2424 else if (!strcmp(arg, "--find-copies-harder"))
2425 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2426 else if (!strcmp(arg, "--follow"))
2427 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2428 else if (!strcmp(arg, "--color"))
2429 DIFF_OPT_SET(options, COLOR_DIFF);
2430 else if (!strcmp(arg, "--no-color"))
2431 DIFF_OPT_CLR(options, COLOR_DIFF);
2432 else if (!strcmp(arg, "--color-words"))
2433 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2434 else if (!strcmp(arg, "--exit-code"))
2435 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2436 else if (!strcmp(arg, "--quiet"))
2437 DIFF_OPT_SET(options, QUIET);
2438 else if (!strcmp(arg, "--ext-diff"))
2439 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2440 else if (!strcmp(arg, "--no-ext-diff"))
2441 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2443 /* misc options */
2444 else if (!strcmp(arg, "-z"))
2445 options->line_termination = 0;
2446 else if (!prefixcmp(arg, "-l"))
2447 options->rename_limit = strtoul(arg+2, NULL, 10);
2448 else if (!prefixcmp(arg, "-S"))
2449 options->pickaxe = arg + 2;
2450 else if (!strcmp(arg, "--pickaxe-all"))
2451 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2452 else if (!strcmp(arg, "--pickaxe-regex"))
2453 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2454 else if (!prefixcmp(arg, "-O"))
2455 options->orderfile = arg + 2;
2456 else if (!prefixcmp(arg, "--diff-filter="))
2457 options->filter = arg + 14;
2458 else if (!strcmp(arg, "--abbrev"))
2459 options->abbrev = DEFAULT_ABBREV;
2460 else if (!prefixcmp(arg, "--abbrev=")) {
2461 options->abbrev = strtoul(arg + 9, NULL, 10);
2462 if (options->abbrev < MINIMUM_ABBREV)
2463 options->abbrev = MINIMUM_ABBREV;
2464 else if (40 < options->abbrev)
2465 options->abbrev = 40;
2467 else if (!prefixcmp(arg, "--src-prefix="))
2468 options->a_prefix = arg + 13;
2469 else if (!prefixcmp(arg, "--dst-prefix="))
2470 options->b_prefix = arg + 13;
2471 else if (!strcmp(arg, "--no-prefix"))
2472 options->a_prefix = options->b_prefix = "";
2473 else
2474 return 0;
2475 return 1;
2478 static int parse_num(const char **cp_p)
2480 unsigned long num, scale;
2481 int ch, dot;
2482 const char *cp = *cp_p;
2484 num = 0;
2485 scale = 1;
2486 dot = 0;
2487 for(;;) {
2488 ch = *cp;
2489 if ( !dot && ch == '.' ) {
2490 scale = 1;
2491 dot = 1;
2492 } else if ( ch == '%' ) {
2493 scale = dot ? scale*100 : 100;
2494 cp++; /* % is always at the end */
2495 break;
2496 } else if ( ch >= '0' && ch <= '9' ) {
2497 if ( scale < 100000 ) {
2498 scale *= 10;
2499 num = (num*10) + (ch-'0');
2501 } else {
2502 break;
2504 cp++;
2506 *cp_p = cp;
2508 /* user says num divided by scale and we say internally that
2509 * is MAX_SCORE * num / scale.
2511 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2514 static int diff_scoreopt_parse(const char *opt)
2516 int opt1, opt2, cmd;
2518 if (*opt++ != '-')
2519 return -1;
2520 cmd = *opt++;
2521 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2522 return -1; /* that is not a -M, -C nor -B option */
2524 opt1 = parse_num(&opt);
2525 if (cmd != 'B')
2526 opt2 = 0;
2527 else {
2528 if (*opt == 0)
2529 opt2 = 0;
2530 else if (*opt != '/')
2531 return -1; /* we expect -B80/99 or -B80 */
2532 else {
2533 opt++;
2534 opt2 = parse_num(&opt);
2537 if (*opt != 0)
2538 return -1;
2539 return opt1 | (opt2 << 16);
2542 struct diff_queue_struct diff_queued_diff;
2544 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2546 if (queue->alloc <= queue->nr) {
2547 queue->alloc = alloc_nr(queue->alloc);
2548 queue->queue = xrealloc(queue->queue,
2549 sizeof(dp) * queue->alloc);
2551 queue->queue[queue->nr++] = dp;
2554 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2555 struct diff_filespec *one,
2556 struct diff_filespec *two)
2558 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2559 dp->one = one;
2560 dp->two = two;
2561 if (queue)
2562 diff_q(queue, dp);
2563 return dp;
2566 void diff_free_filepair(struct diff_filepair *p)
2568 free_filespec(p->one);
2569 free_filespec(p->two);
2570 free(p);
2573 /* This is different from find_unique_abbrev() in that
2574 * it stuffs the result with dots for alignment.
2576 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2578 int abblen;
2579 const char *abbrev;
2580 if (len == 40)
2581 return sha1_to_hex(sha1);
2583 abbrev = find_unique_abbrev(sha1, len);
2584 abblen = strlen(abbrev);
2585 if (abblen < 37) {
2586 static char hex[41];
2587 if (len < abblen && abblen <= len + 2)
2588 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2589 else
2590 sprintf(hex, "%s...", abbrev);
2591 return hex;
2593 return sha1_to_hex(sha1);
2596 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2598 int line_termination = opt->line_termination;
2599 int inter_name_termination = line_termination ? '\t' : '\0';
2601 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2602 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2603 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2604 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2606 if (p->score) {
2607 printf("%c%03d%c", p->status, similarity_index(p),
2608 inter_name_termination);
2609 } else {
2610 printf("%c%c", p->status, inter_name_termination);
2613 if (p->status == DIFF_STATUS_COPIED ||
2614 p->status == DIFF_STATUS_RENAMED) {
2615 const char *name_a, *name_b;
2616 name_a = p->one->path;
2617 name_b = p->two->path;
2618 strip_prefix(opt->prefix_length, &name_a, &name_b);
2619 write_name_quoted(name_a, stdout, inter_name_termination);
2620 write_name_quoted(name_b, stdout, line_termination);
2621 } else {
2622 const char *name_a, *name_b;
2623 name_a = p->one->mode ? p->one->path : p->two->path;
2624 name_b = NULL;
2625 strip_prefix(opt->prefix_length, &name_a, &name_b);
2626 write_name_quoted(name_a, stdout, line_termination);
2630 int diff_unmodified_pair(struct diff_filepair *p)
2632 /* This function is written stricter than necessary to support
2633 * the currently implemented transformers, but the idea is to
2634 * let transformers to produce diff_filepairs any way they want,
2635 * and filter and clean them up here before producing the output.
2637 struct diff_filespec *one = p->one, *two = p->two;
2639 if (DIFF_PAIR_UNMERGED(p))
2640 return 0; /* unmerged is interesting */
2642 /* deletion, addition, mode or type change
2643 * and rename are all interesting.
2645 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2646 DIFF_PAIR_MODE_CHANGED(p) ||
2647 strcmp(one->path, two->path))
2648 return 0;
2650 /* both are valid and point at the same path. that is, we are
2651 * dealing with a change.
2653 if (one->sha1_valid && two->sha1_valid &&
2654 !hashcmp(one->sha1, two->sha1))
2655 return 1; /* no change */
2656 if (!one->sha1_valid && !two->sha1_valid)
2657 return 1; /* both look at the same file on the filesystem. */
2658 return 0;
2661 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2663 if (diff_unmodified_pair(p))
2664 return;
2666 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2667 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2668 return; /* no tree diffs in patch format */
2670 run_diff(p, o);
2673 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2674 struct diffstat_t *diffstat)
2676 if (diff_unmodified_pair(p))
2677 return;
2679 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2680 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2681 return; /* no tree diffs in patch format */
2683 run_diffstat(p, o, diffstat);
2686 static void diff_flush_checkdiff(struct diff_filepair *p,
2687 struct diff_options *o)
2689 if (diff_unmodified_pair(p))
2690 return;
2692 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2693 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2694 return; /* no tree diffs in patch format */
2696 run_checkdiff(p, o);
2699 int diff_queue_is_empty(void)
2701 struct diff_queue_struct *q = &diff_queued_diff;
2702 int i;
2703 for (i = 0; i < q->nr; i++)
2704 if (!diff_unmodified_pair(q->queue[i]))
2705 return 0;
2706 return 1;
2709 #if DIFF_DEBUG
2710 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2712 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2713 x, one ? one : "",
2714 s->path,
2715 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2716 s->mode,
2717 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2718 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2719 x, one ? one : "",
2720 s->size, s->xfrm_flags);
2723 void diff_debug_filepair(const struct diff_filepair *p, int i)
2725 diff_debug_filespec(p->one, i, "one");
2726 diff_debug_filespec(p->two, i, "two");
2727 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2728 p->score, p->status ? p->status : '?',
2729 p->one->rename_used, p->broken_pair);
2732 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2734 int i;
2735 if (msg)
2736 fprintf(stderr, "%s\n", msg);
2737 fprintf(stderr, "q->nr = %d\n", q->nr);
2738 for (i = 0; i < q->nr; i++) {
2739 struct diff_filepair *p = q->queue[i];
2740 diff_debug_filepair(p, i);
2743 #endif
2745 static void diff_resolve_rename_copy(void)
2747 int i;
2748 struct diff_filepair *p;
2749 struct diff_queue_struct *q = &diff_queued_diff;
2751 diff_debug_queue("resolve-rename-copy", q);
2753 for (i = 0; i < q->nr; i++) {
2754 p = q->queue[i];
2755 p->status = 0; /* undecided */
2756 if (DIFF_PAIR_UNMERGED(p))
2757 p->status = DIFF_STATUS_UNMERGED;
2758 else if (!DIFF_FILE_VALID(p->one))
2759 p->status = DIFF_STATUS_ADDED;
2760 else if (!DIFF_FILE_VALID(p->two))
2761 p->status = DIFF_STATUS_DELETED;
2762 else if (DIFF_PAIR_TYPE_CHANGED(p))
2763 p->status = DIFF_STATUS_TYPE_CHANGED;
2765 /* from this point on, we are dealing with a pair
2766 * whose both sides are valid and of the same type, i.e.
2767 * either in-place edit or rename/copy edit.
2769 else if (DIFF_PAIR_RENAME(p)) {
2771 * A rename might have re-connected a broken
2772 * pair up, causing the pathnames to be the
2773 * same again. If so, that's not a rename at
2774 * all, just a modification..
2776 * Otherwise, see if this source was used for
2777 * multiple renames, in which case we decrement
2778 * the count, and call it a copy.
2780 if (!strcmp(p->one->path, p->two->path))
2781 p->status = DIFF_STATUS_MODIFIED;
2782 else if (--p->one->rename_used > 0)
2783 p->status = DIFF_STATUS_COPIED;
2784 else
2785 p->status = DIFF_STATUS_RENAMED;
2787 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2788 p->one->mode != p->two->mode ||
2789 is_null_sha1(p->one->sha1))
2790 p->status = DIFF_STATUS_MODIFIED;
2791 else {
2792 /* This is a "no-change" entry and should not
2793 * happen anymore, but prepare for broken callers.
2795 error("feeding unmodified %s to diffcore",
2796 p->one->path);
2797 p->status = DIFF_STATUS_UNKNOWN;
2800 diff_debug_queue("resolve-rename-copy done", q);
2803 static int check_pair_status(struct diff_filepair *p)
2805 switch (p->status) {
2806 case DIFF_STATUS_UNKNOWN:
2807 return 0;
2808 case 0:
2809 die("internal error in diff-resolve-rename-copy");
2810 default:
2811 return 1;
2815 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2817 int fmt = opt->output_format;
2819 if (fmt & DIFF_FORMAT_CHECKDIFF)
2820 diff_flush_checkdiff(p, opt);
2821 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2822 diff_flush_raw(p, opt);
2823 else if (fmt & DIFF_FORMAT_NAME) {
2824 const char *name_a, *name_b;
2825 name_a = p->two->path;
2826 name_b = NULL;
2827 strip_prefix(opt->prefix_length, &name_a, &name_b);
2828 write_name_quoted(name_a, stdout, opt->line_termination);
2832 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2834 if (fs->mode)
2835 printf(" %s mode %06o ", newdelete, fs->mode);
2836 else
2837 printf(" %s ", newdelete);
2838 write_name_quoted(fs->path, stdout, '\n');
2842 static void show_mode_change(struct diff_filepair *p, int show_name)
2844 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2845 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2846 show_name ? ' ' : '\n');
2847 if (show_name) {
2848 write_name_quoted(p->two->path, stdout, '\n');
2853 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2855 char *names = pprint_rename(p->one->path, p->two->path);
2857 printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2858 free(names);
2859 show_mode_change(p, 0);
2862 static void diff_summary(struct diff_filepair *p)
2864 switch(p->status) {
2865 case DIFF_STATUS_DELETED:
2866 show_file_mode_name("delete", p->one);
2867 break;
2868 case DIFF_STATUS_ADDED:
2869 show_file_mode_name("create", p->two);
2870 break;
2871 case DIFF_STATUS_COPIED:
2872 show_rename_copy("copy", p);
2873 break;
2874 case DIFF_STATUS_RENAMED:
2875 show_rename_copy("rename", p);
2876 break;
2877 default:
2878 if (p->score) {
2879 fputs(" rewrite ", stdout);
2880 write_name_quoted(p->two->path, stdout, ' ');
2881 printf("(%d%%)\n", similarity_index(p));
2883 show_mode_change(p, !p->score);
2884 break;
2888 struct patch_id_t {
2889 struct xdiff_emit_state xm;
2890 SHA_CTX *ctx;
2891 int patchlen;
2894 static int remove_space(char *line, int len)
2896 int i;
2897 char *dst = line;
2898 unsigned char c;
2900 for (i = 0; i < len; i++)
2901 if (!isspace((c = line[i])))
2902 *dst++ = c;
2904 return dst - line;
2907 static void patch_id_consume(void *priv, char *line, unsigned long len)
2909 struct patch_id_t *data = priv;
2910 int new_len;
2912 /* Ignore line numbers when computing the SHA1 of the patch */
2913 if (!prefixcmp(line, "@@ -"))
2914 return;
2916 new_len = remove_space(line, len);
2918 SHA1_Update(data->ctx, line, new_len);
2919 data->patchlen += new_len;
2922 /* returns 0 upon success, and writes result into sha1 */
2923 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2925 struct diff_queue_struct *q = &diff_queued_diff;
2926 int i;
2927 SHA_CTX ctx;
2928 struct patch_id_t data;
2929 char buffer[PATH_MAX * 4 + 20];
2931 SHA1_Init(&ctx);
2932 memset(&data, 0, sizeof(struct patch_id_t));
2933 data.ctx = &ctx;
2934 data.xm.consume = patch_id_consume;
2936 for (i = 0; i < q->nr; i++) {
2937 xpparam_t xpp;
2938 xdemitconf_t xecfg;
2939 xdemitcb_t ecb;
2940 mmfile_t mf1, mf2;
2941 struct diff_filepair *p = q->queue[i];
2942 int len1, len2;
2944 memset(&xecfg, 0, sizeof(xecfg));
2945 if (p->status == 0)
2946 return error("internal diff status error");
2947 if (p->status == DIFF_STATUS_UNKNOWN)
2948 continue;
2949 if (diff_unmodified_pair(p))
2950 continue;
2951 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2952 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2953 continue;
2954 if (DIFF_PAIR_UNMERGED(p))
2955 continue;
2957 diff_fill_sha1_info(p->one);
2958 diff_fill_sha1_info(p->two);
2959 if (fill_mmfile(&mf1, p->one) < 0 ||
2960 fill_mmfile(&mf2, p->two) < 0)
2961 return error("unable to read files to diff");
2963 len1 = remove_space(p->one->path, strlen(p->one->path));
2964 len2 = remove_space(p->two->path, strlen(p->two->path));
2965 if (p->one->mode == 0)
2966 len1 = snprintf(buffer, sizeof(buffer),
2967 "diff--gita/%.*sb/%.*s"
2968 "newfilemode%06o"
2969 "---/dev/null"
2970 "+++b/%.*s",
2971 len1, p->one->path,
2972 len2, p->two->path,
2973 p->two->mode,
2974 len2, p->two->path);
2975 else if (p->two->mode == 0)
2976 len1 = snprintf(buffer, sizeof(buffer),
2977 "diff--gita/%.*sb/%.*s"
2978 "deletedfilemode%06o"
2979 "---a/%.*s"
2980 "+++/dev/null",
2981 len1, p->one->path,
2982 len2, p->two->path,
2983 p->one->mode,
2984 len1, p->one->path);
2985 else
2986 len1 = snprintf(buffer, sizeof(buffer),
2987 "diff--gita/%.*sb/%.*s"
2988 "---a/%.*s"
2989 "+++b/%.*s",
2990 len1, p->one->path,
2991 len2, p->two->path,
2992 len1, p->one->path,
2993 len2, p->two->path);
2994 SHA1_Update(&ctx, buffer, len1);
2996 xpp.flags = XDF_NEED_MINIMAL;
2997 xecfg.ctxlen = 3;
2998 xecfg.flags = XDL_EMIT_FUNCNAMES;
2999 ecb.outf = xdiff_outf;
3000 ecb.priv = &data;
3001 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3004 SHA1_Final(sha1, &ctx);
3005 return 0;
3008 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3010 struct diff_queue_struct *q = &diff_queued_diff;
3011 int i;
3012 int result = diff_get_patch_id(options, sha1);
3014 for (i = 0; i < q->nr; i++)
3015 diff_free_filepair(q->queue[i]);
3017 free(q->queue);
3018 q->queue = NULL;
3019 q->nr = q->alloc = 0;
3021 return result;
3024 static int is_summary_empty(const struct diff_queue_struct *q)
3026 int i;
3028 for (i = 0; i < q->nr; i++) {
3029 const struct diff_filepair *p = q->queue[i];
3031 switch (p->status) {
3032 case DIFF_STATUS_DELETED:
3033 case DIFF_STATUS_ADDED:
3034 case DIFF_STATUS_COPIED:
3035 case DIFF_STATUS_RENAMED:
3036 return 0;
3037 default:
3038 if (p->score)
3039 return 0;
3040 if (p->one->mode && p->two->mode &&
3041 p->one->mode != p->two->mode)
3042 return 0;
3043 break;
3046 return 1;
3049 void diff_flush(struct diff_options *options)
3051 struct diff_queue_struct *q = &diff_queued_diff;
3052 int i, output_format = options->output_format;
3053 int separator = 0;
3056 * Order: raw, stat, summary, patch
3057 * or: name/name-status/checkdiff (other bits clear)
3059 if (!q->nr)
3060 goto free_queue;
3062 if (output_format & (DIFF_FORMAT_RAW |
3063 DIFF_FORMAT_NAME |
3064 DIFF_FORMAT_NAME_STATUS |
3065 DIFF_FORMAT_CHECKDIFF)) {
3066 for (i = 0; i < q->nr; i++) {
3067 struct diff_filepair *p = q->queue[i];
3068 if (check_pair_status(p))
3069 flush_one_pair(p, options);
3071 separator++;
3074 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT|DIFF_FORMAT_DIRSTAT)) {
3075 struct diffstat_t diffstat;
3077 memset(&diffstat, 0, sizeof(struct diffstat_t));
3078 diffstat.xm.consume = diffstat_consume;
3079 for (i = 0; i < q->nr; i++) {
3080 struct diff_filepair *p = q->queue[i];
3081 if (check_pair_status(p))
3082 diff_flush_stat(p, options, &diffstat);
3084 if (output_format & DIFF_FORMAT_DIRSTAT)
3085 show_dirstat(&diffstat, options);
3086 if (output_format & DIFF_FORMAT_NUMSTAT)
3087 show_numstat(&diffstat, options);
3088 if (output_format & DIFF_FORMAT_DIFFSTAT)
3089 show_stats(&diffstat, options);
3090 if (output_format & DIFF_FORMAT_SHORTSTAT)
3091 show_shortstats(&diffstat);
3092 free_diffstat_info(&diffstat);
3093 separator++;
3096 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3097 for (i = 0; i < q->nr; i++)
3098 diff_summary(q->queue[i]);
3099 separator++;
3102 if (output_format & DIFF_FORMAT_PATCH) {
3103 if (separator) {
3104 if (options->stat_sep) {
3105 /* attach patch instead of inline */
3106 fputs(options->stat_sep, stdout);
3107 } else {
3108 putchar(options->line_termination);
3112 for (i = 0; i < q->nr; i++) {
3113 struct diff_filepair *p = q->queue[i];
3114 if (check_pair_status(p))
3115 diff_flush_patch(p, options);
3119 if (output_format & DIFF_FORMAT_CALLBACK)
3120 options->format_callback(q, options, options->format_callback_data);
3122 for (i = 0; i < q->nr; i++)
3123 diff_free_filepair(q->queue[i]);
3124 free_queue:
3125 free(q->queue);
3126 q->queue = NULL;
3127 q->nr = q->alloc = 0;
3130 static void diffcore_apply_filter(const char *filter)
3132 int i;
3133 struct diff_queue_struct *q = &diff_queued_diff;
3134 struct diff_queue_struct outq;
3135 outq.queue = NULL;
3136 outq.nr = outq.alloc = 0;
3138 if (!filter)
3139 return;
3141 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3142 int found;
3143 for (i = found = 0; !found && i < q->nr; i++) {
3144 struct diff_filepair *p = q->queue[i];
3145 if (((p->status == DIFF_STATUS_MODIFIED) &&
3146 ((p->score &&
3147 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3148 (!p->score &&
3149 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3150 ((p->status != DIFF_STATUS_MODIFIED) &&
3151 strchr(filter, p->status)))
3152 found++;
3154 if (found)
3155 return;
3157 /* otherwise we will clear the whole queue
3158 * by copying the empty outq at the end of this
3159 * function, but first clear the current entries
3160 * in the queue.
3162 for (i = 0; i < q->nr; i++)
3163 diff_free_filepair(q->queue[i]);
3165 else {
3166 /* Only the matching ones */
3167 for (i = 0; i < q->nr; i++) {
3168 struct diff_filepair *p = q->queue[i];
3170 if (((p->status == DIFF_STATUS_MODIFIED) &&
3171 ((p->score &&
3172 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3173 (!p->score &&
3174 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3175 ((p->status != DIFF_STATUS_MODIFIED) &&
3176 strchr(filter, p->status)))
3177 diff_q(&outq, p);
3178 else
3179 diff_free_filepair(p);
3182 free(q->queue);
3183 *q = outq;
3186 /* Check whether two filespecs with the same mode and size are identical */
3187 static int diff_filespec_is_identical(struct diff_filespec *one,
3188 struct diff_filespec *two)
3190 if (S_ISGITLINK(one->mode))
3191 return 0;
3192 if (diff_populate_filespec(one, 0))
3193 return 0;
3194 if (diff_populate_filespec(two, 0))
3195 return 0;
3196 return !memcmp(one->data, two->data, one->size);
3199 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3201 int i;
3202 struct diff_queue_struct *q = &diff_queued_diff;
3203 struct diff_queue_struct outq;
3204 outq.queue = NULL;
3205 outq.nr = outq.alloc = 0;
3207 for (i = 0; i < q->nr; i++) {
3208 struct diff_filepair *p = q->queue[i];
3211 * 1. Entries that come from stat info dirtyness
3212 * always have both sides (iow, not create/delete),
3213 * one side of the object name is unknown, with
3214 * the same mode and size. Keep the ones that
3215 * do not match these criteria. They have real
3216 * differences.
3218 * 2. At this point, the file is known to be modified,
3219 * with the same mode and size, and the object
3220 * name of one side is unknown. Need to inspect
3221 * the identical contents.
3223 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3224 !DIFF_FILE_VALID(p->two) ||
3225 (p->one->sha1_valid && p->two->sha1_valid) ||
3226 (p->one->mode != p->two->mode) ||
3227 diff_populate_filespec(p->one, 1) ||
3228 diff_populate_filespec(p->two, 1) ||
3229 (p->one->size != p->two->size) ||
3230 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3231 diff_q(&outq, p);
3232 else {
3234 * The caller can subtract 1 from skip_stat_unmatch
3235 * to determine how many paths were dirty only
3236 * due to stat info mismatch.
3238 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3239 diffopt->skip_stat_unmatch++;
3240 diff_free_filepair(p);
3243 free(q->queue);
3244 *q = outq;
3247 void diffcore_std(struct diff_options *options)
3249 if (DIFF_OPT_TST(options, QUIET))
3250 return;
3252 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3253 diffcore_skip_stat_unmatch(options);
3254 if (options->break_opt != -1)
3255 diffcore_break(options->break_opt);
3256 if (options->detect_rename)
3257 diffcore_rename(options);
3258 if (options->break_opt != -1)
3259 diffcore_merge_broken();
3260 if (options->pickaxe)
3261 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3262 if (options->orderfile)
3263 diffcore_order(options->orderfile);
3264 diff_resolve_rename_copy();
3265 diffcore_apply_filter(options->filter);
3267 if (diff_queued_diff.nr)
3268 DIFF_OPT_SET(options, HAS_CHANGES);
3269 else
3270 DIFF_OPT_CLR(options, HAS_CHANGES);
3273 int diff_result_code(struct diff_options *opt, int status)
3275 int result = 0;
3276 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3277 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3278 return status;
3279 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3280 DIFF_OPT_TST(opt, HAS_CHANGES))
3281 result |= 01;
3282 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3283 DIFF_OPT_TST(opt, CHECK_FAILED))
3284 result |= 02;
3285 return result;
3288 void diff_addremove(struct diff_options *options,
3289 int addremove, unsigned mode,
3290 const unsigned char *sha1,
3291 const char *base, const char *path)
3293 char concatpath[PATH_MAX];
3294 struct diff_filespec *one, *two;
3296 /* This may look odd, but it is a preparation for
3297 * feeding "there are unchanged files which should
3298 * not produce diffs, but when you are doing copy
3299 * detection you would need them, so here they are"
3300 * entries to the diff-core. They will be prefixed
3301 * with something like '=' or '*' (I haven't decided
3302 * which but should not make any difference).
3303 * Feeding the same new and old to diff_change()
3304 * also has the same effect.
3305 * Before the final output happens, they are pruned after
3306 * merged into rename/copy pairs as appropriate.
3308 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3309 addremove = (addremove == '+' ? '-' :
3310 addremove == '-' ? '+' : addremove);
3312 if (!path) path = "";
3313 sprintf(concatpath, "%s%s", base, path);
3315 if (options->prefix &&
3316 strncmp(concatpath, options->prefix, options->prefix_length))
3317 return;
3319 one = alloc_filespec(concatpath);
3320 two = alloc_filespec(concatpath);
3322 if (addremove != '+')
3323 fill_filespec(one, sha1, mode);
3324 if (addremove != '-')
3325 fill_filespec(two, sha1, mode);
3327 diff_queue(&diff_queued_diff, one, two);
3328 DIFF_OPT_SET(options, HAS_CHANGES);
3331 void diff_change(struct diff_options *options,
3332 unsigned old_mode, unsigned new_mode,
3333 const unsigned char *old_sha1,
3334 const unsigned char *new_sha1,
3335 const char *base, const char *path)
3337 char concatpath[PATH_MAX];
3338 struct diff_filespec *one, *two;
3340 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3341 unsigned tmp;
3342 const unsigned char *tmp_c;
3343 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3344 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3346 if (!path) path = "";
3347 sprintf(concatpath, "%s%s", base, path);
3349 if (options->prefix &&
3350 strncmp(concatpath, options->prefix, options->prefix_length))
3351 return;
3353 one = alloc_filespec(concatpath);
3354 two = alloc_filespec(concatpath);
3355 fill_filespec(one, old_sha1, old_mode);
3356 fill_filespec(two, new_sha1, new_mode);
3358 diff_queue(&diff_queued_diff, one, two);
3359 DIFF_OPT_SET(options, HAS_CHANGES);
3362 void diff_unmerge(struct diff_options *options,
3363 const char *path,
3364 unsigned mode, const unsigned char *sha1)
3366 struct diff_filespec *one, *two;
3368 if (options->prefix &&
3369 strncmp(path, options->prefix, options->prefix_length))
3370 return;
3372 one = alloc_filespec(path);
3373 two = alloc_filespec(path);
3374 fill_filespec(one, sha1, mode);
3375 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;