SubmittingPatches: fix a typo
[git/jnareb-git.git] / diff.c
blobfe43407b78150fd7fffe0d1fdff3456346ea27a6
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 = 200;
23 static int diff_suppress_blank_empty;
24 int diff_use_color_default = -1;
25 static const char *external_diff_cmd_cfg;
26 int diff_auto_refresh_index = 1;
28 static char diff_colors[][COLOR_MAXLEN] = {
29 "\033[m", /* reset */
30 "", /* PLAIN (normal) */
31 "\033[1m", /* METAINFO (bold) */
32 "\033[36m", /* FRAGINFO (cyan) */
33 "\033[31m", /* OLD (red) */
34 "\033[32m", /* NEW (green) */
35 "\033[33m", /* COMMIT (yellow) */
36 "\033[41m", /* WHITESPACE (red background) */
39 static int parse_diff_color_slot(const char *var, int ofs)
41 if (!strcasecmp(var+ofs, "plain"))
42 return DIFF_PLAIN;
43 if (!strcasecmp(var+ofs, "meta"))
44 return DIFF_METAINFO;
45 if (!strcasecmp(var+ofs, "frag"))
46 return DIFF_FRAGINFO;
47 if (!strcasecmp(var+ofs, "old"))
48 return DIFF_FILE_OLD;
49 if (!strcasecmp(var+ofs, "new"))
50 return DIFF_FILE_NEW;
51 if (!strcasecmp(var+ofs, "commit"))
52 return DIFF_COMMIT;
53 if (!strcasecmp(var+ofs, "whitespace"))
54 return DIFF_WHITESPACE;
55 die("bad config variable '%s'", var);
58 static struct ll_diff_driver {
59 const char *name;
60 struct ll_diff_driver *next;
61 const char *cmd;
62 } *user_diff, **user_diff_tail;
65 * Currently there is only "diff.<drivername>.command" variable;
66 * because there are "diff.color.<slot>" variables, we are parsing
67 * this in a bit convoluted way to allow low level diff driver
68 * called "color".
70 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
72 const char *name;
73 int namelen;
74 struct ll_diff_driver *drv;
76 name = var + 5;
77 namelen = ep - name;
78 for (drv = user_diff; drv; drv = drv->next)
79 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
80 break;
81 if (!drv) {
82 drv = xcalloc(1, sizeof(struct ll_diff_driver));
83 drv->name = xmemdupz(name, namelen);
84 if (!user_diff_tail)
85 user_diff_tail = &user_diff;
86 *user_diff_tail = drv;
87 user_diff_tail = &(drv->next);
90 return git_config_string(&(drv->cmd), var, value);
94 * 'diff.<what>.funcname' attribute can be specified in the configuration
95 * to define a customized regexp to find the beginning of a function to
96 * be used for hunk header lines of "diff -p" style output.
98 static struct funcname_pattern {
99 char *name;
100 char *pattern;
101 struct funcname_pattern *next;
102 } *funcname_pattern_list;
104 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
106 const char *name;
107 int namelen;
108 struct funcname_pattern *pp;
110 name = var + 5; /* "diff." */
111 namelen = ep - name;
113 for (pp = funcname_pattern_list; pp; pp = pp->next)
114 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
115 break;
116 if (!pp) {
117 pp = xcalloc(1, sizeof(*pp));
118 pp->name = xmemdupz(name, namelen);
119 pp->next = funcname_pattern_list;
120 funcname_pattern_list = pp;
122 free(pp->pattern);
123 pp->pattern = xstrdup(value);
124 return 0;
128 * These are to give UI layer defaults.
129 * The core-level commands such as git-diff-files should
130 * never be affected by the setting of diff.renames
131 * the user happens to have in the configuration file.
133 int git_diff_ui_config(const char *var, const char *value, void *cb)
135 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
136 diff_use_color_default = git_config_colorbool(var, value, -1);
137 return 0;
139 if (!strcmp(var, "diff.renames")) {
140 if (!value)
141 diff_detect_rename_default = DIFF_DETECT_RENAME;
142 else if (!strcasecmp(value, "copies") ||
143 !strcasecmp(value, "copy"))
144 diff_detect_rename_default = DIFF_DETECT_COPY;
145 else if (git_config_bool(var,value))
146 diff_detect_rename_default = DIFF_DETECT_RENAME;
147 return 0;
149 if (!strcmp(var, "diff.autorefreshindex")) {
150 diff_auto_refresh_index = git_config_bool(var, value);
151 return 0;
153 if (!strcmp(var, "diff.external"))
154 return git_config_string(&external_diff_cmd_cfg, var, value);
155 if (!prefixcmp(var, "diff.")) {
156 const char *ep = strrchr(var, '.');
158 if (ep != var + 4 && !strcmp(ep, ".command"))
159 return parse_lldiff_command(var, ep, value);
162 return git_diff_basic_config(var, value, cb);
165 int git_diff_basic_config(const char *var, const char *value, void *cb)
167 if (!strcmp(var, "diff.renamelimit")) {
168 diff_rename_limit_default = git_config_int(var, value);
169 return 0;
172 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
173 int slot = parse_diff_color_slot(var, 11);
174 if (!value)
175 return config_error_nonbool(var);
176 color_parse(value, var, diff_colors[slot]);
177 return 0;
180 /* like GNU diff's --suppress-blank-empty option */
181 if (!strcmp(var, "diff.suppress-blank-empty")) {
182 diff_suppress_blank_empty = git_config_bool(var, value);
183 return 0;
186 if (!prefixcmp(var, "diff.")) {
187 const char *ep = strrchr(var, '.');
188 if (ep != var + 4) {
189 if (!strcmp(ep, ".funcname")) {
190 if (!value)
191 return config_error_nonbool(var);
192 return parse_funcname_pattern(var, ep, value);
197 return git_color_default_config(var, value, cb);
200 static char *quote_two(const char *one, const char *two)
202 int need_one = quote_c_style(one, NULL, NULL, 1);
203 int need_two = quote_c_style(two, NULL, NULL, 1);
204 struct strbuf res;
206 strbuf_init(&res, 0);
207 if (need_one + need_two) {
208 strbuf_addch(&res, '"');
209 quote_c_style(one, &res, NULL, 1);
210 quote_c_style(two, &res, NULL, 1);
211 strbuf_addch(&res, '"');
212 } else {
213 strbuf_addstr(&res, one);
214 strbuf_addstr(&res, two);
216 return strbuf_detach(&res, NULL);
219 static const char *external_diff(void)
221 static const char *external_diff_cmd = NULL;
222 static int done_preparing = 0;
224 if (done_preparing)
225 return external_diff_cmd;
226 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
227 if (!external_diff_cmd)
228 external_diff_cmd = external_diff_cmd_cfg;
229 done_preparing = 1;
230 return external_diff_cmd;
233 static struct diff_tempfile {
234 const char *name; /* filename external diff should read from */
235 char hex[41];
236 char mode[10];
237 char tmp_path[PATH_MAX];
238 } diff_temp[2];
240 static int count_lines(const char *data, int size)
242 int count, ch, completely_empty = 1, nl_just_seen = 0;
243 count = 0;
244 while (0 < size--) {
245 ch = *data++;
246 if (ch == '\n') {
247 count++;
248 nl_just_seen = 1;
249 completely_empty = 0;
251 else {
252 nl_just_seen = 0;
253 completely_empty = 0;
256 if (completely_empty)
257 return 0;
258 if (!nl_just_seen)
259 count++; /* no trailing newline */
260 return count;
263 static void print_line_count(FILE *file, int count)
265 switch (count) {
266 case 0:
267 fprintf(file, "0,0");
268 break;
269 case 1:
270 fprintf(file, "1");
271 break;
272 default:
273 fprintf(file, "1,%d", count);
274 break;
278 static void copy_file_with_prefix(FILE *file,
279 int prefix, const char *data, int size,
280 const char *set, const char *reset)
282 int ch, nl_just_seen = 1;
283 while (0 < size--) {
284 ch = *data++;
285 if (nl_just_seen) {
286 fputs(set, file);
287 putc(prefix, file);
289 if (ch == '\n') {
290 nl_just_seen = 1;
291 fputs(reset, file);
292 } else
293 nl_just_seen = 0;
294 putc(ch, file);
296 if (!nl_just_seen)
297 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
300 static void emit_rewrite_diff(const char *name_a,
301 const char *name_b,
302 struct diff_filespec *one,
303 struct diff_filespec *two,
304 struct diff_options *o)
306 int lc_a, lc_b;
307 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
308 const char *name_a_tab, *name_b_tab;
309 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
310 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
311 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
312 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
313 const char *reset = diff_get_color(color_diff, DIFF_RESET);
314 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
316 name_a += (*name_a == '/');
317 name_b += (*name_b == '/');
318 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
319 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
321 strbuf_reset(&a_name);
322 strbuf_reset(&b_name);
323 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
324 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
326 diff_populate_filespec(one, 0);
327 diff_populate_filespec(two, 0);
328 lc_a = count_lines(one->data, one->size);
329 lc_b = count_lines(two->data, two->size);
330 fprintf(o->file,
331 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
332 metainfo, a_name.buf, name_a_tab, reset,
333 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
334 print_line_count(o->file, lc_a);
335 fprintf(o->file, " +");
336 print_line_count(o->file, lc_b);
337 fprintf(o->file, " @@%s\n", reset);
338 if (lc_a)
339 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
340 if (lc_b)
341 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
344 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
346 if (!DIFF_FILE_VALID(one)) {
347 mf->ptr = (char *)""; /* does not matter */
348 mf->size = 0;
349 return 0;
351 else if (diff_populate_filespec(one, 0))
352 return -1;
353 mf->ptr = one->data;
354 mf->size = one->size;
355 return 0;
358 struct diff_words_buffer {
359 mmfile_t text;
360 long alloc;
361 long current; /* output pointer */
362 int suppressed_newline;
365 static void diff_words_append(char *line, unsigned long len,
366 struct diff_words_buffer *buffer)
368 if (buffer->text.size + len > buffer->alloc) {
369 buffer->alloc = (buffer->text.size + len) * 3 / 2;
370 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
372 line++;
373 len--;
374 memcpy(buffer->text.ptr + buffer->text.size, line, len);
375 buffer->text.size += len;
378 struct diff_words_data {
379 struct xdiff_emit_state xm;
380 struct diff_words_buffer minus, plus;
381 FILE *file;
384 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
385 int suppress_newline)
387 const char *ptr;
388 int eol = 0;
390 if (len == 0)
391 return;
393 ptr = buffer->text.ptr + buffer->current;
394 buffer->current += len;
396 if (ptr[len - 1] == '\n') {
397 eol = 1;
398 len--;
401 fputs(diff_get_color(1, color), file);
402 fwrite(ptr, len, 1, file);
403 fputs(diff_get_color(1, DIFF_RESET), file);
405 if (eol) {
406 if (suppress_newline)
407 buffer->suppressed_newline = 1;
408 else
409 putc('\n', file);
413 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
415 struct diff_words_data *diff_words = priv;
417 if (diff_words->minus.suppressed_newline) {
418 if (line[0] != '+')
419 putc('\n', diff_words->file);
420 diff_words->minus.suppressed_newline = 0;
423 len--;
424 switch (line[0]) {
425 case '-':
426 print_word(diff_words->file,
427 &diff_words->minus, len, DIFF_FILE_OLD, 1);
428 break;
429 case '+':
430 print_word(diff_words->file,
431 &diff_words->plus, len, DIFF_FILE_NEW, 0);
432 break;
433 case ' ':
434 print_word(diff_words->file,
435 &diff_words->plus, len, DIFF_PLAIN, 0);
436 diff_words->minus.current += len;
437 break;
441 /* this executes the word diff on the accumulated buffers */
442 static void diff_words_show(struct diff_words_data *diff_words)
444 xpparam_t xpp;
445 xdemitconf_t xecfg;
446 xdemitcb_t ecb;
447 mmfile_t minus, plus;
448 int i;
450 memset(&xecfg, 0, sizeof(xecfg));
451 minus.size = diff_words->minus.text.size;
452 minus.ptr = xmalloc(minus.size);
453 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
454 for (i = 0; i < minus.size; i++)
455 if (isspace(minus.ptr[i]))
456 minus.ptr[i] = '\n';
457 diff_words->minus.current = 0;
459 plus.size = diff_words->plus.text.size;
460 plus.ptr = xmalloc(plus.size);
461 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
462 for (i = 0; i < plus.size; i++)
463 if (isspace(plus.ptr[i]))
464 plus.ptr[i] = '\n';
465 diff_words->plus.current = 0;
467 xpp.flags = XDF_NEED_MINIMAL;
468 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
469 ecb.outf = xdiff_outf;
470 ecb.priv = diff_words;
471 diff_words->xm.consume = fn_out_diff_words_aux;
472 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
474 free(minus.ptr);
475 free(plus.ptr);
476 diff_words->minus.text.size = diff_words->plus.text.size = 0;
478 if (diff_words->minus.suppressed_newline) {
479 putc('\n', diff_words->file);
480 diff_words->minus.suppressed_newline = 0;
484 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
486 struct emit_callback {
487 struct xdiff_emit_state xm;
488 int nparents, color_diff;
489 unsigned ws_rule;
490 sane_truncate_fn truncate;
491 const char **label_path;
492 struct diff_words_data *diff_words;
493 int *found_changesp;
494 FILE *file;
497 static void free_diff_words_data(struct emit_callback *ecbdata)
499 if (ecbdata->diff_words) {
500 /* flush buffers */
501 if (ecbdata->diff_words->minus.text.size ||
502 ecbdata->diff_words->plus.text.size)
503 diff_words_show(ecbdata->diff_words);
505 free (ecbdata->diff_words->minus.text.ptr);
506 free (ecbdata->diff_words->plus.text.ptr);
507 free(ecbdata->diff_words);
508 ecbdata->diff_words = NULL;
512 const char *diff_get_color(int diff_use_color, enum color_diff ix)
514 if (diff_use_color)
515 return diff_colors[ix];
516 return "";
519 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
521 int has_trailing_newline = (len > 0 && line[len-1] == '\n');
522 if (has_trailing_newline)
523 len--;
525 fputs(set, file);
526 fwrite(line, len, 1, file);
527 fputs(reset, file);
528 if (has_trailing_newline)
529 fputc('\n', file);
532 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
534 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
535 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
537 if (!*ws)
538 emit_line(ecbdata->file, set, reset, line, len);
539 else {
540 /* Emit just the prefix, then the rest. */
541 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
542 ws_check_emit(line + ecbdata->nparents,
543 len - ecbdata->nparents, ecbdata->ws_rule,
544 ecbdata->file, set, reset, ws);
548 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
550 const char *cp;
551 unsigned long allot;
552 size_t l = len;
554 if (ecb->truncate)
555 return ecb->truncate(line, len);
556 cp = line;
557 allot = l;
558 while (0 < l) {
559 (void) utf8_width(&cp, &l);
560 if (!cp)
561 break; /* truncated in the middle? */
563 return allot - l;
566 static void fn_out_consume(void *priv, char *line, unsigned long len)
568 int i;
569 int color;
570 struct emit_callback *ecbdata = priv;
571 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
572 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
573 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
575 *(ecbdata->found_changesp) = 1;
577 if (ecbdata->label_path[0]) {
578 const char *name_a_tab, *name_b_tab;
580 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
581 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
583 fprintf(ecbdata->file, "%s--- %s%s%s\n",
584 meta, ecbdata->label_path[0], reset, name_a_tab);
585 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
586 meta, ecbdata->label_path[1], reset, name_b_tab);
587 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
590 if (diff_suppress_blank_empty
591 && len == 2 && line[0] == ' ' && line[1] == '\n') {
592 line[0] = '\n';
593 len = 1;
596 /* This is not really necessary for now because
597 * this codepath only deals with two-way diffs.
599 for (i = 0; i < len && line[i] == '@'; i++)
601 if (2 <= i && i < len && line[i] == ' ') {
602 ecbdata->nparents = i - 1;
603 len = sane_truncate_line(ecbdata, line, len);
604 emit_line(ecbdata->file,
605 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
606 reset, line, len);
607 if (line[len-1] != '\n')
608 putc('\n', ecbdata->file);
609 return;
612 if (len < ecbdata->nparents) {
613 emit_line(ecbdata->file, reset, reset, line, len);
614 return;
617 color = DIFF_PLAIN;
618 if (ecbdata->diff_words && ecbdata->nparents != 1)
619 /* fall back to normal diff */
620 free_diff_words_data(ecbdata);
621 if (ecbdata->diff_words) {
622 if (line[0] == '-') {
623 diff_words_append(line, len,
624 &ecbdata->diff_words->minus);
625 return;
626 } else if (line[0] == '+') {
627 diff_words_append(line, len,
628 &ecbdata->diff_words->plus);
629 return;
631 if (ecbdata->diff_words->minus.text.size ||
632 ecbdata->diff_words->plus.text.size)
633 diff_words_show(ecbdata->diff_words);
634 line++;
635 len--;
636 emit_line(ecbdata->file, plain, reset, line, len);
637 return;
639 for (i = 0; i < ecbdata->nparents && len; i++) {
640 if (line[i] == '-')
641 color = DIFF_FILE_OLD;
642 else if (line[i] == '+')
643 color = DIFF_FILE_NEW;
646 if (color != DIFF_FILE_NEW) {
647 emit_line(ecbdata->file,
648 diff_get_color(ecbdata->color_diff, color),
649 reset, line, len);
650 return;
652 emit_add_line(reset, ecbdata, line, len);
655 static char *pprint_rename(const char *a, const char *b)
657 const char *old = a;
658 const char *new = b;
659 struct strbuf name;
660 int pfx_length, sfx_length;
661 int len_a = strlen(a);
662 int len_b = strlen(b);
663 int a_midlen, b_midlen;
664 int qlen_a = quote_c_style(a, NULL, NULL, 0);
665 int qlen_b = quote_c_style(b, NULL, NULL, 0);
667 strbuf_init(&name, 0);
668 if (qlen_a || qlen_b) {
669 quote_c_style(a, &name, NULL, 0);
670 strbuf_addstr(&name, " => ");
671 quote_c_style(b, &name, NULL, 0);
672 return strbuf_detach(&name, NULL);
675 /* Find common prefix */
676 pfx_length = 0;
677 while (*old && *new && *old == *new) {
678 if (*old == '/')
679 pfx_length = old - a + 1;
680 old++;
681 new++;
684 /* Find common suffix */
685 old = a + len_a;
686 new = b + len_b;
687 sfx_length = 0;
688 while (a <= old && b <= new && *old == *new) {
689 if (*old == '/')
690 sfx_length = len_a - (old - a);
691 old--;
692 new--;
696 * pfx{mid-a => mid-b}sfx
697 * {pfx-a => pfx-b}sfx
698 * pfx{sfx-a => sfx-b}
699 * name-a => name-b
701 a_midlen = len_a - pfx_length - sfx_length;
702 b_midlen = len_b - pfx_length - sfx_length;
703 if (a_midlen < 0)
704 a_midlen = 0;
705 if (b_midlen < 0)
706 b_midlen = 0;
708 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
709 if (pfx_length + sfx_length) {
710 strbuf_add(&name, a, pfx_length);
711 strbuf_addch(&name, '{');
713 strbuf_add(&name, a + pfx_length, a_midlen);
714 strbuf_addstr(&name, " => ");
715 strbuf_add(&name, b + pfx_length, b_midlen);
716 if (pfx_length + sfx_length) {
717 strbuf_addch(&name, '}');
718 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
720 return strbuf_detach(&name, NULL);
723 struct diffstat_t {
724 struct xdiff_emit_state xm;
726 int nr;
727 int alloc;
728 struct diffstat_file {
729 char *from_name;
730 char *name;
731 char *print_name;
732 unsigned is_unmerged:1;
733 unsigned is_binary:1;
734 unsigned is_renamed:1;
735 unsigned int added, deleted;
736 } **files;
739 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
740 const char *name_a,
741 const char *name_b)
743 struct diffstat_file *x;
744 x = xcalloc(sizeof (*x), 1);
745 if (diffstat->nr == diffstat->alloc) {
746 diffstat->alloc = alloc_nr(diffstat->alloc);
747 diffstat->files = xrealloc(diffstat->files,
748 diffstat->alloc * sizeof(x));
750 diffstat->files[diffstat->nr++] = x;
751 if (name_b) {
752 x->from_name = xstrdup(name_a);
753 x->name = xstrdup(name_b);
754 x->is_renamed = 1;
756 else {
757 x->from_name = NULL;
758 x->name = xstrdup(name_a);
760 return x;
763 static void diffstat_consume(void *priv, char *line, unsigned long len)
765 struct diffstat_t *diffstat = priv;
766 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
768 if (line[0] == '+')
769 x->added++;
770 else if (line[0] == '-')
771 x->deleted++;
774 const char mime_boundary_leader[] = "------------";
776 static int scale_linear(int it, int width, int max_change)
779 * make sure that at least one '-' is printed if there were deletions,
780 * and likewise for '+'.
782 if (max_change < 2)
783 return it;
784 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
787 static void show_name(FILE *file,
788 const char *prefix, const char *name, int len,
789 const char *reset, const char *set)
791 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
794 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
796 if (cnt <= 0)
797 return;
798 fprintf(file, "%s", set);
799 while (cnt--)
800 putc(ch, file);
801 fprintf(file, "%s", reset);
804 static void fill_print_name(struct diffstat_file *file)
806 char *pname;
808 if (file->print_name)
809 return;
811 if (!file->is_renamed) {
812 struct strbuf buf;
813 strbuf_init(&buf, 0);
814 if (quote_c_style(file->name, &buf, NULL, 0)) {
815 pname = strbuf_detach(&buf, NULL);
816 } else {
817 pname = file->name;
818 strbuf_release(&buf);
820 } else {
821 pname = pprint_rename(file->from_name, file->name);
823 file->print_name = pname;
826 static void show_stats(struct diffstat_t* data, struct diff_options *options)
828 int i, len, add, del, total, adds = 0, dels = 0;
829 int max_change = 0, max_len = 0;
830 int total_files = data->nr;
831 int width, name_width;
832 const char *reset, *set, *add_c, *del_c;
834 if (data->nr == 0)
835 return;
837 width = options->stat_width ? options->stat_width : 80;
838 name_width = options->stat_name_width ? options->stat_name_width : 50;
840 /* Sanity: give at least 5 columns to the graph,
841 * but leave at least 10 columns for the name.
843 if (width < 25)
844 width = 25;
845 if (name_width < 10)
846 name_width = 10;
847 else if (width < name_width + 15)
848 name_width = width - 15;
850 /* Find the longest filename and max number of changes */
851 reset = diff_get_color_opt(options, DIFF_RESET);
852 set = diff_get_color_opt(options, DIFF_PLAIN);
853 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
854 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
856 for (i = 0; i < data->nr; i++) {
857 struct diffstat_file *file = data->files[i];
858 int change = file->added + file->deleted;
859 fill_print_name(file);
860 len = strlen(file->print_name);
861 if (max_len < len)
862 max_len = len;
864 if (file->is_binary || file->is_unmerged)
865 continue;
866 if (max_change < change)
867 max_change = change;
870 /* Compute the width of the graph part;
871 * 10 is for one blank at the beginning of the line plus
872 * " | count " between the name and the graph.
874 * From here on, name_width is the width of the name area,
875 * and width is the width of the graph area.
877 name_width = (name_width < max_len) ? name_width : max_len;
878 if (width < (name_width + 10) + max_change)
879 width = width - (name_width + 10);
880 else
881 width = max_change;
883 for (i = 0; i < data->nr; i++) {
884 const char *prefix = "";
885 char *name = data->files[i]->print_name;
886 int added = data->files[i]->added;
887 int deleted = data->files[i]->deleted;
888 int name_len;
891 * "scale" the filename
893 len = name_width;
894 name_len = strlen(name);
895 if (name_width < name_len) {
896 char *slash;
897 prefix = "...";
898 len -= 3;
899 name += name_len - len;
900 slash = strchr(name, '/');
901 if (slash)
902 name = slash;
905 if (data->files[i]->is_binary) {
906 show_name(options->file, prefix, name, len, reset, set);
907 fprintf(options->file, " Bin ");
908 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
909 fprintf(options->file, " -> ");
910 fprintf(options->file, "%s%d%s", add_c, added, reset);
911 fprintf(options->file, " bytes");
912 fprintf(options->file, "\n");
913 continue;
915 else if (data->files[i]->is_unmerged) {
916 show_name(options->file, prefix, name, len, reset, set);
917 fprintf(options->file, " Unmerged\n");
918 continue;
920 else if (!data->files[i]->is_renamed &&
921 (added + deleted == 0)) {
922 total_files--;
923 continue;
927 * scale the add/delete
929 add = added;
930 del = deleted;
931 total = add + del;
932 adds += add;
933 dels += del;
935 if (width <= max_change) {
936 add = scale_linear(add, width, max_change);
937 del = scale_linear(del, width, max_change);
938 total = add + del;
940 show_name(options->file, prefix, name, len, reset, set);
941 fprintf(options->file, "%5d%s", added + deleted,
942 added + deleted ? " " : "");
943 show_graph(options->file, '+', add, add_c, reset);
944 show_graph(options->file, '-', del, del_c, reset);
945 fprintf(options->file, "\n");
947 fprintf(options->file,
948 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
949 set, total_files, adds, dels, reset);
952 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
954 int i, adds = 0, dels = 0, total_files = data->nr;
956 if (data->nr == 0)
957 return;
959 for (i = 0; i < data->nr; i++) {
960 if (!data->files[i]->is_binary &&
961 !data->files[i]->is_unmerged) {
962 int added = data->files[i]->added;
963 int deleted= data->files[i]->deleted;
964 if (!data->files[i]->is_renamed &&
965 (added + deleted == 0)) {
966 total_files--;
967 } else {
968 adds += added;
969 dels += deleted;
973 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
974 total_files, adds, dels);
977 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
979 int i;
981 if (data->nr == 0)
982 return;
984 for (i = 0; i < data->nr; i++) {
985 struct diffstat_file *file = data->files[i];
987 if (file->is_binary)
988 fprintf(options->file, "-\t-\t");
989 else
990 fprintf(options->file,
991 "%d\t%d\t", file->added, file->deleted);
992 if (options->line_termination) {
993 fill_print_name(file);
994 if (!file->is_renamed)
995 write_name_quoted(file->name, options->file,
996 options->line_termination);
997 else {
998 fputs(file->print_name, options->file);
999 putc(options->line_termination, options->file);
1001 } else {
1002 if (file->is_renamed) {
1003 putc('\0', options->file);
1004 write_name_quoted(file->from_name, options->file, '\0');
1006 write_name_quoted(file->name, options->file, '\0');
1011 struct dirstat_file {
1012 const char *name;
1013 unsigned long changed;
1016 struct dirstat_dir {
1017 struct dirstat_file *files;
1018 int alloc, nr, percent, cumulative;
1021 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1023 unsigned long this_dir = 0;
1024 unsigned int sources = 0;
1026 while (dir->nr) {
1027 struct dirstat_file *f = dir->files;
1028 int namelen = strlen(f->name);
1029 unsigned long this;
1030 char *slash;
1032 if (namelen < baselen)
1033 break;
1034 if (memcmp(f->name, base, baselen))
1035 break;
1036 slash = strchr(f->name + baselen, '/');
1037 if (slash) {
1038 int newbaselen = slash + 1 - f->name;
1039 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1040 sources++;
1041 } else {
1042 this = f->changed;
1043 dir->files++;
1044 dir->nr--;
1045 sources += 2;
1047 this_dir += this;
1051 * We don't report dirstat's for
1052 * - the top level
1053 * - or cases where everything came from a single directory
1054 * under this directory (sources == 1).
1056 if (baselen && sources != 1) {
1057 int permille = this_dir * 1000 / changed;
1058 if (permille) {
1059 int percent = permille / 10;
1060 if (percent >= dir->percent) {
1061 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1062 if (!dir->cumulative)
1063 return 0;
1067 return this_dir;
1070 static void show_dirstat(struct diff_options *options)
1072 int i;
1073 unsigned long changed;
1074 struct dirstat_dir dir;
1075 struct diff_queue_struct *q = &diff_queued_diff;
1077 dir.files = NULL;
1078 dir.alloc = 0;
1079 dir.nr = 0;
1080 dir.percent = options->dirstat_percent;
1081 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1083 changed = 0;
1084 for (i = 0; i < q->nr; i++) {
1085 struct diff_filepair *p = q->queue[i];
1086 const char *name;
1087 unsigned long copied, added, damage;
1089 name = p->one->path ? p->one->path : p->two->path;
1091 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1092 diff_populate_filespec(p->one, 0);
1093 diff_populate_filespec(p->two, 0);
1094 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1095 &copied, &added);
1096 diff_free_filespec_data(p->one);
1097 diff_free_filespec_data(p->two);
1098 } else if (DIFF_FILE_VALID(p->one)) {
1099 diff_populate_filespec(p->one, 1);
1100 copied = added = 0;
1101 diff_free_filespec_data(p->one);
1102 } else if (DIFF_FILE_VALID(p->two)) {
1103 diff_populate_filespec(p->two, 1);
1104 copied = 0;
1105 added = p->two->size;
1106 diff_free_filespec_data(p->two);
1107 } else
1108 continue;
1111 * Original minus copied is the removed material,
1112 * added is the new material. They are both damages
1113 * made to the preimage.
1115 damage = (p->one->size - copied) + added;
1117 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1118 dir.files[dir.nr].name = name;
1119 dir.files[dir.nr].changed = damage;
1120 changed += damage;
1121 dir.nr++;
1124 /* This can happen even with many files, if everything was renames */
1125 if (!changed)
1126 return;
1128 /* Show all directories with more than x% of the changes */
1129 gather_dirstat(options->file, &dir, changed, "", 0);
1132 static void free_diffstat_info(struct diffstat_t *diffstat)
1134 int i;
1135 for (i = 0; i < diffstat->nr; i++) {
1136 struct diffstat_file *f = diffstat->files[i];
1137 if (f->name != f->print_name)
1138 free(f->print_name);
1139 free(f->name);
1140 free(f->from_name);
1141 free(f);
1143 free(diffstat->files);
1146 struct checkdiff_t {
1147 struct xdiff_emit_state xm;
1148 const char *filename;
1149 int lineno;
1150 struct diff_options *o;
1151 unsigned ws_rule;
1152 unsigned status;
1153 int trailing_blanks_start;
1156 static int is_conflict_marker(const char *line, unsigned long len)
1158 char firstchar;
1159 int cnt;
1161 if (len < 8)
1162 return 0;
1163 firstchar = line[0];
1164 switch (firstchar) {
1165 case '=': case '>': case '<':
1166 break;
1167 default:
1168 return 0;
1170 for (cnt = 1; cnt < 7; cnt++)
1171 if (line[cnt] != firstchar)
1172 return 0;
1173 /* line[0] thru line[6] are same as firstchar */
1174 if (firstchar == '=') {
1175 /* divider between ours and theirs? */
1176 if (len != 8 || line[7] != '\n')
1177 return 0;
1178 } else if (len < 8 || !isspace(line[7])) {
1179 /* not divider before ours nor after theirs */
1180 return 0;
1182 return 1;
1185 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1187 struct checkdiff_t *data = priv;
1188 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1189 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1190 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1191 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1192 char *err;
1194 if (line[0] == '+') {
1195 unsigned bad;
1196 data->lineno++;
1197 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1198 data->trailing_blanks_start = 0;
1199 else if (!data->trailing_blanks_start)
1200 data->trailing_blanks_start = data->lineno;
1201 if (is_conflict_marker(line + 1, len - 1)) {
1202 data->status |= 1;
1203 fprintf(data->o->file,
1204 "%s:%d: leftover conflict marker\n",
1205 data->filename, data->lineno);
1207 bad = ws_check(line + 1, len - 1, data->ws_rule);
1208 if (!bad)
1209 return;
1210 data->status |= bad;
1211 err = whitespace_error_string(bad);
1212 fprintf(data->o->file, "%s:%d: %s.\n",
1213 data->filename, data->lineno, err);
1214 free(err);
1215 emit_line(data->o->file, set, reset, line, 1);
1216 ws_check_emit(line + 1, len - 1, data->ws_rule,
1217 data->o->file, set, reset, ws);
1218 } else if (line[0] == ' ') {
1219 data->lineno++;
1220 data->trailing_blanks_start = 0;
1221 } else if (line[0] == '@') {
1222 char *plus = strchr(line, '+');
1223 if (plus)
1224 data->lineno = strtol(plus, NULL, 10) - 1;
1225 else
1226 die("invalid diff");
1227 data->trailing_blanks_start = 0;
1231 static unsigned char *deflate_it(char *data,
1232 unsigned long size,
1233 unsigned long *result_size)
1235 int bound;
1236 unsigned char *deflated;
1237 z_stream stream;
1239 memset(&stream, 0, sizeof(stream));
1240 deflateInit(&stream, zlib_compression_level);
1241 bound = deflateBound(&stream, size);
1242 deflated = xmalloc(bound);
1243 stream.next_out = deflated;
1244 stream.avail_out = bound;
1246 stream.next_in = (unsigned char *)data;
1247 stream.avail_in = size;
1248 while (deflate(&stream, Z_FINISH) == Z_OK)
1249 ; /* nothing */
1250 deflateEnd(&stream);
1251 *result_size = stream.total_out;
1252 return deflated;
1255 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1257 void *cp;
1258 void *delta;
1259 void *deflated;
1260 void *data;
1261 unsigned long orig_size;
1262 unsigned long delta_size;
1263 unsigned long deflate_size;
1264 unsigned long data_size;
1266 /* We could do deflated delta, or we could do just deflated two,
1267 * whichever is smaller.
1269 delta = NULL;
1270 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1271 if (one->size && two->size) {
1272 delta = diff_delta(one->ptr, one->size,
1273 two->ptr, two->size,
1274 &delta_size, deflate_size);
1275 if (delta) {
1276 void *to_free = delta;
1277 orig_size = delta_size;
1278 delta = deflate_it(delta, delta_size, &delta_size);
1279 free(to_free);
1283 if (delta && delta_size < deflate_size) {
1284 fprintf(file, "delta %lu\n", orig_size);
1285 free(deflated);
1286 data = delta;
1287 data_size = delta_size;
1289 else {
1290 fprintf(file, "literal %lu\n", two->size);
1291 free(delta);
1292 data = deflated;
1293 data_size = deflate_size;
1296 /* emit data encoded in base85 */
1297 cp = data;
1298 while (data_size) {
1299 int bytes = (52 < data_size) ? 52 : data_size;
1300 char line[70];
1301 data_size -= bytes;
1302 if (bytes <= 26)
1303 line[0] = bytes + 'A' - 1;
1304 else
1305 line[0] = bytes - 26 + 'a' - 1;
1306 encode_85(line + 1, cp, bytes);
1307 cp = (char *) cp + bytes;
1308 fputs(line, file);
1309 fputc('\n', file);
1311 fprintf(file, "\n");
1312 free(data);
1315 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1317 fprintf(file, "GIT binary patch\n");
1318 emit_binary_diff_body(file, one, two);
1319 emit_binary_diff_body(file, two, one);
1322 static void setup_diff_attr_check(struct git_attr_check *check)
1324 static struct git_attr *attr_diff;
1326 if (!attr_diff) {
1327 attr_diff = git_attr("diff", 4);
1329 check[0].attr = attr_diff;
1332 static void diff_filespec_check_attr(struct diff_filespec *one)
1334 struct git_attr_check attr_diff_check;
1335 int check_from_data = 0;
1337 if (one->checked_attr)
1338 return;
1340 setup_diff_attr_check(&attr_diff_check);
1341 one->is_binary = 0;
1342 one->funcname_pattern_ident = NULL;
1344 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1345 const char *value;
1347 /* binaryness */
1348 value = attr_diff_check.value;
1349 if (ATTR_TRUE(value))
1351 else if (ATTR_FALSE(value))
1352 one->is_binary = 1;
1353 else
1354 check_from_data = 1;
1356 /* funcname pattern ident */
1357 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1359 else
1360 one->funcname_pattern_ident = value;
1363 if (check_from_data) {
1364 if (!one->data && DIFF_FILE_VALID(one))
1365 diff_populate_filespec(one, 0);
1367 if (one->data)
1368 one->is_binary = buffer_is_binary(one->data, one->size);
1372 int diff_filespec_is_binary(struct diff_filespec *one)
1374 diff_filespec_check_attr(one);
1375 return one->is_binary;
1378 static const char *funcname_pattern(const char *ident)
1380 struct funcname_pattern *pp;
1382 for (pp = funcname_pattern_list; pp; pp = pp->next)
1383 if (!strcmp(ident, pp->name))
1384 return pp->pattern;
1385 return NULL;
1388 static struct builtin_funcname_pattern {
1389 const char *name;
1390 const char *pattern;
1391 } builtin_funcname_pattern[] = {
1392 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1393 "new\\|return\\|switch\\|throw\\|while\\)\n"
1394 "^[ ]*\\(\\([ ]*"
1395 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1396 "[ ]*([^;]*\\)$" },
1397 { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1398 "destructor\\|interface\\|implementation\\|"
1399 "initialization\\|finalization\\)[ \t]*.*\\)$"
1400 "\\|"
1401 "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1403 { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1404 { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1405 { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1408 static const char *diff_funcname_pattern(struct diff_filespec *one)
1410 const char *ident, *pattern;
1411 int i;
1413 diff_filespec_check_attr(one);
1414 ident = one->funcname_pattern_ident;
1416 if (!ident)
1418 * If the config file has "funcname.default" defined, that
1419 * regexp is used; otherwise NULL is returned and xemit uses
1420 * the built-in default.
1422 return funcname_pattern("default");
1424 /* Look up custom "funcname.$ident" regexp from config. */
1425 pattern = funcname_pattern(ident);
1426 if (pattern)
1427 return pattern;
1430 * And define built-in fallback patterns here. Note that
1431 * these can be overridden by the user's config settings.
1433 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1434 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1435 return builtin_funcname_pattern[i].pattern;
1437 return NULL;
1440 static void builtin_diff(const char *name_a,
1441 const char *name_b,
1442 struct diff_filespec *one,
1443 struct diff_filespec *two,
1444 const char *xfrm_msg,
1445 struct diff_options *o,
1446 int complete_rewrite)
1448 mmfile_t mf1, mf2;
1449 const char *lbl[2];
1450 char *a_one, *b_two;
1451 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1452 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1454 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1455 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1456 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1457 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1458 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1459 if (lbl[0][0] == '/') {
1460 /* /dev/null */
1461 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1462 if (xfrm_msg && xfrm_msg[0])
1463 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1465 else if (lbl[1][0] == '/') {
1466 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1467 if (xfrm_msg && xfrm_msg[0])
1468 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1470 else {
1471 if (one->mode != two->mode) {
1472 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1473 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1475 if (xfrm_msg && xfrm_msg[0])
1476 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1478 * we do not run diff between different kind
1479 * of objects.
1481 if ((one->mode ^ two->mode) & S_IFMT)
1482 goto free_ab_and_return;
1483 if (complete_rewrite) {
1484 emit_rewrite_diff(name_a, name_b, one, two, o);
1485 o->found_changes = 1;
1486 goto free_ab_and_return;
1490 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1491 die("unable to read files to diff");
1493 if (!DIFF_OPT_TST(o, TEXT) &&
1494 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1495 /* Quite common confusing case */
1496 if (mf1.size == mf2.size &&
1497 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1498 goto free_ab_and_return;
1499 if (DIFF_OPT_TST(o, BINARY))
1500 emit_binary_diff(o->file, &mf1, &mf2);
1501 else
1502 fprintf(o->file, "Binary files %s and %s differ\n",
1503 lbl[0], lbl[1]);
1504 o->found_changes = 1;
1506 else {
1507 /* Crazy xdl interfaces.. */
1508 const char *diffopts = getenv("GIT_DIFF_OPTS");
1509 xpparam_t xpp;
1510 xdemitconf_t xecfg;
1511 xdemitcb_t ecb;
1512 struct emit_callback ecbdata;
1513 const char *funcname_pattern;
1515 funcname_pattern = diff_funcname_pattern(one);
1516 if (!funcname_pattern)
1517 funcname_pattern = diff_funcname_pattern(two);
1519 memset(&xecfg, 0, sizeof(xecfg));
1520 memset(&ecbdata, 0, sizeof(ecbdata));
1521 ecbdata.label_path = lbl;
1522 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1523 ecbdata.found_changesp = &o->found_changes;
1524 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1525 ecbdata.file = o->file;
1526 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1527 xecfg.ctxlen = o->context;
1528 xecfg.flags = XDL_EMIT_FUNCNAMES;
1529 if (funcname_pattern)
1530 xdiff_set_find_func(&xecfg, funcname_pattern);
1531 if (!diffopts)
1533 else if (!prefixcmp(diffopts, "--unified="))
1534 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1535 else if (!prefixcmp(diffopts, "-u"))
1536 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1537 ecb.outf = xdiff_outf;
1538 ecb.priv = &ecbdata;
1539 ecbdata.xm.consume = fn_out_consume;
1540 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1541 ecbdata.diff_words =
1542 xcalloc(1, sizeof(struct diff_words_data));
1543 ecbdata.diff_words->file = o->file;
1545 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1546 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1547 free_diff_words_data(&ecbdata);
1550 free_ab_and_return:
1551 diff_free_filespec_data(one);
1552 diff_free_filespec_data(two);
1553 free(a_one);
1554 free(b_two);
1555 return;
1558 static void builtin_diffstat(const char *name_a, const char *name_b,
1559 struct diff_filespec *one,
1560 struct diff_filespec *two,
1561 struct diffstat_t *diffstat,
1562 struct diff_options *o,
1563 int complete_rewrite)
1565 mmfile_t mf1, mf2;
1566 struct diffstat_file *data;
1568 data = diffstat_add(diffstat, name_a, name_b);
1570 if (!one || !two) {
1571 data->is_unmerged = 1;
1572 return;
1574 if (complete_rewrite) {
1575 diff_populate_filespec(one, 0);
1576 diff_populate_filespec(two, 0);
1577 data->deleted = count_lines(one->data, one->size);
1578 data->added = count_lines(two->data, two->size);
1579 goto free_and_return;
1581 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1582 die("unable to read files to diff");
1584 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1585 data->is_binary = 1;
1586 data->added = mf2.size;
1587 data->deleted = mf1.size;
1588 } else {
1589 /* Crazy xdl interfaces.. */
1590 xpparam_t xpp;
1591 xdemitconf_t xecfg;
1592 xdemitcb_t ecb;
1594 memset(&xecfg, 0, sizeof(xecfg));
1595 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1596 ecb.outf = xdiff_outf;
1597 ecb.priv = diffstat;
1598 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1601 free_and_return:
1602 diff_free_filespec_data(one);
1603 diff_free_filespec_data(two);
1606 static void builtin_checkdiff(const char *name_a, const char *name_b,
1607 const char *attr_path,
1608 struct diff_filespec *one,
1609 struct diff_filespec *two,
1610 struct diff_options *o)
1612 mmfile_t mf1, mf2;
1613 struct checkdiff_t data;
1615 if (!two)
1616 return;
1618 memset(&data, 0, sizeof(data));
1619 data.xm.consume = checkdiff_consume;
1620 data.filename = name_b ? name_b : name_a;
1621 data.lineno = 0;
1622 data.o = o;
1623 data.ws_rule = whitespace_rule(attr_path);
1625 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1626 die("unable to read files to diff");
1629 * All the other codepaths check both sides, but not checking
1630 * the "old" side here is deliberate. We are checking the newly
1631 * introduced changes, and as long as the "new" side is text, we
1632 * can and should check what it introduces.
1634 if (diff_filespec_is_binary(two))
1635 goto free_and_return;
1636 else {
1637 /* Crazy xdl interfaces.. */
1638 xpparam_t xpp;
1639 xdemitconf_t xecfg;
1640 xdemitcb_t ecb;
1642 memset(&xecfg, 0, sizeof(xecfg));
1643 xpp.flags = XDF_NEED_MINIMAL;
1644 ecb.outf = xdiff_outf;
1645 ecb.priv = &data;
1646 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1648 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1649 data.trailing_blanks_start) {
1650 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1651 data.filename, data.trailing_blanks_start);
1652 data.status = 1; /* report errors */
1655 free_and_return:
1656 diff_free_filespec_data(one);
1657 diff_free_filespec_data(two);
1658 if (data.status)
1659 DIFF_OPT_SET(o, CHECK_FAILED);
1662 struct diff_filespec *alloc_filespec(const char *path)
1664 int namelen = strlen(path);
1665 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1667 memset(spec, 0, sizeof(*spec));
1668 spec->path = (char *)(spec + 1);
1669 memcpy(spec->path, path, namelen+1);
1670 spec->count = 1;
1671 return spec;
1674 void free_filespec(struct diff_filespec *spec)
1676 if (!--spec->count) {
1677 diff_free_filespec_data(spec);
1678 free(spec);
1682 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1683 unsigned short mode)
1685 if (mode) {
1686 spec->mode = canon_mode(mode);
1687 hashcpy(spec->sha1, sha1);
1688 spec->sha1_valid = !is_null_sha1(sha1);
1693 * Given a name and sha1 pair, if the index tells us the file in
1694 * the work tree has that object contents, return true, so that
1695 * prepare_temp_file() does not have to inflate and extract.
1697 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1699 struct cache_entry *ce;
1700 struct stat st;
1701 int pos, len;
1703 /* We do not read the cache ourselves here, because the
1704 * benchmark with my previous version that always reads cache
1705 * shows that it makes things worse for diff-tree comparing
1706 * two linux-2.6 kernel trees in an already checked out work
1707 * tree. This is because most diff-tree comparisons deal with
1708 * only a small number of files, while reading the cache is
1709 * expensive for a large project, and its cost outweighs the
1710 * savings we get by not inflating the object to a temporary
1711 * file. Practically, this code only helps when we are used
1712 * by diff-cache --cached, which does read the cache before
1713 * calling us.
1715 if (!active_cache)
1716 return 0;
1718 /* We want to avoid the working directory if our caller
1719 * doesn't need the data in a normal file, this system
1720 * is rather slow with its stat/open/mmap/close syscalls,
1721 * and the object is contained in a pack file. The pack
1722 * is probably already open and will be faster to obtain
1723 * the data through than the working directory. Loose
1724 * objects however would tend to be slower as they need
1725 * to be individually opened and inflated.
1727 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1728 return 0;
1730 len = strlen(name);
1731 pos = cache_name_pos(name, len);
1732 if (pos < 0)
1733 return 0;
1734 ce = active_cache[pos];
1737 * This is not the sha1 we are looking for, or
1738 * unreusable because it is not a regular file.
1740 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1741 return 0;
1744 * If ce matches the file in the work tree, we can reuse it.
1746 if (ce_uptodate(ce) ||
1747 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1748 return 1;
1750 return 0;
1753 static int populate_from_stdin(struct diff_filespec *s)
1755 struct strbuf buf;
1756 size_t size = 0;
1758 strbuf_init(&buf, 0);
1759 if (strbuf_read(&buf, 0, 0) < 0)
1760 return error("error while reading from stdin %s",
1761 strerror(errno));
1763 s->should_munmap = 0;
1764 s->data = strbuf_detach(&buf, &size);
1765 s->size = size;
1766 s->should_free = 1;
1767 return 0;
1770 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1772 int len;
1773 char *data = xmalloc(100);
1774 len = snprintf(data, 100,
1775 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1776 s->data = data;
1777 s->size = len;
1778 s->should_free = 1;
1779 if (size_only) {
1780 s->data = NULL;
1781 free(data);
1783 return 0;
1787 * While doing rename detection and pickaxe operation, we may need to
1788 * grab the data for the blob (or file) for our own in-core comparison.
1789 * diff_filespec has data and size fields for this purpose.
1791 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1793 int err = 0;
1794 if (!DIFF_FILE_VALID(s))
1795 die("internal error: asking to populate invalid file.");
1796 if (S_ISDIR(s->mode))
1797 return -1;
1799 if (s->data)
1800 return 0;
1802 if (size_only && 0 < s->size)
1803 return 0;
1805 if (S_ISGITLINK(s->mode))
1806 return diff_populate_gitlink(s, size_only);
1808 if (!s->sha1_valid ||
1809 reuse_worktree_file(s->path, s->sha1, 0)) {
1810 struct strbuf buf;
1811 struct stat st;
1812 int fd;
1814 if (!strcmp(s->path, "-"))
1815 return populate_from_stdin(s);
1817 if (lstat(s->path, &st) < 0) {
1818 if (errno == ENOENT) {
1819 err_empty:
1820 err = -1;
1821 empty:
1822 s->data = (char *)"";
1823 s->size = 0;
1824 return err;
1827 s->size = xsize_t(st.st_size);
1828 if (!s->size)
1829 goto empty;
1830 if (size_only)
1831 return 0;
1832 if (S_ISLNK(st.st_mode)) {
1833 int ret;
1834 s->data = xmalloc(s->size);
1835 s->should_free = 1;
1836 ret = readlink(s->path, s->data, s->size);
1837 if (ret < 0) {
1838 free(s->data);
1839 goto err_empty;
1841 return 0;
1843 fd = open(s->path, O_RDONLY);
1844 if (fd < 0)
1845 goto err_empty;
1846 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1847 close(fd);
1848 s->should_munmap = 1;
1851 * Convert from working tree format to canonical git format
1853 strbuf_init(&buf, 0);
1854 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1855 size_t size = 0;
1856 munmap(s->data, s->size);
1857 s->should_munmap = 0;
1858 s->data = strbuf_detach(&buf, &size);
1859 s->size = size;
1860 s->should_free = 1;
1863 else {
1864 enum object_type type;
1865 if (size_only)
1866 type = sha1_object_info(s->sha1, &s->size);
1867 else {
1868 s->data = read_sha1_file(s->sha1, &type, &s->size);
1869 s->should_free = 1;
1872 return 0;
1875 void diff_free_filespec_blob(struct diff_filespec *s)
1877 if (s->should_free)
1878 free(s->data);
1879 else if (s->should_munmap)
1880 munmap(s->data, s->size);
1882 if (s->should_free || s->should_munmap) {
1883 s->should_free = s->should_munmap = 0;
1884 s->data = NULL;
1888 void diff_free_filespec_data(struct diff_filespec *s)
1890 diff_free_filespec_blob(s);
1891 free(s->cnt_data);
1892 s->cnt_data = NULL;
1895 static void prep_temp_blob(struct diff_tempfile *temp,
1896 void *blob,
1897 unsigned long size,
1898 const unsigned char *sha1,
1899 int mode)
1901 int fd;
1903 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1904 if (fd < 0)
1905 die("unable to create temp-file: %s", strerror(errno));
1906 if (write_in_full(fd, blob, size) != size)
1907 die("unable to write temp-file");
1908 close(fd);
1909 temp->name = temp->tmp_path;
1910 strcpy(temp->hex, sha1_to_hex(sha1));
1911 temp->hex[40] = 0;
1912 sprintf(temp->mode, "%06o", mode);
1915 static void prepare_temp_file(const char *name,
1916 struct diff_tempfile *temp,
1917 struct diff_filespec *one)
1919 if (!DIFF_FILE_VALID(one)) {
1920 not_a_valid_file:
1921 /* A '-' entry produces this for file-2, and
1922 * a '+' entry produces this for file-1.
1924 temp->name = "/dev/null";
1925 strcpy(temp->hex, ".");
1926 strcpy(temp->mode, ".");
1927 return;
1930 if (!one->sha1_valid ||
1931 reuse_worktree_file(name, one->sha1, 1)) {
1932 struct stat st;
1933 if (lstat(name, &st) < 0) {
1934 if (errno == ENOENT)
1935 goto not_a_valid_file;
1936 die("stat(%s): %s", name, strerror(errno));
1938 if (S_ISLNK(st.st_mode)) {
1939 int ret;
1940 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1941 size_t sz = xsize_t(st.st_size);
1942 if (sizeof(buf) <= st.st_size)
1943 die("symlink too long: %s", name);
1944 ret = readlink(name, buf, sz);
1945 if (ret < 0)
1946 die("readlink(%s)", name);
1947 prep_temp_blob(temp, buf, sz,
1948 (one->sha1_valid ?
1949 one->sha1 : null_sha1),
1950 (one->sha1_valid ?
1951 one->mode : S_IFLNK));
1953 else {
1954 /* we can borrow from the file in the work tree */
1955 temp->name = name;
1956 if (!one->sha1_valid)
1957 strcpy(temp->hex, sha1_to_hex(null_sha1));
1958 else
1959 strcpy(temp->hex, sha1_to_hex(one->sha1));
1960 /* Even though we may sometimes borrow the
1961 * contents from the work tree, we always want
1962 * one->mode. mode is trustworthy even when
1963 * !(one->sha1_valid), as long as
1964 * DIFF_FILE_VALID(one).
1966 sprintf(temp->mode, "%06o", one->mode);
1968 return;
1970 else {
1971 if (diff_populate_filespec(one, 0))
1972 die("cannot read data blob for %s", one->path);
1973 prep_temp_blob(temp, one->data, one->size,
1974 one->sha1, one->mode);
1978 static void remove_tempfile(void)
1980 int i;
1982 for (i = 0; i < 2; i++)
1983 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1984 unlink(diff_temp[i].name);
1985 diff_temp[i].name = NULL;
1989 static void remove_tempfile_on_signal(int signo)
1991 remove_tempfile();
1992 signal(SIGINT, SIG_DFL);
1993 raise(signo);
1996 /* An external diff command takes:
1998 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1999 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2002 static void run_external_diff(const char *pgm,
2003 const char *name,
2004 const char *other,
2005 struct diff_filespec *one,
2006 struct diff_filespec *two,
2007 const char *xfrm_msg,
2008 int complete_rewrite)
2010 const char *spawn_arg[10];
2011 struct diff_tempfile *temp = diff_temp;
2012 int retval;
2013 static int atexit_asked = 0;
2014 const char *othername;
2015 const char **arg = &spawn_arg[0];
2017 othername = (other? other : name);
2018 if (one && two) {
2019 prepare_temp_file(name, &temp[0], one);
2020 prepare_temp_file(othername, &temp[1], two);
2021 if (! atexit_asked &&
2022 (temp[0].name == temp[0].tmp_path ||
2023 temp[1].name == temp[1].tmp_path)) {
2024 atexit_asked = 1;
2025 atexit(remove_tempfile);
2027 signal(SIGINT, remove_tempfile_on_signal);
2030 if (one && two) {
2031 *arg++ = pgm;
2032 *arg++ = name;
2033 *arg++ = temp[0].name;
2034 *arg++ = temp[0].hex;
2035 *arg++ = temp[0].mode;
2036 *arg++ = temp[1].name;
2037 *arg++ = temp[1].hex;
2038 *arg++ = temp[1].mode;
2039 if (other) {
2040 *arg++ = other;
2041 *arg++ = xfrm_msg;
2043 } else {
2044 *arg++ = pgm;
2045 *arg++ = name;
2047 *arg = NULL;
2048 fflush(NULL);
2049 retval = run_command_v_opt(spawn_arg, 0);
2050 remove_tempfile();
2051 if (retval) {
2052 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2053 exit(1);
2057 static const char *external_diff_attr(const char *name)
2059 struct git_attr_check attr_diff_check;
2061 if (!name)
2062 return NULL;
2064 setup_diff_attr_check(&attr_diff_check);
2065 if (!git_checkattr(name, 1, &attr_diff_check)) {
2066 const char *value = attr_diff_check.value;
2067 if (!ATTR_TRUE(value) &&
2068 !ATTR_FALSE(value) &&
2069 !ATTR_UNSET(value)) {
2070 struct ll_diff_driver *drv;
2072 for (drv = user_diff; drv; drv = drv->next)
2073 if (!strcmp(drv->name, value))
2074 return drv->cmd;
2077 return NULL;
2080 static void run_diff_cmd(const char *pgm,
2081 const char *name,
2082 const char *other,
2083 const char *attr_path,
2084 struct diff_filespec *one,
2085 struct diff_filespec *two,
2086 const char *xfrm_msg,
2087 struct diff_options *o,
2088 int complete_rewrite)
2090 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2091 pgm = NULL;
2092 else {
2093 const char *cmd = external_diff_attr(attr_path);
2094 if (cmd)
2095 pgm = cmd;
2098 if (pgm) {
2099 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2100 complete_rewrite);
2101 return;
2103 if (one && two)
2104 builtin_diff(name, other ? other : name,
2105 one, two, xfrm_msg, o, complete_rewrite);
2106 else
2107 fprintf(o->file, "* Unmerged path %s\n", name);
2110 static void diff_fill_sha1_info(struct diff_filespec *one)
2112 if (DIFF_FILE_VALID(one)) {
2113 if (!one->sha1_valid) {
2114 struct stat st;
2115 if (!strcmp(one->path, "-")) {
2116 hashcpy(one->sha1, null_sha1);
2117 return;
2119 if (lstat(one->path, &st) < 0)
2120 die("stat %s", one->path);
2121 if (index_path(one->sha1, one->path, &st, 0))
2122 die("cannot hash %s\n", one->path);
2125 else
2126 hashclr(one->sha1);
2129 static int similarity_index(struct diff_filepair *p)
2131 return p->score * 100 / MAX_SCORE;
2134 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2136 /* Strip the prefix but do not molest /dev/null and absolute paths */
2137 if (*namep && **namep != '/')
2138 *namep += prefix_length;
2139 if (*otherp && **otherp != '/')
2140 *otherp += prefix_length;
2143 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2145 const char *pgm = external_diff();
2146 struct strbuf msg;
2147 char *xfrm_msg;
2148 struct diff_filespec *one = p->one;
2149 struct diff_filespec *two = p->two;
2150 const char *name;
2151 const char *other;
2152 const char *attr_path;
2153 int complete_rewrite = 0;
2155 name = p->one->path;
2156 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2157 attr_path = name;
2158 if (o->prefix_length)
2159 strip_prefix(o->prefix_length, &name, &other);
2161 if (DIFF_PAIR_UNMERGED(p)) {
2162 run_diff_cmd(pgm, name, NULL, attr_path,
2163 NULL, NULL, NULL, o, 0);
2164 return;
2167 diff_fill_sha1_info(one);
2168 diff_fill_sha1_info(two);
2170 strbuf_init(&msg, PATH_MAX * 2 + 300);
2171 switch (p->status) {
2172 case DIFF_STATUS_COPIED:
2173 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2174 strbuf_addstr(&msg, "\ncopy from ");
2175 quote_c_style(name, &msg, NULL, 0);
2176 strbuf_addstr(&msg, "\ncopy to ");
2177 quote_c_style(other, &msg, NULL, 0);
2178 strbuf_addch(&msg, '\n');
2179 break;
2180 case DIFF_STATUS_RENAMED:
2181 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2182 strbuf_addstr(&msg, "\nrename from ");
2183 quote_c_style(name, &msg, NULL, 0);
2184 strbuf_addstr(&msg, "\nrename to ");
2185 quote_c_style(other, &msg, NULL, 0);
2186 strbuf_addch(&msg, '\n');
2187 break;
2188 case DIFF_STATUS_MODIFIED:
2189 if (p->score) {
2190 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2191 similarity_index(p));
2192 complete_rewrite = 1;
2193 break;
2195 /* fallthru */
2196 default:
2197 /* nothing */
2201 if (hashcmp(one->sha1, two->sha1)) {
2202 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2204 if (DIFF_OPT_TST(o, BINARY)) {
2205 mmfile_t mf;
2206 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2207 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2208 abbrev = 40;
2210 strbuf_addf(&msg, "index %.*s..%.*s",
2211 abbrev, sha1_to_hex(one->sha1),
2212 abbrev, sha1_to_hex(two->sha1));
2213 if (one->mode == two->mode)
2214 strbuf_addf(&msg, " %06o", one->mode);
2215 strbuf_addch(&msg, '\n');
2218 if (msg.len)
2219 strbuf_setlen(&msg, msg.len - 1);
2220 xfrm_msg = msg.len ? msg.buf : NULL;
2222 if (!pgm &&
2223 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2224 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2225 /* a filepair that changes between file and symlink
2226 * needs to be split into deletion and creation.
2228 struct diff_filespec *null = alloc_filespec(two->path);
2229 run_diff_cmd(NULL, name, other, attr_path,
2230 one, null, xfrm_msg, o, 0);
2231 free(null);
2232 null = alloc_filespec(one->path);
2233 run_diff_cmd(NULL, name, other, attr_path,
2234 null, two, xfrm_msg, o, 0);
2235 free(null);
2237 else
2238 run_diff_cmd(pgm, name, other, attr_path,
2239 one, two, xfrm_msg, o, complete_rewrite);
2241 strbuf_release(&msg);
2244 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2245 struct diffstat_t *diffstat)
2247 const char *name;
2248 const char *other;
2249 int complete_rewrite = 0;
2251 if (DIFF_PAIR_UNMERGED(p)) {
2252 /* unmerged */
2253 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2254 return;
2257 name = p->one->path;
2258 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2260 if (o->prefix_length)
2261 strip_prefix(o->prefix_length, &name, &other);
2263 diff_fill_sha1_info(p->one);
2264 diff_fill_sha1_info(p->two);
2266 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2267 complete_rewrite = 1;
2268 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2271 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2273 const char *name;
2274 const char *other;
2275 const char *attr_path;
2277 if (DIFF_PAIR_UNMERGED(p)) {
2278 /* unmerged */
2279 return;
2282 name = p->one->path;
2283 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2284 attr_path = other ? other : name;
2286 if (o->prefix_length)
2287 strip_prefix(o->prefix_length, &name, &other);
2289 diff_fill_sha1_info(p->one);
2290 diff_fill_sha1_info(p->two);
2292 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2295 void diff_setup(struct diff_options *options)
2297 memset(options, 0, sizeof(*options));
2299 options->file = stdout;
2301 options->line_termination = '\n';
2302 options->break_opt = -1;
2303 options->rename_limit = -1;
2304 options->dirstat_percent = 3;
2305 options->context = 3;
2307 options->change = diff_change;
2308 options->add_remove = diff_addremove;
2309 if (diff_use_color_default > 0)
2310 DIFF_OPT_SET(options, COLOR_DIFF);
2311 else
2312 DIFF_OPT_CLR(options, COLOR_DIFF);
2313 options->detect_rename = diff_detect_rename_default;
2315 options->a_prefix = "a/";
2316 options->b_prefix = "b/";
2319 int diff_setup_done(struct diff_options *options)
2321 int count = 0;
2323 if (options->output_format & DIFF_FORMAT_NAME)
2324 count++;
2325 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2326 count++;
2327 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2328 count++;
2329 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2330 count++;
2331 if (count > 1)
2332 die("--name-only, --name-status, --check and -s are mutually exclusive");
2334 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2335 options->detect_rename = DIFF_DETECT_COPY;
2337 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2338 options->prefix = NULL;
2339 if (options->prefix)
2340 options->prefix_length = strlen(options->prefix);
2341 else
2342 options->prefix_length = 0;
2344 if (options->output_format & (DIFF_FORMAT_NAME |
2345 DIFF_FORMAT_NAME_STATUS |
2346 DIFF_FORMAT_CHECKDIFF |
2347 DIFF_FORMAT_NO_OUTPUT))
2348 options->output_format &= ~(DIFF_FORMAT_RAW |
2349 DIFF_FORMAT_NUMSTAT |
2350 DIFF_FORMAT_DIFFSTAT |
2351 DIFF_FORMAT_SHORTSTAT |
2352 DIFF_FORMAT_DIRSTAT |
2353 DIFF_FORMAT_SUMMARY |
2354 DIFF_FORMAT_PATCH);
2357 * These cases always need recursive; we do not drop caller-supplied
2358 * recursive bits for other formats here.
2360 if (options->output_format & (DIFF_FORMAT_PATCH |
2361 DIFF_FORMAT_NUMSTAT |
2362 DIFF_FORMAT_DIFFSTAT |
2363 DIFF_FORMAT_SHORTSTAT |
2364 DIFF_FORMAT_DIRSTAT |
2365 DIFF_FORMAT_SUMMARY |
2366 DIFF_FORMAT_CHECKDIFF))
2367 DIFF_OPT_SET(options, RECURSIVE);
2369 * Also pickaxe would not work very well if you do not say recursive
2371 if (options->pickaxe)
2372 DIFF_OPT_SET(options, RECURSIVE);
2374 if (options->detect_rename && options->rename_limit < 0)
2375 options->rename_limit = diff_rename_limit_default;
2376 if (options->setup & DIFF_SETUP_USE_CACHE) {
2377 if (!active_cache)
2378 /* read-cache does not die even when it fails
2379 * so it is safe for us to do this here. Also
2380 * it does not smudge active_cache or active_nr
2381 * when it fails, so we do not have to worry about
2382 * cleaning it up ourselves either.
2384 read_cache();
2386 if (options->abbrev <= 0 || 40 < options->abbrev)
2387 options->abbrev = 40; /* full */
2390 * It does not make sense to show the first hit we happened
2391 * to have found. It does not make sense not to return with
2392 * exit code in such a case either.
2394 if (DIFF_OPT_TST(options, QUIET)) {
2395 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2396 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2400 * If we postprocess in diffcore, we cannot simply return
2401 * upon the first hit. We need to run diff as usual.
2403 if (options->pickaxe || options->filter)
2404 DIFF_OPT_CLR(options, QUIET);
2406 return 0;
2409 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2411 char c, *eq;
2412 int len;
2414 if (*arg != '-')
2415 return 0;
2416 c = *++arg;
2417 if (!c)
2418 return 0;
2419 if (c == arg_short) {
2420 c = *++arg;
2421 if (!c)
2422 return 1;
2423 if (val && isdigit(c)) {
2424 char *end;
2425 int n = strtoul(arg, &end, 10);
2426 if (*end)
2427 return 0;
2428 *val = n;
2429 return 1;
2431 return 0;
2433 if (c != '-')
2434 return 0;
2435 arg++;
2436 eq = strchr(arg, '=');
2437 if (eq)
2438 len = eq - arg;
2439 else
2440 len = strlen(arg);
2441 if (!len || strncmp(arg, arg_long, len))
2442 return 0;
2443 if (eq) {
2444 int n;
2445 char *end;
2446 if (!isdigit(*++eq))
2447 return 0;
2448 n = strtoul(eq, &end, 10);
2449 if (*end)
2450 return 0;
2451 *val = n;
2453 return 1;
2456 static int diff_scoreopt_parse(const char *opt);
2458 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2460 const char *arg = av[0];
2462 /* Output format options */
2463 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2464 options->output_format |= DIFF_FORMAT_PATCH;
2465 else if (opt_arg(arg, 'U', "unified", &options->context))
2466 options->output_format |= DIFF_FORMAT_PATCH;
2467 else if (!strcmp(arg, "--raw"))
2468 options->output_format |= DIFF_FORMAT_RAW;
2469 else if (!strcmp(arg, "--patch-with-raw"))
2470 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2471 else if (!strcmp(arg, "--numstat"))
2472 options->output_format |= DIFF_FORMAT_NUMSTAT;
2473 else if (!strcmp(arg, "--shortstat"))
2474 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2475 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2476 options->output_format |= DIFF_FORMAT_DIRSTAT;
2477 else if (!strcmp(arg, "--cumulative"))
2478 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2479 else if (!strcmp(arg, "--check"))
2480 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2481 else if (!strcmp(arg, "--summary"))
2482 options->output_format |= DIFF_FORMAT_SUMMARY;
2483 else if (!strcmp(arg, "--patch-with-stat"))
2484 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2485 else if (!strcmp(arg, "--name-only"))
2486 options->output_format |= DIFF_FORMAT_NAME;
2487 else if (!strcmp(arg, "--name-status"))
2488 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2489 else if (!strcmp(arg, "-s"))
2490 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2491 else if (!prefixcmp(arg, "--stat")) {
2492 char *end;
2493 int width = options->stat_width;
2494 int name_width = options->stat_name_width;
2495 arg += 6;
2496 end = (char *)arg;
2498 switch (*arg) {
2499 case '-':
2500 if (!prefixcmp(arg, "-width="))
2501 width = strtoul(arg + 7, &end, 10);
2502 else if (!prefixcmp(arg, "-name-width="))
2503 name_width = strtoul(arg + 12, &end, 10);
2504 break;
2505 case '=':
2506 width = strtoul(arg+1, &end, 10);
2507 if (*end == ',')
2508 name_width = strtoul(end+1, &end, 10);
2511 /* Important! This checks all the error cases! */
2512 if (*end)
2513 return 0;
2514 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2515 options->stat_name_width = name_width;
2516 options->stat_width = width;
2519 /* renames options */
2520 else if (!prefixcmp(arg, "-B")) {
2521 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2522 return -1;
2524 else if (!prefixcmp(arg, "-M")) {
2525 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2526 return -1;
2527 options->detect_rename = DIFF_DETECT_RENAME;
2529 else if (!prefixcmp(arg, "-C")) {
2530 if (options->detect_rename == DIFF_DETECT_COPY)
2531 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2532 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2533 return -1;
2534 options->detect_rename = DIFF_DETECT_COPY;
2536 else if (!strcmp(arg, "--no-renames"))
2537 options->detect_rename = 0;
2538 else if (!strcmp(arg, "--relative"))
2539 DIFF_OPT_SET(options, RELATIVE_NAME);
2540 else if (!prefixcmp(arg, "--relative=")) {
2541 DIFF_OPT_SET(options, RELATIVE_NAME);
2542 options->prefix = arg + 11;
2545 /* xdiff options */
2546 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2547 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2548 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2549 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2550 else if (!strcmp(arg, "--ignore-space-at-eol"))
2551 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2553 /* flags options */
2554 else if (!strcmp(arg, "--binary")) {
2555 options->output_format |= DIFF_FORMAT_PATCH;
2556 DIFF_OPT_SET(options, BINARY);
2558 else if (!strcmp(arg, "--full-index"))
2559 DIFF_OPT_SET(options, FULL_INDEX);
2560 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2561 DIFF_OPT_SET(options, TEXT);
2562 else if (!strcmp(arg, "-R"))
2563 DIFF_OPT_SET(options, REVERSE_DIFF);
2564 else if (!strcmp(arg, "--find-copies-harder"))
2565 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2566 else if (!strcmp(arg, "--follow"))
2567 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2568 else if (!strcmp(arg, "--color"))
2569 DIFF_OPT_SET(options, COLOR_DIFF);
2570 else if (!strcmp(arg, "--no-color"))
2571 DIFF_OPT_CLR(options, COLOR_DIFF);
2572 else if (!strcmp(arg, "--color-words"))
2573 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2574 else if (!strcmp(arg, "--exit-code"))
2575 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2576 else if (!strcmp(arg, "--quiet"))
2577 DIFF_OPT_SET(options, QUIET);
2578 else if (!strcmp(arg, "--ext-diff"))
2579 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2580 else if (!strcmp(arg, "--no-ext-diff"))
2581 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2582 else if (!strcmp(arg, "--ignore-submodules"))
2583 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2585 /* misc options */
2586 else if (!strcmp(arg, "-z"))
2587 options->line_termination = 0;
2588 else if (!prefixcmp(arg, "-l"))
2589 options->rename_limit = strtoul(arg+2, NULL, 10);
2590 else if (!prefixcmp(arg, "-S"))
2591 options->pickaxe = arg + 2;
2592 else if (!strcmp(arg, "--pickaxe-all"))
2593 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2594 else if (!strcmp(arg, "--pickaxe-regex"))
2595 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2596 else if (!prefixcmp(arg, "-O"))
2597 options->orderfile = arg + 2;
2598 else if (!prefixcmp(arg, "--diff-filter="))
2599 options->filter = arg + 14;
2600 else if (!strcmp(arg, "--abbrev"))
2601 options->abbrev = DEFAULT_ABBREV;
2602 else if (!prefixcmp(arg, "--abbrev=")) {
2603 options->abbrev = strtoul(arg + 9, NULL, 10);
2604 if (options->abbrev < MINIMUM_ABBREV)
2605 options->abbrev = MINIMUM_ABBREV;
2606 else if (40 < options->abbrev)
2607 options->abbrev = 40;
2609 else if (!prefixcmp(arg, "--src-prefix="))
2610 options->a_prefix = arg + 13;
2611 else if (!prefixcmp(arg, "--dst-prefix="))
2612 options->b_prefix = arg + 13;
2613 else if (!strcmp(arg, "--no-prefix"))
2614 options->a_prefix = options->b_prefix = "";
2615 else if (!prefixcmp(arg, "--output=")) {
2616 options->file = fopen(arg + strlen("--output="), "w");
2617 options->close_file = 1;
2618 } else
2619 return 0;
2620 return 1;
2623 static int parse_num(const char **cp_p)
2625 unsigned long num, scale;
2626 int ch, dot;
2627 const char *cp = *cp_p;
2629 num = 0;
2630 scale = 1;
2631 dot = 0;
2632 for(;;) {
2633 ch = *cp;
2634 if ( !dot && ch == '.' ) {
2635 scale = 1;
2636 dot = 1;
2637 } else if ( ch == '%' ) {
2638 scale = dot ? scale*100 : 100;
2639 cp++; /* % is always at the end */
2640 break;
2641 } else if ( ch >= '0' && ch <= '9' ) {
2642 if ( scale < 100000 ) {
2643 scale *= 10;
2644 num = (num*10) + (ch-'0');
2646 } else {
2647 break;
2649 cp++;
2651 *cp_p = cp;
2653 /* user says num divided by scale and we say internally that
2654 * is MAX_SCORE * num / scale.
2656 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2659 static int diff_scoreopt_parse(const char *opt)
2661 int opt1, opt2, cmd;
2663 if (*opt++ != '-')
2664 return -1;
2665 cmd = *opt++;
2666 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2667 return -1; /* that is not a -M, -C nor -B option */
2669 opt1 = parse_num(&opt);
2670 if (cmd != 'B')
2671 opt2 = 0;
2672 else {
2673 if (*opt == 0)
2674 opt2 = 0;
2675 else if (*opt != '/')
2676 return -1; /* we expect -B80/99 or -B80 */
2677 else {
2678 opt++;
2679 opt2 = parse_num(&opt);
2682 if (*opt != 0)
2683 return -1;
2684 return opt1 | (opt2 << 16);
2687 struct diff_queue_struct diff_queued_diff;
2689 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2691 if (queue->alloc <= queue->nr) {
2692 queue->alloc = alloc_nr(queue->alloc);
2693 queue->queue = xrealloc(queue->queue,
2694 sizeof(dp) * queue->alloc);
2696 queue->queue[queue->nr++] = dp;
2699 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2700 struct diff_filespec *one,
2701 struct diff_filespec *two)
2703 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2704 dp->one = one;
2705 dp->two = two;
2706 if (queue)
2707 diff_q(queue, dp);
2708 return dp;
2711 void diff_free_filepair(struct diff_filepair *p)
2713 free_filespec(p->one);
2714 free_filespec(p->two);
2715 free(p);
2718 /* This is different from find_unique_abbrev() in that
2719 * it stuffs the result with dots for alignment.
2721 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2723 int abblen;
2724 const char *abbrev;
2725 if (len == 40)
2726 return sha1_to_hex(sha1);
2728 abbrev = find_unique_abbrev(sha1, len);
2729 abblen = strlen(abbrev);
2730 if (abblen < 37) {
2731 static char hex[41];
2732 if (len < abblen && abblen <= len + 2)
2733 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2734 else
2735 sprintf(hex, "%s...", abbrev);
2736 return hex;
2738 return sha1_to_hex(sha1);
2741 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2743 int line_termination = opt->line_termination;
2744 int inter_name_termination = line_termination ? '\t' : '\0';
2746 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2747 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2748 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2749 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2751 if (p->score) {
2752 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2753 inter_name_termination);
2754 } else {
2755 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2758 if (p->status == DIFF_STATUS_COPIED ||
2759 p->status == DIFF_STATUS_RENAMED) {
2760 const char *name_a, *name_b;
2761 name_a = p->one->path;
2762 name_b = p->two->path;
2763 strip_prefix(opt->prefix_length, &name_a, &name_b);
2764 write_name_quoted(name_a, opt->file, inter_name_termination);
2765 write_name_quoted(name_b, opt->file, line_termination);
2766 } else {
2767 const char *name_a, *name_b;
2768 name_a = p->one->mode ? p->one->path : p->two->path;
2769 name_b = NULL;
2770 strip_prefix(opt->prefix_length, &name_a, &name_b);
2771 write_name_quoted(name_a, opt->file, line_termination);
2775 int diff_unmodified_pair(struct diff_filepair *p)
2777 /* This function is written stricter than necessary to support
2778 * the currently implemented transformers, but the idea is to
2779 * let transformers to produce diff_filepairs any way they want,
2780 * and filter and clean them up here before producing the output.
2782 struct diff_filespec *one = p->one, *two = p->two;
2784 if (DIFF_PAIR_UNMERGED(p))
2785 return 0; /* unmerged is interesting */
2787 /* deletion, addition, mode or type change
2788 * and rename are all interesting.
2790 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2791 DIFF_PAIR_MODE_CHANGED(p) ||
2792 strcmp(one->path, two->path))
2793 return 0;
2795 /* both are valid and point at the same path. that is, we are
2796 * dealing with a change.
2798 if (one->sha1_valid && two->sha1_valid &&
2799 !hashcmp(one->sha1, two->sha1))
2800 return 1; /* no change */
2801 if (!one->sha1_valid && !two->sha1_valid)
2802 return 1; /* both look at the same file on the filesystem. */
2803 return 0;
2806 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2808 if (diff_unmodified_pair(p))
2809 return;
2811 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2812 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2813 return; /* no tree diffs in patch format */
2815 run_diff(p, o);
2818 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2819 struct diffstat_t *diffstat)
2821 if (diff_unmodified_pair(p))
2822 return;
2824 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2825 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2826 return; /* no tree diffs in patch format */
2828 run_diffstat(p, o, diffstat);
2831 static void diff_flush_checkdiff(struct diff_filepair *p,
2832 struct diff_options *o)
2834 if (diff_unmodified_pair(p))
2835 return;
2837 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2838 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2839 return; /* no tree diffs in patch format */
2841 run_checkdiff(p, o);
2844 int diff_queue_is_empty(void)
2846 struct diff_queue_struct *q = &diff_queued_diff;
2847 int i;
2848 for (i = 0; i < q->nr; i++)
2849 if (!diff_unmodified_pair(q->queue[i]))
2850 return 0;
2851 return 1;
2854 #if DIFF_DEBUG
2855 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2857 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2858 x, one ? one : "",
2859 s->path,
2860 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2861 s->mode,
2862 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2863 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2864 x, one ? one : "",
2865 s->size, s->xfrm_flags);
2868 void diff_debug_filepair(const struct diff_filepair *p, int i)
2870 diff_debug_filespec(p->one, i, "one");
2871 diff_debug_filespec(p->two, i, "two");
2872 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2873 p->score, p->status ? p->status : '?',
2874 p->one->rename_used, p->broken_pair);
2877 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2879 int i;
2880 if (msg)
2881 fprintf(stderr, "%s\n", msg);
2882 fprintf(stderr, "q->nr = %d\n", q->nr);
2883 for (i = 0; i < q->nr; i++) {
2884 struct diff_filepair *p = q->queue[i];
2885 diff_debug_filepair(p, i);
2888 #endif
2890 static void diff_resolve_rename_copy(void)
2892 int i;
2893 struct diff_filepair *p;
2894 struct diff_queue_struct *q = &diff_queued_diff;
2896 diff_debug_queue("resolve-rename-copy", q);
2898 for (i = 0; i < q->nr; i++) {
2899 p = q->queue[i];
2900 p->status = 0; /* undecided */
2901 if (DIFF_PAIR_UNMERGED(p))
2902 p->status = DIFF_STATUS_UNMERGED;
2903 else if (!DIFF_FILE_VALID(p->one))
2904 p->status = DIFF_STATUS_ADDED;
2905 else if (!DIFF_FILE_VALID(p->two))
2906 p->status = DIFF_STATUS_DELETED;
2907 else if (DIFF_PAIR_TYPE_CHANGED(p))
2908 p->status = DIFF_STATUS_TYPE_CHANGED;
2910 /* from this point on, we are dealing with a pair
2911 * whose both sides are valid and of the same type, i.e.
2912 * either in-place edit or rename/copy edit.
2914 else if (DIFF_PAIR_RENAME(p)) {
2916 * A rename might have re-connected a broken
2917 * pair up, causing the pathnames to be the
2918 * same again. If so, that's not a rename at
2919 * all, just a modification..
2921 * Otherwise, see if this source was used for
2922 * multiple renames, in which case we decrement
2923 * the count, and call it a copy.
2925 if (!strcmp(p->one->path, p->two->path))
2926 p->status = DIFF_STATUS_MODIFIED;
2927 else if (--p->one->rename_used > 0)
2928 p->status = DIFF_STATUS_COPIED;
2929 else
2930 p->status = DIFF_STATUS_RENAMED;
2932 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2933 p->one->mode != p->two->mode ||
2934 is_null_sha1(p->one->sha1))
2935 p->status = DIFF_STATUS_MODIFIED;
2936 else {
2937 /* This is a "no-change" entry and should not
2938 * happen anymore, but prepare for broken callers.
2940 error("feeding unmodified %s to diffcore",
2941 p->one->path);
2942 p->status = DIFF_STATUS_UNKNOWN;
2945 diff_debug_queue("resolve-rename-copy done", q);
2948 static int check_pair_status(struct diff_filepair *p)
2950 switch (p->status) {
2951 case DIFF_STATUS_UNKNOWN:
2952 return 0;
2953 case 0:
2954 die("internal error in diff-resolve-rename-copy");
2955 default:
2956 return 1;
2960 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2962 int fmt = opt->output_format;
2964 if (fmt & DIFF_FORMAT_CHECKDIFF)
2965 diff_flush_checkdiff(p, opt);
2966 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2967 diff_flush_raw(p, opt);
2968 else if (fmt & DIFF_FORMAT_NAME) {
2969 const char *name_a, *name_b;
2970 name_a = p->two->path;
2971 name_b = NULL;
2972 strip_prefix(opt->prefix_length, &name_a, &name_b);
2973 write_name_quoted(name_a, opt->file, opt->line_termination);
2977 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2979 if (fs->mode)
2980 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2981 else
2982 fprintf(file, " %s ", newdelete);
2983 write_name_quoted(fs->path, file, '\n');
2987 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2989 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2990 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2991 show_name ? ' ' : '\n');
2992 if (show_name) {
2993 write_name_quoted(p->two->path, file, '\n');
2998 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3000 char *names = pprint_rename(p->one->path, p->two->path);
3002 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3003 free(names);
3004 show_mode_change(file, p, 0);
3007 static void diff_summary(FILE *file, struct diff_filepair *p)
3009 switch(p->status) {
3010 case DIFF_STATUS_DELETED:
3011 show_file_mode_name(file, "delete", p->one);
3012 break;
3013 case DIFF_STATUS_ADDED:
3014 show_file_mode_name(file, "create", p->two);
3015 break;
3016 case DIFF_STATUS_COPIED:
3017 show_rename_copy(file, "copy", p);
3018 break;
3019 case DIFF_STATUS_RENAMED:
3020 show_rename_copy(file, "rename", p);
3021 break;
3022 default:
3023 if (p->score) {
3024 fputs(" rewrite ", file);
3025 write_name_quoted(p->two->path, file, ' ');
3026 fprintf(file, "(%d%%)\n", similarity_index(p));
3028 show_mode_change(file, p, !p->score);
3029 break;
3033 struct patch_id_t {
3034 struct xdiff_emit_state xm;
3035 SHA_CTX *ctx;
3036 int patchlen;
3039 static int remove_space(char *line, int len)
3041 int i;
3042 char *dst = line;
3043 unsigned char c;
3045 for (i = 0; i < len; i++)
3046 if (!isspace((c = line[i])))
3047 *dst++ = c;
3049 return dst - line;
3052 static void patch_id_consume(void *priv, char *line, unsigned long len)
3054 struct patch_id_t *data = priv;
3055 int new_len;
3057 /* Ignore line numbers when computing the SHA1 of the patch */
3058 if (!prefixcmp(line, "@@ -"))
3059 return;
3061 new_len = remove_space(line, len);
3063 SHA1_Update(data->ctx, line, new_len);
3064 data->patchlen += new_len;
3067 /* returns 0 upon success, and writes result into sha1 */
3068 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3070 struct diff_queue_struct *q = &diff_queued_diff;
3071 int i;
3072 SHA_CTX ctx;
3073 struct patch_id_t data;
3074 char buffer[PATH_MAX * 4 + 20];
3076 SHA1_Init(&ctx);
3077 memset(&data, 0, sizeof(struct patch_id_t));
3078 data.ctx = &ctx;
3079 data.xm.consume = patch_id_consume;
3081 for (i = 0; i < q->nr; i++) {
3082 xpparam_t xpp;
3083 xdemitconf_t xecfg;
3084 xdemitcb_t ecb;
3085 mmfile_t mf1, mf2;
3086 struct diff_filepair *p = q->queue[i];
3087 int len1, len2;
3089 memset(&xecfg, 0, sizeof(xecfg));
3090 if (p->status == 0)
3091 return error("internal diff status error");
3092 if (p->status == DIFF_STATUS_UNKNOWN)
3093 continue;
3094 if (diff_unmodified_pair(p))
3095 continue;
3096 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3097 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3098 continue;
3099 if (DIFF_PAIR_UNMERGED(p))
3100 continue;
3102 diff_fill_sha1_info(p->one);
3103 diff_fill_sha1_info(p->two);
3104 if (fill_mmfile(&mf1, p->one) < 0 ||
3105 fill_mmfile(&mf2, p->two) < 0)
3106 return error("unable to read files to diff");
3108 len1 = remove_space(p->one->path, strlen(p->one->path));
3109 len2 = remove_space(p->two->path, strlen(p->two->path));
3110 if (p->one->mode == 0)
3111 len1 = snprintf(buffer, sizeof(buffer),
3112 "diff--gita/%.*sb/%.*s"
3113 "newfilemode%06o"
3114 "---/dev/null"
3115 "+++b/%.*s",
3116 len1, p->one->path,
3117 len2, p->two->path,
3118 p->two->mode,
3119 len2, p->two->path);
3120 else if (p->two->mode == 0)
3121 len1 = snprintf(buffer, sizeof(buffer),
3122 "diff--gita/%.*sb/%.*s"
3123 "deletedfilemode%06o"
3124 "---a/%.*s"
3125 "+++/dev/null",
3126 len1, p->one->path,
3127 len2, p->two->path,
3128 p->one->mode,
3129 len1, p->one->path);
3130 else
3131 len1 = snprintf(buffer, sizeof(buffer),
3132 "diff--gita/%.*sb/%.*s"
3133 "---a/%.*s"
3134 "+++b/%.*s",
3135 len1, p->one->path,
3136 len2, p->two->path,
3137 len1, p->one->path,
3138 len2, p->two->path);
3139 SHA1_Update(&ctx, buffer, len1);
3141 xpp.flags = XDF_NEED_MINIMAL;
3142 xecfg.ctxlen = 3;
3143 xecfg.flags = XDL_EMIT_FUNCNAMES;
3144 ecb.outf = xdiff_outf;
3145 ecb.priv = &data;
3146 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3149 SHA1_Final(sha1, &ctx);
3150 return 0;
3153 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3155 struct diff_queue_struct *q = &diff_queued_diff;
3156 int i;
3157 int result = diff_get_patch_id(options, sha1);
3159 for (i = 0; i < q->nr; i++)
3160 diff_free_filepair(q->queue[i]);
3162 free(q->queue);
3163 q->queue = NULL;
3164 q->nr = q->alloc = 0;
3166 return result;
3169 static int is_summary_empty(const struct diff_queue_struct *q)
3171 int i;
3173 for (i = 0; i < q->nr; i++) {
3174 const struct diff_filepair *p = q->queue[i];
3176 switch (p->status) {
3177 case DIFF_STATUS_DELETED:
3178 case DIFF_STATUS_ADDED:
3179 case DIFF_STATUS_COPIED:
3180 case DIFF_STATUS_RENAMED:
3181 return 0;
3182 default:
3183 if (p->score)
3184 return 0;
3185 if (p->one->mode && p->two->mode &&
3186 p->one->mode != p->two->mode)
3187 return 0;
3188 break;
3191 return 1;
3194 void diff_flush(struct diff_options *options)
3196 struct diff_queue_struct *q = &diff_queued_diff;
3197 int i, output_format = options->output_format;
3198 int separator = 0;
3201 * Order: raw, stat, summary, patch
3202 * or: name/name-status/checkdiff (other bits clear)
3204 if (!q->nr)
3205 goto free_queue;
3207 if (output_format & (DIFF_FORMAT_RAW |
3208 DIFF_FORMAT_NAME |
3209 DIFF_FORMAT_NAME_STATUS |
3210 DIFF_FORMAT_CHECKDIFF)) {
3211 for (i = 0; i < q->nr; i++) {
3212 struct diff_filepair *p = q->queue[i];
3213 if (check_pair_status(p))
3214 flush_one_pair(p, options);
3216 separator++;
3219 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3220 struct diffstat_t diffstat;
3222 memset(&diffstat, 0, sizeof(struct diffstat_t));
3223 diffstat.xm.consume = diffstat_consume;
3224 for (i = 0; i < q->nr; i++) {
3225 struct diff_filepair *p = q->queue[i];
3226 if (check_pair_status(p))
3227 diff_flush_stat(p, options, &diffstat);
3229 if (output_format & DIFF_FORMAT_NUMSTAT)
3230 show_numstat(&diffstat, options);
3231 if (output_format & DIFF_FORMAT_DIFFSTAT)
3232 show_stats(&diffstat, options);
3233 if (output_format & DIFF_FORMAT_SHORTSTAT)
3234 show_shortstats(&diffstat, options);
3235 free_diffstat_info(&diffstat);
3236 separator++;
3238 if (output_format & DIFF_FORMAT_DIRSTAT)
3239 show_dirstat(options);
3241 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3242 for (i = 0; i < q->nr; i++)
3243 diff_summary(options->file, q->queue[i]);
3244 separator++;
3247 if (output_format & DIFF_FORMAT_PATCH) {
3248 if (separator) {
3249 putc(options->line_termination, options->file);
3250 if (options->stat_sep) {
3251 /* attach patch instead of inline */
3252 fputs(options->stat_sep, options->file);
3256 for (i = 0; i < q->nr; i++) {
3257 struct diff_filepair *p = q->queue[i];
3258 if (check_pair_status(p))
3259 diff_flush_patch(p, options);
3263 if (output_format & DIFF_FORMAT_CALLBACK)
3264 options->format_callback(q, options, options->format_callback_data);
3266 for (i = 0; i < q->nr; i++)
3267 diff_free_filepair(q->queue[i]);
3268 free_queue:
3269 free(q->queue);
3270 q->queue = NULL;
3271 q->nr = q->alloc = 0;
3272 if (options->close_file)
3273 fclose(options->file);
3276 static void diffcore_apply_filter(const char *filter)
3278 int i;
3279 struct diff_queue_struct *q = &diff_queued_diff;
3280 struct diff_queue_struct outq;
3281 outq.queue = NULL;
3282 outq.nr = outq.alloc = 0;
3284 if (!filter)
3285 return;
3287 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3288 int found;
3289 for (i = found = 0; !found && i < q->nr; i++) {
3290 struct diff_filepair *p = q->queue[i];
3291 if (((p->status == DIFF_STATUS_MODIFIED) &&
3292 ((p->score &&
3293 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3294 (!p->score &&
3295 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3296 ((p->status != DIFF_STATUS_MODIFIED) &&
3297 strchr(filter, p->status)))
3298 found++;
3300 if (found)
3301 return;
3303 /* otherwise we will clear the whole queue
3304 * by copying the empty outq at the end of this
3305 * function, but first clear the current entries
3306 * in the queue.
3308 for (i = 0; i < q->nr; i++)
3309 diff_free_filepair(q->queue[i]);
3311 else {
3312 /* Only the matching ones */
3313 for (i = 0; i < q->nr; i++) {
3314 struct diff_filepair *p = q->queue[i];
3316 if (((p->status == DIFF_STATUS_MODIFIED) &&
3317 ((p->score &&
3318 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3319 (!p->score &&
3320 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3321 ((p->status != DIFF_STATUS_MODIFIED) &&
3322 strchr(filter, p->status)))
3323 diff_q(&outq, p);
3324 else
3325 diff_free_filepair(p);
3328 free(q->queue);
3329 *q = outq;
3332 /* Check whether two filespecs with the same mode and size are identical */
3333 static int diff_filespec_is_identical(struct diff_filespec *one,
3334 struct diff_filespec *two)
3336 if (S_ISGITLINK(one->mode))
3337 return 0;
3338 if (diff_populate_filespec(one, 0))
3339 return 0;
3340 if (diff_populate_filespec(two, 0))
3341 return 0;
3342 return !memcmp(one->data, two->data, one->size);
3345 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3347 int i;
3348 struct diff_queue_struct *q = &diff_queued_diff;
3349 struct diff_queue_struct outq;
3350 outq.queue = NULL;
3351 outq.nr = outq.alloc = 0;
3353 for (i = 0; i < q->nr; i++) {
3354 struct diff_filepair *p = q->queue[i];
3357 * 1. Entries that come from stat info dirtyness
3358 * always have both sides (iow, not create/delete),
3359 * one side of the object name is unknown, with
3360 * the same mode and size. Keep the ones that
3361 * do not match these criteria. They have real
3362 * differences.
3364 * 2. At this point, the file is known to be modified,
3365 * with the same mode and size, and the object
3366 * name of one side is unknown. Need to inspect
3367 * the identical contents.
3369 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3370 !DIFF_FILE_VALID(p->two) ||
3371 (p->one->sha1_valid && p->two->sha1_valid) ||
3372 (p->one->mode != p->two->mode) ||
3373 diff_populate_filespec(p->one, 1) ||
3374 diff_populate_filespec(p->two, 1) ||
3375 (p->one->size != p->two->size) ||
3376 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3377 diff_q(&outq, p);
3378 else {
3380 * The caller can subtract 1 from skip_stat_unmatch
3381 * to determine how many paths were dirty only
3382 * due to stat info mismatch.
3384 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3385 diffopt->skip_stat_unmatch++;
3386 diff_free_filepair(p);
3389 free(q->queue);
3390 *q = outq;
3393 void diffcore_std(struct diff_options *options)
3395 if (DIFF_OPT_TST(options, QUIET))
3396 return;
3398 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3399 diffcore_skip_stat_unmatch(options);
3400 if (options->break_opt != -1)
3401 diffcore_break(options->break_opt);
3402 if (options->detect_rename)
3403 diffcore_rename(options);
3404 if (options->break_opt != -1)
3405 diffcore_merge_broken();
3406 if (options->pickaxe)
3407 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3408 if (options->orderfile)
3409 diffcore_order(options->orderfile);
3410 diff_resolve_rename_copy();
3411 diffcore_apply_filter(options->filter);
3413 if (diff_queued_diff.nr)
3414 DIFF_OPT_SET(options, HAS_CHANGES);
3415 else
3416 DIFF_OPT_CLR(options, HAS_CHANGES);
3419 int diff_result_code(struct diff_options *opt, int status)
3421 int result = 0;
3422 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3423 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3424 return status;
3425 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3426 DIFF_OPT_TST(opt, HAS_CHANGES))
3427 result |= 01;
3428 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3429 DIFF_OPT_TST(opt, CHECK_FAILED))
3430 result |= 02;
3431 return result;
3434 void diff_addremove(struct diff_options *options,
3435 int addremove, unsigned mode,
3436 const unsigned char *sha1,
3437 const char *concatpath)
3439 struct diff_filespec *one, *two;
3441 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3442 return;
3444 /* This may look odd, but it is a preparation for
3445 * feeding "there are unchanged files which should
3446 * not produce diffs, but when you are doing copy
3447 * detection you would need them, so here they are"
3448 * entries to the diff-core. They will be prefixed
3449 * with something like '=' or '*' (I haven't decided
3450 * which but should not make any difference).
3451 * Feeding the same new and old to diff_change()
3452 * also has the same effect.
3453 * Before the final output happens, they are pruned after
3454 * merged into rename/copy pairs as appropriate.
3456 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3457 addremove = (addremove == '+' ? '-' :
3458 addremove == '-' ? '+' : addremove);
3460 if (options->prefix &&
3461 strncmp(concatpath, options->prefix, options->prefix_length))
3462 return;
3464 one = alloc_filespec(concatpath);
3465 two = alloc_filespec(concatpath);
3467 if (addremove != '+')
3468 fill_filespec(one, sha1, mode);
3469 if (addremove != '-')
3470 fill_filespec(two, sha1, mode);
3472 diff_queue(&diff_queued_diff, one, two);
3473 DIFF_OPT_SET(options, HAS_CHANGES);
3476 void diff_change(struct diff_options *options,
3477 unsigned old_mode, unsigned new_mode,
3478 const unsigned char *old_sha1,
3479 const unsigned char *new_sha1,
3480 const char *concatpath)
3482 struct diff_filespec *one, *two;
3484 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3485 && S_ISGITLINK(new_mode))
3486 return;
3488 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3489 unsigned tmp;
3490 const unsigned char *tmp_c;
3491 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3492 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3495 if (options->prefix &&
3496 strncmp(concatpath, options->prefix, options->prefix_length))
3497 return;
3499 one = alloc_filespec(concatpath);
3500 two = alloc_filespec(concatpath);
3501 fill_filespec(one, old_sha1, old_mode);
3502 fill_filespec(two, new_sha1, new_mode);
3504 diff_queue(&diff_queued_diff, one, two);
3505 DIFF_OPT_SET(options, HAS_CHANGES);
3508 void diff_unmerge(struct diff_options *options,
3509 const char *path,
3510 unsigned mode, const unsigned char *sha1)
3512 struct diff_filespec *one, *two;
3514 if (options->prefix &&
3515 strncmp(path, options->prefix, options->prefix_length))
3516 return;
3518 one = alloc_filespec(path);
3519 two = alloc_filespec(path);
3520 fill_filespec(one, sha1, mode);
3521 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;