Use remove_path from dir.c instead of own implementation
[git/jnareb-git.git] / diff.c
blob9385a36f1efb1f52250dfa42dbf994f5fd03856d
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 int diff_use_color_default = -1;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
27 static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
38 static int parse_diff_color_slot(const char *var, int ofs)
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
57 static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 const char *cmd;
61 } *user_diff, **user_diff_tail;
64 * Currently there is only "diff.<drivername>.command" variable;
65 * because there are "diff.color.<slot>" variables, we are parsing
66 * this in a bit convoluted way to allow low level diff driver
67 * called "color".
69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
71 const char *name;
72 int namelen;
73 struct ll_diff_driver *drv;
75 name = var + 5;
76 namelen = ep - name;
77 for (drv = user_diff; drv; drv = drv->next)
78 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79 break;
80 if (!drv) {
81 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82 drv->name = xmemdupz(name, namelen);
83 if (!user_diff_tail)
84 user_diff_tail = &user_diff;
85 *user_diff_tail = drv;
86 user_diff_tail = &(drv->next);
89 return git_config_string(&(drv->cmd), var, value);
93 * 'diff.<what>.funcname' attribute can be specified in the configuration
94 * to define a customized regexp to find the beginning of a function to
95 * be used for hunk header lines of "diff -p" style output.
97 static struct funcname_pattern {
98 char *name;
99 char *pattern;
100 struct funcname_pattern *next;
101 } *funcname_pattern_list;
103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
105 const char *name;
106 int namelen;
107 struct funcname_pattern *pp;
109 name = var + 5; /* "diff." */
110 namelen = ep - name;
112 for (pp = funcname_pattern_list; pp; pp = pp->next)
113 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114 break;
115 if (!pp) {
116 pp = xcalloc(1, sizeof(*pp));
117 pp->name = xmemdupz(name, namelen);
118 pp->next = funcname_pattern_list;
119 funcname_pattern_list = pp;
121 free(pp->pattern);
122 pp->pattern = xstrdup(value);
123 return 0;
127 * These are to give UI layer defaults.
128 * The core-level commands such as git-diff-files should
129 * never be affected by the setting of diff.renames
130 * the user happens to have in the configuration file.
132 int git_diff_ui_config(const char *var, const char *value, void *cb)
134 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
135 diff_use_color_default = git_config_colorbool(var, value, -1);
136 return 0;
138 if (!strcmp(var, "diff.renames")) {
139 if (!value)
140 diff_detect_rename_default = DIFF_DETECT_RENAME;
141 else if (!strcasecmp(value, "copies") ||
142 !strcasecmp(value, "copy"))
143 diff_detect_rename_default = DIFF_DETECT_COPY;
144 else if (git_config_bool(var,value))
145 diff_detect_rename_default = DIFF_DETECT_RENAME;
146 return 0;
148 if (!strcmp(var, "diff.autorefreshindex")) {
149 diff_auto_refresh_index = git_config_bool(var, value);
150 return 0;
152 if (!strcmp(var, "diff.external"))
153 return git_config_string(&external_diff_cmd_cfg, var, value);
154 if (!prefixcmp(var, "diff.")) {
155 const char *ep = strrchr(var, '.');
157 if (ep != var + 4 && !strcmp(ep, ".command"))
158 return parse_lldiff_command(var, ep, value);
161 return git_diff_basic_config(var, value, cb);
164 int git_diff_basic_config(const char *var, const char *value, void *cb)
166 if (!strcmp(var, "diff.renamelimit")) {
167 diff_rename_limit_default = git_config_int(var, value);
168 return 0;
171 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
172 int slot = parse_diff_color_slot(var, 11);
173 if (!value)
174 return config_error_nonbool(var);
175 color_parse(value, var, diff_colors[slot]);
176 return 0;
179 if (!prefixcmp(var, "diff.")) {
180 const char *ep = strrchr(var, '.');
181 if (ep != var + 4) {
182 if (!strcmp(ep, ".funcname")) {
183 if (!value)
184 return config_error_nonbool(var);
185 return parse_funcname_pattern(var, ep, value);
190 return git_color_default_config(var, value, cb);
193 static char *quote_two(const char *one, const char *two)
195 int need_one = quote_c_style(one, NULL, NULL, 1);
196 int need_two = quote_c_style(two, NULL, NULL, 1);
197 struct strbuf res;
199 strbuf_init(&res, 0);
200 if (need_one + need_two) {
201 strbuf_addch(&res, '"');
202 quote_c_style(one, &res, NULL, 1);
203 quote_c_style(two, &res, NULL, 1);
204 strbuf_addch(&res, '"');
205 } else {
206 strbuf_addstr(&res, one);
207 strbuf_addstr(&res, two);
209 return strbuf_detach(&res, NULL);
212 static const char *external_diff(void)
214 static const char *external_diff_cmd = NULL;
215 static int done_preparing = 0;
217 if (done_preparing)
218 return external_diff_cmd;
219 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
220 if (!external_diff_cmd)
221 external_diff_cmd = external_diff_cmd_cfg;
222 done_preparing = 1;
223 return external_diff_cmd;
226 static struct diff_tempfile {
227 const char *name; /* filename external diff should read from */
228 char hex[41];
229 char mode[10];
230 char tmp_path[PATH_MAX];
231 } diff_temp[2];
233 static int count_lines(const char *data, int size)
235 int count, ch, completely_empty = 1, nl_just_seen = 0;
236 count = 0;
237 while (0 < size--) {
238 ch = *data++;
239 if (ch == '\n') {
240 count++;
241 nl_just_seen = 1;
242 completely_empty = 0;
244 else {
245 nl_just_seen = 0;
246 completely_empty = 0;
249 if (completely_empty)
250 return 0;
251 if (!nl_just_seen)
252 count++; /* no trailing newline */
253 return count;
256 static void print_line_count(FILE *file, int count)
258 switch (count) {
259 case 0:
260 fprintf(file, "0,0");
261 break;
262 case 1:
263 fprintf(file, "1");
264 break;
265 default:
266 fprintf(file, "1,%d", count);
267 break;
271 static void copy_file_with_prefix(FILE *file,
272 int prefix, const char *data, int size,
273 const char *set, const char *reset)
275 int ch, nl_just_seen = 1;
276 while (0 < size--) {
277 ch = *data++;
278 if (nl_just_seen) {
279 fputs(set, file);
280 putc(prefix, file);
282 if (ch == '\n') {
283 nl_just_seen = 1;
284 fputs(reset, file);
285 } else
286 nl_just_seen = 0;
287 putc(ch, file);
289 if (!nl_just_seen)
290 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
293 static void emit_rewrite_diff(const char *name_a,
294 const char *name_b,
295 struct diff_filespec *one,
296 struct diff_filespec *two,
297 struct diff_options *o)
299 int lc_a, lc_b;
300 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
301 const char *name_a_tab, *name_b_tab;
302 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
303 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
304 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
305 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
306 const char *reset = diff_get_color(color_diff, DIFF_RESET);
307 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
309 name_a += (*name_a == '/');
310 name_b += (*name_b == '/');
311 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
312 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
314 strbuf_reset(&a_name);
315 strbuf_reset(&b_name);
316 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
317 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
319 diff_populate_filespec(one, 0);
320 diff_populate_filespec(two, 0);
321 lc_a = count_lines(one->data, one->size);
322 lc_b = count_lines(two->data, two->size);
323 fprintf(o->file,
324 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
325 metainfo, a_name.buf, name_a_tab, reset,
326 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
327 print_line_count(o->file, lc_a);
328 fprintf(o->file, " +");
329 print_line_count(o->file, lc_b);
330 fprintf(o->file, " @@%s\n", reset);
331 if (lc_a)
332 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
333 if (lc_b)
334 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
339 if (!DIFF_FILE_VALID(one)) {
340 mf->ptr = (char *)""; /* does not matter */
341 mf->size = 0;
342 return 0;
344 else if (diff_populate_filespec(one, 0))
345 return -1;
346 mf->ptr = one->data;
347 mf->size = one->size;
348 return 0;
351 struct diff_words_buffer {
352 mmfile_t text;
353 long alloc;
354 long current; /* output pointer */
355 int suppressed_newline;
358 static void diff_words_append(char *line, unsigned long len,
359 struct diff_words_buffer *buffer)
361 if (buffer->text.size + len > buffer->alloc) {
362 buffer->alloc = (buffer->text.size + len) * 3 / 2;
363 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
365 line++;
366 len--;
367 memcpy(buffer->text.ptr + buffer->text.size, line, len);
368 buffer->text.size += len;
371 struct diff_words_data {
372 struct xdiff_emit_state xm;
373 struct diff_words_buffer minus, plus;
374 FILE *file;
377 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
378 int suppress_newline)
380 const char *ptr;
381 int eol = 0;
383 if (len == 0)
384 return;
386 ptr = buffer->text.ptr + buffer->current;
387 buffer->current += len;
389 if (ptr[len - 1] == '\n') {
390 eol = 1;
391 len--;
394 fputs(diff_get_color(1, color), file);
395 fwrite(ptr, len, 1, file);
396 fputs(diff_get_color(1, DIFF_RESET), file);
398 if (eol) {
399 if (suppress_newline)
400 buffer->suppressed_newline = 1;
401 else
402 putc('\n', file);
406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
408 struct diff_words_data *diff_words = priv;
410 if (diff_words->minus.suppressed_newline) {
411 if (line[0] != '+')
412 putc('\n', diff_words->file);
413 diff_words->minus.suppressed_newline = 0;
416 len--;
417 switch (line[0]) {
418 case '-':
419 print_word(diff_words->file,
420 &diff_words->minus, len, DIFF_FILE_OLD, 1);
421 break;
422 case '+':
423 print_word(diff_words->file,
424 &diff_words->plus, len, DIFF_FILE_NEW, 0);
425 break;
426 case ' ':
427 print_word(diff_words->file,
428 &diff_words->plus, len, DIFF_PLAIN, 0);
429 diff_words->minus.current += len;
430 break;
434 /* this executes the word diff on the accumulated buffers */
435 static void diff_words_show(struct diff_words_data *diff_words)
437 xpparam_t xpp;
438 xdemitconf_t xecfg;
439 xdemitcb_t ecb;
440 mmfile_t minus, plus;
441 int i;
443 memset(&xecfg, 0, sizeof(xecfg));
444 minus.size = diff_words->minus.text.size;
445 minus.ptr = xmalloc(minus.size);
446 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
447 for (i = 0; i < minus.size; i++)
448 if (isspace(minus.ptr[i]))
449 minus.ptr[i] = '\n';
450 diff_words->minus.current = 0;
452 plus.size = diff_words->plus.text.size;
453 plus.ptr = xmalloc(plus.size);
454 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
455 for (i = 0; i < plus.size; i++)
456 if (isspace(plus.ptr[i]))
457 plus.ptr[i] = '\n';
458 diff_words->plus.current = 0;
460 xpp.flags = XDF_NEED_MINIMAL;
461 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
462 ecb.outf = xdiff_outf;
463 ecb.priv = diff_words;
464 diff_words->xm.consume = fn_out_diff_words_aux;
465 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
467 free(minus.ptr);
468 free(plus.ptr);
469 diff_words->minus.text.size = diff_words->plus.text.size = 0;
471 if (diff_words->minus.suppressed_newline) {
472 putc('\n', diff_words->file);
473 diff_words->minus.suppressed_newline = 0;
477 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
479 struct emit_callback {
480 struct xdiff_emit_state xm;
481 int nparents, color_diff;
482 unsigned ws_rule;
483 sane_truncate_fn truncate;
484 const char **label_path;
485 struct diff_words_data *diff_words;
486 int *found_changesp;
487 FILE *file;
490 static void free_diff_words_data(struct emit_callback *ecbdata)
492 if (ecbdata->diff_words) {
493 /* flush buffers */
494 if (ecbdata->diff_words->minus.text.size ||
495 ecbdata->diff_words->plus.text.size)
496 diff_words_show(ecbdata->diff_words);
498 free (ecbdata->diff_words->minus.text.ptr);
499 free (ecbdata->diff_words->plus.text.ptr);
500 free(ecbdata->diff_words);
501 ecbdata->diff_words = NULL;
505 const char *diff_get_color(int diff_use_color, enum color_diff ix)
507 if (diff_use_color)
508 return diff_colors[ix];
509 return "";
512 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
514 int has_trailing_newline, has_trailing_carriage_return;
516 has_trailing_newline = (len > 0 && line[len-1] == '\n');
517 if (has_trailing_newline)
518 len--;
519 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
520 if (has_trailing_carriage_return)
521 len--;
523 fputs(set, file);
524 fwrite(line, len, 1, file);
525 fputs(reset, file);
526 if (has_trailing_carriage_return)
527 fputc('\r', 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 /* This is not really necessary for now because
591 * this codepath only deals with two-way diffs.
593 for (i = 0; i < len && line[i] == '@'; i++)
595 if (2 <= i && i < len && line[i] == ' ') {
596 ecbdata->nparents = i - 1;
597 len = sane_truncate_line(ecbdata, line, len);
598 emit_line(ecbdata->file,
599 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
600 reset, line, len);
601 if (line[len-1] != '\n')
602 putc('\n', ecbdata->file);
603 return;
606 if (len < ecbdata->nparents) {
607 emit_line(ecbdata->file, reset, reset, line, len);
608 return;
611 color = DIFF_PLAIN;
612 if (ecbdata->diff_words && ecbdata->nparents != 1)
613 /* fall back to normal diff */
614 free_diff_words_data(ecbdata);
615 if (ecbdata->diff_words) {
616 if (line[0] == '-') {
617 diff_words_append(line, len,
618 &ecbdata->diff_words->minus);
619 return;
620 } else if (line[0] == '+') {
621 diff_words_append(line, len,
622 &ecbdata->diff_words->plus);
623 return;
625 if (ecbdata->diff_words->minus.text.size ||
626 ecbdata->diff_words->plus.text.size)
627 diff_words_show(ecbdata->diff_words);
628 line++;
629 len--;
630 emit_line(ecbdata->file, plain, reset, line, len);
631 return;
633 for (i = 0; i < ecbdata->nparents && len; i++) {
634 if (line[i] == '-')
635 color = DIFF_FILE_OLD;
636 else if (line[i] == '+')
637 color = DIFF_FILE_NEW;
640 if (color != DIFF_FILE_NEW) {
641 emit_line(ecbdata->file,
642 diff_get_color(ecbdata->color_diff, color),
643 reset, line, len);
644 return;
646 emit_add_line(reset, ecbdata, line, len);
649 static char *pprint_rename(const char *a, const char *b)
651 const char *old = a;
652 const char *new = b;
653 struct strbuf name;
654 int pfx_length, sfx_length;
655 int len_a = strlen(a);
656 int len_b = strlen(b);
657 int a_midlen, b_midlen;
658 int qlen_a = quote_c_style(a, NULL, NULL, 0);
659 int qlen_b = quote_c_style(b, NULL, NULL, 0);
661 strbuf_init(&name, 0);
662 if (qlen_a || qlen_b) {
663 quote_c_style(a, &name, NULL, 0);
664 strbuf_addstr(&name, " => ");
665 quote_c_style(b, &name, NULL, 0);
666 return strbuf_detach(&name, NULL);
669 /* Find common prefix */
670 pfx_length = 0;
671 while (*old && *new && *old == *new) {
672 if (*old == '/')
673 pfx_length = old - a + 1;
674 old++;
675 new++;
678 /* Find common suffix */
679 old = a + len_a;
680 new = b + len_b;
681 sfx_length = 0;
682 while (a <= old && b <= new && *old == *new) {
683 if (*old == '/')
684 sfx_length = len_a - (old - a);
685 old--;
686 new--;
690 * pfx{mid-a => mid-b}sfx
691 * {pfx-a => pfx-b}sfx
692 * pfx{sfx-a => sfx-b}
693 * name-a => name-b
695 a_midlen = len_a - pfx_length - sfx_length;
696 b_midlen = len_b - pfx_length - sfx_length;
697 if (a_midlen < 0)
698 a_midlen = 0;
699 if (b_midlen < 0)
700 b_midlen = 0;
702 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
703 if (pfx_length + sfx_length) {
704 strbuf_add(&name, a, pfx_length);
705 strbuf_addch(&name, '{');
707 strbuf_add(&name, a + pfx_length, a_midlen);
708 strbuf_addstr(&name, " => ");
709 strbuf_add(&name, b + pfx_length, b_midlen);
710 if (pfx_length + sfx_length) {
711 strbuf_addch(&name, '}');
712 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
714 return strbuf_detach(&name, NULL);
717 struct diffstat_t {
718 struct xdiff_emit_state xm;
720 int nr;
721 int alloc;
722 struct diffstat_file {
723 char *from_name;
724 char *name;
725 char *print_name;
726 unsigned is_unmerged:1;
727 unsigned is_binary:1;
728 unsigned is_renamed:1;
729 unsigned int added, deleted;
730 } **files;
733 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
734 const char *name_a,
735 const char *name_b)
737 struct diffstat_file *x;
738 x = xcalloc(sizeof (*x), 1);
739 if (diffstat->nr == diffstat->alloc) {
740 diffstat->alloc = alloc_nr(diffstat->alloc);
741 diffstat->files = xrealloc(diffstat->files,
742 diffstat->alloc * sizeof(x));
744 diffstat->files[diffstat->nr++] = x;
745 if (name_b) {
746 x->from_name = xstrdup(name_a);
747 x->name = xstrdup(name_b);
748 x->is_renamed = 1;
750 else {
751 x->from_name = NULL;
752 x->name = xstrdup(name_a);
754 return x;
757 static void diffstat_consume(void *priv, char *line, unsigned long len)
759 struct diffstat_t *diffstat = priv;
760 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
762 if (line[0] == '+')
763 x->added++;
764 else if (line[0] == '-')
765 x->deleted++;
768 const char mime_boundary_leader[] = "------------";
770 static int scale_linear(int it, int width, int max_change)
773 * make sure that at least one '-' is printed if there were deletions,
774 * and likewise for '+'.
776 if (max_change < 2)
777 return it;
778 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
781 static void show_name(FILE *file,
782 const char *prefix, const char *name, int len,
783 const char *reset, const char *set)
785 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
788 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
790 if (cnt <= 0)
791 return;
792 fprintf(file, "%s", set);
793 while (cnt--)
794 putc(ch, file);
795 fprintf(file, "%s", reset);
798 static void fill_print_name(struct diffstat_file *file)
800 char *pname;
802 if (file->print_name)
803 return;
805 if (!file->is_renamed) {
806 struct strbuf buf;
807 strbuf_init(&buf, 0);
808 if (quote_c_style(file->name, &buf, NULL, 0)) {
809 pname = strbuf_detach(&buf, NULL);
810 } else {
811 pname = file->name;
812 strbuf_release(&buf);
814 } else {
815 pname = pprint_rename(file->from_name, file->name);
817 file->print_name = pname;
820 static void show_stats(struct diffstat_t* data, struct diff_options *options)
822 int i, len, add, del, total, adds = 0, dels = 0;
823 int max_change = 0, max_len = 0;
824 int total_files = data->nr;
825 int width, name_width;
826 const char *reset, *set, *add_c, *del_c;
828 if (data->nr == 0)
829 return;
831 width = options->stat_width ? options->stat_width : 80;
832 name_width = options->stat_name_width ? options->stat_name_width : 50;
834 /* Sanity: give at least 5 columns to the graph,
835 * but leave at least 10 columns for the name.
837 if (width < 25)
838 width = 25;
839 if (name_width < 10)
840 name_width = 10;
841 else if (width < name_width + 15)
842 name_width = width - 15;
844 /* Find the longest filename and max number of changes */
845 reset = diff_get_color_opt(options, DIFF_RESET);
846 set = diff_get_color_opt(options, DIFF_PLAIN);
847 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
848 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
850 for (i = 0; i < data->nr; i++) {
851 struct diffstat_file *file = data->files[i];
852 int change = file->added + file->deleted;
853 fill_print_name(file);
854 len = strlen(file->print_name);
855 if (max_len < len)
856 max_len = len;
858 if (file->is_binary || file->is_unmerged)
859 continue;
860 if (max_change < change)
861 max_change = change;
864 /* Compute the width of the graph part;
865 * 10 is for one blank at the beginning of the line plus
866 * " | count " between the name and the graph.
868 * From here on, name_width is the width of the name area,
869 * and width is the width of the graph area.
871 name_width = (name_width < max_len) ? name_width : max_len;
872 if (width < (name_width + 10) + max_change)
873 width = width - (name_width + 10);
874 else
875 width = max_change;
877 for (i = 0; i < data->nr; i++) {
878 const char *prefix = "";
879 char *name = data->files[i]->print_name;
880 int added = data->files[i]->added;
881 int deleted = data->files[i]->deleted;
882 int name_len;
885 * "scale" the filename
887 len = name_width;
888 name_len = strlen(name);
889 if (name_width < name_len) {
890 char *slash;
891 prefix = "...";
892 len -= 3;
893 name += name_len - len;
894 slash = strchr(name, '/');
895 if (slash)
896 name = slash;
899 if (data->files[i]->is_binary) {
900 show_name(options->file, prefix, name, len, reset, set);
901 fprintf(options->file, " Bin ");
902 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
903 fprintf(options->file, " -> ");
904 fprintf(options->file, "%s%d%s", add_c, added, reset);
905 fprintf(options->file, " bytes");
906 fprintf(options->file, "\n");
907 continue;
909 else if (data->files[i]->is_unmerged) {
910 show_name(options->file, prefix, name, len, reset, set);
911 fprintf(options->file, " Unmerged\n");
912 continue;
914 else if (!data->files[i]->is_renamed &&
915 (added + deleted == 0)) {
916 total_files--;
917 continue;
921 * scale the add/delete
923 add = added;
924 del = deleted;
925 total = add + del;
926 adds += add;
927 dels += del;
929 if (width <= max_change) {
930 add = scale_linear(add, width, max_change);
931 del = scale_linear(del, width, max_change);
932 total = add + del;
934 show_name(options->file, prefix, name, len, reset, set);
935 fprintf(options->file, "%5d%s", added + deleted,
936 added + deleted ? " " : "");
937 show_graph(options->file, '+', add, add_c, reset);
938 show_graph(options->file, '-', del, del_c, reset);
939 fprintf(options->file, "\n");
941 fprintf(options->file,
942 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
943 set, total_files, adds, dels, reset);
946 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
948 int i, adds = 0, dels = 0, total_files = data->nr;
950 if (data->nr == 0)
951 return;
953 for (i = 0; i < data->nr; i++) {
954 if (!data->files[i]->is_binary &&
955 !data->files[i]->is_unmerged) {
956 int added = data->files[i]->added;
957 int deleted= data->files[i]->deleted;
958 if (!data->files[i]->is_renamed &&
959 (added + deleted == 0)) {
960 total_files--;
961 } else {
962 adds += added;
963 dels += deleted;
967 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
968 total_files, adds, dels);
971 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
973 int i;
975 if (data->nr == 0)
976 return;
978 for (i = 0; i < data->nr; i++) {
979 struct diffstat_file *file = data->files[i];
981 if (file->is_binary)
982 fprintf(options->file, "-\t-\t");
983 else
984 fprintf(options->file,
985 "%d\t%d\t", file->added, file->deleted);
986 if (options->line_termination) {
987 fill_print_name(file);
988 if (!file->is_renamed)
989 write_name_quoted(file->name, options->file,
990 options->line_termination);
991 else {
992 fputs(file->print_name, options->file);
993 putc(options->line_termination, options->file);
995 } else {
996 if (file->is_renamed) {
997 putc('\0', options->file);
998 write_name_quoted(file->from_name, options->file, '\0');
1000 write_name_quoted(file->name, options->file, '\0');
1005 struct dirstat_file {
1006 const char *name;
1007 unsigned long changed;
1010 struct dirstat_dir {
1011 struct dirstat_file *files;
1012 int alloc, nr, percent, cumulative;
1015 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1017 unsigned long this_dir = 0;
1018 unsigned int sources = 0;
1020 while (dir->nr) {
1021 struct dirstat_file *f = dir->files;
1022 int namelen = strlen(f->name);
1023 unsigned long this;
1024 char *slash;
1026 if (namelen < baselen)
1027 break;
1028 if (memcmp(f->name, base, baselen))
1029 break;
1030 slash = strchr(f->name + baselen, '/');
1031 if (slash) {
1032 int newbaselen = slash + 1 - f->name;
1033 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1034 sources++;
1035 } else {
1036 this = f->changed;
1037 dir->files++;
1038 dir->nr--;
1039 sources += 2;
1041 this_dir += this;
1045 * We don't report dirstat's for
1046 * - the top level
1047 * - or cases where everything came from a single directory
1048 * under this directory (sources == 1).
1050 if (baselen && sources != 1) {
1051 int permille = this_dir * 1000 / changed;
1052 if (permille) {
1053 int percent = permille / 10;
1054 if (percent >= dir->percent) {
1055 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1056 if (!dir->cumulative)
1057 return 0;
1061 return this_dir;
1064 static int dirstat_compare(const void *_a, const void *_b)
1066 const struct dirstat_file *a = _a;
1067 const struct dirstat_file *b = _b;
1068 return strcmp(a->name, b->name);
1071 static void show_dirstat(struct diff_options *options)
1073 int i;
1074 unsigned long changed;
1075 struct dirstat_dir dir;
1076 struct diff_queue_struct *q = &diff_queued_diff;
1078 dir.files = NULL;
1079 dir.alloc = 0;
1080 dir.nr = 0;
1081 dir.percent = options->dirstat_percent;
1082 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1084 changed = 0;
1085 for (i = 0; i < q->nr; i++) {
1086 struct diff_filepair *p = q->queue[i];
1087 const char *name;
1088 unsigned long copied, added, damage;
1090 name = p->one->path ? p->one->path : p->two->path;
1092 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1093 diff_populate_filespec(p->one, 0);
1094 diff_populate_filespec(p->two, 0);
1095 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1096 &copied, &added);
1097 diff_free_filespec_data(p->one);
1098 diff_free_filespec_data(p->two);
1099 } else if (DIFF_FILE_VALID(p->one)) {
1100 diff_populate_filespec(p->one, 1);
1101 copied = added = 0;
1102 diff_free_filespec_data(p->one);
1103 } else if (DIFF_FILE_VALID(p->two)) {
1104 diff_populate_filespec(p->two, 1);
1105 copied = 0;
1106 added = p->two->size;
1107 diff_free_filespec_data(p->two);
1108 } else
1109 continue;
1112 * Original minus copied is the removed material,
1113 * added is the new material. They are both damages
1114 * made to the preimage.
1116 damage = (p->one->size - copied) + added;
1118 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1119 dir.files[dir.nr].name = name;
1120 dir.files[dir.nr].changed = damage;
1121 changed += damage;
1122 dir.nr++;
1125 /* This can happen even with many files, if everything was renames */
1126 if (!changed)
1127 return;
1129 /* Show all directories with more than x% of the changes */
1130 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1131 gather_dirstat(options->file, &dir, changed, "", 0);
1134 static void free_diffstat_info(struct diffstat_t *diffstat)
1136 int i;
1137 for (i = 0; i < diffstat->nr; i++) {
1138 struct diffstat_file *f = diffstat->files[i];
1139 if (f->name != f->print_name)
1140 free(f->print_name);
1141 free(f->name);
1142 free(f->from_name);
1143 free(f);
1145 free(diffstat->files);
1148 struct checkdiff_t {
1149 struct xdiff_emit_state xm;
1150 const char *filename;
1151 int lineno;
1152 struct diff_options *o;
1153 unsigned ws_rule;
1154 unsigned status;
1155 int trailing_blanks_start;
1158 static int is_conflict_marker(const char *line, unsigned long len)
1160 char firstchar;
1161 int cnt;
1163 if (len < 8)
1164 return 0;
1165 firstchar = line[0];
1166 switch (firstchar) {
1167 case '=': case '>': case '<':
1168 break;
1169 default:
1170 return 0;
1172 for (cnt = 1; cnt < 7; cnt++)
1173 if (line[cnt] != firstchar)
1174 return 0;
1175 /* line[0] thru line[6] are same as firstchar */
1176 if (firstchar == '=') {
1177 /* divider between ours and theirs? */
1178 if (len != 8 || line[7] != '\n')
1179 return 0;
1180 } else if (len < 8 || !isspace(line[7])) {
1181 /* not divider before ours nor after theirs */
1182 return 0;
1184 return 1;
1187 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1189 struct checkdiff_t *data = priv;
1190 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1191 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1192 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1193 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1194 char *err;
1196 if (line[0] == '+') {
1197 unsigned bad;
1198 data->lineno++;
1199 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1200 data->trailing_blanks_start = 0;
1201 else if (!data->trailing_blanks_start)
1202 data->trailing_blanks_start = data->lineno;
1203 if (is_conflict_marker(line + 1, len - 1)) {
1204 data->status |= 1;
1205 fprintf(data->o->file,
1206 "%s:%d: leftover conflict marker\n",
1207 data->filename, data->lineno);
1209 bad = ws_check(line + 1, len - 1, data->ws_rule);
1210 if (!bad)
1211 return;
1212 data->status |= bad;
1213 err = whitespace_error_string(bad);
1214 fprintf(data->o->file, "%s:%d: %s.\n",
1215 data->filename, data->lineno, err);
1216 free(err);
1217 emit_line(data->o->file, set, reset, line, 1);
1218 ws_check_emit(line + 1, len - 1, data->ws_rule,
1219 data->o->file, set, reset, ws);
1220 } else if (line[0] == ' ') {
1221 data->lineno++;
1222 data->trailing_blanks_start = 0;
1223 } else if (line[0] == '@') {
1224 char *plus = strchr(line, '+');
1225 if (plus)
1226 data->lineno = strtol(plus, NULL, 10) - 1;
1227 else
1228 die("invalid diff");
1229 data->trailing_blanks_start = 0;
1233 static unsigned char *deflate_it(char *data,
1234 unsigned long size,
1235 unsigned long *result_size)
1237 int bound;
1238 unsigned char *deflated;
1239 z_stream stream;
1241 memset(&stream, 0, sizeof(stream));
1242 deflateInit(&stream, zlib_compression_level);
1243 bound = deflateBound(&stream, size);
1244 deflated = xmalloc(bound);
1245 stream.next_out = deflated;
1246 stream.avail_out = bound;
1248 stream.next_in = (unsigned char *)data;
1249 stream.avail_in = size;
1250 while (deflate(&stream, Z_FINISH) == Z_OK)
1251 ; /* nothing */
1252 deflateEnd(&stream);
1253 *result_size = stream.total_out;
1254 return deflated;
1257 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1259 void *cp;
1260 void *delta;
1261 void *deflated;
1262 void *data;
1263 unsigned long orig_size;
1264 unsigned long delta_size;
1265 unsigned long deflate_size;
1266 unsigned long data_size;
1268 /* We could do deflated delta, or we could do just deflated two,
1269 * whichever is smaller.
1271 delta = NULL;
1272 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1273 if (one->size && two->size) {
1274 delta = diff_delta(one->ptr, one->size,
1275 two->ptr, two->size,
1276 &delta_size, deflate_size);
1277 if (delta) {
1278 void *to_free = delta;
1279 orig_size = delta_size;
1280 delta = deflate_it(delta, delta_size, &delta_size);
1281 free(to_free);
1285 if (delta && delta_size < deflate_size) {
1286 fprintf(file, "delta %lu\n", orig_size);
1287 free(deflated);
1288 data = delta;
1289 data_size = delta_size;
1291 else {
1292 fprintf(file, "literal %lu\n", two->size);
1293 free(delta);
1294 data = deflated;
1295 data_size = deflate_size;
1298 /* emit data encoded in base85 */
1299 cp = data;
1300 while (data_size) {
1301 int bytes = (52 < data_size) ? 52 : data_size;
1302 char line[70];
1303 data_size -= bytes;
1304 if (bytes <= 26)
1305 line[0] = bytes + 'A' - 1;
1306 else
1307 line[0] = bytes - 26 + 'a' - 1;
1308 encode_85(line + 1, cp, bytes);
1309 cp = (char *) cp + bytes;
1310 fputs(line, file);
1311 fputc('\n', file);
1313 fprintf(file, "\n");
1314 free(data);
1317 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1319 fprintf(file, "GIT binary patch\n");
1320 emit_binary_diff_body(file, one, two);
1321 emit_binary_diff_body(file, two, one);
1324 static void setup_diff_attr_check(struct git_attr_check *check)
1326 static struct git_attr *attr_diff;
1328 if (!attr_diff) {
1329 attr_diff = git_attr("diff", 4);
1331 check[0].attr = attr_diff;
1334 static void diff_filespec_check_attr(struct diff_filespec *one)
1336 struct git_attr_check attr_diff_check;
1337 int check_from_data = 0;
1339 if (one->checked_attr)
1340 return;
1342 setup_diff_attr_check(&attr_diff_check);
1343 one->is_binary = 0;
1344 one->funcname_pattern_ident = NULL;
1346 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1347 const char *value;
1349 /* binaryness */
1350 value = attr_diff_check.value;
1351 if (ATTR_TRUE(value))
1353 else if (ATTR_FALSE(value))
1354 one->is_binary = 1;
1355 else
1356 check_from_data = 1;
1358 /* funcname pattern ident */
1359 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1361 else
1362 one->funcname_pattern_ident = value;
1365 if (check_from_data) {
1366 if (!one->data && DIFF_FILE_VALID(one))
1367 diff_populate_filespec(one, 0);
1369 if (one->data)
1370 one->is_binary = buffer_is_binary(one->data, one->size);
1374 int diff_filespec_is_binary(struct diff_filespec *one)
1376 diff_filespec_check_attr(one);
1377 return one->is_binary;
1380 static const char *funcname_pattern(const char *ident)
1382 struct funcname_pattern *pp;
1384 for (pp = funcname_pattern_list; pp; pp = pp->next)
1385 if (!strcmp(ident, pp->name))
1386 return pp->pattern;
1387 return NULL;
1390 static struct builtin_funcname_pattern {
1391 const char *name;
1392 const char *pattern;
1393 } builtin_funcname_pattern[] = {
1394 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1395 "new\\|return\\|switch\\|throw\\|while\\)\n"
1396 "^[ ]*\\(\\([ ]*"
1397 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1398 "[ ]*([^;]*\\)$" },
1399 { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1400 "destructor\\|interface\\|implementation\\|"
1401 "initialization\\|finalization\\)[ \t]*.*\\)$"
1402 "\\|"
1403 "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1405 { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1406 { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1407 { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1410 static const char *diff_funcname_pattern(struct diff_filespec *one)
1412 const char *ident, *pattern;
1413 int i;
1415 diff_filespec_check_attr(one);
1416 ident = one->funcname_pattern_ident;
1418 if (!ident)
1420 * If the config file has "funcname.default" defined, that
1421 * regexp is used; otherwise NULL is returned and xemit uses
1422 * the built-in default.
1424 return funcname_pattern("default");
1426 /* Look up custom "funcname.$ident" regexp from config. */
1427 pattern = funcname_pattern(ident);
1428 if (pattern)
1429 return pattern;
1432 * And define built-in fallback patterns here. Note that
1433 * these can be overridden by the user's config settings.
1435 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1436 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1437 return builtin_funcname_pattern[i].pattern;
1439 return NULL;
1442 static void builtin_diff(const char *name_a,
1443 const char *name_b,
1444 struct diff_filespec *one,
1445 struct diff_filespec *two,
1446 const char *xfrm_msg,
1447 struct diff_options *o,
1448 int complete_rewrite)
1450 mmfile_t mf1, mf2;
1451 const char *lbl[2];
1452 char *a_one, *b_two;
1453 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1454 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1456 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1457 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1458 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1459 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1460 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1461 if (lbl[0][0] == '/') {
1462 /* /dev/null */
1463 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1464 if (xfrm_msg && xfrm_msg[0])
1465 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1467 else if (lbl[1][0] == '/') {
1468 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1469 if (xfrm_msg && xfrm_msg[0])
1470 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1472 else {
1473 if (one->mode != two->mode) {
1474 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1475 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1477 if (xfrm_msg && xfrm_msg[0])
1478 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1480 * we do not run diff between different kind
1481 * of objects.
1483 if ((one->mode ^ two->mode) & S_IFMT)
1484 goto free_ab_and_return;
1485 if (complete_rewrite) {
1486 emit_rewrite_diff(name_a, name_b, one, two, o);
1487 o->found_changes = 1;
1488 goto free_ab_and_return;
1492 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1493 die("unable to read files to diff");
1495 if (!DIFF_OPT_TST(o, TEXT) &&
1496 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1497 /* Quite common confusing case */
1498 if (mf1.size == mf2.size &&
1499 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1500 goto free_ab_and_return;
1501 if (DIFF_OPT_TST(o, BINARY))
1502 emit_binary_diff(o->file, &mf1, &mf2);
1503 else
1504 fprintf(o->file, "Binary files %s and %s differ\n",
1505 lbl[0], lbl[1]);
1506 o->found_changes = 1;
1508 else {
1509 /* Crazy xdl interfaces.. */
1510 const char *diffopts = getenv("GIT_DIFF_OPTS");
1511 xpparam_t xpp;
1512 xdemitconf_t xecfg;
1513 xdemitcb_t ecb;
1514 struct emit_callback ecbdata;
1515 const char *funcname_pattern;
1517 funcname_pattern = diff_funcname_pattern(one);
1518 if (!funcname_pattern)
1519 funcname_pattern = diff_funcname_pattern(two);
1521 memset(&xecfg, 0, sizeof(xecfg));
1522 memset(&ecbdata, 0, sizeof(ecbdata));
1523 ecbdata.label_path = lbl;
1524 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1525 ecbdata.found_changesp = &o->found_changes;
1526 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1527 ecbdata.file = o->file;
1528 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1529 xecfg.ctxlen = o->context;
1530 xecfg.flags = XDL_EMIT_FUNCNAMES;
1531 if (funcname_pattern)
1532 xdiff_set_find_func(&xecfg, funcname_pattern);
1533 if (!diffopts)
1535 else if (!prefixcmp(diffopts, "--unified="))
1536 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1537 else if (!prefixcmp(diffopts, "-u"))
1538 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1539 ecb.outf = xdiff_outf;
1540 ecb.priv = &ecbdata;
1541 ecbdata.xm.consume = fn_out_consume;
1542 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1543 ecbdata.diff_words =
1544 xcalloc(1, sizeof(struct diff_words_data));
1545 ecbdata.diff_words->file = o->file;
1547 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1548 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1549 free_diff_words_data(&ecbdata);
1552 free_ab_and_return:
1553 diff_free_filespec_data(one);
1554 diff_free_filespec_data(two);
1555 free(a_one);
1556 free(b_two);
1557 return;
1560 static void builtin_diffstat(const char *name_a, const char *name_b,
1561 struct diff_filespec *one,
1562 struct diff_filespec *two,
1563 struct diffstat_t *diffstat,
1564 struct diff_options *o,
1565 int complete_rewrite)
1567 mmfile_t mf1, mf2;
1568 struct diffstat_file *data;
1570 data = diffstat_add(diffstat, name_a, name_b);
1572 if (!one || !two) {
1573 data->is_unmerged = 1;
1574 return;
1576 if (complete_rewrite) {
1577 diff_populate_filespec(one, 0);
1578 diff_populate_filespec(two, 0);
1579 data->deleted = count_lines(one->data, one->size);
1580 data->added = count_lines(two->data, two->size);
1581 goto free_and_return;
1583 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1584 die("unable to read files to diff");
1586 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1587 data->is_binary = 1;
1588 data->added = mf2.size;
1589 data->deleted = mf1.size;
1590 } else {
1591 /* Crazy xdl interfaces.. */
1592 xpparam_t xpp;
1593 xdemitconf_t xecfg;
1594 xdemitcb_t ecb;
1596 memset(&xecfg, 0, sizeof(xecfg));
1597 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1598 ecb.outf = xdiff_outf;
1599 ecb.priv = diffstat;
1600 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1603 free_and_return:
1604 diff_free_filespec_data(one);
1605 diff_free_filespec_data(two);
1608 static void builtin_checkdiff(const char *name_a, const char *name_b,
1609 const char *attr_path,
1610 struct diff_filespec *one,
1611 struct diff_filespec *two,
1612 struct diff_options *o)
1614 mmfile_t mf1, mf2;
1615 struct checkdiff_t data;
1617 if (!two)
1618 return;
1620 memset(&data, 0, sizeof(data));
1621 data.xm.consume = checkdiff_consume;
1622 data.filename = name_b ? name_b : name_a;
1623 data.lineno = 0;
1624 data.o = o;
1625 data.ws_rule = whitespace_rule(attr_path);
1627 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1628 die("unable to read files to diff");
1631 * All the other codepaths check both sides, but not checking
1632 * the "old" side here is deliberate. We are checking the newly
1633 * introduced changes, and as long as the "new" side is text, we
1634 * can and should check what it introduces.
1636 if (diff_filespec_is_binary(two))
1637 goto free_and_return;
1638 else {
1639 /* Crazy xdl interfaces.. */
1640 xpparam_t xpp;
1641 xdemitconf_t xecfg;
1642 xdemitcb_t ecb;
1644 memset(&xecfg, 0, sizeof(xecfg));
1645 xecfg.ctxlen = 1; /* at least one context line */
1646 xpp.flags = XDF_NEED_MINIMAL;
1647 ecb.outf = xdiff_outf;
1648 ecb.priv = &data;
1649 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1651 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1652 data.trailing_blanks_start) {
1653 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1654 data.filename, data.trailing_blanks_start);
1655 data.status = 1; /* report errors */
1658 free_and_return:
1659 diff_free_filespec_data(one);
1660 diff_free_filespec_data(two);
1661 if (data.status)
1662 DIFF_OPT_SET(o, CHECK_FAILED);
1665 struct diff_filespec *alloc_filespec(const char *path)
1667 int namelen = strlen(path);
1668 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1670 memset(spec, 0, sizeof(*spec));
1671 spec->path = (char *)(spec + 1);
1672 memcpy(spec->path, path, namelen+1);
1673 spec->count = 1;
1674 return spec;
1677 void free_filespec(struct diff_filespec *spec)
1679 if (!--spec->count) {
1680 diff_free_filespec_data(spec);
1681 free(spec);
1685 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1686 unsigned short mode)
1688 if (mode) {
1689 spec->mode = canon_mode(mode);
1690 hashcpy(spec->sha1, sha1);
1691 spec->sha1_valid = !is_null_sha1(sha1);
1696 * Given a name and sha1 pair, if the index tells us the file in
1697 * the work tree has that object contents, return true, so that
1698 * prepare_temp_file() does not have to inflate and extract.
1700 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1702 struct cache_entry *ce;
1703 struct stat st;
1704 int pos, len;
1706 /* We do not read the cache ourselves here, because the
1707 * benchmark with my previous version that always reads cache
1708 * shows that it makes things worse for diff-tree comparing
1709 * two linux-2.6 kernel trees in an already checked out work
1710 * tree. This is because most diff-tree comparisons deal with
1711 * only a small number of files, while reading the cache is
1712 * expensive for a large project, and its cost outweighs the
1713 * savings we get by not inflating the object to a temporary
1714 * file. Practically, this code only helps when we are used
1715 * by diff-cache --cached, which does read the cache before
1716 * calling us.
1718 if (!active_cache)
1719 return 0;
1721 /* We want to avoid the working directory if our caller
1722 * doesn't need the data in a normal file, this system
1723 * is rather slow with its stat/open/mmap/close syscalls,
1724 * and the object is contained in a pack file. The pack
1725 * is probably already open and will be faster to obtain
1726 * the data through than the working directory. Loose
1727 * objects however would tend to be slower as they need
1728 * to be individually opened and inflated.
1730 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1731 return 0;
1733 len = strlen(name);
1734 pos = cache_name_pos(name, len);
1735 if (pos < 0)
1736 return 0;
1737 ce = active_cache[pos];
1740 * This is not the sha1 we are looking for, or
1741 * unreusable because it is not a regular file.
1743 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1744 return 0;
1747 * If ce matches the file in the work tree, we can reuse it.
1749 if (ce_uptodate(ce) ||
1750 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1751 return 1;
1753 return 0;
1756 static int populate_from_stdin(struct diff_filespec *s)
1758 struct strbuf buf;
1759 size_t size = 0;
1761 strbuf_init(&buf, 0);
1762 if (strbuf_read(&buf, 0, 0) < 0)
1763 return error("error while reading from stdin %s",
1764 strerror(errno));
1766 s->should_munmap = 0;
1767 s->data = strbuf_detach(&buf, &size);
1768 s->size = size;
1769 s->should_free = 1;
1770 return 0;
1773 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1775 int len;
1776 char *data = xmalloc(100);
1777 len = snprintf(data, 100,
1778 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1779 s->data = data;
1780 s->size = len;
1781 s->should_free = 1;
1782 if (size_only) {
1783 s->data = NULL;
1784 free(data);
1786 return 0;
1790 * While doing rename detection and pickaxe operation, we may need to
1791 * grab the data for the blob (or file) for our own in-core comparison.
1792 * diff_filespec has data and size fields for this purpose.
1794 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1796 int err = 0;
1797 if (!DIFF_FILE_VALID(s))
1798 die("internal error: asking to populate invalid file.");
1799 if (S_ISDIR(s->mode))
1800 return -1;
1802 if (s->data)
1803 return 0;
1805 if (size_only && 0 < s->size)
1806 return 0;
1808 if (S_ISGITLINK(s->mode))
1809 return diff_populate_gitlink(s, size_only);
1811 if (!s->sha1_valid ||
1812 reuse_worktree_file(s->path, s->sha1, 0)) {
1813 struct strbuf buf;
1814 struct stat st;
1815 int fd;
1817 if (!strcmp(s->path, "-"))
1818 return populate_from_stdin(s);
1820 if (lstat(s->path, &st) < 0) {
1821 if (errno == ENOENT) {
1822 err_empty:
1823 err = -1;
1824 empty:
1825 s->data = (char *)"";
1826 s->size = 0;
1827 return err;
1830 s->size = xsize_t(st.st_size);
1831 if (!s->size)
1832 goto empty;
1833 if (size_only)
1834 return 0;
1835 if (S_ISLNK(st.st_mode)) {
1836 int ret;
1837 s->data = xmalloc(s->size);
1838 s->should_free = 1;
1839 ret = readlink(s->path, s->data, s->size);
1840 if (ret < 0) {
1841 free(s->data);
1842 goto err_empty;
1844 return 0;
1846 fd = open(s->path, O_RDONLY);
1847 if (fd < 0)
1848 goto err_empty;
1849 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1850 close(fd);
1851 s->should_munmap = 1;
1854 * Convert from working tree format to canonical git format
1856 strbuf_init(&buf, 0);
1857 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1858 size_t size = 0;
1859 munmap(s->data, s->size);
1860 s->should_munmap = 0;
1861 s->data = strbuf_detach(&buf, &size);
1862 s->size = size;
1863 s->should_free = 1;
1866 else {
1867 enum object_type type;
1868 if (size_only)
1869 type = sha1_object_info(s->sha1, &s->size);
1870 else {
1871 s->data = read_sha1_file(s->sha1, &type, &s->size);
1872 s->should_free = 1;
1875 return 0;
1878 void diff_free_filespec_blob(struct diff_filespec *s)
1880 if (s->should_free)
1881 free(s->data);
1882 else if (s->should_munmap)
1883 munmap(s->data, s->size);
1885 if (s->should_free || s->should_munmap) {
1886 s->should_free = s->should_munmap = 0;
1887 s->data = NULL;
1891 void diff_free_filespec_data(struct diff_filespec *s)
1893 diff_free_filespec_blob(s);
1894 free(s->cnt_data);
1895 s->cnt_data = NULL;
1898 static void prep_temp_blob(struct diff_tempfile *temp,
1899 void *blob,
1900 unsigned long size,
1901 const unsigned char *sha1,
1902 int mode)
1904 int fd;
1906 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1907 if (fd < 0)
1908 die("unable to create temp-file: %s", strerror(errno));
1909 if (write_in_full(fd, blob, size) != size)
1910 die("unable to write temp-file");
1911 close(fd);
1912 temp->name = temp->tmp_path;
1913 strcpy(temp->hex, sha1_to_hex(sha1));
1914 temp->hex[40] = 0;
1915 sprintf(temp->mode, "%06o", mode);
1918 static void prepare_temp_file(const char *name,
1919 struct diff_tempfile *temp,
1920 struct diff_filespec *one)
1922 if (!DIFF_FILE_VALID(one)) {
1923 not_a_valid_file:
1924 /* A '-' entry produces this for file-2, and
1925 * a '+' entry produces this for file-1.
1927 temp->name = "/dev/null";
1928 strcpy(temp->hex, ".");
1929 strcpy(temp->mode, ".");
1930 return;
1933 if (!one->sha1_valid ||
1934 reuse_worktree_file(name, one->sha1, 1)) {
1935 struct stat st;
1936 if (lstat(name, &st) < 0) {
1937 if (errno == ENOENT)
1938 goto not_a_valid_file;
1939 die("stat(%s): %s", name, strerror(errno));
1941 if (S_ISLNK(st.st_mode)) {
1942 int ret;
1943 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1944 size_t sz = xsize_t(st.st_size);
1945 if (sizeof(buf) <= st.st_size)
1946 die("symlink too long: %s", name);
1947 ret = readlink(name, buf, sz);
1948 if (ret < 0)
1949 die("readlink(%s)", name);
1950 prep_temp_blob(temp, buf, sz,
1951 (one->sha1_valid ?
1952 one->sha1 : null_sha1),
1953 (one->sha1_valid ?
1954 one->mode : S_IFLNK));
1956 else {
1957 /* we can borrow from the file in the work tree */
1958 temp->name = name;
1959 if (!one->sha1_valid)
1960 strcpy(temp->hex, sha1_to_hex(null_sha1));
1961 else
1962 strcpy(temp->hex, sha1_to_hex(one->sha1));
1963 /* Even though we may sometimes borrow the
1964 * contents from the work tree, we always want
1965 * one->mode. mode is trustworthy even when
1966 * !(one->sha1_valid), as long as
1967 * DIFF_FILE_VALID(one).
1969 sprintf(temp->mode, "%06o", one->mode);
1971 return;
1973 else {
1974 if (diff_populate_filespec(one, 0))
1975 die("cannot read data blob for %s", one->path);
1976 prep_temp_blob(temp, one->data, one->size,
1977 one->sha1, one->mode);
1981 static void remove_tempfile(void)
1983 int i;
1985 for (i = 0; i < 2; i++)
1986 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1987 unlink(diff_temp[i].name);
1988 diff_temp[i].name = NULL;
1992 static void remove_tempfile_on_signal(int signo)
1994 remove_tempfile();
1995 signal(SIGINT, SIG_DFL);
1996 raise(signo);
1999 /* An external diff command takes:
2001 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2002 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2005 static void run_external_diff(const char *pgm,
2006 const char *name,
2007 const char *other,
2008 struct diff_filespec *one,
2009 struct diff_filespec *two,
2010 const char *xfrm_msg,
2011 int complete_rewrite)
2013 const char *spawn_arg[10];
2014 struct diff_tempfile *temp = diff_temp;
2015 int retval;
2016 static int atexit_asked = 0;
2017 const char *othername;
2018 const char **arg = &spawn_arg[0];
2020 othername = (other? other : name);
2021 if (one && two) {
2022 prepare_temp_file(name, &temp[0], one);
2023 prepare_temp_file(othername, &temp[1], two);
2024 if (! atexit_asked &&
2025 (temp[0].name == temp[0].tmp_path ||
2026 temp[1].name == temp[1].tmp_path)) {
2027 atexit_asked = 1;
2028 atexit(remove_tempfile);
2030 signal(SIGINT, remove_tempfile_on_signal);
2033 if (one && two) {
2034 *arg++ = pgm;
2035 *arg++ = name;
2036 *arg++ = temp[0].name;
2037 *arg++ = temp[0].hex;
2038 *arg++ = temp[0].mode;
2039 *arg++ = temp[1].name;
2040 *arg++ = temp[1].hex;
2041 *arg++ = temp[1].mode;
2042 if (other) {
2043 *arg++ = other;
2044 *arg++ = xfrm_msg;
2046 } else {
2047 *arg++ = pgm;
2048 *arg++ = name;
2050 *arg = NULL;
2051 fflush(NULL);
2052 retval = run_command_v_opt(spawn_arg, 0);
2053 remove_tempfile();
2054 if (retval) {
2055 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2056 exit(1);
2060 static const char *external_diff_attr(const char *name)
2062 struct git_attr_check attr_diff_check;
2064 if (!name)
2065 return NULL;
2067 setup_diff_attr_check(&attr_diff_check);
2068 if (!git_checkattr(name, 1, &attr_diff_check)) {
2069 const char *value = attr_diff_check.value;
2070 if (!ATTR_TRUE(value) &&
2071 !ATTR_FALSE(value) &&
2072 !ATTR_UNSET(value)) {
2073 struct ll_diff_driver *drv;
2075 for (drv = user_diff; drv; drv = drv->next)
2076 if (!strcmp(drv->name, value))
2077 return drv->cmd;
2080 return NULL;
2083 static void run_diff_cmd(const char *pgm,
2084 const char *name,
2085 const char *other,
2086 const char *attr_path,
2087 struct diff_filespec *one,
2088 struct diff_filespec *two,
2089 const char *xfrm_msg,
2090 struct diff_options *o,
2091 int complete_rewrite)
2093 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2094 pgm = NULL;
2095 else {
2096 const char *cmd = external_diff_attr(attr_path);
2097 if (cmd)
2098 pgm = cmd;
2101 if (pgm) {
2102 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2103 complete_rewrite);
2104 return;
2106 if (one && two)
2107 builtin_diff(name, other ? other : name,
2108 one, two, xfrm_msg, o, complete_rewrite);
2109 else
2110 fprintf(o->file, "* Unmerged path %s\n", name);
2113 static void diff_fill_sha1_info(struct diff_filespec *one)
2115 if (DIFF_FILE_VALID(one)) {
2116 if (!one->sha1_valid) {
2117 struct stat st;
2118 if (!strcmp(one->path, "-")) {
2119 hashcpy(one->sha1, null_sha1);
2120 return;
2122 if (lstat(one->path, &st) < 0)
2123 die("stat %s", one->path);
2124 if (index_path(one->sha1, one->path, &st, 0))
2125 die("cannot hash %s\n", one->path);
2128 else
2129 hashclr(one->sha1);
2132 static int similarity_index(struct diff_filepair *p)
2134 return p->score * 100 / MAX_SCORE;
2137 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2139 /* Strip the prefix but do not molest /dev/null and absolute paths */
2140 if (*namep && **namep != '/')
2141 *namep += prefix_length;
2142 if (*otherp && **otherp != '/')
2143 *otherp += prefix_length;
2146 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2148 const char *pgm = external_diff();
2149 struct strbuf msg;
2150 char *xfrm_msg;
2151 struct diff_filespec *one = p->one;
2152 struct diff_filespec *two = p->two;
2153 const char *name;
2154 const char *other;
2155 const char *attr_path;
2156 int complete_rewrite = 0;
2158 name = p->one->path;
2159 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2160 attr_path = name;
2161 if (o->prefix_length)
2162 strip_prefix(o->prefix_length, &name, &other);
2164 if (DIFF_PAIR_UNMERGED(p)) {
2165 run_diff_cmd(pgm, name, NULL, attr_path,
2166 NULL, NULL, NULL, o, 0);
2167 return;
2170 diff_fill_sha1_info(one);
2171 diff_fill_sha1_info(two);
2173 strbuf_init(&msg, PATH_MAX * 2 + 300);
2174 switch (p->status) {
2175 case DIFF_STATUS_COPIED:
2176 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2177 strbuf_addstr(&msg, "\ncopy from ");
2178 quote_c_style(name, &msg, NULL, 0);
2179 strbuf_addstr(&msg, "\ncopy to ");
2180 quote_c_style(other, &msg, NULL, 0);
2181 strbuf_addch(&msg, '\n');
2182 break;
2183 case DIFF_STATUS_RENAMED:
2184 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2185 strbuf_addstr(&msg, "\nrename from ");
2186 quote_c_style(name, &msg, NULL, 0);
2187 strbuf_addstr(&msg, "\nrename to ");
2188 quote_c_style(other, &msg, NULL, 0);
2189 strbuf_addch(&msg, '\n');
2190 break;
2191 case DIFF_STATUS_MODIFIED:
2192 if (p->score) {
2193 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2194 similarity_index(p));
2195 complete_rewrite = 1;
2196 break;
2198 /* fallthru */
2199 default:
2200 /* nothing */
2204 if (hashcmp(one->sha1, two->sha1)) {
2205 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2207 if (DIFF_OPT_TST(o, BINARY)) {
2208 mmfile_t mf;
2209 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2210 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2211 abbrev = 40;
2213 strbuf_addf(&msg, "index %.*s..%.*s",
2214 abbrev, sha1_to_hex(one->sha1),
2215 abbrev, sha1_to_hex(two->sha1));
2216 if (one->mode == two->mode)
2217 strbuf_addf(&msg, " %06o", one->mode);
2218 strbuf_addch(&msg, '\n');
2221 if (msg.len)
2222 strbuf_setlen(&msg, msg.len - 1);
2223 xfrm_msg = msg.len ? msg.buf : NULL;
2225 if (!pgm &&
2226 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2227 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2228 /* a filepair that changes between file and symlink
2229 * needs to be split into deletion and creation.
2231 struct diff_filespec *null = alloc_filespec(two->path);
2232 run_diff_cmd(NULL, name, other, attr_path,
2233 one, null, xfrm_msg, o, 0);
2234 free(null);
2235 null = alloc_filespec(one->path);
2236 run_diff_cmd(NULL, name, other, attr_path,
2237 null, two, xfrm_msg, o, 0);
2238 free(null);
2240 else
2241 run_diff_cmd(pgm, name, other, attr_path,
2242 one, two, xfrm_msg, o, complete_rewrite);
2244 strbuf_release(&msg);
2247 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2248 struct diffstat_t *diffstat)
2250 const char *name;
2251 const char *other;
2252 int complete_rewrite = 0;
2254 if (DIFF_PAIR_UNMERGED(p)) {
2255 /* unmerged */
2256 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2257 return;
2260 name = p->one->path;
2261 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2263 if (o->prefix_length)
2264 strip_prefix(o->prefix_length, &name, &other);
2266 diff_fill_sha1_info(p->one);
2267 diff_fill_sha1_info(p->two);
2269 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2270 complete_rewrite = 1;
2271 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2274 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2276 const char *name;
2277 const char *other;
2278 const char *attr_path;
2280 if (DIFF_PAIR_UNMERGED(p)) {
2281 /* unmerged */
2282 return;
2285 name = p->one->path;
2286 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2287 attr_path = other ? other : name;
2289 if (o->prefix_length)
2290 strip_prefix(o->prefix_length, &name, &other);
2292 diff_fill_sha1_info(p->one);
2293 diff_fill_sha1_info(p->two);
2295 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2298 void diff_setup(struct diff_options *options)
2300 memset(options, 0, sizeof(*options));
2302 options->file = stdout;
2304 options->line_termination = '\n';
2305 options->break_opt = -1;
2306 options->rename_limit = -1;
2307 options->dirstat_percent = 3;
2308 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2309 options->context = 3;
2311 options->change = diff_change;
2312 options->add_remove = diff_addremove;
2313 if (diff_use_color_default > 0)
2314 DIFF_OPT_SET(options, COLOR_DIFF);
2315 else
2316 DIFF_OPT_CLR(options, COLOR_DIFF);
2317 options->detect_rename = diff_detect_rename_default;
2319 options->a_prefix = "a/";
2320 options->b_prefix = "b/";
2323 int diff_setup_done(struct diff_options *options)
2325 int count = 0;
2327 if (options->output_format & DIFF_FORMAT_NAME)
2328 count++;
2329 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2330 count++;
2331 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2332 count++;
2333 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2334 count++;
2335 if (count > 1)
2336 die("--name-only, --name-status, --check and -s are mutually exclusive");
2338 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2339 options->detect_rename = DIFF_DETECT_COPY;
2341 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2342 options->prefix = NULL;
2343 if (options->prefix)
2344 options->prefix_length = strlen(options->prefix);
2345 else
2346 options->prefix_length = 0;
2348 if (options->output_format & (DIFF_FORMAT_NAME |
2349 DIFF_FORMAT_NAME_STATUS |
2350 DIFF_FORMAT_CHECKDIFF |
2351 DIFF_FORMAT_NO_OUTPUT))
2352 options->output_format &= ~(DIFF_FORMAT_RAW |
2353 DIFF_FORMAT_NUMSTAT |
2354 DIFF_FORMAT_DIFFSTAT |
2355 DIFF_FORMAT_SHORTSTAT |
2356 DIFF_FORMAT_DIRSTAT |
2357 DIFF_FORMAT_SUMMARY |
2358 DIFF_FORMAT_PATCH);
2361 * These cases always need recursive; we do not drop caller-supplied
2362 * recursive bits for other formats here.
2364 if (options->output_format & (DIFF_FORMAT_PATCH |
2365 DIFF_FORMAT_NUMSTAT |
2366 DIFF_FORMAT_DIFFSTAT |
2367 DIFF_FORMAT_SHORTSTAT |
2368 DIFF_FORMAT_DIRSTAT |
2369 DIFF_FORMAT_SUMMARY |
2370 DIFF_FORMAT_CHECKDIFF))
2371 DIFF_OPT_SET(options, RECURSIVE);
2373 * Also pickaxe would not work very well if you do not say recursive
2375 if (options->pickaxe)
2376 DIFF_OPT_SET(options, RECURSIVE);
2378 if (options->detect_rename && options->rename_limit < 0)
2379 options->rename_limit = diff_rename_limit_default;
2380 if (options->setup & DIFF_SETUP_USE_CACHE) {
2381 if (!active_cache)
2382 /* read-cache does not die even when it fails
2383 * so it is safe for us to do this here. Also
2384 * it does not smudge active_cache or active_nr
2385 * when it fails, so we do not have to worry about
2386 * cleaning it up ourselves either.
2388 read_cache();
2390 if (options->abbrev <= 0 || 40 < options->abbrev)
2391 options->abbrev = 40; /* full */
2394 * It does not make sense to show the first hit we happened
2395 * to have found. It does not make sense not to return with
2396 * exit code in such a case either.
2398 if (DIFF_OPT_TST(options, QUIET)) {
2399 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2400 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2403 return 0;
2406 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2408 char c, *eq;
2409 int len;
2411 if (*arg != '-')
2412 return 0;
2413 c = *++arg;
2414 if (!c)
2415 return 0;
2416 if (c == arg_short) {
2417 c = *++arg;
2418 if (!c)
2419 return 1;
2420 if (val && isdigit(c)) {
2421 char *end;
2422 int n = strtoul(arg, &end, 10);
2423 if (*end)
2424 return 0;
2425 *val = n;
2426 return 1;
2428 return 0;
2430 if (c != '-')
2431 return 0;
2432 arg++;
2433 eq = strchr(arg, '=');
2434 if (eq)
2435 len = eq - arg;
2436 else
2437 len = strlen(arg);
2438 if (!len || strncmp(arg, arg_long, len))
2439 return 0;
2440 if (eq) {
2441 int n;
2442 char *end;
2443 if (!isdigit(*++eq))
2444 return 0;
2445 n = strtoul(eq, &end, 10);
2446 if (*end)
2447 return 0;
2448 *val = n;
2450 return 1;
2453 static int diff_scoreopt_parse(const char *opt);
2455 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2457 const char *arg = av[0];
2459 /* Output format options */
2460 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2461 options->output_format |= DIFF_FORMAT_PATCH;
2462 else if (opt_arg(arg, 'U', "unified", &options->context))
2463 options->output_format |= DIFF_FORMAT_PATCH;
2464 else if (!strcmp(arg, "--raw"))
2465 options->output_format |= DIFF_FORMAT_RAW;
2466 else if (!strcmp(arg, "--patch-with-raw"))
2467 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2468 else if (!strcmp(arg, "--numstat"))
2469 options->output_format |= DIFF_FORMAT_NUMSTAT;
2470 else if (!strcmp(arg, "--shortstat"))
2471 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2472 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2473 options->output_format |= DIFF_FORMAT_DIRSTAT;
2474 else if (!strcmp(arg, "--cumulative")) {
2475 options->output_format |= DIFF_FORMAT_DIRSTAT;
2476 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2478 else if (!strcmp(arg, "--check"))
2479 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2480 else if (!strcmp(arg, "--summary"))
2481 options->output_format |= DIFF_FORMAT_SUMMARY;
2482 else if (!strcmp(arg, "--patch-with-stat"))
2483 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2484 else if (!strcmp(arg, "--name-only"))
2485 options->output_format |= DIFF_FORMAT_NAME;
2486 else if (!strcmp(arg, "--name-status"))
2487 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2488 else if (!strcmp(arg, "-s"))
2489 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2490 else if (!prefixcmp(arg, "--stat")) {
2491 char *end;
2492 int width = options->stat_width;
2493 int name_width = options->stat_name_width;
2494 arg += 6;
2495 end = (char *)arg;
2497 switch (*arg) {
2498 case '-':
2499 if (!prefixcmp(arg, "-width="))
2500 width = strtoul(arg + 7, &end, 10);
2501 else if (!prefixcmp(arg, "-name-width="))
2502 name_width = strtoul(arg + 12, &end, 10);
2503 break;
2504 case '=':
2505 width = strtoul(arg+1, &end, 10);
2506 if (*end == ',')
2507 name_width = strtoul(end+1, &end, 10);
2510 /* Important! This checks all the error cases! */
2511 if (*end)
2512 return 0;
2513 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2514 options->stat_name_width = name_width;
2515 options->stat_width = width;
2518 /* renames options */
2519 else if (!prefixcmp(arg, "-B")) {
2520 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2521 return -1;
2523 else if (!prefixcmp(arg, "-M")) {
2524 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2525 return -1;
2526 options->detect_rename = DIFF_DETECT_RENAME;
2528 else if (!prefixcmp(arg, "-C")) {
2529 if (options->detect_rename == DIFF_DETECT_COPY)
2530 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2531 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2532 return -1;
2533 options->detect_rename = DIFF_DETECT_COPY;
2535 else if (!strcmp(arg, "--no-renames"))
2536 options->detect_rename = 0;
2537 else if (!strcmp(arg, "--relative"))
2538 DIFF_OPT_SET(options, RELATIVE_NAME);
2539 else if (!prefixcmp(arg, "--relative=")) {
2540 DIFF_OPT_SET(options, RELATIVE_NAME);
2541 options->prefix = arg + 11;
2544 /* xdiff options */
2545 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2546 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2547 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2548 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2549 else if (!strcmp(arg, "--ignore-space-at-eol"))
2550 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2552 /* flags options */
2553 else if (!strcmp(arg, "--binary")) {
2554 options->output_format |= DIFF_FORMAT_PATCH;
2555 DIFF_OPT_SET(options, BINARY);
2557 else if (!strcmp(arg, "--full-index"))
2558 DIFF_OPT_SET(options, FULL_INDEX);
2559 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2560 DIFF_OPT_SET(options, TEXT);
2561 else if (!strcmp(arg, "-R"))
2562 DIFF_OPT_SET(options, REVERSE_DIFF);
2563 else if (!strcmp(arg, "--find-copies-harder"))
2564 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2565 else if (!strcmp(arg, "--follow"))
2566 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2567 else if (!strcmp(arg, "--color"))
2568 DIFF_OPT_SET(options, COLOR_DIFF);
2569 else if (!strcmp(arg, "--no-color"))
2570 DIFF_OPT_CLR(options, COLOR_DIFF);
2571 else if (!strcmp(arg, "--color-words"))
2572 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2573 else if (!strcmp(arg, "--exit-code"))
2574 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2575 else if (!strcmp(arg, "--quiet"))
2576 DIFF_OPT_SET(options, QUIET);
2577 else if (!strcmp(arg, "--ext-diff"))
2578 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2579 else if (!strcmp(arg, "--no-ext-diff"))
2580 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2581 else if (!strcmp(arg, "--ignore-submodules"))
2582 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2584 /* misc options */
2585 else if (!strcmp(arg, "-z"))
2586 options->line_termination = 0;
2587 else if (!prefixcmp(arg, "-l"))
2588 options->rename_limit = strtoul(arg+2, NULL, 10);
2589 else if (!prefixcmp(arg, "-S"))
2590 options->pickaxe = arg + 2;
2591 else if (!strcmp(arg, "--pickaxe-all"))
2592 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2593 else if (!strcmp(arg, "--pickaxe-regex"))
2594 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2595 else if (!prefixcmp(arg, "-O"))
2596 options->orderfile = arg + 2;
2597 else if (!prefixcmp(arg, "--diff-filter="))
2598 options->filter = arg + 14;
2599 else if (!strcmp(arg, "--abbrev"))
2600 options->abbrev = DEFAULT_ABBREV;
2601 else if (!prefixcmp(arg, "--abbrev=")) {
2602 options->abbrev = strtoul(arg + 9, NULL, 10);
2603 if (options->abbrev < MINIMUM_ABBREV)
2604 options->abbrev = MINIMUM_ABBREV;
2605 else if (40 < options->abbrev)
2606 options->abbrev = 40;
2608 else if (!prefixcmp(arg, "--src-prefix="))
2609 options->a_prefix = arg + 13;
2610 else if (!prefixcmp(arg, "--dst-prefix="))
2611 options->b_prefix = arg + 13;
2612 else if (!strcmp(arg, "--no-prefix"))
2613 options->a_prefix = options->b_prefix = "";
2614 else if (!prefixcmp(arg, "--output=")) {
2615 options->file = fopen(arg + strlen("--output="), "w");
2616 options->close_file = 1;
2617 } else
2618 return 0;
2619 return 1;
2622 static int parse_num(const char **cp_p)
2624 unsigned long num, scale;
2625 int ch, dot;
2626 const char *cp = *cp_p;
2628 num = 0;
2629 scale = 1;
2630 dot = 0;
2631 for(;;) {
2632 ch = *cp;
2633 if ( !dot && ch == '.' ) {
2634 scale = 1;
2635 dot = 1;
2636 } else if ( ch == '%' ) {
2637 scale = dot ? scale*100 : 100;
2638 cp++; /* % is always at the end */
2639 break;
2640 } else if ( ch >= '0' && ch <= '9' ) {
2641 if ( scale < 100000 ) {
2642 scale *= 10;
2643 num = (num*10) + (ch-'0');
2645 } else {
2646 break;
2648 cp++;
2650 *cp_p = cp;
2652 /* user says num divided by scale and we say internally that
2653 * is MAX_SCORE * num / scale.
2655 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2658 static int diff_scoreopt_parse(const char *opt)
2660 int opt1, opt2, cmd;
2662 if (*opt++ != '-')
2663 return -1;
2664 cmd = *opt++;
2665 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2666 return -1; /* that is not a -M, -C nor -B option */
2668 opt1 = parse_num(&opt);
2669 if (cmd != 'B')
2670 opt2 = 0;
2671 else {
2672 if (*opt == 0)
2673 opt2 = 0;
2674 else if (*opt != '/')
2675 return -1; /* we expect -B80/99 or -B80 */
2676 else {
2677 opt++;
2678 opt2 = parse_num(&opt);
2681 if (*opt != 0)
2682 return -1;
2683 return opt1 | (opt2 << 16);
2686 struct diff_queue_struct diff_queued_diff;
2688 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2690 if (queue->alloc <= queue->nr) {
2691 queue->alloc = alloc_nr(queue->alloc);
2692 queue->queue = xrealloc(queue->queue,
2693 sizeof(dp) * queue->alloc);
2695 queue->queue[queue->nr++] = dp;
2698 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2699 struct diff_filespec *one,
2700 struct diff_filespec *two)
2702 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2703 dp->one = one;
2704 dp->two = two;
2705 if (queue)
2706 diff_q(queue, dp);
2707 return dp;
2710 void diff_free_filepair(struct diff_filepair *p)
2712 free_filespec(p->one);
2713 free_filespec(p->two);
2714 free(p);
2717 /* This is different from find_unique_abbrev() in that
2718 * it stuffs the result with dots for alignment.
2720 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2722 int abblen;
2723 const char *abbrev;
2724 if (len == 40)
2725 return sha1_to_hex(sha1);
2727 abbrev = find_unique_abbrev(sha1, len);
2728 abblen = strlen(abbrev);
2729 if (abblen < 37) {
2730 static char hex[41];
2731 if (len < abblen && abblen <= len + 2)
2732 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2733 else
2734 sprintf(hex, "%s...", abbrev);
2735 return hex;
2737 return sha1_to_hex(sha1);
2740 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2742 int line_termination = opt->line_termination;
2743 int inter_name_termination = line_termination ? '\t' : '\0';
2745 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2746 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2747 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2748 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2750 if (p->score) {
2751 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2752 inter_name_termination);
2753 } else {
2754 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2757 if (p->status == DIFF_STATUS_COPIED ||
2758 p->status == DIFF_STATUS_RENAMED) {
2759 const char *name_a, *name_b;
2760 name_a = p->one->path;
2761 name_b = p->two->path;
2762 strip_prefix(opt->prefix_length, &name_a, &name_b);
2763 write_name_quoted(name_a, opt->file, inter_name_termination);
2764 write_name_quoted(name_b, opt->file, line_termination);
2765 } else {
2766 const char *name_a, *name_b;
2767 name_a = p->one->mode ? p->one->path : p->two->path;
2768 name_b = NULL;
2769 strip_prefix(opt->prefix_length, &name_a, &name_b);
2770 write_name_quoted(name_a, opt->file, line_termination);
2774 int diff_unmodified_pair(struct diff_filepair *p)
2776 /* This function is written stricter than necessary to support
2777 * the currently implemented transformers, but the idea is to
2778 * let transformers to produce diff_filepairs any way they want,
2779 * and filter and clean them up here before producing the output.
2781 struct diff_filespec *one = p->one, *two = p->two;
2783 if (DIFF_PAIR_UNMERGED(p))
2784 return 0; /* unmerged is interesting */
2786 /* deletion, addition, mode or type change
2787 * and rename are all interesting.
2789 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2790 DIFF_PAIR_MODE_CHANGED(p) ||
2791 strcmp(one->path, two->path))
2792 return 0;
2794 /* both are valid and point at the same path. that is, we are
2795 * dealing with a change.
2797 if (one->sha1_valid && two->sha1_valid &&
2798 !hashcmp(one->sha1, two->sha1))
2799 return 1; /* no change */
2800 if (!one->sha1_valid && !two->sha1_valid)
2801 return 1; /* both look at the same file on the filesystem. */
2802 return 0;
2805 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2807 if (diff_unmodified_pair(p))
2808 return;
2810 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2811 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2812 return; /* no tree diffs in patch format */
2814 run_diff(p, o);
2817 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2818 struct diffstat_t *diffstat)
2820 if (diff_unmodified_pair(p))
2821 return;
2823 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2824 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2825 return; /* no tree diffs in patch format */
2827 run_diffstat(p, o, diffstat);
2830 static void diff_flush_checkdiff(struct diff_filepair *p,
2831 struct diff_options *o)
2833 if (diff_unmodified_pair(p))
2834 return;
2836 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2837 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2838 return; /* no tree diffs in patch format */
2840 run_checkdiff(p, o);
2843 int diff_queue_is_empty(void)
2845 struct diff_queue_struct *q = &diff_queued_diff;
2846 int i;
2847 for (i = 0; i < q->nr; i++)
2848 if (!diff_unmodified_pair(q->queue[i]))
2849 return 0;
2850 return 1;
2853 #if DIFF_DEBUG
2854 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2856 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2857 x, one ? one : "",
2858 s->path,
2859 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2860 s->mode,
2861 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2862 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2863 x, one ? one : "",
2864 s->size, s->xfrm_flags);
2867 void diff_debug_filepair(const struct diff_filepair *p, int i)
2869 diff_debug_filespec(p->one, i, "one");
2870 diff_debug_filespec(p->two, i, "two");
2871 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2872 p->score, p->status ? p->status : '?',
2873 p->one->rename_used, p->broken_pair);
2876 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2878 int i;
2879 if (msg)
2880 fprintf(stderr, "%s\n", msg);
2881 fprintf(stderr, "q->nr = %d\n", q->nr);
2882 for (i = 0; i < q->nr; i++) {
2883 struct diff_filepair *p = q->queue[i];
2884 diff_debug_filepair(p, i);
2887 #endif
2889 static void diff_resolve_rename_copy(void)
2891 int i;
2892 struct diff_filepair *p;
2893 struct diff_queue_struct *q = &diff_queued_diff;
2895 diff_debug_queue("resolve-rename-copy", q);
2897 for (i = 0; i < q->nr; i++) {
2898 p = q->queue[i];
2899 p->status = 0; /* undecided */
2900 if (DIFF_PAIR_UNMERGED(p))
2901 p->status = DIFF_STATUS_UNMERGED;
2902 else if (!DIFF_FILE_VALID(p->one))
2903 p->status = DIFF_STATUS_ADDED;
2904 else if (!DIFF_FILE_VALID(p->two))
2905 p->status = DIFF_STATUS_DELETED;
2906 else if (DIFF_PAIR_TYPE_CHANGED(p))
2907 p->status = DIFF_STATUS_TYPE_CHANGED;
2909 /* from this point on, we are dealing with a pair
2910 * whose both sides are valid and of the same type, i.e.
2911 * either in-place edit or rename/copy edit.
2913 else if (DIFF_PAIR_RENAME(p)) {
2915 * A rename might have re-connected a broken
2916 * pair up, causing the pathnames to be the
2917 * same again. If so, that's not a rename at
2918 * all, just a modification..
2920 * Otherwise, see if this source was used for
2921 * multiple renames, in which case we decrement
2922 * the count, and call it a copy.
2924 if (!strcmp(p->one->path, p->two->path))
2925 p->status = DIFF_STATUS_MODIFIED;
2926 else if (--p->one->rename_used > 0)
2927 p->status = DIFF_STATUS_COPIED;
2928 else
2929 p->status = DIFF_STATUS_RENAMED;
2931 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2932 p->one->mode != p->two->mode ||
2933 is_null_sha1(p->one->sha1))
2934 p->status = DIFF_STATUS_MODIFIED;
2935 else {
2936 /* This is a "no-change" entry and should not
2937 * happen anymore, but prepare for broken callers.
2939 error("feeding unmodified %s to diffcore",
2940 p->one->path);
2941 p->status = DIFF_STATUS_UNKNOWN;
2944 diff_debug_queue("resolve-rename-copy done", q);
2947 static int check_pair_status(struct diff_filepair *p)
2949 switch (p->status) {
2950 case DIFF_STATUS_UNKNOWN:
2951 return 0;
2952 case 0:
2953 die("internal error in diff-resolve-rename-copy");
2954 default:
2955 return 1;
2959 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2961 int fmt = opt->output_format;
2963 if (fmt & DIFF_FORMAT_CHECKDIFF)
2964 diff_flush_checkdiff(p, opt);
2965 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2966 diff_flush_raw(p, opt);
2967 else if (fmt & DIFF_FORMAT_NAME) {
2968 const char *name_a, *name_b;
2969 name_a = p->two->path;
2970 name_b = NULL;
2971 strip_prefix(opt->prefix_length, &name_a, &name_b);
2972 write_name_quoted(name_a, opt->file, opt->line_termination);
2976 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2978 if (fs->mode)
2979 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2980 else
2981 fprintf(file, " %s ", newdelete);
2982 write_name_quoted(fs->path, file, '\n');
2986 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2988 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2989 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2990 show_name ? ' ' : '\n');
2991 if (show_name) {
2992 write_name_quoted(p->two->path, file, '\n');
2997 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2999 char *names = pprint_rename(p->one->path, p->two->path);
3001 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3002 free(names);
3003 show_mode_change(file, p, 0);
3006 static void diff_summary(FILE *file, struct diff_filepair *p)
3008 switch(p->status) {
3009 case DIFF_STATUS_DELETED:
3010 show_file_mode_name(file, "delete", p->one);
3011 break;
3012 case DIFF_STATUS_ADDED:
3013 show_file_mode_name(file, "create", p->two);
3014 break;
3015 case DIFF_STATUS_COPIED:
3016 show_rename_copy(file, "copy", p);
3017 break;
3018 case DIFF_STATUS_RENAMED:
3019 show_rename_copy(file, "rename", p);
3020 break;
3021 default:
3022 if (p->score) {
3023 fputs(" rewrite ", file);
3024 write_name_quoted(p->two->path, file, ' ');
3025 fprintf(file, "(%d%%)\n", similarity_index(p));
3027 show_mode_change(file, p, !p->score);
3028 break;
3032 struct patch_id_t {
3033 struct xdiff_emit_state xm;
3034 SHA_CTX *ctx;
3035 int patchlen;
3038 static int remove_space(char *line, int len)
3040 int i;
3041 char *dst = line;
3042 unsigned char c;
3044 for (i = 0; i < len; i++)
3045 if (!isspace((c = line[i])))
3046 *dst++ = c;
3048 return dst - line;
3051 static void patch_id_consume(void *priv, char *line, unsigned long len)
3053 struct patch_id_t *data = priv;
3054 int new_len;
3056 /* Ignore line numbers when computing the SHA1 of the patch */
3057 if (!prefixcmp(line, "@@ -"))
3058 return;
3060 new_len = remove_space(line, len);
3062 SHA1_Update(data->ctx, line, new_len);
3063 data->patchlen += new_len;
3066 /* returns 0 upon success, and writes result into sha1 */
3067 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3069 struct diff_queue_struct *q = &diff_queued_diff;
3070 int i;
3071 SHA_CTX ctx;
3072 struct patch_id_t data;
3073 char buffer[PATH_MAX * 4 + 20];
3075 SHA1_Init(&ctx);
3076 memset(&data, 0, sizeof(struct patch_id_t));
3077 data.ctx = &ctx;
3078 data.xm.consume = patch_id_consume;
3080 for (i = 0; i < q->nr; i++) {
3081 xpparam_t xpp;
3082 xdemitconf_t xecfg;
3083 xdemitcb_t ecb;
3084 mmfile_t mf1, mf2;
3085 struct diff_filepair *p = q->queue[i];
3086 int len1, len2;
3088 memset(&xecfg, 0, sizeof(xecfg));
3089 if (p->status == 0)
3090 return error("internal diff status error");
3091 if (p->status == DIFF_STATUS_UNKNOWN)
3092 continue;
3093 if (diff_unmodified_pair(p))
3094 continue;
3095 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3096 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3097 continue;
3098 if (DIFF_PAIR_UNMERGED(p))
3099 continue;
3101 diff_fill_sha1_info(p->one);
3102 diff_fill_sha1_info(p->two);
3103 if (fill_mmfile(&mf1, p->one) < 0 ||
3104 fill_mmfile(&mf2, p->two) < 0)
3105 return error("unable to read files to diff");
3107 len1 = remove_space(p->one->path, strlen(p->one->path));
3108 len2 = remove_space(p->two->path, strlen(p->two->path));
3109 if (p->one->mode == 0)
3110 len1 = snprintf(buffer, sizeof(buffer),
3111 "diff--gita/%.*sb/%.*s"
3112 "newfilemode%06o"
3113 "---/dev/null"
3114 "+++b/%.*s",
3115 len1, p->one->path,
3116 len2, p->two->path,
3117 p->two->mode,
3118 len2, p->two->path);
3119 else if (p->two->mode == 0)
3120 len1 = snprintf(buffer, sizeof(buffer),
3121 "diff--gita/%.*sb/%.*s"
3122 "deletedfilemode%06o"
3123 "---a/%.*s"
3124 "+++/dev/null",
3125 len1, p->one->path,
3126 len2, p->two->path,
3127 p->one->mode,
3128 len1, p->one->path);
3129 else
3130 len1 = snprintf(buffer, sizeof(buffer),
3131 "diff--gita/%.*sb/%.*s"
3132 "---a/%.*s"
3133 "+++b/%.*s",
3134 len1, p->one->path,
3135 len2, p->two->path,
3136 len1, p->one->path,
3137 len2, p->two->path);
3138 SHA1_Update(&ctx, buffer, len1);
3140 xpp.flags = XDF_NEED_MINIMAL;
3141 xecfg.ctxlen = 3;
3142 xecfg.flags = XDL_EMIT_FUNCNAMES;
3143 ecb.outf = xdiff_outf;
3144 ecb.priv = &data;
3145 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3148 SHA1_Final(sha1, &ctx);
3149 return 0;
3152 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3154 struct diff_queue_struct *q = &diff_queued_diff;
3155 int i;
3156 int result = diff_get_patch_id(options, sha1);
3158 for (i = 0; i < q->nr; i++)
3159 diff_free_filepair(q->queue[i]);
3161 free(q->queue);
3162 q->queue = NULL;
3163 q->nr = q->alloc = 0;
3165 return result;
3168 static int is_summary_empty(const struct diff_queue_struct *q)
3170 int i;
3172 for (i = 0; i < q->nr; i++) {
3173 const struct diff_filepair *p = q->queue[i];
3175 switch (p->status) {
3176 case DIFF_STATUS_DELETED:
3177 case DIFF_STATUS_ADDED:
3178 case DIFF_STATUS_COPIED:
3179 case DIFF_STATUS_RENAMED:
3180 return 0;
3181 default:
3182 if (p->score)
3183 return 0;
3184 if (p->one->mode && p->two->mode &&
3185 p->one->mode != p->two->mode)
3186 return 0;
3187 break;
3190 return 1;
3193 void diff_flush(struct diff_options *options)
3195 struct diff_queue_struct *q = &diff_queued_diff;
3196 int i, output_format = options->output_format;
3197 int separator = 0;
3200 * Order: raw, stat, summary, patch
3201 * or: name/name-status/checkdiff (other bits clear)
3203 if (!q->nr)
3204 goto free_queue;
3206 if (output_format & (DIFF_FORMAT_RAW |
3207 DIFF_FORMAT_NAME |
3208 DIFF_FORMAT_NAME_STATUS |
3209 DIFF_FORMAT_CHECKDIFF)) {
3210 for (i = 0; i < q->nr; i++) {
3211 struct diff_filepair *p = q->queue[i];
3212 if (check_pair_status(p))
3213 flush_one_pair(p, options);
3215 separator++;
3218 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3219 struct diffstat_t diffstat;
3221 memset(&diffstat, 0, sizeof(struct diffstat_t));
3222 diffstat.xm.consume = diffstat_consume;
3223 for (i = 0; i < q->nr; i++) {
3224 struct diff_filepair *p = q->queue[i];
3225 if (check_pair_status(p))
3226 diff_flush_stat(p, options, &diffstat);
3228 if (output_format & DIFF_FORMAT_NUMSTAT)
3229 show_numstat(&diffstat, options);
3230 if (output_format & DIFF_FORMAT_DIFFSTAT)
3231 show_stats(&diffstat, options);
3232 if (output_format & DIFF_FORMAT_SHORTSTAT)
3233 show_shortstats(&diffstat, options);
3234 free_diffstat_info(&diffstat);
3235 separator++;
3237 if (output_format & DIFF_FORMAT_DIRSTAT)
3238 show_dirstat(options);
3240 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3241 for (i = 0; i < q->nr; i++)
3242 diff_summary(options->file, q->queue[i]);
3243 separator++;
3246 if (output_format & DIFF_FORMAT_PATCH) {
3247 if (separator) {
3248 putc(options->line_termination, options->file);
3249 if (options->stat_sep) {
3250 /* attach patch instead of inline */
3251 fputs(options->stat_sep, options->file);
3255 for (i = 0; i < q->nr; i++) {
3256 struct diff_filepair *p = q->queue[i];
3257 if (check_pair_status(p))
3258 diff_flush_patch(p, options);
3262 if (output_format & DIFF_FORMAT_CALLBACK)
3263 options->format_callback(q, options, options->format_callback_data);
3265 for (i = 0; i < q->nr; i++)
3266 diff_free_filepair(q->queue[i]);
3267 free_queue:
3268 free(q->queue);
3269 q->queue = NULL;
3270 q->nr = q->alloc = 0;
3271 if (options->close_file)
3272 fclose(options->file);
3275 static void diffcore_apply_filter(const char *filter)
3277 int i;
3278 struct diff_queue_struct *q = &diff_queued_diff;
3279 struct diff_queue_struct outq;
3280 outq.queue = NULL;
3281 outq.nr = outq.alloc = 0;
3283 if (!filter)
3284 return;
3286 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3287 int found;
3288 for (i = found = 0; !found && i < q->nr; i++) {
3289 struct diff_filepair *p = q->queue[i];
3290 if (((p->status == DIFF_STATUS_MODIFIED) &&
3291 ((p->score &&
3292 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3293 (!p->score &&
3294 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3295 ((p->status != DIFF_STATUS_MODIFIED) &&
3296 strchr(filter, p->status)))
3297 found++;
3299 if (found)
3300 return;
3302 /* otherwise we will clear the whole queue
3303 * by copying the empty outq at the end of this
3304 * function, but first clear the current entries
3305 * in the queue.
3307 for (i = 0; i < q->nr; i++)
3308 diff_free_filepair(q->queue[i]);
3310 else {
3311 /* Only the matching ones */
3312 for (i = 0; i < q->nr; i++) {
3313 struct diff_filepair *p = q->queue[i];
3315 if (((p->status == DIFF_STATUS_MODIFIED) &&
3316 ((p->score &&
3317 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3318 (!p->score &&
3319 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3320 ((p->status != DIFF_STATUS_MODIFIED) &&
3321 strchr(filter, p->status)))
3322 diff_q(&outq, p);
3323 else
3324 diff_free_filepair(p);
3327 free(q->queue);
3328 *q = outq;
3331 /* Check whether two filespecs with the same mode and size are identical */
3332 static int diff_filespec_is_identical(struct diff_filespec *one,
3333 struct diff_filespec *two)
3335 if (S_ISGITLINK(one->mode))
3336 return 0;
3337 if (diff_populate_filespec(one, 0))
3338 return 0;
3339 if (diff_populate_filespec(two, 0))
3340 return 0;
3341 return !memcmp(one->data, two->data, one->size);
3344 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3346 int i;
3347 struct diff_queue_struct *q = &diff_queued_diff;
3348 struct diff_queue_struct outq;
3349 outq.queue = NULL;
3350 outq.nr = outq.alloc = 0;
3352 for (i = 0; i < q->nr; i++) {
3353 struct diff_filepair *p = q->queue[i];
3356 * 1. Entries that come from stat info dirtyness
3357 * always have both sides (iow, not create/delete),
3358 * one side of the object name is unknown, with
3359 * the same mode and size. Keep the ones that
3360 * do not match these criteria. They have real
3361 * differences.
3363 * 2. At this point, the file is known to be modified,
3364 * with the same mode and size, and the object
3365 * name of one side is unknown. Need to inspect
3366 * the identical contents.
3368 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3369 !DIFF_FILE_VALID(p->two) ||
3370 (p->one->sha1_valid && p->two->sha1_valid) ||
3371 (p->one->mode != p->two->mode) ||
3372 diff_populate_filespec(p->one, 1) ||
3373 diff_populate_filespec(p->two, 1) ||
3374 (p->one->size != p->two->size) ||
3375 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3376 diff_q(&outq, p);
3377 else {
3379 * The caller can subtract 1 from skip_stat_unmatch
3380 * to determine how many paths were dirty only
3381 * due to stat info mismatch.
3383 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3384 diffopt->skip_stat_unmatch++;
3385 diff_free_filepair(p);
3388 free(q->queue);
3389 *q = outq;
3392 void diffcore_std(struct diff_options *options)
3394 if (options->skip_stat_unmatch)
3395 diffcore_skip_stat_unmatch(options);
3396 if (options->break_opt != -1)
3397 diffcore_break(options->break_opt);
3398 if (options->detect_rename)
3399 diffcore_rename(options);
3400 if (options->break_opt != -1)
3401 diffcore_merge_broken();
3402 if (options->pickaxe)
3403 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3404 if (options->orderfile)
3405 diffcore_order(options->orderfile);
3406 diff_resolve_rename_copy();
3407 diffcore_apply_filter(options->filter);
3409 if (diff_queued_diff.nr)
3410 DIFF_OPT_SET(options, HAS_CHANGES);
3411 else
3412 DIFF_OPT_CLR(options, HAS_CHANGES);
3415 int diff_result_code(struct diff_options *opt, int status)
3417 int result = 0;
3418 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3419 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3420 return status;
3421 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3422 DIFF_OPT_TST(opt, HAS_CHANGES))
3423 result |= 01;
3424 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3425 DIFF_OPT_TST(opt, CHECK_FAILED))
3426 result |= 02;
3427 return result;
3430 void diff_addremove(struct diff_options *options,
3431 int addremove, unsigned mode,
3432 const unsigned char *sha1,
3433 const char *concatpath)
3435 struct diff_filespec *one, *two;
3437 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3438 return;
3440 /* This may look odd, but it is a preparation for
3441 * feeding "there are unchanged files which should
3442 * not produce diffs, but when you are doing copy
3443 * detection you would need them, so here they are"
3444 * entries to the diff-core. They will be prefixed
3445 * with something like '=' or '*' (I haven't decided
3446 * which but should not make any difference).
3447 * Feeding the same new and old to diff_change()
3448 * also has the same effect.
3449 * Before the final output happens, they are pruned after
3450 * merged into rename/copy pairs as appropriate.
3452 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3453 addremove = (addremove == '+' ? '-' :
3454 addremove == '-' ? '+' : addremove);
3456 if (options->prefix &&
3457 strncmp(concatpath, options->prefix, options->prefix_length))
3458 return;
3460 one = alloc_filespec(concatpath);
3461 two = alloc_filespec(concatpath);
3463 if (addremove != '+')
3464 fill_filespec(one, sha1, mode);
3465 if (addremove != '-')
3466 fill_filespec(two, sha1, mode);
3468 diff_queue(&diff_queued_diff, one, two);
3469 DIFF_OPT_SET(options, HAS_CHANGES);
3472 void diff_change(struct diff_options *options,
3473 unsigned old_mode, unsigned new_mode,
3474 const unsigned char *old_sha1,
3475 const unsigned char *new_sha1,
3476 const char *concatpath)
3478 struct diff_filespec *one, *two;
3480 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3481 && S_ISGITLINK(new_mode))
3482 return;
3484 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3485 unsigned tmp;
3486 const unsigned char *tmp_c;
3487 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3488 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3491 if (options->prefix &&
3492 strncmp(concatpath, options->prefix, options->prefix_length))
3493 return;
3495 one = alloc_filespec(concatpath);
3496 two = alloc_filespec(concatpath);
3497 fill_filespec(one, old_sha1, old_mode);
3498 fill_filespec(two, new_sha1, new_mode);
3500 diff_queue(&diff_queued_diff, one, two);
3501 DIFF_OPT_SET(options, HAS_CHANGES);
3504 void diff_unmerge(struct diff_options *options,
3505 const char *path,
3506 unsigned mode, const unsigned char *sha1)
3508 struct diff_filespec *one, *two;
3510 if (options->prefix &&
3511 strncmp(path, options->prefix, options->prefix_length))
3512 return;
3514 one = alloc_filespec(path);
3515 two = alloc_filespec(path);
3516 fill_filespec(one, sha1, mode);
3517 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;