Merge branch 'ml/submodule-foreach'
[git/spearce.git] / diff.c
blob5923fe281b6ab421f785a1d47afd6dee5711e52d
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 200;
23 static int diff_suppress_blank_empty;
24 int diff_use_color_default = -1;
25 static const char *external_diff_cmd_cfg;
26 int diff_auto_refresh_index = 1;
28 static char diff_colors[][COLOR_MAXLEN] = {
29 "\033[m", /* reset */
30 "", /* PLAIN (normal) */
31 "\033[1m", /* METAINFO (bold) */
32 "\033[36m", /* FRAGINFO (cyan) */
33 "\033[31m", /* OLD (red) */
34 "\033[32m", /* NEW (green) */
35 "\033[33m", /* COMMIT (yellow) */
36 "\033[41m", /* WHITESPACE (red background) */
39 static int parse_diff_color_slot(const char *var, int ofs)
41 if (!strcasecmp(var+ofs, "plain"))
42 return DIFF_PLAIN;
43 if (!strcasecmp(var+ofs, "meta"))
44 return DIFF_METAINFO;
45 if (!strcasecmp(var+ofs, "frag"))
46 return DIFF_FRAGINFO;
47 if (!strcasecmp(var+ofs, "old"))
48 return DIFF_FILE_OLD;
49 if (!strcasecmp(var+ofs, "new"))
50 return DIFF_FILE_NEW;
51 if (!strcasecmp(var+ofs, "commit"))
52 return DIFF_COMMIT;
53 if (!strcasecmp(var+ofs, "whitespace"))
54 return DIFF_WHITESPACE;
55 die("bad config variable '%s'", var);
58 static struct ll_diff_driver {
59 const char *name;
60 struct ll_diff_driver *next;
61 const char *cmd;
62 } *user_diff, **user_diff_tail;
65 * Currently there is only "diff.<drivername>.command" variable;
66 * because there are "diff.color.<slot>" variables, we are parsing
67 * this in a bit convoluted way to allow low level diff driver
68 * called "color".
70 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
72 const char *name;
73 int namelen;
74 struct ll_diff_driver *drv;
76 name = var + 5;
77 namelen = ep - name;
78 for (drv = user_diff; drv; drv = drv->next)
79 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
80 break;
81 if (!drv) {
82 drv = xcalloc(1, sizeof(struct ll_diff_driver));
83 drv->name = xmemdupz(name, namelen);
84 if (!user_diff_tail)
85 user_diff_tail = &user_diff;
86 *user_diff_tail = drv;
87 user_diff_tail = &(drv->next);
90 return git_config_string(&(drv->cmd), var, value);
94 * 'diff.<what>.funcname' attribute can be specified in the configuration
95 * to define a customized regexp to find the beginning of a function to
96 * be used for hunk header lines of "diff -p" style output.
98 static struct funcname_pattern {
99 char *name;
100 char *pattern;
101 struct funcname_pattern *next;
102 } *funcname_pattern_list;
104 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
106 const char *name;
107 int namelen;
108 struct funcname_pattern *pp;
110 name = var + 5; /* "diff." */
111 namelen = ep - name;
113 for (pp = funcname_pattern_list; pp; pp = pp->next)
114 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
115 break;
116 if (!pp) {
117 pp = xcalloc(1, sizeof(*pp));
118 pp->name = xmemdupz(name, namelen);
119 pp->next = funcname_pattern_list;
120 funcname_pattern_list = pp;
122 free(pp->pattern);
123 pp->pattern = xstrdup(value);
124 return 0;
128 * These are to give UI layer defaults.
129 * The core-level commands such as git-diff-files should
130 * never be affected by the setting of diff.renames
131 * the user happens to have in the configuration file.
133 int git_diff_ui_config(const char *var, const char *value, void *cb)
135 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
136 diff_use_color_default = git_config_colorbool(var, value, -1);
137 return 0;
139 if (!strcmp(var, "diff.renames")) {
140 if (!value)
141 diff_detect_rename_default = DIFF_DETECT_RENAME;
142 else if (!strcasecmp(value, "copies") ||
143 !strcasecmp(value, "copy"))
144 diff_detect_rename_default = DIFF_DETECT_COPY;
145 else if (git_config_bool(var,value))
146 diff_detect_rename_default = DIFF_DETECT_RENAME;
147 return 0;
149 if (!strcmp(var, "diff.autorefreshindex")) {
150 diff_auto_refresh_index = git_config_bool(var, value);
151 return 0;
153 if (!strcmp(var, "diff.external"))
154 return git_config_string(&external_diff_cmd_cfg, var, value);
155 if (!prefixcmp(var, "diff.")) {
156 const char *ep = strrchr(var, '.');
158 if (ep != var + 4 && !strcmp(ep, ".command"))
159 return parse_lldiff_command(var, ep, value);
162 return git_diff_basic_config(var, value, cb);
165 int git_diff_basic_config(const char *var, const char *value, void *cb)
167 if (!strcmp(var, "diff.renamelimit")) {
168 diff_rename_limit_default = git_config_int(var, value);
169 return 0;
172 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
173 int slot = parse_diff_color_slot(var, 11);
174 if (!value)
175 return config_error_nonbool(var);
176 color_parse(value, var, diff_colors[slot]);
177 return 0;
180 /* like GNU diff's --suppress-blank-empty option */
181 if (!strcmp(var, "diff.suppress-blank-empty")) {
182 diff_suppress_blank_empty = git_config_bool(var, value);
183 return 0;
186 if (!prefixcmp(var, "diff.")) {
187 const char *ep = strrchr(var, '.');
188 if (ep != var + 4) {
189 if (!strcmp(ep, ".funcname")) {
190 if (!value)
191 return config_error_nonbool(var);
192 return parse_funcname_pattern(var, ep, value);
197 return git_color_default_config(var, value, cb);
200 static char *quote_two(const char *one, const char *two)
202 int need_one = quote_c_style(one, NULL, NULL, 1);
203 int need_two = quote_c_style(two, NULL, NULL, 1);
204 struct strbuf res;
206 strbuf_init(&res, 0);
207 if (need_one + need_two) {
208 strbuf_addch(&res, '"');
209 quote_c_style(one, &res, NULL, 1);
210 quote_c_style(two, &res, NULL, 1);
211 strbuf_addch(&res, '"');
212 } else {
213 strbuf_addstr(&res, one);
214 strbuf_addstr(&res, two);
216 return strbuf_detach(&res, NULL);
219 static const char *external_diff(void)
221 static const char *external_diff_cmd = NULL;
222 static int done_preparing = 0;
224 if (done_preparing)
225 return external_diff_cmd;
226 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
227 if (!external_diff_cmd)
228 external_diff_cmd = external_diff_cmd_cfg;
229 done_preparing = 1;
230 return external_diff_cmd;
233 static struct diff_tempfile {
234 const char *name; /* filename external diff should read from */
235 char hex[41];
236 char mode[10];
237 char tmp_path[PATH_MAX];
238 } diff_temp[2];
240 static int count_lines(const char *data, int size)
242 int count, ch, completely_empty = 1, nl_just_seen = 0;
243 count = 0;
244 while (0 < size--) {
245 ch = *data++;
246 if (ch == '\n') {
247 count++;
248 nl_just_seen = 1;
249 completely_empty = 0;
251 else {
252 nl_just_seen = 0;
253 completely_empty = 0;
256 if (completely_empty)
257 return 0;
258 if (!nl_just_seen)
259 count++; /* no trailing newline */
260 return count;
263 static void print_line_count(FILE *file, int count)
265 switch (count) {
266 case 0:
267 fprintf(file, "0,0");
268 break;
269 case 1:
270 fprintf(file, "1");
271 break;
272 default:
273 fprintf(file, "1,%d", count);
274 break;
278 static void copy_file_with_prefix(FILE *file,
279 int prefix, const char *data, int size,
280 const char *set, const char *reset)
282 int ch, nl_just_seen = 1;
283 while (0 < size--) {
284 ch = *data++;
285 if (nl_just_seen) {
286 fputs(set, file);
287 putc(prefix, file);
289 if (ch == '\n') {
290 nl_just_seen = 1;
291 fputs(reset, file);
292 } else
293 nl_just_seen = 0;
294 putc(ch, file);
296 if (!nl_just_seen)
297 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
300 static void emit_rewrite_diff(const char *name_a,
301 const char *name_b,
302 struct diff_filespec *one,
303 struct diff_filespec *two,
304 struct diff_options *o)
306 int lc_a, lc_b;
307 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
308 const char *name_a_tab, *name_b_tab;
309 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
310 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
311 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
312 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
313 const char *reset = diff_get_color(color_diff, DIFF_RESET);
314 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
316 name_a += (*name_a == '/');
317 name_b += (*name_b == '/');
318 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
319 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
321 strbuf_reset(&a_name);
322 strbuf_reset(&b_name);
323 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
324 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
326 diff_populate_filespec(one, 0);
327 diff_populate_filespec(two, 0);
328 lc_a = count_lines(one->data, one->size);
329 lc_b = count_lines(two->data, two->size);
330 fprintf(o->file,
331 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
332 metainfo, a_name.buf, name_a_tab, reset,
333 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
334 print_line_count(o->file, lc_a);
335 fprintf(o->file, " +");
336 print_line_count(o->file, lc_b);
337 fprintf(o->file, " @@%s\n", reset);
338 if (lc_a)
339 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
340 if (lc_b)
341 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
344 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
346 if (!DIFF_FILE_VALID(one)) {
347 mf->ptr = (char *)""; /* does not matter */
348 mf->size = 0;
349 return 0;
351 else if (diff_populate_filespec(one, 0))
352 return -1;
353 mf->ptr = one->data;
354 mf->size = one->size;
355 return 0;
358 struct diff_words_buffer {
359 mmfile_t text;
360 long alloc;
361 long current; /* output pointer */
362 int suppressed_newline;
365 static void diff_words_append(char *line, unsigned long len,
366 struct diff_words_buffer *buffer)
368 if (buffer->text.size + len > buffer->alloc) {
369 buffer->alloc = (buffer->text.size + len) * 3 / 2;
370 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
372 line++;
373 len--;
374 memcpy(buffer->text.ptr + buffer->text.size, line, len);
375 buffer->text.size += len;
378 struct diff_words_data {
379 struct diff_words_buffer minus, plus;
380 FILE *file;
383 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
384 int suppress_newline)
386 const char *ptr;
387 int eol = 0;
389 if (len == 0)
390 return;
392 ptr = buffer->text.ptr + buffer->current;
393 buffer->current += len;
395 if (ptr[len - 1] == '\n') {
396 eol = 1;
397 len--;
400 fputs(diff_get_color(1, color), file);
401 fwrite(ptr, len, 1, file);
402 fputs(diff_get_color(1, DIFF_RESET), file);
404 if (eol) {
405 if (suppress_newline)
406 buffer->suppressed_newline = 1;
407 else
408 putc('\n', file);
412 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
414 struct diff_words_data *diff_words = priv;
416 if (diff_words->minus.suppressed_newline) {
417 if (line[0] != '+')
418 putc('\n', diff_words->file);
419 diff_words->minus.suppressed_newline = 0;
422 len--;
423 switch (line[0]) {
424 case '-':
425 print_word(diff_words->file,
426 &diff_words->minus, len, DIFF_FILE_OLD, 1);
427 break;
428 case '+':
429 print_word(diff_words->file,
430 &diff_words->plus, len, DIFF_FILE_NEW, 0);
431 break;
432 case ' ':
433 print_word(diff_words->file,
434 &diff_words->plus, len, DIFF_PLAIN, 0);
435 diff_words->minus.current += len;
436 break;
440 /* this executes the word diff on the accumulated buffers */
441 static void diff_words_show(struct diff_words_data *diff_words)
443 xpparam_t xpp;
444 xdemitconf_t xecfg;
445 xdemitcb_t ecb;
446 mmfile_t minus, plus;
447 int i;
449 memset(&xecfg, 0, sizeof(xecfg));
450 minus.size = diff_words->minus.text.size;
451 minus.ptr = xmalloc(minus.size);
452 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
453 for (i = 0; i < minus.size; i++)
454 if (isspace(minus.ptr[i]))
455 minus.ptr[i] = '\n';
456 diff_words->minus.current = 0;
458 plus.size = diff_words->plus.text.size;
459 plus.ptr = xmalloc(plus.size);
460 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
461 for (i = 0; i < plus.size; i++)
462 if (isspace(plus.ptr[i]))
463 plus.ptr[i] = '\n';
464 diff_words->plus.current = 0;
466 xpp.flags = XDF_NEED_MINIMAL;
467 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
468 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
469 &xpp, &xecfg, &ecb);
470 free(minus.ptr);
471 free(plus.ptr);
472 diff_words->minus.text.size = diff_words->plus.text.size = 0;
474 if (diff_words->minus.suppressed_newline) {
475 putc('\n', diff_words->file);
476 diff_words->minus.suppressed_newline = 0;
480 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
482 struct emit_callback {
483 int nparents, color_diff;
484 unsigned ws_rule;
485 sane_truncate_fn truncate;
486 const char **label_path;
487 struct diff_words_data *diff_words;
488 int *found_changesp;
489 FILE *file;
492 static void free_diff_words_data(struct emit_callback *ecbdata)
494 if (ecbdata->diff_words) {
495 /* flush buffers */
496 if (ecbdata->diff_words->minus.text.size ||
497 ecbdata->diff_words->plus.text.size)
498 diff_words_show(ecbdata->diff_words);
500 free (ecbdata->diff_words->minus.text.ptr);
501 free (ecbdata->diff_words->plus.text.ptr);
502 free(ecbdata->diff_words);
503 ecbdata->diff_words = NULL;
507 const char *diff_get_color(int diff_use_color, enum color_diff ix)
509 if (diff_use_color)
510 return diff_colors[ix];
511 return "";
514 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
516 int has_trailing_newline = (len > 0 && line[len-1] == '\n');
517 if (has_trailing_newline)
518 len--;
520 fputs(set, file);
521 fwrite(line, len, 1, file);
522 fputs(reset, file);
523 if (has_trailing_newline)
524 fputc('\n', file);
527 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
529 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
530 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
532 if (!*ws)
533 emit_line(ecbdata->file, set, reset, line, len);
534 else {
535 /* Emit just the prefix, then the rest. */
536 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
537 ws_check_emit(line + ecbdata->nparents,
538 len - ecbdata->nparents, ecbdata->ws_rule,
539 ecbdata->file, set, reset, ws);
543 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
545 const char *cp;
546 unsigned long allot;
547 size_t l = len;
549 if (ecb->truncate)
550 return ecb->truncate(line, len);
551 cp = line;
552 allot = l;
553 while (0 < l) {
554 (void) utf8_width(&cp, &l);
555 if (!cp)
556 break; /* truncated in the middle? */
558 return allot - l;
561 static void fn_out_consume(void *priv, char *line, unsigned long len)
563 int i;
564 int color;
565 struct emit_callback *ecbdata = priv;
566 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
567 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
568 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
570 *(ecbdata->found_changesp) = 1;
572 if (ecbdata->label_path[0]) {
573 const char *name_a_tab, *name_b_tab;
575 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
576 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
578 fprintf(ecbdata->file, "%s--- %s%s%s\n",
579 meta, ecbdata->label_path[0], reset, name_a_tab);
580 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
581 meta, ecbdata->label_path[1], reset, name_b_tab);
582 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
585 if (diff_suppress_blank_empty
586 && len == 2 && line[0] == ' ' && line[1] == '\n') {
587 line[0] = '\n';
588 len = 1;
591 /* This is not really necessary for now because
592 * this codepath only deals with two-way diffs.
594 for (i = 0; i < len && line[i] == '@'; i++)
596 if (2 <= i && i < len && line[i] == ' ') {
597 ecbdata->nparents = i - 1;
598 len = sane_truncate_line(ecbdata, line, len);
599 emit_line(ecbdata->file,
600 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
601 reset, line, len);
602 if (line[len-1] != '\n')
603 putc('\n', ecbdata->file);
604 return;
607 if (len < ecbdata->nparents) {
608 emit_line(ecbdata->file, reset, reset, line, len);
609 return;
612 color = DIFF_PLAIN;
613 if (ecbdata->diff_words && ecbdata->nparents != 1)
614 /* fall back to normal diff */
615 free_diff_words_data(ecbdata);
616 if (ecbdata->diff_words) {
617 if (line[0] == '-') {
618 diff_words_append(line, len,
619 &ecbdata->diff_words->minus);
620 return;
621 } else if (line[0] == '+') {
622 diff_words_append(line, len,
623 &ecbdata->diff_words->plus);
624 return;
626 if (ecbdata->diff_words->minus.text.size ||
627 ecbdata->diff_words->plus.text.size)
628 diff_words_show(ecbdata->diff_words);
629 line++;
630 len--;
631 emit_line(ecbdata->file, plain, reset, line, len);
632 return;
634 for (i = 0; i < ecbdata->nparents && len; i++) {
635 if (line[i] == '-')
636 color = DIFF_FILE_OLD;
637 else if (line[i] == '+')
638 color = DIFF_FILE_NEW;
641 if (color != DIFF_FILE_NEW) {
642 emit_line(ecbdata->file,
643 diff_get_color(ecbdata->color_diff, color),
644 reset, line, len);
645 return;
647 emit_add_line(reset, ecbdata, line, len);
650 static char *pprint_rename(const char *a, const char *b)
652 const char *old = a;
653 const char *new = b;
654 struct strbuf name;
655 int pfx_length, sfx_length;
656 int len_a = strlen(a);
657 int len_b = strlen(b);
658 int a_midlen, b_midlen;
659 int qlen_a = quote_c_style(a, NULL, NULL, 0);
660 int qlen_b = quote_c_style(b, NULL, NULL, 0);
662 strbuf_init(&name, 0);
663 if (qlen_a || qlen_b) {
664 quote_c_style(a, &name, NULL, 0);
665 strbuf_addstr(&name, " => ");
666 quote_c_style(b, &name, NULL, 0);
667 return strbuf_detach(&name, NULL);
670 /* Find common prefix */
671 pfx_length = 0;
672 while (*old && *new && *old == *new) {
673 if (*old == '/')
674 pfx_length = old - a + 1;
675 old++;
676 new++;
679 /* Find common suffix */
680 old = a + len_a;
681 new = b + len_b;
682 sfx_length = 0;
683 while (a <= old && b <= new && *old == *new) {
684 if (*old == '/')
685 sfx_length = len_a - (old - a);
686 old--;
687 new--;
691 * pfx{mid-a => mid-b}sfx
692 * {pfx-a => pfx-b}sfx
693 * pfx{sfx-a => sfx-b}
694 * name-a => name-b
696 a_midlen = len_a - pfx_length - sfx_length;
697 b_midlen = len_b - pfx_length - sfx_length;
698 if (a_midlen < 0)
699 a_midlen = 0;
700 if (b_midlen < 0)
701 b_midlen = 0;
703 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
704 if (pfx_length + sfx_length) {
705 strbuf_add(&name, a, pfx_length);
706 strbuf_addch(&name, '{');
708 strbuf_add(&name, a + pfx_length, a_midlen);
709 strbuf_addstr(&name, " => ");
710 strbuf_add(&name, b + pfx_length, b_midlen);
711 if (pfx_length + sfx_length) {
712 strbuf_addch(&name, '}');
713 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
715 return strbuf_detach(&name, NULL);
718 struct diffstat_t {
719 int nr;
720 int alloc;
721 struct diffstat_file {
722 char *from_name;
723 char *name;
724 char *print_name;
725 unsigned is_unmerged:1;
726 unsigned is_binary:1;
727 unsigned is_renamed:1;
728 unsigned int added, deleted;
729 } **files;
732 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
733 const char *name_a,
734 const char *name_b)
736 struct diffstat_file *x;
737 x = xcalloc(sizeof (*x), 1);
738 if (diffstat->nr == diffstat->alloc) {
739 diffstat->alloc = alloc_nr(diffstat->alloc);
740 diffstat->files = xrealloc(diffstat->files,
741 diffstat->alloc * sizeof(x));
743 diffstat->files[diffstat->nr++] = x;
744 if (name_b) {
745 x->from_name = xstrdup(name_a);
746 x->name = xstrdup(name_b);
747 x->is_renamed = 1;
749 else {
750 x->from_name = NULL;
751 x->name = xstrdup(name_a);
753 return x;
756 static void diffstat_consume(void *priv, char *line, unsigned long len)
758 struct diffstat_t *diffstat = priv;
759 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
761 if (line[0] == '+')
762 x->added++;
763 else if (line[0] == '-')
764 x->deleted++;
767 const char mime_boundary_leader[] = "------------";
769 static int scale_linear(int it, int width, int max_change)
772 * make sure that at least one '-' is printed if there were deletions,
773 * and likewise for '+'.
775 if (max_change < 2)
776 return it;
777 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
780 static void show_name(FILE *file,
781 const char *prefix, const char *name, int len,
782 const char *reset, const char *set)
784 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
787 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
789 if (cnt <= 0)
790 return;
791 fprintf(file, "%s", set);
792 while (cnt--)
793 putc(ch, file);
794 fprintf(file, "%s", reset);
797 static void fill_print_name(struct diffstat_file *file)
799 char *pname;
801 if (file->print_name)
802 return;
804 if (!file->is_renamed) {
805 struct strbuf buf;
806 strbuf_init(&buf, 0);
807 if (quote_c_style(file->name, &buf, NULL, 0)) {
808 pname = strbuf_detach(&buf, NULL);
809 } else {
810 pname = file->name;
811 strbuf_release(&buf);
813 } else {
814 pname = pprint_rename(file->from_name, file->name);
816 file->print_name = pname;
819 static void show_stats(struct diffstat_t* data, struct diff_options *options)
821 int i, len, add, del, total, adds = 0, dels = 0;
822 int max_change = 0, max_len = 0;
823 int total_files = data->nr;
824 int width, name_width;
825 const char *reset, *set, *add_c, *del_c;
827 if (data->nr == 0)
828 return;
830 width = options->stat_width ? options->stat_width : 80;
831 name_width = options->stat_name_width ? options->stat_name_width : 50;
833 /* Sanity: give at least 5 columns to the graph,
834 * but leave at least 10 columns for the name.
836 if (width < 25)
837 width = 25;
838 if (name_width < 10)
839 name_width = 10;
840 else if (width < name_width + 15)
841 name_width = width - 15;
843 /* Find the longest filename and max number of changes */
844 reset = diff_get_color_opt(options, DIFF_RESET);
845 set = diff_get_color_opt(options, DIFF_PLAIN);
846 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
847 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
849 for (i = 0; i < data->nr; i++) {
850 struct diffstat_file *file = data->files[i];
851 int change = file->added + file->deleted;
852 fill_print_name(file);
853 len = strlen(file->print_name);
854 if (max_len < len)
855 max_len = len;
857 if (file->is_binary || file->is_unmerged)
858 continue;
859 if (max_change < change)
860 max_change = change;
863 /* Compute the width of the graph part;
864 * 10 is for one blank at the beginning of the line plus
865 * " | count " between the name and the graph.
867 * From here on, name_width is the width of the name area,
868 * and width is the width of the graph area.
870 name_width = (name_width < max_len) ? name_width : max_len;
871 if (width < (name_width + 10) + max_change)
872 width = width - (name_width + 10);
873 else
874 width = max_change;
876 for (i = 0; i < data->nr; i++) {
877 const char *prefix = "";
878 char *name = data->files[i]->print_name;
879 int added = data->files[i]->added;
880 int deleted = data->files[i]->deleted;
881 int name_len;
884 * "scale" the filename
886 len = name_width;
887 name_len = strlen(name);
888 if (name_width < name_len) {
889 char *slash;
890 prefix = "...";
891 len -= 3;
892 name += name_len - len;
893 slash = strchr(name, '/');
894 if (slash)
895 name = slash;
898 if (data->files[i]->is_binary) {
899 show_name(options->file, prefix, name, len, reset, set);
900 fprintf(options->file, " Bin ");
901 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
902 fprintf(options->file, " -> ");
903 fprintf(options->file, "%s%d%s", add_c, added, reset);
904 fprintf(options->file, " bytes");
905 fprintf(options->file, "\n");
906 continue;
908 else if (data->files[i]->is_unmerged) {
909 show_name(options->file, prefix, name, len, reset, set);
910 fprintf(options->file, " Unmerged\n");
911 continue;
913 else if (!data->files[i]->is_renamed &&
914 (added + deleted == 0)) {
915 total_files--;
916 continue;
920 * scale the add/delete
922 add = added;
923 del = deleted;
924 total = add + del;
925 adds += add;
926 dels += del;
928 if (width <= max_change) {
929 add = scale_linear(add, width, max_change);
930 del = scale_linear(del, width, max_change);
931 total = add + del;
933 show_name(options->file, prefix, name, len, reset, set);
934 fprintf(options->file, "%5d%s", added + deleted,
935 added + deleted ? " " : "");
936 show_graph(options->file, '+', add, add_c, reset);
937 show_graph(options->file, '-', del, del_c, reset);
938 fprintf(options->file, "\n");
940 fprintf(options->file,
941 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
942 set, total_files, adds, dels, reset);
945 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
947 int i, adds = 0, dels = 0, total_files = data->nr;
949 if (data->nr == 0)
950 return;
952 for (i = 0; i < data->nr; i++) {
953 if (!data->files[i]->is_binary &&
954 !data->files[i]->is_unmerged) {
955 int added = data->files[i]->added;
956 int deleted= data->files[i]->deleted;
957 if (!data->files[i]->is_renamed &&
958 (added + deleted == 0)) {
959 total_files--;
960 } else {
961 adds += added;
962 dels += deleted;
966 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
967 total_files, adds, dels);
970 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
972 int i;
974 if (data->nr == 0)
975 return;
977 for (i = 0; i < data->nr; i++) {
978 struct diffstat_file *file = data->files[i];
980 if (file->is_binary)
981 fprintf(options->file, "-\t-\t");
982 else
983 fprintf(options->file,
984 "%d\t%d\t", file->added, file->deleted);
985 if (options->line_termination) {
986 fill_print_name(file);
987 if (!file->is_renamed)
988 write_name_quoted(file->name, options->file,
989 options->line_termination);
990 else {
991 fputs(file->print_name, options->file);
992 putc(options->line_termination, options->file);
994 } else {
995 if (file->is_renamed) {
996 putc('\0', options->file);
997 write_name_quoted(file->from_name, options->file, '\0');
999 write_name_quoted(file->name, options->file, '\0');
1004 struct dirstat_file {
1005 const char *name;
1006 unsigned long changed;
1009 struct dirstat_dir {
1010 struct dirstat_file *files;
1011 int alloc, nr, percent, cumulative;
1014 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1016 unsigned long this_dir = 0;
1017 unsigned int sources = 0;
1019 while (dir->nr) {
1020 struct dirstat_file *f = dir->files;
1021 int namelen = strlen(f->name);
1022 unsigned long this;
1023 char *slash;
1025 if (namelen < baselen)
1026 break;
1027 if (memcmp(f->name, base, baselen))
1028 break;
1029 slash = strchr(f->name + baselen, '/');
1030 if (slash) {
1031 int newbaselen = slash + 1 - f->name;
1032 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1033 sources++;
1034 } else {
1035 this = f->changed;
1036 dir->files++;
1037 dir->nr--;
1038 sources += 2;
1040 this_dir += this;
1044 * We don't report dirstat's for
1045 * - the top level
1046 * - or cases where everything came from a single directory
1047 * under this directory (sources == 1).
1049 if (baselen && sources != 1) {
1050 int permille = this_dir * 1000 / changed;
1051 if (permille) {
1052 int percent = permille / 10;
1053 if (percent >= dir->percent) {
1054 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1055 if (!dir->cumulative)
1056 return 0;
1060 return this_dir;
1063 static void show_dirstat(struct diff_options *options)
1065 int i;
1066 unsigned long changed;
1067 struct dirstat_dir dir;
1068 struct diff_queue_struct *q = &diff_queued_diff;
1070 dir.files = NULL;
1071 dir.alloc = 0;
1072 dir.nr = 0;
1073 dir.percent = options->dirstat_percent;
1074 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1076 changed = 0;
1077 for (i = 0; i < q->nr; i++) {
1078 struct diff_filepair *p = q->queue[i];
1079 const char *name;
1080 unsigned long copied, added, damage;
1082 name = p->one->path ? p->one->path : p->two->path;
1084 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1085 diff_populate_filespec(p->one, 0);
1086 diff_populate_filespec(p->two, 0);
1087 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1088 &copied, &added);
1089 diff_free_filespec_data(p->one);
1090 diff_free_filespec_data(p->two);
1091 } else if (DIFF_FILE_VALID(p->one)) {
1092 diff_populate_filespec(p->one, 1);
1093 copied = added = 0;
1094 diff_free_filespec_data(p->one);
1095 } else if (DIFF_FILE_VALID(p->two)) {
1096 diff_populate_filespec(p->two, 1);
1097 copied = 0;
1098 added = p->two->size;
1099 diff_free_filespec_data(p->two);
1100 } else
1101 continue;
1104 * Original minus copied is the removed material,
1105 * added is the new material. They are both damages
1106 * made to the preimage.
1108 damage = (p->one->size - copied) + added;
1110 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1111 dir.files[dir.nr].name = name;
1112 dir.files[dir.nr].changed = damage;
1113 changed += damage;
1114 dir.nr++;
1117 /* This can happen even with many files, if everything was renames */
1118 if (!changed)
1119 return;
1121 /* Show all directories with more than x% of the changes */
1122 gather_dirstat(options->file, &dir, changed, "", 0);
1125 static void free_diffstat_info(struct diffstat_t *diffstat)
1127 int i;
1128 for (i = 0; i < diffstat->nr; i++) {
1129 struct diffstat_file *f = diffstat->files[i];
1130 if (f->name != f->print_name)
1131 free(f->print_name);
1132 free(f->name);
1133 free(f->from_name);
1134 free(f);
1136 free(diffstat->files);
1139 struct checkdiff_t {
1140 const char *filename;
1141 int lineno;
1142 struct diff_options *o;
1143 unsigned ws_rule;
1144 unsigned status;
1145 int trailing_blanks_start;
1148 static int is_conflict_marker(const char *line, unsigned long len)
1150 char firstchar;
1151 int cnt;
1153 if (len < 8)
1154 return 0;
1155 firstchar = line[0];
1156 switch (firstchar) {
1157 case '=': case '>': case '<':
1158 break;
1159 default:
1160 return 0;
1162 for (cnt = 1; cnt < 7; cnt++)
1163 if (line[cnt] != firstchar)
1164 return 0;
1165 /* line[0] thru line[6] are same as firstchar */
1166 if (firstchar == '=') {
1167 /* divider between ours and theirs? */
1168 if (len != 8 || line[7] != '\n')
1169 return 0;
1170 } else if (len < 8 || !isspace(line[7])) {
1171 /* not divider before ours nor after theirs */
1172 return 0;
1174 return 1;
1177 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1179 struct checkdiff_t *data = priv;
1180 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1181 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1182 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1183 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1184 char *err;
1186 if (line[0] == '+') {
1187 unsigned bad;
1188 data->lineno++;
1189 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1190 data->trailing_blanks_start = 0;
1191 else if (!data->trailing_blanks_start)
1192 data->trailing_blanks_start = data->lineno;
1193 if (is_conflict_marker(line + 1, len - 1)) {
1194 data->status |= 1;
1195 fprintf(data->o->file,
1196 "%s:%d: leftover conflict marker\n",
1197 data->filename, data->lineno);
1199 bad = ws_check(line + 1, len - 1, data->ws_rule);
1200 if (!bad)
1201 return;
1202 data->status |= bad;
1203 err = whitespace_error_string(bad);
1204 fprintf(data->o->file, "%s:%d: %s.\n",
1205 data->filename, data->lineno, err);
1206 free(err);
1207 emit_line(data->o->file, set, reset, line, 1);
1208 ws_check_emit(line + 1, len - 1, data->ws_rule,
1209 data->o->file, set, reset, ws);
1210 } else if (line[0] == ' ') {
1211 data->lineno++;
1212 data->trailing_blanks_start = 0;
1213 } else if (line[0] == '@') {
1214 char *plus = strchr(line, '+');
1215 if (plus)
1216 data->lineno = strtol(plus, NULL, 10) - 1;
1217 else
1218 die("invalid diff");
1219 data->trailing_blanks_start = 0;
1223 static unsigned char *deflate_it(char *data,
1224 unsigned long size,
1225 unsigned long *result_size)
1227 int bound;
1228 unsigned char *deflated;
1229 z_stream stream;
1231 memset(&stream, 0, sizeof(stream));
1232 deflateInit(&stream, zlib_compression_level);
1233 bound = deflateBound(&stream, size);
1234 deflated = xmalloc(bound);
1235 stream.next_out = deflated;
1236 stream.avail_out = bound;
1238 stream.next_in = (unsigned char *)data;
1239 stream.avail_in = size;
1240 while (deflate(&stream, Z_FINISH) == Z_OK)
1241 ; /* nothing */
1242 deflateEnd(&stream);
1243 *result_size = stream.total_out;
1244 return deflated;
1247 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1249 void *cp;
1250 void *delta;
1251 void *deflated;
1252 void *data;
1253 unsigned long orig_size;
1254 unsigned long delta_size;
1255 unsigned long deflate_size;
1256 unsigned long data_size;
1258 /* We could do deflated delta, or we could do just deflated two,
1259 * whichever is smaller.
1261 delta = NULL;
1262 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1263 if (one->size && two->size) {
1264 delta = diff_delta(one->ptr, one->size,
1265 two->ptr, two->size,
1266 &delta_size, deflate_size);
1267 if (delta) {
1268 void *to_free = delta;
1269 orig_size = delta_size;
1270 delta = deflate_it(delta, delta_size, &delta_size);
1271 free(to_free);
1275 if (delta && delta_size < deflate_size) {
1276 fprintf(file, "delta %lu\n", orig_size);
1277 free(deflated);
1278 data = delta;
1279 data_size = delta_size;
1281 else {
1282 fprintf(file, "literal %lu\n", two->size);
1283 free(delta);
1284 data = deflated;
1285 data_size = deflate_size;
1288 /* emit data encoded in base85 */
1289 cp = data;
1290 while (data_size) {
1291 int bytes = (52 < data_size) ? 52 : data_size;
1292 char line[70];
1293 data_size -= bytes;
1294 if (bytes <= 26)
1295 line[0] = bytes + 'A' - 1;
1296 else
1297 line[0] = bytes - 26 + 'a' - 1;
1298 encode_85(line + 1, cp, bytes);
1299 cp = (char *) cp + bytes;
1300 fputs(line, file);
1301 fputc('\n', file);
1303 fprintf(file, "\n");
1304 free(data);
1307 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1309 fprintf(file, "GIT binary patch\n");
1310 emit_binary_diff_body(file, one, two);
1311 emit_binary_diff_body(file, two, one);
1314 static void setup_diff_attr_check(struct git_attr_check *check)
1316 static struct git_attr *attr_diff;
1318 if (!attr_diff) {
1319 attr_diff = git_attr("diff", 4);
1321 check[0].attr = attr_diff;
1324 static void diff_filespec_check_attr(struct diff_filespec *one)
1326 struct git_attr_check attr_diff_check;
1327 int check_from_data = 0;
1329 if (one->checked_attr)
1330 return;
1332 setup_diff_attr_check(&attr_diff_check);
1333 one->is_binary = 0;
1334 one->funcname_pattern_ident = NULL;
1336 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1337 const char *value;
1339 /* binaryness */
1340 value = attr_diff_check.value;
1341 if (ATTR_TRUE(value))
1343 else if (ATTR_FALSE(value))
1344 one->is_binary = 1;
1345 else
1346 check_from_data = 1;
1348 /* funcname pattern ident */
1349 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1351 else
1352 one->funcname_pattern_ident = value;
1355 if (check_from_data) {
1356 if (!one->data && DIFF_FILE_VALID(one))
1357 diff_populate_filespec(one, 0);
1359 if (one->data)
1360 one->is_binary = buffer_is_binary(one->data, one->size);
1364 int diff_filespec_is_binary(struct diff_filespec *one)
1366 diff_filespec_check_attr(one);
1367 return one->is_binary;
1370 static const char *funcname_pattern(const char *ident)
1372 struct funcname_pattern *pp;
1374 for (pp = funcname_pattern_list; pp; pp = pp->next)
1375 if (!strcmp(ident, pp->name))
1376 return pp->pattern;
1377 return NULL;
1380 static struct builtin_funcname_pattern {
1381 const char *name;
1382 const char *pattern;
1383 } builtin_funcname_pattern[] = {
1384 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1385 "new\\|return\\|switch\\|throw\\|while\\)\n"
1386 "^[ ]*\\(\\([ ]*"
1387 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1388 "[ ]*([^;]*\\)$" },
1389 { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1390 "destructor\\|interface\\|implementation\\|"
1391 "initialization\\|finalization\\)[ \t]*.*\\)$"
1392 "\\|"
1393 "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1395 { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1396 { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1397 { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1400 static const char *diff_funcname_pattern(struct diff_filespec *one)
1402 const char *ident, *pattern;
1403 int i;
1405 diff_filespec_check_attr(one);
1406 ident = one->funcname_pattern_ident;
1408 if (!ident)
1410 * If the config file has "funcname.default" defined, that
1411 * regexp is used; otherwise NULL is returned and xemit uses
1412 * the built-in default.
1414 return funcname_pattern("default");
1416 /* Look up custom "funcname.$ident" regexp from config. */
1417 pattern = funcname_pattern(ident);
1418 if (pattern)
1419 return pattern;
1422 * And define built-in fallback patterns here. Note that
1423 * these can be overridden by the user's config settings.
1425 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1426 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1427 return builtin_funcname_pattern[i].pattern;
1429 return NULL;
1432 static void builtin_diff(const char *name_a,
1433 const char *name_b,
1434 struct diff_filespec *one,
1435 struct diff_filespec *two,
1436 const char *xfrm_msg,
1437 struct diff_options *o,
1438 int complete_rewrite)
1440 mmfile_t mf1, mf2;
1441 const char *lbl[2];
1442 char *a_one, *b_two;
1443 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1444 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1446 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1447 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1448 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1449 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1450 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1451 if (lbl[0][0] == '/') {
1452 /* /dev/null */
1453 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1454 if (xfrm_msg && xfrm_msg[0])
1455 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1457 else if (lbl[1][0] == '/') {
1458 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1459 if (xfrm_msg && xfrm_msg[0])
1460 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1462 else {
1463 if (one->mode != two->mode) {
1464 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1465 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1467 if (xfrm_msg && xfrm_msg[0])
1468 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1470 * we do not run diff between different kind
1471 * of objects.
1473 if ((one->mode ^ two->mode) & S_IFMT)
1474 goto free_ab_and_return;
1475 if (complete_rewrite) {
1476 emit_rewrite_diff(name_a, name_b, one, two, o);
1477 o->found_changes = 1;
1478 goto free_ab_and_return;
1482 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1483 die("unable to read files to diff");
1485 if (!DIFF_OPT_TST(o, TEXT) &&
1486 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1487 /* Quite common confusing case */
1488 if (mf1.size == mf2.size &&
1489 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1490 goto free_ab_and_return;
1491 if (DIFF_OPT_TST(o, BINARY))
1492 emit_binary_diff(o->file, &mf1, &mf2);
1493 else
1494 fprintf(o->file, "Binary files %s and %s differ\n",
1495 lbl[0], lbl[1]);
1496 o->found_changes = 1;
1498 else {
1499 /* Crazy xdl interfaces.. */
1500 const char *diffopts = getenv("GIT_DIFF_OPTS");
1501 xpparam_t xpp;
1502 xdemitconf_t xecfg;
1503 xdemitcb_t ecb;
1504 struct emit_callback ecbdata;
1505 const char *funcname_pattern;
1507 funcname_pattern = diff_funcname_pattern(one);
1508 if (!funcname_pattern)
1509 funcname_pattern = diff_funcname_pattern(two);
1511 memset(&xecfg, 0, sizeof(xecfg));
1512 memset(&ecbdata, 0, sizeof(ecbdata));
1513 ecbdata.label_path = lbl;
1514 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1515 ecbdata.found_changesp = &o->found_changes;
1516 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1517 ecbdata.file = o->file;
1518 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1519 xecfg.ctxlen = o->context;
1520 xecfg.flags = XDL_EMIT_FUNCNAMES;
1521 if (funcname_pattern)
1522 xdiff_set_find_func(&xecfg, funcname_pattern);
1523 if (!diffopts)
1525 else if (!prefixcmp(diffopts, "--unified="))
1526 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1527 else if (!prefixcmp(diffopts, "-u"))
1528 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1529 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1530 ecbdata.diff_words =
1531 xcalloc(1, sizeof(struct diff_words_data));
1532 ecbdata.diff_words->file = o->file;
1534 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1535 &xpp, &xecfg, &ecb);
1536 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1537 free_diff_words_data(&ecbdata);
1540 free_ab_and_return:
1541 diff_free_filespec_data(one);
1542 diff_free_filespec_data(two);
1543 free(a_one);
1544 free(b_two);
1545 return;
1548 static void builtin_diffstat(const char *name_a, const char *name_b,
1549 struct diff_filespec *one,
1550 struct diff_filespec *two,
1551 struct diffstat_t *diffstat,
1552 struct diff_options *o,
1553 int complete_rewrite)
1555 mmfile_t mf1, mf2;
1556 struct diffstat_file *data;
1558 data = diffstat_add(diffstat, name_a, name_b);
1560 if (!one || !two) {
1561 data->is_unmerged = 1;
1562 return;
1564 if (complete_rewrite) {
1565 diff_populate_filespec(one, 0);
1566 diff_populate_filespec(two, 0);
1567 data->deleted = count_lines(one->data, one->size);
1568 data->added = count_lines(two->data, two->size);
1569 goto free_and_return;
1571 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1572 die("unable to read files to diff");
1574 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1575 data->is_binary = 1;
1576 data->added = mf2.size;
1577 data->deleted = mf1.size;
1578 } else {
1579 /* Crazy xdl interfaces.. */
1580 xpparam_t xpp;
1581 xdemitconf_t xecfg;
1582 xdemitcb_t ecb;
1584 memset(&xecfg, 0, sizeof(xecfg));
1585 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1586 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1587 &xpp, &xecfg, &ecb);
1590 free_and_return:
1591 diff_free_filespec_data(one);
1592 diff_free_filespec_data(two);
1595 static void builtin_checkdiff(const char *name_a, const char *name_b,
1596 const char *attr_path,
1597 struct diff_filespec *one,
1598 struct diff_filespec *two,
1599 struct diff_options *o)
1601 mmfile_t mf1, mf2;
1602 struct checkdiff_t data;
1604 if (!two)
1605 return;
1607 memset(&data, 0, sizeof(data));
1608 data.filename = name_b ? name_b : name_a;
1609 data.lineno = 0;
1610 data.o = o;
1611 data.ws_rule = whitespace_rule(attr_path);
1613 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1614 die("unable to read files to diff");
1617 * All the other codepaths check both sides, but not checking
1618 * the "old" side here is deliberate. We are checking the newly
1619 * introduced changes, and as long as the "new" side is text, we
1620 * can and should check what it introduces.
1622 if (diff_filespec_is_binary(two))
1623 goto free_and_return;
1624 else {
1625 /* Crazy xdl interfaces.. */
1626 xpparam_t xpp;
1627 xdemitconf_t xecfg;
1628 xdemitcb_t ecb;
1630 memset(&xecfg, 0, sizeof(xecfg));
1631 xecfg.ctxlen = 1; /* at least one context line */
1632 xpp.flags = XDF_NEED_MINIMAL;
1633 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1634 &xpp, &xecfg, &ecb);
1636 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1637 data.trailing_blanks_start) {
1638 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1639 data.filename, data.trailing_blanks_start);
1640 data.status = 1; /* report errors */
1643 free_and_return:
1644 diff_free_filespec_data(one);
1645 diff_free_filespec_data(two);
1646 if (data.status)
1647 DIFF_OPT_SET(o, CHECK_FAILED);
1650 struct diff_filespec *alloc_filespec(const char *path)
1652 int namelen = strlen(path);
1653 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1655 memset(spec, 0, sizeof(*spec));
1656 spec->path = (char *)(spec + 1);
1657 memcpy(spec->path, path, namelen+1);
1658 spec->count = 1;
1659 return spec;
1662 void free_filespec(struct diff_filespec *spec)
1664 if (!--spec->count) {
1665 diff_free_filespec_data(spec);
1666 free(spec);
1670 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1671 unsigned short mode)
1673 if (mode) {
1674 spec->mode = canon_mode(mode);
1675 hashcpy(spec->sha1, sha1);
1676 spec->sha1_valid = !is_null_sha1(sha1);
1681 * Given a name and sha1 pair, if the index tells us the file in
1682 * the work tree has that object contents, return true, so that
1683 * prepare_temp_file() does not have to inflate and extract.
1685 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1687 struct cache_entry *ce;
1688 struct stat st;
1689 int pos, len;
1691 /* We do not read the cache ourselves here, because the
1692 * benchmark with my previous version that always reads cache
1693 * shows that it makes things worse for diff-tree comparing
1694 * two linux-2.6 kernel trees in an already checked out work
1695 * tree. This is because most diff-tree comparisons deal with
1696 * only a small number of files, while reading the cache is
1697 * expensive for a large project, and its cost outweighs the
1698 * savings we get by not inflating the object to a temporary
1699 * file. Practically, this code only helps when we are used
1700 * by diff-cache --cached, which does read the cache before
1701 * calling us.
1703 if (!active_cache)
1704 return 0;
1706 /* We want to avoid the working directory if our caller
1707 * doesn't need the data in a normal file, this system
1708 * is rather slow with its stat/open/mmap/close syscalls,
1709 * and the object is contained in a pack file. The pack
1710 * is probably already open and will be faster to obtain
1711 * the data through than the working directory. Loose
1712 * objects however would tend to be slower as they need
1713 * to be individually opened and inflated.
1715 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1716 return 0;
1718 len = strlen(name);
1719 pos = cache_name_pos(name, len);
1720 if (pos < 0)
1721 return 0;
1722 ce = active_cache[pos];
1725 * This is not the sha1 we are looking for, or
1726 * unreusable because it is not a regular file.
1728 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1729 return 0;
1732 * If ce matches the file in the work tree, we can reuse it.
1734 if (ce_uptodate(ce) ||
1735 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1736 return 1;
1738 return 0;
1741 static int populate_from_stdin(struct diff_filespec *s)
1743 struct strbuf buf;
1744 size_t size = 0;
1746 strbuf_init(&buf, 0);
1747 if (strbuf_read(&buf, 0, 0) < 0)
1748 return error("error while reading from stdin %s",
1749 strerror(errno));
1751 s->should_munmap = 0;
1752 s->data = strbuf_detach(&buf, &size);
1753 s->size = size;
1754 s->should_free = 1;
1755 return 0;
1758 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1760 int len;
1761 char *data = xmalloc(100);
1762 len = snprintf(data, 100,
1763 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1764 s->data = data;
1765 s->size = len;
1766 s->should_free = 1;
1767 if (size_only) {
1768 s->data = NULL;
1769 free(data);
1771 return 0;
1775 * While doing rename detection and pickaxe operation, we may need to
1776 * grab the data for the blob (or file) for our own in-core comparison.
1777 * diff_filespec has data and size fields for this purpose.
1779 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1781 int err = 0;
1782 if (!DIFF_FILE_VALID(s))
1783 die("internal error: asking to populate invalid file.");
1784 if (S_ISDIR(s->mode))
1785 return -1;
1787 if (s->data)
1788 return 0;
1790 if (size_only && 0 < s->size)
1791 return 0;
1793 if (S_ISGITLINK(s->mode))
1794 return diff_populate_gitlink(s, size_only);
1796 if (!s->sha1_valid ||
1797 reuse_worktree_file(s->path, s->sha1, 0)) {
1798 struct strbuf buf;
1799 struct stat st;
1800 int fd;
1802 if (!strcmp(s->path, "-"))
1803 return populate_from_stdin(s);
1805 if (lstat(s->path, &st) < 0) {
1806 if (errno == ENOENT) {
1807 err_empty:
1808 err = -1;
1809 empty:
1810 s->data = (char *)"";
1811 s->size = 0;
1812 return err;
1815 s->size = xsize_t(st.st_size);
1816 if (!s->size)
1817 goto empty;
1818 if (size_only)
1819 return 0;
1820 if (S_ISLNK(st.st_mode)) {
1821 int ret;
1822 s->data = xmalloc(s->size);
1823 s->should_free = 1;
1824 ret = readlink(s->path, s->data, s->size);
1825 if (ret < 0) {
1826 free(s->data);
1827 goto err_empty;
1829 return 0;
1831 fd = open(s->path, O_RDONLY);
1832 if (fd < 0)
1833 goto err_empty;
1834 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1835 close(fd);
1836 s->should_munmap = 1;
1839 * Convert from working tree format to canonical git format
1841 strbuf_init(&buf, 0);
1842 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1843 size_t size = 0;
1844 munmap(s->data, s->size);
1845 s->should_munmap = 0;
1846 s->data = strbuf_detach(&buf, &size);
1847 s->size = size;
1848 s->should_free = 1;
1851 else {
1852 enum object_type type;
1853 if (size_only)
1854 type = sha1_object_info(s->sha1, &s->size);
1855 else {
1856 s->data = read_sha1_file(s->sha1, &type, &s->size);
1857 s->should_free = 1;
1860 return 0;
1863 void diff_free_filespec_blob(struct diff_filespec *s)
1865 if (s->should_free)
1866 free(s->data);
1867 else if (s->should_munmap)
1868 munmap(s->data, s->size);
1870 if (s->should_free || s->should_munmap) {
1871 s->should_free = s->should_munmap = 0;
1872 s->data = NULL;
1876 void diff_free_filespec_data(struct diff_filespec *s)
1878 diff_free_filespec_blob(s);
1879 free(s->cnt_data);
1880 s->cnt_data = NULL;
1883 static void prep_temp_blob(struct diff_tempfile *temp,
1884 void *blob,
1885 unsigned long size,
1886 const unsigned char *sha1,
1887 int mode)
1889 int fd;
1891 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1892 if (fd < 0)
1893 die("unable to create temp-file: %s", strerror(errno));
1894 if (write_in_full(fd, blob, size) != size)
1895 die("unable to write temp-file");
1896 close(fd);
1897 temp->name = temp->tmp_path;
1898 strcpy(temp->hex, sha1_to_hex(sha1));
1899 temp->hex[40] = 0;
1900 sprintf(temp->mode, "%06o", mode);
1903 static void prepare_temp_file(const char *name,
1904 struct diff_tempfile *temp,
1905 struct diff_filespec *one)
1907 if (!DIFF_FILE_VALID(one)) {
1908 not_a_valid_file:
1909 /* A '-' entry produces this for file-2, and
1910 * a '+' entry produces this for file-1.
1912 temp->name = "/dev/null";
1913 strcpy(temp->hex, ".");
1914 strcpy(temp->mode, ".");
1915 return;
1918 if (!one->sha1_valid ||
1919 reuse_worktree_file(name, one->sha1, 1)) {
1920 struct stat st;
1921 if (lstat(name, &st) < 0) {
1922 if (errno == ENOENT)
1923 goto not_a_valid_file;
1924 die("stat(%s): %s", name, strerror(errno));
1926 if (S_ISLNK(st.st_mode)) {
1927 int ret;
1928 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1929 size_t sz = xsize_t(st.st_size);
1930 if (sizeof(buf) <= st.st_size)
1931 die("symlink too long: %s", name);
1932 ret = readlink(name, buf, sz);
1933 if (ret < 0)
1934 die("readlink(%s)", name);
1935 prep_temp_blob(temp, buf, sz,
1936 (one->sha1_valid ?
1937 one->sha1 : null_sha1),
1938 (one->sha1_valid ?
1939 one->mode : S_IFLNK));
1941 else {
1942 /* we can borrow from the file in the work tree */
1943 temp->name = name;
1944 if (!one->sha1_valid)
1945 strcpy(temp->hex, sha1_to_hex(null_sha1));
1946 else
1947 strcpy(temp->hex, sha1_to_hex(one->sha1));
1948 /* Even though we may sometimes borrow the
1949 * contents from the work tree, we always want
1950 * one->mode. mode is trustworthy even when
1951 * !(one->sha1_valid), as long as
1952 * DIFF_FILE_VALID(one).
1954 sprintf(temp->mode, "%06o", one->mode);
1956 return;
1958 else {
1959 if (diff_populate_filespec(one, 0))
1960 die("cannot read data blob for %s", one->path);
1961 prep_temp_blob(temp, one->data, one->size,
1962 one->sha1, one->mode);
1966 static void remove_tempfile(void)
1968 int i;
1970 for (i = 0; i < 2; i++)
1971 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1972 unlink(diff_temp[i].name);
1973 diff_temp[i].name = NULL;
1977 static void remove_tempfile_on_signal(int signo)
1979 remove_tempfile();
1980 signal(SIGINT, SIG_DFL);
1981 raise(signo);
1984 /* An external diff command takes:
1986 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1987 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1990 static void run_external_diff(const char *pgm,
1991 const char *name,
1992 const char *other,
1993 struct diff_filespec *one,
1994 struct diff_filespec *two,
1995 const char *xfrm_msg,
1996 int complete_rewrite)
1998 const char *spawn_arg[10];
1999 struct diff_tempfile *temp = diff_temp;
2000 int retval;
2001 static int atexit_asked = 0;
2002 const char *othername;
2003 const char **arg = &spawn_arg[0];
2005 othername = (other? other : name);
2006 if (one && two) {
2007 prepare_temp_file(name, &temp[0], one);
2008 prepare_temp_file(othername, &temp[1], two);
2009 if (! atexit_asked &&
2010 (temp[0].name == temp[0].tmp_path ||
2011 temp[1].name == temp[1].tmp_path)) {
2012 atexit_asked = 1;
2013 atexit(remove_tempfile);
2015 signal(SIGINT, remove_tempfile_on_signal);
2018 if (one && two) {
2019 *arg++ = pgm;
2020 *arg++ = name;
2021 *arg++ = temp[0].name;
2022 *arg++ = temp[0].hex;
2023 *arg++ = temp[0].mode;
2024 *arg++ = temp[1].name;
2025 *arg++ = temp[1].hex;
2026 *arg++ = temp[1].mode;
2027 if (other) {
2028 *arg++ = other;
2029 *arg++ = xfrm_msg;
2031 } else {
2032 *arg++ = pgm;
2033 *arg++ = name;
2035 *arg = NULL;
2036 fflush(NULL);
2037 retval = run_command_v_opt(spawn_arg, 0);
2038 remove_tempfile();
2039 if (retval) {
2040 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2041 exit(1);
2045 static const char *external_diff_attr(const char *name)
2047 struct git_attr_check attr_diff_check;
2049 if (!name)
2050 return NULL;
2052 setup_diff_attr_check(&attr_diff_check);
2053 if (!git_checkattr(name, 1, &attr_diff_check)) {
2054 const char *value = attr_diff_check.value;
2055 if (!ATTR_TRUE(value) &&
2056 !ATTR_FALSE(value) &&
2057 !ATTR_UNSET(value)) {
2058 struct ll_diff_driver *drv;
2060 for (drv = user_diff; drv; drv = drv->next)
2061 if (!strcmp(drv->name, value))
2062 return drv->cmd;
2065 return NULL;
2068 static void run_diff_cmd(const char *pgm,
2069 const char *name,
2070 const char *other,
2071 const char *attr_path,
2072 struct diff_filespec *one,
2073 struct diff_filespec *two,
2074 const char *xfrm_msg,
2075 struct diff_options *o,
2076 int complete_rewrite)
2078 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2079 pgm = NULL;
2080 else {
2081 const char *cmd = external_diff_attr(attr_path);
2082 if (cmd)
2083 pgm = cmd;
2086 if (pgm) {
2087 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2088 complete_rewrite);
2089 return;
2091 if (one && two)
2092 builtin_diff(name, other ? other : name,
2093 one, two, xfrm_msg, o, complete_rewrite);
2094 else
2095 fprintf(o->file, "* Unmerged path %s\n", name);
2098 static void diff_fill_sha1_info(struct diff_filespec *one)
2100 if (DIFF_FILE_VALID(one)) {
2101 if (!one->sha1_valid) {
2102 struct stat st;
2103 if (!strcmp(one->path, "-")) {
2104 hashcpy(one->sha1, null_sha1);
2105 return;
2107 if (lstat(one->path, &st) < 0)
2108 die("stat %s", one->path);
2109 if (index_path(one->sha1, one->path, &st, 0))
2110 die("cannot hash %s\n", one->path);
2113 else
2114 hashclr(one->sha1);
2117 static int similarity_index(struct diff_filepair *p)
2119 return p->score * 100 / MAX_SCORE;
2122 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2124 /* Strip the prefix but do not molest /dev/null and absolute paths */
2125 if (*namep && **namep != '/')
2126 *namep += prefix_length;
2127 if (*otherp && **otherp != '/')
2128 *otherp += prefix_length;
2131 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2133 const char *pgm = external_diff();
2134 struct strbuf msg;
2135 char *xfrm_msg;
2136 struct diff_filespec *one = p->one;
2137 struct diff_filespec *two = p->two;
2138 const char *name;
2139 const char *other;
2140 const char *attr_path;
2141 int complete_rewrite = 0;
2143 name = p->one->path;
2144 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2145 attr_path = name;
2146 if (o->prefix_length)
2147 strip_prefix(o->prefix_length, &name, &other);
2149 if (DIFF_PAIR_UNMERGED(p)) {
2150 run_diff_cmd(pgm, name, NULL, attr_path,
2151 NULL, NULL, NULL, o, 0);
2152 return;
2155 diff_fill_sha1_info(one);
2156 diff_fill_sha1_info(two);
2158 strbuf_init(&msg, PATH_MAX * 2 + 300);
2159 switch (p->status) {
2160 case DIFF_STATUS_COPIED:
2161 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2162 strbuf_addstr(&msg, "\ncopy from ");
2163 quote_c_style(name, &msg, NULL, 0);
2164 strbuf_addstr(&msg, "\ncopy to ");
2165 quote_c_style(other, &msg, NULL, 0);
2166 strbuf_addch(&msg, '\n');
2167 break;
2168 case DIFF_STATUS_RENAMED:
2169 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2170 strbuf_addstr(&msg, "\nrename from ");
2171 quote_c_style(name, &msg, NULL, 0);
2172 strbuf_addstr(&msg, "\nrename to ");
2173 quote_c_style(other, &msg, NULL, 0);
2174 strbuf_addch(&msg, '\n');
2175 break;
2176 case DIFF_STATUS_MODIFIED:
2177 if (p->score) {
2178 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2179 similarity_index(p));
2180 complete_rewrite = 1;
2181 break;
2183 /* fallthru */
2184 default:
2185 /* nothing */
2189 if (hashcmp(one->sha1, two->sha1)) {
2190 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2192 if (DIFF_OPT_TST(o, BINARY)) {
2193 mmfile_t mf;
2194 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2195 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2196 abbrev = 40;
2198 strbuf_addf(&msg, "index %.*s..%.*s",
2199 abbrev, sha1_to_hex(one->sha1),
2200 abbrev, sha1_to_hex(two->sha1));
2201 if (one->mode == two->mode)
2202 strbuf_addf(&msg, " %06o", one->mode);
2203 strbuf_addch(&msg, '\n');
2206 if (msg.len)
2207 strbuf_setlen(&msg, msg.len - 1);
2208 xfrm_msg = msg.len ? msg.buf : NULL;
2210 if (!pgm &&
2211 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2212 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2213 /* a filepair that changes between file and symlink
2214 * needs to be split into deletion and creation.
2216 struct diff_filespec *null = alloc_filespec(two->path);
2217 run_diff_cmd(NULL, name, other, attr_path,
2218 one, null, xfrm_msg, o, 0);
2219 free(null);
2220 null = alloc_filespec(one->path);
2221 run_diff_cmd(NULL, name, other, attr_path,
2222 null, two, xfrm_msg, o, 0);
2223 free(null);
2225 else
2226 run_diff_cmd(pgm, name, other, attr_path,
2227 one, two, xfrm_msg, o, complete_rewrite);
2229 strbuf_release(&msg);
2232 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2233 struct diffstat_t *diffstat)
2235 const char *name;
2236 const char *other;
2237 int complete_rewrite = 0;
2239 if (DIFF_PAIR_UNMERGED(p)) {
2240 /* unmerged */
2241 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2242 return;
2245 name = p->one->path;
2246 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2248 if (o->prefix_length)
2249 strip_prefix(o->prefix_length, &name, &other);
2251 diff_fill_sha1_info(p->one);
2252 diff_fill_sha1_info(p->two);
2254 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2255 complete_rewrite = 1;
2256 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2259 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2261 const char *name;
2262 const char *other;
2263 const char *attr_path;
2265 if (DIFF_PAIR_UNMERGED(p)) {
2266 /* unmerged */
2267 return;
2270 name = p->one->path;
2271 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2272 attr_path = other ? other : name;
2274 if (o->prefix_length)
2275 strip_prefix(o->prefix_length, &name, &other);
2277 diff_fill_sha1_info(p->one);
2278 diff_fill_sha1_info(p->two);
2280 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2283 void diff_setup(struct diff_options *options)
2285 memset(options, 0, sizeof(*options));
2287 options->file = stdout;
2289 options->line_termination = '\n';
2290 options->break_opt = -1;
2291 options->rename_limit = -1;
2292 options->dirstat_percent = 3;
2293 options->context = 3;
2295 options->change = diff_change;
2296 options->add_remove = diff_addremove;
2297 if (diff_use_color_default > 0)
2298 DIFF_OPT_SET(options, COLOR_DIFF);
2299 else
2300 DIFF_OPT_CLR(options, COLOR_DIFF);
2301 options->detect_rename = diff_detect_rename_default;
2303 options->a_prefix = "a/";
2304 options->b_prefix = "b/";
2307 int diff_setup_done(struct diff_options *options)
2309 int count = 0;
2311 if (options->output_format & DIFF_FORMAT_NAME)
2312 count++;
2313 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2314 count++;
2315 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2316 count++;
2317 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2318 count++;
2319 if (count > 1)
2320 die("--name-only, --name-status, --check and -s are mutually exclusive");
2322 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2323 options->detect_rename = DIFF_DETECT_COPY;
2325 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2326 options->prefix = NULL;
2327 if (options->prefix)
2328 options->prefix_length = strlen(options->prefix);
2329 else
2330 options->prefix_length = 0;
2332 if (options->output_format & (DIFF_FORMAT_NAME |
2333 DIFF_FORMAT_NAME_STATUS |
2334 DIFF_FORMAT_CHECKDIFF |
2335 DIFF_FORMAT_NO_OUTPUT))
2336 options->output_format &= ~(DIFF_FORMAT_RAW |
2337 DIFF_FORMAT_NUMSTAT |
2338 DIFF_FORMAT_DIFFSTAT |
2339 DIFF_FORMAT_SHORTSTAT |
2340 DIFF_FORMAT_DIRSTAT |
2341 DIFF_FORMAT_SUMMARY |
2342 DIFF_FORMAT_PATCH);
2345 * These cases always need recursive; we do not drop caller-supplied
2346 * recursive bits for other formats here.
2348 if (options->output_format & (DIFF_FORMAT_PATCH |
2349 DIFF_FORMAT_NUMSTAT |
2350 DIFF_FORMAT_DIFFSTAT |
2351 DIFF_FORMAT_SHORTSTAT |
2352 DIFF_FORMAT_DIRSTAT |
2353 DIFF_FORMAT_SUMMARY |
2354 DIFF_FORMAT_CHECKDIFF))
2355 DIFF_OPT_SET(options, RECURSIVE);
2357 * Also pickaxe would not work very well if you do not say recursive
2359 if (options->pickaxe)
2360 DIFF_OPT_SET(options, RECURSIVE);
2362 if (options->detect_rename && options->rename_limit < 0)
2363 options->rename_limit = diff_rename_limit_default;
2364 if (options->setup & DIFF_SETUP_USE_CACHE) {
2365 if (!active_cache)
2366 /* read-cache does not die even when it fails
2367 * so it is safe for us to do this here. Also
2368 * it does not smudge active_cache or active_nr
2369 * when it fails, so we do not have to worry about
2370 * cleaning it up ourselves either.
2372 read_cache();
2374 if (options->abbrev <= 0 || 40 < options->abbrev)
2375 options->abbrev = 40; /* full */
2378 * It does not make sense to show the first hit we happened
2379 * to have found. It does not make sense not to return with
2380 * exit code in such a case either.
2382 if (DIFF_OPT_TST(options, QUIET)) {
2383 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2384 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2388 * If we postprocess in diffcore, we cannot simply return
2389 * upon the first hit. We need to run diff as usual.
2391 if (options->pickaxe || options->filter)
2392 DIFF_OPT_CLR(options, QUIET);
2394 return 0;
2397 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2399 char c, *eq;
2400 int len;
2402 if (*arg != '-')
2403 return 0;
2404 c = *++arg;
2405 if (!c)
2406 return 0;
2407 if (c == arg_short) {
2408 c = *++arg;
2409 if (!c)
2410 return 1;
2411 if (val && isdigit(c)) {
2412 char *end;
2413 int n = strtoul(arg, &end, 10);
2414 if (*end)
2415 return 0;
2416 *val = n;
2417 return 1;
2419 return 0;
2421 if (c != '-')
2422 return 0;
2423 arg++;
2424 eq = strchr(arg, '=');
2425 if (eq)
2426 len = eq - arg;
2427 else
2428 len = strlen(arg);
2429 if (!len || strncmp(arg, arg_long, len))
2430 return 0;
2431 if (eq) {
2432 int n;
2433 char *end;
2434 if (!isdigit(*++eq))
2435 return 0;
2436 n = strtoul(eq, &end, 10);
2437 if (*end)
2438 return 0;
2439 *val = n;
2441 return 1;
2444 static int diff_scoreopt_parse(const char *opt);
2446 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2448 const char *arg = av[0];
2450 /* Output format options */
2451 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2452 options->output_format |= DIFF_FORMAT_PATCH;
2453 else if (opt_arg(arg, 'U', "unified", &options->context))
2454 options->output_format |= DIFF_FORMAT_PATCH;
2455 else if (!strcmp(arg, "--raw"))
2456 options->output_format |= DIFF_FORMAT_RAW;
2457 else if (!strcmp(arg, "--patch-with-raw"))
2458 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2459 else if (!strcmp(arg, "--numstat"))
2460 options->output_format |= DIFF_FORMAT_NUMSTAT;
2461 else if (!strcmp(arg, "--shortstat"))
2462 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2463 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2464 options->output_format |= DIFF_FORMAT_DIRSTAT;
2465 else if (!strcmp(arg, "--cumulative"))
2466 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2467 else if (!strcmp(arg, "--check"))
2468 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2469 else if (!strcmp(arg, "--summary"))
2470 options->output_format |= DIFF_FORMAT_SUMMARY;
2471 else if (!strcmp(arg, "--patch-with-stat"))
2472 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2473 else if (!strcmp(arg, "--name-only"))
2474 options->output_format |= DIFF_FORMAT_NAME;
2475 else if (!strcmp(arg, "--name-status"))
2476 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2477 else if (!strcmp(arg, "-s"))
2478 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2479 else if (!prefixcmp(arg, "--stat")) {
2480 char *end;
2481 int width = options->stat_width;
2482 int name_width = options->stat_name_width;
2483 arg += 6;
2484 end = (char *)arg;
2486 switch (*arg) {
2487 case '-':
2488 if (!prefixcmp(arg, "-width="))
2489 width = strtoul(arg + 7, &end, 10);
2490 else if (!prefixcmp(arg, "-name-width="))
2491 name_width = strtoul(arg + 12, &end, 10);
2492 break;
2493 case '=':
2494 width = strtoul(arg+1, &end, 10);
2495 if (*end == ',')
2496 name_width = strtoul(end+1, &end, 10);
2499 /* Important! This checks all the error cases! */
2500 if (*end)
2501 return 0;
2502 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2503 options->stat_name_width = name_width;
2504 options->stat_width = width;
2507 /* renames options */
2508 else if (!prefixcmp(arg, "-B")) {
2509 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2510 return -1;
2512 else if (!prefixcmp(arg, "-M")) {
2513 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2514 return -1;
2515 options->detect_rename = DIFF_DETECT_RENAME;
2517 else if (!prefixcmp(arg, "-C")) {
2518 if (options->detect_rename == DIFF_DETECT_COPY)
2519 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2520 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2521 return -1;
2522 options->detect_rename = DIFF_DETECT_COPY;
2524 else if (!strcmp(arg, "--no-renames"))
2525 options->detect_rename = 0;
2526 else if (!strcmp(arg, "--relative"))
2527 DIFF_OPT_SET(options, RELATIVE_NAME);
2528 else if (!prefixcmp(arg, "--relative=")) {
2529 DIFF_OPT_SET(options, RELATIVE_NAME);
2530 options->prefix = arg + 11;
2533 /* xdiff options */
2534 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2535 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2536 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2537 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2538 else if (!strcmp(arg, "--ignore-space-at-eol"))
2539 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2541 /* flags options */
2542 else if (!strcmp(arg, "--binary")) {
2543 options->output_format |= DIFF_FORMAT_PATCH;
2544 DIFF_OPT_SET(options, BINARY);
2546 else if (!strcmp(arg, "--full-index"))
2547 DIFF_OPT_SET(options, FULL_INDEX);
2548 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2549 DIFF_OPT_SET(options, TEXT);
2550 else if (!strcmp(arg, "-R"))
2551 DIFF_OPT_SET(options, REVERSE_DIFF);
2552 else if (!strcmp(arg, "--find-copies-harder"))
2553 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2554 else if (!strcmp(arg, "--follow"))
2555 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2556 else if (!strcmp(arg, "--color"))
2557 DIFF_OPT_SET(options, COLOR_DIFF);
2558 else if (!strcmp(arg, "--no-color"))
2559 DIFF_OPT_CLR(options, COLOR_DIFF);
2560 else if (!strcmp(arg, "--color-words"))
2561 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2562 else if (!strcmp(arg, "--exit-code"))
2563 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2564 else if (!strcmp(arg, "--quiet"))
2565 DIFF_OPT_SET(options, QUIET);
2566 else if (!strcmp(arg, "--ext-diff"))
2567 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2568 else if (!strcmp(arg, "--no-ext-diff"))
2569 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2570 else if (!strcmp(arg, "--ignore-submodules"))
2571 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2573 /* misc options */
2574 else if (!strcmp(arg, "-z"))
2575 options->line_termination = 0;
2576 else if (!prefixcmp(arg, "-l"))
2577 options->rename_limit = strtoul(arg+2, NULL, 10);
2578 else if (!prefixcmp(arg, "-S"))
2579 options->pickaxe = arg + 2;
2580 else if (!strcmp(arg, "--pickaxe-all"))
2581 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2582 else if (!strcmp(arg, "--pickaxe-regex"))
2583 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2584 else if (!prefixcmp(arg, "-O"))
2585 options->orderfile = arg + 2;
2586 else if (!prefixcmp(arg, "--diff-filter="))
2587 options->filter = arg + 14;
2588 else if (!strcmp(arg, "--abbrev"))
2589 options->abbrev = DEFAULT_ABBREV;
2590 else if (!prefixcmp(arg, "--abbrev=")) {
2591 options->abbrev = strtoul(arg + 9, NULL, 10);
2592 if (options->abbrev < MINIMUM_ABBREV)
2593 options->abbrev = MINIMUM_ABBREV;
2594 else if (40 < options->abbrev)
2595 options->abbrev = 40;
2597 else if (!prefixcmp(arg, "--src-prefix="))
2598 options->a_prefix = arg + 13;
2599 else if (!prefixcmp(arg, "--dst-prefix="))
2600 options->b_prefix = arg + 13;
2601 else if (!strcmp(arg, "--no-prefix"))
2602 options->a_prefix = options->b_prefix = "";
2603 else if (!prefixcmp(arg, "--output=")) {
2604 options->file = fopen(arg + strlen("--output="), "w");
2605 options->close_file = 1;
2606 } else
2607 return 0;
2608 return 1;
2611 static int parse_num(const char **cp_p)
2613 unsigned long num, scale;
2614 int ch, dot;
2615 const char *cp = *cp_p;
2617 num = 0;
2618 scale = 1;
2619 dot = 0;
2620 for(;;) {
2621 ch = *cp;
2622 if ( !dot && ch == '.' ) {
2623 scale = 1;
2624 dot = 1;
2625 } else if ( ch == '%' ) {
2626 scale = dot ? scale*100 : 100;
2627 cp++; /* % is always at the end */
2628 break;
2629 } else if ( ch >= '0' && ch <= '9' ) {
2630 if ( scale < 100000 ) {
2631 scale *= 10;
2632 num = (num*10) + (ch-'0');
2634 } else {
2635 break;
2637 cp++;
2639 *cp_p = cp;
2641 /* user says num divided by scale and we say internally that
2642 * is MAX_SCORE * num / scale.
2644 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2647 static int diff_scoreopt_parse(const char *opt)
2649 int opt1, opt2, cmd;
2651 if (*opt++ != '-')
2652 return -1;
2653 cmd = *opt++;
2654 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2655 return -1; /* that is not a -M, -C nor -B option */
2657 opt1 = parse_num(&opt);
2658 if (cmd != 'B')
2659 opt2 = 0;
2660 else {
2661 if (*opt == 0)
2662 opt2 = 0;
2663 else if (*opt != '/')
2664 return -1; /* we expect -B80/99 or -B80 */
2665 else {
2666 opt++;
2667 opt2 = parse_num(&opt);
2670 if (*opt != 0)
2671 return -1;
2672 return opt1 | (opt2 << 16);
2675 struct diff_queue_struct diff_queued_diff;
2677 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2679 if (queue->alloc <= queue->nr) {
2680 queue->alloc = alloc_nr(queue->alloc);
2681 queue->queue = xrealloc(queue->queue,
2682 sizeof(dp) * queue->alloc);
2684 queue->queue[queue->nr++] = dp;
2687 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2688 struct diff_filespec *one,
2689 struct diff_filespec *two)
2691 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2692 dp->one = one;
2693 dp->two = two;
2694 if (queue)
2695 diff_q(queue, dp);
2696 return dp;
2699 void diff_free_filepair(struct diff_filepair *p)
2701 free_filespec(p->one);
2702 free_filespec(p->two);
2703 free(p);
2706 /* This is different from find_unique_abbrev() in that
2707 * it stuffs the result with dots for alignment.
2709 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2711 int abblen;
2712 const char *abbrev;
2713 if (len == 40)
2714 return sha1_to_hex(sha1);
2716 abbrev = find_unique_abbrev(sha1, len);
2717 abblen = strlen(abbrev);
2718 if (abblen < 37) {
2719 static char hex[41];
2720 if (len < abblen && abblen <= len + 2)
2721 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2722 else
2723 sprintf(hex, "%s...", abbrev);
2724 return hex;
2726 return sha1_to_hex(sha1);
2729 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2731 int line_termination = opt->line_termination;
2732 int inter_name_termination = line_termination ? '\t' : '\0';
2734 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2735 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2736 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2737 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2739 if (p->score) {
2740 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2741 inter_name_termination);
2742 } else {
2743 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2746 if (p->status == DIFF_STATUS_COPIED ||
2747 p->status == DIFF_STATUS_RENAMED) {
2748 const char *name_a, *name_b;
2749 name_a = p->one->path;
2750 name_b = p->two->path;
2751 strip_prefix(opt->prefix_length, &name_a, &name_b);
2752 write_name_quoted(name_a, opt->file, inter_name_termination);
2753 write_name_quoted(name_b, opt->file, line_termination);
2754 } else {
2755 const char *name_a, *name_b;
2756 name_a = p->one->mode ? p->one->path : p->two->path;
2757 name_b = NULL;
2758 strip_prefix(opt->prefix_length, &name_a, &name_b);
2759 write_name_quoted(name_a, opt->file, line_termination);
2763 int diff_unmodified_pair(struct diff_filepair *p)
2765 /* This function is written stricter than necessary to support
2766 * the currently implemented transformers, but the idea is to
2767 * let transformers to produce diff_filepairs any way they want,
2768 * and filter and clean them up here before producing the output.
2770 struct diff_filespec *one = p->one, *two = p->two;
2772 if (DIFF_PAIR_UNMERGED(p))
2773 return 0; /* unmerged is interesting */
2775 /* deletion, addition, mode or type change
2776 * and rename are all interesting.
2778 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2779 DIFF_PAIR_MODE_CHANGED(p) ||
2780 strcmp(one->path, two->path))
2781 return 0;
2783 /* both are valid and point at the same path. that is, we are
2784 * dealing with a change.
2786 if (one->sha1_valid && two->sha1_valid &&
2787 !hashcmp(one->sha1, two->sha1))
2788 return 1; /* no change */
2789 if (!one->sha1_valid && !two->sha1_valid)
2790 return 1; /* both look at the same file on the filesystem. */
2791 return 0;
2794 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2796 if (diff_unmodified_pair(p))
2797 return;
2799 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2800 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2801 return; /* no tree diffs in patch format */
2803 run_diff(p, o);
2806 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2807 struct diffstat_t *diffstat)
2809 if (diff_unmodified_pair(p))
2810 return;
2812 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2813 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2814 return; /* no tree diffs in patch format */
2816 run_diffstat(p, o, diffstat);
2819 static void diff_flush_checkdiff(struct diff_filepair *p,
2820 struct diff_options *o)
2822 if (diff_unmodified_pair(p))
2823 return;
2825 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2826 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2827 return; /* no tree diffs in patch format */
2829 run_checkdiff(p, o);
2832 int diff_queue_is_empty(void)
2834 struct diff_queue_struct *q = &diff_queued_diff;
2835 int i;
2836 for (i = 0; i < q->nr; i++)
2837 if (!diff_unmodified_pair(q->queue[i]))
2838 return 0;
2839 return 1;
2842 #if DIFF_DEBUG
2843 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2845 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2846 x, one ? one : "",
2847 s->path,
2848 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2849 s->mode,
2850 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2851 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2852 x, one ? one : "",
2853 s->size, s->xfrm_flags);
2856 void diff_debug_filepair(const struct diff_filepair *p, int i)
2858 diff_debug_filespec(p->one, i, "one");
2859 diff_debug_filespec(p->two, i, "two");
2860 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2861 p->score, p->status ? p->status : '?',
2862 p->one->rename_used, p->broken_pair);
2865 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2867 int i;
2868 if (msg)
2869 fprintf(stderr, "%s\n", msg);
2870 fprintf(stderr, "q->nr = %d\n", q->nr);
2871 for (i = 0; i < q->nr; i++) {
2872 struct diff_filepair *p = q->queue[i];
2873 diff_debug_filepair(p, i);
2876 #endif
2878 static void diff_resolve_rename_copy(void)
2880 int i;
2881 struct diff_filepair *p;
2882 struct diff_queue_struct *q = &diff_queued_diff;
2884 diff_debug_queue("resolve-rename-copy", q);
2886 for (i = 0; i < q->nr; i++) {
2887 p = q->queue[i];
2888 p->status = 0; /* undecided */
2889 if (DIFF_PAIR_UNMERGED(p))
2890 p->status = DIFF_STATUS_UNMERGED;
2891 else if (!DIFF_FILE_VALID(p->one))
2892 p->status = DIFF_STATUS_ADDED;
2893 else if (!DIFF_FILE_VALID(p->two))
2894 p->status = DIFF_STATUS_DELETED;
2895 else if (DIFF_PAIR_TYPE_CHANGED(p))
2896 p->status = DIFF_STATUS_TYPE_CHANGED;
2898 /* from this point on, we are dealing with a pair
2899 * whose both sides are valid and of the same type, i.e.
2900 * either in-place edit or rename/copy edit.
2902 else if (DIFF_PAIR_RENAME(p)) {
2904 * A rename might have re-connected a broken
2905 * pair up, causing the pathnames to be the
2906 * same again. If so, that's not a rename at
2907 * all, just a modification..
2909 * Otherwise, see if this source was used for
2910 * multiple renames, in which case we decrement
2911 * the count, and call it a copy.
2913 if (!strcmp(p->one->path, p->two->path))
2914 p->status = DIFF_STATUS_MODIFIED;
2915 else if (--p->one->rename_used > 0)
2916 p->status = DIFF_STATUS_COPIED;
2917 else
2918 p->status = DIFF_STATUS_RENAMED;
2920 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2921 p->one->mode != p->two->mode ||
2922 is_null_sha1(p->one->sha1))
2923 p->status = DIFF_STATUS_MODIFIED;
2924 else {
2925 /* This is a "no-change" entry and should not
2926 * happen anymore, but prepare for broken callers.
2928 error("feeding unmodified %s to diffcore",
2929 p->one->path);
2930 p->status = DIFF_STATUS_UNKNOWN;
2933 diff_debug_queue("resolve-rename-copy done", q);
2936 static int check_pair_status(struct diff_filepair *p)
2938 switch (p->status) {
2939 case DIFF_STATUS_UNKNOWN:
2940 return 0;
2941 case 0:
2942 die("internal error in diff-resolve-rename-copy");
2943 default:
2944 return 1;
2948 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2950 int fmt = opt->output_format;
2952 if (fmt & DIFF_FORMAT_CHECKDIFF)
2953 diff_flush_checkdiff(p, opt);
2954 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2955 diff_flush_raw(p, opt);
2956 else if (fmt & DIFF_FORMAT_NAME) {
2957 const char *name_a, *name_b;
2958 name_a = p->two->path;
2959 name_b = NULL;
2960 strip_prefix(opt->prefix_length, &name_a, &name_b);
2961 write_name_quoted(name_a, opt->file, opt->line_termination);
2965 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2967 if (fs->mode)
2968 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2969 else
2970 fprintf(file, " %s ", newdelete);
2971 write_name_quoted(fs->path, file, '\n');
2975 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2977 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2978 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2979 show_name ? ' ' : '\n');
2980 if (show_name) {
2981 write_name_quoted(p->two->path, file, '\n');
2986 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2988 char *names = pprint_rename(p->one->path, p->two->path);
2990 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2991 free(names);
2992 show_mode_change(file, p, 0);
2995 static void diff_summary(FILE *file, struct diff_filepair *p)
2997 switch(p->status) {
2998 case DIFF_STATUS_DELETED:
2999 show_file_mode_name(file, "delete", p->one);
3000 break;
3001 case DIFF_STATUS_ADDED:
3002 show_file_mode_name(file, "create", p->two);
3003 break;
3004 case DIFF_STATUS_COPIED:
3005 show_rename_copy(file, "copy", p);
3006 break;
3007 case DIFF_STATUS_RENAMED:
3008 show_rename_copy(file, "rename", p);
3009 break;
3010 default:
3011 if (p->score) {
3012 fputs(" rewrite ", file);
3013 write_name_quoted(p->two->path, file, ' ');
3014 fprintf(file, "(%d%%)\n", similarity_index(p));
3016 show_mode_change(file, p, !p->score);
3017 break;
3021 struct patch_id_t {
3022 SHA_CTX *ctx;
3023 int patchlen;
3026 static int remove_space(char *line, int len)
3028 int i;
3029 char *dst = line;
3030 unsigned char c;
3032 for (i = 0; i < len; i++)
3033 if (!isspace((c = line[i])))
3034 *dst++ = c;
3036 return dst - line;
3039 static void patch_id_consume(void *priv, char *line, unsigned long len)
3041 struct patch_id_t *data = priv;
3042 int new_len;
3044 /* Ignore line numbers when computing the SHA1 of the patch */
3045 if (!prefixcmp(line, "@@ -"))
3046 return;
3048 new_len = remove_space(line, len);
3050 SHA1_Update(data->ctx, line, new_len);
3051 data->patchlen += new_len;
3054 /* returns 0 upon success, and writes result into sha1 */
3055 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3057 struct diff_queue_struct *q = &diff_queued_diff;
3058 int i;
3059 SHA_CTX ctx;
3060 struct patch_id_t data;
3061 char buffer[PATH_MAX * 4 + 20];
3063 SHA1_Init(&ctx);
3064 memset(&data, 0, sizeof(struct patch_id_t));
3065 data.ctx = &ctx;
3067 for (i = 0; i < q->nr; i++) {
3068 xpparam_t xpp;
3069 xdemitconf_t xecfg;
3070 xdemitcb_t ecb;
3071 mmfile_t mf1, mf2;
3072 struct diff_filepair *p = q->queue[i];
3073 int len1, len2;
3075 memset(&xecfg, 0, sizeof(xecfg));
3076 if (p->status == 0)
3077 return error("internal diff status error");
3078 if (p->status == DIFF_STATUS_UNKNOWN)
3079 continue;
3080 if (diff_unmodified_pair(p))
3081 continue;
3082 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3083 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3084 continue;
3085 if (DIFF_PAIR_UNMERGED(p))
3086 continue;
3088 diff_fill_sha1_info(p->one);
3089 diff_fill_sha1_info(p->two);
3090 if (fill_mmfile(&mf1, p->one) < 0 ||
3091 fill_mmfile(&mf2, p->two) < 0)
3092 return error("unable to read files to diff");
3094 len1 = remove_space(p->one->path, strlen(p->one->path));
3095 len2 = remove_space(p->two->path, strlen(p->two->path));
3096 if (p->one->mode == 0)
3097 len1 = snprintf(buffer, sizeof(buffer),
3098 "diff--gita/%.*sb/%.*s"
3099 "newfilemode%06o"
3100 "---/dev/null"
3101 "+++b/%.*s",
3102 len1, p->one->path,
3103 len2, p->two->path,
3104 p->two->mode,
3105 len2, p->two->path);
3106 else if (p->two->mode == 0)
3107 len1 = snprintf(buffer, sizeof(buffer),
3108 "diff--gita/%.*sb/%.*s"
3109 "deletedfilemode%06o"
3110 "---a/%.*s"
3111 "+++/dev/null",
3112 len1, p->one->path,
3113 len2, p->two->path,
3114 p->one->mode,
3115 len1, p->one->path);
3116 else
3117 len1 = snprintf(buffer, sizeof(buffer),
3118 "diff--gita/%.*sb/%.*s"
3119 "---a/%.*s"
3120 "+++b/%.*s",
3121 len1, p->one->path,
3122 len2, p->two->path,
3123 len1, p->one->path,
3124 len2, p->two->path);
3125 SHA1_Update(&ctx, buffer, len1);
3127 xpp.flags = XDF_NEED_MINIMAL;
3128 xecfg.ctxlen = 3;
3129 xecfg.flags = XDL_EMIT_FUNCNAMES;
3130 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3131 &xpp, &xecfg, &ecb);
3134 SHA1_Final(sha1, &ctx);
3135 return 0;
3138 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3140 struct diff_queue_struct *q = &diff_queued_diff;
3141 int i;
3142 int result = diff_get_patch_id(options, sha1);
3144 for (i = 0; i < q->nr; i++)
3145 diff_free_filepair(q->queue[i]);
3147 free(q->queue);
3148 q->queue = NULL;
3149 q->nr = q->alloc = 0;
3151 return result;
3154 static int is_summary_empty(const struct diff_queue_struct *q)
3156 int i;
3158 for (i = 0; i < q->nr; i++) {
3159 const struct diff_filepair *p = q->queue[i];
3161 switch (p->status) {
3162 case DIFF_STATUS_DELETED:
3163 case DIFF_STATUS_ADDED:
3164 case DIFF_STATUS_COPIED:
3165 case DIFF_STATUS_RENAMED:
3166 return 0;
3167 default:
3168 if (p->score)
3169 return 0;
3170 if (p->one->mode && p->two->mode &&
3171 p->one->mode != p->two->mode)
3172 return 0;
3173 break;
3176 return 1;
3179 void diff_flush(struct diff_options *options)
3181 struct diff_queue_struct *q = &diff_queued_diff;
3182 int i, output_format = options->output_format;
3183 int separator = 0;
3186 * Order: raw, stat, summary, patch
3187 * or: name/name-status/checkdiff (other bits clear)
3189 if (!q->nr)
3190 goto free_queue;
3192 if (output_format & (DIFF_FORMAT_RAW |
3193 DIFF_FORMAT_NAME |
3194 DIFF_FORMAT_NAME_STATUS |
3195 DIFF_FORMAT_CHECKDIFF)) {
3196 for (i = 0; i < q->nr; i++) {
3197 struct diff_filepair *p = q->queue[i];
3198 if (check_pair_status(p))
3199 flush_one_pair(p, options);
3201 separator++;
3204 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3205 struct diffstat_t diffstat;
3207 memset(&diffstat, 0, sizeof(struct diffstat_t));
3208 for (i = 0; i < q->nr; i++) {
3209 struct diff_filepair *p = q->queue[i];
3210 if (check_pair_status(p))
3211 diff_flush_stat(p, options, &diffstat);
3213 if (output_format & DIFF_FORMAT_NUMSTAT)
3214 show_numstat(&diffstat, options);
3215 if (output_format & DIFF_FORMAT_DIFFSTAT)
3216 show_stats(&diffstat, options);
3217 if (output_format & DIFF_FORMAT_SHORTSTAT)
3218 show_shortstats(&diffstat, options);
3219 free_diffstat_info(&diffstat);
3220 separator++;
3222 if (output_format & DIFF_FORMAT_DIRSTAT)
3223 show_dirstat(options);
3225 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3226 for (i = 0; i < q->nr; i++)
3227 diff_summary(options->file, q->queue[i]);
3228 separator++;
3231 if (output_format & DIFF_FORMAT_PATCH) {
3232 if (separator) {
3233 putc(options->line_termination, options->file);
3234 if (options->stat_sep) {
3235 /* attach patch instead of inline */
3236 fputs(options->stat_sep, options->file);
3240 for (i = 0; i < q->nr; i++) {
3241 struct diff_filepair *p = q->queue[i];
3242 if (check_pair_status(p))
3243 diff_flush_patch(p, options);
3247 if (output_format & DIFF_FORMAT_CALLBACK)
3248 options->format_callback(q, options, options->format_callback_data);
3250 for (i = 0; i < q->nr; i++)
3251 diff_free_filepair(q->queue[i]);
3252 free_queue:
3253 free(q->queue);
3254 q->queue = NULL;
3255 q->nr = q->alloc = 0;
3256 if (options->close_file)
3257 fclose(options->file);
3260 static void diffcore_apply_filter(const char *filter)
3262 int i;
3263 struct diff_queue_struct *q = &diff_queued_diff;
3264 struct diff_queue_struct outq;
3265 outq.queue = NULL;
3266 outq.nr = outq.alloc = 0;
3268 if (!filter)
3269 return;
3271 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3272 int found;
3273 for (i = found = 0; !found && i < q->nr; i++) {
3274 struct diff_filepair *p = q->queue[i];
3275 if (((p->status == DIFF_STATUS_MODIFIED) &&
3276 ((p->score &&
3277 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3278 (!p->score &&
3279 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3280 ((p->status != DIFF_STATUS_MODIFIED) &&
3281 strchr(filter, p->status)))
3282 found++;
3284 if (found)
3285 return;
3287 /* otherwise we will clear the whole queue
3288 * by copying the empty outq at the end of this
3289 * function, but first clear the current entries
3290 * in the queue.
3292 for (i = 0; i < q->nr; i++)
3293 diff_free_filepair(q->queue[i]);
3295 else {
3296 /* Only the matching ones */
3297 for (i = 0; i < q->nr; i++) {
3298 struct diff_filepair *p = q->queue[i];
3300 if (((p->status == DIFF_STATUS_MODIFIED) &&
3301 ((p->score &&
3302 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3303 (!p->score &&
3304 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3305 ((p->status != DIFF_STATUS_MODIFIED) &&
3306 strchr(filter, p->status)))
3307 diff_q(&outq, p);
3308 else
3309 diff_free_filepair(p);
3312 free(q->queue);
3313 *q = outq;
3316 /* Check whether two filespecs with the same mode and size are identical */
3317 static int diff_filespec_is_identical(struct diff_filespec *one,
3318 struct diff_filespec *two)
3320 if (S_ISGITLINK(one->mode))
3321 return 0;
3322 if (diff_populate_filespec(one, 0))
3323 return 0;
3324 if (diff_populate_filespec(two, 0))
3325 return 0;
3326 return !memcmp(one->data, two->data, one->size);
3329 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3331 int i;
3332 struct diff_queue_struct *q = &diff_queued_diff;
3333 struct diff_queue_struct outq;
3334 outq.queue = NULL;
3335 outq.nr = outq.alloc = 0;
3337 for (i = 0; i < q->nr; i++) {
3338 struct diff_filepair *p = q->queue[i];
3341 * 1. Entries that come from stat info dirtyness
3342 * always have both sides (iow, not create/delete),
3343 * one side of the object name is unknown, with
3344 * the same mode and size. Keep the ones that
3345 * do not match these criteria. They have real
3346 * differences.
3348 * 2. At this point, the file is known to be modified,
3349 * with the same mode and size, and the object
3350 * name of one side is unknown. Need to inspect
3351 * the identical contents.
3353 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3354 !DIFF_FILE_VALID(p->two) ||
3355 (p->one->sha1_valid && p->two->sha1_valid) ||
3356 (p->one->mode != p->two->mode) ||
3357 diff_populate_filespec(p->one, 1) ||
3358 diff_populate_filespec(p->two, 1) ||
3359 (p->one->size != p->two->size) ||
3360 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3361 diff_q(&outq, p);
3362 else {
3364 * The caller can subtract 1 from skip_stat_unmatch
3365 * to determine how many paths were dirty only
3366 * due to stat info mismatch.
3368 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3369 diffopt->skip_stat_unmatch++;
3370 diff_free_filepair(p);
3373 free(q->queue);
3374 *q = outq;
3377 void diffcore_std(struct diff_options *options)
3379 if (DIFF_OPT_TST(options, QUIET))
3380 return;
3382 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3383 diffcore_skip_stat_unmatch(options);
3384 if (options->break_opt != -1)
3385 diffcore_break(options->break_opt);
3386 if (options->detect_rename)
3387 diffcore_rename(options);
3388 if (options->break_opt != -1)
3389 diffcore_merge_broken();
3390 if (options->pickaxe)
3391 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3392 if (options->orderfile)
3393 diffcore_order(options->orderfile);
3394 diff_resolve_rename_copy();
3395 diffcore_apply_filter(options->filter);
3397 if (diff_queued_diff.nr)
3398 DIFF_OPT_SET(options, HAS_CHANGES);
3399 else
3400 DIFF_OPT_CLR(options, HAS_CHANGES);
3403 int diff_result_code(struct diff_options *opt, int status)
3405 int result = 0;
3406 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3407 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3408 return status;
3409 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3410 DIFF_OPT_TST(opt, HAS_CHANGES))
3411 result |= 01;
3412 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3413 DIFF_OPT_TST(opt, CHECK_FAILED))
3414 result |= 02;
3415 return result;
3418 void diff_addremove(struct diff_options *options,
3419 int addremove, unsigned mode,
3420 const unsigned char *sha1,
3421 const char *concatpath)
3423 struct diff_filespec *one, *two;
3425 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3426 return;
3428 /* This may look odd, but it is a preparation for
3429 * feeding "there are unchanged files which should
3430 * not produce diffs, but when you are doing copy
3431 * detection you would need them, so here they are"
3432 * entries to the diff-core. They will be prefixed
3433 * with something like '=' or '*' (I haven't decided
3434 * which but should not make any difference).
3435 * Feeding the same new and old to diff_change()
3436 * also has the same effect.
3437 * Before the final output happens, they are pruned after
3438 * merged into rename/copy pairs as appropriate.
3440 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3441 addremove = (addremove == '+' ? '-' :
3442 addremove == '-' ? '+' : addremove);
3444 if (options->prefix &&
3445 strncmp(concatpath, options->prefix, options->prefix_length))
3446 return;
3448 one = alloc_filespec(concatpath);
3449 two = alloc_filespec(concatpath);
3451 if (addremove != '+')
3452 fill_filespec(one, sha1, mode);
3453 if (addremove != '-')
3454 fill_filespec(two, sha1, mode);
3456 diff_queue(&diff_queued_diff, one, two);
3457 DIFF_OPT_SET(options, HAS_CHANGES);
3460 void diff_change(struct diff_options *options,
3461 unsigned old_mode, unsigned new_mode,
3462 const unsigned char *old_sha1,
3463 const unsigned char *new_sha1,
3464 const char *concatpath)
3466 struct diff_filespec *one, *two;
3468 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3469 && S_ISGITLINK(new_mode))
3470 return;
3472 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3473 unsigned tmp;
3474 const unsigned char *tmp_c;
3475 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3476 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3479 if (options->prefix &&
3480 strncmp(concatpath, options->prefix, options->prefix_length))
3481 return;
3483 one = alloc_filespec(concatpath);
3484 two = alloc_filespec(concatpath);
3485 fill_filespec(one, old_sha1, old_mode);
3486 fill_filespec(two, new_sha1, new_mode);
3488 diff_queue(&diff_queued_diff, one, two);
3489 DIFF_OPT_SET(options, HAS_CHANGES);
3492 void diff_unmerge(struct diff_options *options,
3493 const char *path,
3494 unsigned mode, const unsigned char *sha1)
3496 struct diff_filespec *one, *two;
3498 if (options->prefix &&
3499 strncmp(path, options->prefix, options->prefix_length))
3500 return;
3502 one = alloc_filespec(path);
3503 two = alloc_filespec(path);
3504 fill_filespec(one, sha1, mode);
3505 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;