cherry-pick/revert: make direct internal call to merge_tree()
[git/spearce.git] / diff.c
blob135dec459a0f5d0abba812d773d7962d460e078f
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 int dirstat_compare(const void *_a, const void *_b)
1065 const struct dirstat_file *a = _a;
1066 const struct dirstat_file *b = _b;
1067 return strcmp(a->name, b->name);
1070 static void show_dirstat(struct diff_options *options)
1072 int i;
1073 unsigned long changed;
1074 struct dirstat_dir dir;
1075 struct diff_queue_struct *q = &diff_queued_diff;
1077 dir.files = NULL;
1078 dir.alloc = 0;
1079 dir.nr = 0;
1080 dir.percent = options->dirstat_percent;
1081 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1083 changed = 0;
1084 for (i = 0; i < q->nr; i++) {
1085 struct diff_filepair *p = q->queue[i];
1086 const char *name;
1087 unsigned long copied, added, damage;
1089 name = p->one->path ? p->one->path : p->two->path;
1091 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1092 diff_populate_filespec(p->one, 0);
1093 diff_populate_filespec(p->two, 0);
1094 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1095 &copied, &added);
1096 diff_free_filespec_data(p->one);
1097 diff_free_filespec_data(p->two);
1098 } else if (DIFF_FILE_VALID(p->one)) {
1099 diff_populate_filespec(p->one, 1);
1100 copied = added = 0;
1101 diff_free_filespec_data(p->one);
1102 } else if (DIFF_FILE_VALID(p->two)) {
1103 diff_populate_filespec(p->two, 1);
1104 copied = 0;
1105 added = p->two->size;
1106 diff_free_filespec_data(p->two);
1107 } else
1108 continue;
1111 * Original minus copied is the removed material,
1112 * added is the new material. They are both damages
1113 * made to the preimage.
1115 damage = (p->one->size - copied) + added;
1117 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1118 dir.files[dir.nr].name = name;
1119 dir.files[dir.nr].changed = damage;
1120 changed += damage;
1121 dir.nr++;
1124 /* This can happen even with many files, if everything was renames */
1125 if (!changed)
1126 return;
1128 /* Show all directories with more than x% of the changes */
1129 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1130 gather_dirstat(options->file, &dir, changed, "", 0);
1133 static void free_diffstat_info(struct diffstat_t *diffstat)
1135 int i;
1136 for (i = 0; i < diffstat->nr; i++) {
1137 struct diffstat_file *f = diffstat->files[i];
1138 if (f->name != f->print_name)
1139 free(f->print_name);
1140 free(f->name);
1141 free(f->from_name);
1142 free(f);
1144 free(diffstat->files);
1147 struct checkdiff_t {
1148 const char *filename;
1149 int lineno;
1150 struct diff_options *o;
1151 unsigned ws_rule;
1152 unsigned status;
1153 int trailing_blanks_start;
1156 static int is_conflict_marker(const char *line, unsigned long len)
1158 char firstchar;
1159 int cnt;
1161 if (len < 8)
1162 return 0;
1163 firstchar = line[0];
1164 switch (firstchar) {
1165 case '=': case '>': case '<':
1166 break;
1167 default:
1168 return 0;
1170 for (cnt = 1; cnt < 7; cnt++)
1171 if (line[cnt] != firstchar)
1172 return 0;
1173 /* line[0] thru line[6] are same as firstchar */
1174 if (firstchar == '=') {
1175 /* divider between ours and theirs? */
1176 if (len != 8 || line[7] != '\n')
1177 return 0;
1178 } else if (len < 8 || !isspace(line[7])) {
1179 /* not divider before ours nor after theirs */
1180 return 0;
1182 return 1;
1185 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1187 struct checkdiff_t *data = priv;
1188 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1189 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1190 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1191 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1192 char *err;
1194 if (line[0] == '+') {
1195 unsigned bad;
1196 data->lineno++;
1197 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1198 data->trailing_blanks_start = 0;
1199 else if (!data->trailing_blanks_start)
1200 data->trailing_blanks_start = data->lineno;
1201 if (is_conflict_marker(line + 1, len - 1)) {
1202 data->status |= 1;
1203 fprintf(data->o->file,
1204 "%s:%d: leftover conflict marker\n",
1205 data->filename, data->lineno);
1207 bad = ws_check(line + 1, len - 1, data->ws_rule);
1208 if (!bad)
1209 return;
1210 data->status |= bad;
1211 err = whitespace_error_string(bad);
1212 fprintf(data->o->file, "%s:%d: %s.\n",
1213 data->filename, data->lineno, err);
1214 free(err);
1215 emit_line(data->o->file, set, reset, line, 1);
1216 ws_check_emit(line + 1, len - 1, data->ws_rule,
1217 data->o->file, set, reset, ws);
1218 } else if (line[0] == ' ') {
1219 data->lineno++;
1220 data->trailing_blanks_start = 0;
1221 } else if (line[0] == '@') {
1222 char *plus = strchr(line, '+');
1223 if (plus)
1224 data->lineno = strtol(plus, NULL, 10) - 1;
1225 else
1226 die("invalid diff");
1227 data->trailing_blanks_start = 0;
1231 static unsigned char *deflate_it(char *data,
1232 unsigned long size,
1233 unsigned long *result_size)
1235 int bound;
1236 unsigned char *deflated;
1237 z_stream stream;
1239 memset(&stream, 0, sizeof(stream));
1240 deflateInit(&stream, zlib_compression_level);
1241 bound = deflateBound(&stream, size);
1242 deflated = xmalloc(bound);
1243 stream.next_out = deflated;
1244 stream.avail_out = bound;
1246 stream.next_in = (unsigned char *)data;
1247 stream.avail_in = size;
1248 while (deflate(&stream, Z_FINISH) == Z_OK)
1249 ; /* nothing */
1250 deflateEnd(&stream);
1251 *result_size = stream.total_out;
1252 return deflated;
1255 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1257 void *cp;
1258 void *delta;
1259 void *deflated;
1260 void *data;
1261 unsigned long orig_size;
1262 unsigned long delta_size;
1263 unsigned long deflate_size;
1264 unsigned long data_size;
1266 /* We could do deflated delta, or we could do just deflated two,
1267 * whichever is smaller.
1269 delta = NULL;
1270 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1271 if (one->size && two->size) {
1272 delta = diff_delta(one->ptr, one->size,
1273 two->ptr, two->size,
1274 &delta_size, deflate_size);
1275 if (delta) {
1276 void *to_free = delta;
1277 orig_size = delta_size;
1278 delta = deflate_it(delta, delta_size, &delta_size);
1279 free(to_free);
1283 if (delta && delta_size < deflate_size) {
1284 fprintf(file, "delta %lu\n", orig_size);
1285 free(deflated);
1286 data = delta;
1287 data_size = delta_size;
1289 else {
1290 fprintf(file, "literal %lu\n", two->size);
1291 free(delta);
1292 data = deflated;
1293 data_size = deflate_size;
1296 /* emit data encoded in base85 */
1297 cp = data;
1298 while (data_size) {
1299 int bytes = (52 < data_size) ? 52 : data_size;
1300 char line[70];
1301 data_size -= bytes;
1302 if (bytes <= 26)
1303 line[0] = bytes + 'A' - 1;
1304 else
1305 line[0] = bytes - 26 + 'a' - 1;
1306 encode_85(line + 1, cp, bytes);
1307 cp = (char *) cp + bytes;
1308 fputs(line, file);
1309 fputc('\n', file);
1311 fprintf(file, "\n");
1312 free(data);
1315 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1317 fprintf(file, "GIT binary patch\n");
1318 emit_binary_diff_body(file, one, two);
1319 emit_binary_diff_body(file, two, one);
1322 static void setup_diff_attr_check(struct git_attr_check *check)
1324 static struct git_attr *attr_diff;
1326 if (!attr_diff) {
1327 attr_diff = git_attr("diff", 4);
1329 check[0].attr = attr_diff;
1332 static void diff_filespec_check_attr(struct diff_filespec *one)
1334 struct git_attr_check attr_diff_check;
1335 int check_from_data = 0;
1337 if (one->checked_attr)
1338 return;
1340 setup_diff_attr_check(&attr_diff_check);
1341 one->is_binary = 0;
1342 one->funcname_pattern_ident = NULL;
1344 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1345 const char *value;
1347 /* binaryness */
1348 value = attr_diff_check.value;
1349 if (ATTR_TRUE(value))
1351 else if (ATTR_FALSE(value))
1352 one->is_binary = 1;
1353 else
1354 check_from_data = 1;
1356 /* funcname pattern ident */
1357 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1359 else
1360 one->funcname_pattern_ident = value;
1363 if (check_from_data) {
1364 if (!one->data && DIFF_FILE_VALID(one))
1365 diff_populate_filespec(one, 0);
1367 if (one->data)
1368 one->is_binary = buffer_is_binary(one->data, one->size);
1372 int diff_filespec_is_binary(struct diff_filespec *one)
1374 diff_filespec_check_attr(one);
1375 return one->is_binary;
1378 static const char *funcname_pattern(const char *ident)
1380 struct funcname_pattern *pp;
1382 for (pp = funcname_pattern_list; pp; pp = pp->next)
1383 if (!strcmp(ident, pp->name))
1384 return pp->pattern;
1385 return NULL;
1388 static struct builtin_funcname_pattern {
1389 const char *name;
1390 const char *pattern;
1391 } builtin_funcname_pattern[] = {
1392 { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1393 { "html", "^\\s*\\(<[Hh][1-6]\\s.*>.*\\)$" },
1394 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1395 "new\\|return\\|switch\\|throw\\|while\\)\n"
1396 "^[ ]*\\(\\([ ]*"
1397 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1398 "[ ]*([^;]*\\)$" },
1399 { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1400 "destructor\\|interface\\|implementation\\|"
1401 "initialization\\|finalization\\)[ \t]*.*\\)$"
1402 "\\|"
1403 "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1405 { "python", "^\\s*\\(\\(class\\|def\\)\\s.*\\)$" },
1406 { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1407 { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1410 static const char *diff_funcname_pattern(struct diff_filespec *one)
1412 const char *ident, *pattern;
1413 int i;
1415 diff_filespec_check_attr(one);
1416 ident = one->funcname_pattern_ident;
1418 if (!ident)
1420 * If the config file has "funcname.default" defined, that
1421 * regexp is used; otherwise NULL is returned and xemit uses
1422 * the built-in default.
1424 return funcname_pattern("default");
1426 /* Look up custom "funcname.$ident" regexp from config. */
1427 pattern = funcname_pattern(ident);
1428 if (pattern)
1429 return pattern;
1432 * And define built-in fallback patterns here. Note that
1433 * these can be overridden by the user's config settings.
1435 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1436 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1437 return builtin_funcname_pattern[i].pattern;
1439 return NULL;
1442 static void builtin_diff(const char *name_a,
1443 const char *name_b,
1444 struct diff_filespec *one,
1445 struct diff_filespec *two,
1446 const char *xfrm_msg,
1447 struct diff_options *o,
1448 int complete_rewrite)
1450 mmfile_t mf1, mf2;
1451 const char *lbl[2];
1452 char *a_one, *b_two;
1453 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1454 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1456 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1457 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1458 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1459 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1460 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1461 if (lbl[0][0] == '/') {
1462 /* /dev/null */
1463 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1464 if (xfrm_msg && xfrm_msg[0])
1465 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1467 else if (lbl[1][0] == '/') {
1468 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1469 if (xfrm_msg && xfrm_msg[0])
1470 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1472 else {
1473 if (one->mode != two->mode) {
1474 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1475 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1477 if (xfrm_msg && xfrm_msg[0])
1478 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1480 * we do not run diff between different kind
1481 * of objects.
1483 if ((one->mode ^ two->mode) & S_IFMT)
1484 goto free_ab_and_return;
1485 if (complete_rewrite) {
1486 emit_rewrite_diff(name_a, name_b, one, two, o);
1487 o->found_changes = 1;
1488 goto free_ab_and_return;
1492 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1493 die("unable to read files to diff");
1495 if (!DIFF_OPT_TST(o, TEXT) &&
1496 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1497 /* Quite common confusing case */
1498 if (mf1.size == mf2.size &&
1499 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1500 goto free_ab_and_return;
1501 if (DIFF_OPT_TST(o, BINARY))
1502 emit_binary_diff(o->file, &mf1, &mf2);
1503 else
1504 fprintf(o->file, "Binary files %s and %s differ\n",
1505 lbl[0], lbl[1]);
1506 o->found_changes = 1;
1508 else {
1509 /* Crazy xdl interfaces.. */
1510 const char *diffopts = getenv("GIT_DIFF_OPTS");
1511 xpparam_t xpp;
1512 xdemitconf_t xecfg;
1513 xdemitcb_t ecb;
1514 struct emit_callback ecbdata;
1515 const char *funcname_pattern;
1517 funcname_pattern = diff_funcname_pattern(one);
1518 if (!funcname_pattern)
1519 funcname_pattern = diff_funcname_pattern(two);
1521 memset(&xecfg, 0, sizeof(xecfg));
1522 memset(&ecbdata, 0, sizeof(ecbdata));
1523 ecbdata.label_path = lbl;
1524 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1525 ecbdata.found_changesp = &o->found_changes;
1526 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1527 ecbdata.file = o->file;
1528 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1529 xecfg.ctxlen = o->context;
1530 xecfg.flags = XDL_EMIT_FUNCNAMES;
1531 if (funcname_pattern)
1532 xdiff_set_find_func(&xecfg, funcname_pattern);
1533 if (!diffopts)
1535 else if (!prefixcmp(diffopts, "--unified="))
1536 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1537 else if (!prefixcmp(diffopts, "-u"))
1538 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1539 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1540 ecbdata.diff_words =
1541 xcalloc(1, sizeof(struct diff_words_data));
1542 ecbdata.diff_words->file = o->file;
1544 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1545 &xpp, &xecfg, &ecb);
1546 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1547 free_diff_words_data(&ecbdata);
1550 free_ab_and_return:
1551 diff_free_filespec_data(one);
1552 diff_free_filespec_data(two);
1553 free(a_one);
1554 free(b_two);
1555 return;
1558 static void builtin_diffstat(const char *name_a, const char *name_b,
1559 struct diff_filespec *one,
1560 struct diff_filespec *two,
1561 struct diffstat_t *diffstat,
1562 struct diff_options *o,
1563 int complete_rewrite)
1565 mmfile_t mf1, mf2;
1566 struct diffstat_file *data;
1568 data = diffstat_add(diffstat, name_a, name_b);
1570 if (!one || !two) {
1571 data->is_unmerged = 1;
1572 return;
1574 if (complete_rewrite) {
1575 diff_populate_filespec(one, 0);
1576 diff_populate_filespec(two, 0);
1577 data->deleted = count_lines(one->data, one->size);
1578 data->added = count_lines(two->data, two->size);
1579 goto free_and_return;
1581 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1582 die("unable to read files to diff");
1584 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1585 data->is_binary = 1;
1586 data->added = mf2.size;
1587 data->deleted = mf1.size;
1588 } else {
1589 /* Crazy xdl interfaces.. */
1590 xpparam_t xpp;
1591 xdemitconf_t xecfg;
1592 xdemitcb_t ecb;
1594 memset(&xecfg, 0, sizeof(xecfg));
1595 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1596 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1597 &xpp, &xecfg, &ecb);
1600 free_and_return:
1601 diff_free_filespec_data(one);
1602 diff_free_filespec_data(two);
1605 static void builtin_checkdiff(const char *name_a, const char *name_b,
1606 const char *attr_path,
1607 struct diff_filespec *one,
1608 struct diff_filespec *two,
1609 struct diff_options *o)
1611 mmfile_t mf1, mf2;
1612 struct checkdiff_t data;
1614 if (!two)
1615 return;
1617 memset(&data, 0, sizeof(data));
1618 data.filename = name_b ? name_b : name_a;
1619 data.lineno = 0;
1620 data.o = o;
1621 data.ws_rule = whitespace_rule(attr_path);
1623 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1624 die("unable to read files to diff");
1627 * All the other codepaths check both sides, but not checking
1628 * the "old" side here is deliberate. We are checking the newly
1629 * introduced changes, and as long as the "new" side is text, we
1630 * can and should check what it introduces.
1632 if (diff_filespec_is_binary(two))
1633 goto free_and_return;
1634 else {
1635 /* Crazy xdl interfaces.. */
1636 xpparam_t xpp;
1637 xdemitconf_t xecfg;
1638 xdemitcb_t ecb;
1640 memset(&xecfg, 0, sizeof(xecfg));
1641 xecfg.ctxlen = 1; /* at least one context line */
1642 xpp.flags = XDF_NEED_MINIMAL;
1643 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1644 &xpp, &xecfg, &ecb);
1646 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1647 data.trailing_blanks_start) {
1648 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1649 data.filename, data.trailing_blanks_start);
1650 data.status = 1; /* report errors */
1653 free_and_return:
1654 diff_free_filespec_data(one);
1655 diff_free_filespec_data(two);
1656 if (data.status)
1657 DIFF_OPT_SET(o, CHECK_FAILED);
1660 struct diff_filespec *alloc_filespec(const char *path)
1662 int namelen = strlen(path);
1663 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1665 memset(spec, 0, sizeof(*spec));
1666 spec->path = (char *)(spec + 1);
1667 memcpy(spec->path, path, namelen+1);
1668 spec->count = 1;
1669 return spec;
1672 void free_filespec(struct diff_filespec *spec)
1674 if (!--spec->count) {
1675 diff_free_filespec_data(spec);
1676 free(spec);
1680 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1681 unsigned short mode)
1683 if (mode) {
1684 spec->mode = canon_mode(mode);
1685 hashcpy(spec->sha1, sha1);
1686 spec->sha1_valid = !is_null_sha1(sha1);
1691 * Given a name and sha1 pair, if the index tells us the file in
1692 * the work tree has that object contents, return true, so that
1693 * prepare_temp_file() does not have to inflate and extract.
1695 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1697 struct cache_entry *ce;
1698 struct stat st;
1699 int pos, len;
1701 /* We do not read the cache ourselves here, because the
1702 * benchmark with my previous version that always reads cache
1703 * shows that it makes things worse for diff-tree comparing
1704 * two linux-2.6 kernel trees in an already checked out work
1705 * tree. This is because most diff-tree comparisons deal with
1706 * only a small number of files, while reading the cache is
1707 * expensive for a large project, and its cost outweighs the
1708 * savings we get by not inflating the object to a temporary
1709 * file. Practically, this code only helps when we are used
1710 * by diff-cache --cached, which does read the cache before
1711 * calling us.
1713 if (!active_cache)
1714 return 0;
1716 /* We want to avoid the working directory if our caller
1717 * doesn't need the data in a normal file, this system
1718 * is rather slow with its stat/open/mmap/close syscalls,
1719 * and the object is contained in a pack file. The pack
1720 * is probably already open and will be faster to obtain
1721 * the data through than the working directory. Loose
1722 * objects however would tend to be slower as they need
1723 * to be individually opened and inflated.
1725 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1726 return 0;
1728 len = strlen(name);
1729 pos = cache_name_pos(name, len);
1730 if (pos < 0)
1731 return 0;
1732 ce = active_cache[pos];
1735 * This is not the sha1 we are looking for, or
1736 * unreusable because it is not a regular file.
1738 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1739 return 0;
1742 * If ce matches the file in the work tree, we can reuse it.
1744 if (ce_uptodate(ce) ||
1745 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1746 return 1;
1748 return 0;
1751 static int populate_from_stdin(struct diff_filespec *s)
1753 struct strbuf buf;
1754 size_t size = 0;
1756 strbuf_init(&buf, 0);
1757 if (strbuf_read(&buf, 0, 0) < 0)
1758 return error("error while reading from stdin %s",
1759 strerror(errno));
1761 s->should_munmap = 0;
1762 s->data = strbuf_detach(&buf, &size);
1763 s->size = size;
1764 s->should_free = 1;
1765 return 0;
1768 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1770 int len;
1771 char *data = xmalloc(100);
1772 len = snprintf(data, 100,
1773 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1774 s->data = data;
1775 s->size = len;
1776 s->should_free = 1;
1777 if (size_only) {
1778 s->data = NULL;
1779 free(data);
1781 return 0;
1785 * While doing rename detection and pickaxe operation, we may need to
1786 * grab the data for the blob (or file) for our own in-core comparison.
1787 * diff_filespec has data and size fields for this purpose.
1789 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1791 int err = 0;
1792 if (!DIFF_FILE_VALID(s))
1793 die("internal error: asking to populate invalid file.");
1794 if (S_ISDIR(s->mode))
1795 return -1;
1797 if (s->data)
1798 return 0;
1800 if (size_only && 0 < s->size)
1801 return 0;
1803 if (S_ISGITLINK(s->mode))
1804 return diff_populate_gitlink(s, size_only);
1806 if (!s->sha1_valid ||
1807 reuse_worktree_file(s->path, s->sha1, 0)) {
1808 struct strbuf buf;
1809 struct stat st;
1810 int fd;
1812 if (!strcmp(s->path, "-"))
1813 return populate_from_stdin(s);
1815 if (lstat(s->path, &st) < 0) {
1816 if (errno == ENOENT) {
1817 err_empty:
1818 err = -1;
1819 empty:
1820 s->data = (char *)"";
1821 s->size = 0;
1822 return err;
1825 s->size = xsize_t(st.st_size);
1826 if (!s->size)
1827 goto empty;
1828 if (size_only)
1829 return 0;
1830 if (S_ISLNK(st.st_mode)) {
1831 int ret;
1832 s->data = xmalloc(s->size);
1833 s->should_free = 1;
1834 ret = readlink(s->path, s->data, s->size);
1835 if (ret < 0) {
1836 free(s->data);
1837 goto err_empty;
1839 return 0;
1841 fd = open(s->path, O_RDONLY);
1842 if (fd < 0)
1843 goto err_empty;
1844 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1845 close(fd);
1846 s->should_munmap = 1;
1849 * Convert from working tree format to canonical git format
1851 strbuf_init(&buf, 0);
1852 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1853 size_t size = 0;
1854 munmap(s->data, s->size);
1855 s->should_munmap = 0;
1856 s->data = strbuf_detach(&buf, &size);
1857 s->size = size;
1858 s->should_free = 1;
1861 else {
1862 enum object_type type;
1863 if (size_only)
1864 type = sha1_object_info(s->sha1, &s->size);
1865 else {
1866 s->data = read_sha1_file(s->sha1, &type, &s->size);
1867 s->should_free = 1;
1870 return 0;
1873 void diff_free_filespec_blob(struct diff_filespec *s)
1875 if (s->should_free)
1876 free(s->data);
1877 else if (s->should_munmap)
1878 munmap(s->data, s->size);
1880 if (s->should_free || s->should_munmap) {
1881 s->should_free = s->should_munmap = 0;
1882 s->data = NULL;
1886 void diff_free_filespec_data(struct diff_filespec *s)
1888 diff_free_filespec_blob(s);
1889 free(s->cnt_data);
1890 s->cnt_data = NULL;
1893 static void prep_temp_blob(struct diff_tempfile *temp,
1894 void *blob,
1895 unsigned long size,
1896 const unsigned char *sha1,
1897 int mode)
1899 int fd;
1901 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1902 if (fd < 0)
1903 die("unable to create temp-file: %s", strerror(errno));
1904 if (write_in_full(fd, blob, size) != size)
1905 die("unable to write temp-file");
1906 close(fd);
1907 temp->name = temp->tmp_path;
1908 strcpy(temp->hex, sha1_to_hex(sha1));
1909 temp->hex[40] = 0;
1910 sprintf(temp->mode, "%06o", mode);
1913 static void prepare_temp_file(const char *name,
1914 struct diff_tempfile *temp,
1915 struct diff_filespec *one)
1917 if (!DIFF_FILE_VALID(one)) {
1918 not_a_valid_file:
1919 /* A '-' entry produces this for file-2, and
1920 * a '+' entry produces this for file-1.
1922 temp->name = "/dev/null";
1923 strcpy(temp->hex, ".");
1924 strcpy(temp->mode, ".");
1925 return;
1928 if (!one->sha1_valid ||
1929 reuse_worktree_file(name, one->sha1, 1)) {
1930 struct stat st;
1931 if (lstat(name, &st) < 0) {
1932 if (errno == ENOENT)
1933 goto not_a_valid_file;
1934 die("stat(%s): %s", name, strerror(errno));
1936 if (S_ISLNK(st.st_mode)) {
1937 int ret;
1938 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1939 size_t sz = xsize_t(st.st_size);
1940 if (sizeof(buf) <= st.st_size)
1941 die("symlink too long: %s", name);
1942 ret = readlink(name, buf, sz);
1943 if (ret < 0)
1944 die("readlink(%s)", name);
1945 prep_temp_blob(temp, buf, sz,
1946 (one->sha1_valid ?
1947 one->sha1 : null_sha1),
1948 (one->sha1_valid ?
1949 one->mode : S_IFLNK));
1951 else {
1952 /* we can borrow from the file in the work tree */
1953 temp->name = name;
1954 if (!one->sha1_valid)
1955 strcpy(temp->hex, sha1_to_hex(null_sha1));
1956 else
1957 strcpy(temp->hex, sha1_to_hex(one->sha1));
1958 /* Even though we may sometimes borrow the
1959 * contents from the work tree, we always want
1960 * one->mode. mode is trustworthy even when
1961 * !(one->sha1_valid), as long as
1962 * DIFF_FILE_VALID(one).
1964 sprintf(temp->mode, "%06o", one->mode);
1966 return;
1968 else {
1969 if (diff_populate_filespec(one, 0))
1970 die("cannot read data blob for %s", one->path);
1971 prep_temp_blob(temp, one->data, one->size,
1972 one->sha1, one->mode);
1976 static void remove_tempfile(void)
1978 int i;
1980 for (i = 0; i < 2; i++)
1981 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1982 unlink(diff_temp[i].name);
1983 diff_temp[i].name = NULL;
1987 static void remove_tempfile_on_signal(int signo)
1989 remove_tempfile();
1990 signal(SIGINT, SIG_DFL);
1991 raise(signo);
1994 /* An external diff command takes:
1996 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1997 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2000 static void run_external_diff(const char *pgm,
2001 const char *name,
2002 const char *other,
2003 struct diff_filespec *one,
2004 struct diff_filespec *two,
2005 const char *xfrm_msg,
2006 int complete_rewrite)
2008 const char *spawn_arg[10];
2009 struct diff_tempfile *temp = diff_temp;
2010 int retval;
2011 static int atexit_asked = 0;
2012 const char *othername;
2013 const char **arg = &spawn_arg[0];
2015 othername = (other? other : name);
2016 if (one && two) {
2017 prepare_temp_file(name, &temp[0], one);
2018 prepare_temp_file(othername, &temp[1], two);
2019 if (! atexit_asked &&
2020 (temp[0].name == temp[0].tmp_path ||
2021 temp[1].name == temp[1].tmp_path)) {
2022 atexit_asked = 1;
2023 atexit(remove_tempfile);
2025 signal(SIGINT, remove_tempfile_on_signal);
2028 if (one && two) {
2029 *arg++ = pgm;
2030 *arg++ = name;
2031 *arg++ = temp[0].name;
2032 *arg++ = temp[0].hex;
2033 *arg++ = temp[0].mode;
2034 *arg++ = temp[1].name;
2035 *arg++ = temp[1].hex;
2036 *arg++ = temp[1].mode;
2037 if (other) {
2038 *arg++ = other;
2039 *arg++ = xfrm_msg;
2041 } else {
2042 *arg++ = pgm;
2043 *arg++ = name;
2045 *arg = NULL;
2046 fflush(NULL);
2047 retval = run_command_v_opt(spawn_arg, 0);
2048 remove_tempfile();
2049 if (retval) {
2050 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2051 exit(1);
2055 static const char *external_diff_attr(const char *name)
2057 struct git_attr_check attr_diff_check;
2059 if (!name)
2060 return NULL;
2062 setup_diff_attr_check(&attr_diff_check);
2063 if (!git_checkattr(name, 1, &attr_diff_check)) {
2064 const char *value = attr_diff_check.value;
2065 if (!ATTR_TRUE(value) &&
2066 !ATTR_FALSE(value) &&
2067 !ATTR_UNSET(value)) {
2068 struct ll_diff_driver *drv;
2070 for (drv = user_diff; drv; drv = drv->next)
2071 if (!strcmp(drv->name, value))
2072 return drv->cmd;
2075 return NULL;
2078 static void run_diff_cmd(const char *pgm,
2079 const char *name,
2080 const char *other,
2081 const char *attr_path,
2082 struct diff_filespec *one,
2083 struct diff_filespec *two,
2084 const char *xfrm_msg,
2085 struct diff_options *o,
2086 int complete_rewrite)
2088 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2089 pgm = NULL;
2090 else {
2091 const char *cmd = external_diff_attr(attr_path);
2092 if (cmd)
2093 pgm = cmd;
2096 if (pgm) {
2097 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2098 complete_rewrite);
2099 return;
2101 if (one && two)
2102 builtin_diff(name, other ? other : name,
2103 one, two, xfrm_msg, o, complete_rewrite);
2104 else
2105 fprintf(o->file, "* Unmerged path %s\n", name);
2108 static void diff_fill_sha1_info(struct diff_filespec *one)
2110 if (DIFF_FILE_VALID(one)) {
2111 if (!one->sha1_valid) {
2112 struct stat st;
2113 if (!strcmp(one->path, "-")) {
2114 hashcpy(one->sha1, null_sha1);
2115 return;
2117 if (lstat(one->path, &st) < 0)
2118 die("stat %s", one->path);
2119 if (index_path(one->sha1, one->path, &st, 0))
2120 die("cannot hash %s\n", one->path);
2123 else
2124 hashclr(one->sha1);
2127 static int similarity_index(struct diff_filepair *p)
2129 return p->score * 100 / MAX_SCORE;
2132 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2134 /* Strip the prefix but do not molest /dev/null and absolute paths */
2135 if (*namep && **namep != '/')
2136 *namep += prefix_length;
2137 if (*otherp && **otherp != '/')
2138 *otherp += prefix_length;
2141 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2143 const char *pgm = external_diff();
2144 struct strbuf msg;
2145 char *xfrm_msg;
2146 struct diff_filespec *one = p->one;
2147 struct diff_filespec *two = p->two;
2148 const char *name;
2149 const char *other;
2150 const char *attr_path;
2151 int complete_rewrite = 0;
2153 name = p->one->path;
2154 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2155 attr_path = name;
2156 if (o->prefix_length)
2157 strip_prefix(o->prefix_length, &name, &other);
2159 if (DIFF_PAIR_UNMERGED(p)) {
2160 run_diff_cmd(pgm, name, NULL, attr_path,
2161 NULL, NULL, NULL, o, 0);
2162 return;
2165 diff_fill_sha1_info(one);
2166 diff_fill_sha1_info(two);
2168 strbuf_init(&msg, PATH_MAX * 2 + 300);
2169 switch (p->status) {
2170 case DIFF_STATUS_COPIED:
2171 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2172 strbuf_addstr(&msg, "\ncopy from ");
2173 quote_c_style(name, &msg, NULL, 0);
2174 strbuf_addstr(&msg, "\ncopy to ");
2175 quote_c_style(other, &msg, NULL, 0);
2176 strbuf_addch(&msg, '\n');
2177 break;
2178 case DIFF_STATUS_RENAMED:
2179 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2180 strbuf_addstr(&msg, "\nrename from ");
2181 quote_c_style(name, &msg, NULL, 0);
2182 strbuf_addstr(&msg, "\nrename to ");
2183 quote_c_style(other, &msg, NULL, 0);
2184 strbuf_addch(&msg, '\n');
2185 break;
2186 case DIFF_STATUS_MODIFIED:
2187 if (p->score) {
2188 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2189 similarity_index(p));
2190 complete_rewrite = 1;
2191 break;
2193 /* fallthru */
2194 default:
2195 /* nothing */
2199 if (hashcmp(one->sha1, two->sha1)) {
2200 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2202 if (DIFF_OPT_TST(o, BINARY)) {
2203 mmfile_t mf;
2204 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2205 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2206 abbrev = 40;
2208 strbuf_addf(&msg, "index %.*s..%.*s",
2209 abbrev, sha1_to_hex(one->sha1),
2210 abbrev, sha1_to_hex(two->sha1));
2211 if (one->mode == two->mode)
2212 strbuf_addf(&msg, " %06o", one->mode);
2213 strbuf_addch(&msg, '\n');
2216 if (msg.len)
2217 strbuf_setlen(&msg, msg.len - 1);
2218 xfrm_msg = msg.len ? msg.buf : NULL;
2220 if (!pgm &&
2221 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2222 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2223 /* a filepair that changes between file and symlink
2224 * needs to be split into deletion and creation.
2226 struct diff_filespec *null = alloc_filespec(two->path);
2227 run_diff_cmd(NULL, name, other, attr_path,
2228 one, null, xfrm_msg, o, 0);
2229 free(null);
2230 null = alloc_filespec(one->path);
2231 run_diff_cmd(NULL, name, other, attr_path,
2232 null, two, xfrm_msg, o, 0);
2233 free(null);
2235 else
2236 run_diff_cmd(pgm, name, other, attr_path,
2237 one, two, xfrm_msg, o, complete_rewrite);
2239 strbuf_release(&msg);
2242 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2243 struct diffstat_t *diffstat)
2245 const char *name;
2246 const char *other;
2247 int complete_rewrite = 0;
2249 if (DIFF_PAIR_UNMERGED(p)) {
2250 /* unmerged */
2251 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2252 return;
2255 name = p->one->path;
2256 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2258 if (o->prefix_length)
2259 strip_prefix(o->prefix_length, &name, &other);
2261 diff_fill_sha1_info(p->one);
2262 diff_fill_sha1_info(p->two);
2264 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2265 complete_rewrite = 1;
2266 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2269 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2271 const char *name;
2272 const char *other;
2273 const char *attr_path;
2275 if (DIFF_PAIR_UNMERGED(p)) {
2276 /* unmerged */
2277 return;
2280 name = p->one->path;
2281 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2282 attr_path = other ? other : name;
2284 if (o->prefix_length)
2285 strip_prefix(o->prefix_length, &name, &other);
2287 diff_fill_sha1_info(p->one);
2288 diff_fill_sha1_info(p->two);
2290 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2293 void diff_setup(struct diff_options *options)
2295 memset(options, 0, sizeof(*options));
2297 options->file = stdout;
2299 options->line_termination = '\n';
2300 options->break_opt = -1;
2301 options->rename_limit = -1;
2302 options->dirstat_percent = 3;
2303 options->context = 3;
2305 options->change = diff_change;
2306 options->add_remove = diff_addremove;
2307 if (diff_use_color_default > 0)
2308 DIFF_OPT_SET(options, COLOR_DIFF);
2309 else
2310 DIFF_OPT_CLR(options, COLOR_DIFF);
2311 options->detect_rename = diff_detect_rename_default;
2313 options->a_prefix = "a/";
2314 options->b_prefix = "b/";
2317 int diff_setup_done(struct diff_options *options)
2319 int count = 0;
2321 if (options->output_format & DIFF_FORMAT_NAME)
2322 count++;
2323 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2324 count++;
2325 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2326 count++;
2327 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2328 count++;
2329 if (count > 1)
2330 die("--name-only, --name-status, --check and -s are mutually exclusive");
2332 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2333 options->detect_rename = DIFF_DETECT_COPY;
2335 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2336 options->prefix = NULL;
2337 if (options->prefix)
2338 options->prefix_length = strlen(options->prefix);
2339 else
2340 options->prefix_length = 0;
2342 if (options->output_format & (DIFF_FORMAT_NAME |
2343 DIFF_FORMAT_NAME_STATUS |
2344 DIFF_FORMAT_CHECKDIFF |
2345 DIFF_FORMAT_NO_OUTPUT))
2346 options->output_format &= ~(DIFF_FORMAT_RAW |
2347 DIFF_FORMAT_NUMSTAT |
2348 DIFF_FORMAT_DIFFSTAT |
2349 DIFF_FORMAT_SHORTSTAT |
2350 DIFF_FORMAT_DIRSTAT |
2351 DIFF_FORMAT_SUMMARY |
2352 DIFF_FORMAT_PATCH);
2355 * These cases always need recursive; we do not drop caller-supplied
2356 * recursive bits for other formats here.
2358 if (options->output_format & (DIFF_FORMAT_PATCH |
2359 DIFF_FORMAT_NUMSTAT |
2360 DIFF_FORMAT_DIFFSTAT |
2361 DIFF_FORMAT_SHORTSTAT |
2362 DIFF_FORMAT_DIRSTAT |
2363 DIFF_FORMAT_SUMMARY |
2364 DIFF_FORMAT_CHECKDIFF))
2365 DIFF_OPT_SET(options, RECURSIVE);
2367 * Also pickaxe would not work very well if you do not say recursive
2369 if (options->pickaxe)
2370 DIFF_OPT_SET(options, RECURSIVE);
2372 if (options->detect_rename && options->rename_limit < 0)
2373 options->rename_limit = diff_rename_limit_default;
2374 if (options->setup & DIFF_SETUP_USE_CACHE) {
2375 if (!active_cache)
2376 /* read-cache does not die even when it fails
2377 * so it is safe for us to do this here. Also
2378 * it does not smudge active_cache or active_nr
2379 * when it fails, so we do not have to worry about
2380 * cleaning it up ourselves either.
2382 read_cache();
2384 if (options->abbrev <= 0 || 40 < options->abbrev)
2385 options->abbrev = 40; /* full */
2388 * It does not make sense to show the first hit we happened
2389 * to have found. It does not make sense not to return with
2390 * exit code in such a case either.
2392 if (DIFF_OPT_TST(options, QUIET)) {
2393 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2394 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2398 * If we postprocess in diffcore, we cannot simply return
2399 * upon the first hit. We need to run diff as usual.
2401 if (options->pickaxe || options->filter)
2402 DIFF_OPT_CLR(options, QUIET);
2404 return 0;
2407 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2409 char c, *eq;
2410 int len;
2412 if (*arg != '-')
2413 return 0;
2414 c = *++arg;
2415 if (!c)
2416 return 0;
2417 if (c == arg_short) {
2418 c = *++arg;
2419 if (!c)
2420 return 1;
2421 if (val && isdigit(c)) {
2422 char *end;
2423 int n = strtoul(arg, &end, 10);
2424 if (*end)
2425 return 0;
2426 *val = n;
2427 return 1;
2429 return 0;
2431 if (c != '-')
2432 return 0;
2433 arg++;
2434 eq = strchr(arg, '=');
2435 if (eq)
2436 len = eq - arg;
2437 else
2438 len = strlen(arg);
2439 if (!len || strncmp(arg, arg_long, len))
2440 return 0;
2441 if (eq) {
2442 int n;
2443 char *end;
2444 if (!isdigit(*++eq))
2445 return 0;
2446 n = strtoul(eq, &end, 10);
2447 if (*end)
2448 return 0;
2449 *val = n;
2451 return 1;
2454 static int diff_scoreopt_parse(const char *opt);
2456 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2458 const char *arg = av[0];
2460 /* Output format options */
2461 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2462 options->output_format |= DIFF_FORMAT_PATCH;
2463 else if (opt_arg(arg, 'U', "unified", &options->context))
2464 options->output_format |= DIFF_FORMAT_PATCH;
2465 else if (!strcmp(arg, "--raw"))
2466 options->output_format |= DIFF_FORMAT_RAW;
2467 else if (!strcmp(arg, "--patch-with-raw"))
2468 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2469 else if (!strcmp(arg, "--numstat"))
2470 options->output_format |= DIFF_FORMAT_NUMSTAT;
2471 else if (!strcmp(arg, "--shortstat"))
2472 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2473 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2474 options->output_format |= DIFF_FORMAT_DIRSTAT;
2475 else if (!strcmp(arg, "--cumulative"))
2476 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2477 else if (!strcmp(arg, "--check"))
2478 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2479 else if (!strcmp(arg, "--summary"))
2480 options->output_format |= DIFF_FORMAT_SUMMARY;
2481 else if (!strcmp(arg, "--patch-with-stat"))
2482 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2483 else if (!strcmp(arg, "--name-only"))
2484 options->output_format |= DIFF_FORMAT_NAME;
2485 else if (!strcmp(arg, "--name-status"))
2486 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2487 else if (!strcmp(arg, "-s"))
2488 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2489 else if (!prefixcmp(arg, "--stat")) {
2490 char *end;
2491 int width = options->stat_width;
2492 int name_width = options->stat_name_width;
2493 arg += 6;
2494 end = (char *)arg;
2496 switch (*arg) {
2497 case '-':
2498 if (!prefixcmp(arg, "-width="))
2499 width = strtoul(arg + 7, &end, 10);
2500 else if (!prefixcmp(arg, "-name-width="))
2501 name_width = strtoul(arg + 12, &end, 10);
2502 break;
2503 case '=':
2504 width = strtoul(arg+1, &end, 10);
2505 if (*end == ',')
2506 name_width = strtoul(end+1, &end, 10);
2509 /* Important! This checks all the error cases! */
2510 if (*end)
2511 return 0;
2512 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2513 options->stat_name_width = name_width;
2514 options->stat_width = width;
2517 /* renames options */
2518 else if (!prefixcmp(arg, "-B")) {
2519 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2520 return -1;
2522 else if (!prefixcmp(arg, "-M")) {
2523 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2524 return -1;
2525 options->detect_rename = DIFF_DETECT_RENAME;
2527 else if (!prefixcmp(arg, "-C")) {
2528 if (options->detect_rename == DIFF_DETECT_COPY)
2529 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2530 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2531 return -1;
2532 options->detect_rename = DIFF_DETECT_COPY;
2534 else if (!strcmp(arg, "--no-renames"))
2535 options->detect_rename = 0;
2536 else if (!strcmp(arg, "--relative"))
2537 DIFF_OPT_SET(options, RELATIVE_NAME);
2538 else if (!prefixcmp(arg, "--relative=")) {
2539 DIFF_OPT_SET(options, RELATIVE_NAME);
2540 options->prefix = arg + 11;
2543 /* xdiff options */
2544 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2545 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2546 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2547 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2548 else if (!strcmp(arg, "--ignore-space-at-eol"))
2549 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2551 /* flags options */
2552 else if (!strcmp(arg, "--binary")) {
2553 options->output_format |= DIFF_FORMAT_PATCH;
2554 DIFF_OPT_SET(options, BINARY);
2556 else if (!strcmp(arg, "--full-index"))
2557 DIFF_OPT_SET(options, FULL_INDEX);
2558 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2559 DIFF_OPT_SET(options, TEXT);
2560 else if (!strcmp(arg, "-R"))
2561 DIFF_OPT_SET(options, REVERSE_DIFF);
2562 else if (!strcmp(arg, "--find-copies-harder"))
2563 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2564 else if (!strcmp(arg, "--follow"))
2565 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2566 else if (!strcmp(arg, "--color"))
2567 DIFF_OPT_SET(options, COLOR_DIFF);
2568 else if (!strcmp(arg, "--no-color"))
2569 DIFF_OPT_CLR(options, COLOR_DIFF);
2570 else if (!strcmp(arg, "--color-words"))
2571 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2572 else if (!strcmp(arg, "--exit-code"))
2573 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2574 else if (!strcmp(arg, "--quiet"))
2575 DIFF_OPT_SET(options, QUIET);
2576 else if (!strcmp(arg, "--ext-diff"))
2577 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2578 else if (!strcmp(arg, "--no-ext-diff"))
2579 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2580 else if (!strcmp(arg, "--ignore-submodules"))
2581 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2583 /* misc options */
2584 else if (!strcmp(arg, "-z"))
2585 options->line_termination = 0;
2586 else if (!prefixcmp(arg, "-l"))
2587 options->rename_limit = strtoul(arg+2, NULL, 10);
2588 else if (!prefixcmp(arg, "-S"))
2589 options->pickaxe = arg + 2;
2590 else if (!strcmp(arg, "--pickaxe-all"))
2591 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2592 else if (!strcmp(arg, "--pickaxe-regex"))
2593 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2594 else if (!prefixcmp(arg, "-O"))
2595 options->orderfile = arg + 2;
2596 else if (!prefixcmp(arg, "--diff-filter="))
2597 options->filter = arg + 14;
2598 else if (!strcmp(arg, "--abbrev"))
2599 options->abbrev = DEFAULT_ABBREV;
2600 else if (!prefixcmp(arg, "--abbrev=")) {
2601 options->abbrev = strtoul(arg + 9, NULL, 10);
2602 if (options->abbrev < MINIMUM_ABBREV)
2603 options->abbrev = MINIMUM_ABBREV;
2604 else if (40 < options->abbrev)
2605 options->abbrev = 40;
2607 else if (!prefixcmp(arg, "--src-prefix="))
2608 options->a_prefix = arg + 13;
2609 else if (!prefixcmp(arg, "--dst-prefix="))
2610 options->b_prefix = arg + 13;
2611 else if (!strcmp(arg, "--no-prefix"))
2612 options->a_prefix = options->b_prefix = "";
2613 else if (!prefixcmp(arg, "--output=")) {
2614 options->file = fopen(arg + strlen("--output="), "w");
2615 options->close_file = 1;
2616 } else
2617 return 0;
2618 return 1;
2621 static int parse_num(const char **cp_p)
2623 unsigned long num, scale;
2624 int ch, dot;
2625 const char *cp = *cp_p;
2627 num = 0;
2628 scale = 1;
2629 dot = 0;
2630 for(;;) {
2631 ch = *cp;
2632 if ( !dot && ch == '.' ) {
2633 scale = 1;
2634 dot = 1;
2635 } else if ( ch == '%' ) {
2636 scale = dot ? scale*100 : 100;
2637 cp++; /* % is always at the end */
2638 break;
2639 } else if ( ch >= '0' && ch <= '9' ) {
2640 if ( scale < 100000 ) {
2641 scale *= 10;
2642 num = (num*10) + (ch-'0');
2644 } else {
2645 break;
2647 cp++;
2649 *cp_p = cp;
2651 /* user says num divided by scale and we say internally that
2652 * is MAX_SCORE * num / scale.
2654 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2657 static int diff_scoreopt_parse(const char *opt)
2659 int opt1, opt2, cmd;
2661 if (*opt++ != '-')
2662 return -1;
2663 cmd = *opt++;
2664 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2665 return -1; /* that is not a -M, -C nor -B option */
2667 opt1 = parse_num(&opt);
2668 if (cmd != 'B')
2669 opt2 = 0;
2670 else {
2671 if (*opt == 0)
2672 opt2 = 0;
2673 else if (*opt != '/')
2674 return -1; /* we expect -B80/99 or -B80 */
2675 else {
2676 opt++;
2677 opt2 = parse_num(&opt);
2680 if (*opt != 0)
2681 return -1;
2682 return opt1 | (opt2 << 16);
2685 struct diff_queue_struct diff_queued_diff;
2687 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2689 if (queue->alloc <= queue->nr) {
2690 queue->alloc = alloc_nr(queue->alloc);
2691 queue->queue = xrealloc(queue->queue,
2692 sizeof(dp) * queue->alloc);
2694 queue->queue[queue->nr++] = dp;
2697 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2698 struct diff_filespec *one,
2699 struct diff_filespec *two)
2701 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2702 dp->one = one;
2703 dp->two = two;
2704 if (queue)
2705 diff_q(queue, dp);
2706 return dp;
2709 void diff_free_filepair(struct diff_filepair *p)
2711 free_filespec(p->one);
2712 free_filespec(p->two);
2713 free(p);
2716 /* This is different from find_unique_abbrev() in that
2717 * it stuffs the result with dots for alignment.
2719 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2721 int abblen;
2722 const char *abbrev;
2723 if (len == 40)
2724 return sha1_to_hex(sha1);
2726 abbrev = find_unique_abbrev(sha1, len);
2727 abblen = strlen(abbrev);
2728 if (abblen < 37) {
2729 static char hex[41];
2730 if (len < abblen && abblen <= len + 2)
2731 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2732 else
2733 sprintf(hex, "%s...", abbrev);
2734 return hex;
2736 return sha1_to_hex(sha1);
2739 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2741 int line_termination = opt->line_termination;
2742 int inter_name_termination = line_termination ? '\t' : '\0';
2744 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2745 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2746 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2747 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2749 if (p->score) {
2750 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2751 inter_name_termination);
2752 } else {
2753 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2756 if (p->status == DIFF_STATUS_COPIED ||
2757 p->status == DIFF_STATUS_RENAMED) {
2758 const char *name_a, *name_b;
2759 name_a = p->one->path;
2760 name_b = p->two->path;
2761 strip_prefix(opt->prefix_length, &name_a, &name_b);
2762 write_name_quoted(name_a, opt->file, inter_name_termination);
2763 write_name_quoted(name_b, opt->file, line_termination);
2764 } else {
2765 const char *name_a, *name_b;
2766 name_a = p->one->mode ? p->one->path : p->two->path;
2767 name_b = NULL;
2768 strip_prefix(opt->prefix_length, &name_a, &name_b);
2769 write_name_quoted(name_a, opt->file, line_termination);
2773 int diff_unmodified_pair(struct diff_filepair *p)
2775 /* This function is written stricter than necessary to support
2776 * the currently implemented transformers, but the idea is to
2777 * let transformers to produce diff_filepairs any way they want,
2778 * and filter and clean them up here before producing the output.
2780 struct diff_filespec *one = p->one, *two = p->two;
2782 if (DIFF_PAIR_UNMERGED(p))
2783 return 0; /* unmerged is interesting */
2785 /* deletion, addition, mode or type change
2786 * and rename are all interesting.
2788 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2789 DIFF_PAIR_MODE_CHANGED(p) ||
2790 strcmp(one->path, two->path))
2791 return 0;
2793 /* both are valid and point at the same path. that is, we are
2794 * dealing with a change.
2796 if (one->sha1_valid && two->sha1_valid &&
2797 !hashcmp(one->sha1, two->sha1))
2798 return 1; /* no change */
2799 if (!one->sha1_valid && !two->sha1_valid)
2800 return 1; /* both look at the same file on the filesystem. */
2801 return 0;
2804 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2806 if (diff_unmodified_pair(p))
2807 return;
2809 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2810 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2811 return; /* no tree diffs in patch format */
2813 run_diff(p, o);
2816 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2817 struct diffstat_t *diffstat)
2819 if (diff_unmodified_pair(p))
2820 return;
2822 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2823 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2824 return; /* no tree diffs in patch format */
2826 run_diffstat(p, o, diffstat);
2829 static void diff_flush_checkdiff(struct diff_filepair *p,
2830 struct diff_options *o)
2832 if (diff_unmodified_pair(p))
2833 return;
2835 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2836 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2837 return; /* no tree diffs in patch format */
2839 run_checkdiff(p, o);
2842 int diff_queue_is_empty(void)
2844 struct diff_queue_struct *q = &diff_queued_diff;
2845 int i;
2846 for (i = 0; i < q->nr; i++)
2847 if (!diff_unmodified_pair(q->queue[i]))
2848 return 0;
2849 return 1;
2852 #if DIFF_DEBUG
2853 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2855 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2856 x, one ? one : "",
2857 s->path,
2858 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2859 s->mode,
2860 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2861 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2862 x, one ? one : "",
2863 s->size, s->xfrm_flags);
2866 void diff_debug_filepair(const struct diff_filepair *p, int i)
2868 diff_debug_filespec(p->one, i, "one");
2869 diff_debug_filespec(p->two, i, "two");
2870 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2871 p->score, p->status ? p->status : '?',
2872 p->one->rename_used, p->broken_pair);
2875 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2877 int i;
2878 if (msg)
2879 fprintf(stderr, "%s\n", msg);
2880 fprintf(stderr, "q->nr = %d\n", q->nr);
2881 for (i = 0; i < q->nr; i++) {
2882 struct diff_filepair *p = q->queue[i];
2883 diff_debug_filepair(p, i);
2886 #endif
2888 static void diff_resolve_rename_copy(void)
2890 int i;
2891 struct diff_filepair *p;
2892 struct diff_queue_struct *q = &diff_queued_diff;
2894 diff_debug_queue("resolve-rename-copy", q);
2896 for (i = 0; i < q->nr; i++) {
2897 p = q->queue[i];
2898 p->status = 0; /* undecided */
2899 if (DIFF_PAIR_UNMERGED(p))
2900 p->status = DIFF_STATUS_UNMERGED;
2901 else if (!DIFF_FILE_VALID(p->one))
2902 p->status = DIFF_STATUS_ADDED;
2903 else if (!DIFF_FILE_VALID(p->two))
2904 p->status = DIFF_STATUS_DELETED;
2905 else if (DIFF_PAIR_TYPE_CHANGED(p))
2906 p->status = DIFF_STATUS_TYPE_CHANGED;
2908 /* from this point on, we are dealing with a pair
2909 * whose both sides are valid and of the same type, i.e.
2910 * either in-place edit or rename/copy edit.
2912 else if (DIFF_PAIR_RENAME(p)) {
2914 * A rename might have re-connected a broken
2915 * pair up, causing the pathnames to be the
2916 * same again. If so, that's not a rename at
2917 * all, just a modification..
2919 * Otherwise, see if this source was used for
2920 * multiple renames, in which case we decrement
2921 * the count, and call it a copy.
2923 if (!strcmp(p->one->path, p->two->path))
2924 p->status = DIFF_STATUS_MODIFIED;
2925 else if (--p->one->rename_used > 0)
2926 p->status = DIFF_STATUS_COPIED;
2927 else
2928 p->status = DIFF_STATUS_RENAMED;
2930 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2931 p->one->mode != p->two->mode ||
2932 is_null_sha1(p->one->sha1))
2933 p->status = DIFF_STATUS_MODIFIED;
2934 else {
2935 /* This is a "no-change" entry and should not
2936 * happen anymore, but prepare for broken callers.
2938 error("feeding unmodified %s to diffcore",
2939 p->one->path);
2940 p->status = DIFF_STATUS_UNKNOWN;
2943 diff_debug_queue("resolve-rename-copy done", q);
2946 static int check_pair_status(struct diff_filepair *p)
2948 switch (p->status) {
2949 case DIFF_STATUS_UNKNOWN:
2950 return 0;
2951 case 0:
2952 die("internal error in diff-resolve-rename-copy");
2953 default:
2954 return 1;
2958 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2960 int fmt = opt->output_format;
2962 if (fmt & DIFF_FORMAT_CHECKDIFF)
2963 diff_flush_checkdiff(p, opt);
2964 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2965 diff_flush_raw(p, opt);
2966 else if (fmt & DIFF_FORMAT_NAME) {
2967 const char *name_a, *name_b;
2968 name_a = p->two->path;
2969 name_b = NULL;
2970 strip_prefix(opt->prefix_length, &name_a, &name_b);
2971 write_name_quoted(name_a, opt->file, opt->line_termination);
2975 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2977 if (fs->mode)
2978 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2979 else
2980 fprintf(file, " %s ", newdelete);
2981 write_name_quoted(fs->path, file, '\n');
2985 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2987 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2988 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2989 show_name ? ' ' : '\n');
2990 if (show_name) {
2991 write_name_quoted(p->two->path, file, '\n');
2996 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2998 char *names = pprint_rename(p->one->path, p->two->path);
3000 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3001 free(names);
3002 show_mode_change(file, p, 0);
3005 static void diff_summary(FILE *file, struct diff_filepair *p)
3007 switch(p->status) {
3008 case DIFF_STATUS_DELETED:
3009 show_file_mode_name(file, "delete", p->one);
3010 break;
3011 case DIFF_STATUS_ADDED:
3012 show_file_mode_name(file, "create", p->two);
3013 break;
3014 case DIFF_STATUS_COPIED:
3015 show_rename_copy(file, "copy", p);
3016 break;
3017 case DIFF_STATUS_RENAMED:
3018 show_rename_copy(file, "rename", p);
3019 break;
3020 default:
3021 if (p->score) {
3022 fputs(" rewrite ", file);
3023 write_name_quoted(p->two->path, file, ' ');
3024 fprintf(file, "(%d%%)\n", similarity_index(p));
3026 show_mode_change(file, p, !p->score);
3027 break;
3031 struct patch_id_t {
3032 SHA_CTX *ctx;
3033 int patchlen;
3036 static int remove_space(char *line, int len)
3038 int i;
3039 char *dst = line;
3040 unsigned char c;
3042 for (i = 0; i < len; i++)
3043 if (!isspace((c = line[i])))
3044 *dst++ = c;
3046 return dst - line;
3049 static void patch_id_consume(void *priv, char *line, unsigned long len)
3051 struct patch_id_t *data = priv;
3052 int new_len;
3054 /* Ignore line numbers when computing the SHA1 of the patch */
3055 if (!prefixcmp(line, "@@ -"))
3056 return;
3058 new_len = remove_space(line, len);
3060 SHA1_Update(data->ctx, line, new_len);
3061 data->patchlen += new_len;
3064 /* returns 0 upon success, and writes result into sha1 */
3065 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3067 struct diff_queue_struct *q = &diff_queued_diff;
3068 int i;
3069 SHA_CTX ctx;
3070 struct patch_id_t data;
3071 char buffer[PATH_MAX * 4 + 20];
3073 SHA1_Init(&ctx);
3074 memset(&data, 0, sizeof(struct patch_id_t));
3075 data.ctx = &ctx;
3077 for (i = 0; i < q->nr; i++) {
3078 xpparam_t xpp;
3079 xdemitconf_t xecfg;
3080 xdemitcb_t ecb;
3081 mmfile_t mf1, mf2;
3082 struct diff_filepair *p = q->queue[i];
3083 int len1, len2;
3085 memset(&xecfg, 0, sizeof(xecfg));
3086 if (p->status == 0)
3087 return error("internal diff status error");
3088 if (p->status == DIFF_STATUS_UNKNOWN)
3089 continue;
3090 if (diff_unmodified_pair(p))
3091 continue;
3092 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3093 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3094 continue;
3095 if (DIFF_PAIR_UNMERGED(p))
3096 continue;
3098 diff_fill_sha1_info(p->one);
3099 diff_fill_sha1_info(p->two);
3100 if (fill_mmfile(&mf1, p->one) < 0 ||
3101 fill_mmfile(&mf2, p->two) < 0)
3102 return error("unable to read files to diff");
3104 len1 = remove_space(p->one->path, strlen(p->one->path));
3105 len2 = remove_space(p->two->path, strlen(p->two->path));
3106 if (p->one->mode == 0)
3107 len1 = snprintf(buffer, sizeof(buffer),
3108 "diff--gita/%.*sb/%.*s"
3109 "newfilemode%06o"
3110 "---/dev/null"
3111 "+++b/%.*s",
3112 len1, p->one->path,
3113 len2, p->two->path,
3114 p->two->mode,
3115 len2, p->two->path);
3116 else if (p->two->mode == 0)
3117 len1 = snprintf(buffer, sizeof(buffer),
3118 "diff--gita/%.*sb/%.*s"
3119 "deletedfilemode%06o"
3120 "---a/%.*s"
3121 "+++/dev/null",
3122 len1, p->one->path,
3123 len2, p->two->path,
3124 p->one->mode,
3125 len1, p->one->path);
3126 else
3127 len1 = snprintf(buffer, sizeof(buffer),
3128 "diff--gita/%.*sb/%.*s"
3129 "---a/%.*s"
3130 "+++b/%.*s",
3131 len1, p->one->path,
3132 len2, p->two->path,
3133 len1, p->one->path,
3134 len2, p->two->path);
3135 SHA1_Update(&ctx, buffer, len1);
3137 xpp.flags = XDF_NEED_MINIMAL;
3138 xecfg.ctxlen = 3;
3139 xecfg.flags = XDL_EMIT_FUNCNAMES;
3140 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3141 &xpp, &xecfg, &ecb);
3144 SHA1_Final(sha1, &ctx);
3145 return 0;
3148 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3150 struct diff_queue_struct *q = &diff_queued_diff;
3151 int i;
3152 int result = diff_get_patch_id(options, sha1);
3154 for (i = 0; i < q->nr; i++)
3155 diff_free_filepair(q->queue[i]);
3157 free(q->queue);
3158 q->queue = NULL;
3159 q->nr = q->alloc = 0;
3161 return result;
3164 static int is_summary_empty(const struct diff_queue_struct *q)
3166 int i;
3168 for (i = 0; i < q->nr; i++) {
3169 const struct diff_filepair *p = q->queue[i];
3171 switch (p->status) {
3172 case DIFF_STATUS_DELETED:
3173 case DIFF_STATUS_ADDED:
3174 case DIFF_STATUS_COPIED:
3175 case DIFF_STATUS_RENAMED:
3176 return 0;
3177 default:
3178 if (p->score)
3179 return 0;
3180 if (p->one->mode && p->two->mode &&
3181 p->one->mode != p->two->mode)
3182 return 0;
3183 break;
3186 return 1;
3189 void diff_flush(struct diff_options *options)
3191 struct diff_queue_struct *q = &diff_queued_diff;
3192 int i, output_format = options->output_format;
3193 int separator = 0;
3196 * Order: raw, stat, summary, patch
3197 * or: name/name-status/checkdiff (other bits clear)
3199 if (!q->nr)
3200 goto free_queue;
3202 if (output_format & (DIFF_FORMAT_RAW |
3203 DIFF_FORMAT_NAME |
3204 DIFF_FORMAT_NAME_STATUS |
3205 DIFF_FORMAT_CHECKDIFF)) {
3206 for (i = 0; i < q->nr; i++) {
3207 struct diff_filepair *p = q->queue[i];
3208 if (check_pair_status(p))
3209 flush_one_pair(p, options);
3211 separator++;
3214 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3215 struct diffstat_t diffstat;
3217 memset(&diffstat, 0, sizeof(struct diffstat_t));
3218 for (i = 0; i < q->nr; i++) {
3219 struct diff_filepair *p = q->queue[i];
3220 if (check_pair_status(p))
3221 diff_flush_stat(p, options, &diffstat);
3223 if (output_format & DIFF_FORMAT_NUMSTAT)
3224 show_numstat(&diffstat, options);
3225 if (output_format & DIFF_FORMAT_DIFFSTAT)
3226 show_stats(&diffstat, options);
3227 if (output_format & DIFF_FORMAT_SHORTSTAT)
3228 show_shortstats(&diffstat, options);
3229 free_diffstat_info(&diffstat);
3230 separator++;
3232 if (output_format & DIFF_FORMAT_DIRSTAT)
3233 show_dirstat(options);
3235 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3236 for (i = 0; i < q->nr; i++)
3237 diff_summary(options->file, q->queue[i]);
3238 separator++;
3241 if (output_format & DIFF_FORMAT_PATCH) {
3242 if (separator) {
3243 putc(options->line_termination, options->file);
3244 if (options->stat_sep) {
3245 /* attach patch instead of inline */
3246 fputs(options->stat_sep, options->file);
3250 for (i = 0; i < q->nr; i++) {
3251 struct diff_filepair *p = q->queue[i];
3252 if (check_pair_status(p))
3253 diff_flush_patch(p, options);
3257 if (output_format & DIFF_FORMAT_CALLBACK)
3258 options->format_callback(q, options, options->format_callback_data);
3260 for (i = 0; i < q->nr; i++)
3261 diff_free_filepair(q->queue[i]);
3262 free_queue:
3263 free(q->queue);
3264 q->queue = NULL;
3265 q->nr = q->alloc = 0;
3266 if (options->close_file)
3267 fclose(options->file);
3270 static void diffcore_apply_filter(const char *filter)
3272 int i;
3273 struct diff_queue_struct *q = &diff_queued_diff;
3274 struct diff_queue_struct outq;
3275 outq.queue = NULL;
3276 outq.nr = outq.alloc = 0;
3278 if (!filter)
3279 return;
3281 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3282 int found;
3283 for (i = found = 0; !found && i < q->nr; i++) {
3284 struct diff_filepair *p = q->queue[i];
3285 if (((p->status == DIFF_STATUS_MODIFIED) &&
3286 ((p->score &&
3287 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3288 (!p->score &&
3289 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3290 ((p->status != DIFF_STATUS_MODIFIED) &&
3291 strchr(filter, p->status)))
3292 found++;
3294 if (found)
3295 return;
3297 /* otherwise we will clear the whole queue
3298 * by copying the empty outq at the end of this
3299 * function, but first clear the current entries
3300 * in the queue.
3302 for (i = 0; i < q->nr; i++)
3303 diff_free_filepair(q->queue[i]);
3305 else {
3306 /* Only the matching ones */
3307 for (i = 0; i < q->nr; i++) {
3308 struct diff_filepair *p = q->queue[i];
3310 if (((p->status == DIFF_STATUS_MODIFIED) &&
3311 ((p->score &&
3312 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3313 (!p->score &&
3314 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3315 ((p->status != DIFF_STATUS_MODIFIED) &&
3316 strchr(filter, p->status)))
3317 diff_q(&outq, p);
3318 else
3319 diff_free_filepair(p);
3322 free(q->queue);
3323 *q = outq;
3326 /* Check whether two filespecs with the same mode and size are identical */
3327 static int diff_filespec_is_identical(struct diff_filespec *one,
3328 struct diff_filespec *two)
3330 if (S_ISGITLINK(one->mode))
3331 return 0;
3332 if (diff_populate_filespec(one, 0))
3333 return 0;
3334 if (diff_populate_filespec(two, 0))
3335 return 0;
3336 return !memcmp(one->data, two->data, one->size);
3339 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3341 int i;
3342 struct diff_queue_struct *q = &diff_queued_diff;
3343 struct diff_queue_struct outq;
3344 outq.queue = NULL;
3345 outq.nr = outq.alloc = 0;
3347 for (i = 0; i < q->nr; i++) {
3348 struct diff_filepair *p = q->queue[i];
3351 * 1. Entries that come from stat info dirtyness
3352 * always have both sides (iow, not create/delete),
3353 * one side of the object name is unknown, with
3354 * the same mode and size. Keep the ones that
3355 * do not match these criteria. They have real
3356 * differences.
3358 * 2. At this point, the file is known to be modified,
3359 * with the same mode and size, and the object
3360 * name of one side is unknown. Need to inspect
3361 * the identical contents.
3363 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3364 !DIFF_FILE_VALID(p->two) ||
3365 (p->one->sha1_valid && p->two->sha1_valid) ||
3366 (p->one->mode != p->two->mode) ||
3367 diff_populate_filespec(p->one, 1) ||
3368 diff_populate_filespec(p->two, 1) ||
3369 (p->one->size != p->two->size) ||
3370 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3371 diff_q(&outq, p);
3372 else {
3374 * The caller can subtract 1 from skip_stat_unmatch
3375 * to determine how many paths were dirty only
3376 * due to stat info mismatch.
3378 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3379 diffopt->skip_stat_unmatch++;
3380 diff_free_filepair(p);
3383 free(q->queue);
3384 *q = outq;
3387 void diffcore_std(struct diff_options *options)
3389 if (DIFF_OPT_TST(options, QUIET))
3390 return;
3392 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3393 diffcore_skip_stat_unmatch(options);
3394 if (options->break_opt != -1)
3395 diffcore_break(options->break_opt);
3396 if (options->detect_rename)
3397 diffcore_rename(options);
3398 if (options->break_opt != -1)
3399 diffcore_merge_broken();
3400 if (options->pickaxe)
3401 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3402 if (options->orderfile)
3403 diffcore_order(options->orderfile);
3404 diff_resolve_rename_copy();
3405 diffcore_apply_filter(options->filter);
3407 if (diff_queued_diff.nr)
3408 DIFF_OPT_SET(options, HAS_CHANGES);
3409 else
3410 DIFF_OPT_CLR(options, HAS_CHANGES);
3413 int diff_result_code(struct diff_options *opt, int status)
3415 int result = 0;
3416 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3417 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3418 return status;
3419 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3420 DIFF_OPT_TST(opt, HAS_CHANGES))
3421 result |= 01;
3422 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3423 DIFF_OPT_TST(opt, CHECK_FAILED))
3424 result |= 02;
3425 return result;
3428 void diff_addremove(struct diff_options *options,
3429 int addremove, unsigned mode,
3430 const unsigned char *sha1,
3431 const char *concatpath)
3433 struct diff_filespec *one, *two;
3435 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3436 return;
3438 /* This may look odd, but it is a preparation for
3439 * feeding "there are unchanged files which should
3440 * not produce diffs, but when you are doing copy
3441 * detection you would need them, so here they are"
3442 * entries to the diff-core. They will be prefixed
3443 * with something like '=' or '*' (I haven't decided
3444 * which but should not make any difference).
3445 * Feeding the same new and old to diff_change()
3446 * also has the same effect.
3447 * Before the final output happens, they are pruned after
3448 * merged into rename/copy pairs as appropriate.
3450 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3451 addremove = (addremove == '+' ? '-' :
3452 addremove == '-' ? '+' : addremove);
3454 if (options->prefix &&
3455 strncmp(concatpath, options->prefix, options->prefix_length))
3456 return;
3458 one = alloc_filespec(concatpath);
3459 two = alloc_filespec(concatpath);
3461 if (addremove != '+')
3462 fill_filespec(one, sha1, mode);
3463 if (addremove != '-')
3464 fill_filespec(two, sha1, mode);
3466 diff_queue(&diff_queued_diff, one, two);
3467 DIFF_OPT_SET(options, HAS_CHANGES);
3470 void diff_change(struct diff_options *options,
3471 unsigned old_mode, unsigned new_mode,
3472 const unsigned char *old_sha1,
3473 const unsigned char *new_sha1,
3474 const char *concatpath)
3476 struct diff_filespec *one, *two;
3478 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3479 && S_ISGITLINK(new_mode))
3480 return;
3482 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3483 unsigned tmp;
3484 const unsigned char *tmp_c;
3485 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3486 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3489 if (options->prefix &&
3490 strncmp(concatpath, options->prefix, options->prefix_length))
3491 return;
3493 one = alloc_filespec(concatpath);
3494 two = alloc_filespec(concatpath);
3495 fill_filespec(one, old_sha1, old_mode);
3496 fill_filespec(two, new_sha1, new_mode);
3498 diff_queue(&diff_queued_diff, one, two);
3499 DIFF_OPT_SET(options, HAS_CHANGES);
3502 void diff_unmerge(struct diff_options *options,
3503 const char *path,
3504 unsigned mode, const unsigned char *sha1)
3506 struct diff_filespec *one, *two;
3508 if (options->prefix &&
3509 strncmp(path, options->prefix, options->prefix_length))
3510 return;
3512 one = alloc_filespec(path);
3513 two = alloc_filespec(path);
3514 fill_filespec(one, sha1, mode);
3515 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;