Teach git diff about BibTeX head hunk patterns
[git/mjg.git] / diff.c
blobbf5d5f15a3b54f1e1a0a2068990e899fd869d435
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 200;
23 int diff_use_color_default = -1;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
27 static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
38 static int parse_diff_color_slot(const char *var, int ofs)
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
57 static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 const char *cmd;
61 } *user_diff, **user_diff_tail;
64 * Currently there is only "diff.<drivername>.command" variable;
65 * because there are "diff.color.<slot>" variables, we are parsing
66 * this in a bit convoluted way to allow low level diff driver
67 * called "color".
69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
71 const char *name;
72 int namelen;
73 struct ll_diff_driver *drv;
75 name = var + 5;
76 namelen = ep - name;
77 for (drv = user_diff; drv; drv = drv->next)
78 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79 break;
80 if (!drv) {
81 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82 drv->name = xmemdupz(name, namelen);
83 if (!user_diff_tail)
84 user_diff_tail = &user_diff;
85 *user_diff_tail = drv;
86 user_diff_tail = &(drv->next);
89 return git_config_string(&(drv->cmd), var, value);
93 * 'diff.<what>.funcname' attribute can be specified in the configuration
94 * to define a customized regexp to find the beginning of a function to
95 * be used for hunk header lines of "diff -p" style output.
97 static struct funcname_pattern {
98 char *name;
99 char *pattern;
100 struct funcname_pattern *next;
101 } *funcname_pattern_list;
103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
105 const char *name;
106 int namelen;
107 struct funcname_pattern *pp;
109 name = var + 5; /* "diff." */
110 namelen = ep - name;
112 for (pp = funcname_pattern_list; pp; pp = pp->next)
113 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114 break;
115 if (!pp) {
116 pp = xcalloc(1, sizeof(*pp));
117 pp->name = xmemdupz(name, namelen);
118 pp->next = funcname_pattern_list;
119 funcname_pattern_list = pp;
121 free(pp->pattern);
122 pp->pattern = xstrdup(value);
123 return 0;
127 * These are to give UI layer defaults.
128 * The core-level commands such as git-diff-files should
129 * never be affected by the setting of diff.renames
130 * the user happens to have in the configuration file.
132 int git_diff_ui_config(const char *var, const char *value, void *cb)
134 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
135 diff_use_color_default = git_config_colorbool(var, value, -1);
136 return 0;
138 if (!strcmp(var, "diff.renames")) {
139 if (!value)
140 diff_detect_rename_default = DIFF_DETECT_RENAME;
141 else if (!strcasecmp(value, "copies") ||
142 !strcasecmp(value, "copy"))
143 diff_detect_rename_default = DIFF_DETECT_COPY;
144 else if (git_config_bool(var,value))
145 diff_detect_rename_default = DIFF_DETECT_RENAME;
146 return 0;
148 if (!strcmp(var, "diff.autorefreshindex")) {
149 diff_auto_refresh_index = git_config_bool(var, value);
150 return 0;
152 if (!strcmp(var, "diff.external"))
153 return git_config_string(&external_diff_cmd_cfg, var, value);
154 if (!prefixcmp(var, "diff.")) {
155 const char *ep = strrchr(var, '.');
157 if (ep != var + 4 && !strcmp(ep, ".command"))
158 return parse_lldiff_command(var, ep, value);
161 return git_diff_basic_config(var, value, cb);
164 int git_diff_basic_config(const char *var, const char *value, void *cb)
166 if (!strcmp(var, "diff.renamelimit")) {
167 diff_rename_limit_default = git_config_int(var, value);
168 return 0;
171 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
172 int slot = parse_diff_color_slot(var, 11);
173 if (!value)
174 return config_error_nonbool(var);
175 color_parse(value, var, diff_colors[slot]);
176 return 0;
179 if (!prefixcmp(var, "diff.")) {
180 const char *ep = strrchr(var, '.');
181 if (ep != var + 4) {
182 if (!strcmp(ep, ".funcname")) {
183 if (!value)
184 return config_error_nonbool(var);
185 return parse_funcname_pattern(var, ep, value);
190 return git_color_default_config(var, value, cb);
193 static char *quote_two(const char *one, const char *two)
195 int need_one = quote_c_style(one, NULL, NULL, 1);
196 int need_two = quote_c_style(two, NULL, NULL, 1);
197 struct strbuf res;
199 strbuf_init(&res, 0);
200 if (need_one + need_two) {
201 strbuf_addch(&res, '"');
202 quote_c_style(one, &res, NULL, 1);
203 quote_c_style(two, &res, NULL, 1);
204 strbuf_addch(&res, '"');
205 } else {
206 strbuf_addstr(&res, one);
207 strbuf_addstr(&res, two);
209 return strbuf_detach(&res, NULL);
212 static const char *external_diff(void)
214 static const char *external_diff_cmd = NULL;
215 static int done_preparing = 0;
217 if (done_preparing)
218 return external_diff_cmd;
219 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
220 if (!external_diff_cmd)
221 external_diff_cmd = external_diff_cmd_cfg;
222 done_preparing = 1;
223 return external_diff_cmd;
226 static struct diff_tempfile {
227 const char *name; /* filename external diff should read from */
228 char hex[41];
229 char mode[10];
230 char tmp_path[PATH_MAX];
231 } diff_temp[2];
233 static int count_lines(const char *data, int size)
235 int count, ch, completely_empty = 1, nl_just_seen = 0;
236 count = 0;
237 while (0 < size--) {
238 ch = *data++;
239 if (ch == '\n') {
240 count++;
241 nl_just_seen = 1;
242 completely_empty = 0;
244 else {
245 nl_just_seen = 0;
246 completely_empty = 0;
249 if (completely_empty)
250 return 0;
251 if (!nl_just_seen)
252 count++; /* no trailing newline */
253 return count;
256 static void print_line_count(FILE *file, int count)
258 switch (count) {
259 case 0:
260 fprintf(file, "0,0");
261 break;
262 case 1:
263 fprintf(file, "1");
264 break;
265 default:
266 fprintf(file, "1,%d", count);
267 break;
271 static void copy_file_with_prefix(FILE *file,
272 int prefix, const char *data, int size,
273 const char *set, const char *reset)
275 int ch, nl_just_seen = 1;
276 while (0 < size--) {
277 ch = *data++;
278 if (nl_just_seen) {
279 fputs(set, file);
280 putc(prefix, file);
282 if (ch == '\n') {
283 nl_just_seen = 1;
284 fputs(reset, file);
285 } else
286 nl_just_seen = 0;
287 putc(ch, file);
289 if (!nl_just_seen)
290 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
293 static void emit_rewrite_diff(const char *name_a,
294 const char *name_b,
295 struct diff_filespec *one,
296 struct diff_filespec *two,
297 struct diff_options *o)
299 int lc_a, lc_b;
300 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
301 const char *name_a_tab, *name_b_tab;
302 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
303 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
304 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
305 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
306 const char *reset = diff_get_color(color_diff, DIFF_RESET);
307 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
309 name_a += (*name_a == '/');
310 name_b += (*name_b == '/');
311 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
312 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
314 strbuf_reset(&a_name);
315 strbuf_reset(&b_name);
316 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
317 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
319 diff_populate_filespec(one, 0);
320 diff_populate_filespec(two, 0);
321 lc_a = count_lines(one->data, one->size);
322 lc_b = count_lines(two->data, two->size);
323 fprintf(o->file,
324 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
325 metainfo, a_name.buf, name_a_tab, reset,
326 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
327 print_line_count(o->file, lc_a);
328 fprintf(o->file, " +");
329 print_line_count(o->file, lc_b);
330 fprintf(o->file, " @@%s\n", reset);
331 if (lc_a)
332 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
333 if (lc_b)
334 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
339 if (!DIFF_FILE_VALID(one)) {
340 mf->ptr = (char *)""; /* does not matter */
341 mf->size = 0;
342 return 0;
344 else if (diff_populate_filespec(one, 0))
345 return -1;
346 mf->ptr = one->data;
347 mf->size = one->size;
348 return 0;
351 struct diff_words_buffer {
352 mmfile_t text;
353 long alloc;
354 long current; /* output pointer */
355 int suppressed_newline;
358 static void diff_words_append(char *line, unsigned long len,
359 struct diff_words_buffer *buffer)
361 if (buffer->text.size + len > buffer->alloc) {
362 buffer->alloc = (buffer->text.size + len) * 3 / 2;
363 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
365 line++;
366 len--;
367 memcpy(buffer->text.ptr + buffer->text.size, line, len);
368 buffer->text.size += len;
371 struct diff_words_data {
372 struct xdiff_emit_state xm;
373 struct diff_words_buffer minus, plus;
374 FILE *file;
377 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
378 int suppress_newline)
380 const char *ptr;
381 int eol = 0;
383 if (len == 0)
384 return;
386 ptr = buffer->text.ptr + buffer->current;
387 buffer->current += len;
389 if (ptr[len - 1] == '\n') {
390 eol = 1;
391 len--;
394 fputs(diff_get_color(1, color), file);
395 fwrite(ptr, len, 1, file);
396 fputs(diff_get_color(1, DIFF_RESET), file);
398 if (eol) {
399 if (suppress_newline)
400 buffer->suppressed_newline = 1;
401 else
402 putc('\n', file);
406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
408 struct diff_words_data *diff_words = priv;
410 if (diff_words->minus.suppressed_newline) {
411 if (line[0] != '+')
412 putc('\n', diff_words->file);
413 diff_words->minus.suppressed_newline = 0;
416 len--;
417 switch (line[0]) {
418 case '-':
419 print_word(diff_words->file,
420 &diff_words->minus, len, DIFF_FILE_OLD, 1);
421 break;
422 case '+':
423 print_word(diff_words->file,
424 &diff_words->plus, len, DIFF_FILE_NEW, 0);
425 break;
426 case ' ':
427 print_word(diff_words->file,
428 &diff_words->plus, len, DIFF_PLAIN, 0);
429 diff_words->minus.current += len;
430 break;
434 /* this executes the word diff on the accumulated buffers */
435 static void diff_words_show(struct diff_words_data *diff_words)
437 xpparam_t xpp;
438 xdemitconf_t xecfg;
439 xdemitcb_t ecb;
440 mmfile_t minus, plus;
441 int i;
443 memset(&xecfg, 0, sizeof(xecfg));
444 minus.size = diff_words->minus.text.size;
445 minus.ptr = xmalloc(minus.size);
446 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
447 for (i = 0; i < minus.size; i++)
448 if (isspace(minus.ptr[i]))
449 minus.ptr[i] = '\n';
450 diff_words->minus.current = 0;
452 plus.size = diff_words->plus.text.size;
453 plus.ptr = xmalloc(plus.size);
454 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
455 for (i = 0; i < plus.size; i++)
456 if (isspace(plus.ptr[i]))
457 plus.ptr[i] = '\n';
458 diff_words->plus.current = 0;
460 xpp.flags = XDF_NEED_MINIMAL;
461 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
462 ecb.outf = xdiff_outf;
463 ecb.priv = diff_words;
464 diff_words->xm.consume = fn_out_diff_words_aux;
465 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
467 free(minus.ptr);
468 free(plus.ptr);
469 diff_words->minus.text.size = diff_words->plus.text.size = 0;
471 if (diff_words->minus.suppressed_newline) {
472 putc('\n', diff_words->file);
473 diff_words->minus.suppressed_newline = 0;
477 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
479 struct emit_callback {
480 struct xdiff_emit_state xm;
481 int nparents, color_diff;
482 unsigned ws_rule;
483 sane_truncate_fn truncate;
484 const char **label_path;
485 struct diff_words_data *diff_words;
486 int *found_changesp;
487 FILE *file;
490 static void free_diff_words_data(struct emit_callback *ecbdata)
492 if (ecbdata->diff_words) {
493 /* flush buffers */
494 if (ecbdata->diff_words->minus.text.size ||
495 ecbdata->diff_words->plus.text.size)
496 diff_words_show(ecbdata->diff_words);
498 free (ecbdata->diff_words->minus.text.ptr);
499 free (ecbdata->diff_words->plus.text.ptr);
500 free(ecbdata->diff_words);
501 ecbdata->diff_words = NULL;
505 const char *diff_get_color(int diff_use_color, enum color_diff ix)
507 if (diff_use_color)
508 return diff_colors[ix];
509 return "";
512 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
514 int has_trailing_newline = (len > 0 && line[len-1] == '\n');
515 if (has_trailing_newline)
516 len--;
518 fputs(set, file);
519 fwrite(line, len, 1, file);
520 fputs(reset, file);
521 if (has_trailing_newline)
522 fputc('\n', file);
525 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
527 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
528 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
530 if (!*ws)
531 emit_line(ecbdata->file, set, reset, line, len);
532 else {
533 /* Emit just the prefix, then the rest. */
534 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
535 ws_check_emit(line + ecbdata->nparents,
536 len - ecbdata->nparents, ecbdata->ws_rule,
537 ecbdata->file, set, reset, ws);
541 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
543 const char *cp;
544 unsigned long allot;
545 size_t l = len;
547 if (ecb->truncate)
548 return ecb->truncate(line, len);
549 cp = line;
550 allot = l;
551 while (0 < l) {
552 (void) utf8_width(&cp, &l);
553 if (!cp)
554 break; /* truncated in the middle? */
556 return allot - l;
559 static void fn_out_consume(void *priv, char *line, unsigned long len)
561 int i;
562 int color;
563 struct emit_callback *ecbdata = priv;
564 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
565 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
566 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
568 *(ecbdata->found_changesp) = 1;
570 if (ecbdata->label_path[0]) {
571 const char *name_a_tab, *name_b_tab;
573 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
574 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
576 fprintf(ecbdata->file, "%s--- %s%s%s\n",
577 meta, ecbdata->label_path[0], reset, name_a_tab);
578 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
579 meta, ecbdata->label_path[1], reset, name_b_tab);
580 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
583 /* This is not really necessary for now because
584 * this codepath only deals with two-way diffs.
586 for (i = 0; i < len && line[i] == '@'; i++)
588 if (2 <= i && i < len && line[i] == ' ') {
589 ecbdata->nparents = i - 1;
590 len = sane_truncate_line(ecbdata, line, len);
591 emit_line(ecbdata->file,
592 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
593 reset, line, len);
594 if (line[len-1] != '\n')
595 putc('\n', ecbdata->file);
596 return;
599 if (len < ecbdata->nparents) {
600 emit_line(ecbdata->file, reset, reset, line, len);
601 return;
604 color = DIFF_PLAIN;
605 if (ecbdata->diff_words && ecbdata->nparents != 1)
606 /* fall back to normal diff */
607 free_diff_words_data(ecbdata);
608 if (ecbdata->diff_words) {
609 if (line[0] == '-') {
610 diff_words_append(line, len,
611 &ecbdata->diff_words->minus);
612 return;
613 } else if (line[0] == '+') {
614 diff_words_append(line, len,
615 &ecbdata->diff_words->plus);
616 return;
618 if (ecbdata->diff_words->minus.text.size ||
619 ecbdata->diff_words->plus.text.size)
620 diff_words_show(ecbdata->diff_words);
621 line++;
622 len--;
623 emit_line(ecbdata->file, plain, reset, line, len);
624 return;
626 for (i = 0; i < ecbdata->nparents && len; i++) {
627 if (line[i] == '-')
628 color = DIFF_FILE_OLD;
629 else if (line[i] == '+')
630 color = DIFF_FILE_NEW;
633 if (color != DIFF_FILE_NEW) {
634 emit_line(ecbdata->file,
635 diff_get_color(ecbdata->color_diff, color),
636 reset, line, len);
637 return;
639 emit_add_line(reset, ecbdata, line, len);
642 static char *pprint_rename(const char *a, const char *b)
644 const char *old = a;
645 const char *new = b;
646 struct strbuf name;
647 int pfx_length, sfx_length;
648 int len_a = strlen(a);
649 int len_b = strlen(b);
650 int a_midlen, b_midlen;
651 int qlen_a = quote_c_style(a, NULL, NULL, 0);
652 int qlen_b = quote_c_style(b, NULL, NULL, 0);
654 strbuf_init(&name, 0);
655 if (qlen_a || qlen_b) {
656 quote_c_style(a, &name, NULL, 0);
657 strbuf_addstr(&name, " => ");
658 quote_c_style(b, &name, NULL, 0);
659 return strbuf_detach(&name, NULL);
662 /* Find common prefix */
663 pfx_length = 0;
664 while (*old && *new && *old == *new) {
665 if (*old == '/')
666 pfx_length = old - a + 1;
667 old++;
668 new++;
671 /* Find common suffix */
672 old = a + len_a;
673 new = b + len_b;
674 sfx_length = 0;
675 while (a <= old && b <= new && *old == *new) {
676 if (*old == '/')
677 sfx_length = len_a - (old - a);
678 old--;
679 new--;
683 * pfx{mid-a => mid-b}sfx
684 * {pfx-a => pfx-b}sfx
685 * pfx{sfx-a => sfx-b}
686 * name-a => name-b
688 a_midlen = len_a - pfx_length - sfx_length;
689 b_midlen = len_b - pfx_length - sfx_length;
690 if (a_midlen < 0)
691 a_midlen = 0;
692 if (b_midlen < 0)
693 b_midlen = 0;
695 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
696 if (pfx_length + sfx_length) {
697 strbuf_add(&name, a, pfx_length);
698 strbuf_addch(&name, '{');
700 strbuf_add(&name, a + pfx_length, a_midlen);
701 strbuf_addstr(&name, " => ");
702 strbuf_add(&name, b + pfx_length, b_midlen);
703 if (pfx_length + sfx_length) {
704 strbuf_addch(&name, '}');
705 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
707 return strbuf_detach(&name, NULL);
710 struct diffstat_t {
711 struct xdiff_emit_state xm;
713 int nr;
714 int alloc;
715 struct diffstat_file {
716 char *from_name;
717 char *name;
718 char *print_name;
719 unsigned is_unmerged:1;
720 unsigned is_binary:1;
721 unsigned is_renamed:1;
722 unsigned int added, deleted;
723 } **files;
726 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
727 const char *name_a,
728 const char *name_b)
730 struct diffstat_file *x;
731 x = xcalloc(sizeof (*x), 1);
732 if (diffstat->nr == diffstat->alloc) {
733 diffstat->alloc = alloc_nr(diffstat->alloc);
734 diffstat->files = xrealloc(diffstat->files,
735 diffstat->alloc * sizeof(x));
737 diffstat->files[diffstat->nr++] = x;
738 if (name_b) {
739 x->from_name = xstrdup(name_a);
740 x->name = xstrdup(name_b);
741 x->is_renamed = 1;
743 else {
744 x->from_name = NULL;
745 x->name = xstrdup(name_a);
747 return x;
750 static void diffstat_consume(void *priv, char *line, unsigned long len)
752 struct diffstat_t *diffstat = priv;
753 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
755 if (line[0] == '+')
756 x->added++;
757 else if (line[0] == '-')
758 x->deleted++;
761 const char mime_boundary_leader[] = "------------";
763 static int scale_linear(int it, int width, int max_change)
766 * make sure that at least one '-' is printed if there were deletions,
767 * and likewise for '+'.
769 if (max_change < 2)
770 return it;
771 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
774 static void show_name(FILE *file,
775 const char *prefix, const char *name, int len,
776 const char *reset, const char *set)
778 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
781 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
783 if (cnt <= 0)
784 return;
785 fprintf(file, "%s", set);
786 while (cnt--)
787 putc(ch, file);
788 fprintf(file, "%s", reset);
791 static void fill_print_name(struct diffstat_file *file)
793 char *pname;
795 if (file->print_name)
796 return;
798 if (!file->is_renamed) {
799 struct strbuf buf;
800 strbuf_init(&buf, 0);
801 if (quote_c_style(file->name, &buf, NULL, 0)) {
802 pname = strbuf_detach(&buf, NULL);
803 } else {
804 pname = file->name;
805 strbuf_release(&buf);
807 } else {
808 pname = pprint_rename(file->from_name, file->name);
810 file->print_name = pname;
813 static void show_stats(struct diffstat_t* data, struct diff_options *options)
815 int i, len, add, del, total, adds = 0, dels = 0;
816 int max_change = 0, max_len = 0;
817 int total_files = data->nr;
818 int width, name_width;
819 const char *reset, *set, *add_c, *del_c;
821 if (data->nr == 0)
822 return;
824 width = options->stat_width ? options->stat_width : 80;
825 name_width = options->stat_name_width ? options->stat_name_width : 50;
827 /* Sanity: give at least 5 columns to the graph,
828 * but leave at least 10 columns for the name.
830 if (width < 25)
831 width = 25;
832 if (name_width < 10)
833 name_width = 10;
834 else if (width < name_width + 15)
835 name_width = width - 15;
837 /* Find the longest filename and max number of changes */
838 reset = diff_get_color_opt(options, DIFF_RESET);
839 set = diff_get_color_opt(options, DIFF_PLAIN);
840 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
841 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
843 for (i = 0; i < data->nr; i++) {
844 struct diffstat_file *file = data->files[i];
845 int change = file->added + file->deleted;
846 fill_print_name(file);
847 len = strlen(file->print_name);
848 if (max_len < len)
849 max_len = len;
851 if (file->is_binary || file->is_unmerged)
852 continue;
853 if (max_change < change)
854 max_change = change;
857 /* Compute the width of the graph part;
858 * 10 is for one blank at the beginning of the line plus
859 * " | count " between the name and the graph.
861 * From here on, name_width is the width of the name area,
862 * and width is the width of the graph area.
864 name_width = (name_width < max_len) ? name_width : max_len;
865 if (width < (name_width + 10) + max_change)
866 width = width - (name_width + 10);
867 else
868 width = max_change;
870 for (i = 0; i < data->nr; i++) {
871 const char *prefix = "";
872 char *name = data->files[i]->print_name;
873 int added = data->files[i]->added;
874 int deleted = data->files[i]->deleted;
875 int name_len;
878 * "scale" the filename
880 len = name_width;
881 name_len = strlen(name);
882 if (name_width < name_len) {
883 char *slash;
884 prefix = "...";
885 len -= 3;
886 name += name_len - len;
887 slash = strchr(name, '/');
888 if (slash)
889 name = slash;
892 if (data->files[i]->is_binary) {
893 show_name(options->file, prefix, name, len, reset, set);
894 fprintf(options->file, " Bin ");
895 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
896 fprintf(options->file, " -> ");
897 fprintf(options->file, "%s%d%s", add_c, added, reset);
898 fprintf(options->file, " bytes");
899 fprintf(options->file, "\n");
900 continue;
902 else if (data->files[i]->is_unmerged) {
903 show_name(options->file, prefix, name, len, reset, set);
904 fprintf(options->file, " Unmerged\n");
905 continue;
907 else if (!data->files[i]->is_renamed &&
908 (added + deleted == 0)) {
909 total_files--;
910 continue;
914 * scale the add/delete
916 add = added;
917 del = deleted;
918 total = add + del;
919 adds += add;
920 dels += del;
922 if (width <= max_change) {
923 add = scale_linear(add, width, max_change);
924 del = scale_linear(del, width, max_change);
925 total = add + del;
927 show_name(options->file, prefix, name, len, reset, set);
928 fprintf(options->file, "%5d%s", added + deleted,
929 added + deleted ? " " : "");
930 show_graph(options->file, '+', add, add_c, reset);
931 show_graph(options->file, '-', del, del_c, reset);
932 fprintf(options->file, "\n");
934 fprintf(options->file,
935 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
936 set, total_files, adds, dels, reset);
939 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
941 int i, adds = 0, dels = 0, total_files = data->nr;
943 if (data->nr == 0)
944 return;
946 for (i = 0; i < data->nr; i++) {
947 if (!data->files[i]->is_binary &&
948 !data->files[i]->is_unmerged) {
949 int added = data->files[i]->added;
950 int deleted= data->files[i]->deleted;
951 if (!data->files[i]->is_renamed &&
952 (added + deleted == 0)) {
953 total_files--;
954 } else {
955 adds += added;
956 dels += deleted;
960 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
961 total_files, adds, dels);
964 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
966 int i;
968 if (data->nr == 0)
969 return;
971 for (i = 0; i < data->nr; i++) {
972 struct diffstat_file *file = data->files[i];
974 if (file->is_binary)
975 fprintf(options->file, "-\t-\t");
976 else
977 fprintf(options->file,
978 "%d\t%d\t", file->added, file->deleted);
979 if (options->line_termination) {
980 fill_print_name(file);
981 if (!file->is_renamed)
982 write_name_quoted(file->name, options->file,
983 options->line_termination);
984 else {
985 fputs(file->print_name, options->file);
986 putc(options->line_termination, options->file);
988 } else {
989 if (file->is_renamed) {
990 putc('\0', options->file);
991 write_name_quoted(file->from_name, options->file, '\0');
993 write_name_quoted(file->name, options->file, '\0');
998 struct dirstat_file {
999 const char *name;
1000 unsigned long changed;
1003 struct dirstat_dir {
1004 struct dirstat_file *files;
1005 int alloc, nr, percent, cumulative;
1008 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1010 unsigned long this_dir = 0;
1011 unsigned int sources = 0;
1013 while (dir->nr) {
1014 struct dirstat_file *f = dir->files;
1015 int namelen = strlen(f->name);
1016 unsigned long this;
1017 char *slash;
1019 if (namelen < baselen)
1020 break;
1021 if (memcmp(f->name, base, baselen))
1022 break;
1023 slash = strchr(f->name + baselen, '/');
1024 if (slash) {
1025 int newbaselen = slash + 1 - f->name;
1026 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1027 sources++;
1028 } else {
1029 this = f->changed;
1030 dir->files++;
1031 dir->nr--;
1032 sources += 2;
1034 this_dir += this;
1038 * We don't report dirstat's for
1039 * - the top level
1040 * - or cases where everything came from a single directory
1041 * under this directory (sources == 1).
1043 if (baselen && sources != 1) {
1044 int permille = this_dir * 1000 / changed;
1045 if (permille) {
1046 int percent = permille / 10;
1047 if (percent >= dir->percent) {
1048 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1049 if (!dir->cumulative)
1050 return 0;
1054 return this_dir;
1057 static void show_dirstat(struct diff_options *options)
1059 int i;
1060 unsigned long changed;
1061 struct dirstat_dir dir;
1062 struct diff_queue_struct *q = &diff_queued_diff;
1064 dir.files = NULL;
1065 dir.alloc = 0;
1066 dir.nr = 0;
1067 dir.percent = options->dirstat_percent;
1068 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1070 changed = 0;
1071 for (i = 0; i < q->nr; i++) {
1072 struct diff_filepair *p = q->queue[i];
1073 const char *name;
1074 unsigned long copied, added, damage;
1076 name = p->one->path ? p->one->path : p->two->path;
1078 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1079 diff_populate_filespec(p->one, 0);
1080 diff_populate_filespec(p->two, 0);
1081 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1082 &copied, &added);
1083 diff_free_filespec_data(p->one);
1084 diff_free_filespec_data(p->two);
1085 } else if (DIFF_FILE_VALID(p->one)) {
1086 diff_populate_filespec(p->one, 1);
1087 copied = added = 0;
1088 diff_free_filespec_data(p->one);
1089 } else if (DIFF_FILE_VALID(p->two)) {
1090 diff_populate_filespec(p->two, 1);
1091 copied = 0;
1092 added = p->two->size;
1093 diff_free_filespec_data(p->two);
1094 } else
1095 continue;
1098 * Original minus copied is the removed material,
1099 * added is the new material. They are both damages
1100 * made to the preimage.
1102 damage = (p->one->size - copied) + added;
1104 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1105 dir.files[dir.nr].name = name;
1106 dir.files[dir.nr].changed = damage;
1107 changed += damage;
1108 dir.nr++;
1111 /* This can happen even with many files, if everything was renames */
1112 if (!changed)
1113 return;
1115 /* Show all directories with more than x% of the changes */
1116 gather_dirstat(options->file, &dir, changed, "", 0);
1119 static void free_diffstat_info(struct diffstat_t *diffstat)
1121 int i;
1122 for (i = 0; i < diffstat->nr; i++) {
1123 struct diffstat_file *f = diffstat->files[i];
1124 if (f->name != f->print_name)
1125 free(f->print_name);
1126 free(f->name);
1127 free(f->from_name);
1128 free(f);
1130 free(diffstat->files);
1133 struct checkdiff_t {
1134 struct xdiff_emit_state xm;
1135 const char *filename;
1136 int lineno;
1137 struct diff_options *o;
1138 unsigned ws_rule;
1139 unsigned status;
1140 int trailing_blanks_start;
1143 static int is_conflict_marker(const char *line, unsigned long len)
1145 char firstchar;
1146 int cnt;
1148 if (len < 8)
1149 return 0;
1150 firstchar = line[0];
1151 switch (firstchar) {
1152 case '=': case '>': case '<':
1153 break;
1154 default:
1155 return 0;
1157 for (cnt = 1; cnt < 7; cnt++)
1158 if (line[cnt] != firstchar)
1159 return 0;
1160 /* line[0] thru line[6] are same as firstchar */
1161 if (firstchar == '=') {
1162 /* divider between ours and theirs? */
1163 if (len != 8 || line[7] != '\n')
1164 return 0;
1165 } else if (len < 8 || !isspace(line[7])) {
1166 /* not divider before ours nor after theirs */
1167 return 0;
1169 return 1;
1172 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1174 struct checkdiff_t *data = priv;
1175 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1176 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1177 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1178 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1179 char *err;
1181 if (line[0] == '+') {
1182 unsigned bad;
1183 data->lineno++;
1184 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1185 data->trailing_blanks_start = 0;
1186 else if (!data->trailing_blanks_start)
1187 data->trailing_blanks_start = data->lineno;
1188 if (is_conflict_marker(line + 1, len - 1)) {
1189 data->status |= 1;
1190 fprintf(data->o->file,
1191 "%s:%d: leftover conflict marker\n",
1192 data->filename, data->lineno);
1194 bad = ws_check(line + 1, len - 1, data->ws_rule);
1195 if (!bad)
1196 return;
1197 data->status |= bad;
1198 err = whitespace_error_string(bad);
1199 fprintf(data->o->file, "%s:%d: %s.\n",
1200 data->filename, data->lineno, err);
1201 free(err);
1202 emit_line(data->o->file, set, reset, line, 1);
1203 ws_check_emit(line + 1, len - 1, data->ws_rule,
1204 data->o->file, set, reset, ws);
1205 } else if (line[0] == ' ') {
1206 data->lineno++;
1207 data->trailing_blanks_start = 0;
1208 } else if (line[0] == '@') {
1209 char *plus = strchr(line, '+');
1210 if (plus)
1211 data->lineno = strtol(plus, NULL, 10) - 1;
1212 else
1213 die("invalid diff");
1214 data->trailing_blanks_start = 0;
1218 static unsigned char *deflate_it(char *data,
1219 unsigned long size,
1220 unsigned long *result_size)
1222 int bound;
1223 unsigned char *deflated;
1224 z_stream stream;
1226 memset(&stream, 0, sizeof(stream));
1227 deflateInit(&stream, zlib_compression_level);
1228 bound = deflateBound(&stream, size);
1229 deflated = xmalloc(bound);
1230 stream.next_out = deflated;
1231 stream.avail_out = bound;
1233 stream.next_in = (unsigned char *)data;
1234 stream.avail_in = size;
1235 while (deflate(&stream, Z_FINISH) == Z_OK)
1236 ; /* nothing */
1237 deflateEnd(&stream);
1238 *result_size = stream.total_out;
1239 return deflated;
1242 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1244 void *cp;
1245 void *delta;
1246 void *deflated;
1247 void *data;
1248 unsigned long orig_size;
1249 unsigned long delta_size;
1250 unsigned long deflate_size;
1251 unsigned long data_size;
1253 /* We could do deflated delta, or we could do just deflated two,
1254 * whichever is smaller.
1256 delta = NULL;
1257 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1258 if (one->size && two->size) {
1259 delta = diff_delta(one->ptr, one->size,
1260 two->ptr, two->size,
1261 &delta_size, deflate_size);
1262 if (delta) {
1263 void *to_free = delta;
1264 orig_size = delta_size;
1265 delta = deflate_it(delta, delta_size, &delta_size);
1266 free(to_free);
1270 if (delta && delta_size < deflate_size) {
1271 fprintf(file, "delta %lu\n", orig_size);
1272 free(deflated);
1273 data = delta;
1274 data_size = delta_size;
1276 else {
1277 fprintf(file, "literal %lu\n", two->size);
1278 free(delta);
1279 data = deflated;
1280 data_size = deflate_size;
1283 /* emit data encoded in base85 */
1284 cp = data;
1285 while (data_size) {
1286 int bytes = (52 < data_size) ? 52 : data_size;
1287 char line[70];
1288 data_size -= bytes;
1289 if (bytes <= 26)
1290 line[0] = bytes + 'A' - 1;
1291 else
1292 line[0] = bytes - 26 + 'a' - 1;
1293 encode_85(line + 1, cp, bytes);
1294 cp = (char *) cp + bytes;
1295 fputs(line, file);
1296 fputc('\n', file);
1298 fprintf(file, "\n");
1299 free(data);
1302 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1304 fprintf(file, "GIT binary patch\n");
1305 emit_binary_diff_body(file, one, two);
1306 emit_binary_diff_body(file, two, one);
1309 static void setup_diff_attr_check(struct git_attr_check *check)
1311 static struct git_attr *attr_diff;
1313 if (!attr_diff) {
1314 attr_diff = git_attr("diff", 4);
1316 check[0].attr = attr_diff;
1319 static void diff_filespec_check_attr(struct diff_filespec *one)
1321 struct git_attr_check attr_diff_check;
1322 int check_from_data = 0;
1324 if (one->checked_attr)
1325 return;
1327 setup_diff_attr_check(&attr_diff_check);
1328 one->is_binary = 0;
1329 one->funcname_pattern_ident = NULL;
1331 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1332 const char *value;
1334 /* binaryness */
1335 value = attr_diff_check.value;
1336 if (ATTR_TRUE(value))
1338 else if (ATTR_FALSE(value))
1339 one->is_binary = 1;
1340 else
1341 check_from_data = 1;
1343 /* funcname pattern ident */
1344 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1346 else
1347 one->funcname_pattern_ident = value;
1350 if (check_from_data) {
1351 if (!one->data && DIFF_FILE_VALID(one))
1352 diff_populate_filespec(one, 0);
1354 if (one->data)
1355 one->is_binary = buffer_is_binary(one->data, one->size);
1359 int diff_filespec_is_binary(struct diff_filespec *one)
1361 diff_filespec_check_attr(one);
1362 return one->is_binary;
1365 static const char *funcname_pattern(const char *ident)
1367 struct funcname_pattern *pp;
1369 for (pp = funcname_pattern_list; pp; pp = pp->next)
1370 if (!strcmp(ident, pp->name))
1371 return pp->pattern;
1372 return NULL;
1375 static struct builtin_funcname_pattern {
1376 const char *name;
1377 const char *pattern;
1378 } builtin_funcname_pattern[] = {
1379 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1380 "new\\|return\\|switch\\|throw\\|while\\)\n"
1381 "^[ ]*\\(\\([ ]*"
1382 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1383 "[ ]*([^;]*\\)$" },
1384 { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1385 "destructor\\|interface\\|implementation\\|"
1386 "initialization\\|finalization\\)[ \t]*.*\\)$"
1387 "\\|"
1388 "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1390 { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
1391 { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1392 { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1395 static const char *diff_funcname_pattern(struct diff_filespec *one)
1397 const char *ident, *pattern;
1398 int i;
1400 diff_filespec_check_attr(one);
1401 ident = one->funcname_pattern_ident;
1403 if (!ident)
1405 * If the config file has "funcname.default" defined, that
1406 * regexp is used; otherwise NULL is returned and xemit uses
1407 * the built-in default.
1409 return funcname_pattern("default");
1411 /* Look up custom "funcname.$ident" regexp from config. */
1412 pattern = funcname_pattern(ident);
1413 if (pattern)
1414 return pattern;
1417 * And define built-in fallback patterns here. Note that
1418 * these can be overridden by the user's config settings.
1420 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1421 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1422 return builtin_funcname_pattern[i].pattern;
1424 return NULL;
1427 static void builtin_diff(const char *name_a,
1428 const char *name_b,
1429 struct diff_filespec *one,
1430 struct diff_filespec *two,
1431 const char *xfrm_msg,
1432 struct diff_options *o,
1433 int complete_rewrite)
1435 mmfile_t mf1, mf2;
1436 const char *lbl[2];
1437 char *a_one, *b_two;
1438 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1439 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1441 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1442 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1443 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1444 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1445 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1446 if (lbl[0][0] == '/') {
1447 /* /dev/null */
1448 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1449 if (xfrm_msg && xfrm_msg[0])
1450 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1452 else if (lbl[1][0] == '/') {
1453 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1454 if (xfrm_msg && xfrm_msg[0])
1455 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1457 else {
1458 if (one->mode != two->mode) {
1459 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1460 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1462 if (xfrm_msg && xfrm_msg[0])
1463 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1465 * we do not run diff between different kind
1466 * of objects.
1468 if ((one->mode ^ two->mode) & S_IFMT)
1469 goto free_ab_and_return;
1470 if (complete_rewrite) {
1471 emit_rewrite_diff(name_a, name_b, one, two, o);
1472 o->found_changes = 1;
1473 goto free_ab_and_return;
1477 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1478 die("unable to read files to diff");
1480 if (!DIFF_OPT_TST(o, TEXT) &&
1481 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1482 /* Quite common confusing case */
1483 if (mf1.size == mf2.size &&
1484 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1485 goto free_ab_and_return;
1486 if (DIFF_OPT_TST(o, BINARY))
1487 emit_binary_diff(o->file, &mf1, &mf2);
1488 else
1489 fprintf(o->file, "Binary files %s and %s differ\n",
1490 lbl[0], lbl[1]);
1491 o->found_changes = 1;
1493 else {
1494 /* Crazy xdl interfaces.. */
1495 const char *diffopts = getenv("GIT_DIFF_OPTS");
1496 xpparam_t xpp;
1497 xdemitconf_t xecfg;
1498 xdemitcb_t ecb;
1499 struct emit_callback ecbdata;
1500 const char *funcname_pattern;
1502 funcname_pattern = diff_funcname_pattern(one);
1503 if (!funcname_pattern)
1504 funcname_pattern = diff_funcname_pattern(two);
1506 memset(&xecfg, 0, sizeof(xecfg));
1507 memset(&ecbdata, 0, sizeof(ecbdata));
1508 ecbdata.label_path = lbl;
1509 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1510 ecbdata.found_changesp = &o->found_changes;
1511 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1512 ecbdata.file = o->file;
1513 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1514 xecfg.ctxlen = o->context;
1515 xecfg.flags = XDL_EMIT_FUNCNAMES;
1516 if (funcname_pattern)
1517 xdiff_set_find_func(&xecfg, funcname_pattern);
1518 if (!diffopts)
1520 else if (!prefixcmp(diffopts, "--unified="))
1521 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1522 else if (!prefixcmp(diffopts, "-u"))
1523 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1524 ecb.outf = xdiff_outf;
1525 ecb.priv = &ecbdata;
1526 ecbdata.xm.consume = fn_out_consume;
1527 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1528 ecbdata.diff_words =
1529 xcalloc(1, sizeof(struct diff_words_data));
1530 ecbdata.diff_words->file = o->file;
1532 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1533 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1534 free_diff_words_data(&ecbdata);
1537 free_ab_and_return:
1538 diff_free_filespec_data(one);
1539 diff_free_filespec_data(two);
1540 free(a_one);
1541 free(b_two);
1542 return;
1545 static void builtin_diffstat(const char *name_a, const char *name_b,
1546 struct diff_filespec *one,
1547 struct diff_filespec *two,
1548 struct diffstat_t *diffstat,
1549 struct diff_options *o,
1550 int complete_rewrite)
1552 mmfile_t mf1, mf2;
1553 struct diffstat_file *data;
1555 data = diffstat_add(diffstat, name_a, name_b);
1557 if (!one || !two) {
1558 data->is_unmerged = 1;
1559 return;
1561 if (complete_rewrite) {
1562 diff_populate_filespec(one, 0);
1563 diff_populate_filespec(two, 0);
1564 data->deleted = count_lines(one->data, one->size);
1565 data->added = count_lines(two->data, two->size);
1566 goto free_and_return;
1568 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1569 die("unable to read files to diff");
1571 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1572 data->is_binary = 1;
1573 data->added = mf2.size;
1574 data->deleted = mf1.size;
1575 } else {
1576 /* Crazy xdl interfaces.. */
1577 xpparam_t xpp;
1578 xdemitconf_t xecfg;
1579 xdemitcb_t ecb;
1581 memset(&xecfg, 0, sizeof(xecfg));
1582 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1583 ecb.outf = xdiff_outf;
1584 ecb.priv = diffstat;
1585 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1588 free_and_return:
1589 diff_free_filespec_data(one);
1590 diff_free_filespec_data(two);
1593 static void builtin_checkdiff(const char *name_a, const char *name_b,
1594 const char *attr_path,
1595 struct diff_filespec *one,
1596 struct diff_filespec *two,
1597 struct diff_options *o)
1599 mmfile_t mf1, mf2;
1600 struct checkdiff_t data;
1602 if (!two)
1603 return;
1605 memset(&data, 0, sizeof(data));
1606 data.xm.consume = checkdiff_consume;
1607 data.filename = name_b ? name_b : name_a;
1608 data.lineno = 0;
1609 data.o = o;
1610 data.ws_rule = whitespace_rule(attr_path);
1612 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1613 die("unable to read files to diff");
1616 * All the other codepaths check both sides, but not checking
1617 * the "old" side here is deliberate. We are checking the newly
1618 * introduced changes, and as long as the "new" side is text, we
1619 * can and should check what it introduces.
1621 if (diff_filespec_is_binary(two))
1622 goto free_and_return;
1623 else {
1624 /* Crazy xdl interfaces.. */
1625 xpparam_t xpp;
1626 xdemitconf_t xecfg;
1627 xdemitcb_t ecb;
1629 memset(&xecfg, 0, sizeof(xecfg));
1630 xpp.flags = XDF_NEED_MINIMAL;
1631 ecb.outf = xdiff_outf;
1632 ecb.priv = &data;
1633 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1635 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1636 data.trailing_blanks_start) {
1637 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1638 data.filename, data.trailing_blanks_start);
1639 data.status = 1; /* report errors */
1642 free_and_return:
1643 diff_free_filespec_data(one);
1644 diff_free_filespec_data(two);
1645 if (data.status)
1646 DIFF_OPT_SET(o, CHECK_FAILED);
1649 struct diff_filespec *alloc_filespec(const char *path)
1651 int namelen = strlen(path);
1652 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1654 memset(spec, 0, sizeof(*spec));
1655 spec->path = (char *)(spec + 1);
1656 memcpy(spec->path, path, namelen+1);
1657 spec->count = 1;
1658 return spec;
1661 void free_filespec(struct diff_filespec *spec)
1663 if (!--spec->count) {
1664 diff_free_filespec_data(spec);
1665 free(spec);
1669 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1670 unsigned short mode)
1672 if (mode) {
1673 spec->mode = canon_mode(mode);
1674 hashcpy(spec->sha1, sha1);
1675 spec->sha1_valid = !is_null_sha1(sha1);
1680 * Given a name and sha1 pair, if the index tells us the file in
1681 * the work tree has that object contents, return true, so that
1682 * prepare_temp_file() does not have to inflate and extract.
1684 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1686 struct cache_entry *ce;
1687 struct stat st;
1688 int pos, len;
1690 /* We do not read the cache ourselves here, because the
1691 * benchmark with my previous version that always reads cache
1692 * shows that it makes things worse for diff-tree comparing
1693 * two linux-2.6 kernel trees in an already checked out work
1694 * tree. This is because most diff-tree comparisons deal with
1695 * only a small number of files, while reading the cache is
1696 * expensive for a large project, and its cost outweighs the
1697 * savings we get by not inflating the object to a temporary
1698 * file. Practically, this code only helps when we are used
1699 * by diff-cache --cached, which does read the cache before
1700 * calling us.
1702 if (!active_cache)
1703 return 0;
1705 /* We want to avoid the working directory if our caller
1706 * doesn't need the data in a normal file, this system
1707 * is rather slow with its stat/open/mmap/close syscalls,
1708 * and the object is contained in a pack file. The pack
1709 * is probably already open and will be faster to obtain
1710 * the data through than the working directory. Loose
1711 * objects however would tend to be slower as they need
1712 * to be individually opened and inflated.
1714 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1715 return 0;
1717 len = strlen(name);
1718 pos = cache_name_pos(name, len);
1719 if (pos < 0)
1720 return 0;
1721 ce = active_cache[pos];
1724 * This is not the sha1 we are looking for, or
1725 * unreusable because it is not a regular file.
1727 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1728 return 0;
1731 * If ce matches the file in the work tree, we can reuse it.
1733 if (ce_uptodate(ce) ||
1734 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1735 return 1;
1737 return 0;
1740 static int populate_from_stdin(struct diff_filespec *s)
1742 struct strbuf buf;
1743 size_t size = 0;
1745 strbuf_init(&buf, 0);
1746 if (strbuf_read(&buf, 0, 0) < 0)
1747 return error("error while reading from stdin %s",
1748 strerror(errno));
1750 s->should_munmap = 0;
1751 s->data = strbuf_detach(&buf, &size);
1752 s->size = size;
1753 s->should_free = 1;
1754 return 0;
1757 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1759 int len;
1760 char *data = xmalloc(100);
1761 len = snprintf(data, 100,
1762 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1763 s->data = data;
1764 s->size = len;
1765 s->should_free = 1;
1766 if (size_only) {
1767 s->data = NULL;
1768 free(data);
1770 return 0;
1774 * While doing rename detection and pickaxe operation, we may need to
1775 * grab the data for the blob (or file) for our own in-core comparison.
1776 * diff_filespec has data and size fields for this purpose.
1778 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1780 int err = 0;
1781 if (!DIFF_FILE_VALID(s))
1782 die("internal error: asking to populate invalid file.");
1783 if (S_ISDIR(s->mode))
1784 return -1;
1786 if (s->data)
1787 return 0;
1789 if (size_only && 0 < s->size)
1790 return 0;
1792 if (S_ISGITLINK(s->mode))
1793 return diff_populate_gitlink(s, size_only);
1795 if (!s->sha1_valid ||
1796 reuse_worktree_file(s->path, s->sha1, 0)) {
1797 struct strbuf buf;
1798 struct stat st;
1799 int fd;
1801 if (!strcmp(s->path, "-"))
1802 return populate_from_stdin(s);
1804 if (lstat(s->path, &st) < 0) {
1805 if (errno == ENOENT) {
1806 err_empty:
1807 err = -1;
1808 empty:
1809 s->data = (char *)"";
1810 s->size = 0;
1811 return err;
1814 s->size = xsize_t(st.st_size);
1815 if (!s->size)
1816 goto empty;
1817 if (size_only)
1818 return 0;
1819 if (S_ISLNK(st.st_mode)) {
1820 int ret;
1821 s->data = xmalloc(s->size);
1822 s->should_free = 1;
1823 ret = readlink(s->path, s->data, s->size);
1824 if (ret < 0) {
1825 free(s->data);
1826 goto err_empty;
1828 return 0;
1830 fd = open(s->path, O_RDONLY);
1831 if (fd < 0)
1832 goto err_empty;
1833 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1834 close(fd);
1835 s->should_munmap = 1;
1838 * Convert from working tree format to canonical git format
1840 strbuf_init(&buf, 0);
1841 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1842 size_t size = 0;
1843 munmap(s->data, s->size);
1844 s->should_munmap = 0;
1845 s->data = strbuf_detach(&buf, &size);
1846 s->size = size;
1847 s->should_free = 1;
1850 else {
1851 enum object_type type;
1852 if (size_only)
1853 type = sha1_object_info(s->sha1, &s->size);
1854 else {
1855 s->data = read_sha1_file(s->sha1, &type, &s->size);
1856 s->should_free = 1;
1859 return 0;
1862 void diff_free_filespec_blob(struct diff_filespec *s)
1864 if (s->should_free)
1865 free(s->data);
1866 else if (s->should_munmap)
1867 munmap(s->data, s->size);
1869 if (s->should_free || s->should_munmap) {
1870 s->should_free = s->should_munmap = 0;
1871 s->data = NULL;
1875 void diff_free_filespec_data(struct diff_filespec *s)
1877 diff_free_filespec_blob(s);
1878 free(s->cnt_data);
1879 s->cnt_data = NULL;
1882 static void prep_temp_blob(struct diff_tempfile *temp,
1883 void *blob,
1884 unsigned long size,
1885 const unsigned char *sha1,
1886 int mode)
1888 int fd;
1890 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1891 if (fd < 0)
1892 die("unable to create temp-file: %s", strerror(errno));
1893 if (write_in_full(fd, blob, size) != size)
1894 die("unable to write temp-file");
1895 close(fd);
1896 temp->name = temp->tmp_path;
1897 strcpy(temp->hex, sha1_to_hex(sha1));
1898 temp->hex[40] = 0;
1899 sprintf(temp->mode, "%06o", mode);
1902 static void prepare_temp_file(const char *name,
1903 struct diff_tempfile *temp,
1904 struct diff_filespec *one)
1906 if (!DIFF_FILE_VALID(one)) {
1907 not_a_valid_file:
1908 /* A '-' entry produces this for file-2, and
1909 * a '+' entry produces this for file-1.
1911 temp->name = "/dev/null";
1912 strcpy(temp->hex, ".");
1913 strcpy(temp->mode, ".");
1914 return;
1917 if (!one->sha1_valid ||
1918 reuse_worktree_file(name, one->sha1, 1)) {
1919 struct stat st;
1920 if (lstat(name, &st) < 0) {
1921 if (errno == ENOENT)
1922 goto not_a_valid_file;
1923 die("stat(%s): %s", name, strerror(errno));
1925 if (S_ISLNK(st.st_mode)) {
1926 int ret;
1927 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1928 size_t sz = xsize_t(st.st_size);
1929 if (sizeof(buf) <= st.st_size)
1930 die("symlink too long: %s", name);
1931 ret = readlink(name, buf, sz);
1932 if (ret < 0)
1933 die("readlink(%s)", name);
1934 prep_temp_blob(temp, buf, sz,
1935 (one->sha1_valid ?
1936 one->sha1 : null_sha1),
1937 (one->sha1_valid ?
1938 one->mode : S_IFLNK));
1940 else {
1941 /* we can borrow from the file in the work tree */
1942 temp->name = name;
1943 if (!one->sha1_valid)
1944 strcpy(temp->hex, sha1_to_hex(null_sha1));
1945 else
1946 strcpy(temp->hex, sha1_to_hex(one->sha1));
1947 /* Even though we may sometimes borrow the
1948 * contents from the work tree, we always want
1949 * one->mode. mode is trustworthy even when
1950 * !(one->sha1_valid), as long as
1951 * DIFF_FILE_VALID(one).
1953 sprintf(temp->mode, "%06o", one->mode);
1955 return;
1957 else {
1958 if (diff_populate_filespec(one, 0))
1959 die("cannot read data blob for %s", one->path);
1960 prep_temp_blob(temp, one->data, one->size,
1961 one->sha1, one->mode);
1965 static void remove_tempfile(void)
1967 int i;
1969 for (i = 0; i < 2; i++)
1970 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1971 unlink(diff_temp[i].name);
1972 diff_temp[i].name = NULL;
1976 static void remove_tempfile_on_signal(int signo)
1978 remove_tempfile();
1979 signal(SIGINT, SIG_DFL);
1980 raise(signo);
1983 /* An external diff command takes:
1985 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1986 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1989 static void run_external_diff(const char *pgm,
1990 const char *name,
1991 const char *other,
1992 struct diff_filespec *one,
1993 struct diff_filespec *two,
1994 const char *xfrm_msg,
1995 int complete_rewrite)
1997 const char *spawn_arg[10];
1998 struct diff_tempfile *temp = diff_temp;
1999 int retval;
2000 static int atexit_asked = 0;
2001 const char *othername;
2002 const char **arg = &spawn_arg[0];
2004 othername = (other? other : name);
2005 if (one && two) {
2006 prepare_temp_file(name, &temp[0], one);
2007 prepare_temp_file(othername, &temp[1], two);
2008 if (! atexit_asked &&
2009 (temp[0].name == temp[0].tmp_path ||
2010 temp[1].name == temp[1].tmp_path)) {
2011 atexit_asked = 1;
2012 atexit(remove_tempfile);
2014 signal(SIGINT, remove_tempfile_on_signal);
2017 if (one && two) {
2018 *arg++ = pgm;
2019 *arg++ = name;
2020 *arg++ = temp[0].name;
2021 *arg++ = temp[0].hex;
2022 *arg++ = temp[0].mode;
2023 *arg++ = temp[1].name;
2024 *arg++ = temp[1].hex;
2025 *arg++ = temp[1].mode;
2026 if (other) {
2027 *arg++ = other;
2028 *arg++ = xfrm_msg;
2030 } else {
2031 *arg++ = pgm;
2032 *arg++ = name;
2034 *arg = NULL;
2035 fflush(NULL);
2036 retval = run_command_v_opt(spawn_arg, 0);
2037 remove_tempfile();
2038 if (retval) {
2039 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2040 exit(1);
2044 static const char *external_diff_attr(const char *name)
2046 struct git_attr_check attr_diff_check;
2048 if (!name)
2049 return NULL;
2051 setup_diff_attr_check(&attr_diff_check);
2052 if (!git_checkattr(name, 1, &attr_diff_check)) {
2053 const char *value = attr_diff_check.value;
2054 if (!ATTR_TRUE(value) &&
2055 !ATTR_FALSE(value) &&
2056 !ATTR_UNSET(value)) {
2057 struct ll_diff_driver *drv;
2059 for (drv = user_diff; drv; drv = drv->next)
2060 if (!strcmp(drv->name, value))
2061 return drv->cmd;
2064 return NULL;
2067 static void run_diff_cmd(const char *pgm,
2068 const char *name,
2069 const char *other,
2070 const char *attr_path,
2071 struct diff_filespec *one,
2072 struct diff_filespec *two,
2073 const char *xfrm_msg,
2074 struct diff_options *o,
2075 int complete_rewrite)
2077 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2078 pgm = NULL;
2079 else {
2080 const char *cmd = external_diff_attr(attr_path);
2081 if (cmd)
2082 pgm = cmd;
2085 if (pgm) {
2086 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2087 complete_rewrite);
2088 return;
2090 if (one && two)
2091 builtin_diff(name, other ? other : name,
2092 one, two, xfrm_msg, o, complete_rewrite);
2093 else
2094 fprintf(o->file, "* Unmerged path %s\n", name);
2097 static void diff_fill_sha1_info(struct diff_filespec *one)
2099 if (DIFF_FILE_VALID(one)) {
2100 if (!one->sha1_valid) {
2101 struct stat st;
2102 if (!strcmp(one->path, "-")) {
2103 hashcpy(one->sha1, null_sha1);
2104 return;
2106 if (lstat(one->path, &st) < 0)
2107 die("stat %s", one->path);
2108 if (index_path(one->sha1, one->path, &st, 0))
2109 die("cannot hash %s\n", one->path);
2112 else
2113 hashclr(one->sha1);
2116 static int similarity_index(struct diff_filepair *p)
2118 return p->score * 100 / MAX_SCORE;
2121 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2123 /* Strip the prefix but do not molest /dev/null and absolute paths */
2124 if (*namep && **namep != '/')
2125 *namep += prefix_length;
2126 if (*otherp && **otherp != '/')
2127 *otherp += prefix_length;
2130 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2132 const char *pgm = external_diff();
2133 struct strbuf msg;
2134 char *xfrm_msg;
2135 struct diff_filespec *one = p->one;
2136 struct diff_filespec *two = p->two;
2137 const char *name;
2138 const char *other;
2139 const char *attr_path;
2140 int complete_rewrite = 0;
2142 name = p->one->path;
2143 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2144 attr_path = name;
2145 if (o->prefix_length)
2146 strip_prefix(o->prefix_length, &name, &other);
2148 if (DIFF_PAIR_UNMERGED(p)) {
2149 run_diff_cmd(pgm, name, NULL, attr_path,
2150 NULL, NULL, NULL, o, 0);
2151 return;
2154 diff_fill_sha1_info(one);
2155 diff_fill_sha1_info(two);
2157 strbuf_init(&msg, PATH_MAX * 2 + 300);
2158 switch (p->status) {
2159 case DIFF_STATUS_COPIED:
2160 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2161 strbuf_addstr(&msg, "\ncopy from ");
2162 quote_c_style(name, &msg, NULL, 0);
2163 strbuf_addstr(&msg, "\ncopy to ");
2164 quote_c_style(other, &msg, NULL, 0);
2165 strbuf_addch(&msg, '\n');
2166 break;
2167 case DIFF_STATUS_RENAMED:
2168 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2169 strbuf_addstr(&msg, "\nrename from ");
2170 quote_c_style(name, &msg, NULL, 0);
2171 strbuf_addstr(&msg, "\nrename to ");
2172 quote_c_style(other, &msg, NULL, 0);
2173 strbuf_addch(&msg, '\n');
2174 break;
2175 case DIFF_STATUS_MODIFIED:
2176 if (p->score) {
2177 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2178 similarity_index(p));
2179 complete_rewrite = 1;
2180 break;
2182 /* fallthru */
2183 default:
2184 /* nothing */
2188 if (hashcmp(one->sha1, two->sha1)) {
2189 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2191 if (DIFF_OPT_TST(o, BINARY)) {
2192 mmfile_t mf;
2193 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2194 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2195 abbrev = 40;
2197 strbuf_addf(&msg, "index %.*s..%.*s",
2198 abbrev, sha1_to_hex(one->sha1),
2199 abbrev, sha1_to_hex(two->sha1));
2200 if (one->mode == two->mode)
2201 strbuf_addf(&msg, " %06o", one->mode);
2202 strbuf_addch(&msg, '\n');
2205 if (msg.len)
2206 strbuf_setlen(&msg, msg.len - 1);
2207 xfrm_msg = msg.len ? msg.buf : NULL;
2209 if (!pgm &&
2210 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2211 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2212 /* a filepair that changes between file and symlink
2213 * needs to be split into deletion and creation.
2215 struct diff_filespec *null = alloc_filespec(two->path);
2216 run_diff_cmd(NULL, name, other, attr_path,
2217 one, null, xfrm_msg, o, 0);
2218 free(null);
2219 null = alloc_filespec(one->path);
2220 run_diff_cmd(NULL, name, other, attr_path,
2221 null, two, xfrm_msg, o, 0);
2222 free(null);
2224 else
2225 run_diff_cmd(pgm, name, other, attr_path,
2226 one, two, xfrm_msg, o, complete_rewrite);
2228 strbuf_release(&msg);
2231 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2232 struct diffstat_t *diffstat)
2234 const char *name;
2235 const char *other;
2236 int complete_rewrite = 0;
2238 if (DIFF_PAIR_UNMERGED(p)) {
2239 /* unmerged */
2240 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2241 return;
2244 name = p->one->path;
2245 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2247 if (o->prefix_length)
2248 strip_prefix(o->prefix_length, &name, &other);
2250 diff_fill_sha1_info(p->one);
2251 diff_fill_sha1_info(p->two);
2253 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2254 complete_rewrite = 1;
2255 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2258 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2260 const char *name;
2261 const char *other;
2262 const char *attr_path;
2264 if (DIFF_PAIR_UNMERGED(p)) {
2265 /* unmerged */
2266 return;
2269 name = p->one->path;
2270 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2271 attr_path = other ? other : name;
2273 if (o->prefix_length)
2274 strip_prefix(o->prefix_length, &name, &other);
2276 diff_fill_sha1_info(p->one);
2277 diff_fill_sha1_info(p->two);
2279 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2282 void diff_setup(struct diff_options *options)
2284 memset(options, 0, sizeof(*options));
2286 options->file = stdout;
2288 options->line_termination = '\n';
2289 options->break_opt = -1;
2290 options->rename_limit = -1;
2291 options->dirstat_percent = 3;
2292 options->context = 3;
2294 options->change = diff_change;
2295 options->add_remove = diff_addremove;
2296 if (diff_use_color_default > 0)
2297 DIFF_OPT_SET(options, COLOR_DIFF);
2298 else
2299 DIFF_OPT_CLR(options, COLOR_DIFF);
2300 options->detect_rename = diff_detect_rename_default;
2302 options->a_prefix = "a/";
2303 options->b_prefix = "b/";
2306 int diff_setup_done(struct diff_options *options)
2308 int count = 0;
2310 if (options->output_format & DIFF_FORMAT_NAME)
2311 count++;
2312 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2313 count++;
2314 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2315 count++;
2316 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2317 count++;
2318 if (count > 1)
2319 die("--name-only, --name-status, --check and -s are mutually exclusive");
2321 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2322 options->detect_rename = DIFF_DETECT_COPY;
2324 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2325 options->prefix = NULL;
2326 if (options->prefix)
2327 options->prefix_length = strlen(options->prefix);
2328 else
2329 options->prefix_length = 0;
2331 if (options->output_format & (DIFF_FORMAT_NAME |
2332 DIFF_FORMAT_NAME_STATUS |
2333 DIFF_FORMAT_CHECKDIFF |
2334 DIFF_FORMAT_NO_OUTPUT))
2335 options->output_format &= ~(DIFF_FORMAT_RAW |
2336 DIFF_FORMAT_NUMSTAT |
2337 DIFF_FORMAT_DIFFSTAT |
2338 DIFF_FORMAT_SHORTSTAT |
2339 DIFF_FORMAT_DIRSTAT |
2340 DIFF_FORMAT_SUMMARY |
2341 DIFF_FORMAT_PATCH);
2344 * These cases always need recursive; we do not drop caller-supplied
2345 * recursive bits for other formats here.
2347 if (options->output_format & (DIFF_FORMAT_PATCH |
2348 DIFF_FORMAT_NUMSTAT |
2349 DIFF_FORMAT_DIFFSTAT |
2350 DIFF_FORMAT_SHORTSTAT |
2351 DIFF_FORMAT_DIRSTAT |
2352 DIFF_FORMAT_SUMMARY |
2353 DIFF_FORMAT_CHECKDIFF))
2354 DIFF_OPT_SET(options, RECURSIVE);
2356 * Also pickaxe would not work very well if you do not say recursive
2358 if (options->pickaxe)
2359 DIFF_OPT_SET(options, RECURSIVE);
2361 if (options->detect_rename && options->rename_limit < 0)
2362 options->rename_limit = diff_rename_limit_default;
2363 if (options->setup & DIFF_SETUP_USE_CACHE) {
2364 if (!active_cache)
2365 /* read-cache does not die even when it fails
2366 * so it is safe for us to do this here. Also
2367 * it does not smudge active_cache or active_nr
2368 * when it fails, so we do not have to worry about
2369 * cleaning it up ourselves either.
2371 read_cache();
2373 if (options->abbrev <= 0 || 40 < options->abbrev)
2374 options->abbrev = 40; /* full */
2377 * It does not make sense to show the first hit we happened
2378 * to have found. It does not make sense not to return with
2379 * exit code in such a case either.
2381 if (DIFF_OPT_TST(options, QUIET)) {
2382 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2383 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2387 * If we postprocess in diffcore, we cannot simply return
2388 * upon the first hit. We need to run diff as usual.
2390 if (options->pickaxe || options->filter)
2391 DIFF_OPT_CLR(options, QUIET);
2393 return 0;
2396 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2398 char c, *eq;
2399 int len;
2401 if (*arg != '-')
2402 return 0;
2403 c = *++arg;
2404 if (!c)
2405 return 0;
2406 if (c == arg_short) {
2407 c = *++arg;
2408 if (!c)
2409 return 1;
2410 if (val && isdigit(c)) {
2411 char *end;
2412 int n = strtoul(arg, &end, 10);
2413 if (*end)
2414 return 0;
2415 *val = n;
2416 return 1;
2418 return 0;
2420 if (c != '-')
2421 return 0;
2422 arg++;
2423 eq = strchr(arg, '=');
2424 if (eq)
2425 len = eq - arg;
2426 else
2427 len = strlen(arg);
2428 if (!len || strncmp(arg, arg_long, len))
2429 return 0;
2430 if (eq) {
2431 int n;
2432 char *end;
2433 if (!isdigit(*++eq))
2434 return 0;
2435 n = strtoul(eq, &end, 10);
2436 if (*end)
2437 return 0;
2438 *val = n;
2440 return 1;
2443 static int diff_scoreopt_parse(const char *opt);
2445 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2447 const char *arg = av[0];
2449 /* Output format options */
2450 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2451 options->output_format |= DIFF_FORMAT_PATCH;
2452 else if (opt_arg(arg, 'U', "unified", &options->context))
2453 options->output_format |= DIFF_FORMAT_PATCH;
2454 else if (!strcmp(arg, "--raw"))
2455 options->output_format |= DIFF_FORMAT_RAW;
2456 else if (!strcmp(arg, "--patch-with-raw"))
2457 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2458 else if (!strcmp(arg, "--numstat"))
2459 options->output_format |= DIFF_FORMAT_NUMSTAT;
2460 else if (!strcmp(arg, "--shortstat"))
2461 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2462 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2463 options->output_format |= DIFF_FORMAT_DIRSTAT;
2464 else if (!strcmp(arg, "--cumulative"))
2465 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2466 else if (!strcmp(arg, "--check"))
2467 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2468 else if (!strcmp(arg, "--summary"))
2469 options->output_format |= DIFF_FORMAT_SUMMARY;
2470 else if (!strcmp(arg, "--patch-with-stat"))
2471 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2472 else if (!strcmp(arg, "--name-only"))
2473 options->output_format |= DIFF_FORMAT_NAME;
2474 else if (!strcmp(arg, "--name-status"))
2475 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2476 else if (!strcmp(arg, "-s"))
2477 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2478 else if (!prefixcmp(arg, "--stat")) {
2479 char *end;
2480 int width = options->stat_width;
2481 int name_width = options->stat_name_width;
2482 arg += 6;
2483 end = (char *)arg;
2485 switch (*arg) {
2486 case '-':
2487 if (!prefixcmp(arg, "-width="))
2488 width = strtoul(arg + 7, &end, 10);
2489 else if (!prefixcmp(arg, "-name-width="))
2490 name_width = strtoul(arg + 12, &end, 10);
2491 break;
2492 case '=':
2493 width = strtoul(arg+1, &end, 10);
2494 if (*end == ',')
2495 name_width = strtoul(end+1, &end, 10);
2498 /* Important! This checks all the error cases! */
2499 if (*end)
2500 return 0;
2501 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2502 options->stat_name_width = name_width;
2503 options->stat_width = width;
2506 /* renames options */
2507 else if (!prefixcmp(arg, "-B")) {
2508 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2509 return -1;
2511 else if (!prefixcmp(arg, "-M")) {
2512 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2513 return -1;
2514 options->detect_rename = DIFF_DETECT_RENAME;
2516 else if (!prefixcmp(arg, "-C")) {
2517 if (options->detect_rename == DIFF_DETECT_COPY)
2518 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2519 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2520 return -1;
2521 options->detect_rename = DIFF_DETECT_COPY;
2523 else if (!strcmp(arg, "--no-renames"))
2524 options->detect_rename = 0;
2525 else if (!strcmp(arg, "--relative"))
2526 DIFF_OPT_SET(options, RELATIVE_NAME);
2527 else if (!prefixcmp(arg, "--relative=")) {
2528 DIFF_OPT_SET(options, RELATIVE_NAME);
2529 options->prefix = arg + 11;
2532 /* xdiff options */
2533 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2534 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2535 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2536 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2537 else if (!strcmp(arg, "--ignore-space-at-eol"))
2538 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2540 /* flags options */
2541 else if (!strcmp(arg, "--binary")) {
2542 options->output_format |= DIFF_FORMAT_PATCH;
2543 DIFF_OPT_SET(options, BINARY);
2545 else if (!strcmp(arg, "--full-index"))
2546 DIFF_OPT_SET(options, FULL_INDEX);
2547 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2548 DIFF_OPT_SET(options, TEXT);
2549 else if (!strcmp(arg, "-R"))
2550 DIFF_OPT_SET(options, REVERSE_DIFF);
2551 else if (!strcmp(arg, "--find-copies-harder"))
2552 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2553 else if (!strcmp(arg, "--follow"))
2554 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2555 else if (!strcmp(arg, "--color"))
2556 DIFF_OPT_SET(options, COLOR_DIFF);
2557 else if (!strcmp(arg, "--no-color"))
2558 DIFF_OPT_CLR(options, COLOR_DIFF);
2559 else if (!strcmp(arg, "--color-words"))
2560 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2561 else if (!strcmp(arg, "--exit-code"))
2562 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2563 else if (!strcmp(arg, "--quiet"))
2564 DIFF_OPT_SET(options, QUIET);
2565 else if (!strcmp(arg, "--ext-diff"))
2566 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2567 else if (!strcmp(arg, "--no-ext-diff"))
2568 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2569 else if (!strcmp(arg, "--ignore-submodules"))
2570 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2572 /* misc options */
2573 else if (!strcmp(arg, "-z"))
2574 options->line_termination = 0;
2575 else if (!prefixcmp(arg, "-l"))
2576 options->rename_limit = strtoul(arg+2, NULL, 10);
2577 else if (!prefixcmp(arg, "-S"))
2578 options->pickaxe = arg + 2;
2579 else if (!strcmp(arg, "--pickaxe-all"))
2580 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2581 else if (!strcmp(arg, "--pickaxe-regex"))
2582 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2583 else if (!prefixcmp(arg, "-O"))
2584 options->orderfile = arg + 2;
2585 else if (!prefixcmp(arg, "--diff-filter="))
2586 options->filter = arg + 14;
2587 else if (!strcmp(arg, "--abbrev"))
2588 options->abbrev = DEFAULT_ABBREV;
2589 else if (!prefixcmp(arg, "--abbrev=")) {
2590 options->abbrev = strtoul(arg + 9, NULL, 10);
2591 if (options->abbrev < MINIMUM_ABBREV)
2592 options->abbrev = MINIMUM_ABBREV;
2593 else if (40 < options->abbrev)
2594 options->abbrev = 40;
2596 else if (!prefixcmp(arg, "--src-prefix="))
2597 options->a_prefix = arg + 13;
2598 else if (!prefixcmp(arg, "--dst-prefix="))
2599 options->b_prefix = arg + 13;
2600 else if (!strcmp(arg, "--no-prefix"))
2601 options->a_prefix = options->b_prefix = "";
2602 else if (!prefixcmp(arg, "--output=")) {
2603 options->file = fopen(arg + strlen("--output="), "w");
2604 options->close_file = 1;
2605 } else
2606 return 0;
2607 return 1;
2610 static int parse_num(const char **cp_p)
2612 unsigned long num, scale;
2613 int ch, dot;
2614 const char *cp = *cp_p;
2616 num = 0;
2617 scale = 1;
2618 dot = 0;
2619 for(;;) {
2620 ch = *cp;
2621 if ( !dot && ch == '.' ) {
2622 scale = 1;
2623 dot = 1;
2624 } else if ( ch == '%' ) {
2625 scale = dot ? scale*100 : 100;
2626 cp++; /* % is always at the end */
2627 break;
2628 } else if ( ch >= '0' && ch <= '9' ) {
2629 if ( scale < 100000 ) {
2630 scale *= 10;
2631 num = (num*10) + (ch-'0');
2633 } else {
2634 break;
2636 cp++;
2638 *cp_p = cp;
2640 /* user says num divided by scale and we say internally that
2641 * is MAX_SCORE * num / scale.
2643 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2646 static int diff_scoreopt_parse(const char *opt)
2648 int opt1, opt2, cmd;
2650 if (*opt++ != '-')
2651 return -1;
2652 cmd = *opt++;
2653 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2654 return -1; /* that is not a -M, -C nor -B option */
2656 opt1 = parse_num(&opt);
2657 if (cmd != 'B')
2658 opt2 = 0;
2659 else {
2660 if (*opt == 0)
2661 opt2 = 0;
2662 else if (*opt != '/')
2663 return -1; /* we expect -B80/99 or -B80 */
2664 else {
2665 opt++;
2666 opt2 = parse_num(&opt);
2669 if (*opt != 0)
2670 return -1;
2671 return opt1 | (opt2 << 16);
2674 struct diff_queue_struct diff_queued_diff;
2676 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2678 if (queue->alloc <= queue->nr) {
2679 queue->alloc = alloc_nr(queue->alloc);
2680 queue->queue = xrealloc(queue->queue,
2681 sizeof(dp) * queue->alloc);
2683 queue->queue[queue->nr++] = dp;
2686 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2687 struct diff_filespec *one,
2688 struct diff_filespec *two)
2690 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2691 dp->one = one;
2692 dp->two = two;
2693 if (queue)
2694 diff_q(queue, dp);
2695 return dp;
2698 void diff_free_filepair(struct diff_filepair *p)
2700 free_filespec(p->one);
2701 free_filespec(p->two);
2702 free(p);
2705 /* This is different from find_unique_abbrev() in that
2706 * it stuffs the result with dots for alignment.
2708 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2710 int abblen;
2711 const char *abbrev;
2712 if (len == 40)
2713 return sha1_to_hex(sha1);
2715 abbrev = find_unique_abbrev(sha1, len);
2716 abblen = strlen(abbrev);
2717 if (abblen < 37) {
2718 static char hex[41];
2719 if (len < abblen && abblen <= len + 2)
2720 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2721 else
2722 sprintf(hex, "%s...", abbrev);
2723 return hex;
2725 return sha1_to_hex(sha1);
2728 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2730 int line_termination = opt->line_termination;
2731 int inter_name_termination = line_termination ? '\t' : '\0';
2733 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2734 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2735 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2736 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2738 if (p->score) {
2739 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2740 inter_name_termination);
2741 } else {
2742 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2745 if (p->status == DIFF_STATUS_COPIED ||
2746 p->status == DIFF_STATUS_RENAMED) {
2747 const char *name_a, *name_b;
2748 name_a = p->one->path;
2749 name_b = p->two->path;
2750 strip_prefix(opt->prefix_length, &name_a, &name_b);
2751 write_name_quoted(name_a, opt->file, inter_name_termination);
2752 write_name_quoted(name_b, opt->file, line_termination);
2753 } else {
2754 const char *name_a, *name_b;
2755 name_a = p->one->mode ? p->one->path : p->two->path;
2756 name_b = NULL;
2757 strip_prefix(opt->prefix_length, &name_a, &name_b);
2758 write_name_quoted(name_a, opt->file, line_termination);
2762 int diff_unmodified_pair(struct diff_filepair *p)
2764 /* This function is written stricter than necessary to support
2765 * the currently implemented transformers, but the idea is to
2766 * let transformers to produce diff_filepairs any way they want,
2767 * and filter and clean them up here before producing the output.
2769 struct diff_filespec *one = p->one, *two = p->two;
2771 if (DIFF_PAIR_UNMERGED(p))
2772 return 0; /* unmerged is interesting */
2774 /* deletion, addition, mode or type change
2775 * and rename are all interesting.
2777 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2778 DIFF_PAIR_MODE_CHANGED(p) ||
2779 strcmp(one->path, two->path))
2780 return 0;
2782 /* both are valid and point at the same path. that is, we are
2783 * dealing with a change.
2785 if (one->sha1_valid && two->sha1_valid &&
2786 !hashcmp(one->sha1, two->sha1))
2787 return 1; /* no change */
2788 if (!one->sha1_valid && !two->sha1_valid)
2789 return 1; /* both look at the same file on the filesystem. */
2790 return 0;
2793 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2795 if (diff_unmodified_pair(p))
2796 return;
2798 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2799 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2800 return; /* no tree diffs in patch format */
2802 run_diff(p, o);
2805 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2806 struct diffstat_t *diffstat)
2808 if (diff_unmodified_pair(p))
2809 return;
2811 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2812 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2813 return; /* no tree diffs in patch format */
2815 run_diffstat(p, o, diffstat);
2818 static void diff_flush_checkdiff(struct diff_filepair *p,
2819 struct diff_options *o)
2821 if (diff_unmodified_pair(p))
2822 return;
2824 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2825 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2826 return; /* no tree diffs in patch format */
2828 run_checkdiff(p, o);
2831 int diff_queue_is_empty(void)
2833 struct diff_queue_struct *q = &diff_queued_diff;
2834 int i;
2835 for (i = 0; i < q->nr; i++)
2836 if (!diff_unmodified_pair(q->queue[i]))
2837 return 0;
2838 return 1;
2841 #if DIFF_DEBUG
2842 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2844 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2845 x, one ? one : "",
2846 s->path,
2847 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2848 s->mode,
2849 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2850 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2851 x, one ? one : "",
2852 s->size, s->xfrm_flags);
2855 void diff_debug_filepair(const struct diff_filepair *p, int i)
2857 diff_debug_filespec(p->one, i, "one");
2858 diff_debug_filespec(p->two, i, "two");
2859 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2860 p->score, p->status ? p->status : '?',
2861 p->one->rename_used, p->broken_pair);
2864 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2866 int i;
2867 if (msg)
2868 fprintf(stderr, "%s\n", msg);
2869 fprintf(stderr, "q->nr = %d\n", q->nr);
2870 for (i = 0; i < q->nr; i++) {
2871 struct diff_filepair *p = q->queue[i];
2872 diff_debug_filepair(p, i);
2875 #endif
2877 static void diff_resolve_rename_copy(void)
2879 int i;
2880 struct diff_filepair *p;
2881 struct diff_queue_struct *q = &diff_queued_diff;
2883 diff_debug_queue("resolve-rename-copy", q);
2885 for (i = 0; i < q->nr; i++) {
2886 p = q->queue[i];
2887 p->status = 0; /* undecided */
2888 if (DIFF_PAIR_UNMERGED(p))
2889 p->status = DIFF_STATUS_UNMERGED;
2890 else if (!DIFF_FILE_VALID(p->one))
2891 p->status = DIFF_STATUS_ADDED;
2892 else if (!DIFF_FILE_VALID(p->two))
2893 p->status = DIFF_STATUS_DELETED;
2894 else if (DIFF_PAIR_TYPE_CHANGED(p))
2895 p->status = DIFF_STATUS_TYPE_CHANGED;
2897 /* from this point on, we are dealing with a pair
2898 * whose both sides are valid and of the same type, i.e.
2899 * either in-place edit or rename/copy edit.
2901 else if (DIFF_PAIR_RENAME(p)) {
2903 * A rename might have re-connected a broken
2904 * pair up, causing the pathnames to be the
2905 * same again. If so, that's not a rename at
2906 * all, just a modification..
2908 * Otherwise, see if this source was used for
2909 * multiple renames, in which case we decrement
2910 * the count, and call it a copy.
2912 if (!strcmp(p->one->path, p->two->path))
2913 p->status = DIFF_STATUS_MODIFIED;
2914 else if (--p->one->rename_used > 0)
2915 p->status = DIFF_STATUS_COPIED;
2916 else
2917 p->status = DIFF_STATUS_RENAMED;
2919 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2920 p->one->mode != p->two->mode ||
2921 is_null_sha1(p->one->sha1))
2922 p->status = DIFF_STATUS_MODIFIED;
2923 else {
2924 /* This is a "no-change" entry and should not
2925 * happen anymore, but prepare for broken callers.
2927 error("feeding unmodified %s to diffcore",
2928 p->one->path);
2929 p->status = DIFF_STATUS_UNKNOWN;
2932 diff_debug_queue("resolve-rename-copy done", q);
2935 static int check_pair_status(struct diff_filepair *p)
2937 switch (p->status) {
2938 case DIFF_STATUS_UNKNOWN:
2939 return 0;
2940 case 0:
2941 die("internal error in diff-resolve-rename-copy");
2942 default:
2943 return 1;
2947 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2949 int fmt = opt->output_format;
2951 if (fmt & DIFF_FORMAT_CHECKDIFF)
2952 diff_flush_checkdiff(p, opt);
2953 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2954 diff_flush_raw(p, opt);
2955 else if (fmt & DIFF_FORMAT_NAME) {
2956 const char *name_a, *name_b;
2957 name_a = p->two->path;
2958 name_b = NULL;
2959 strip_prefix(opt->prefix_length, &name_a, &name_b);
2960 write_name_quoted(name_a, opt->file, opt->line_termination);
2964 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2966 if (fs->mode)
2967 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2968 else
2969 fprintf(file, " %s ", newdelete);
2970 write_name_quoted(fs->path, file, '\n');
2974 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2976 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2977 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2978 show_name ? ' ' : '\n');
2979 if (show_name) {
2980 write_name_quoted(p->two->path, file, '\n');
2985 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2987 char *names = pprint_rename(p->one->path, p->two->path);
2989 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2990 free(names);
2991 show_mode_change(file, p, 0);
2994 static void diff_summary(FILE *file, struct diff_filepair *p)
2996 switch(p->status) {
2997 case DIFF_STATUS_DELETED:
2998 show_file_mode_name(file, "delete", p->one);
2999 break;
3000 case DIFF_STATUS_ADDED:
3001 show_file_mode_name(file, "create", p->two);
3002 break;
3003 case DIFF_STATUS_COPIED:
3004 show_rename_copy(file, "copy", p);
3005 break;
3006 case DIFF_STATUS_RENAMED:
3007 show_rename_copy(file, "rename", p);
3008 break;
3009 default:
3010 if (p->score) {
3011 fputs(" rewrite ", file);
3012 write_name_quoted(p->two->path, file, ' ');
3013 fprintf(file, "(%d%%)\n", similarity_index(p));
3015 show_mode_change(file, p, !p->score);
3016 break;
3020 struct patch_id_t {
3021 struct xdiff_emit_state xm;
3022 SHA_CTX *ctx;
3023 int patchlen;
3026 static int remove_space(char *line, int len)
3028 int i;
3029 char *dst = line;
3030 unsigned char c;
3032 for (i = 0; i < len; i++)
3033 if (!isspace((c = line[i])))
3034 *dst++ = c;
3036 return dst - line;
3039 static void patch_id_consume(void *priv, char *line, unsigned long len)
3041 struct patch_id_t *data = priv;
3042 int new_len;
3044 /* Ignore line numbers when computing the SHA1 of the patch */
3045 if (!prefixcmp(line, "@@ -"))
3046 return;
3048 new_len = remove_space(line, len);
3050 SHA1_Update(data->ctx, line, new_len);
3051 data->patchlen += new_len;
3054 /* returns 0 upon success, and writes result into sha1 */
3055 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3057 struct diff_queue_struct *q = &diff_queued_diff;
3058 int i;
3059 SHA_CTX ctx;
3060 struct patch_id_t data;
3061 char buffer[PATH_MAX * 4 + 20];
3063 SHA1_Init(&ctx);
3064 memset(&data, 0, sizeof(struct patch_id_t));
3065 data.ctx = &ctx;
3066 data.xm.consume = patch_id_consume;
3068 for (i = 0; i < q->nr; i++) {
3069 xpparam_t xpp;
3070 xdemitconf_t xecfg;
3071 xdemitcb_t ecb;
3072 mmfile_t mf1, mf2;
3073 struct diff_filepair *p = q->queue[i];
3074 int len1, len2;
3076 memset(&xecfg, 0, sizeof(xecfg));
3077 if (p->status == 0)
3078 return error("internal diff status error");
3079 if (p->status == DIFF_STATUS_UNKNOWN)
3080 continue;
3081 if (diff_unmodified_pair(p))
3082 continue;
3083 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3084 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3085 continue;
3086 if (DIFF_PAIR_UNMERGED(p))
3087 continue;
3089 diff_fill_sha1_info(p->one);
3090 diff_fill_sha1_info(p->two);
3091 if (fill_mmfile(&mf1, p->one) < 0 ||
3092 fill_mmfile(&mf2, p->two) < 0)
3093 return error("unable to read files to diff");
3095 len1 = remove_space(p->one->path, strlen(p->one->path));
3096 len2 = remove_space(p->two->path, strlen(p->two->path));
3097 if (p->one->mode == 0)
3098 len1 = snprintf(buffer, sizeof(buffer),
3099 "diff--gita/%.*sb/%.*s"
3100 "newfilemode%06o"
3101 "---/dev/null"
3102 "+++b/%.*s",
3103 len1, p->one->path,
3104 len2, p->two->path,
3105 p->two->mode,
3106 len2, p->two->path);
3107 else if (p->two->mode == 0)
3108 len1 = snprintf(buffer, sizeof(buffer),
3109 "diff--gita/%.*sb/%.*s"
3110 "deletedfilemode%06o"
3111 "---a/%.*s"
3112 "+++/dev/null",
3113 len1, p->one->path,
3114 len2, p->two->path,
3115 p->one->mode,
3116 len1, p->one->path);
3117 else
3118 len1 = snprintf(buffer, sizeof(buffer),
3119 "diff--gita/%.*sb/%.*s"
3120 "---a/%.*s"
3121 "+++b/%.*s",
3122 len1, p->one->path,
3123 len2, p->two->path,
3124 len1, p->one->path,
3125 len2, p->two->path);
3126 SHA1_Update(&ctx, buffer, len1);
3128 xpp.flags = XDF_NEED_MINIMAL;
3129 xecfg.ctxlen = 3;
3130 xecfg.flags = XDL_EMIT_FUNCNAMES;
3131 ecb.outf = xdiff_outf;
3132 ecb.priv = &data;
3133 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3136 SHA1_Final(sha1, &ctx);
3137 return 0;
3140 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3142 struct diff_queue_struct *q = &diff_queued_diff;
3143 int i;
3144 int result = diff_get_patch_id(options, sha1);
3146 for (i = 0; i < q->nr; i++)
3147 diff_free_filepair(q->queue[i]);
3149 free(q->queue);
3150 q->queue = NULL;
3151 q->nr = q->alloc = 0;
3153 return result;
3156 static int is_summary_empty(const struct diff_queue_struct *q)
3158 int i;
3160 for (i = 0; i < q->nr; i++) {
3161 const struct diff_filepair *p = q->queue[i];
3163 switch (p->status) {
3164 case DIFF_STATUS_DELETED:
3165 case DIFF_STATUS_ADDED:
3166 case DIFF_STATUS_COPIED:
3167 case DIFF_STATUS_RENAMED:
3168 return 0;
3169 default:
3170 if (p->score)
3171 return 0;
3172 if (p->one->mode && p->two->mode &&
3173 p->one->mode != p->two->mode)
3174 return 0;
3175 break;
3178 return 1;
3181 void diff_flush(struct diff_options *options)
3183 struct diff_queue_struct *q = &diff_queued_diff;
3184 int i, output_format = options->output_format;
3185 int separator = 0;
3188 * Order: raw, stat, summary, patch
3189 * or: name/name-status/checkdiff (other bits clear)
3191 if (!q->nr)
3192 goto free_queue;
3194 if (output_format & (DIFF_FORMAT_RAW |
3195 DIFF_FORMAT_NAME |
3196 DIFF_FORMAT_NAME_STATUS |
3197 DIFF_FORMAT_CHECKDIFF)) {
3198 for (i = 0; i < q->nr; i++) {
3199 struct diff_filepair *p = q->queue[i];
3200 if (check_pair_status(p))
3201 flush_one_pair(p, options);
3203 separator++;
3206 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3207 struct diffstat_t diffstat;
3209 memset(&diffstat, 0, sizeof(struct diffstat_t));
3210 diffstat.xm.consume = diffstat_consume;
3211 for (i = 0; i < q->nr; i++) {
3212 struct diff_filepair *p = q->queue[i];
3213 if (check_pair_status(p))
3214 diff_flush_stat(p, options, &diffstat);
3216 if (output_format & DIFF_FORMAT_NUMSTAT)
3217 show_numstat(&diffstat, options);
3218 if (output_format & DIFF_FORMAT_DIFFSTAT)
3219 show_stats(&diffstat, options);
3220 if (output_format & DIFF_FORMAT_SHORTSTAT)
3221 show_shortstats(&diffstat, options);
3222 free_diffstat_info(&diffstat);
3223 separator++;
3225 if (output_format & DIFF_FORMAT_DIRSTAT)
3226 show_dirstat(options);
3228 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3229 for (i = 0; i < q->nr; i++)
3230 diff_summary(options->file, q->queue[i]);
3231 separator++;
3234 if (output_format & DIFF_FORMAT_PATCH) {
3235 if (separator) {
3236 putc(options->line_termination, options->file);
3237 if (options->stat_sep) {
3238 /* attach patch instead of inline */
3239 fputs(options->stat_sep, options->file);
3243 for (i = 0; i < q->nr; i++) {
3244 struct diff_filepair *p = q->queue[i];
3245 if (check_pair_status(p))
3246 diff_flush_patch(p, options);
3250 if (output_format & DIFF_FORMAT_CALLBACK)
3251 options->format_callback(q, options, options->format_callback_data);
3253 for (i = 0; i < q->nr; i++)
3254 diff_free_filepair(q->queue[i]);
3255 free_queue:
3256 free(q->queue);
3257 q->queue = NULL;
3258 q->nr = q->alloc = 0;
3259 if (options->close_file)
3260 fclose(options->file);
3263 static void diffcore_apply_filter(const char *filter)
3265 int i;
3266 struct diff_queue_struct *q = &diff_queued_diff;
3267 struct diff_queue_struct outq;
3268 outq.queue = NULL;
3269 outq.nr = outq.alloc = 0;
3271 if (!filter)
3272 return;
3274 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3275 int found;
3276 for (i = found = 0; !found && i < q->nr; i++) {
3277 struct diff_filepair *p = q->queue[i];
3278 if (((p->status == DIFF_STATUS_MODIFIED) &&
3279 ((p->score &&
3280 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3281 (!p->score &&
3282 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3283 ((p->status != DIFF_STATUS_MODIFIED) &&
3284 strchr(filter, p->status)))
3285 found++;
3287 if (found)
3288 return;
3290 /* otherwise we will clear the whole queue
3291 * by copying the empty outq at the end of this
3292 * function, but first clear the current entries
3293 * in the queue.
3295 for (i = 0; i < q->nr; i++)
3296 diff_free_filepair(q->queue[i]);
3298 else {
3299 /* Only the matching ones */
3300 for (i = 0; i < q->nr; i++) {
3301 struct diff_filepair *p = q->queue[i];
3303 if (((p->status == DIFF_STATUS_MODIFIED) &&
3304 ((p->score &&
3305 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3306 (!p->score &&
3307 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3308 ((p->status != DIFF_STATUS_MODIFIED) &&
3309 strchr(filter, p->status)))
3310 diff_q(&outq, p);
3311 else
3312 diff_free_filepair(p);
3315 free(q->queue);
3316 *q = outq;
3319 /* Check whether two filespecs with the same mode and size are identical */
3320 static int diff_filespec_is_identical(struct diff_filespec *one,
3321 struct diff_filespec *two)
3323 if (S_ISGITLINK(one->mode))
3324 return 0;
3325 if (diff_populate_filespec(one, 0))
3326 return 0;
3327 if (diff_populate_filespec(two, 0))
3328 return 0;
3329 return !memcmp(one->data, two->data, one->size);
3332 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3334 int i;
3335 struct diff_queue_struct *q = &diff_queued_diff;
3336 struct diff_queue_struct outq;
3337 outq.queue = NULL;
3338 outq.nr = outq.alloc = 0;
3340 for (i = 0; i < q->nr; i++) {
3341 struct diff_filepair *p = q->queue[i];
3344 * 1. Entries that come from stat info dirtyness
3345 * always have both sides (iow, not create/delete),
3346 * one side of the object name is unknown, with
3347 * the same mode and size. Keep the ones that
3348 * do not match these criteria. They have real
3349 * differences.
3351 * 2. At this point, the file is known to be modified,
3352 * with the same mode and size, and the object
3353 * name of one side is unknown. Need to inspect
3354 * the identical contents.
3356 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3357 !DIFF_FILE_VALID(p->two) ||
3358 (p->one->sha1_valid && p->two->sha1_valid) ||
3359 (p->one->mode != p->two->mode) ||
3360 diff_populate_filespec(p->one, 1) ||
3361 diff_populate_filespec(p->two, 1) ||
3362 (p->one->size != p->two->size) ||
3363 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3364 diff_q(&outq, p);
3365 else {
3367 * The caller can subtract 1 from skip_stat_unmatch
3368 * to determine how many paths were dirty only
3369 * due to stat info mismatch.
3371 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3372 diffopt->skip_stat_unmatch++;
3373 diff_free_filepair(p);
3376 free(q->queue);
3377 *q = outq;
3380 void diffcore_std(struct diff_options *options)
3382 if (DIFF_OPT_TST(options, QUIET))
3383 return;
3385 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3386 diffcore_skip_stat_unmatch(options);
3387 if (options->break_opt != -1)
3388 diffcore_break(options->break_opt);
3389 if (options->detect_rename)
3390 diffcore_rename(options);
3391 if (options->break_opt != -1)
3392 diffcore_merge_broken();
3393 if (options->pickaxe)
3394 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3395 if (options->orderfile)
3396 diffcore_order(options->orderfile);
3397 diff_resolve_rename_copy();
3398 diffcore_apply_filter(options->filter);
3400 if (diff_queued_diff.nr)
3401 DIFF_OPT_SET(options, HAS_CHANGES);
3402 else
3403 DIFF_OPT_CLR(options, HAS_CHANGES);
3406 int diff_result_code(struct diff_options *opt, int status)
3408 int result = 0;
3409 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3410 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3411 return status;
3412 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3413 DIFF_OPT_TST(opt, HAS_CHANGES))
3414 result |= 01;
3415 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3416 DIFF_OPT_TST(opt, CHECK_FAILED))
3417 result |= 02;
3418 return result;
3421 void diff_addremove(struct diff_options *options,
3422 int addremove, unsigned mode,
3423 const unsigned char *sha1,
3424 const char *concatpath)
3426 struct diff_filespec *one, *two;
3428 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3429 return;
3431 /* This may look odd, but it is a preparation for
3432 * feeding "there are unchanged files which should
3433 * not produce diffs, but when you are doing copy
3434 * detection you would need them, so here they are"
3435 * entries to the diff-core. They will be prefixed
3436 * with something like '=' or '*' (I haven't decided
3437 * which but should not make any difference).
3438 * Feeding the same new and old to diff_change()
3439 * also has the same effect.
3440 * Before the final output happens, they are pruned after
3441 * merged into rename/copy pairs as appropriate.
3443 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3444 addremove = (addremove == '+' ? '-' :
3445 addremove == '-' ? '+' : addremove);
3447 if (options->prefix &&
3448 strncmp(concatpath, options->prefix, options->prefix_length))
3449 return;
3451 one = alloc_filespec(concatpath);
3452 two = alloc_filespec(concatpath);
3454 if (addremove != '+')
3455 fill_filespec(one, sha1, mode);
3456 if (addremove != '-')
3457 fill_filespec(two, sha1, mode);
3459 diff_queue(&diff_queued_diff, one, two);
3460 DIFF_OPT_SET(options, HAS_CHANGES);
3463 void diff_change(struct diff_options *options,
3464 unsigned old_mode, unsigned new_mode,
3465 const unsigned char *old_sha1,
3466 const unsigned char *new_sha1,
3467 const char *concatpath)
3469 struct diff_filespec *one, *two;
3471 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3472 && S_ISGITLINK(new_mode))
3473 return;
3475 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3476 unsigned tmp;
3477 const unsigned char *tmp_c;
3478 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3479 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3482 if (options->prefix &&
3483 strncmp(concatpath, options->prefix, options->prefix_length))
3484 return;
3486 one = alloc_filespec(concatpath);
3487 two = alloc_filespec(concatpath);
3488 fill_filespec(one, old_sha1, old_mode);
3489 fill_filespec(two, new_sha1, new_mode);
3491 diff_queue(&diff_queued_diff, one, two);
3492 DIFF_OPT_SET(options, HAS_CHANGES);
3495 void diff_unmerge(struct diff_options *options,
3496 const char *path,
3497 unsigned mode, const unsigned char *sha1)
3499 struct diff_filespec *one, *two;
3501 if (options->prefix &&
3502 strncmp(path, options->prefix, options->prefix_length))
3503 return;
3505 one = alloc_filespec(path);
3506 two = alloc_filespec(path);
3507 fill_filespec(one, sha1, mode);
3508 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;