diff --color: color blank-at-eof
[git/jrn.git] / diff.c
blob2b285b8ce90faed3218d5af1524283fc65c9540d
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 struct funcname_pattern_entry {
98 char *name;
99 char *pattern;
100 int cflags;
102 static struct funcname_pattern_list {
103 struct funcname_pattern_list *next;
104 struct funcname_pattern_entry e;
105 } *funcname_pattern_list;
107 static int parse_funcname_pattern(const char *var, const char *ep, const char *value, int cflags)
109 const char *name;
110 int namelen;
111 struct funcname_pattern_list *pp;
113 name = var + 5; /* "diff." */
114 namelen = ep - name;
116 for (pp = funcname_pattern_list; pp; pp = pp->next)
117 if (!strncmp(pp->e.name, name, namelen) && !pp->e.name[namelen])
118 break;
119 if (!pp) {
120 pp = xcalloc(1, sizeof(*pp));
121 pp->e.name = xmemdupz(name, namelen);
122 pp->next = funcname_pattern_list;
123 funcname_pattern_list = pp;
125 free(pp->e.pattern);
126 pp->e.pattern = xstrdup(value);
127 pp->e.cflags = cflags;
128 return 0;
132 * These are to give UI layer defaults.
133 * The core-level commands such as git-diff-files should
134 * never be affected by the setting of diff.renames
135 * the user happens to have in the configuration file.
137 int git_diff_ui_config(const char *var, const char *value, void *cb)
139 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
140 diff_use_color_default = git_config_colorbool(var, value, -1);
141 return 0;
143 if (!strcmp(var, "diff.renames")) {
144 if (!value)
145 diff_detect_rename_default = DIFF_DETECT_RENAME;
146 else if (!strcasecmp(value, "copies") ||
147 !strcasecmp(value, "copy"))
148 diff_detect_rename_default = DIFF_DETECT_COPY;
149 else if (git_config_bool(var,value))
150 diff_detect_rename_default = DIFF_DETECT_RENAME;
151 return 0;
153 if (!strcmp(var, "diff.autorefreshindex")) {
154 diff_auto_refresh_index = git_config_bool(var, value);
155 return 0;
157 if (!strcmp(var, "diff.external"))
158 return git_config_string(&external_diff_cmd_cfg, var, value);
159 if (!prefixcmp(var, "diff.")) {
160 const char *ep = strrchr(var, '.');
162 if (ep != var + 4 && !strcmp(ep, ".command"))
163 return parse_lldiff_command(var, ep, value);
166 return git_diff_basic_config(var, value, cb);
169 int git_diff_basic_config(const char *var, const char *value, void *cb)
171 if (!strcmp(var, "diff.renamelimit")) {
172 diff_rename_limit_default = git_config_int(var, value);
173 return 0;
176 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
177 int slot = parse_diff_color_slot(var, 11);
178 if (!value)
179 return config_error_nonbool(var);
180 color_parse(value, var, diff_colors[slot]);
181 return 0;
184 if (!prefixcmp(var, "diff.")) {
185 const char *ep = strrchr(var, '.');
186 if (ep != var + 4) {
187 if (!strcmp(ep, ".funcname")) {
188 if (!value)
189 return config_error_nonbool(var);
190 return parse_funcname_pattern(var, ep, value,
192 } else if (!strcmp(ep, ".xfuncname")) {
193 if (!value)
194 return config_error_nonbool(var);
195 return parse_funcname_pattern(var, ep, value,
196 REG_EXTENDED);
201 return git_color_default_config(var, value, cb);
204 static char *quote_two(const char *one, const char *two)
206 int need_one = quote_c_style(one, NULL, NULL, 1);
207 int need_two = quote_c_style(two, NULL, NULL, 1);
208 struct strbuf res;
210 strbuf_init(&res, 0);
211 if (need_one + need_two) {
212 strbuf_addch(&res, '"');
213 quote_c_style(one, &res, NULL, 1);
214 quote_c_style(two, &res, NULL, 1);
215 strbuf_addch(&res, '"');
216 } else {
217 strbuf_addstr(&res, one);
218 strbuf_addstr(&res, two);
220 return strbuf_detach(&res, NULL);
223 static const char *external_diff(void)
225 static const char *external_diff_cmd = NULL;
226 static int done_preparing = 0;
228 if (done_preparing)
229 return external_diff_cmd;
230 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
231 if (!external_diff_cmd)
232 external_diff_cmd = external_diff_cmd_cfg;
233 done_preparing = 1;
234 return external_diff_cmd;
237 static struct diff_tempfile {
238 const char *name; /* filename external diff should read from */
239 char hex[41];
240 char mode[10];
241 char tmp_path[PATH_MAX];
242 } diff_temp[2];
244 static int count_lines(const char *data, int size)
246 int count, ch, completely_empty = 1, nl_just_seen = 0;
247 count = 0;
248 while (0 < size--) {
249 ch = *data++;
250 if (ch == '\n') {
251 count++;
252 nl_just_seen = 1;
253 completely_empty = 0;
255 else {
256 nl_just_seen = 0;
257 completely_empty = 0;
260 if (completely_empty)
261 return 0;
262 if (!nl_just_seen)
263 count++; /* no trailing newline */
264 return count;
267 static void print_line_count(FILE *file, int count)
269 switch (count) {
270 case 0:
271 fprintf(file, "0,0");
272 break;
273 case 1:
274 fprintf(file, "1");
275 break;
276 default:
277 fprintf(file, "1,%d", count);
278 break;
282 static void copy_file_with_prefix(FILE *file,
283 int prefix, const char *data, int size,
284 const char *set, const char *reset)
286 int ch, nl_just_seen = 1;
287 while (0 < size--) {
288 ch = *data++;
289 if (nl_just_seen) {
290 fputs(set, file);
291 putc(prefix, file);
293 if (ch == '\n') {
294 nl_just_seen = 1;
295 fputs(reset, file);
296 } else
297 nl_just_seen = 0;
298 putc(ch, file);
300 if (!nl_just_seen)
301 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
304 static void emit_rewrite_diff(const char *name_a,
305 const char *name_b,
306 struct diff_filespec *one,
307 struct diff_filespec *two,
308 struct diff_options *o)
310 int lc_a, lc_b;
311 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
312 const char *name_a_tab, *name_b_tab;
313 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
314 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
315 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
316 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
317 const char *reset = diff_get_color(color_diff, DIFF_RESET);
318 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
320 name_a += (*name_a == '/');
321 name_b += (*name_b == '/');
322 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
323 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
325 strbuf_reset(&a_name);
326 strbuf_reset(&b_name);
327 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
328 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
330 diff_populate_filespec(one, 0);
331 diff_populate_filespec(two, 0);
332 lc_a = count_lines(one->data, one->size);
333 lc_b = count_lines(two->data, two->size);
334 fprintf(o->file,
335 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
336 metainfo, a_name.buf, name_a_tab, reset,
337 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
338 print_line_count(o->file, lc_a);
339 fprintf(o->file, " +");
340 print_line_count(o->file, lc_b);
341 fprintf(o->file, " @@%s\n", reset);
342 if (lc_a)
343 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
344 if (lc_b)
345 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
348 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
350 if (!DIFF_FILE_VALID(one)) {
351 mf->ptr = (char *)""; /* does not matter */
352 mf->size = 0;
353 return 0;
355 else if (diff_populate_filespec(one, 0))
356 return -1;
357 mf->ptr = one->data;
358 mf->size = one->size;
359 return 0;
362 struct diff_words_buffer {
363 mmfile_t text;
364 long alloc;
365 long current; /* output pointer */
366 int suppressed_newline;
369 static void diff_words_append(char *line, unsigned long len,
370 struct diff_words_buffer *buffer)
372 if (buffer->text.size + len > buffer->alloc) {
373 buffer->alloc = (buffer->text.size + len) * 3 / 2;
374 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
376 line++;
377 len--;
378 memcpy(buffer->text.ptr + buffer->text.size, line, len);
379 buffer->text.size += len;
382 struct diff_words_data {
383 struct xdiff_emit_state xm;
384 struct diff_words_buffer minus, plus;
385 FILE *file;
388 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
389 int suppress_newline)
391 const char *ptr;
392 int eol = 0;
394 if (len == 0)
395 return;
397 ptr = buffer->text.ptr + buffer->current;
398 buffer->current += len;
400 if (ptr[len - 1] == '\n') {
401 eol = 1;
402 len--;
405 fputs(diff_get_color(1, color), file);
406 fwrite(ptr, len, 1, file);
407 fputs(diff_get_color(1, DIFF_RESET), file);
409 if (eol) {
410 if (suppress_newline)
411 buffer->suppressed_newline = 1;
412 else
413 putc('\n', file);
417 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
419 struct diff_words_data *diff_words = priv;
421 if (diff_words->minus.suppressed_newline) {
422 if (line[0] != '+')
423 putc('\n', diff_words->file);
424 diff_words->minus.suppressed_newline = 0;
427 len--;
428 switch (line[0]) {
429 case '-':
430 print_word(diff_words->file,
431 &diff_words->minus, len, DIFF_FILE_OLD, 1);
432 break;
433 case '+':
434 print_word(diff_words->file,
435 &diff_words->plus, len, DIFF_FILE_NEW, 0);
436 break;
437 case ' ':
438 print_word(diff_words->file,
439 &diff_words->plus, len, DIFF_PLAIN, 0);
440 diff_words->minus.current += len;
441 break;
445 /* this executes the word diff on the accumulated buffers */
446 static void diff_words_show(struct diff_words_data *diff_words)
448 xpparam_t xpp;
449 xdemitconf_t xecfg;
450 xdemitcb_t ecb;
451 mmfile_t minus, plus;
452 int i;
454 memset(&xecfg, 0, sizeof(xecfg));
455 minus.size = diff_words->minus.text.size;
456 minus.ptr = xmalloc(minus.size);
457 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
458 for (i = 0; i < minus.size; i++)
459 if (isspace(minus.ptr[i]))
460 minus.ptr[i] = '\n';
461 diff_words->minus.current = 0;
463 plus.size = diff_words->plus.text.size;
464 plus.ptr = xmalloc(plus.size);
465 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
466 for (i = 0; i < plus.size; i++)
467 if (isspace(plus.ptr[i]))
468 plus.ptr[i] = '\n';
469 diff_words->plus.current = 0;
471 xpp.flags = XDF_NEED_MINIMAL;
472 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
473 ecb.outf = xdiff_outf;
474 ecb.priv = diff_words;
475 diff_words->xm.consume = fn_out_diff_words_aux;
476 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
478 free(minus.ptr);
479 free(plus.ptr);
480 diff_words->minus.text.size = diff_words->plus.text.size = 0;
482 if (diff_words->minus.suppressed_newline) {
483 putc('\n', diff_words->file);
484 diff_words->minus.suppressed_newline = 0;
488 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
490 struct emit_callback {
491 struct xdiff_emit_state xm;
492 int color_diff;
493 unsigned ws_rule;
494 int blank_at_eof;
495 int lno_in_preimage;
496 sane_truncate_fn truncate;
497 const char **label_path;
498 struct diff_words_data *diff_words;
499 int *found_changesp;
500 FILE *file;
503 static void free_diff_words_data(struct emit_callback *ecbdata)
505 if (ecbdata->diff_words) {
506 /* flush buffers */
507 if (ecbdata->diff_words->minus.text.size ||
508 ecbdata->diff_words->plus.text.size)
509 diff_words_show(ecbdata->diff_words);
511 free (ecbdata->diff_words->minus.text.ptr);
512 free (ecbdata->diff_words->plus.text.ptr);
513 free(ecbdata->diff_words);
514 ecbdata->diff_words = NULL;
518 const char *diff_get_color(int diff_use_color, enum color_diff ix)
520 if (diff_use_color)
521 return diff_colors[ix];
522 return "";
525 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
527 int has_trailing_newline, has_trailing_carriage_return;
529 has_trailing_newline = (len > 0 && line[len-1] == '\n');
530 if (has_trailing_newline)
531 len--;
532 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
533 if (has_trailing_carriage_return)
534 len--;
536 fputs(set, file);
537 fwrite(line, len, 1, file);
538 fputs(reset, file);
539 if (has_trailing_carriage_return)
540 fputc('\r', file);
541 if (has_trailing_newline)
542 fputc('\n', file);
545 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
547 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
548 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
550 if (!*ws)
551 emit_line(ecbdata->file, set, reset, line, len);
552 else if ((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
553 ecbdata->blank_at_eof &&
554 (ecbdata->blank_at_eof <= ecbdata->lno_in_preimage) &&
555 ws_blank_line(line + 1, len - 1, ecbdata->ws_rule))
556 /* Blank line at EOF */
557 emit_line(ecbdata->file, ws, reset, line, len);
558 else {
559 /* Emit just the prefix, then the rest. */
560 emit_line(ecbdata->file, set, reset, line, 1);
561 ws_check_emit(line + 1, len - 1, ecbdata->ws_rule,
562 ecbdata->file, set, reset, ws);
566 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
568 const char *cp;
569 unsigned long allot;
570 size_t l = len;
572 if (ecb->truncate)
573 return ecb->truncate(line, len);
574 cp = line;
575 allot = l;
576 while (0 < l) {
577 (void) utf8_width(&cp, &l);
578 if (!cp)
579 break; /* truncated in the middle? */
581 return allot - l;
584 static int find_preimage_lno(const char *line)
586 char *p = strchr(line, '-');
587 if (!p)
588 return 0; /* should not happen */
589 return strtol(p+1, NULL, 10);
592 static void fn_out_consume(void *priv, char *line, unsigned long len)
594 struct emit_callback *ecbdata = priv;
595 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
596 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
597 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
599 *(ecbdata->found_changesp) = 1;
601 if (ecbdata->label_path[0]) {
602 const char *name_a_tab, *name_b_tab;
604 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
605 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
607 fprintf(ecbdata->file, "%s--- %s%s%s\n",
608 meta, ecbdata->label_path[0], reset, name_a_tab);
609 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
610 meta, ecbdata->label_path[1], reset, name_b_tab);
611 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
614 if (line[0] == '@') {
615 len = sane_truncate_line(ecbdata, line, len);
616 ecbdata->lno_in_preimage = find_preimage_lno(line);
617 emit_line(ecbdata->file,
618 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
619 reset, line, len);
620 if (line[len-1] != '\n')
621 putc('\n', ecbdata->file);
622 return;
625 if (len < 1) {
626 emit_line(ecbdata->file, reset, reset, line, len);
627 return;
630 if (ecbdata->diff_words) {
631 if (line[0] == '-') {
632 diff_words_append(line, len,
633 &ecbdata->diff_words->minus);
634 return;
635 } else if (line[0] == '+') {
636 diff_words_append(line, len,
637 &ecbdata->diff_words->plus);
638 return;
640 if (ecbdata->diff_words->minus.text.size ||
641 ecbdata->diff_words->plus.text.size)
642 diff_words_show(ecbdata->diff_words);
643 line++;
644 len--;
645 emit_line(ecbdata->file, plain, reset, line, len);
646 return;
649 if (line[0] != '+') {
650 const char *color =
651 diff_get_color(ecbdata->color_diff,
652 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
653 ecbdata->lno_in_preimage++;
654 emit_line(ecbdata->file, color, reset, line, len);
655 return;
657 emit_add_line(reset, ecbdata, line, len);
660 static char *pprint_rename(const char *a, const char *b)
662 const char *old = a;
663 const char *new = b;
664 struct strbuf name;
665 int pfx_length, sfx_length;
666 int len_a = strlen(a);
667 int len_b = strlen(b);
668 int a_midlen, b_midlen;
669 int qlen_a = quote_c_style(a, NULL, NULL, 0);
670 int qlen_b = quote_c_style(b, NULL, NULL, 0);
672 strbuf_init(&name, 0);
673 if (qlen_a || qlen_b) {
674 quote_c_style(a, &name, NULL, 0);
675 strbuf_addstr(&name, " => ");
676 quote_c_style(b, &name, NULL, 0);
677 return strbuf_detach(&name, NULL);
680 /* Find common prefix */
681 pfx_length = 0;
682 while (*old && *new && *old == *new) {
683 if (*old == '/')
684 pfx_length = old - a + 1;
685 old++;
686 new++;
689 /* Find common suffix */
690 old = a + len_a;
691 new = b + len_b;
692 sfx_length = 0;
693 while (a <= old && b <= new && *old == *new) {
694 if (*old == '/')
695 sfx_length = len_a - (old - a);
696 old--;
697 new--;
701 * pfx{mid-a => mid-b}sfx
702 * {pfx-a => pfx-b}sfx
703 * pfx{sfx-a => sfx-b}
704 * name-a => name-b
706 a_midlen = len_a - pfx_length - sfx_length;
707 b_midlen = len_b - pfx_length - sfx_length;
708 if (a_midlen < 0)
709 a_midlen = 0;
710 if (b_midlen < 0)
711 b_midlen = 0;
713 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
714 if (pfx_length + sfx_length) {
715 strbuf_add(&name, a, pfx_length);
716 strbuf_addch(&name, '{');
718 strbuf_add(&name, a + pfx_length, a_midlen);
719 strbuf_addstr(&name, " => ");
720 strbuf_add(&name, b + pfx_length, b_midlen);
721 if (pfx_length + sfx_length) {
722 strbuf_addch(&name, '}');
723 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
725 return strbuf_detach(&name, NULL);
728 struct diffstat_t {
729 struct xdiff_emit_state xm;
731 int nr;
732 int alloc;
733 struct diffstat_file {
734 char *from_name;
735 char *name;
736 char *print_name;
737 unsigned is_unmerged:1;
738 unsigned is_binary:1;
739 unsigned is_renamed:1;
740 unsigned int added, deleted;
741 } **files;
744 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
745 const char *name_a,
746 const char *name_b)
748 struct diffstat_file *x;
749 x = xcalloc(sizeof (*x), 1);
750 if (diffstat->nr == diffstat->alloc) {
751 diffstat->alloc = alloc_nr(diffstat->alloc);
752 diffstat->files = xrealloc(diffstat->files,
753 diffstat->alloc * sizeof(x));
755 diffstat->files[diffstat->nr++] = x;
756 if (name_b) {
757 x->from_name = xstrdup(name_a);
758 x->name = xstrdup(name_b);
759 x->is_renamed = 1;
761 else {
762 x->from_name = NULL;
763 x->name = xstrdup(name_a);
765 return x;
768 static void diffstat_consume(void *priv, char *line, unsigned long len)
770 struct diffstat_t *diffstat = priv;
771 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
773 if (line[0] == '+')
774 x->added++;
775 else if (line[0] == '-')
776 x->deleted++;
779 const char mime_boundary_leader[] = "------------";
781 static int scale_linear(int it, int width, int max_change)
784 * make sure that at least one '-' is printed if there were deletions,
785 * and likewise for '+'.
787 if (max_change < 2)
788 return it;
789 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
792 static void show_name(FILE *file,
793 const char *prefix, const char *name, int len,
794 const char *reset, const char *set)
796 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
799 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
801 if (cnt <= 0)
802 return;
803 fprintf(file, "%s", set);
804 while (cnt--)
805 putc(ch, file);
806 fprintf(file, "%s", reset);
809 static void fill_print_name(struct diffstat_file *file)
811 char *pname;
813 if (file->print_name)
814 return;
816 if (!file->is_renamed) {
817 struct strbuf buf;
818 strbuf_init(&buf, 0);
819 if (quote_c_style(file->name, &buf, NULL, 0)) {
820 pname = strbuf_detach(&buf, NULL);
821 } else {
822 pname = file->name;
823 strbuf_release(&buf);
825 } else {
826 pname = pprint_rename(file->from_name, file->name);
828 file->print_name = pname;
831 static void show_stats(struct diffstat_t* data, struct diff_options *options)
833 int i, len, add, del, total, adds = 0, dels = 0;
834 int max_change = 0, max_len = 0;
835 int total_files = data->nr;
836 int width, name_width;
837 const char *reset, *set, *add_c, *del_c;
839 if (data->nr == 0)
840 return;
842 width = options->stat_width ? options->stat_width : 80;
843 name_width = options->stat_name_width ? options->stat_name_width : 50;
845 /* Sanity: give at least 5 columns to the graph,
846 * but leave at least 10 columns for the name.
848 if (width < 25)
849 width = 25;
850 if (name_width < 10)
851 name_width = 10;
852 else if (width < name_width + 15)
853 name_width = width - 15;
855 /* Find the longest filename and max number of changes */
856 reset = diff_get_color_opt(options, DIFF_RESET);
857 set = diff_get_color_opt(options, DIFF_PLAIN);
858 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
859 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
861 for (i = 0; i < data->nr; i++) {
862 struct diffstat_file *file = data->files[i];
863 int change = file->added + file->deleted;
864 fill_print_name(file);
865 len = strlen(file->print_name);
866 if (max_len < len)
867 max_len = len;
869 if (file->is_binary || file->is_unmerged)
870 continue;
871 if (max_change < change)
872 max_change = change;
875 /* Compute the width of the graph part;
876 * 10 is for one blank at the beginning of the line plus
877 * " | count " between the name and the graph.
879 * From here on, name_width is the width of the name area,
880 * and width is the width of the graph area.
882 name_width = (name_width < max_len) ? name_width : max_len;
883 if (width < (name_width + 10) + max_change)
884 width = width - (name_width + 10);
885 else
886 width = max_change;
888 for (i = 0; i < data->nr; i++) {
889 const char *prefix = "";
890 char *name = data->files[i]->print_name;
891 int added = data->files[i]->added;
892 int deleted = data->files[i]->deleted;
893 int name_len;
896 * "scale" the filename
898 len = name_width;
899 name_len = strlen(name);
900 if (name_width < name_len) {
901 char *slash;
902 prefix = "...";
903 len -= 3;
904 name += name_len - len;
905 slash = strchr(name, '/');
906 if (slash)
907 name = slash;
910 if (data->files[i]->is_binary) {
911 show_name(options->file, prefix, name, len, reset, set);
912 fprintf(options->file, " Bin ");
913 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
914 fprintf(options->file, " -> ");
915 fprintf(options->file, "%s%d%s", add_c, added, reset);
916 fprintf(options->file, " bytes");
917 fprintf(options->file, "\n");
918 continue;
920 else if (data->files[i]->is_unmerged) {
921 show_name(options->file, prefix, name, len, reset, set);
922 fprintf(options->file, " Unmerged\n");
923 continue;
925 else if (!data->files[i]->is_renamed &&
926 (added + deleted == 0)) {
927 total_files--;
928 continue;
932 * scale the add/delete
934 add = added;
935 del = deleted;
936 total = add + del;
937 adds += add;
938 dels += del;
940 if (width <= max_change) {
941 add = scale_linear(add, width, max_change);
942 del = scale_linear(del, width, max_change);
943 total = add + del;
945 show_name(options->file, prefix, name, len, reset, set);
946 fprintf(options->file, "%5d%s", added + deleted,
947 added + deleted ? " " : "");
948 show_graph(options->file, '+', add, add_c, reset);
949 show_graph(options->file, '-', del, del_c, reset);
950 fprintf(options->file, "\n");
952 fprintf(options->file,
953 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
954 set, total_files, adds, dels, reset);
957 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
959 int i, adds = 0, dels = 0, total_files = data->nr;
961 if (data->nr == 0)
962 return;
964 for (i = 0; i < data->nr; i++) {
965 if (!data->files[i]->is_binary &&
966 !data->files[i]->is_unmerged) {
967 int added = data->files[i]->added;
968 int deleted= data->files[i]->deleted;
969 if (!data->files[i]->is_renamed &&
970 (added + deleted == 0)) {
971 total_files--;
972 } else {
973 adds += added;
974 dels += deleted;
978 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
979 total_files, adds, dels);
982 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
984 int i;
986 if (data->nr == 0)
987 return;
989 for (i = 0; i < data->nr; i++) {
990 struct diffstat_file *file = data->files[i];
992 if (file->is_binary)
993 fprintf(options->file, "-\t-\t");
994 else
995 fprintf(options->file,
996 "%d\t%d\t", file->added, file->deleted);
997 if (options->line_termination) {
998 fill_print_name(file);
999 if (!file->is_renamed)
1000 write_name_quoted(file->name, options->file,
1001 options->line_termination);
1002 else {
1003 fputs(file->print_name, options->file);
1004 putc(options->line_termination, options->file);
1006 } else {
1007 if (file->is_renamed) {
1008 putc('\0', options->file);
1009 write_name_quoted(file->from_name, options->file, '\0');
1011 write_name_quoted(file->name, options->file, '\0');
1016 struct dirstat_file {
1017 const char *name;
1018 unsigned long changed;
1021 struct dirstat_dir {
1022 struct dirstat_file *files;
1023 int alloc, nr, percent, cumulative;
1026 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1028 unsigned long this_dir = 0;
1029 unsigned int sources = 0;
1031 while (dir->nr) {
1032 struct dirstat_file *f = dir->files;
1033 int namelen = strlen(f->name);
1034 unsigned long this;
1035 char *slash;
1037 if (namelen < baselen)
1038 break;
1039 if (memcmp(f->name, base, baselen))
1040 break;
1041 slash = strchr(f->name + baselen, '/');
1042 if (slash) {
1043 int newbaselen = slash + 1 - f->name;
1044 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1045 sources++;
1046 } else {
1047 this = f->changed;
1048 dir->files++;
1049 dir->nr--;
1050 sources += 2;
1052 this_dir += this;
1056 * We don't report dirstat's for
1057 * - the top level
1058 * - or cases where everything came from a single directory
1059 * under this directory (sources == 1).
1061 if (baselen && sources != 1) {
1062 int permille = this_dir * 1000 / changed;
1063 if (permille) {
1064 int percent = permille / 10;
1065 if (percent >= dir->percent) {
1066 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1067 if (!dir->cumulative)
1068 return 0;
1072 return this_dir;
1075 static int dirstat_compare(const void *_a, const void *_b)
1077 const struct dirstat_file *a = _a;
1078 const struct dirstat_file *b = _b;
1079 return strcmp(a->name, b->name);
1082 static void show_dirstat(struct diff_options *options)
1084 int i;
1085 unsigned long changed;
1086 struct dirstat_dir dir;
1087 struct diff_queue_struct *q = &diff_queued_diff;
1089 dir.files = NULL;
1090 dir.alloc = 0;
1091 dir.nr = 0;
1092 dir.percent = options->dirstat_percent;
1093 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1095 changed = 0;
1096 for (i = 0; i < q->nr; i++) {
1097 struct diff_filepair *p = q->queue[i];
1098 const char *name;
1099 unsigned long copied, added, damage;
1101 name = p->one->path ? p->one->path : p->two->path;
1103 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1104 diff_populate_filespec(p->one, 0);
1105 diff_populate_filespec(p->two, 0);
1106 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1107 &copied, &added);
1108 diff_free_filespec_data(p->one);
1109 diff_free_filespec_data(p->two);
1110 } else if (DIFF_FILE_VALID(p->one)) {
1111 diff_populate_filespec(p->one, 1);
1112 copied = added = 0;
1113 diff_free_filespec_data(p->one);
1114 } else if (DIFF_FILE_VALID(p->two)) {
1115 diff_populate_filespec(p->two, 1);
1116 copied = 0;
1117 added = p->two->size;
1118 diff_free_filespec_data(p->two);
1119 } else
1120 continue;
1123 * Original minus copied is the removed material,
1124 * added is the new material. They are both damages
1125 * made to the preimage.
1127 damage = (p->one->size - copied) + added;
1129 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1130 dir.files[dir.nr].name = name;
1131 dir.files[dir.nr].changed = damage;
1132 changed += damage;
1133 dir.nr++;
1136 /* This can happen even with many files, if everything was renames */
1137 if (!changed)
1138 return;
1140 /* Show all directories with more than x% of the changes */
1141 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1142 gather_dirstat(options->file, &dir, changed, "", 0);
1145 static void free_diffstat_info(struct diffstat_t *diffstat)
1147 int i;
1148 for (i = 0; i < diffstat->nr; i++) {
1149 struct diffstat_file *f = diffstat->files[i];
1150 if (f->name != f->print_name)
1151 free(f->print_name);
1152 free(f->name);
1153 free(f->from_name);
1154 free(f);
1156 free(diffstat->files);
1159 struct checkdiff_t {
1160 struct xdiff_emit_state xm;
1161 const char *filename;
1162 int lineno;
1163 struct diff_options *o;
1164 unsigned ws_rule;
1165 unsigned status;
1168 static int is_conflict_marker(const char *line, unsigned long len)
1170 char firstchar;
1171 int cnt;
1173 if (len < 8)
1174 return 0;
1175 firstchar = line[0];
1176 switch (firstchar) {
1177 case '=': case '>': case '<':
1178 break;
1179 default:
1180 return 0;
1182 for (cnt = 1; cnt < 7; cnt++)
1183 if (line[cnt] != firstchar)
1184 return 0;
1185 /* line[0] thru line[6] are same as firstchar */
1186 if (firstchar == '=') {
1187 /* divider between ours and theirs? */
1188 if (len != 8 || line[7] != '\n')
1189 return 0;
1190 } else if (len < 8 || !isspace(line[7])) {
1191 /* not divider before ours nor after theirs */
1192 return 0;
1194 return 1;
1197 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1199 struct checkdiff_t *data = priv;
1200 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1201 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1202 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1203 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1204 char *err;
1206 if (line[0] == '+') {
1207 unsigned bad;
1208 data->lineno++;
1209 if (is_conflict_marker(line + 1, len - 1)) {
1210 data->status |= 1;
1211 fprintf(data->o->file,
1212 "%s:%d: leftover conflict marker\n",
1213 data->filename, data->lineno);
1215 bad = ws_check(line + 1, len - 1, data->ws_rule);
1216 if (!bad)
1217 return;
1218 data->status |= bad;
1219 err = whitespace_error_string(bad);
1220 fprintf(data->o->file, "%s:%d: %s.\n",
1221 data->filename, data->lineno, err);
1222 free(err);
1223 emit_line(data->o->file, set, reset, line, 1);
1224 ws_check_emit(line + 1, len - 1, data->ws_rule,
1225 data->o->file, set, reset, ws);
1226 } else if (line[0] == ' ') {
1227 data->lineno++;
1228 } else if (line[0] == '@') {
1229 char *plus = strchr(line, '+');
1230 if (plus)
1231 data->lineno = strtol(plus, NULL, 10) - 1;
1232 else
1233 die("invalid diff");
1237 static unsigned char *deflate_it(char *data,
1238 unsigned long size,
1239 unsigned long *result_size)
1241 int bound;
1242 unsigned char *deflated;
1243 z_stream stream;
1245 memset(&stream, 0, sizeof(stream));
1246 deflateInit(&stream, zlib_compression_level);
1247 bound = deflateBound(&stream, size);
1248 deflated = xmalloc(bound);
1249 stream.next_out = deflated;
1250 stream.avail_out = bound;
1252 stream.next_in = (unsigned char *)data;
1253 stream.avail_in = size;
1254 while (deflate(&stream, Z_FINISH) == Z_OK)
1255 ; /* nothing */
1256 deflateEnd(&stream);
1257 *result_size = stream.total_out;
1258 return deflated;
1261 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1263 void *cp;
1264 void *delta;
1265 void *deflated;
1266 void *data;
1267 unsigned long orig_size;
1268 unsigned long delta_size;
1269 unsigned long deflate_size;
1270 unsigned long data_size;
1272 /* We could do deflated delta, or we could do just deflated two,
1273 * whichever is smaller.
1275 delta = NULL;
1276 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1277 if (one->size && two->size) {
1278 delta = diff_delta(one->ptr, one->size,
1279 two->ptr, two->size,
1280 &delta_size, deflate_size);
1281 if (delta) {
1282 void *to_free = delta;
1283 orig_size = delta_size;
1284 delta = deflate_it(delta, delta_size, &delta_size);
1285 free(to_free);
1289 if (delta && delta_size < deflate_size) {
1290 fprintf(file, "delta %lu\n", orig_size);
1291 free(deflated);
1292 data = delta;
1293 data_size = delta_size;
1295 else {
1296 fprintf(file, "literal %lu\n", two->size);
1297 free(delta);
1298 data = deflated;
1299 data_size = deflate_size;
1302 /* emit data encoded in base85 */
1303 cp = data;
1304 while (data_size) {
1305 int bytes = (52 < data_size) ? 52 : data_size;
1306 char line[70];
1307 data_size -= bytes;
1308 if (bytes <= 26)
1309 line[0] = bytes + 'A' - 1;
1310 else
1311 line[0] = bytes - 26 + 'a' - 1;
1312 encode_85(line + 1, cp, bytes);
1313 cp = (char *) cp + bytes;
1314 fputs(line, file);
1315 fputc('\n', file);
1317 fprintf(file, "\n");
1318 free(data);
1321 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1323 fprintf(file, "GIT binary patch\n");
1324 emit_binary_diff_body(file, one, two);
1325 emit_binary_diff_body(file, two, one);
1328 static void setup_diff_attr_check(struct git_attr_check *check)
1330 static struct git_attr *attr_diff;
1332 if (!attr_diff) {
1333 attr_diff = git_attr("diff", 4);
1335 check[0].attr = attr_diff;
1338 static void diff_filespec_check_attr(struct diff_filespec *one)
1340 struct git_attr_check attr_diff_check;
1341 int check_from_data = 0;
1343 if (one->checked_attr)
1344 return;
1346 setup_diff_attr_check(&attr_diff_check);
1347 one->is_binary = 0;
1348 one->funcname_pattern_ident = NULL;
1350 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1351 const char *value;
1353 /* binaryness */
1354 value = attr_diff_check.value;
1355 if (ATTR_TRUE(value))
1357 else if (ATTR_FALSE(value))
1358 one->is_binary = 1;
1359 else
1360 check_from_data = 1;
1362 /* funcname pattern ident */
1363 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1365 else
1366 one->funcname_pattern_ident = value;
1369 if (check_from_data) {
1370 if (!one->data && DIFF_FILE_VALID(one))
1371 diff_populate_filespec(one, 0);
1373 if (one->data)
1374 one->is_binary = buffer_is_binary(one->data, one->size);
1378 int diff_filespec_is_binary(struct diff_filespec *one)
1380 diff_filespec_check_attr(one);
1381 return one->is_binary;
1384 static const struct funcname_pattern_entry *funcname_pattern(const char *ident)
1386 struct funcname_pattern_list *pp;
1388 for (pp = funcname_pattern_list; pp; pp = pp->next)
1389 if (!strcmp(ident, pp->e.name))
1390 return &pp->e;
1391 return NULL;
1394 static const struct funcname_pattern_entry builtin_funcname_pattern[] = {
1395 { "java",
1396 "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
1397 "^[ \t]*(([ \t]*[A-Za-z_][A-Za-z_0-9]*){2,}[ \t]*\\([^;]*)$",
1398 REG_EXTENDED },
1399 { "pascal",
1400 "^((procedure|function|constructor|destructor|interface|"
1401 "implementation|initialization|finalization)[ \t]*.*)$"
1403 "^(.*=[ \t]*(class|record).*)$",
1404 REG_EXTENDED },
1405 { "bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
1406 REG_EXTENDED },
1407 { "tex",
1408 "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
1409 REG_EXTENDED },
1410 { "ruby", "^[ \t]*((class|module|def)[ \t].*)$",
1411 REG_EXTENDED },
1414 static const struct funcname_pattern_entry *diff_funcname_pattern(struct diff_filespec *one)
1416 const char *ident;
1417 const struct funcname_pattern_entry *pe;
1418 int i;
1420 diff_filespec_check_attr(one);
1421 ident = one->funcname_pattern_ident;
1423 if (!ident)
1425 * If the config file has "funcname.default" defined, that
1426 * regexp is used; otherwise NULL is returned and xemit uses
1427 * the built-in default.
1429 return funcname_pattern("default");
1431 /* Look up custom "funcname.$ident" regexp from config. */
1432 pe = funcname_pattern(ident);
1433 if (pe)
1434 return pe;
1437 * And define built-in fallback patterns here. Note that
1438 * these can be overridden by the user's config settings.
1440 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1441 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1442 return &builtin_funcname_pattern[i];
1444 return NULL;
1447 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
1449 char *ptr = mf->ptr;
1450 long size = mf->size;
1451 int cnt = 0;
1453 if (!size)
1454 return cnt;
1455 ptr += size - 1; /* pointing at the very end */
1456 if (*ptr != '\n')
1457 ; /* incomplete line */
1458 else
1459 ptr--; /* skip the last LF */
1460 while (mf->ptr < ptr) {
1461 char *prev_eol;
1462 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
1463 if (*prev_eol == '\n')
1464 break;
1465 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
1466 break;
1467 cnt++;
1468 ptr = prev_eol - 1;
1470 return cnt;
1473 static int adds_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2, unsigned ws_rule)
1475 int l1, l2, at;
1476 l1 = count_trailing_blank(mf1, ws_rule);
1477 l2 = count_trailing_blank(mf2, ws_rule);
1478 if (l2 <= l1)
1479 return 0;
1480 /* starting where? */
1481 at = count_lines(mf1->ptr, mf1->size);
1482 return (at - l1) + 1; /* the line number counts from 1 */
1485 static void builtin_diff(const char *name_a,
1486 const char *name_b,
1487 struct diff_filespec *one,
1488 struct diff_filespec *two,
1489 const char *xfrm_msg,
1490 struct diff_options *o,
1491 int complete_rewrite)
1493 mmfile_t mf1, mf2;
1494 const char *lbl[2];
1495 char *a_one, *b_two;
1496 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1497 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1499 /* Never use a non-valid filename anywhere if at all possible */
1500 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1501 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1503 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1504 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1505 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1506 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1507 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1508 if (lbl[0][0] == '/') {
1509 /* /dev/null */
1510 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1511 if (xfrm_msg && xfrm_msg[0])
1512 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1514 else if (lbl[1][0] == '/') {
1515 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1516 if (xfrm_msg && xfrm_msg[0])
1517 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1519 else {
1520 if (one->mode != two->mode) {
1521 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1522 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1524 if (xfrm_msg && xfrm_msg[0])
1525 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1527 * we do not run diff between different kind
1528 * of objects.
1530 if ((one->mode ^ two->mode) & S_IFMT)
1531 goto free_ab_and_return;
1532 if (complete_rewrite) {
1533 emit_rewrite_diff(name_a, name_b, one, two, o);
1534 o->found_changes = 1;
1535 goto free_ab_and_return;
1539 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1540 die("unable to read files to diff");
1542 if (!DIFF_OPT_TST(o, TEXT) &&
1543 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1544 /* Quite common confusing case */
1545 if (mf1.size == mf2.size &&
1546 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1547 goto free_ab_and_return;
1548 if (DIFF_OPT_TST(o, BINARY))
1549 emit_binary_diff(o->file, &mf1, &mf2);
1550 else
1551 fprintf(o->file, "Binary files %s and %s differ\n",
1552 lbl[0], lbl[1]);
1553 o->found_changes = 1;
1555 else {
1556 /* Crazy xdl interfaces.. */
1557 const char *diffopts = getenv("GIT_DIFF_OPTS");
1558 xpparam_t xpp;
1559 xdemitconf_t xecfg;
1560 xdemitcb_t ecb;
1561 struct emit_callback ecbdata;
1562 const struct funcname_pattern_entry *pe;
1564 pe = diff_funcname_pattern(one);
1565 if (!pe)
1566 pe = diff_funcname_pattern(two);
1568 memset(&xecfg, 0, sizeof(xecfg));
1569 memset(&ecbdata, 0, sizeof(ecbdata));
1570 ecbdata.label_path = lbl;
1571 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1572 ecbdata.found_changesp = &o->found_changes;
1573 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1574 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1575 ecbdata.blank_at_eof =
1576 adds_blank_at_eof(&mf1, &mf2, ecbdata.ws_rule);
1577 ecbdata.file = o->file;
1578 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1579 xecfg.ctxlen = o->context;
1580 xecfg.flags = XDL_EMIT_FUNCNAMES;
1581 if (pe)
1582 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1583 if (!diffopts)
1585 else if (!prefixcmp(diffopts, "--unified="))
1586 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1587 else if (!prefixcmp(diffopts, "-u"))
1588 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1589 ecb.outf = xdiff_outf;
1590 ecb.priv = &ecbdata;
1591 ecbdata.xm.consume = fn_out_consume;
1592 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1593 ecbdata.diff_words =
1594 xcalloc(1, sizeof(struct diff_words_data));
1595 ecbdata.diff_words->file = o->file;
1597 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1598 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1599 free_diff_words_data(&ecbdata);
1602 free_ab_and_return:
1603 diff_free_filespec_data(one);
1604 diff_free_filespec_data(two);
1605 free(a_one);
1606 free(b_two);
1607 return;
1610 static void builtin_diffstat(const char *name_a, const char *name_b,
1611 struct diff_filespec *one,
1612 struct diff_filespec *two,
1613 struct diffstat_t *diffstat,
1614 struct diff_options *o,
1615 int complete_rewrite)
1617 mmfile_t mf1, mf2;
1618 struct diffstat_file *data;
1620 data = diffstat_add(diffstat, name_a, name_b);
1622 if (!one || !two) {
1623 data->is_unmerged = 1;
1624 return;
1626 if (complete_rewrite) {
1627 diff_populate_filespec(one, 0);
1628 diff_populate_filespec(two, 0);
1629 data->deleted = count_lines(one->data, one->size);
1630 data->added = count_lines(two->data, two->size);
1631 goto free_and_return;
1633 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1634 die("unable to read files to diff");
1636 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1637 data->is_binary = 1;
1638 data->added = mf2.size;
1639 data->deleted = mf1.size;
1640 } else {
1641 /* Crazy xdl interfaces.. */
1642 xpparam_t xpp;
1643 xdemitconf_t xecfg;
1644 xdemitcb_t ecb;
1646 memset(&xecfg, 0, sizeof(xecfg));
1647 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1648 ecb.outf = xdiff_outf;
1649 ecb.priv = diffstat;
1650 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1653 free_and_return:
1654 diff_free_filespec_data(one);
1655 diff_free_filespec_data(two);
1658 static void builtin_checkdiff(const char *name_a, const char *name_b,
1659 const char *attr_path,
1660 struct diff_filespec *one,
1661 struct diff_filespec *two,
1662 struct diff_options *o)
1664 mmfile_t mf1, mf2;
1665 struct checkdiff_t data;
1667 if (!two)
1668 return;
1670 memset(&data, 0, sizeof(data));
1671 data.xm.consume = checkdiff_consume;
1672 data.filename = name_b ? name_b : name_a;
1673 data.lineno = 0;
1674 data.o = o;
1675 data.ws_rule = whitespace_rule(attr_path);
1677 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1678 die("unable to read files to diff");
1681 * All the other codepaths check both sides, but not checking
1682 * the "old" side here is deliberate. We are checking the newly
1683 * introduced changes, and as long as the "new" side is text, we
1684 * can and should check what it introduces.
1686 if (diff_filespec_is_binary(two))
1687 goto free_and_return;
1688 else {
1689 /* Crazy xdl interfaces.. */
1690 xpparam_t xpp;
1691 xdemitconf_t xecfg;
1692 xdemitcb_t ecb;
1694 memset(&xecfg, 0, sizeof(xecfg));
1695 xecfg.ctxlen = 1; /* at least one context line */
1696 xpp.flags = XDF_NEED_MINIMAL;
1697 ecb.outf = xdiff_outf;
1698 ecb.priv = &data;
1699 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1701 if (data.ws_rule & WS_BLANK_AT_EOF) {
1702 int blank_at_eof = adds_blank_at_eof(&mf1, &mf2, data.ws_rule);
1703 if (blank_at_eof) {
1704 static char *err;
1705 if (!err)
1706 err = whitespace_error_string(WS_BLANK_AT_EOF);
1707 fprintf(o->file, "%s:%d: %s.\n",
1708 data.filename, blank_at_eof, err);
1709 data.status = 1; /* report errors */
1713 free_and_return:
1714 diff_free_filespec_data(one);
1715 diff_free_filespec_data(two);
1716 if (data.status)
1717 DIFF_OPT_SET(o, CHECK_FAILED);
1720 struct diff_filespec *alloc_filespec(const char *path)
1722 int namelen = strlen(path);
1723 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1725 memset(spec, 0, sizeof(*spec));
1726 spec->path = (char *)(spec + 1);
1727 memcpy(spec->path, path, namelen+1);
1728 spec->count = 1;
1729 return spec;
1732 void free_filespec(struct diff_filespec *spec)
1734 if (!--spec->count) {
1735 diff_free_filespec_data(spec);
1736 free(spec);
1740 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1741 unsigned short mode)
1743 if (mode) {
1744 spec->mode = canon_mode(mode);
1745 hashcpy(spec->sha1, sha1);
1746 spec->sha1_valid = !is_null_sha1(sha1);
1751 * Given a name and sha1 pair, if the index tells us the file in
1752 * the work tree has that object contents, return true, so that
1753 * prepare_temp_file() does not have to inflate and extract.
1755 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1757 struct cache_entry *ce;
1758 struct stat st;
1759 int pos, len;
1761 /* We do not read the cache ourselves here, because the
1762 * benchmark with my previous version that always reads cache
1763 * shows that it makes things worse for diff-tree comparing
1764 * two linux-2.6 kernel trees in an already checked out work
1765 * tree. This is because most diff-tree comparisons deal with
1766 * only a small number of files, while reading the cache is
1767 * expensive for a large project, and its cost outweighs the
1768 * savings we get by not inflating the object to a temporary
1769 * file. Practically, this code only helps when we are used
1770 * by diff-cache --cached, which does read the cache before
1771 * calling us.
1773 if (!active_cache)
1774 return 0;
1776 /* We want to avoid the working directory if our caller
1777 * doesn't need the data in a normal file, this system
1778 * is rather slow with its stat/open/mmap/close syscalls,
1779 * and the object is contained in a pack file. The pack
1780 * is probably already open and will be faster to obtain
1781 * the data through than the working directory. Loose
1782 * objects however would tend to be slower as they need
1783 * to be individually opened and inflated.
1785 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1786 return 0;
1788 len = strlen(name);
1789 pos = cache_name_pos(name, len);
1790 if (pos < 0)
1791 return 0;
1792 ce = active_cache[pos];
1795 * This is not the sha1 we are looking for, or
1796 * unreusable because it is not a regular file.
1798 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1799 return 0;
1802 * If ce matches the file in the work tree, we can reuse it.
1804 if (ce_uptodate(ce) ||
1805 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1806 return 1;
1808 return 0;
1811 static int populate_from_stdin(struct diff_filespec *s)
1813 struct strbuf buf;
1814 size_t size = 0;
1816 strbuf_init(&buf, 0);
1817 if (strbuf_read(&buf, 0, 0) < 0)
1818 return error("error while reading from stdin %s",
1819 strerror(errno));
1821 s->should_munmap = 0;
1822 s->data = strbuf_detach(&buf, &size);
1823 s->size = size;
1824 s->should_free = 1;
1825 return 0;
1828 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1830 int len;
1831 char *data = xmalloc(100);
1832 len = snprintf(data, 100,
1833 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1834 s->data = data;
1835 s->size = len;
1836 s->should_free = 1;
1837 if (size_only) {
1838 s->data = NULL;
1839 free(data);
1841 return 0;
1845 * While doing rename detection and pickaxe operation, we may need to
1846 * grab the data for the blob (or file) for our own in-core comparison.
1847 * diff_filespec has data and size fields for this purpose.
1849 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1851 int err = 0;
1852 if (!DIFF_FILE_VALID(s))
1853 die("internal error: asking to populate invalid file.");
1854 if (S_ISDIR(s->mode))
1855 return -1;
1857 if (s->data)
1858 return 0;
1860 if (size_only && 0 < s->size)
1861 return 0;
1863 if (S_ISGITLINK(s->mode))
1864 return diff_populate_gitlink(s, size_only);
1866 if (!s->sha1_valid ||
1867 reuse_worktree_file(s->path, s->sha1, 0)) {
1868 struct strbuf buf;
1869 struct stat st;
1870 int fd;
1872 if (!strcmp(s->path, "-"))
1873 return populate_from_stdin(s);
1875 if (lstat(s->path, &st) < 0) {
1876 if (errno == ENOENT) {
1877 err_empty:
1878 err = -1;
1879 empty:
1880 s->data = (char *)"";
1881 s->size = 0;
1882 return err;
1885 s->size = xsize_t(st.st_size);
1886 if (!s->size)
1887 goto empty;
1888 if (size_only)
1889 return 0;
1890 if (S_ISLNK(st.st_mode)) {
1891 int ret;
1892 s->data = xmalloc(s->size);
1893 s->should_free = 1;
1894 ret = readlink(s->path, s->data, s->size);
1895 if (ret < 0) {
1896 free(s->data);
1897 goto err_empty;
1899 return 0;
1901 fd = open(s->path, O_RDONLY);
1902 if (fd < 0)
1903 goto err_empty;
1904 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1905 close(fd);
1906 s->should_munmap = 1;
1909 * Convert from working tree format to canonical git format
1911 strbuf_init(&buf, 0);
1912 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1913 size_t size = 0;
1914 munmap(s->data, s->size);
1915 s->should_munmap = 0;
1916 s->data = strbuf_detach(&buf, &size);
1917 s->size = size;
1918 s->should_free = 1;
1921 else {
1922 enum object_type type;
1923 if (size_only)
1924 type = sha1_object_info(s->sha1, &s->size);
1925 else {
1926 s->data = read_sha1_file(s->sha1, &type, &s->size);
1927 s->should_free = 1;
1930 return 0;
1933 void diff_free_filespec_blob(struct diff_filespec *s)
1935 if (s->should_free)
1936 free(s->data);
1937 else if (s->should_munmap)
1938 munmap(s->data, s->size);
1940 if (s->should_free || s->should_munmap) {
1941 s->should_free = s->should_munmap = 0;
1942 s->data = NULL;
1946 void diff_free_filespec_data(struct diff_filespec *s)
1948 diff_free_filespec_blob(s);
1949 free(s->cnt_data);
1950 s->cnt_data = NULL;
1953 static void prep_temp_blob(struct diff_tempfile *temp,
1954 void *blob,
1955 unsigned long size,
1956 const unsigned char *sha1,
1957 int mode)
1959 int fd;
1961 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1962 if (fd < 0)
1963 die("unable to create temp-file: %s", strerror(errno));
1964 if (write_in_full(fd, blob, size) != size)
1965 die("unable to write temp-file");
1966 close(fd);
1967 temp->name = temp->tmp_path;
1968 strcpy(temp->hex, sha1_to_hex(sha1));
1969 temp->hex[40] = 0;
1970 sprintf(temp->mode, "%06o", mode);
1973 static void prepare_temp_file(const char *name,
1974 struct diff_tempfile *temp,
1975 struct diff_filespec *one)
1977 if (!DIFF_FILE_VALID(one)) {
1978 not_a_valid_file:
1979 /* A '-' entry produces this for file-2, and
1980 * a '+' entry produces this for file-1.
1982 temp->name = "/dev/null";
1983 strcpy(temp->hex, ".");
1984 strcpy(temp->mode, ".");
1985 return;
1988 if (!one->sha1_valid ||
1989 reuse_worktree_file(name, one->sha1, 1)) {
1990 struct stat st;
1991 if (lstat(name, &st) < 0) {
1992 if (errno == ENOENT)
1993 goto not_a_valid_file;
1994 die("stat(%s): %s", name, strerror(errno));
1996 if (S_ISLNK(st.st_mode)) {
1997 int ret;
1998 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1999 size_t sz = xsize_t(st.st_size);
2000 if (sizeof(buf) <= st.st_size)
2001 die("symlink too long: %s", name);
2002 ret = readlink(name, buf, sz);
2003 if (ret < 0)
2004 die("readlink(%s)", name);
2005 prep_temp_blob(temp, buf, sz,
2006 (one->sha1_valid ?
2007 one->sha1 : null_sha1),
2008 (one->sha1_valid ?
2009 one->mode : S_IFLNK));
2011 else {
2012 /* we can borrow from the file in the work tree */
2013 temp->name = name;
2014 if (!one->sha1_valid)
2015 strcpy(temp->hex, sha1_to_hex(null_sha1));
2016 else
2017 strcpy(temp->hex, sha1_to_hex(one->sha1));
2018 /* Even though we may sometimes borrow the
2019 * contents from the work tree, we always want
2020 * one->mode. mode is trustworthy even when
2021 * !(one->sha1_valid), as long as
2022 * DIFF_FILE_VALID(one).
2024 sprintf(temp->mode, "%06o", one->mode);
2026 return;
2028 else {
2029 if (diff_populate_filespec(one, 0))
2030 die("cannot read data blob for %s", one->path);
2031 prep_temp_blob(temp, one->data, one->size,
2032 one->sha1, one->mode);
2036 static void remove_tempfile(void)
2038 int i;
2040 for (i = 0; i < 2; i++)
2041 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2042 unlink(diff_temp[i].name);
2043 diff_temp[i].name = NULL;
2047 static void remove_tempfile_on_signal(int signo)
2049 remove_tempfile();
2050 signal(SIGINT, SIG_DFL);
2051 raise(signo);
2054 /* An external diff command takes:
2056 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2057 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2060 static void run_external_diff(const char *pgm,
2061 const char *name,
2062 const char *other,
2063 struct diff_filespec *one,
2064 struct diff_filespec *two,
2065 const char *xfrm_msg,
2066 int complete_rewrite)
2068 const char *spawn_arg[10];
2069 struct diff_tempfile *temp = diff_temp;
2070 int retval;
2071 static int atexit_asked = 0;
2072 const char *othername;
2073 const char **arg = &spawn_arg[0];
2075 othername = (other? other : name);
2076 if (one && two) {
2077 prepare_temp_file(name, &temp[0], one);
2078 prepare_temp_file(othername, &temp[1], two);
2079 if (! atexit_asked &&
2080 (temp[0].name == temp[0].tmp_path ||
2081 temp[1].name == temp[1].tmp_path)) {
2082 atexit_asked = 1;
2083 atexit(remove_tempfile);
2085 signal(SIGINT, remove_tempfile_on_signal);
2088 if (one && two) {
2089 *arg++ = pgm;
2090 *arg++ = name;
2091 *arg++ = temp[0].name;
2092 *arg++ = temp[0].hex;
2093 *arg++ = temp[0].mode;
2094 *arg++ = temp[1].name;
2095 *arg++ = temp[1].hex;
2096 *arg++ = temp[1].mode;
2097 if (other) {
2098 *arg++ = other;
2099 *arg++ = xfrm_msg;
2101 } else {
2102 *arg++ = pgm;
2103 *arg++ = name;
2105 *arg = NULL;
2106 fflush(NULL);
2107 retval = run_command_v_opt(spawn_arg, 0);
2108 remove_tempfile();
2109 if (retval) {
2110 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2111 exit(1);
2115 static const char *external_diff_attr(const char *name)
2117 struct git_attr_check attr_diff_check;
2119 if (!name)
2120 return NULL;
2122 setup_diff_attr_check(&attr_diff_check);
2123 if (!git_checkattr(name, 1, &attr_diff_check)) {
2124 const char *value = attr_diff_check.value;
2125 if (!ATTR_TRUE(value) &&
2126 !ATTR_FALSE(value) &&
2127 !ATTR_UNSET(value)) {
2128 struct ll_diff_driver *drv;
2130 for (drv = user_diff; drv; drv = drv->next)
2131 if (!strcmp(drv->name, value))
2132 return drv->cmd;
2135 return NULL;
2138 static int similarity_index(struct diff_filepair *p)
2140 return p->score * 100 / MAX_SCORE;
2143 static void fill_metainfo(struct strbuf *msg,
2144 const char *name,
2145 const char *other,
2146 struct diff_filespec *one,
2147 struct diff_filespec *two,
2148 struct diff_options *o,
2149 struct diff_filepair *p)
2151 strbuf_init(msg, PATH_MAX * 2 + 300);
2152 switch (p->status) {
2153 case DIFF_STATUS_COPIED:
2154 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2155 strbuf_addstr(msg, "\ncopy from ");
2156 quote_c_style(name, msg, NULL, 0);
2157 strbuf_addstr(msg, "\ncopy to ");
2158 quote_c_style(other, msg, NULL, 0);
2159 strbuf_addch(msg, '\n');
2160 break;
2161 case DIFF_STATUS_RENAMED:
2162 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2163 strbuf_addstr(msg, "\nrename from ");
2164 quote_c_style(name, msg, NULL, 0);
2165 strbuf_addstr(msg, "\nrename to ");
2166 quote_c_style(other, msg, NULL, 0);
2167 strbuf_addch(msg, '\n');
2168 break;
2169 case DIFF_STATUS_MODIFIED:
2170 if (p->score) {
2171 strbuf_addf(msg, "dissimilarity index %d%%\n",
2172 similarity_index(p));
2173 break;
2175 /* fallthru */
2176 default:
2177 /* nothing */
2180 if (one && two && hashcmp(one->sha1, two->sha1)) {
2181 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2183 if (DIFF_OPT_TST(o, BINARY)) {
2184 mmfile_t mf;
2185 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2186 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2187 abbrev = 40;
2189 strbuf_addf(msg, "index %.*s..%.*s",
2190 abbrev, sha1_to_hex(one->sha1),
2191 abbrev, sha1_to_hex(two->sha1));
2192 if (one->mode == two->mode)
2193 strbuf_addf(msg, " %06o", one->mode);
2194 strbuf_addch(msg, '\n');
2196 if (msg->len)
2197 strbuf_setlen(msg, msg->len - 1);
2200 static void run_diff_cmd(const char *pgm,
2201 const char *name,
2202 const char *other,
2203 const char *attr_path,
2204 struct diff_filespec *one,
2205 struct diff_filespec *two,
2206 struct strbuf *msg,
2207 struct diff_options *o,
2208 struct diff_filepair *p)
2210 const char *xfrm_msg = NULL;
2211 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2213 if (msg) {
2214 fill_metainfo(msg, name, other, one, two, o, p);
2215 xfrm_msg = msg->len ? msg->buf : NULL;
2218 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2219 pgm = NULL;
2220 else {
2221 const char *cmd = external_diff_attr(attr_path);
2222 if (cmd)
2223 pgm = cmd;
2226 if (pgm) {
2227 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2228 complete_rewrite);
2229 return;
2231 if (one && two)
2232 builtin_diff(name, other ? other : name,
2233 one, two, xfrm_msg, o, complete_rewrite);
2234 else
2235 fprintf(o->file, "* Unmerged path %s\n", name);
2238 static void diff_fill_sha1_info(struct diff_filespec *one)
2240 if (DIFF_FILE_VALID(one)) {
2241 if (!one->sha1_valid) {
2242 struct stat st;
2243 if (!strcmp(one->path, "-")) {
2244 hashcpy(one->sha1, null_sha1);
2245 return;
2247 if (lstat(one->path, &st) < 0)
2248 die("stat %s", one->path);
2249 if (index_path(one->sha1, one->path, &st, 0))
2250 die("cannot hash %s\n", one->path);
2253 else
2254 hashclr(one->sha1);
2257 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2259 /* Strip the prefix but do not molest /dev/null and absolute paths */
2260 if (*namep && **namep != '/')
2261 *namep += prefix_length;
2262 if (*otherp && **otherp != '/')
2263 *otherp += prefix_length;
2266 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2268 const char *pgm = external_diff();
2269 struct strbuf msg;
2270 struct diff_filespec *one = p->one;
2271 struct diff_filespec *two = p->two;
2272 const char *name;
2273 const char *other;
2274 const char *attr_path;
2276 name = p->one->path;
2277 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2278 attr_path = name;
2279 if (o->prefix_length)
2280 strip_prefix(o->prefix_length, &name, &other);
2282 if (DIFF_PAIR_UNMERGED(p)) {
2283 run_diff_cmd(pgm, name, NULL, attr_path,
2284 NULL, NULL, NULL, o, p);
2285 return;
2288 diff_fill_sha1_info(one);
2289 diff_fill_sha1_info(two);
2291 if (!pgm &&
2292 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2293 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2295 * a filepair that changes between file and symlink
2296 * needs to be split into deletion and creation.
2298 struct diff_filespec *null = alloc_filespec(two->path);
2299 run_diff_cmd(NULL, name, other, attr_path,
2300 one, null, &msg, o, p);
2301 free(null);
2302 strbuf_release(&msg);
2304 null = alloc_filespec(one->path);
2305 run_diff_cmd(NULL, name, other, attr_path,
2306 null, two, &msg, o, p);
2307 free(null);
2309 else
2310 run_diff_cmd(pgm, name, other, attr_path,
2311 one, two, &msg, o, p);
2313 strbuf_release(&msg);
2316 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2317 struct diffstat_t *diffstat)
2319 const char *name;
2320 const char *other;
2321 int complete_rewrite = 0;
2323 if (DIFF_PAIR_UNMERGED(p)) {
2324 /* unmerged */
2325 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2326 return;
2329 name = p->one->path;
2330 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2332 if (o->prefix_length)
2333 strip_prefix(o->prefix_length, &name, &other);
2335 diff_fill_sha1_info(p->one);
2336 diff_fill_sha1_info(p->two);
2338 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2339 complete_rewrite = 1;
2340 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2343 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2345 const char *name;
2346 const char *other;
2347 const char *attr_path;
2349 if (DIFF_PAIR_UNMERGED(p)) {
2350 /* unmerged */
2351 return;
2354 name = p->one->path;
2355 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2356 attr_path = other ? other : name;
2358 if (o->prefix_length)
2359 strip_prefix(o->prefix_length, &name, &other);
2361 diff_fill_sha1_info(p->one);
2362 diff_fill_sha1_info(p->two);
2364 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2367 void diff_setup(struct diff_options *options)
2369 memset(options, 0, sizeof(*options));
2371 options->file = stdout;
2373 options->line_termination = '\n';
2374 options->break_opt = -1;
2375 options->rename_limit = -1;
2376 options->dirstat_percent = 3;
2377 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2378 options->context = 3;
2380 options->change = diff_change;
2381 options->add_remove = diff_addremove;
2382 if (diff_use_color_default > 0)
2383 DIFF_OPT_SET(options, COLOR_DIFF);
2384 else
2385 DIFF_OPT_CLR(options, COLOR_DIFF);
2386 options->detect_rename = diff_detect_rename_default;
2388 options->a_prefix = "a/";
2389 options->b_prefix = "b/";
2392 int diff_setup_done(struct diff_options *options)
2394 int count = 0;
2396 if (options->output_format & DIFF_FORMAT_NAME)
2397 count++;
2398 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2399 count++;
2400 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2401 count++;
2402 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2403 count++;
2404 if (count > 1)
2405 die("--name-only, --name-status, --check and -s are mutually exclusive");
2407 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2408 options->detect_rename = DIFF_DETECT_COPY;
2410 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2411 options->prefix = NULL;
2412 if (options->prefix)
2413 options->prefix_length = strlen(options->prefix);
2414 else
2415 options->prefix_length = 0;
2417 if (options->output_format & (DIFF_FORMAT_NAME |
2418 DIFF_FORMAT_NAME_STATUS |
2419 DIFF_FORMAT_CHECKDIFF |
2420 DIFF_FORMAT_NO_OUTPUT))
2421 options->output_format &= ~(DIFF_FORMAT_RAW |
2422 DIFF_FORMAT_NUMSTAT |
2423 DIFF_FORMAT_DIFFSTAT |
2424 DIFF_FORMAT_SHORTSTAT |
2425 DIFF_FORMAT_DIRSTAT |
2426 DIFF_FORMAT_SUMMARY |
2427 DIFF_FORMAT_PATCH);
2430 * These cases always need recursive; we do not drop caller-supplied
2431 * recursive bits for other formats here.
2433 if (options->output_format & (DIFF_FORMAT_PATCH |
2434 DIFF_FORMAT_NUMSTAT |
2435 DIFF_FORMAT_DIFFSTAT |
2436 DIFF_FORMAT_SHORTSTAT |
2437 DIFF_FORMAT_DIRSTAT |
2438 DIFF_FORMAT_SUMMARY |
2439 DIFF_FORMAT_CHECKDIFF))
2440 DIFF_OPT_SET(options, RECURSIVE);
2442 * Also pickaxe would not work very well if you do not say recursive
2444 if (options->pickaxe)
2445 DIFF_OPT_SET(options, RECURSIVE);
2447 if (options->detect_rename && options->rename_limit < 0)
2448 options->rename_limit = diff_rename_limit_default;
2449 if (options->setup & DIFF_SETUP_USE_CACHE) {
2450 if (!active_cache)
2451 /* read-cache does not die even when it fails
2452 * so it is safe for us to do this here. Also
2453 * it does not smudge active_cache or active_nr
2454 * when it fails, so we do not have to worry about
2455 * cleaning it up ourselves either.
2457 read_cache();
2459 if (options->abbrev <= 0 || 40 < options->abbrev)
2460 options->abbrev = 40; /* full */
2463 * It does not make sense to show the first hit we happened
2464 * to have found. It does not make sense not to return with
2465 * exit code in such a case either.
2467 if (DIFF_OPT_TST(options, QUIET)) {
2468 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2469 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2472 return 0;
2475 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2477 char c, *eq;
2478 int len;
2480 if (*arg != '-')
2481 return 0;
2482 c = *++arg;
2483 if (!c)
2484 return 0;
2485 if (c == arg_short) {
2486 c = *++arg;
2487 if (!c)
2488 return 1;
2489 if (val && isdigit(c)) {
2490 char *end;
2491 int n = strtoul(arg, &end, 10);
2492 if (*end)
2493 return 0;
2494 *val = n;
2495 return 1;
2497 return 0;
2499 if (c != '-')
2500 return 0;
2501 arg++;
2502 eq = strchr(arg, '=');
2503 if (eq)
2504 len = eq - arg;
2505 else
2506 len = strlen(arg);
2507 if (!len || strncmp(arg, arg_long, len))
2508 return 0;
2509 if (eq) {
2510 int n;
2511 char *end;
2512 if (!isdigit(*++eq))
2513 return 0;
2514 n = strtoul(eq, &end, 10);
2515 if (*end)
2516 return 0;
2517 *val = n;
2519 return 1;
2522 static int diff_scoreopt_parse(const char *opt);
2524 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2526 const char *arg = av[0];
2528 /* Output format options */
2529 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2530 options->output_format |= DIFF_FORMAT_PATCH;
2531 else if (opt_arg(arg, 'U', "unified", &options->context))
2532 options->output_format |= DIFF_FORMAT_PATCH;
2533 else if (!strcmp(arg, "--raw"))
2534 options->output_format |= DIFF_FORMAT_RAW;
2535 else if (!strcmp(arg, "--patch-with-raw"))
2536 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2537 else if (!strcmp(arg, "--numstat"))
2538 options->output_format |= DIFF_FORMAT_NUMSTAT;
2539 else if (!strcmp(arg, "--shortstat"))
2540 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2541 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2542 options->output_format |= DIFF_FORMAT_DIRSTAT;
2543 else if (!strcmp(arg, "--cumulative")) {
2544 options->output_format |= DIFF_FORMAT_DIRSTAT;
2545 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2547 else if (!strcmp(arg, "--check"))
2548 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2549 else if (!strcmp(arg, "--summary"))
2550 options->output_format |= DIFF_FORMAT_SUMMARY;
2551 else if (!strcmp(arg, "--patch-with-stat"))
2552 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2553 else if (!strcmp(arg, "--name-only"))
2554 options->output_format |= DIFF_FORMAT_NAME;
2555 else if (!strcmp(arg, "--name-status"))
2556 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2557 else if (!strcmp(arg, "-s"))
2558 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2559 else if (!prefixcmp(arg, "--stat")) {
2560 char *end;
2561 int width = options->stat_width;
2562 int name_width = options->stat_name_width;
2563 arg += 6;
2564 end = (char *)arg;
2566 switch (*arg) {
2567 case '-':
2568 if (!prefixcmp(arg, "-width="))
2569 width = strtoul(arg + 7, &end, 10);
2570 else if (!prefixcmp(arg, "-name-width="))
2571 name_width = strtoul(arg + 12, &end, 10);
2572 break;
2573 case '=':
2574 width = strtoul(arg+1, &end, 10);
2575 if (*end == ',')
2576 name_width = strtoul(end+1, &end, 10);
2579 /* Important! This checks all the error cases! */
2580 if (*end)
2581 return 0;
2582 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2583 options->stat_name_width = name_width;
2584 options->stat_width = width;
2587 /* renames options */
2588 else if (!prefixcmp(arg, "-B")) {
2589 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2590 return -1;
2592 else if (!prefixcmp(arg, "-M")) {
2593 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2594 return -1;
2595 options->detect_rename = DIFF_DETECT_RENAME;
2597 else if (!prefixcmp(arg, "-C")) {
2598 if (options->detect_rename == DIFF_DETECT_COPY)
2599 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2600 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2601 return -1;
2602 options->detect_rename = DIFF_DETECT_COPY;
2604 else if (!strcmp(arg, "--no-renames"))
2605 options->detect_rename = 0;
2606 else if (!strcmp(arg, "--relative"))
2607 DIFF_OPT_SET(options, RELATIVE_NAME);
2608 else if (!prefixcmp(arg, "--relative=")) {
2609 DIFF_OPT_SET(options, RELATIVE_NAME);
2610 options->prefix = arg + 11;
2613 /* xdiff options */
2614 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2615 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2616 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2617 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2618 else if (!strcmp(arg, "--ignore-space-at-eol"))
2619 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2621 /* flags options */
2622 else if (!strcmp(arg, "--binary")) {
2623 options->output_format |= DIFF_FORMAT_PATCH;
2624 DIFF_OPT_SET(options, BINARY);
2626 else if (!strcmp(arg, "--full-index"))
2627 DIFF_OPT_SET(options, FULL_INDEX);
2628 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2629 DIFF_OPT_SET(options, TEXT);
2630 else if (!strcmp(arg, "-R"))
2631 DIFF_OPT_SET(options, REVERSE_DIFF);
2632 else if (!strcmp(arg, "--find-copies-harder"))
2633 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2634 else if (!strcmp(arg, "--follow"))
2635 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2636 else if (!strcmp(arg, "--color"))
2637 DIFF_OPT_SET(options, COLOR_DIFF);
2638 else if (!strcmp(arg, "--no-color"))
2639 DIFF_OPT_CLR(options, COLOR_DIFF);
2640 else if (!strcmp(arg, "--color-words"))
2641 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2642 else if (!strcmp(arg, "--exit-code"))
2643 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2644 else if (!strcmp(arg, "--quiet"))
2645 DIFF_OPT_SET(options, QUIET);
2646 else if (!strcmp(arg, "--ext-diff"))
2647 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2648 else if (!strcmp(arg, "--no-ext-diff"))
2649 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2650 else if (!strcmp(arg, "--ignore-submodules"))
2651 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2653 /* misc options */
2654 else if (!strcmp(arg, "-z"))
2655 options->line_termination = 0;
2656 else if (!prefixcmp(arg, "-l"))
2657 options->rename_limit = strtoul(arg+2, NULL, 10);
2658 else if (!prefixcmp(arg, "-S"))
2659 options->pickaxe = arg + 2;
2660 else if (!strcmp(arg, "--pickaxe-all"))
2661 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2662 else if (!strcmp(arg, "--pickaxe-regex"))
2663 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2664 else if (!prefixcmp(arg, "-O"))
2665 options->orderfile = arg + 2;
2666 else if (!prefixcmp(arg, "--diff-filter="))
2667 options->filter = arg + 14;
2668 else if (!strcmp(arg, "--abbrev"))
2669 options->abbrev = DEFAULT_ABBREV;
2670 else if (!prefixcmp(arg, "--abbrev=")) {
2671 options->abbrev = strtoul(arg + 9, NULL, 10);
2672 if (options->abbrev < MINIMUM_ABBREV)
2673 options->abbrev = MINIMUM_ABBREV;
2674 else if (40 < options->abbrev)
2675 options->abbrev = 40;
2677 else if (!prefixcmp(arg, "--src-prefix="))
2678 options->a_prefix = arg + 13;
2679 else if (!prefixcmp(arg, "--dst-prefix="))
2680 options->b_prefix = arg + 13;
2681 else if (!strcmp(arg, "--no-prefix"))
2682 options->a_prefix = options->b_prefix = "";
2683 else if (!prefixcmp(arg, "--output=")) {
2684 options->file = fopen(arg + strlen("--output="), "w");
2685 options->close_file = 1;
2686 } else
2687 return 0;
2688 return 1;
2691 static int parse_num(const char **cp_p)
2693 unsigned long num, scale;
2694 int ch, dot;
2695 const char *cp = *cp_p;
2697 num = 0;
2698 scale = 1;
2699 dot = 0;
2700 for(;;) {
2701 ch = *cp;
2702 if ( !dot && ch == '.' ) {
2703 scale = 1;
2704 dot = 1;
2705 } else if ( ch == '%' ) {
2706 scale = dot ? scale*100 : 100;
2707 cp++; /* % is always at the end */
2708 break;
2709 } else if ( ch >= '0' && ch <= '9' ) {
2710 if ( scale < 100000 ) {
2711 scale *= 10;
2712 num = (num*10) + (ch-'0');
2714 } else {
2715 break;
2717 cp++;
2719 *cp_p = cp;
2721 /* user says num divided by scale and we say internally that
2722 * is MAX_SCORE * num / scale.
2724 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2727 static int diff_scoreopt_parse(const char *opt)
2729 int opt1, opt2, cmd;
2731 if (*opt++ != '-')
2732 return -1;
2733 cmd = *opt++;
2734 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2735 return -1; /* that is not a -M, -C nor -B option */
2737 opt1 = parse_num(&opt);
2738 if (cmd != 'B')
2739 opt2 = 0;
2740 else {
2741 if (*opt == 0)
2742 opt2 = 0;
2743 else if (*opt != '/')
2744 return -1; /* we expect -B80/99 or -B80 */
2745 else {
2746 opt++;
2747 opt2 = parse_num(&opt);
2750 if (*opt != 0)
2751 return -1;
2752 return opt1 | (opt2 << 16);
2755 struct diff_queue_struct diff_queued_diff;
2757 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2759 if (queue->alloc <= queue->nr) {
2760 queue->alloc = alloc_nr(queue->alloc);
2761 queue->queue = xrealloc(queue->queue,
2762 sizeof(dp) * queue->alloc);
2764 queue->queue[queue->nr++] = dp;
2767 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2768 struct diff_filespec *one,
2769 struct diff_filespec *two)
2771 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2772 dp->one = one;
2773 dp->two = two;
2774 if (queue)
2775 diff_q(queue, dp);
2776 return dp;
2779 void diff_free_filepair(struct diff_filepair *p)
2781 free_filespec(p->one);
2782 free_filespec(p->two);
2783 free(p);
2786 /* This is different from find_unique_abbrev() in that
2787 * it stuffs the result with dots for alignment.
2789 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2791 int abblen;
2792 const char *abbrev;
2793 if (len == 40)
2794 return sha1_to_hex(sha1);
2796 abbrev = find_unique_abbrev(sha1, len);
2797 abblen = strlen(abbrev);
2798 if (abblen < 37) {
2799 static char hex[41];
2800 if (len < abblen && abblen <= len + 2)
2801 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2802 else
2803 sprintf(hex, "%s...", abbrev);
2804 return hex;
2806 return sha1_to_hex(sha1);
2809 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2811 int line_termination = opt->line_termination;
2812 int inter_name_termination = line_termination ? '\t' : '\0';
2814 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2815 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2816 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2817 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2819 if (p->score) {
2820 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2821 inter_name_termination);
2822 } else {
2823 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2826 if (p->status == DIFF_STATUS_COPIED ||
2827 p->status == DIFF_STATUS_RENAMED) {
2828 const char *name_a, *name_b;
2829 name_a = p->one->path;
2830 name_b = p->two->path;
2831 strip_prefix(opt->prefix_length, &name_a, &name_b);
2832 write_name_quoted(name_a, opt->file, inter_name_termination);
2833 write_name_quoted(name_b, opt->file, line_termination);
2834 } else {
2835 const char *name_a, *name_b;
2836 name_a = p->one->mode ? p->one->path : p->two->path;
2837 name_b = NULL;
2838 strip_prefix(opt->prefix_length, &name_a, &name_b);
2839 write_name_quoted(name_a, opt->file, line_termination);
2843 int diff_unmodified_pair(struct diff_filepair *p)
2845 /* This function is written stricter than necessary to support
2846 * the currently implemented transformers, but the idea is to
2847 * let transformers to produce diff_filepairs any way they want,
2848 * and filter and clean them up here before producing the output.
2850 struct diff_filespec *one = p->one, *two = p->two;
2852 if (DIFF_PAIR_UNMERGED(p))
2853 return 0; /* unmerged is interesting */
2855 /* deletion, addition, mode or type change
2856 * and rename are all interesting.
2858 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2859 DIFF_PAIR_MODE_CHANGED(p) ||
2860 strcmp(one->path, two->path))
2861 return 0;
2863 /* both are valid and point at the same path. that is, we are
2864 * dealing with a change.
2866 if (one->sha1_valid && two->sha1_valid &&
2867 !hashcmp(one->sha1, two->sha1))
2868 return 1; /* no change */
2869 if (!one->sha1_valid && !two->sha1_valid)
2870 return 1; /* both look at the same file on the filesystem. */
2871 return 0;
2874 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2876 if (diff_unmodified_pair(p))
2877 return;
2879 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2880 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2881 return; /* no tree diffs in patch format */
2883 run_diff(p, o);
2886 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2887 struct diffstat_t *diffstat)
2889 if (diff_unmodified_pair(p))
2890 return;
2892 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2893 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2894 return; /* no tree diffs in patch format */
2896 run_diffstat(p, o, diffstat);
2899 static void diff_flush_checkdiff(struct diff_filepair *p,
2900 struct diff_options *o)
2902 if (diff_unmodified_pair(p))
2903 return;
2905 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2906 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2907 return; /* no tree diffs in patch format */
2909 run_checkdiff(p, o);
2912 int diff_queue_is_empty(void)
2914 struct diff_queue_struct *q = &diff_queued_diff;
2915 int i;
2916 for (i = 0; i < q->nr; i++)
2917 if (!diff_unmodified_pair(q->queue[i]))
2918 return 0;
2919 return 1;
2922 #if DIFF_DEBUG
2923 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2925 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2926 x, one ? one : "",
2927 s->path,
2928 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2929 s->mode,
2930 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2931 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2932 x, one ? one : "",
2933 s->size, s->xfrm_flags);
2936 void diff_debug_filepair(const struct diff_filepair *p, int i)
2938 diff_debug_filespec(p->one, i, "one");
2939 diff_debug_filespec(p->two, i, "two");
2940 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2941 p->score, p->status ? p->status : '?',
2942 p->one->rename_used, p->broken_pair);
2945 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2947 int i;
2948 if (msg)
2949 fprintf(stderr, "%s\n", msg);
2950 fprintf(stderr, "q->nr = %d\n", q->nr);
2951 for (i = 0; i < q->nr; i++) {
2952 struct diff_filepair *p = q->queue[i];
2953 diff_debug_filepair(p, i);
2956 #endif
2958 static void diff_resolve_rename_copy(void)
2960 int i;
2961 struct diff_filepair *p;
2962 struct diff_queue_struct *q = &diff_queued_diff;
2964 diff_debug_queue("resolve-rename-copy", q);
2966 for (i = 0; i < q->nr; i++) {
2967 p = q->queue[i];
2968 p->status = 0; /* undecided */
2969 if (DIFF_PAIR_UNMERGED(p))
2970 p->status = DIFF_STATUS_UNMERGED;
2971 else if (!DIFF_FILE_VALID(p->one))
2972 p->status = DIFF_STATUS_ADDED;
2973 else if (!DIFF_FILE_VALID(p->two))
2974 p->status = DIFF_STATUS_DELETED;
2975 else if (DIFF_PAIR_TYPE_CHANGED(p))
2976 p->status = DIFF_STATUS_TYPE_CHANGED;
2978 /* from this point on, we are dealing with a pair
2979 * whose both sides are valid and of the same type, i.e.
2980 * either in-place edit or rename/copy edit.
2982 else if (DIFF_PAIR_RENAME(p)) {
2984 * A rename might have re-connected a broken
2985 * pair up, causing the pathnames to be the
2986 * same again. If so, that's not a rename at
2987 * all, just a modification..
2989 * Otherwise, see if this source was used for
2990 * multiple renames, in which case we decrement
2991 * the count, and call it a copy.
2993 if (!strcmp(p->one->path, p->two->path))
2994 p->status = DIFF_STATUS_MODIFIED;
2995 else if (--p->one->rename_used > 0)
2996 p->status = DIFF_STATUS_COPIED;
2997 else
2998 p->status = DIFF_STATUS_RENAMED;
3000 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3001 p->one->mode != p->two->mode ||
3002 is_null_sha1(p->one->sha1))
3003 p->status = DIFF_STATUS_MODIFIED;
3004 else {
3005 /* This is a "no-change" entry and should not
3006 * happen anymore, but prepare for broken callers.
3008 error("feeding unmodified %s to diffcore",
3009 p->one->path);
3010 p->status = DIFF_STATUS_UNKNOWN;
3013 diff_debug_queue("resolve-rename-copy done", q);
3016 static int check_pair_status(struct diff_filepair *p)
3018 switch (p->status) {
3019 case DIFF_STATUS_UNKNOWN:
3020 return 0;
3021 case 0:
3022 die("internal error in diff-resolve-rename-copy");
3023 default:
3024 return 1;
3028 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3030 int fmt = opt->output_format;
3032 if (fmt & DIFF_FORMAT_CHECKDIFF)
3033 diff_flush_checkdiff(p, opt);
3034 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3035 diff_flush_raw(p, opt);
3036 else if (fmt & DIFF_FORMAT_NAME) {
3037 const char *name_a, *name_b;
3038 name_a = p->two->path;
3039 name_b = NULL;
3040 strip_prefix(opt->prefix_length, &name_a, &name_b);
3041 write_name_quoted(name_a, opt->file, opt->line_termination);
3045 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3047 if (fs->mode)
3048 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3049 else
3050 fprintf(file, " %s ", newdelete);
3051 write_name_quoted(fs->path, file, '\n');
3055 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3057 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3058 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3059 show_name ? ' ' : '\n');
3060 if (show_name) {
3061 write_name_quoted(p->two->path, file, '\n');
3066 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3068 char *names = pprint_rename(p->one->path, p->two->path);
3070 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3071 free(names);
3072 show_mode_change(file, p, 0);
3075 static void diff_summary(FILE *file, struct diff_filepair *p)
3077 switch(p->status) {
3078 case DIFF_STATUS_DELETED:
3079 show_file_mode_name(file, "delete", p->one);
3080 break;
3081 case DIFF_STATUS_ADDED:
3082 show_file_mode_name(file, "create", p->two);
3083 break;
3084 case DIFF_STATUS_COPIED:
3085 show_rename_copy(file, "copy", p);
3086 break;
3087 case DIFF_STATUS_RENAMED:
3088 show_rename_copy(file, "rename", p);
3089 break;
3090 default:
3091 if (p->score) {
3092 fputs(" rewrite ", file);
3093 write_name_quoted(p->two->path, file, ' ');
3094 fprintf(file, "(%d%%)\n", similarity_index(p));
3096 show_mode_change(file, p, !p->score);
3097 break;
3101 struct patch_id_t {
3102 struct xdiff_emit_state xm;
3103 SHA_CTX *ctx;
3104 int patchlen;
3107 static int remove_space(char *line, int len)
3109 int i;
3110 char *dst = line;
3111 unsigned char c;
3113 for (i = 0; i < len; i++)
3114 if (!isspace((c = line[i])))
3115 *dst++ = c;
3117 return dst - line;
3120 static void patch_id_consume(void *priv, char *line, unsigned long len)
3122 struct patch_id_t *data = priv;
3123 int new_len;
3125 /* Ignore line numbers when computing the SHA1 of the patch */
3126 if (!prefixcmp(line, "@@ -"))
3127 return;
3129 new_len = remove_space(line, len);
3131 SHA1_Update(data->ctx, line, new_len);
3132 data->patchlen += new_len;
3135 /* returns 0 upon success, and writes result into sha1 */
3136 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3138 struct diff_queue_struct *q = &diff_queued_diff;
3139 int i;
3140 SHA_CTX ctx;
3141 struct patch_id_t data;
3142 char buffer[PATH_MAX * 4 + 20];
3144 SHA1_Init(&ctx);
3145 memset(&data, 0, sizeof(struct patch_id_t));
3146 data.ctx = &ctx;
3147 data.xm.consume = patch_id_consume;
3149 for (i = 0; i < q->nr; i++) {
3150 xpparam_t xpp;
3151 xdemitconf_t xecfg;
3152 xdemitcb_t ecb;
3153 mmfile_t mf1, mf2;
3154 struct diff_filepair *p = q->queue[i];
3155 int len1, len2;
3157 memset(&xecfg, 0, sizeof(xecfg));
3158 if (p->status == 0)
3159 return error("internal diff status error");
3160 if (p->status == DIFF_STATUS_UNKNOWN)
3161 continue;
3162 if (diff_unmodified_pair(p))
3163 continue;
3164 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3165 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3166 continue;
3167 if (DIFF_PAIR_UNMERGED(p))
3168 continue;
3170 diff_fill_sha1_info(p->one);
3171 diff_fill_sha1_info(p->two);
3172 if (fill_mmfile(&mf1, p->one) < 0 ||
3173 fill_mmfile(&mf2, p->two) < 0)
3174 return error("unable to read files to diff");
3176 len1 = remove_space(p->one->path, strlen(p->one->path));
3177 len2 = remove_space(p->two->path, strlen(p->two->path));
3178 if (p->one->mode == 0)
3179 len1 = snprintf(buffer, sizeof(buffer),
3180 "diff--gita/%.*sb/%.*s"
3181 "newfilemode%06o"
3182 "---/dev/null"
3183 "+++b/%.*s",
3184 len1, p->one->path,
3185 len2, p->two->path,
3186 p->two->mode,
3187 len2, p->two->path);
3188 else if (p->two->mode == 0)
3189 len1 = snprintf(buffer, sizeof(buffer),
3190 "diff--gita/%.*sb/%.*s"
3191 "deletedfilemode%06o"
3192 "---a/%.*s"
3193 "+++/dev/null",
3194 len1, p->one->path,
3195 len2, p->two->path,
3196 p->one->mode,
3197 len1, p->one->path);
3198 else
3199 len1 = snprintf(buffer, sizeof(buffer),
3200 "diff--gita/%.*sb/%.*s"
3201 "---a/%.*s"
3202 "+++b/%.*s",
3203 len1, p->one->path,
3204 len2, p->two->path,
3205 len1, p->one->path,
3206 len2, p->two->path);
3207 SHA1_Update(&ctx, buffer, len1);
3209 xpp.flags = XDF_NEED_MINIMAL;
3210 xecfg.ctxlen = 3;
3211 xecfg.flags = XDL_EMIT_FUNCNAMES;
3212 ecb.outf = xdiff_outf;
3213 ecb.priv = &data;
3214 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3217 SHA1_Final(sha1, &ctx);
3218 return 0;
3221 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3223 struct diff_queue_struct *q = &diff_queued_diff;
3224 int i;
3225 int result = diff_get_patch_id(options, sha1);
3227 for (i = 0; i < q->nr; i++)
3228 diff_free_filepair(q->queue[i]);
3230 free(q->queue);
3231 q->queue = NULL;
3232 q->nr = q->alloc = 0;
3234 return result;
3237 static int is_summary_empty(const struct diff_queue_struct *q)
3239 int i;
3241 for (i = 0; i < q->nr; i++) {
3242 const struct diff_filepair *p = q->queue[i];
3244 switch (p->status) {
3245 case DIFF_STATUS_DELETED:
3246 case DIFF_STATUS_ADDED:
3247 case DIFF_STATUS_COPIED:
3248 case DIFF_STATUS_RENAMED:
3249 return 0;
3250 default:
3251 if (p->score)
3252 return 0;
3253 if (p->one->mode && p->two->mode &&
3254 p->one->mode != p->two->mode)
3255 return 0;
3256 break;
3259 return 1;
3262 void diff_flush(struct diff_options *options)
3264 struct diff_queue_struct *q = &diff_queued_diff;
3265 int i, output_format = options->output_format;
3266 int separator = 0;
3269 * Order: raw, stat, summary, patch
3270 * or: name/name-status/checkdiff (other bits clear)
3272 if (!q->nr)
3273 goto free_queue;
3275 if (output_format & (DIFF_FORMAT_RAW |
3276 DIFF_FORMAT_NAME |
3277 DIFF_FORMAT_NAME_STATUS |
3278 DIFF_FORMAT_CHECKDIFF)) {
3279 for (i = 0; i < q->nr; i++) {
3280 struct diff_filepair *p = q->queue[i];
3281 if (check_pair_status(p))
3282 flush_one_pair(p, options);
3284 separator++;
3287 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3288 struct diffstat_t diffstat;
3290 memset(&diffstat, 0, sizeof(struct diffstat_t));
3291 diffstat.xm.consume = diffstat_consume;
3292 for (i = 0; i < q->nr; i++) {
3293 struct diff_filepair *p = q->queue[i];
3294 if (check_pair_status(p))
3295 diff_flush_stat(p, options, &diffstat);
3297 if (output_format & DIFF_FORMAT_NUMSTAT)
3298 show_numstat(&diffstat, options);
3299 if (output_format & DIFF_FORMAT_DIFFSTAT)
3300 show_stats(&diffstat, options);
3301 if (output_format & DIFF_FORMAT_SHORTSTAT)
3302 show_shortstats(&diffstat, options);
3303 free_diffstat_info(&diffstat);
3304 separator++;
3306 if (output_format & DIFF_FORMAT_DIRSTAT)
3307 show_dirstat(options);
3309 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3310 for (i = 0; i < q->nr; i++)
3311 diff_summary(options->file, q->queue[i]);
3312 separator++;
3315 if (output_format & DIFF_FORMAT_PATCH) {
3316 if (separator) {
3317 putc(options->line_termination, options->file);
3318 if (options->stat_sep) {
3319 /* attach patch instead of inline */
3320 fputs(options->stat_sep, options->file);
3324 for (i = 0; i < q->nr; i++) {
3325 struct diff_filepair *p = q->queue[i];
3326 if (check_pair_status(p))
3327 diff_flush_patch(p, options);
3331 if (output_format & DIFF_FORMAT_CALLBACK)
3332 options->format_callback(q, options, options->format_callback_data);
3334 for (i = 0; i < q->nr; i++)
3335 diff_free_filepair(q->queue[i]);
3336 free_queue:
3337 free(q->queue);
3338 q->queue = NULL;
3339 q->nr = q->alloc = 0;
3340 if (options->close_file)
3341 fclose(options->file);
3344 static void diffcore_apply_filter(const char *filter)
3346 int i;
3347 struct diff_queue_struct *q = &diff_queued_diff;
3348 struct diff_queue_struct outq;
3349 outq.queue = NULL;
3350 outq.nr = outq.alloc = 0;
3352 if (!filter)
3353 return;
3355 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3356 int found;
3357 for (i = found = 0; !found && i < q->nr; i++) {
3358 struct diff_filepair *p = q->queue[i];
3359 if (((p->status == DIFF_STATUS_MODIFIED) &&
3360 ((p->score &&
3361 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3362 (!p->score &&
3363 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3364 ((p->status != DIFF_STATUS_MODIFIED) &&
3365 strchr(filter, p->status)))
3366 found++;
3368 if (found)
3369 return;
3371 /* otherwise we will clear the whole queue
3372 * by copying the empty outq at the end of this
3373 * function, but first clear the current entries
3374 * in the queue.
3376 for (i = 0; i < q->nr; i++)
3377 diff_free_filepair(q->queue[i]);
3379 else {
3380 /* Only the matching ones */
3381 for (i = 0; i < q->nr; i++) {
3382 struct diff_filepair *p = q->queue[i];
3384 if (((p->status == DIFF_STATUS_MODIFIED) &&
3385 ((p->score &&
3386 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3387 (!p->score &&
3388 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3389 ((p->status != DIFF_STATUS_MODIFIED) &&
3390 strchr(filter, p->status)))
3391 diff_q(&outq, p);
3392 else
3393 diff_free_filepair(p);
3396 free(q->queue);
3397 *q = outq;
3400 /* Check whether two filespecs with the same mode and size are identical */
3401 static int diff_filespec_is_identical(struct diff_filespec *one,
3402 struct diff_filespec *two)
3404 if (S_ISGITLINK(one->mode))
3405 return 0;
3406 if (diff_populate_filespec(one, 0))
3407 return 0;
3408 if (diff_populate_filespec(two, 0))
3409 return 0;
3410 return !memcmp(one->data, two->data, one->size);
3413 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3415 int i;
3416 struct diff_queue_struct *q = &diff_queued_diff;
3417 struct diff_queue_struct outq;
3418 outq.queue = NULL;
3419 outq.nr = outq.alloc = 0;
3421 for (i = 0; i < q->nr; i++) {
3422 struct diff_filepair *p = q->queue[i];
3425 * 1. Entries that come from stat info dirtyness
3426 * always have both sides (iow, not create/delete),
3427 * one side of the object name is unknown, with
3428 * the same mode and size. Keep the ones that
3429 * do not match these criteria. They have real
3430 * differences.
3432 * 2. At this point, the file is known to be modified,
3433 * with the same mode and size, and the object
3434 * name of one side is unknown. Need to inspect
3435 * the identical contents.
3437 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3438 !DIFF_FILE_VALID(p->two) ||
3439 (p->one->sha1_valid && p->two->sha1_valid) ||
3440 (p->one->mode != p->two->mode) ||
3441 diff_populate_filespec(p->one, 1) ||
3442 diff_populate_filespec(p->two, 1) ||
3443 (p->one->size != p->two->size) ||
3444 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3445 diff_q(&outq, p);
3446 else {
3448 * The caller can subtract 1 from skip_stat_unmatch
3449 * to determine how many paths were dirty only
3450 * due to stat info mismatch.
3452 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3453 diffopt->skip_stat_unmatch++;
3454 diff_free_filepair(p);
3457 free(q->queue);
3458 *q = outq;
3461 void diffcore_std(struct diff_options *options)
3463 if (options->skip_stat_unmatch)
3464 diffcore_skip_stat_unmatch(options);
3465 if (options->break_opt != -1)
3466 diffcore_break(options->break_opt);
3467 if (options->detect_rename)
3468 diffcore_rename(options);
3469 if (options->break_opt != -1)
3470 diffcore_merge_broken();
3471 if (options->pickaxe)
3472 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3473 if (options->orderfile)
3474 diffcore_order(options->orderfile);
3475 diff_resolve_rename_copy();
3476 diffcore_apply_filter(options->filter);
3478 if (diff_queued_diff.nr)
3479 DIFF_OPT_SET(options, HAS_CHANGES);
3480 else
3481 DIFF_OPT_CLR(options, HAS_CHANGES);
3484 int diff_result_code(struct diff_options *opt, int status)
3486 int result = 0;
3487 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3488 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3489 return status;
3490 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3491 DIFF_OPT_TST(opt, HAS_CHANGES))
3492 result |= 01;
3493 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3494 DIFF_OPT_TST(opt, CHECK_FAILED))
3495 result |= 02;
3496 return result;
3499 void diff_addremove(struct diff_options *options,
3500 int addremove, unsigned mode,
3501 const unsigned char *sha1,
3502 const char *concatpath)
3504 struct diff_filespec *one, *two;
3506 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3507 return;
3509 /* This may look odd, but it is a preparation for
3510 * feeding "there are unchanged files which should
3511 * not produce diffs, but when you are doing copy
3512 * detection you would need them, so here they are"
3513 * entries to the diff-core. They will be prefixed
3514 * with something like '=' or '*' (I haven't decided
3515 * which but should not make any difference).
3516 * Feeding the same new and old to diff_change()
3517 * also has the same effect.
3518 * Before the final output happens, they are pruned after
3519 * merged into rename/copy pairs as appropriate.
3521 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3522 addremove = (addremove == '+' ? '-' :
3523 addremove == '-' ? '+' : addremove);
3525 if (options->prefix &&
3526 strncmp(concatpath, options->prefix, options->prefix_length))
3527 return;
3529 one = alloc_filespec(concatpath);
3530 two = alloc_filespec(concatpath);
3532 if (addremove != '+')
3533 fill_filespec(one, sha1, mode);
3534 if (addremove != '-')
3535 fill_filespec(two, sha1, mode);
3537 diff_queue(&diff_queued_diff, one, two);
3538 DIFF_OPT_SET(options, HAS_CHANGES);
3541 void diff_change(struct diff_options *options,
3542 unsigned old_mode, unsigned new_mode,
3543 const unsigned char *old_sha1,
3544 const unsigned char *new_sha1,
3545 const char *concatpath)
3547 struct diff_filespec *one, *two;
3549 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3550 && S_ISGITLINK(new_mode))
3551 return;
3553 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3554 unsigned tmp;
3555 const unsigned char *tmp_c;
3556 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3557 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3560 if (options->prefix &&
3561 strncmp(concatpath, options->prefix, options->prefix_length))
3562 return;
3564 one = alloc_filespec(concatpath);
3565 two = alloc_filespec(concatpath);
3566 fill_filespec(one, old_sha1, old_mode);
3567 fill_filespec(two, new_sha1, new_mode);
3569 diff_queue(&diff_queued_diff, one, two);
3570 DIFF_OPT_SET(options, HAS_CHANGES);
3573 void diff_unmerge(struct diff_options *options,
3574 const char *path,
3575 unsigned mode, const unsigned char *sha1)
3577 struct diff_filespec *one, *two;
3579 if (options->prefix &&
3580 strncmp(path, options->prefix, options->prefix_length))
3581 return;
3583 one = alloc_filespec(path);
3584 two = alloc_filespec(path);
3585 fill_filespec(one, sha1, mode);
3586 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;