mergetool: respect autocrlf by using checkout-index
[git/debian.git] / diff.c
blob073131316041fa25790c1da27c0a526d1c1b3d66
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"
14 #include "userdiff.h"
16 #ifdef NO_FAST_WORKING_DIRECTORY
17 #define FAST_WORKING_DIRECTORY 0
18 #else
19 #define FAST_WORKING_DIRECTORY 1
20 #endif
22 static int diff_detect_rename_default;
23 static int diff_rename_limit_default = 200;
24 static int diff_suppress_blank_empty;
25 int diff_use_color_default = -1;
26 static const char *external_diff_cmd_cfg;
27 int diff_auto_refresh_index = 1;
28 static int diff_mnemonic_prefix;
30 static char diff_colors[][COLOR_MAXLEN] = {
31 "\033[m", /* reset */
32 "", /* PLAIN (normal) */
33 "\033[1m", /* METAINFO (bold) */
34 "\033[36m", /* FRAGINFO (cyan) */
35 "\033[31m", /* OLD (red) */
36 "\033[32m", /* NEW (green) */
37 "\033[33m", /* COMMIT (yellow) */
38 "\033[41m", /* WHITESPACE (red background) */
41 static void diff_filespec_load_driver(struct diff_filespec *one);
42 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
44 static int parse_diff_color_slot(const char *var, int ofs)
46 if (!strcasecmp(var+ofs, "plain"))
47 return DIFF_PLAIN;
48 if (!strcasecmp(var+ofs, "meta"))
49 return DIFF_METAINFO;
50 if (!strcasecmp(var+ofs, "frag"))
51 return DIFF_FRAGINFO;
52 if (!strcasecmp(var+ofs, "old"))
53 return DIFF_FILE_OLD;
54 if (!strcasecmp(var+ofs, "new"))
55 return DIFF_FILE_NEW;
56 if (!strcasecmp(var+ofs, "commit"))
57 return DIFF_COMMIT;
58 if (!strcasecmp(var+ofs, "whitespace"))
59 return DIFF_WHITESPACE;
60 die("bad config variable '%s'", var);
64 * These are to give UI layer defaults.
65 * The core-level commands such as git-diff-files should
66 * never be affected by the setting of diff.renames
67 * the user happens to have in the configuration file.
69 int git_diff_ui_config(const char *var, const char *value, void *cb)
71 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
72 diff_use_color_default = git_config_colorbool(var, value, -1);
73 return 0;
75 if (!strcmp(var, "diff.renames")) {
76 if (!value)
77 diff_detect_rename_default = DIFF_DETECT_RENAME;
78 else if (!strcasecmp(value, "copies") ||
79 !strcasecmp(value, "copy"))
80 diff_detect_rename_default = DIFF_DETECT_COPY;
81 else if (git_config_bool(var,value))
82 diff_detect_rename_default = DIFF_DETECT_RENAME;
83 return 0;
85 if (!strcmp(var, "diff.autorefreshindex")) {
86 diff_auto_refresh_index = git_config_bool(var, value);
87 return 0;
89 if (!strcmp(var, "diff.mnemonicprefix")) {
90 diff_mnemonic_prefix = git_config_bool(var, value);
91 return 0;
93 if (!strcmp(var, "diff.external"))
94 return git_config_string(&external_diff_cmd_cfg, var, value);
96 return git_diff_basic_config(var, value, cb);
99 int git_diff_basic_config(const char *var, const char *value, void *cb)
101 if (!strcmp(var, "diff.renamelimit")) {
102 diff_rename_limit_default = git_config_int(var, value);
103 return 0;
106 switch (userdiff_config(var, value)) {
107 case 0: break;
108 case -1: return -1;
109 default: return 0;
112 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
113 int slot = parse_diff_color_slot(var, 11);
114 if (!value)
115 return config_error_nonbool(var);
116 color_parse(value, var, diff_colors[slot]);
117 return 0;
120 /* like GNU diff's --suppress-blank-empty option */
121 if (!strcmp(var, "diff.suppressblankempty") ||
122 /* for backwards compatibility */
123 !strcmp(var, "diff.suppress-blank-empty")) {
124 diff_suppress_blank_empty = git_config_bool(var, value);
125 return 0;
128 return git_color_default_config(var, value, cb);
131 static char *quote_two(const char *one, const char *two)
133 int need_one = quote_c_style(one, NULL, NULL, 1);
134 int need_two = quote_c_style(two, NULL, NULL, 1);
135 struct strbuf res = STRBUF_INIT;
137 if (need_one + need_two) {
138 strbuf_addch(&res, '"');
139 quote_c_style(one, &res, NULL, 1);
140 quote_c_style(two, &res, NULL, 1);
141 strbuf_addch(&res, '"');
142 } else {
143 strbuf_addstr(&res, one);
144 strbuf_addstr(&res, two);
146 return strbuf_detach(&res, NULL);
149 static const char *external_diff(void)
151 static const char *external_diff_cmd = NULL;
152 static int done_preparing = 0;
154 if (done_preparing)
155 return external_diff_cmd;
156 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
157 if (!external_diff_cmd)
158 external_diff_cmd = external_diff_cmd_cfg;
159 done_preparing = 1;
160 return external_diff_cmd;
163 static struct diff_tempfile {
164 const char *name; /* filename external diff should read from */
165 char hex[41];
166 char mode[10];
167 char tmp_path[PATH_MAX];
168 } diff_temp[2];
170 static int count_lines(const char *data, int size)
172 int count, ch, completely_empty = 1, nl_just_seen = 0;
173 count = 0;
174 while (0 < size--) {
175 ch = *data++;
176 if (ch == '\n') {
177 count++;
178 nl_just_seen = 1;
179 completely_empty = 0;
181 else {
182 nl_just_seen = 0;
183 completely_empty = 0;
186 if (completely_empty)
187 return 0;
188 if (!nl_just_seen)
189 count++; /* no trailing newline */
190 return count;
193 static void print_line_count(FILE *file, int count)
195 switch (count) {
196 case 0:
197 fprintf(file, "0,0");
198 break;
199 case 1:
200 fprintf(file, "1");
201 break;
202 default:
203 fprintf(file, "1,%d", count);
204 break;
208 static void copy_file_with_prefix(FILE *file,
209 int prefix, const char *data, int size,
210 const char *set, const char *reset)
212 int ch, nl_just_seen = 1;
213 while (0 < size--) {
214 ch = *data++;
215 if (nl_just_seen) {
216 fputs(set, file);
217 putc(prefix, file);
219 if (ch == '\n') {
220 nl_just_seen = 1;
221 fputs(reset, file);
222 } else
223 nl_just_seen = 0;
224 putc(ch, file);
226 if (!nl_just_seen)
227 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
230 static void emit_rewrite_diff(const char *name_a,
231 const char *name_b,
232 struct diff_filespec *one,
233 struct diff_filespec *two,
234 const char *textconv_one,
235 const char *textconv_two,
236 struct diff_options *o)
238 int lc_a, lc_b;
239 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
240 const char *name_a_tab, *name_b_tab;
241 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
242 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
243 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
244 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
245 const char *reset = diff_get_color(color_diff, DIFF_RESET);
246 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
247 const char *a_prefix, *b_prefix;
248 const char *data_one, *data_two;
249 size_t size_one, size_two;
251 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
252 a_prefix = o->b_prefix;
253 b_prefix = o->a_prefix;
254 } else {
255 a_prefix = o->a_prefix;
256 b_prefix = o->b_prefix;
259 name_a += (*name_a == '/');
260 name_b += (*name_b == '/');
261 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
262 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
264 strbuf_reset(&a_name);
265 strbuf_reset(&b_name);
266 quote_two_c_style(&a_name, a_prefix, name_a, 0);
267 quote_two_c_style(&b_name, b_prefix, name_b, 0);
269 diff_populate_filespec(one, 0);
270 diff_populate_filespec(two, 0);
271 if (textconv_one) {
272 data_one = run_textconv(textconv_one, one, &size_one);
273 if (!data_one)
274 die("unable to read files to diff");
276 else {
277 data_one = one->data;
278 size_one = one->size;
280 if (textconv_two) {
281 data_two = run_textconv(textconv_two, two, &size_two);
282 if (!data_two)
283 die("unable to read files to diff");
285 else {
286 data_two = two->data;
287 size_two = two->size;
290 lc_a = count_lines(data_one, size_one);
291 lc_b = count_lines(data_two, size_two);
292 fprintf(o->file,
293 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
294 metainfo, a_name.buf, name_a_tab, reset,
295 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
296 print_line_count(o->file, lc_a);
297 fprintf(o->file, " +");
298 print_line_count(o->file, lc_b);
299 fprintf(o->file, " @@%s\n", reset);
300 if (lc_a)
301 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
302 if (lc_b)
303 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
306 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
308 if (!DIFF_FILE_VALID(one)) {
309 mf->ptr = (char *)""; /* does not matter */
310 mf->size = 0;
311 return 0;
313 else if (diff_populate_filespec(one, 0))
314 return -1;
316 mf->ptr = one->data;
317 mf->size = one->size;
318 return 0;
321 struct diff_words_buffer {
322 mmfile_t text;
323 long alloc;
324 long current; /* output pointer */
325 int suppressed_newline;
328 static void diff_words_append(char *line, unsigned long len,
329 struct diff_words_buffer *buffer)
331 if (buffer->text.size + len > buffer->alloc) {
332 buffer->alloc = (buffer->text.size + len) * 3 / 2;
333 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
335 line++;
336 len--;
337 memcpy(buffer->text.ptr + buffer->text.size, line, len);
338 buffer->text.size += len;
341 struct diff_words_data {
342 struct diff_words_buffer minus, plus;
343 FILE *file;
346 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
347 int suppress_newline)
349 const char *ptr;
350 int eol = 0;
352 if (len == 0)
353 return;
355 ptr = buffer->text.ptr + buffer->current;
356 buffer->current += len;
358 if (ptr[len - 1] == '\n') {
359 eol = 1;
360 len--;
363 fputs(diff_get_color(1, color), file);
364 fwrite(ptr, len, 1, file);
365 fputs(diff_get_color(1, DIFF_RESET), file);
367 if (eol) {
368 if (suppress_newline)
369 buffer->suppressed_newline = 1;
370 else
371 putc('\n', file);
375 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
377 struct diff_words_data *diff_words = priv;
379 if (diff_words->minus.suppressed_newline) {
380 if (line[0] != '+')
381 putc('\n', diff_words->file);
382 diff_words->minus.suppressed_newline = 0;
385 len--;
386 switch (line[0]) {
387 case '-':
388 print_word(diff_words->file,
389 &diff_words->minus, len, DIFF_FILE_OLD, 1);
390 break;
391 case '+':
392 print_word(diff_words->file,
393 &diff_words->plus, len, DIFF_FILE_NEW, 0);
394 break;
395 case ' ':
396 print_word(diff_words->file,
397 &diff_words->plus, len, DIFF_PLAIN, 0);
398 diff_words->minus.current += len;
399 break;
403 /* this executes the word diff on the accumulated buffers */
404 static void diff_words_show(struct diff_words_data *diff_words)
406 xpparam_t xpp;
407 xdemitconf_t xecfg;
408 xdemitcb_t ecb;
409 mmfile_t minus, plus;
410 int i;
412 memset(&xpp, 0, sizeof(xpp));
413 memset(&xecfg, 0, sizeof(xecfg));
414 minus.size = diff_words->minus.text.size;
415 minus.ptr = xmalloc(minus.size);
416 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
417 for (i = 0; i < minus.size; i++)
418 if (isspace(minus.ptr[i]))
419 minus.ptr[i] = '\n';
420 diff_words->minus.current = 0;
422 plus.size = diff_words->plus.text.size;
423 plus.ptr = xmalloc(plus.size);
424 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
425 for (i = 0; i < plus.size; i++)
426 if (isspace(plus.ptr[i]))
427 plus.ptr[i] = '\n';
428 diff_words->plus.current = 0;
430 xpp.flags = XDF_NEED_MINIMAL;
431 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
432 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
433 &xpp, &xecfg, &ecb);
434 free(minus.ptr);
435 free(plus.ptr);
436 diff_words->minus.text.size = diff_words->plus.text.size = 0;
438 if (diff_words->minus.suppressed_newline) {
439 putc('\n', diff_words->file);
440 diff_words->minus.suppressed_newline = 0;
444 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
446 struct emit_callback {
447 int nparents, color_diff;
448 unsigned ws_rule;
449 sane_truncate_fn truncate;
450 const char **label_path;
451 struct diff_words_data *diff_words;
452 int *found_changesp;
453 FILE *file;
456 static void free_diff_words_data(struct emit_callback *ecbdata)
458 if (ecbdata->diff_words) {
459 /* flush buffers */
460 if (ecbdata->diff_words->minus.text.size ||
461 ecbdata->diff_words->plus.text.size)
462 diff_words_show(ecbdata->diff_words);
464 free (ecbdata->diff_words->minus.text.ptr);
465 free (ecbdata->diff_words->plus.text.ptr);
466 free(ecbdata->diff_words);
467 ecbdata->diff_words = NULL;
471 const char *diff_get_color(int diff_use_color, enum color_diff ix)
473 if (diff_use_color)
474 return diff_colors[ix];
475 return "";
478 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
480 int has_trailing_newline, has_trailing_carriage_return;
482 has_trailing_newline = (len > 0 && line[len-1] == '\n');
483 if (has_trailing_newline)
484 len--;
485 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
486 if (has_trailing_carriage_return)
487 len--;
489 fputs(set, file);
490 fwrite(line, len, 1, file);
491 fputs(reset, file);
492 if (has_trailing_carriage_return)
493 fputc('\r', file);
494 if (has_trailing_newline)
495 fputc('\n', file);
498 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
500 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
501 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
503 if (!*ws)
504 emit_line(ecbdata->file, set, reset, line, len);
505 else {
506 /* Emit just the prefix, then the rest. */
507 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
508 ws_check_emit(line + ecbdata->nparents,
509 len - ecbdata->nparents, ecbdata->ws_rule,
510 ecbdata->file, set, reset, ws);
514 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
516 const char *cp;
517 unsigned long allot;
518 size_t l = len;
520 if (ecb->truncate)
521 return ecb->truncate(line, len);
522 cp = line;
523 allot = l;
524 while (0 < l) {
525 (void) utf8_width(&cp, &l);
526 if (!cp)
527 break; /* truncated in the middle? */
529 return allot - l;
532 static void fn_out_consume(void *priv, char *line, unsigned long len)
534 int i;
535 int color;
536 struct emit_callback *ecbdata = priv;
537 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
538 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
539 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
541 *(ecbdata->found_changesp) = 1;
543 if (ecbdata->label_path[0]) {
544 const char *name_a_tab, *name_b_tab;
546 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
547 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
549 fprintf(ecbdata->file, "%s--- %s%s%s\n",
550 meta, ecbdata->label_path[0], reset, name_a_tab);
551 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
552 meta, ecbdata->label_path[1], reset, name_b_tab);
553 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
556 if (diff_suppress_blank_empty
557 && len == 2 && line[0] == ' ' && line[1] == '\n') {
558 line[0] = '\n';
559 len = 1;
562 /* This is not really necessary for now because
563 * this codepath only deals with two-way diffs.
565 for (i = 0; i < len && line[i] == '@'; i++)
567 if (2 <= i && i < len && line[i] == ' ') {
568 ecbdata->nparents = i - 1;
569 len = sane_truncate_line(ecbdata, line, len);
570 emit_line(ecbdata->file,
571 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
572 reset, line, len);
573 if (line[len-1] != '\n')
574 putc('\n', ecbdata->file);
575 return;
578 if (len < ecbdata->nparents) {
579 emit_line(ecbdata->file, reset, reset, line, len);
580 return;
583 color = DIFF_PLAIN;
584 if (ecbdata->diff_words && ecbdata->nparents != 1)
585 /* fall back to normal diff */
586 free_diff_words_data(ecbdata);
587 if (ecbdata->diff_words) {
588 if (line[0] == '-') {
589 diff_words_append(line, len,
590 &ecbdata->diff_words->minus);
591 return;
592 } else if (line[0] == '+') {
593 diff_words_append(line, len,
594 &ecbdata->diff_words->plus);
595 return;
597 if (ecbdata->diff_words->minus.text.size ||
598 ecbdata->diff_words->plus.text.size)
599 diff_words_show(ecbdata->diff_words);
600 line++;
601 len--;
602 emit_line(ecbdata->file, plain, reset, line, len);
603 return;
605 for (i = 0; i < ecbdata->nparents && len; i++) {
606 if (line[i] == '-')
607 color = DIFF_FILE_OLD;
608 else if (line[i] == '+')
609 color = DIFF_FILE_NEW;
612 if (color != DIFF_FILE_NEW) {
613 emit_line(ecbdata->file,
614 diff_get_color(ecbdata->color_diff, color),
615 reset, line, len);
616 return;
618 emit_add_line(reset, ecbdata, line, len);
621 static char *pprint_rename(const char *a, const char *b)
623 const char *old = a;
624 const char *new = b;
625 struct strbuf name = STRBUF_INIT;
626 int pfx_length, sfx_length;
627 int len_a = strlen(a);
628 int len_b = strlen(b);
629 int a_midlen, b_midlen;
630 int qlen_a = quote_c_style(a, NULL, NULL, 0);
631 int qlen_b = quote_c_style(b, NULL, NULL, 0);
633 if (qlen_a || qlen_b) {
634 quote_c_style(a, &name, NULL, 0);
635 strbuf_addstr(&name, " => ");
636 quote_c_style(b, &name, NULL, 0);
637 return strbuf_detach(&name, NULL);
640 /* Find common prefix */
641 pfx_length = 0;
642 while (*old && *new && *old == *new) {
643 if (*old == '/')
644 pfx_length = old - a + 1;
645 old++;
646 new++;
649 /* Find common suffix */
650 old = a + len_a;
651 new = b + len_b;
652 sfx_length = 0;
653 while (a <= old && b <= new && *old == *new) {
654 if (*old == '/')
655 sfx_length = len_a - (old - a);
656 old--;
657 new--;
661 * pfx{mid-a => mid-b}sfx
662 * {pfx-a => pfx-b}sfx
663 * pfx{sfx-a => sfx-b}
664 * name-a => name-b
666 a_midlen = len_a - pfx_length - sfx_length;
667 b_midlen = len_b - pfx_length - sfx_length;
668 if (a_midlen < 0)
669 a_midlen = 0;
670 if (b_midlen < 0)
671 b_midlen = 0;
673 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
674 if (pfx_length + sfx_length) {
675 strbuf_add(&name, a, pfx_length);
676 strbuf_addch(&name, '{');
678 strbuf_add(&name, a + pfx_length, a_midlen);
679 strbuf_addstr(&name, " => ");
680 strbuf_add(&name, b + pfx_length, b_midlen);
681 if (pfx_length + sfx_length) {
682 strbuf_addch(&name, '}');
683 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
685 return strbuf_detach(&name, NULL);
688 struct diffstat_t {
689 int nr;
690 int alloc;
691 struct diffstat_file {
692 char *from_name;
693 char *name;
694 char *print_name;
695 unsigned is_unmerged:1;
696 unsigned is_binary:1;
697 unsigned is_renamed:1;
698 unsigned int added, deleted;
699 } **files;
702 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
703 const char *name_a,
704 const char *name_b)
706 struct diffstat_file *x;
707 x = xcalloc(sizeof (*x), 1);
708 if (diffstat->nr == diffstat->alloc) {
709 diffstat->alloc = alloc_nr(diffstat->alloc);
710 diffstat->files = xrealloc(diffstat->files,
711 diffstat->alloc * sizeof(x));
713 diffstat->files[diffstat->nr++] = x;
714 if (name_b) {
715 x->from_name = xstrdup(name_a);
716 x->name = xstrdup(name_b);
717 x->is_renamed = 1;
719 else {
720 x->from_name = NULL;
721 x->name = xstrdup(name_a);
723 return x;
726 static void diffstat_consume(void *priv, char *line, unsigned long len)
728 struct diffstat_t *diffstat = priv;
729 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
731 if (line[0] == '+')
732 x->added++;
733 else if (line[0] == '-')
734 x->deleted++;
737 const char mime_boundary_leader[] = "------------";
739 static int scale_linear(int it, int width, int max_change)
742 * make sure that at least one '-' is printed if there were deletions,
743 * and likewise for '+'.
745 if (max_change < 2)
746 return it;
747 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
750 static void show_name(FILE *file,
751 const char *prefix, const char *name, int len,
752 const char *reset, const char *set)
754 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
757 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
759 if (cnt <= 0)
760 return;
761 fprintf(file, "%s", set);
762 while (cnt--)
763 putc(ch, file);
764 fprintf(file, "%s", reset);
767 static void fill_print_name(struct diffstat_file *file)
769 char *pname;
771 if (file->print_name)
772 return;
774 if (!file->is_renamed) {
775 struct strbuf buf = STRBUF_INIT;
776 if (quote_c_style(file->name, &buf, NULL, 0)) {
777 pname = strbuf_detach(&buf, NULL);
778 } else {
779 pname = file->name;
780 strbuf_release(&buf);
782 } else {
783 pname = pprint_rename(file->from_name, file->name);
785 file->print_name = pname;
788 static void show_stats(struct diffstat_t* data, struct diff_options *options)
790 int i, len, add, del, total, adds = 0, dels = 0;
791 int max_change = 0, max_len = 0;
792 int total_files = data->nr;
793 int width, name_width;
794 const char *reset, *set, *add_c, *del_c;
796 if (data->nr == 0)
797 return;
799 width = options->stat_width ? options->stat_width : 80;
800 name_width = options->stat_name_width ? options->stat_name_width : 50;
802 /* Sanity: give at least 5 columns to the graph,
803 * but leave at least 10 columns for the name.
805 if (width < 25)
806 width = 25;
807 if (name_width < 10)
808 name_width = 10;
809 else if (width < name_width + 15)
810 name_width = width - 15;
812 /* Find the longest filename and max number of changes */
813 reset = diff_get_color_opt(options, DIFF_RESET);
814 set = diff_get_color_opt(options, DIFF_PLAIN);
815 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
816 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
818 for (i = 0; i < data->nr; i++) {
819 struct diffstat_file *file = data->files[i];
820 int change = file->added + file->deleted;
821 fill_print_name(file);
822 len = strlen(file->print_name);
823 if (max_len < len)
824 max_len = len;
826 if (file->is_binary || file->is_unmerged)
827 continue;
828 if (max_change < change)
829 max_change = change;
832 /* Compute the width of the graph part;
833 * 10 is for one blank at the beginning of the line plus
834 * " | count " between the name and the graph.
836 * From here on, name_width is the width of the name area,
837 * and width is the width of the graph area.
839 name_width = (name_width < max_len) ? name_width : max_len;
840 if (width < (name_width + 10) + max_change)
841 width = width - (name_width + 10);
842 else
843 width = max_change;
845 for (i = 0; i < data->nr; i++) {
846 const char *prefix = "";
847 char *name = data->files[i]->print_name;
848 int added = data->files[i]->added;
849 int deleted = data->files[i]->deleted;
850 int name_len;
853 * "scale" the filename
855 len = name_width;
856 name_len = strlen(name);
857 if (name_width < name_len) {
858 char *slash;
859 prefix = "...";
860 len -= 3;
861 name += name_len - len;
862 slash = strchr(name, '/');
863 if (slash)
864 name = slash;
867 if (data->files[i]->is_binary) {
868 show_name(options->file, prefix, name, len, reset, set);
869 fprintf(options->file, " Bin ");
870 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
871 fprintf(options->file, " -> ");
872 fprintf(options->file, "%s%d%s", add_c, added, reset);
873 fprintf(options->file, " bytes");
874 fprintf(options->file, "\n");
875 continue;
877 else if (data->files[i]->is_unmerged) {
878 show_name(options->file, prefix, name, len, reset, set);
879 fprintf(options->file, " Unmerged\n");
880 continue;
882 else if (!data->files[i]->is_renamed &&
883 (added + deleted == 0)) {
884 total_files--;
885 continue;
889 * scale the add/delete
891 add = added;
892 del = deleted;
893 total = add + del;
894 adds += add;
895 dels += del;
897 if (width <= max_change) {
898 add = scale_linear(add, width, max_change);
899 del = scale_linear(del, width, max_change);
900 total = add + del;
902 show_name(options->file, prefix, name, len, reset, set);
903 fprintf(options->file, "%5d%s", added + deleted,
904 added + deleted ? " " : "");
905 show_graph(options->file, '+', add, add_c, reset);
906 show_graph(options->file, '-', del, del_c, reset);
907 fprintf(options->file, "\n");
909 fprintf(options->file,
910 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
911 set, total_files, adds, dels, reset);
914 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
916 int i, adds = 0, dels = 0, total_files = data->nr;
918 if (data->nr == 0)
919 return;
921 for (i = 0; i < data->nr; i++) {
922 if (!data->files[i]->is_binary &&
923 !data->files[i]->is_unmerged) {
924 int added = data->files[i]->added;
925 int deleted= data->files[i]->deleted;
926 if (!data->files[i]->is_renamed &&
927 (added + deleted == 0)) {
928 total_files--;
929 } else {
930 adds += added;
931 dels += deleted;
935 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
936 total_files, adds, dels);
939 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
941 int i;
943 if (data->nr == 0)
944 return;
946 for (i = 0; i < data->nr; i++) {
947 struct diffstat_file *file = data->files[i];
949 if (file->is_binary)
950 fprintf(options->file, "-\t-\t");
951 else
952 fprintf(options->file,
953 "%d\t%d\t", file->added, file->deleted);
954 if (options->line_termination) {
955 fill_print_name(file);
956 if (!file->is_renamed)
957 write_name_quoted(file->name, options->file,
958 options->line_termination);
959 else {
960 fputs(file->print_name, options->file);
961 putc(options->line_termination, options->file);
963 } else {
964 if (file->is_renamed) {
965 putc('\0', options->file);
966 write_name_quoted(file->from_name, options->file, '\0');
968 write_name_quoted(file->name, options->file, '\0');
973 struct dirstat_file {
974 const char *name;
975 unsigned long changed;
978 struct dirstat_dir {
979 struct dirstat_file *files;
980 int alloc, nr, percent, cumulative;
983 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
985 unsigned long this_dir = 0;
986 unsigned int sources = 0;
988 while (dir->nr) {
989 struct dirstat_file *f = dir->files;
990 int namelen = strlen(f->name);
991 unsigned long this;
992 char *slash;
994 if (namelen < baselen)
995 break;
996 if (memcmp(f->name, base, baselen))
997 break;
998 slash = strchr(f->name + baselen, '/');
999 if (slash) {
1000 int newbaselen = slash + 1 - f->name;
1001 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1002 sources++;
1003 } else {
1004 this = f->changed;
1005 dir->files++;
1006 dir->nr--;
1007 sources += 2;
1009 this_dir += this;
1013 * We don't report dirstat's for
1014 * - the top level
1015 * - or cases where everything came from a single directory
1016 * under this directory (sources == 1).
1018 if (baselen && sources != 1) {
1019 int permille = this_dir * 1000 / changed;
1020 if (permille) {
1021 int percent = permille / 10;
1022 if (percent >= dir->percent) {
1023 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1024 if (!dir->cumulative)
1025 return 0;
1029 return this_dir;
1032 static int dirstat_compare(const void *_a, const void *_b)
1034 const struct dirstat_file *a = _a;
1035 const struct dirstat_file *b = _b;
1036 return strcmp(a->name, b->name);
1039 static void show_dirstat(struct diff_options *options)
1041 int i;
1042 unsigned long changed;
1043 struct dirstat_dir dir;
1044 struct diff_queue_struct *q = &diff_queued_diff;
1046 dir.files = NULL;
1047 dir.alloc = 0;
1048 dir.nr = 0;
1049 dir.percent = options->dirstat_percent;
1050 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1052 changed = 0;
1053 for (i = 0; i < q->nr; i++) {
1054 struct diff_filepair *p = q->queue[i];
1055 const char *name;
1056 unsigned long copied, added, damage;
1058 name = p->one->path ? p->one->path : p->two->path;
1060 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1061 diff_populate_filespec(p->one, 0);
1062 diff_populate_filespec(p->two, 0);
1063 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1064 &copied, &added);
1065 diff_free_filespec_data(p->one);
1066 diff_free_filespec_data(p->two);
1067 } else if (DIFF_FILE_VALID(p->one)) {
1068 diff_populate_filespec(p->one, 1);
1069 copied = added = 0;
1070 diff_free_filespec_data(p->one);
1071 } else if (DIFF_FILE_VALID(p->two)) {
1072 diff_populate_filespec(p->two, 1);
1073 copied = 0;
1074 added = p->two->size;
1075 diff_free_filespec_data(p->two);
1076 } else
1077 continue;
1080 * Original minus copied is the removed material,
1081 * added is the new material. They are both damages
1082 * made to the preimage. In --dirstat-by-file mode, count
1083 * damaged files, not damaged lines. This is done by
1084 * counting only a single damaged line per file.
1086 damage = (p->one->size - copied) + added;
1087 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1088 damage = 1;
1090 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1091 dir.files[dir.nr].name = name;
1092 dir.files[dir.nr].changed = damage;
1093 changed += damage;
1094 dir.nr++;
1097 /* This can happen even with many files, if everything was renames */
1098 if (!changed)
1099 return;
1101 /* Show all directories with more than x% of the changes */
1102 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1103 gather_dirstat(options->file, &dir, changed, "", 0);
1106 static void free_diffstat_info(struct diffstat_t *diffstat)
1108 int i;
1109 for (i = 0; i < diffstat->nr; i++) {
1110 struct diffstat_file *f = diffstat->files[i];
1111 if (f->name != f->print_name)
1112 free(f->print_name);
1113 free(f->name);
1114 free(f->from_name);
1115 free(f);
1117 free(diffstat->files);
1120 struct checkdiff_t {
1121 const char *filename;
1122 int lineno;
1123 struct diff_options *o;
1124 unsigned ws_rule;
1125 unsigned status;
1126 int trailing_blanks_start;
1129 static int is_conflict_marker(const char *line, unsigned long len)
1131 char firstchar;
1132 int cnt;
1134 if (len < 8)
1135 return 0;
1136 firstchar = line[0];
1137 switch (firstchar) {
1138 case '=': case '>': case '<':
1139 break;
1140 default:
1141 return 0;
1143 for (cnt = 1; cnt < 7; cnt++)
1144 if (line[cnt] != firstchar)
1145 return 0;
1146 /* line[0] thru line[6] are same as firstchar */
1147 if (firstchar == '=') {
1148 /* divider between ours and theirs? */
1149 if (len != 8 || line[7] != '\n')
1150 return 0;
1151 } else if (len < 8 || !isspace(line[7])) {
1152 /* not divider before ours nor after theirs */
1153 return 0;
1155 return 1;
1158 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1160 struct checkdiff_t *data = priv;
1161 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1162 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1163 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1164 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1165 char *err;
1167 if (line[0] == '+') {
1168 unsigned bad;
1169 data->lineno++;
1170 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1171 data->trailing_blanks_start = 0;
1172 else if (!data->trailing_blanks_start)
1173 data->trailing_blanks_start = data->lineno;
1174 if (is_conflict_marker(line + 1, len - 1)) {
1175 data->status |= 1;
1176 fprintf(data->o->file,
1177 "%s:%d: leftover conflict marker\n",
1178 data->filename, data->lineno);
1180 bad = ws_check(line + 1, len - 1, data->ws_rule);
1181 if (!bad)
1182 return;
1183 data->status |= bad;
1184 err = whitespace_error_string(bad);
1185 fprintf(data->o->file, "%s:%d: %s.\n",
1186 data->filename, data->lineno, err);
1187 free(err);
1188 emit_line(data->o->file, set, reset, line, 1);
1189 ws_check_emit(line + 1, len - 1, data->ws_rule,
1190 data->o->file, set, reset, ws);
1191 } else if (line[0] == ' ') {
1192 data->lineno++;
1193 data->trailing_blanks_start = 0;
1194 } else if (line[0] == '@') {
1195 char *plus = strchr(line, '+');
1196 if (plus)
1197 data->lineno = strtol(plus, NULL, 10) - 1;
1198 else
1199 die("invalid diff");
1200 data->trailing_blanks_start = 0;
1204 static unsigned char *deflate_it(char *data,
1205 unsigned long size,
1206 unsigned long *result_size)
1208 int bound;
1209 unsigned char *deflated;
1210 z_stream stream;
1212 memset(&stream, 0, sizeof(stream));
1213 deflateInit(&stream, zlib_compression_level);
1214 bound = deflateBound(&stream, size);
1215 deflated = xmalloc(bound);
1216 stream.next_out = deflated;
1217 stream.avail_out = bound;
1219 stream.next_in = (unsigned char *)data;
1220 stream.avail_in = size;
1221 while (deflate(&stream, Z_FINISH) == Z_OK)
1222 ; /* nothing */
1223 deflateEnd(&stream);
1224 *result_size = stream.total_out;
1225 return deflated;
1228 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1230 void *cp;
1231 void *delta;
1232 void *deflated;
1233 void *data;
1234 unsigned long orig_size;
1235 unsigned long delta_size;
1236 unsigned long deflate_size;
1237 unsigned long data_size;
1239 /* We could do deflated delta, or we could do just deflated two,
1240 * whichever is smaller.
1242 delta = NULL;
1243 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1244 if (one->size && two->size) {
1245 delta = diff_delta(one->ptr, one->size,
1246 two->ptr, two->size,
1247 &delta_size, deflate_size);
1248 if (delta) {
1249 void *to_free = delta;
1250 orig_size = delta_size;
1251 delta = deflate_it(delta, delta_size, &delta_size);
1252 free(to_free);
1256 if (delta && delta_size < deflate_size) {
1257 fprintf(file, "delta %lu\n", orig_size);
1258 free(deflated);
1259 data = delta;
1260 data_size = delta_size;
1262 else {
1263 fprintf(file, "literal %lu\n", two->size);
1264 free(delta);
1265 data = deflated;
1266 data_size = deflate_size;
1269 /* emit data encoded in base85 */
1270 cp = data;
1271 while (data_size) {
1272 int bytes = (52 < data_size) ? 52 : data_size;
1273 char line[70];
1274 data_size -= bytes;
1275 if (bytes <= 26)
1276 line[0] = bytes + 'A' - 1;
1277 else
1278 line[0] = bytes - 26 + 'a' - 1;
1279 encode_85(line + 1, cp, bytes);
1280 cp = (char *) cp + bytes;
1281 fputs(line, file);
1282 fputc('\n', file);
1284 fprintf(file, "\n");
1285 free(data);
1288 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1290 fprintf(file, "GIT binary patch\n");
1291 emit_binary_diff_body(file, one, two);
1292 emit_binary_diff_body(file, two, one);
1295 static void diff_filespec_load_driver(struct diff_filespec *one)
1297 if (!one->driver)
1298 one->driver = userdiff_find_by_path(one->path);
1299 if (!one->driver)
1300 one->driver = userdiff_find_by_name("default");
1303 int diff_filespec_is_binary(struct diff_filespec *one)
1305 if (one->is_binary == -1) {
1306 diff_filespec_load_driver(one);
1307 if (one->driver->binary != -1)
1308 one->is_binary = one->driver->binary;
1309 else {
1310 if (!one->data && DIFF_FILE_VALID(one))
1311 diff_populate_filespec(one, 0);
1312 if (one->data)
1313 one->is_binary = buffer_is_binary(one->data,
1314 one->size);
1315 if (one->is_binary == -1)
1316 one->is_binary = 0;
1319 return one->is_binary;
1322 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1324 diff_filespec_load_driver(one);
1325 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1328 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1330 if (!options->a_prefix)
1331 options->a_prefix = a;
1332 if (!options->b_prefix)
1333 options->b_prefix = b;
1336 static const char *get_textconv(struct diff_filespec *one)
1338 if (!DIFF_FILE_VALID(one))
1339 return NULL;
1340 if (!S_ISREG(one->mode))
1341 return NULL;
1342 diff_filespec_load_driver(one);
1343 return one->driver->textconv;
1346 static void builtin_diff(const char *name_a,
1347 const char *name_b,
1348 struct diff_filespec *one,
1349 struct diff_filespec *two,
1350 const char *xfrm_msg,
1351 struct diff_options *o,
1352 int complete_rewrite)
1354 mmfile_t mf1, mf2;
1355 const char *lbl[2];
1356 char *a_one, *b_two;
1357 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1358 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1359 const char *a_prefix, *b_prefix;
1360 const char *textconv_one = NULL, *textconv_two = NULL;
1362 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1363 textconv_one = get_textconv(one);
1364 textconv_two = get_textconv(two);
1367 diff_set_mnemonic_prefix(o, "a/", "b/");
1368 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1369 a_prefix = o->b_prefix;
1370 b_prefix = o->a_prefix;
1371 } else {
1372 a_prefix = o->a_prefix;
1373 b_prefix = o->b_prefix;
1376 /* Never use a non-valid filename anywhere if at all possible */
1377 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1378 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1380 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1381 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1382 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1383 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1384 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1385 if (lbl[0][0] == '/') {
1386 /* /dev/null */
1387 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1388 if (xfrm_msg && xfrm_msg[0])
1389 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1391 else if (lbl[1][0] == '/') {
1392 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1393 if (xfrm_msg && xfrm_msg[0])
1394 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1396 else {
1397 if (one->mode != two->mode) {
1398 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1399 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1401 if (xfrm_msg && xfrm_msg[0])
1402 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1404 * we do not run diff between different kind
1405 * of objects.
1407 if ((one->mode ^ two->mode) & S_IFMT)
1408 goto free_ab_and_return;
1409 if (complete_rewrite &&
1410 (textconv_one || !diff_filespec_is_binary(one)) &&
1411 (textconv_two || !diff_filespec_is_binary(two))) {
1412 emit_rewrite_diff(name_a, name_b, one, two,
1413 textconv_one, textconv_two, o);
1414 o->found_changes = 1;
1415 goto free_ab_and_return;
1419 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1420 die("unable to read files to diff");
1422 if (!DIFF_OPT_TST(o, TEXT) &&
1423 ( (diff_filespec_is_binary(one) && !textconv_one) ||
1424 (diff_filespec_is_binary(two) && !textconv_two) )) {
1425 /* Quite common confusing case */
1426 if (mf1.size == mf2.size &&
1427 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1428 goto free_ab_and_return;
1429 if (DIFF_OPT_TST(o, BINARY))
1430 emit_binary_diff(o->file, &mf1, &mf2);
1431 else
1432 fprintf(o->file, "Binary files %s and %s differ\n",
1433 lbl[0], lbl[1]);
1434 o->found_changes = 1;
1436 else {
1437 /* Crazy xdl interfaces.. */
1438 const char *diffopts = getenv("GIT_DIFF_OPTS");
1439 xpparam_t xpp;
1440 xdemitconf_t xecfg;
1441 xdemitcb_t ecb;
1442 struct emit_callback ecbdata;
1443 const struct userdiff_funcname *pe;
1445 if (textconv_one) {
1446 size_t size;
1447 mf1.ptr = run_textconv(textconv_one, one, &size);
1448 if (!mf1.ptr)
1449 die("unable to read files to diff");
1450 mf1.size = size;
1452 if (textconv_two) {
1453 size_t size;
1454 mf2.ptr = run_textconv(textconv_two, two, &size);
1455 if (!mf2.ptr)
1456 die("unable to read files to diff");
1457 mf2.size = size;
1460 pe = diff_funcname_pattern(one);
1461 if (!pe)
1462 pe = diff_funcname_pattern(two);
1464 memset(&xpp, 0, sizeof(xpp));
1465 memset(&xecfg, 0, sizeof(xecfg));
1466 memset(&ecbdata, 0, sizeof(ecbdata));
1467 ecbdata.label_path = lbl;
1468 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1469 ecbdata.found_changesp = &o->found_changes;
1470 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1471 ecbdata.file = o->file;
1472 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1473 xecfg.ctxlen = o->context;
1474 xecfg.interhunkctxlen = o->interhunkcontext;
1475 xecfg.flags = XDL_EMIT_FUNCNAMES;
1476 if (pe)
1477 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1478 if (!diffopts)
1480 else if (!prefixcmp(diffopts, "--unified="))
1481 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1482 else if (!prefixcmp(diffopts, "-u"))
1483 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1484 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1485 ecbdata.diff_words =
1486 xcalloc(1, sizeof(struct diff_words_data));
1487 ecbdata.diff_words->file = o->file;
1489 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1490 &xpp, &xecfg, &ecb);
1491 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1492 free_diff_words_data(&ecbdata);
1493 if (textconv_one)
1494 free(mf1.ptr);
1495 if (textconv_two)
1496 free(mf2.ptr);
1499 free_ab_and_return:
1500 diff_free_filespec_data(one);
1501 diff_free_filespec_data(two);
1502 free(a_one);
1503 free(b_two);
1504 return;
1507 static void builtin_diffstat(const char *name_a, const char *name_b,
1508 struct diff_filespec *one,
1509 struct diff_filespec *two,
1510 struct diffstat_t *diffstat,
1511 struct diff_options *o,
1512 int complete_rewrite)
1514 mmfile_t mf1, mf2;
1515 struct diffstat_file *data;
1517 data = diffstat_add(diffstat, name_a, name_b);
1519 if (!one || !two) {
1520 data->is_unmerged = 1;
1521 return;
1523 if (complete_rewrite) {
1524 diff_populate_filespec(one, 0);
1525 diff_populate_filespec(two, 0);
1526 data->deleted = count_lines(one->data, one->size);
1527 data->added = count_lines(two->data, two->size);
1528 goto free_and_return;
1530 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1531 die("unable to read files to diff");
1533 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1534 data->is_binary = 1;
1535 data->added = mf2.size;
1536 data->deleted = mf1.size;
1537 } else {
1538 /* Crazy xdl interfaces.. */
1539 xpparam_t xpp;
1540 xdemitconf_t xecfg;
1541 xdemitcb_t ecb;
1543 memset(&xpp, 0, sizeof(xpp));
1544 memset(&xecfg, 0, sizeof(xecfg));
1545 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1546 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1547 &xpp, &xecfg, &ecb);
1550 free_and_return:
1551 diff_free_filespec_data(one);
1552 diff_free_filespec_data(two);
1555 static void builtin_checkdiff(const char *name_a, const char *name_b,
1556 const char *attr_path,
1557 struct diff_filespec *one,
1558 struct diff_filespec *two,
1559 struct diff_options *o)
1561 mmfile_t mf1, mf2;
1562 struct checkdiff_t data;
1564 if (!two)
1565 return;
1567 memset(&data, 0, sizeof(data));
1568 data.filename = name_b ? name_b : name_a;
1569 data.lineno = 0;
1570 data.o = o;
1571 data.ws_rule = whitespace_rule(attr_path);
1573 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1574 die("unable to read files to diff");
1577 * All the other codepaths check both sides, but not checking
1578 * the "old" side here is deliberate. We are checking the newly
1579 * introduced changes, and as long as the "new" side is text, we
1580 * can and should check what it introduces.
1582 if (diff_filespec_is_binary(two))
1583 goto free_and_return;
1584 else {
1585 /* Crazy xdl interfaces.. */
1586 xpparam_t xpp;
1587 xdemitconf_t xecfg;
1588 xdemitcb_t ecb;
1590 memset(&xpp, 0, sizeof(xpp));
1591 memset(&xecfg, 0, sizeof(xecfg));
1592 xecfg.ctxlen = 1; /* at least one context line */
1593 xpp.flags = XDF_NEED_MINIMAL;
1594 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1595 &xpp, &xecfg, &ecb);
1597 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1598 data.trailing_blanks_start) {
1599 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1600 data.filename, data.trailing_blanks_start);
1601 data.status = 1; /* report errors */
1604 free_and_return:
1605 diff_free_filespec_data(one);
1606 diff_free_filespec_data(two);
1607 if (data.status)
1608 DIFF_OPT_SET(o, CHECK_FAILED);
1611 struct diff_filespec *alloc_filespec(const char *path)
1613 int namelen = strlen(path);
1614 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1616 memset(spec, 0, sizeof(*spec));
1617 spec->path = (char *)(spec + 1);
1618 memcpy(spec->path, path, namelen+1);
1619 spec->count = 1;
1620 spec->is_binary = -1;
1621 return spec;
1624 void free_filespec(struct diff_filespec *spec)
1626 if (!--spec->count) {
1627 diff_free_filespec_data(spec);
1628 free(spec);
1632 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1633 unsigned short mode)
1635 if (mode) {
1636 spec->mode = canon_mode(mode);
1637 hashcpy(spec->sha1, sha1);
1638 spec->sha1_valid = !is_null_sha1(sha1);
1643 * Given a name and sha1 pair, if the index tells us the file in
1644 * the work tree has that object contents, return true, so that
1645 * prepare_temp_file() does not have to inflate and extract.
1647 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1649 struct cache_entry *ce;
1650 struct stat st;
1651 int pos, len;
1653 /* We do not read the cache ourselves here, because the
1654 * benchmark with my previous version that always reads cache
1655 * shows that it makes things worse for diff-tree comparing
1656 * two linux-2.6 kernel trees in an already checked out work
1657 * tree. This is because most diff-tree comparisons deal with
1658 * only a small number of files, while reading the cache is
1659 * expensive for a large project, and its cost outweighs the
1660 * savings we get by not inflating the object to a temporary
1661 * file. Practically, this code only helps when we are used
1662 * by diff-cache --cached, which does read the cache before
1663 * calling us.
1665 if (!active_cache)
1666 return 0;
1668 /* We want to avoid the working directory if our caller
1669 * doesn't need the data in a normal file, this system
1670 * is rather slow with its stat/open/mmap/close syscalls,
1671 * and the object is contained in a pack file. The pack
1672 * is probably already open and will be faster to obtain
1673 * the data through than the working directory. Loose
1674 * objects however would tend to be slower as they need
1675 * to be individually opened and inflated.
1677 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1678 return 0;
1680 len = strlen(name);
1681 pos = cache_name_pos(name, len);
1682 if (pos < 0)
1683 return 0;
1684 ce = active_cache[pos];
1687 * This is not the sha1 we are looking for, or
1688 * unreusable because it is not a regular file.
1690 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1691 return 0;
1694 * If ce matches the file in the work tree, we can reuse it.
1696 if (ce_uptodate(ce) ||
1697 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1698 return 1;
1700 return 0;
1703 static int populate_from_stdin(struct diff_filespec *s)
1705 struct strbuf buf = STRBUF_INIT;
1706 size_t size = 0;
1708 if (strbuf_read(&buf, 0, 0) < 0)
1709 return error("error while reading from stdin %s",
1710 strerror(errno));
1712 s->should_munmap = 0;
1713 s->data = strbuf_detach(&buf, &size);
1714 s->size = size;
1715 s->should_free = 1;
1716 return 0;
1719 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1721 int len;
1722 char *data = xmalloc(100);
1723 len = snprintf(data, 100,
1724 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1725 s->data = data;
1726 s->size = len;
1727 s->should_free = 1;
1728 if (size_only) {
1729 s->data = NULL;
1730 free(data);
1732 return 0;
1736 * While doing rename detection and pickaxe operation, we may need to
1737 * grab the data for the blob (or file) for our own in-core comparison.
1738 * diff_filespec has data and size fields for this purpose.
1740 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1742 int err = 0;
1743 if (!DIFF_FILE_VALID(s))
1744 die("internal error: asking to populate invalid file.");
1745 if (S_ISDIR(s->mode))
1746 return -1;
1748 if (s->data)
1749 return 0;
1751 if (size_only && 0 < s->size)
1752 return 0;
1754 if (S_ISGITLINK(s->mode))
1755 return diff_populate_gitlink(s, size_only);
1757 if (!s->sha1_valid ||
1758 reuse_worktree_file(s->path, s->sha1, 0)) {
1759 struct strbuf buf = STRBUF_INIT;
1760 struct stat st;
1761 int fd;
1763 if (!strcmp(s->path, "-"))
1764 return populate_from_stdin(s);
1766 if (lstat(s->path, &st) < 0) {
1767 if (errno == ENOENT) {
1768 err_empty:
1769 err = -1;
1770 empty:
1771 s->data = (char *)"";
1772 s->size = 0;
1773 return err;
1776 s->size = xsize_t(st.st_size);
1777 if (!s->size)
1778 goto empty;
1779 if (S_ISLNK(st.st_mode)) {
1780 struct strbuf sb = STRBUF_INIT;
1782 if (strbuf_readlink(&sb, s->path, s->size))
1783 goto err_empty;
1784 s->size = sb.len;
1785 s->data = strbuf_detach(&sb, NULL);
1786 s->should_free = 1;
1787 return 0;
1789 if (size_only)
1790 return 0;
1791 fd = open(s->path, O_RDONLY);
1792 if (fd < 0)
1793 goto err_empty;
1794 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1795 close(fd);
1796 s->should_munmap = 1;
1799 * Convert from working tree format to canonical git format
1801 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1802 size_t size = 0;
1803 munmap(s->data, s->size);
1804 s->should_munmap = 0;
1805 s->data = strbuf_detach(&buf, &size);
1806 s->size = size;
1807 s->should_free = 1;
1810 else {
1811 enum object_type type;
1812 if (size_only)
1813 type = sha1_object_info(s->sha1, &s->size);
1814 else {
1815 s->data = read_sha1_file(s->sha1, &type, &s->size);
1816 s->should_free = 1;
1819 return 0;
1822 void diff_free_filespec_blob(struct diff_filespec *s)
1824 if (s->should_free)
1825 free(s->data);
1826 else if (s->should_munmap)
1827 munmap(s->data, s->size);
1829 if (s->should_free || s->should_munmap) {
1830 s->should_free = s->should_munmap = 0;
1831 s->data = NULL;
1835 void diff_free_filespec_data(struct diff_filespec *s)
1837 diff_free_filespec_blob(s);
1838 free(s->cnt_data);
1839 s->cnt_data = NULL;
1842 static void prep_temp_blob(struct diff_tempfile *temp,
1843 void *blob,
1844 unsigned long size,
1845 const unsigned char *sha1,
1846 int mode)
1848 int fd;
1850 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1851 if (fd < 0)
1852 die("unable to create temp-file: %s", strerror(errno));
1853 if (write_in_full(fd, blob, size) != size)
1854 die("unable to write temp-file");
1855 close(fd);
1856 temp->name = temp->tmp_path;
1857 strcpy(temp->hex, sha1_to_hex(sha1));
1858 temp->hex[40] = 0;
1859 sprintf(temp->mode, "%06o", mode);
1862 static void prepare_temp_file(const char *name,
1863 struct diff_tempfile *temp,
1864 struct diff_filespec *one)
1866 if (!DIFF_FILE_VALID(one)) {
1867 not_a_valid_file:
1868 /* A '-' entry produces this for file-2, and
1869 * a '+' entry produces this for file-1.
1871 temp->name = "/dev/null";
1872 strcpy(temp->hex, ".");
1873 strcpy(temp->mode, ".");
1874 return;
1877 if (!one->sha1_valid ||
1878 reuse_worktree_file(name, one->sha1, 1)) {
1879 struct stat st;
1880 if (lstat(name, &st) < 0) {
1881 if (errno == ENOENT)
1882 goto not_a_valid_file;
1883 die("stat(%s): %s", name, strerror(errno));
1885 if (S_ISLNK(st.st_mode)) {
1886 int ret;
1887 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1888 ret = readlink(name, buf, sizeof(buf));
1889 if (ret < 0)
1890 die("readlink(%s)", name);
1891 if (ret == sizeof(buf))
1892 die("symlink too long: %s", name);
1893 prep_temp_blob(temp, buf, ret,
1894 (one->sha1_valid ?
1895 one->sha1 : null_sha1),
1896 (one->sha1_valid ?
1897 one->mode : S_IFLNK));
1899 else {
1900 /* we can borrow from the file in the work tree */
1901 temp->name = name;
1902 if (!one->sha1_valid)
1903 strcpy(temp->hex, sha1_to_hex(null_sha1));
1904 else
1905 strcpy(temp->hex, sha1_to_hex(one->sha1));
1906 /* Even though we may sometimes borrow the
1907 * contents from the work tree, we always want
1908 * one->mode. mode is trustworthy even when
1909 * !(one->sha1_valid), as long as
1910 * DIFF_FILE_VALID(one).
1912 sprintf(temp->mode, "%06o", one->mode);
1914 return;
1916 else {
1917 if (diff_populate_filespec(one, 0))
1918 die("cannot read data blob for %s", one->path);
1919 prep_temp_blob(temp, one->data, one->size,
1920 one->sha1, one->mode);
1924 static void remove_tempfile(void)
1926 int i;
1928 for (i = 0; i < 2; i++)
1929 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1930 unlink(diff_temp[i].name);
1931 diff_temp[i].name = NULL;
1935 static void remove_tempfile_on_signal(int signo)
1937 remove_tempfile();
1938 signal(SIGINT, SIG_DFL);
1939 raise(signo);
1942 /* An external diff command takes:
1944 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1945 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1948 static void run_external_diff(const char *pgm,
1949 const char *name,
1950 const char *other,
1951 struct diff_filespec *one,
1952 struct diff_filespec *two,
1953 const char *xfrm_msg,
1954 int complete_rewrite)
1956 const char *spawn_arg[10];
1957 struct diff_tempfile *temp = diff_temp;
1958 int retval;
1959 static int atexit_asked = 0;
1960 const char *othername;
1961 const char **arg = &spawn_arg[0];
1963 othername = (other? other : name);
1964 if (one && two) {
1965 prepare_temp_file(name, &temp[0], one);
1966 prepare_temp_file(othername, &temp[1], two);
1967 if (! atexit_asked &&
1968 (temp[0].name == temp[0].tmp_path ||
1969 temp[1].name == temp[1].tmp_path)) {
1970 atexit_asked = 1;
1971 atexit(remove_tempfile);
1973 signal(SIGINT, remove_tempfile_on_signal);
1976 if (one && two) {
1977 *arg++ = pgm;
1978 *arg++ = name;
1979 *arg++ = temp[0].name;
1980 *arg++ = temp[0].hex;
1981 *arg++ = temp[0].mode;
1982 *arg++ = temp[1].name;
1983 *arg++ = temp[1].hex;
1984 *arg++ = temp[1].mode;
1985 if (other) {
1986 *arg++ = other;
1987 *arg++ = xfrm_msg;
1989 } else {
1990 *arg++ = pgm;
1991 *arg++ = name;
1993 *arg = NULL;
1994 fflush(NULL);
1995 retval = run_command_v_opt(spawn_arg, 0);
1996 remove_tempfile();
1997 if (retval) {
1998 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1999 exit(1);
2003 static void run_diff_cmd(const char *pgm,
2004 const char *name,
2005 const char *other,
2006 const char *attr_path,
2007 struct diff_filespec *one,
2008 struct diff_filespec *two,
2009 const char *xfrm_msg,
2010 struct diff_options *o,
2011 int complete_rewrite)
2013 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2014 pgm = NULL;
2015 else {
2016 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2017 if (drv && drv->external)
2018 pgm = drv->external;
2021 if (pgm) {
2022 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2023 complete_rewrite);
2024 return;
2026 if (one && two)
2027 builtin_diff(name, other ? other : name,
2028 one, two, xfrm_msg, o, complete_rewrite);
2029 else
2030 fprintf(o->file, "* Unmerged path %s\n", name);
2033 static void diff_fill_sha1_info(struct diff_filespec *one)
2035 if (DIFF_FILE_VALID(one)) {
2036 if (!one->sha1_valid) {
2037 struct stat st;
2038 if (!strcmp(one->path, "-")) {
2039 hashcpy(one->sha1, null_sha1);
2040 return;
2042 if (lstat(one->path, &st) < 0)
2043 die("stat %s", one->path);
2044 if (index_path(one->sha1, one->path, &st, 0))
2045 die("cannot hash %s", one->path);
2048 else
2049 hashclr(one->sha1);
2052 static int similarity_index(struct diff_filepair *p)
2054 return p->score * 100 / MAX_SCORE;
2057 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2059 /* Strip the prefix but do not molest /dev/null and absolute paths */
2060 if (*namep && **namep != '/')
2061 *namep += prefix_length;
2062 if (*otherp && **otherp != '/')
2063 *otherp += prefix_length;
2066 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2068 const char *pgm = external_diff();
2069 struct strbuf msg;
2070 char *xfrm_msg;
2071 struct diff_filespec *one = p->one;
2072 struct diff_filespec *two = p->two;
2073 const char *name;
2074 const char *other;
2075 const char *attr_path;
2076 int complete_rewrite = 0;
2078 name = p->one->path;
2079 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2080 attr_path = name;
2081 if (o->prefix_length)
2082 strip_prefix(o->prefix_length, &name, &other);
2084 if (DIFF_PAIR_UNMERGED(p)) {
2085 run_diff_cmd(pgm, name, NULL, attr_path,
2086 NULL, NULL, NULL, o, 0);
2087 return;
2090 diff_fill_sha1_info(one);
2091 diff_fill_sha1_info(two);
2093 strbuf_init(&msg, PATH_MAX * 2 + 300);
2094 switch (p->status) {
2095 case DIFF_STATUS_COPIED:
2096 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2097 strbuf_addstr(&msg, "\ncopy from ");
2098 quote_c_style(name, &msg, NULL, 0);
2099 strbuf_addstr(&msg, "\ncopy to ");
2100 quote_c_style(other, &msg, NULL, 0);
2101 strbuf_addch(&msg, '\n');
2102 break;
2103 case DIFF_STATUS_RENAMED:
2104 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2105 strbuf_addstr(&msg, "\nrename from ");
2106 quote_c_style(name, &msg, NULL, 0);
2107 strbuf_addstr(&msg, "\nrename to ");
2108 quote_c_style(other, &msg, NULL, 0);
2109 strbuf_addch(&msg, '\n');
2110 break;
2111 case DIFF_STATUS_MODIFIED:
2112 if (p->score) {
2113 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2114 similarity_index(p));
2115 complete_rewrite = 1;
2116 break;
2118 /* fallthru */
2119 default:
2120 /* nothing */
2124 if (hashcmp(one->sha1, two->sha1)) {
2125 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2127 if (DIFF_OPT_TST(o, BINARY)) {
2128 mmfile_t mf;
2129 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2130 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2131 abbrev = 40;
2133 strbuf_addf(&msg, "index %.*s..%.*s",
2134 abbrev, sha1_to_hex(one->sha1),
2135 abbrev, sha1_to_hex(two->sha1));
2136 if (one->mode == two->mode)
2137 strbuf_addf(&msg, " %06o", one->mode);
2138 strbuf_addch(&msg, '\n');
2141 if (msg.len)
2142 strbuf_setlen(&msg, msg.len - 1);
2143 xfrm_msg = msg.len ? msg.buf : NULL;
2145 if (!pgm &&
2146 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2147 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2148 /* a filepair that changes between file and symlink
2149 * needs to be split into deletion and creation.
2151 struct diff_filespec *null = alloc_filespec(two->path);
2152 run_diff_cmd(NULL, name, other, attr_path,
2153 one, null, xfrm_msg, o, 0);
2154 free(null);
2155 null = alloc_filespec(one->path);
2156 run_diff_cmd(NULL, name, other, attr_path,
2157 null, two, xfrm_msg, o, 0);
2158 free(null);
2160 else
2161 run_diff_cmd(pgm, name, other, attr_path,
2162 one, two, xfrm_msg, o, complete_rewrite);
2164 strbuf_release(&msg);
2167 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2168 struct diffstat_t *diffstat)
2170 const char *name;
2171 const char *other;
2172 int complete_rewrite = 0;
2174 if (DIFF_PAIR_UNMERGED(p)) {
2175 /* unmerged */
2176 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2177 return;
2180 name = p->one->path;
2181 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2183 if (o->prefix_length)
2184 strip_prefix(o->prefix_length, &name, &other);
2186 diff_fill_sha1_info(p->one);
2187 diff_fill_sha1_info(p->two);
2189 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2190 complete_rewrite = 1;
2191 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2194 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2196 const char *name;
2197 const char *other;
2198 const char *attr_path;
2200 if (DIFF_PAIR_UNMERGED(p)) {
2201 /* unmerged */
2202 return;
2205 name = p->one->path;
2206 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2207 attr_path = other ? other : name;
2209 if (o->prefix_length)
2210 strip_prefix(o->prefix_length, &name, &other);
2212 diff_fill_sha1_info(p->one);
2213 diff_fill_sha1_info(p->two);
2215 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2218 void diff_setup(struct diff_options *options)
2220 memset(options, 0, sizeof(*options));
2222 options->file = stdout;
2224 options->line_termination = '\n';
2225 options->break_opt = -1;
2226 options->rename_limit = -1;
2227 options->dirstat_percent = 3;
2228 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2229 options->context = 3;
2231 options->change = diff_change;
2232 options->add_remove = diff_addremove;
2233 if (diff_use_color_default > 0)
2234 DIFF_OPT_SET(options, COLOR_DIFF);
2235 else
2236 DIFF_OPT_CLR(options, COLOR_DIFF);
2237 options->detect_rename = diff_detect_rename_default;
2239 if (!diff_mnemonic_prefix) {
2240 options->a_prefix = "a/";
2241 options->b_prefix = "b/";
2245 int diff_setup_done(struct diff_options *options)
2247 int count = 0;
2249 if (options->output_format & DIFF_FORMAT_NAME)
2250 count++;
2251 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2252 count++;
2253 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2254 count++;
2255 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2256 count++;
2257 if (count > 1)
2258 die("--name-only, --name-status, --check and -s are mutually exclusive");
2260 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2261 options->detect_rename = DIFF_DETECT_COPY;
2263 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2264 options->prefix = NULL;
2265 if (options->prefix)
2266 options->prefix_length = strlen(options->prefix);
2267 else
2268 options->prefix_length = 0;
2270 if (options->output_format & (DIFF_FORMAT_NAME |
2271 DIFF_FORMAT_NAME_STATUS |
2272 DIFF_FORMAT_CHECKDIFF |
2273 DIFF_FORMAT_NO_OUTPUT))
2274 options->output_format &= ~(DIFF_FORMAT_RAW |
2275 DIFF_FORMAT_NUMSTAT |
2276 DIFF_FORMAT_DIFFSTAT |
2277 DIFF_FORMAT_SHORTSTAT |
2278 DIFF_FORMAT_DIRSTAT |
2279 DIFF_FORMAT_SUMMARY |
2280 DIFF_FORMAT_PATCH);
2283 * These cases always need recursive; we do not drop caller-supplied
2284 * recursive bits for other formats here.
2286 if (options->output_format & (DIFF_FORMAT_PATCH |
2287 DIFF_FORMAT_NUMSTAT |
2288 DIFF_FORMAT_DIFFSTAT |
2289 DIFF_FORMAT_SHORTSTAT |
2290 DIFF_FORMAT_DIRSTAT |
2291 DIFF_FORMAT_SUMMARY |
2292 DIFF_FORMAT_CHECKDIFF))
2293 DIFF_OPT_SET(options, RECURSIVE);
2295 * Also pickaxe would not work very well if you do not say recursive
2297 if (options->pickaxe)
2298 DIFF_OPT_SET(options, RECURSIVE);
2300 if (options->detect_rename && options->rename_limit < 0)
2301 options->rename_limit = diff_rename_limit_default;
2302 if (options->setup & DIFF_SETUP_USE_CACHE) {
2303 if (!active_cache)
2304 /* read-cache does not die even when it fails
2305 * so it is safe for us to do this here. Also
2306 * it does not smudge active_cache or active_nr
2307 * when it fails, so we do not have to worry about
2308 * cleaning it up ourselves either.
2310 read_cache();
2312 if (options->abbrev <= 0 || 40 < options->abbrev)
2313 options->abbrev = 40; /* full */
2316 * It does not make sense to show the first hit we happened
2317 * to have found. It does not make sense not to return with
2318 * exit code in such a case either.
2320 if (DIFF_OPT_TST(options, QUIET)) {
2321 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2322 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2325 return 0;
2328 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2330 char c, *eq;
2331 int len;
2333 if (*arg != '-')
2334 return 0;
2335 c = *++arg;
2336 if (!c)
2337 return 0;
2338 if (c == arg_short) {
2339 c = *++arg;
2340 if (!c)
2341 return 1;
2342 if (val && isdigit(c)) {
2343 char *end;
2344 int n = strtoul(arg, &end, 10);
2345 if (*end)
2346 return 0;
2347 *val = n;
2348 return 1;
2350 return 0;
2352 if (c != '-')
2353 return 0;
2354 arg++;
2355 eq = strchr(arg, '=');
2356 if (eq)
2357 len = eq - arg;
2358 else
2359 len = strlen(arg);
2360 if (!len || strncmp(arg, arg_long, len))
2361 return 0;
2362 if (eq) {
2363 int n;
2364 char *end;
2365 if (!isdigit(*++eq))
2366 return 0;
2367 n = strtoul(eq, &end, 10);
2368 if (*end)
2369 return 0;
2370 *val = n;
2372 return 1;
2375 static int diff_scoreopt_parse(const char *opt);
2377 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2379 const char *arg = av[0];
2381 /* Output format options */
2382 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2383 options->output_format |= DIFF_FORMAT_PATCH;
2384 else if (opt_arg(arg, 'U', "unified", &options->context))
2385 options->output_format |= DIFF_FORMAT_PATCH;
2386 else if (!strcmp(arg, "--raw"))
2387 options->output_format |= DIFF_FORMAT_RAW;
2388 else if (!strcmp(arg, "--patch-with-raw"))
2389 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2390 else if (!strcmp(arg, "--numstat"))
2391 options->output_format |= DIFF_FORMAT_NUMSTAT;
2392 else if (!strcmp(arg, "--shortstat"))
2393 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2394 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2395 options->output_format |= DIFF_FORMAT_DIRSTAT;
2396 else if (!strcmp(arg, "--cumulative")) {
2397 options->output_format |= DIFF_FORMAT_DIRSTAT;
2398 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2399 } else if (opt_arg(arg, 0, "dirstat-by-file",
2400 &options->dirstat_percent)) {
2401 options->output_format |= DIFF_FORMAT_DIRSTAT;
2402 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2404 else if (!strcmp(arg, "--check"))
2405 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2406 else if (!strcmp(arg, "--summary"))
2407 options->output_format |= DIFF_FORMAT_SUMMARY;
2408 else if (!strcmp(arg, "--patch-with-stat"))
2409 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2410 else if (!strcmp(arg, "--name-only"))
2411 options->output_format |= DIFF_FORMAT_NAME;
2412 else if (!strcmp(arg, "--name-status"))
2413 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2414 else if (!strcmp(arg, "-s"))
2415 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2416 else if (!prefixcmp(arg, "--stat")) {
2417 char *end;
2418 int width = options->stat_width;
2419 int name_width = options->stat_name_width;
2420 arg += 6;
2421 end = (char *)arg;
2423 switch (*arg) {
2424 case '-':
2425 if (!prefixcmp(arg, "-width="))
2426 width = strtoul(arg + 7, &end, 10);
2427 else if (!prefixcmp(arg, "-name-width="))
2428 name_width = strtoul(arg + 12, &end, 10);
2429 break;
2430 case '=':
2431 width = strtoul(arg+1, &end, 10);
2432 if (*end == ',')
2433 name_width = strtoul(end+1, &end, 10);
2436 /* Important! This checks all the error cases! */
2437 if (*end)
2438 return 0;
2439 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2440 options->stat_name_width = name_width;
2441 options->stat_width = width;
2444 /* renames options */
2445 else if (!prefixcmp(arg, "-B")) {
2446 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2447 return -1;
2449 else if (!prefixcmp(arg, "-M")) {
2450 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2451 return -1;
2452 options->detect_rename = DIFF_DETECT_RENAME;
2454 else if (!prefixcmp(arg, "-C")) {
2455 if (options->detect_rename == DIFF_DETECT_COPY)
2456 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2457 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2458 return -1;
2459 options->detect_rename = DIFF_DETECT_COPY;
2461 else if (!strcmp(arg, "--no-renames"))
2462 options->detect_rename = 0;
2463 else if (!strcmp(arg, "--relative"))
2464 DIFF_OPT_SET(options, RELATIVE_NAME);
2465 else if (!prefixcmp(arg, "--relative=")) {
2466 DIFF_OPT_SET(options, RELATIVE_NAME);
2467 options->prefix = arg + 11;
2470 /* xdiff options */
2471 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2472 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2473 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2474 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2475 else if (!strcmp(arg, "--ignore-space-at-eol"))
2476 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2478 /* flags options */
2479 else if (!strcmp(arg, "--binary")) {
2480 options->output_format |= DIFF_FORMAT_PATCH;
2481 DIFF_OPT_SET(options, BINARY);
2483 else if (!strcmp(arg, "--full-index"))
2484 DIFF_OPT_SET(options, FULL_INDEX);
2485 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2486 DIFF_OPT_SET(options, TEXT);
2487 else if (!strcmp(arg, "-R"))
2488 DIFF_OPT_SET(options, REVERSE_DIFF);
2489 else if (!strcmp(arg, "--find-copies-harder"))
2490 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2491 else if (!strcmp(arg, "--follow"))
2492 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2493 else if (!strcmp(arg, "--color"))
2494 DIFF_OPT_SET(options, COLOR_DIFF);
2495 else if (!strcmp(arg, "--no-color"))
2496 DIFF_OPT_CLR(options, COLOR_DIFF);
2497 else if (!strcmp(arg, "--color-words"))
2498 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2499 else if (!strcmp(arg, "--exit-code"))
2500 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2501 else if (!strcmp(arg, "--quiet"))
2502 DIFF_OPT_SET(options, QUIET);
2503 else if (!strcmp(arg, "--ext-diff"))
2504 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2505 else if (!strcmp(arg, "--no-ext-diff"))
2506 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2507 else if (!strcmp(arg, "--textconv"))
2508 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2509 else if (!strcmp(arg, "--no-textconv"))
2510 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2511 else if (!strcmp(arg, "--ignore-submodules"))
2512 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2514 /* misc options */
2515 else if (!strcmp(arg, "-z"))
2516 options->line_termination = 0;
2517 else if (!prefixcmp(arg, "-l"))
2518 options->rename_limit = strtoul(arg+2, NULL, 10);
2519 else if (!prefixcmp(arg, "-S"))
2520 options->pickaxe = arg + 2;
2521 else if (!strcmp(arg, "--pickaxe-all"))
2522 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2523 else if (!strcmp(arg, "--pickaxe-regex"))
2524 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2525 else if (!prefixcmp(arg, "-O"))
2526 options->orderfile = arg + 2;
2527 else if (!prefixcmp(arg, "--diff-filter="))
2528 options->filter = arg + 14;
2529 else if (!strcmp(arg, "--abbrev"))
2530 options->abbrev = DEFAULT_ABBREV;
2531 else if (!prefixcmp(arg, "--abbrev=")) {
2532 options->abbrev = strtoul(arg + 9, NULL, 10);
2533 if (options->abbrev < MINIMUM_ABBREV)
2534 options->abbrev = MINIMUM_ABBREV;
2535 else if (40 < options->abbrev)
2536 options->abbrev = 40;
2538 else if (!prefixcmp(arg, "--src-prefix="))
2539 options->a_prefix = arg + 13;
2540 else if (!prefixcmp(arg, "--dst-prefix="))
2541 options->b_prefix = arg + 13;
2542 else if (!strcmp(arg, "--no-prefix"))
2543 options->a_prefix = options->b_prefix = "";
2544 else if (opt_arg(arg, '\0', "inter-hunk-context",
2545 &options->interhunkcontext))
2547 else if (!prefixcmp(arg, "--output=")) {
2548 options->file = fopen(arg + strlen("--output="), "w");
2549 options->close_file = 1;
2550 } else
2551 return 0;
2552 return 1;
2555 static int parse_num(const char **cp_p)
2557 unsigned long num, scale;
2558 int ch, dot;
2559 const char *cp = *cp_p;
2561 num = 0;
2562 scale = 1;
2563 dot = 0;
2564 for(;;) {
2565 ch = *cp;
2566 if ( !dot && ch == '.' ) {
2567 scale = 1;
2568 dot = 1;
2569 } else if ( ch == '%' ) {
2570 scale = dot ? scale*100 : 100;
2571 cp++; /* % is always at the end */
2572 break;
2573 } else if ( ch >= '0' && ch <= '9' ) {
2574 if ( scale < 100000 ) {
2575 scale *= 10;
2576 num = (num*10) + (ch-'0');
2578 } else {
2579 break;
2581 cp++;
2583 *cp_p = cp;
2585 /* user says num divided by scale and we say internally that
2586 * is MAX_SCORE * num / scale.
2588 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2591 static int diff_scoreopt_parse(const char *opt)
2593 int opt1, opt2, cmd;
2595 if (*opt++ != '-')
2596 return -1;
2597 cmd = *opt++;
2598 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2599 return -1; /* that is not a -M, -C nor -B option */
2601 opt1 = parse_num(&opt);
2602 if (cmd != 'B')
2603 opt2 = 0;
2604 else {
2605 if (*opt == 0)
2606 opt2 = 0;
2607 else if (*opt != '/')
2608 return -1; /* we expect -B80/99 or -B80 */
2609 else {
2610 opt++;
2611 opt2 = parse_num(&opt);
2614 if (*opt != 0)
2615 return -1;
2616 return opt1 | (opt2 << 16);
2619 struct diff_queue_struct diff_queued_diff;
2621 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2623 if (queue->alloc <= queue->nr) {
2624 queue->alloc = alloc_nr(queue->alloc);
2625 queue->queue = xrealloc(queue->queue,
2626 sizeof(dp) * queue->alloc);
2628 queue->queue[queue->nr++] = dp;
2631 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2632 struct diff_filespec *one,
2633 struct diff_filespec *two)
2635 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2636 dp->one = one;
2637 dp->two = two;
2638 if (queue)
2639 diff_q(queue, dp);
2640 return dp;
2643 void diff_free_filepair(struct diff_filepair *p)
2645 free_filespec(p->one);
2646 free_filespec(p->two);
2647 free(p);
2650 /* This is different from find_unique_abbrev() in that
2651 * it stuffs the result with dots for alignment.
2653 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2655 int abblen;
2656 const char *abbrev;
2657 if (len == 40)
2658 return sha1_to_hex(sha1);
2660 abbrev = find_unique_abbrev(sha1, len);
2661 abblen = strlen(abbrev);
2662 if (abblen < 37) {
2663 static char hex[41];
2664 if (len < abblen && abblen <= len + 2)
2665 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2666 else
2667 sprintf(hex, "%s...", abbrev);
2668 return hex;
2670 return sha1_to_hex(sha1);
2673 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2675 int line_termination = opt->line_termination;
2676 int inter_name_termination = line_termination ? '\t' : '\0';
2678 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2679 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2680 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2681 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2683 if (p->score) {
2684 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2685 inter_name_termination);
2686 } else {
2687 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2690 if (p->status == DIFF_STATUS_COPIED ||
2691 p->status == DIFF_STATUS_RENAMED) {
2692 const char *name_a, *name_b;
2693 name_a = p->one->path;
2694 name_b = p->two->path;
2695 strip_prefix(opt->prefix_length, &name_a, &name_b);
2696 write_name_quoted(name_a, opt->file, inter_name_termination);
2697 write_name_quoted(name_b, opt->file, line_termination);
2698 } else {
2699 const char *name_a, *name_b;
2700 name_a = p->one->mode ? p->one->path : p->two->path;
2701 name_b = NULL;
2702 strip_prefix(opt->prefix_length, &name_a, &name_b);
2703 write_name_quoted(name_a, opt->file, line_termination);
2707 int diff_unmodified_pair(struct diff_filepair *p)
2709 /* This function is written stricter than necessary to support
2710 * the currently implemented transformers, but the idea is to
2711 * let transformers to produce diff_filepairs any way they want,
2712 * and filter and clean them up here before producing the output.
2714 struct diff_filespec *one = p->one, *two = p->two;
2716 if (DIFF_PAIR_UNMERGED(p))
2717 return 0; /* unmerged is interesting */
2719 /* deletion, addition, mode or type change
2720 * and rename are all interesting.
2722 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2723 DIFF_PAIR_MODE_CHANGED(p) ||
2724 strcmp(one->path, two->path))
2725 return 0;
2727 /* both are valid and point at the same path. that is, we are
2728 * dealing with a change.
2730 if (one->sha1_valid && two->sha1_valid &&
2731 !hashcmp(one->sha1, two->sha1))
2732 return 1; /* no change */
2733 if (!one->sha1_valid && !two->sha1_valid)
2734 return 1; /* both look at the same file on the filesystem. */
2735 return 0;
2738 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2740 if (diff_unmodified_pair(p))
2741 return;
2743 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2744 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2745 return; /* no tree diffs in patch format */
2747 run_diff(p, o);
2750 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2751 struct diffstat_t *diffstat)
2753 if (diff_unmodified_pair(p))
2754 return;
2756 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2757 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2758 return; /* no tree diffs in patch format */
2760 run_diffstat(p, o, diffstat);
2763 static void diff_flush_checkdiff(struct diff_filepair *p,
2764 struct diff_options *o)
2766 if (diff_unmodified_pair(p))
2767 return;
2769 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2770 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2771 return; /* no tree diffs in patch format */
2773 run_checkdiff(p, o);
2776 int diff_queue_is_empty(void)
2778 struct diff_queue_struct *q = &diff_queued_diff;
2779 int i;
2780 for (i = 0; i < q->nr; i++)
2781 if (!diff_unmodified_pair(q->queue[i]))
2782 return 0;
2783 return 1;
2786 #if DIFF_DEBUG
2787 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2789 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2790 x, one ? one : "",
2791 s->path,
2792 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2793 s->mode,
2794 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2795 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2796 x, one ? one : "",
2797 s->size, s->xfrm_flags);
2800 void diff_debug_filepair(const struct diff_filepair *p, int i)
2802 diff_debug_filespec(p->one, i, "one");
2803 diff_debug_filespec(p->two, i, "two");
2804 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2805 p->score, p->status ? p->status : '?',
2806 p->one->rename_used, p->broken_pair);
2809 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2811 int i;
2812 if (msg)
2813 fprintf(stderr, "%s\n", msg);
2814 fprintf(stderr, "q->nr = %d\n", q->nr);
2815 for (i = 0; i < q->nr; i++) {
2816 struct diff_filepair *p = q->queue[i];
2817 diff_debug_filepair(p, i);
2820 #endif
2822 static void diff_resolve_rename_copy(void)
2824 int i;
2825 struct diff_filepair *p;
2826 struct diff_queue_struct *q = &diff_queued_diff;
2828 diff_debug_queue("resolve-rename-copy", q);
2830 for (i = 0; i < q->nr; i++) {
2831 p = q->queue[i];
2832 p->status = 0; /* undecided */
2833 if (DIFF_PAIR_UNMERGED(p))
2834 p->status = DIFF_STATUS_UNMERGED;
2835 else if (!DIFF_FILE_VALID(p->one))
2836 p->status = DIFF_STATUS_ADDED;
2837 else if (!DIFF_FILE_VALID(p->two))
2838 p->status = DIFF_STATUS_DELETED;
2839 else if (DIFF_PAIR_TYPE_CHANGED(p))
2840 p->status = DIFF_STATUS_TYPE_CHANGED;
2842 /* from this point on, we are dealing with a pair
2843 * whose both sides are valid and of the same type, i.e.
2844 * either in-place edit or rename/copy edit.
2846 else if (DIFF_PAIR_RENAME(p)) {
2848 * A rename might have re-connected a broken
2849 * pair up, causing the pathnames to be the
2850 * same again. If so, that's not a rename at
2851 * all, just a modification..
2853 * Otherwise, see if this source was used for
2854 * multiple renames, in which case we decrement
2855 * the count, and call it a copy.
2857 if (!strcmp(p->one->path, p->two->path))
2858 p->status = DIFF_STATUS_MODIFIED;
2859 else if (--p->one->rename_used > 0)
2860 p->status = DIFF_STATUS_COPIED;
2861 else
2862 p->status = DIFF_STATUS_RENAMED;
2864 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2865 p->one->mode != p->two->mode ||
2866 is_null_sha1(p->one->sha1))
2867 p->status = DIFF_STATUS_MODIFIED;
2868 else {
2869 /* This is a "no-change" entry and should not
2870 * happen anymore, but prepare for broken callers.
2872 error("feeding unmodified %s to diffcore",
2873 p->one->path);
2874 p->status = DIFF_STATUS_UNKNOWN;
2877 diff_debug_queue("resolve-rename-copy done", q);
2880 static int check_pair_status(struct diff_filepair *p)
2882 switch (p->status) {
2883 case DIFF_STATUS_UNKNOWN:
2884 return 0;
2885 case 0:
2886 die("internal error in diff-resolve-rename-copy");
2887 default:
2888 return 1;
2892 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2894 int fmt = opt->output_format;
2896 if (fmt & DIFF_FORMAT_CHECKDIFF)
2897 diff_flush_checkdiff(p, opt);
2898 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2899 diff_flush_raw(p, opt);
2900 else if (fmt & DIFF_FORMAT_NAME) {
2901 const char *name_a, *name_b;
2902 name_a = p->two->path;
2903 name_b = NULL;
2904 strip_prefix(opt->prefix_length, &name_a, &name_b);
2905 write_name_quoted(name_a, opt->file, opt->line_termination);
2909 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2911 if (fs->mode)
2912 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2913 else
2914 fprintf(file, " %s ", newdelete);
2915 write_name_quoted(fs->path, file, '\n');
2919 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2921 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2922 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2923 show_name ? ' ' : '\n');
2924 if (show_name) {
2925 write_name_quoted(p->two->path, file, '\n');
2930 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2932 char *names = pprint_rename(p->one->path, p->two->path);
2934 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2935 free(names);
2936 show_mode_change(file, p, 0);
2939 static void diff_summary(FILE *file, struct diff_filepair *p)
2941 switch(p->status) {
2942 case DIFF_STATUS_DELETED:
2943 show_file_mode_name(file, "delete", p->one);
2944 break;
2945 case DIFF_STATUS_ADDED:
2946 show_file_mode_name(file, "create", p->two);
2947 break;
2948 case DIFF_STATUS_COPIED:
2949 show_rename_copy(file, "copy", p);
2950 break;
2951 case DIFF_STATUS_RENAMED:
2952 show_rename_copy(file, "rename", p);
2953 break;
2954 default:
2955 if (p->score) {
2956 fputs(" rewrite ", file);
2957 write_name_quoted(p->two->path, file, ' ');
2958 fprintf(file, "(%d%%)\n", similarity_index(p));
2960 show_mode_change(file, p, !p->score);
2961 break;
2965 struct patch_id_t {
2966 git_SHA_CTX *ctx;
2967 int patchlen;
2970 static int remove_space(char *line, int len)
2972 int i;
2973 char *dst = line;
2974 unsigned char c;
2976 for (i = 0; i < len; i++)
2977 if (!isspace((c = line[i])))
2978 *dst++ = c;
2980 return dst - line;
2983 static void patch_id_consume(void *priv, char *line, unsigned long len)
2985 struct patch_id_t *data = priv;
2986 int new_len;
2988 /* Ignore line numbers when computing the SHA1 of the patch */
2989 if (!prefixcmp(line, "@@ -"))
2990 return;
2992 new_len = remove_space(line, len);
2994 git_SHA1_Update(data->ctx, line, new_len);
2995 data->patchlen += new_len;
2998 /* returns 0 upon success, and writes result into sha1 */
2999 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3001 struct diff_queue_struct *q = &diff_queued_diff;
3002 int i;
3003 git_SHA_CTX ctx;
3004 struct patch_id_t data;
3005 char buffer[PATH_MAX * 4 + 20];
3007 git_SHA1_Init(&ctx);
3008 memset(&data, 0, sizeof(struct patch_id_t));
3009 data.ctx = &ctx;
3011 for (i = 0; i < q->nr; i++) {
3012 xpparam_t xpp;
3013 xdemitconf_t xecfg;
3014 xdemitcb_t ecb;
3015 mmfile_t mf1, mf2;
3016 struct diff_filepair *p = q->queue[i];
3017 int len1, len2;
3019 memset(&xpp, 0, sizeof(xpp));
3020 memset(&xecfg, 0, sizeof(xecfg));
3021 if (p->status == 0)
3022 return error("internal diff status error");
3023 if (p->status == DIFF_STATUS_UNKNOWN)
3024 continue;
3025 if (diff_unmodified_pair(p))
3026 continue;
3027 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3028 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3029 continue;
3030 if (DIFF_PAIR_UNMERGED(p))
3031 continue;
3033 diff_fill_sha1_info(p->one);
3034 diff_fill_sha1_info(p->two);
3035 if (fill_mmfile(&mf1, p->one) < 0 ||
3036 fill_mmfile(&mf2, p->two) < 0)
3037 return error("unable to read files to diff");
3039 len1 = remove_space(p->one->path, strlen(p->one->path));
3040 len2 = remove_space(p->two->path, strlen(p->two->path));
3041 if (p->one->mode == 0)
3042 len1 = snprintf(buffer, sizeof(buffer),
3043 "diff--gita/%.*sb/%.*s"
3044 "newfilemode%06o"
3045 "---/dev/null"
3046 "+++b/%.*s",
3047 len1, p->one->path,
3048 len2, p->two->path,
3049 p->two->mode,
3050 len2, p->two->path);
3051 else if (p->two->mode == 0)
3052 len1 = snprintf(buffer, sizeof(buffer),
3053 "diff--gita/%.*sb/%.*s"
3054 "deletedfilemode%06o"
3055 "---a/%.*s"
3056 "+++/dev/null",
3057 len1, p->one->path,
3058 len2, p->two->path,
3059 p->one->mode,
3060 len1, p->one->path);
3061 else
3062 len1 = snprintf(buffer, sizeof(buffer),
3063 "diff--gita/%.*sb/%.*s"
3064 "---a/%.*s"
3065 "+++b/%.*s",
3066 len1, p->one->path,
3067 len2, p->two->path,
3068 len1, p->one->path,
3069 len2, p->two->path);
3070 git_SHA1_Update(&ctx, buffer, len1);
3072 xpp.flags = XDF_NEED_MINIMAL;
3073 xecfg.ctxlen = 3;
3074 xecfg.flags = XDL_EMIT_FUNCNAMES;
3075 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3076 &xpp, &xecfg, &ecb);
3079 git_SHA1_Final(sha1, &ctx);
3080 return 0;
3083 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3085 struct diff_queue_struct *q = &diff_queued_diff;
3086 int i;
3087 int result = diff_get_patch_id(options, sha1);
3089 for (i = 0; i < q->nr; i++)
3090 diff_free_filepair(q->queue[i]);
3092 free(q->queue);
3093 q->queue = NULL;
3094 q->nr = q->alloc = 0;
3096 return result;
3099 static int is_summary_empty(const struct diff_queue_struct *q)
3101 int i;
3103 for (i = 0; i < q->nr; i++) {
3104 const struct diff_filepair *p = q->queue[i];
3106 switch (p->status) {
3107 case DIFF_STATUS_DELETED:
3108 case DIFF_STATUS_ADDED:
3109 case DIFF_STATUS_COPIED:
3110 case DIFF_STATUS_RENAMED:
3111 return 0;
3112 default:
3113 if (p->score)
3114 return 0;
3115 if (p->one->mode && p->two->mode &&
3116 p->one->mode != p->two->mode)
3117 return 0;
3118 break;
3121 return 1;
3124 void diff_flush(struct diff_options *options)
3126 struct diff_queue_struct *q = &diff_queued_diff;
3127 int i, output_format = options->output_format;
3128 int separator = 0;
3131 * Order: raw, stat, summary, patch
3132 * or: name/name-status/checkdiff (other bits clear)
3134 if (!q->nr)
3135 goto free_queue;
3137 if (output_format & (DIFF_FORMAT_RAW |
3138 DIFF_FORMAT_NAME |
3139 DIFF_FORMAT_NAME_STATUS |
3140 DIFF_FORMAT_CHECKDIFF)) {
3141 for (i = 0; i < q->nr; i++) {
3142 struct diff_filepair *p = q->queue[i];
3143 if (check_pair_status(p))
3144 flush_one_pair(p, options);
3146 separator++;
3149 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3150 struct diffstat_t diffstat;
3152 memset(&diffstat, 0, sizeof(struct diffstat_t));
3153 for (i = 0; i < q->nr; i++) {
3154 struct diff_filepair *p = q->queue[i];
3155 if (check_pair_status(p))
3156 diff_flush_stat(p, options, &diffstat);
3158 if (output_format & DIFF_FORMAT_NUMSTAT)
3159 show_numstat(&diffstat, options);
3160 if (output_format & DIFF_FORMAT_DIFFSTAT)
3161 show_stats(&diffstat, options);
3162 if (output_format & DIFF_FORMAT_SHORTSTAT)
3163 show_shortstats(&diffstat, options);
3164 free_diffstat_info(&diffstat);
3165 separator++;
3167 if (output_format & DIFF_FORMAT_DIRSTAT)
3168 show_dirstat(options);
3170 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3171 for (i = 0; i < q->nr; i++)
3172 diff_summary(options->file, q->queue[i]);
3173 separator++;
3176 if (output_format & DIFF_FORMAT_PATCH) {
3177 if (separator) {
3178 putc(options->line_termination, options->file);
3179 if (options->stat_sep) {
3180 /* attach patch instead of inline */
3181 fputs(options->stat_sep, options->file);
3185 for (i = 0; i < q->nr; i++) {
3186 struct diff_filepair *p = q->queue[i];
3187 if (check_pair_status(p))
3188 diff_flush_patch(p, options);
3192 if (output_format & DIFF_FORMAT_CALLBACK)
3193 options->format_callback(q, options, options->format_callback_data);
3195 for (i = 0; i < q->nr; i++)
3196 diff_free_filepair(q->queue[i]);
3197 free_queue:
3198 free(q->queue);
3199 q->queue = NULL;
3200 q->nr = q->alloc = 0;
3201 if (options->close_file)
3202 fclose(options->file);
3205 static void diffcore_apply_filter(const char *filter)
3207 int i;
3208 struct diff_queue_struct *q = &diff_queued_diff;
3209 struct diff_queue_struct outq;
3210 outq.queue = NULL;
3211 outq.nr = outq.alloc = 0;
3213 if (!filter)
3214 return;
3216 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3217 int found;
3218 for (i = found = 0; !found && i < q->nr; i++) {
3219 struct diff_filepair *p = q->queue[i];
3220 if (((p->status == DIFF_STATUS_MODIFIED) &&
3221 ((p->score &&
3222 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3223 (!p->score &&
3224 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3225 ((p->status != DIFF_STATUS_MODIFIED) &&
3226 strchr(filter, p->status)))
3227 found++;
3229 if (found)
3230 return;
3232 /* otherwise we will clear the whole queue
3233 * by copying the empty outq at the end of this
3234 * function, but first clear the current entries
3235 * in the queue.
3237 for (i = 0; i < q->nr; i++)
3238 diff_free_filepair(q->queue[i]);
3240 else {
3241 /* Only the matching ones */
3242 for (i = 0; i < q->nr; i++) {
3243 struct diff_filepair *p = q->queue[i];
3245 if (((p->status == DIFF_STATUS_MODIFIED) &&
3246 ((p->score &&
3247 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3248 (!p->score &&
3249 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3250 ((p->status != DIFF_STATUS_MODIFIED) &&
3251 strchr(filter, p->status)))
3252 diff_q(&outq, p);
3253 else
3254 diff_free_filepair(p);
3257 free(q->queue);
3258 *q = outq;
3261 /* Check whether two filespecs with the same mode and size are identical */
3262 static int diff_filespec_is_identical(struct diff_filespec *one,
3263 struct diff_filespec *two)
3265 if (S_ISGITLINK(one->mode))
3266 return 0;
3267 if (diff_populate_filespec(one, 0))
3268 return 0;
3269 if (diff_populate_filespec(two, 0))
3270 return 0;
3271 return !memcmp(one->data, two->data, one->size);
3274 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3276 int i;
3277 struct diff_queue_struct *q = &diff_queued_diff;
3278 struct diff_queue_struct outq;
3279 outq.queue = NULL;
3280 outq.nr = outq.alloc = 0;
3282 for (i = 0; i < q->nr; i++) {
3283 struct diff_filepair *p = q->queue[i];
3286 * 1. Entries that come from stat info dirtyness
3287 * always have both sides (iow, not create/delete),
3288 * one side of the object name is unknown, with
3289 * the same mode and size. Keep the ones that
3290 * do not match these criteria. They have real
3291 * differences.
3293 * 2. At this point, the file is known to be modified,
3294 * with the same mode and size, and the object
3295 * name of one side is unknown. Need to inspect
3296 * the identical contents.
3298 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3299 !DIFF_FILE_VALID(p->two) ||
3300 (p->one->sha1_valid && p->two->sha1_valid) ||
3301 (p->one->mode != p->two->mode) ||
3302 diff_populate_filespec(p->one, 1) ||
3303 diff_populate_filespec(p->two, 1) ||
3304 (p->one->size != p->two->size) ||
3305 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3306 diff_q(&outq, p);
3307 else {
3309 * The caller can subtract 1 from skip_stat_unmatch
3310 * to determine how many paths were dirty only
3311 * due to stat info mismatch.
3313 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3314 diffopt->skip_stat_unmatch++;
3315 diff_free_filepair(p);
3318 free(q->queue);
3319 *q = outq;
3322 void diffcore_std(struct diff_options *options)
3324 if (options->skip_stat_unmatch)
3325 diffcore_skip_stat_unmatch(options);
3326 if (options->break_opt != -1)
3327 diffcore_break(options->break_opt);
3328 if (options->detect_rename)
3329 diffcore_rename(options);
3330 if (options->break_opt != -1)
3331 diffcore_merge_broken();
3332 if (options->pickaxe)
3333 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3334 if (options->orderfile)
3335 diffcore_order(options->orderfile);
3336 diff_resolve_rename_copy();
3337 diffcore_apply_filter(options->filter);
3339 if (diff_queued_diff.nr)
3340 DIFF_OPT_SET(options, HAS_CHANGES);
3341 else
3342 DIFF_OPT_CLR(options, HAS_CHANGES);
3345 int diff_result_code(struct diff_options *opt, int status)
3347 int result = 0;
3348 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3349 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3350 return status;
3351 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3352 DIFF_OPT_TST(opt, HAS_CHANGES))
3353 result |= 01;
3354 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3355 DIFF_OPT_TST(opt, CHECK_FAILED))
3356 result |= 02;
3357 return result;
3360 void diff_addremove(struct diff_options *options,
3361 int addremove, unsigned mode,
3362 const unsigned char *sha1,
3363 const char *concatpath)
3365 struct diff_filespec *one, *two;
3367 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3368 return;
3370 /* This may look odd, but it is a preparation for
3371 * feeding "there are unchanged files which should
3372 * not produce diffs, but when you are doing copy
3373 * detection you would need them, so here they are"
3374 * entries to the diff-core. They will be prefixed
3375 * with something like '=' or '*' (I haven't decided
3376 * which but should not make any difference).
3377 * Feeding the same new and old to diff_change()
3378 * also has the same effect.
3379 * Before the final output happens, they are pruned after
3380 * merged into rename/copy pairs as appropriate.
3382 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3383 addremove = (addremove == '+' ? '-' :
3384 addremove == '-' ? '+' : addremove);
3386 if (options->prefix &&
3387 strncmp(concatpath, options->prefix, options->prefix_length))
3388 return;
3390 one = alloc_filespec(concatpath);
3391 two = alloc_filespec(concatpath);
3393 if (addremove != '+')
3394 fill_filespec(one, sha1, mode);
3395 if (addremove != '-')
3396 fill_filespec(two, sha1, mode);
3398 diff_queue(&diff_queued_diff, one, two);
3399 DIFF_OPT_SET(options, HAS_CHANGES);
3402 void diff_change(struct diff_options *options,
3403 unsigned old_mode, unsigned new_mode,
3404 const unsigned char *old_sha1,
3405 const unsigned char *new_sha1,
3406 const char *concatpath)
3408 struct diff_filespec *one, *two;
3410 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3411 && S_ISGITLINK(new_mode))
3412 return;
3414 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3415 unsigned tmp;
3416 const unsigned char *tmp_c;
3417 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3418 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3421 if (options->prefix &&
3422 strncmp(concatpath, options->prefix, options->prefix_length))
3423 return;
3425 one = alloc_filespec(concatpath);
3426 two = alloc_filespec(concatpath);
3427 fill_filespec(one, old_sha1, old_mode);
3428 fill_filespec(two, new_sha1, new_mode);
3430 diff_queue(&diff_queued_diff, one, two);
3431 DIFF_OPT_SET(options, HAS_CHANGES);
3434 void diff_unmerge(struct diff_options *options,
3435 const char *path,
3436 unsigned mode, const unsigned char *sha1)
3438 struct diff_filespec *one, *two;
3440 if (options->prefix &&
3441 strncmp(path, options->prefix, options->prefix_length))
3442 return;
3444 one = alloc_filespec(path);
3445 two = alloc_filespec(path);
3446 fill_filespec(one, sha1, mode);
3447 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3450 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3451 size_t *outsize)
3453 struct diff_tempfile temp;
3454 const char *argv[3];
3455 const char **arg = argv;
3456 struct child_process child;
3457 struct strbuf buf = STRBUF_INIT;
3459 prepare_temp_file(spec->path, &temp, spec);
3460 *arg++ = pgm;
3461 *arg++ = temp.name;
3462 *arg = NULL;
3464 memset(&child, 0, sizeof(child));
3465 child.argv = argv;
3466 child.out = -1;
3467 if (start_command(&child) != 0 ||
3468 strbuf_read(&buf, child.out, 0) < 0 ||
3469 finish_command(&child) != 0) {
3470 if (temp.name == temp.tmp_path)
3471 unlink(temp.name);
3472 error("error running textconv command '%s'", pgm);
3473 return NULL;
3475 if (temp.name == temp.tmp_path)
3476 unlink(temp.name);
3478 return strbuf_detach(&buf, outsize);