Merge git://git.kernel.org/pub/scm/gitk/gitk
[git/gitweb.git] / diff.c
blob33153787b8117396cf906e69e656849ac04f3257
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "quote.h"
9 #include "diff.h"
10 #include "diffcore.h"
11 #include "delta.h"
12 #include "xdiff-interface.h"
13 #include "color.h"
15 static int use_size_cache;
17 static int diff_detect_rename_default;
18 static int diff_rename_limit_default = -1;
19 static int diff_use_color_default;
21 static char diff_colors[][COLOR_MAXLEN] = {
22 "\033[m", /* reset */
23 "", /* PLAIN (normal) */
24 "\033[1m", /* METAINFO (bold) */
25 "\033[36m", /* FRAGINFO (cyan) */
26 "\033[31m", /* OLD (red) */
27 "\033[32m", /* NEW (green) */
28 "\033[33m", /* COMMIT (yellow) */
29 "\033[41m", /* WHITESPACE (red background) */
32 static int parse_diff_color_slot(const char *var, int ofs)
34 if (!strcasecmp(var+ofs, "plain"))
35 return DIFF_PLAIN;
36 if (!strcasecmp(var+ofs, "meta"))
37 return DIFF_METAINFO;
38 if (!strcasecmp(var+ofs, "frag"))
39 return DIFF_FRAGINFO;
40 if (!strcasecmp(var+ofs, "old"))
41 return DIFF_FILE_OLD;
42 if (!strcasecmp(var+ofs, "new"))
43 return DIFF_FILE_NEW;
44 if (!strcasecmp(var+ofs, "commit"))
45 return DIFF_COMMIT;
46 if (!strcasecmp(var+ofs, "whitespace"))
47 return DIFF_WHITESPACE;
48 die("bad config variable '%s'", var);
52 * These are to give UI layer defaults.
53 * The core-level commands such as git-diff-files should
54 * never be affected by the setting of diff.renames
55 * the user happens to have in the configuration file.
57 int git_diff_ui_config(const char *var, const char *value)
59 if (!strcmp(var, "diff.renamelimit")) {
60 diff_rename_limit_default = git_config_int(var, value);
61 return 0;
63 if (!strcmp(var, "diff.color")) {
64 diff_use_color_default = git_config_colorbool(var, value);
65 return 0;
67 if (!strcmp(var, "diff.renames")) {
68 if (!value)
69 diff_detect_rename_default = DIFF_DETECT_RENAME;
70 else if (!strcasecmp(value, "copies") ||
71 !strcasecmp(value, "copy"))
72 diff_detect_rename_default = DIFF_DETECT_COPY;
73 else if (git_config_bool(var,value))
74 diff_detect_rename_default = DIFF_DETECT_RENAME;
75 return 0;
77 if (!strncmp(var, "diff.color.", 11)) {
78 int slot = parse_diff_color_slot(var, 11);
79 color_parse(value, var, diff_colors[slot]);
80 return 0;
82 return git_default_config(var, value);
85 static char *quote_one(const char *str)
87 int needlen;
88 char *xp;
90 if (!str)
91 return NULL;
92 needlen = quote_c_style(str, NULL, NULL, 0);
93 if (!needlen)
94 return xstrdup(str);
95 xp = xmalloc(needlen + 1);
96 quote_c_style(str, xp, NULL, 0);
97 return xp;
100 static char *quote_two(const char *one, const char *two)
102 int need_one = quote_c_style(one, NULL, NULL, 1);
103 int need_two = quote_c_style(two, NULL, NULL, 1);
104 char *xp;
106 if (need_one + need_two) {
107 if (!need_one) need_one = strlen(one);
108 if (!need_two) need_one = strlen(two);
110 xp = xmalloc(need_one + need_two + 3);
111 xp[0] = '"';
112 quote_c_style(one, xp + 1, NULL, 1);
113 quote_c_style(two, xp + need_one + 1, NULL, 1);
114 strcpy(xp + need_one + need_two + 1, "\"");
115 return xp;
117 need_one = strlen(one);
118 need_two = strlen(two);
119 xp = xmalloc(need_one + need_two + 1);
120 strcpy(xp, one);
121 strcpy(xp + need_one, two);
122 return xp;
125 static const char *external_diff(void)
127 static const char *external_diff_cmd = NULL;
128 static int done_preparing = 0;
130 if (done_preparing)
131 return external_diff_cmd;
132 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
133 done_preparing = 1;
134 return external_diff_cmd;
137 #define TEMPFILE_PATH_LEN 50
139 static struct diff_tempfile {
140 const char *name; /* filename external diff should read from */
141 char hex[41];
142 char mode[10];
143 char tmp_path[TEMPFILE_PATH_LEN];
144 } diff_temp[2];
146 static int count_lines(const char *data, int size)
148 int count, ch, completely_empty = 1, nl_just_seen = 0;
149 count = 0;
150 while (0 < size--) {
151 ch = *data++;
152 if (ch == '\n') {
153 count++;
154 nl_just_seen = 1;
155 completely_empty = 0;
157 else {
158 nl_just_seen = 0;
159 completely_empty = 0;
162 if (completely_empty)
163 return 0;
164 if (!nl_just_seen)
165 count++; /* no trailing newline */
166 return count;
169 static void print_line_count(int count)
171 switch (count) {
172 case 0:
173 printf("0,0");
174 break;
175 case 1:
176 printf("1");
177 break;
178 default:
179 printf("1,%d", count);
180 break;
184 static void copy_file(int prefix, const char *data, int size)
186 int ch, nl_just_seen = 1;
187 while (0 < size--) {
188 ch = *data++;
189 if (nl_just_seen)
190 putchar(prefix);
191 putchar(ch);
192 if (ch == '\n')
193 nl_just_seen = 1;
194 else
195 nl_just_seen = 0;
197 if (!nl_just_seen)
198 printf("\n\\ No newline at end of file\n");
201 static void emit_rewrite_diff(const char *name_a,
202 const char *name_b,
203 struct diff_filespec *one,
204 struct diff_filespec *two)
206 int lc_a, lc_b;
207 diff_populate_filespec(one, 0);
208 diff_populate_filespec(two, 0);
209 lc_a = count_lines(one->data, one->size);
210 lc_b = count_lines(two->data, two->size);
211 printf("--- a/%s\n+++ b/%s\n@@ -", name_a, name_b);
212 print_line_count(lc_a);
213 printf(" +");
214 print_line_count(lc_b);
215 printf(" @@\n");
216 if (lc_a)
217 copy_file('-', one->data, one->size);
218 if (lc_b)
219 copy_file('+', two->data, two->size);
222 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
224 if (!DIFF_FILE_VALID(one)) {
225 mf->ptr = (char *)""; /* does not matter */
226 mf->size = 0;
227 return 0;
229 else if (diff_populate_filespec(one, 0))
230 return -1;
231 mf->ptr = one->data;
232 mf->size = one->size;
233 return 0;
236 struct diff_words_buffer {
237 mmfile_t text;
238 long alloc;
239 long current; /* output pointer */
240 int suppressed_newline;
243 static void diff_words_append(char *line, unsigned long len,
244 struct diff_words_buffer *buffer)
246 if (buffer->text.size + len > buffer->alloc) {
247 buffer->alloc = (buffer->text.size + len) * 3 / 2;
248 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
250 line++;
251 len--;
252 memcpy(buffer->text.ptr + buffer->text.size, line, len);
253 buffer->text.size += len;
256 struct diff_words_data {
257 struct xdiff_emit_state xm;
258 struct diff_words_buffer minus, plus;
261 static void print_word(struct diff_words_buffer *buffer, int len, int color,
262 int suppress_newline)
264 const char *ptr;
265 int eol = 0;
267 if (len == 0)
268 return;
270 ptr = buffer->text.ptr + buffer->current;
271 buffer->current += len;
273 if (ptr[len - 1] == '\n') {
274 eol = 1;
275 len--;
278 fputs(diff_get_color(1, color), stdout);
279 fwrite(ptr, len, 1, stdout);
280 fputs(diff_get_color(1, DIFF_RESET), stdout);
282 if (eol) {
283 if (suppress_newline)
284 buffer->suppressed_newline = 1;
285 else
286 putchar('\n');
290 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
292 struct diff_words_data *diff_words = priv;
294 if (diff_words->minus.suppressed_newline) {
295 if (line[0] != '+')
296 putchar('\n');
297 diff_words->minus.suppressed_newline = 0;
300 len--;
301 switch (line[0]) {
302 case '-':
303 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
304 break;
305 case '+':
306 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
307 break;
308 case ' ':
309 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
310 diff_words->minus.current += len;
311 break;
315 /* this executes the word diff on the accumulated buffers */
316 static void diff_words_show(struct diff_words_data *diff_words)
318 xpparam_t xpp;
319 xdemitconf_t xecfg;
320 xdemitcb_t ecb;
321 mmfile_t minus, plus;
322 int i;
324 minus.size = diff_words->minus.text.size;
325 minus.ptr = xmalloc(minus.size);
326 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
327 for (i = 0; i < minus.size; i++)
328 if (isspace(minus.ptr[i]))
329 minus.ptr[i] = '\n';
330 diff_words->minus.current = 0;
332 plus.size = diff_words->plus.text.size;
333 plus.ptr = xmalloc(plus.size);
334 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
335 for (i = 0; i < plus.size; i++)
336 if (isspace(plus.ptr[i]))
337 plus.ptr[i] = '\n';
338 diff_words->plus.current = 0;
340 xpp.flags = XDF_NEED_MINIMAL;
341 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
342 xecfg.flags = 0;
343 ecb.outf = xdiff_outf;
344 ecb.priv = diff_words;
345 diff_words->xm.consume = fn_out_diff_words_aux;
346 xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
348 free(minus.ptr);
349 free(plus.ptr);
350 diff_words->minus.text.size = diff_words->plus.text.size = 0;
352 if (diff_words->minus.suppressed_newline) {
353 putchar('\n');
354 diff_words->minus.suppressed_newline = 0;
358 struct emit_callback {
359 struct xdiff_emit_state xm;
360 int nparents, color_diff;
361 const char **label_path;
362 struct diff_words_data *diff_words;
365 static void free_diff_words_data(struct emit_callback *ecbdata)
367 if (ecbdata->diff_words) {
368 /* flush buffers */
369 if (ecbdata->diff_words->minus.text.size ||
370 ecbdata->diff_words->plus.text.size)
371 diff_words_show(ecbdata->diff_words);
373 if (ecbdata->diff_words->minus.text.ptr)
374 free (ecbdata->diff_words->minus.text.ptr);
375 if (ecbdata->diff_words->plus.text.ptr)
376 free (ecbdata->diff_words->plus.text.ptr);
377 free(ecbdata->diff_words);
378 ecbdata->diff_words = NULL;
382 const char *diff_get_color(int diff_use_color, enum color_diff ix)
384 if (diff_use_color)
385 return diff_colors[ix];
386 return "";
389 static void emit_line(const char *set, const char *reset, const char *line, int len)
391 if (len > 0 && line[len-1] == '\n')
392 len--;
393 fputs(set, stdout);
394 fwrite(line, len, 1, stdout);
395 puts(reset);
398 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
400 int col0 = ecbdata->nparents;
401 int last_tab_in_indent = -1;
402 int last_space_in_indent = -1;
403 int i;
404 int tail = len;
405 int need_highlight_leading_space = 0;
406 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
407 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
409 if (!*ws) {
410 emit_line(set, reset, line, len);
411 return;
414 /* The line is a newly added line. Does it have funny leading
415 * whitespaces? In indent, SP should never precede a TAB.
417 for (i = col0; i < len; i++) {
418 if (line[i] == '\t') {
419 last_tab_in_indent = i;
420 if (0 <= last_space_in_indent)
421 need_highlight_leading_space = 1;
423 else if (line[i] == ' ')
424 last_space_in_indent = i;
425 else
426 break;
428 fputs(set, stdout);
429 fwrite(line, col0, 1, stdout);
430 fputs(reset, stdout);
431 if (((i == len) || line[i] == '\n') && i != col0) {
432 /* The whole line was indent */
433 emit_line(ws, reset, line + col0, len - col0);
434 return;
436 i = col0;
437 if (need_highlight_leading_space) {
438 while (i < last_tab_in_indent) {
439 if (line[i] == ' ') {
440 fputs(ws, stdout);
441 putchar(' ');
442 fputs(reset, stdout);
444 else
445 putchar(line[i]);
446 i++;
449 tail = len - 1;
450 if (line[tail] == '\n' && i < tail)
451 tail--;
452 while (i < tail) {
453 if (!isspace(line[tail]))
454 break;
455 tail--;
457 if ((i < tail && line[tail + 1] != '\n')) {
458 /* This has whitespace between tail+1..len */
459 fputs(set, stdout);
460 fwrite(line + i, tail - i + 1, 1, stdout);
461 fputs(reset, stdout);
462 emit_line(ws, reset, line + tail + 1, len - tail - 1);
464 else
465 emit_line(set, reset, line + i, len - i);
468 static void fn_out_consume(void *priv, char *line, unsigned long len)
470 int i;
471 int color;
472 struct emit_callback *ecbdata = priv;
473 const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
474 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
476 if (ecbdata->label_path[0]) {
477 printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
478 printf("%s+++ %s%s\n", set, ecbdata->label_path[1], reset);
479 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
482 /* This is not really necessary for now because
483 * this codepath only deals with two-way diffs.
485 for (i = 0; i < len && line[i] == '@'; i++)
487 if (2 <= i && i < len && line[i] == ' ') {
488 ecbdata->nparents = i - 1;
489 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
490 reset, line, len);
491 return;
494 if (len < ecbdata->nparents) {
495 set = reset;
496 emit_line(reset, reset, line, len);
497 return;
500 color = DIFF_PLAIN;
501 if (ecbdata->diff_words && ecbdata->nparents != 1)
502 /* fall back to normal diff */
503 free_diff_words_data(ecbdata);
504 if (ecbdata->diff_words) {
505 if (line[0] == '-') {
506 diff_words_append(line, len,
507 &ecbdata->diff_words->minus);
508 return;
509 } else if (line[0] == '+') {
510 diff_words_append(line, len,
511 &ecbdata->diff_words->plus);
512 return;
514 if (ecbdata->diff_words->minus.text.size ||
515 ecbdata->diff_words->plus.text.size)
516 diff_words_show(ecbdata->diff_words);
517 line++;
518 len--;
519 emit_line(set, reset, line, len);
520 return;
522 for (i = 0; i < ecbdata->nparents && len; i++) {
523 if (line[i] == '-')
524 color = DIFF_FILE_OLD;
525 else if (line[i] == '+')
526 color = DIFF_FILE_NEW;
529 if (color != DIFF_FILE_NEW) {
530 emit_line(diff_get_color(ecbdata->color_diff, color),
531 reset, line, len);
532 return;
534 emit_add_line(reset, ecbdata, line, len);
537 static char *pprint_rename(const char *a, const char *b)
539 const char *old = a;
540 const char *new = b;
541 char *name = NULL;
542 int pfx_length, sfx_length;
543 int len_a = strlen(a);
544 int len_b = strlen(b);
546 /* Find common prefix */
547 pfx_length = 0;
548 while (*old && *new && *old == *new) {
549 if (*old == '/')
550 pfx_length = old - a + 1;
551 old++;
552 new++;
555 /* Find common suffix */
556 old = a + len_a;
557 new = b + len_b;
558 sfx_length = 0;
559 while (a <= old && b <= new && *old == *new) {
560 if (*old == '/')
561 sfx_length = len_a - (old - a);
562 old--;
563 new--;
567 * pfx{mid-a => mid-b}sfx
568 * {pfx-a => pfx-b}sfx
569 * pfx{sfx-a => sfx-b}
570 * name-a => name-b
572 if (pfx_length + sfx_length) {
573 int a_midlen = len_a - pfx_length - sfx_length;
574 int b_midlen = len_b - pfx_length - sfx_length;
575 if (a_midlen < 0) a_midlen = 0;
576 if (b_midlen < 0) b_midlen = 0;
578 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
579 sprintf(name, "%.*s{%.*s => %.*s}%s",
580 pfx_length, a,
581 a_midlen, a + pfx_length,
582 b_midlen, b + pfx_length,
583 a + len_a - sfx_length);
585 else {
586 name = xmalloc(len_a + len_b + 5);
587 sprintf(name, "%s => %s", a, b);
589 return name;
592 struct diffstat_t {
593 struct xdiff_emit_state xm;
595 int nr;
596 int alloc;
597 struct diffstat_file {
598 char *name;
599 unsigned is_unmerged:1;
600 unsigned is_binary:1;
601 unsigned is_renamed:1;
602 unsigned int added, deleted;
603 } **files;
606 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
607 const char *name_a,
608 const char *name_b)
610 struct diffstat_file *x;
611 x = xcalloc(sizeof (*x), 1);
612 if (diffstat->nr == diffstat->alloc) {
613 diffstat->alloc = alloc_nr(diffstat->alloc);
614 diffstat->files = xrealloc(diffstat->files,
615 diffstat->alloc * sizeof(x));
617 diffstat->files[diffstat->nr++] = x;
618 if (name_b) {
619 x->name = pprint_rename(name_a, name_b);
620 x->is_renamed = 1;
622 else
623 x->name = xstrdup(name_a);
624 return x;
627 static void diffstat_consume(void *priv, char *line, unsigned long len)
629 struct diffstat_t *diffstat = priv;
630 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
632 if (line[0] == '+')
633 x->added++;
634 else if (line[0] == '-')
635 x->deleted++;
638 const char mime_boundary_leader[] = "------------";
640 static int scale_linear(int it, int width, int max_change)
643 * make sure that at least one '-' is printed if there were deletions,
644 * and likewise for '+'.
646 if (max_change < 2)
647 return it;
648 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
651 static void show_name(const char *prefix, const char *name, int len,
652 const char *reset, const char *set)
654 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
657 static void show_graph(char ch, int cnt, const char *set, const char *reset)
659 if (cnt <= 0)
660 return;
661 printf("%s", set);
662 while (cnt--)
663 putchar(ch);
664 printf("%s", reset);
667 static void show_stats(struct diffstat_t* data, struct diff_options *options)
669 int i, len, add, del, total, adds = 0, dels = 0;
670 int max_change = 0, max_len = 0;
671 int total_files = data->nr;
672 int width, name_width;
673 const char *reset, *set, *add_c, *del_c;
675 if (data->nr == 0)
676 return;
678 width = options->stat_width ? options->stat_width : 80;
679 name_width = options->stat_name_width ? options->stat_name_width : 50;
681 /* Sanity: give at least 5 columns to the graph,
682 * but leave at least 10 columns for the name.
684 if (width < name_width + 15) {
685 if (name_width <= 25)
686 width = name_width + 15;
687 else
688 name_width = width - 15;
691 /* Find the longest filename and max number of changes */
692 reset = diff_get_color(options->color_diff, DIFF_RESET);
693 set = diff_get_color(options->color_diff, DIFF_PLAIN);
694 add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
695 del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
697 for (i = 0; i < data->nr; i++) {
698 struct diffstat_file *file = data->files[i];
699 int change = file->added + file->deleted;
701 len = quote_c_style(file->name, NULL, NULL, 0);
702 if (len) {
703 char *qname = xmalloc(len + 1);
704 quote_c_style(file->name, qname, NULL, 0);
705 free(file->name);
706 file->name = qname;
709 len = strlen(file->name);
710 if (max_len < len)
711 max_len = len;
713 if (file->is_binary || file->is_unmerged)
714 continue;
715 if (max_change < change)
716 max_change = change;
719 /* Compute the width of the graph part;
720 * 10 is for one blank at the beginning of the line plus
721 * " | count " between the name and the graph.
723 * From here on, name_width is the width of the name area,
724 * and width is the width of the graph area.
726 name_width = (name_width < max_len) ? name_width : max_len;
727 if (width < (name_width + 10) + max_change)
728 width = width - (name_width + 10);
729 else
730 width = max_change;
732 for (i = 0; i < data->nr; i++) {
733 const char *prefix = "";
734 char *name = data->files[i]->name;
735 int added = data->files[i]->added;
736 int deleted = data->files[i]->deleted;
737 int name_len;
740 * "scale" the filename
742 len = name_width;
743 name_len = strlen(name);
744 if (name_width < name_len) {
745 char *slash;
746 prefix = "...";
747 len -= 3;
748 name += name_len - len;
749 slash = strchr(name, '/');
750 if (slash)
751 name = slash;
754 if (data->files[i]->is_binary) {
755 show_name(prefix, name, len, reset, set);
756 printf(" Bin\n");
757 goto free_diffstat_file;
759 else if (data->files[i]->is_unmerged) {
760 show_name(prefix, name, len, reset, set);
761 printf(" Unmerged\n");
762 goto free_diffstat_file;
764 else if (!data->files[i]->is_renamed &&
765 (added + deleted == 0)) {
766 total_files--;
767 goto free_diffstat_file;
771 * scale the add/delete
773 add = added;
774 del = deleted;
775 total = add + del;
776 adds += add;
777 dels += del;
779 if (width <= max_change) {
780 add = scale_linear(add, width, max_change);
781 del = scale_linear(del, width, max_change);
782 total = add + del;
784 show_name(prefix, name, len, reset, set);
785 printf("%5d ", added + deleted);
786 show_graph('+', add, add_c, reset);
787 show_graph('-', del, del_c, reset);
788 putchar('\n');
789 free_diffstat_file:
790 free(data->files[i]->name);
791 free(data->files[i]);
793 free(data->files);
794 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
795 set, total_files, adds, dels, reset);
798 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
800 int i;
802 for (i = 0; i < data->nr; i++) {
803 struct diffstat_file *file = data->files[i];
805 printf("%d\t%d\t", file->added, file->deleted);
806 if (options->line_termination &&
807 quote_c_style(file->name, NULL, NULL, 0))
808 quote_c_style(file->name, NULL, stdout, 0);
809 else
810 fputs(file->name, stdout);
811 putchar(options->line_termination);
815 struct checkdiff_t {
816 struct xdiff_emit_state xm;
817 const char *filename;
818 int lineno;
821 static void checkdiff_consume(void *priv, char *line, unsigned long len)
823 struct checkdiff_t *data = priv;
825 if (line[0] == '+') {
826 int i, spaces = 0;
828 data->lineno++;
830 /* check space before tab */
831 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
832 if (line[i] == ' ')
833 spaces++;
834 if (line[i - 1] == '\t' && spaces)
835 printf("%s:%d: space before tab:%.*s\n",
836 data->filename, data->lineno, (int)len, line);
838 /* check white space at line end */
839 if (line[len - 1] == '\n')
840 len--;
841 if (isspace(line[len - 1]))
842 printf("%s:%d: white space at end: %.*s\n",
843 data->filename, data->lineno, (int)len, line);
844 } else if (line[0] == ' ')
845 data->lineno++;
846 else if (line[0] == '@') {
847 char *plus = strchr(line, '+');
848 if (plus)
849 data->lineno = strtol(plus, NULL, 10);
850 else
851 die("invalid diff");
855 static unsigned char *deflate_it(char *data,
856 unsigned long size,
857 unsigned long *result_size)
859 int bound;
860 unsigned char *deflated;
861 z_stream stream;
863 memset(&stream, 0, sizeof(stream));
864 deflateInit(&stream, zlib_compression_level);
865 bound = deflateBound(&stream, size);
866 deflated = xmalloc(bound);
867 stream.next_out = deflated;
868 stream.avail_out = bound;
870 stream.next_in = (unsigned char *)data;
871 stream.avail_in = size;
872 while (deflate(&stream, Z_FINISH) == Z_OK)
873 ; /* nothing */
874 deflateEnd(&stream);
875 *result_size = stream.total_out;
876 return deflated;
879 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
881 void *cp;
882 void *delta;
883 void *deflated;
884 void *data;
885 unsigned long orig_size;
886 unsigned long delta_size;
887 unsigned long deflate_size;
888 unsigned long data_size;
890 /* We could do deflated delta, or we could do just deflated two,
891 * whichever is smaller.
893 delta = NULL;
894 deflated = deflate_it(two->ptr, two->size, &deflate_size);
895 if (one->size && two->size) {
896 delta = diff_delta(one->ptr, one->size,
897 two->ptr, two->size,
898 &delta_size, deflate_size);
899 if (delta) {
900 void *to_free = delta;
901 orig_size = delta_size;
902 delta = deflate_it(delta, delta_size, &delta_size);
903 free(to_free);
907 if (delta && delta_size < deflate_size) {
908 printf("delta %lu\n", orig_size);
909 free(deflated);
910 data = delta;
911 data_size = delta_size;
913 else {
914 printf("literal %lu\n", two->size);
915 free(delta);
916 data = deflated;
917 data_size = deflate_size;
920 /* emit data encoded in base85 */
921 cp = data;
922 while (data_size) {
923 int bytes = (52 < data_size) ? 52 : data_size;
924 char line[70];
925 data_size -= bytes;
926 if (bytes <= 26)
927 line[0] = bytes + 'A' - 1;
928 else
929 line[0] = bytes - 26 + 'a' - 1;
930 encode_85(line + 1, cp, bytes);
931 cp = (char *) cp + bytes;
932 puts(line);
934 printf("\n");
935 free(data);
938 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
940 printf("GIT binary patch\n");
941 emit_binary_diff_body(one, two);
942 emit_binary_diff_body(two, one);
945 #define FIRST_FEW_BYTES 8000
946 static int mmfile_is_binary(mmfile_t *mf)
948 long sz = mf->size;
949 if (FIRST_FEW_BYTES < sz)
950 sz = FIRST_FEW_BYTES;
951 return !!memchr(mf->ptr, 0, sz);
954 static void builtin_diff(const char *name_a,
955 const char *name_b,
956 struct diff_filespec *one,
957 struct diff_filespec *two,
958 const char *xfrm_msg,
959 struct diff_options *o,
960 int complete_rewrite)
962 mmfile_t mf1, mf2;
963 const char *lbl[2];
964 char *a_one, *b_two;
965 const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
966 const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
968 a_one = quote_two("a/", name_a);
969 b_two = quote_two("b/", name_b);
970 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
971 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
972 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
973 if (lbl[0][0] == '/') {
974 /* /dev/null */
975 printf("%snew file mode %06o%s\n", set, two->mode, reset);
976 if (xfrm_msg && xfrm_msg[0])
977 printf("%s%s%s\n", set, xfrm_msg, reset);
979 else if (lbl[1][0] == '/') {
980 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
981 if (xfrm_msg && xfrm_msg[0])
982 printf("%s%s%s\n", set, xfrm_msg, reset);
984 else {
985 if (one->mode != two->mode) {
986 printf("%sold mode %06o%s\n", set, one->mode, reset);
987 printf("%snew mode %06o%s\n", set, two->mode, reset);
989 if (xfrm_msg && xfrm_msg[0])
990 printf("%s%s%s\n", set, xfrm_msg, reset);
992 * we do not run diff between different kind
993 * of objects.
995 if ((one->mode ^ two->mode) & S_IFMT)
996 goto free_ab_and_return;
997 if (complete_rewrite) {
998 emit_rewrite_diff(name_a, name_b, one, two);
999 goto free_ab_and_return;
1003 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1004 die("unable to read files to diff");
1006 if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
1007 /* Quite common confusing case */
1008 if (mf1.size == mf2.size &&
1009 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1010 goto free_ab_and_return;
1011 if (o->binary)
1012 emit_binary_diff(&mf1, &mf2);
1013 else
1014 printf("Binary files %s and %s differ\n",
1015 lbl[0], lbl[1]);
1017 else {
1018 /* Crazy xdl interfaces.. */
1019 const char *diffopts = getenv("GIT_DIFF_OPTS");
1020 xpparam_t xpp;
1021 xdemitconf_t xecfg;
1022 xdemitcb_t ecb;
1023 struct emit_callback ecbdata;
1025 memset(&ecbdata, 0, sizeof(ecbdata));
1026 ecbdata.label_path = lbl;
1027 ecbdata.color_diff = o->color_diff;
1028 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1029 xecfg.ctxlen = o->context;
1030 xecfg.flags = XDL_EMIT_FUNCNAMES;
1031 if (!diffopts)
1033 else if (!strncmp(diffopts, "--unified=", 10))
1034 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1035 else if (!strncmp(diffopts, "-u", 2))
1036 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1037 ecb.outf = xdiff_outf;
1038 ecb.priv = &ecbdata;
1039 ecbdata.xm.consume = fn_out_consume;
1040 if (o->color_diff_words)
1041 ecbdata.diff_words =
1042 xcalloc(1, sizeof(struct diff_words_data));
1043 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1044 if (o->color_diff_words)
1045 free_diff_words_data(&ecbdata);
1048 free_ab_and_return:
1049 free(a_one);
1050 free(b_two);
1051 return;
1054 static void builtin_diffstat(const char *name_a, const char *name_b,
1055 struct diff_filespec *one,
1056 struct diff_filespec *two,
1057 struct diffstat_t *diffstat,
1058 struct diff_options *o,
1059 int complete_rewrite)
1061 mmfile_t mf1, mf2;
1062 struct diffstat_file *data;
1064 data = diffstat_add(diffstat, name_a, name_b);
1066 if (!one || !two) {
1067 data->is_unmerged = 1;
1068 return;
1070 if (complete_rewrite) {
1071 diff_populate_filespec(one, 0);
1072 diff_populate_filespec(two, 0);
1073 data->deleted = count_lines(one->data, one->size);
1074 data->added = count_lines(two->data, two->size);
1075 return;
1077 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1078 die("unable to read files to diff");
1080 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
1081 data->is_binary = 1;
1082 else {
1083 /* Crazy xdl interfaces.. */
1084 xpparam_t xpp;
1085 xdemitconf_t xecfg;
1086 xdemitcb_t ecb;
1088 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1089 xecfg.ctxlen = 0;
1090 xecfg.flags = 0;
1091 ecb.outf = xdiff_outf;
1092 ecb.priv = diffstat;
1093 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1097 static void builtin_checkdiff(const char *name_a, const char *name_b,
1098 struct diff_filespec *one,
1099 struct diff_filespec *two)
1101 mmfile_t mf1, mf2;
1102 struct checkdiff_t data;
1104 if (!two)
1105 return;
1107 memset(&data, 0, sizeof(data));
1108 data.xm.consume = checkdiff_consume;
1109 data.filename = name_b ? name_b : name_a;
1110 data.lineno = 0;
1112 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1113 die("unable to read files to diff");
1115 if (mmfile_is_binary(&mf2))
1116 return;
1117 else {
1118 /* Crazy xdl interfaces.. */
1119 xpparam_t xpp;
1120 xdemitconf_t xecfg;
1121 xdemitcb_t ecb;
1123 xpp.flags = XDF_NEED_MINIMAL;
1124 xecfg.ctxlen = 0;
1125 xecfg.flags = 0;
1126 ecb.outf = xdiff_outf;
1127 ecb.priv = &data;
1128 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1132 struct diff_filespec *alloc_filespec(const char *path)
1134 int namelen = strlen(path);
1135 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1137 memset(spec, 0, sizeof(*spec));
1138 spec->path = (char *)(spec + 1);
1139 memcpy(spec->path, path, namelen+1);
1140 return spec;
1143 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1144 unsigned short mode)
1146 if (mode) {
1147 spec->mode = canon_mode(mode);
1148 hashcpy(spec->sha1, sha1);
1149 spec->sha1_valid = !is_null_sha1(sha1);
1154 * Given a name and sha1 pair, if the dircache tells us the file in
1155 * the work tree has that object contents, return true, so that
1156 * prepare_temp_file() does not have to inflate and extract.
1158 static int work_tree_matches(const char *name, const unsigned char *sha1)
1160 struct cache_entry *ce;
1161 struct stat st;
1162 int pos, len;
1164 /* We do not read the cache ourselves here, because the
1165 * benchmark with my previous version that always reads cache
1166 * shows that it makes things worse for diff-tree comparing
1167 * two linux-2.6 kernel trees in an already checked out work
1168 * tree. This is because most diff-tree comparisons deal with
1169 * only a small number of files, while reading the cache is
1170 * expensive for a large project, and its cost outweighs the
1171 * savings we get by not inflating the object to a temporary
1172 * file. Practically, this code only helps when we are used
1173 * by diff-cache --cached, which does read the cache before
1174 * calling us.
1176 if (!active_cache)
1177 return 0;
1179 len = strlen(name);
1180 pos = cache_name_pos(name, len);
1181 if (pos < 0)
1182 return 0;
1183 ce = active_cache[pos];
1184 if ((lstat(name, &st) < 0) ||
1185 !S_ISREG(st.st_mode) || /* careful! */
1186 ce_match_stat(ce, &st, 0) ||
1187 hashcmp(sha1, ce->sha1))
1188 return 0;
1189 /* we return 1 only when we can stat, it is a regular file,
1190 * stat information matches, and sha1 recorded in the cache
1191 * matches. I.e. we know the file in the work tree really is
1192 * the same as the <name, sha1> pair.
1194 return 1;
1197 static struct sha1_size_cache {
1198 unsigned char sha1[20];
1199 unsigned long size;
1200 } **sha1_size_cache;
1201 static int sha1_size_cache_nr, sha1_size_cache_alloc;
1203 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
1204 int find_only,
1205 unsigned long size)
1207 int first, last;
1208 struct sha1_size_cache *e;
1210 first = 0;
1211 last = sha1_size_cache_nr;
1212 while (last > first) {
1213 int cmp, next = (last + first) >> 1;
1214 e = sha1_size_cache[next];
1215 cmp = hashcmp(e->sha1, sha1);
1216 if (!cmp)
1217 return e;
1218 if (cmp < 0) {
1219 last = next;
1220 continue;
1222 first = next+1;
1224 /* not found */
1225 if (find_only)
1226 return NULL;
1227 /* insert to make it at "first" */
1228 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1229 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1230 sha1_size_cache = xrealloc(sha1_size_cache,
1231 sha1_size_cache_alloc *
1232 sizeof(*sha1_size_cache));
1234 sha1_size_cache_nr++;
1235 if (first < sha1_size_cache_nr)
1236 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1237 (sha1_size_cache_nr - first - 1) *
1238 sizeof(*sha1_size_cache));
1239 e = xmalloc(sizeof(struct sha1_size_cache));
1240 sha1_size_cache[first] = e;
1241 hashcpy(e->sha1, sha1);
1242 e->size = size;
1243 return e;
1247 * While doing rename detection and pickaxe operation, we may need to
1248 * grab the data for the blob (or file) for our own in-core comparison.
1249 * diff_filespec has data and size fields for this purpose.
1251 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1253 int err = 0;
1254 if (!DIFF_FILE_VALID(s))
1255 die("internal error: asking to populate invalid file.");
1256 if (S_ISDIR(s->mode))
1257 return -1;
1259 if (!use_size_cache)
1260 size_only = 0;
1262 if (s->data)
1263 return err;
1264 if (!s->sha1_valid ||
1265 work_tree_matches(s->path, s->sha1)) {
1266 struct stat st;
1267 int fd;
1268 if (lstat(s->path, &st) < 0) {
1269 if (errno == ENOENT) {
1270 err_empty:
1271 err = -1;
1272 empty:
1273 s->data = (char *)"";
1274 s->size = 0;
1275 return err;
1278 s->size = st.st_size;
1279 if (!s->size)
1280 goto empty;
1281 if (size_only)
1282 return 0;
1283 if (S_ISLNK(st.st_mode)) {
1284 int ret;
1285 s->data = xmalloc(s->size);
1286 s->should_free = 1;
1287 ret = readlink(s->path, s->data, s->size);
1288 if (ret < 0) {
1289 free(s->data);
1290 goto err_empty;
1292 return 0;
1294 fd = open(s->path, O_RDONLY);
1295 if (fd < 0)
1296 goto err_empty;
1297 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1298 close(fd);
1299 if (s->data == MAP_FAILED)
1300 goto err_empty;
1301 s->should_munmap = 1;
1303 else {
1304 char type[20];
1305 struct sha1_size_cache *e;
1307 if (size_only) {
1308 e = locate_size_cache(s->sha1, 1, 0);
1309 if (e) {
1310 s->size = e->size;
1311 return 0;
1313 if (!sha1_object_info(s->sha1, type, &s->size))
1314 locate_size_cache(s->sha1, 0, s->size);
1316 else {
1317 s->data = read_sha1_file(s->sha1, type, &s->size);
1318 s->should_free = 1;
1321 return 0;
1324 void diff_free_filespec_data(struct diff_filespec *s)
1326 if (s->should_free)
1327 free(s->data);
1328 else if (s->should_munmap)
1329 munmap(s->data, s->size);
1330 s->should_free = s->should_munmap = 0;
1331 s->data = NULL;
1332 free(s->cnt_data);
1333 s->cnt_data = NULL;
1336 static void prep_temp_blob(struct diff_tempfile *temp,
1337 void *blob,
1338 unsigned long size,
1339 const unsigned char *sha1,
1340 int mode)
1342 int fd;
1344 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1345 if (fd < 0)
1346 die("unable to create temp-file");
1347 if (write(fd, blob, size) != size)
1348 die("unable to write temp-file");
1349 close(fd);
1350 temp->name = temp->tmp_path;
1351 strcpy(temp->hex, sha1_to_hex(sha1));
1352 temp->hex[40] = 0;
1353 sprintf(temp->mode, "%06o", mode);
1356 static void prepare_temp_file(const char *name,
1357 struct diff_tempfile *temp,
1358 struct diff_filespec *one)
1360 if (!DIFF_FILE_VALID(one)) {
1361 not_a_valid_file:
1362 /* A '-' entry produces this for file-2, and
1363 * a '+' entry produces this for file-1.
1365 temp->name = "/dev/null";
1366 strcpy(temp->hex, ".");
1367 strcpy(temp->mode, ".");
1368 return;
1371 if (!one->sha1_valid ||
1372 work_tree_matches(name, one->sha1)) {
1373 struct stat st;
1374 if (lstat(name, &st) < 0) {
1375 if (errno == ENOENT)
1376 goto not_a_valid_file;
1377 die("stat(%s): %s", name, strerror(errno));
1379 if (S_ISLNK(st.st_mode)) {
1380 int ret;
1381 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1382 if (sizeof(buf) <= st.st_size)
1383 die("symlink too long: %s", name);
1384 ret = readlink(name, buf, st.st_size);
1385 if (ret < 0)
1386 die("readlink(%s)", name);
1387 prep_temp_blob(temp, buf, st.st_size,
1388 (one->sha1_valid ?
1389 one->sha1 : null_sha1),
1390 (one->sha1_valid ?
1391 one->mode : S_IFLNK));
1393 else {
1394 /* we can borrow from the file in the work tree */
1395 temp->name = name;
1396 if (!one->sha1_valid)
1397 strcpy(temp->hex, sha1_to_hex(null_sha1));
1398 else
1399 strcpy(temp->hex, sha1_to_hex(one->sha1));
1400 /* Even though we may sometimes borrow the
1401 * contents from the work tree, we always want
1402 * one->mode. mode is trustworthy even when
1403 * !(one->sha1_valid), as long as
1404 * DIFF_FILE_VALID(one).
1406 sprintf(temp->mode, "%06o", one->mode);
1408 return;
1410 else {
1411 if (diff_populate_filespec(one, 0))
1412 die("cannot read data blob for %s", one->path);
1413 prep_temp_blob(temp, one->data, one->size,
1414 one->sha1, one->mode);
1418 static void remove_tempfile(void)
1420 int i;
1422 for (i = 0; i < 2; i++)
1423 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1424 unlink(diff_temp[i].name);
1425 diff_temp[i].name = NULL;
1429 static void remove_tempfile_on_signal(int signo)
1431 remove_tempfile();
1432 signal(SIGINT, SIG_DFL);
1433 raise(signo);
1436 static int spawn_prog(const char *pgm, const char **arg)
1438 pid_t pid;
1439 int status;
1441 fflush(NULL);
1442 pid = fork();
1443 if (pid < 0)
1444 die("unable to fork");
1445 if (!pid) {
1446 execvp(pgm, (char *const*) arg);
1447 exit(255);
1450 while (waitpid(pid, &status, 0) < 0) {
1451 if (errno == EINTR)
1452 continue;
1453 return -1;
1456 /* Earlier we did not check the exit status because
1457 * diff exits non-zero if files are different, and
1458 * we are not interested in knowing that. It was a
1459 * mistake which made it harder to quit a diff-*
1460 * session that uses the git-apply-patch-script as
1461 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1462 * should also exit non-zero only when it wants to
1463 * abort the entire diff-* session.
1465 if (WIFEXITED(status) && !WEXITSTATUS(status))
1466 return 0;
1467 return -1;
1470 /* An external diff command takes:
1472 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1473 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1476 static void run_external_diff(const char *pgm,
1477 const char *name,
1478 const char *other,
1479 struct diff_filespec *one,
1480 struct diff_filespec *two,
1481 const char *xfrm_msg,
1482 int complete_rewrite)
1484 const char *spawn_arg[10];
1485 struct diff_tempfile *temp = diff_temp;
1486 int retval;
1487 static int atexit_asked = 0;
1488 const char *othername;
1489 const char **arg = &spawn_arg[0];
1491 othername = (other? other : name);
1492 if (one && two) {
1493 prepare_temp_file(name, &temp[0], one);
1494 prepare_temp_file(othername, &temp[1], two);
1495 if (! atexit_asked &&
1496 (temp[0].name == temp[0].tmp_path ||
1497 temp[1].name == temp[1].tmp_path)) {
1498 atexit_asked = 1;
1499 atexit(remove_tempfile);
1501 signal(SIGINT, remove_tempfile_on_signal);
1504 if (one && two) {
1505 *arg++ = pgm;
1506 *arg++ = name;
1507 *arg++ = temp[0].name;
1508 *arg++ = temp[0].hex;
1509 *arg++ = temp[0].mode;
1510 *arg++ = temp[1].name;
1511 *arg++ = temp[1].hex;
1512 *arg++ = temp[1].mode;
1513 if (other) {
1514 *arg++ = other;
1515 *arg++ = xfrm_msg;
1517 } else {
1518 *arg++ = pgm;
1519 *arg++ = name;
1521 *arg = NULL;
1522 retval = spawn_prog(pgm, spawn_arg);
1523 remove_tempfile();
1524 if (retval) {
1525 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1526 exit(1);
1530 static void run_diff_cmd(const char *pgm,
1531 const char *name,
1532 const char *other,
1533 struct diff_filespec *one,
1534 struct diff_filespec *two,
1535 const char *xfrm_msg,
1536 struct diff_options *o,
1537 int complete_rewrite)
1539 if (pgm) {
1540 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1541 complete_rewrite);
1542 return;
1544 if (one && two)
1545 builtin_diff(name, other ? other : name,
1546 one, two, xfrm_msg, o, complete_rewrite);
1547 else
1548 printf("* Unmerged path %s\n", name);
1551 static void diff_fill_sha1_info(struct diff_filespec *one)
1553 if (DIFF_FILE_VALID(one)) {
1554 if (!one->sha1_valid) {
1555 struct stat st;
1556 if (lstat(one->path, &st) < 0)
1557 die("stat %s", one->path);
1558 if (index_path(one->sha1, one->path, &st, 0))
1559 die("cannot hash %s\n", one->path);
1562 else
1563 hashclr(one->sha1);
1566 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1568 const char *pgm = external_diff();
1569 char msg[PATH_MAX*2+300], *xfrm_msg;
1570 struct diff_filespec *one;
1571 struct diff_filespec *two;
1572 const char *name;
1573 const char *other;
1574 char *name_munged, *other_munged;
1575 int complete_rewrite = 0;
1576 int len;
1578 if (DIFF_PAIR_UNMERGED(p)) {
1579 /* unmerged */
1580 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1581 return;
1584 name = p->one->path;
1585 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1586 name_munged = quote_one(name);
1587 other_munged = quote_one(other);
1588 one = p->one; two = p->two;
1590 diff_fill_sha1_info(one);
1591 diff_fill_sha1_info(two);
1593 len = 0;
1594 switch (p->status) {
1595 case DIFF_STATUS_COPIED:
1596 len += snprintf(msg + len, sizeof(msg) - len,
1597 "similarity index %d%%\n"
1598 "copy from %s\n"
1599 "copy to %s\n",
1600 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1601 name_munged, other_munged);
1602 break;
1603 case DIFF_STATUS_RENAMED:
1604 len += snprintf(msg + len, sizeof(msg) - len,
1605 "similarity index %d%%\n"
1606 "rename from %s\n"
1607 "rename to %s\n",
1608 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1609 name_munged, other_munged);
1610 break;
1611 case DIFF_STATUS_MODIFIED:
1612 if (p->score) {
1613 len += snprintf(msg + len, sizeof(msg) - len,
1614 "dissimilarity index %d%%\n",
1615 (int)(0.5 + p->score *
1616 100.0/MAX_SCORE));
1617 complete_rewrite = 1;
1618 break;
1620 /* fallthru */
1621 default:
1622 /* nothing */
1626 if (hashcmp(one->sha1, two->sha1)) {
1627 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1629 if (o->binary) {
1630 mmfile_t mf;
1631 if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
1632 (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
1633 abbrev = 40;
1635 len += snprintf(msg + len, sizeof(msg) - len,
1636 "index %.*s..%.*s",
1637 abbrev, sha1_to_hex(one->sha1),
1638 abbrev, sha1_to_hex(two->sha1));
1639 if (one->mode == two->mode)
1640 len += snprintf(msg + len, sizeof(msg) - len,
1641 " %06o", one->mode);
1642 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1645 if (len)
1646 msg[--len] = 0;
1647 xfrm_msg = len ? msg : NULL;
1649 if (!pgm &&
1650 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1651 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1652 /* a filepair that changes between file and symlink
1653 * needs to be split into deletion and creation.
1655 struct diff_filespec *null = alloc_filespec(two->path);
1656 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1657 free(null);
1658 null = alloc_filespec(one->path);
1659 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1660 free(null);
1662 else
1663 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1664 complete_rewrite);
1666 free(name_munged);
1667 free(other_munged);
1670 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1671 struct diffstat_t *diffstat)
1673 const char *name;
1674 const char *other;
1675 int complete_rewrite = 0;
1677 if (DIFF_PAIR_UNMERGED(p)) {
1678 /* unmerged */
1679 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1680 return;
1683 name = p->one->path;
1684 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1686 diff_fill_sha1_info(p->one);
1687 diff_fill_sha1_info(p->two);
1689 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1690 complete_rewrite = 1;
1691 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1694 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1696 const char *name;
1697 const char *other;
1699 if (DIFF_PAIR_UNMERGED(p)) {
1700 /* unmerged */
1701 return;
1704 name = p->one->path;
1705 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1707 diff_fill_sha1_info(p->one);
1708 diff_fill_sha1_info(p->two);
1710 builtin_checkdiff(name, other, p->one, p->two);
1713 void diff_setup(struct diff_options *options)
1715 memset(options, 0, sizeof(*options));
1716 options->line_termination = '\n';
1717 options->break_opt = -1;
1718 options->rename_limit = -1;
1719 options->context = 3;
1720 options->msg_sep = "";
1722 options->change = diff_change;
1723 options->add_remove = diff_addremove;
1724 options->color_diff = diff_use_color_default;
1725 options->detect_rename = diff_detect_rename_default;
1728 int diff_setup_done(struct diff_options *options)
1730 int count = 0;
1732 if (options->output_format & DIFF_FORMAT_NAME)
1733 count++;
1734 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
1735 count++;
1736 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
1737 count++;
1738 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
1739 count++;
1740 if (count > 1)
1741 die("--name-only, --name-status, --check and -s are mutually exclusive");
1743 if (options->find_copies_harder)
1744 options->detect_rename = DIFF_DETECT_COPY;
1746 if (options->output_format & (DIFF_FORMAT_NAME |
1747 DIFF_FORMAT_NAME_STATUS |
1748 DIFF_FORMAT_CHECKDIFF |
1749 DIFF_FORMAT_NO_OUTPUT))
1750 options->output_format &= ~(DIFF_FORMAT_RAW |
1751 DIFF_FORMAT_NUMSTAT |
1752 DIFF_FORMAT_DIFFSTAT |
1753 DIFF_FORMAT_SUMMARY |
1754 DIFF_FORMAT_PATCH);
1757 * These cases always need recursive; we do not drop caller-supplied
1758 * recursive bits for other formats here.
1760 if (options->output_format & (DIFF_FORMAT_PATCH |
1761 DIFF_FORMAT_NUMSTAT |
1762 DIFF_FORMAT_DIFFSTAT |
1763 DIFF_FORMAT_SUMMARY |
1764 DIFF_FORMAT_CHECKDIFF))
1765 options->recursive = 1;
1767 * Also pickaxe would not work very well if you do not say recursive
1769 if (options->pickaxe)
1770 options->recursive = 1;
1772 if (options->detect_rename && options->rename_limit < 0)
1773 options->rename_limit = diff_rename_limit_default;
1774 if (options->setup & DIFF_SETUP_USE_CACHE) {
1775 if (!active_cache)
1776 /* read-cache does not die even when it fails
1777 * so it is safe for us to do this here. Also
1778 * it does not smudge active_cache or active_nr
1779 * when it fails, so we do not have to worry about
1780 * cleaning it up ourselves either.
1782 read_cache();
1784 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1785 use_size_cache = 1;
1786 if (options->abbrev <= 0 || 40 < options->abbrev)
1787 options->abbrev = 40; /* full */
1789 return 0;
1792 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1794 char c, *eq;
1795 int len;
1797 if (*arg != '-')
1798 return 0;
1799 c = *++arg;
1800 if (!c)
1801 return 0;
1802 if (c == arg_short) {
1803 c = *++arg;
1804 if (!c)
1805 return 1;
1806 if (val && isdigit(c)) {
1807 char *end;
1808 int n = strtoul(arg, &end, 10);
1809 if (*end)
1810 return 0;
1811 *val = n;
1812 return 1;
1814 return 0;
1816 if (c != '-')
1817 return 0;
1818 arg++;
1819 eq = strchr(arg, '=');
1820 if (eq)
1821 len = eq - arg;
1822 else
1823 len = strlen(arg);
1824 if (!len || strncmp(arg, arg_long, len))
1825 return 0;
1826 if (eq) {
1827 int n;
1828 char *end;
1829 if (!isdigit(*++eq))
1830 return 0;
1831 n = strtoul(eq, &end, 10);
1832 if (*end)
1833 return 0;
1834 *val = n;
1836 return 1;
1839 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1841 const char *arg = av[0];
1842 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1843 options->output_format |= DIFF_FORMAT_PATCH;
1844 else if (opt_arg(arg, 'U', "unified", &options->context))
1845 options->output_format |= DIFF_FORMAT_PATCH;
1846 else if (!strcmp(arg, "--raw"))
1847 options->output_format |= DIFF_FORMAT_RAW;
1848 else if (!strcmp(arg, "--patch-with-raw")) {
1849 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1851 else if (!strcmp(arg, "--numstat")) {
1852 options->output_format |= DIFF_FORMAT_NUMSTAT;
1854 else if (!strncmp(arg, "--stat", 6)) {
1855 char *end;
1856 int width = options->stat_width;
1857 int name_width = options->stat_name_width;
1858 arg += 6;
1859 end = (char *)arg;
1861 switch (*arg) {
1862 case '-':
1863 if (!strncmp(arg, "-width=", 7))
1864 width = strtoul(arg + 7, &end, 10);
1865 else if (!strncmp(arg, "-name-width=", 12))
1866 name_width = strtoul(arg + 12, &end, 10);
1867 break;
1868 case '=':
1869 width = strtoul(arg+1, &end, 10);
1870 if (*end == ',')
1871 name_width = strtoul(end+1, &end, 10);
1874 /* Important! This checks all the error cases! */
1875 if (*end)
1876 return 0;
1877 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1878 options->stat_name_width = name_width;
1879 options->stat_width = width;
1881 else if (!strcmp(arg, "--check"))
1882 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1883 else if (!strcmp(arg, "--summary"))
1884 options->output_format |= DIFF_FORMAT_SUMMARY;
1885 else if (!strcmp(arg, "--patch-with-stat")) {
1886 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1888 else if (!strcmp(arg, "-z"))
1889 options->line_termination = 0;
1890 else if (!strncmp(arg, "-l", 2))
1891 options->rename_limit = strtoul(arg+2, NULL, 10);
1892 else if (!strcmp(arg, "--full-index"))
1893 options->full_index = 1;
1894 else if (!strcmp(arg, "--binary")) {
1895 options->output_format |= DIFF_FORMAT_PATCH;
1896 options->binary = 1;
1898 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
1899 options->text = 1;
1901 else if (!strcmp(arg, "--name-only"))
1902 options->output_format |= DIFF_FORMAT_NAME;
1903 else if (!strcmp(arg, "--name-status"))
1904 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1905 else if (!strcmp(arg, "-R"))
1906 options->reverse_diff = 1;
1907 else if (!strncmp(arg, "-S", 2))
1908 options->pickaxe = arg + 2;
1909 else if (!strcmp(arg, "-s")) {
1910 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1912 else if (!strncmp(arg, "-O", 2))
1913 options->orderfile = arg + 2;
1914 else if (!strncmp(arg, "--diff-filter=", 14))
1915 options->filter = arg + 14;
1916 else if (!strcmp(arg, "--pickaxe-all"))
1917 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1918 else if (!strcmp(arg, "--pickaxe-regex"))
1919 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1920 else if (!strncmp(arg, "-B", 2)) {
1921 if ((options->break_opt =
1922 diff_scoreopt_parse(arg)) == -1)
1923 return -1;
1925 else if (!strncmp(arg, "-M", 2)) {
1926 if ((options->rename_score =
1927 diff_scoreopt_parse(arg)) == -1)
1928 return -1;
1929 options->detect_rename = DIFF_DETECT_RENAME;
1931 else if (!strncmp(arg, "-C", 2)) {
1932 if ((options->rename_score =
1933 diff_scoreopt_parse(arg)) == -1)
1934 return -1;
1935 options->detect_rename = DIFF_DETECT_COPY;
1937 else if (!strcmp(arg, "--find-copies-harder"))
1938 options->find_copies_harder = 1;
1939 else if (!strcmp(arg, "--abbrev"))
1940 options->abbrev = DEFAULT_ABBREV;
1941 else if (!strncmp(arg, "--abbrev=", 9)) {
1942 options->abbrev = strtoul(arg + 9, NULL, 10);
1943 if (options->abbrev < MINIMUM_ABBREV)
1944 options->abbrev = MINIMUM_ABBREV;
1945 else if (40 < options->abbrev)
1946 options->abbrev = 40;
1948 else if (!strcmp(arg, "--color"))
1949 options->color_diff = 1;
1950 else if (!strcmp(arg, "--no-color"))
1951 options->color_diff = 0;
1952 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
1953 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
1954 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
1955 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
1956 else if (!strcmp(arg, "--color-words"))
1957 options->color_diff = options->color_diff_words = 1;
1958 else if (!strcmp(arg, "--no-renames"))
1959 options->detect_rename = 0;
1960 else
1961 return 0;
1962 return 1;
1965 static int parse_num(const char **cp_p)
1967 unsigned long num, scale;
1968 int ch, dot;
1969 const char *cp = *cp_p;
1971 num = 0;
1972 scale = 1;
1973 dot = 0;
1974 for(;;) {
1975 ch = *cp;
1976 if ( !dot && ch == '.' ) {
1977 scale = 1;
1978 dot = 1;
1979 } else if ( ch == '%' ) {
1980 scale = dot ? scale*100 : 100;
1981 cp++; /* % is always at the end */
1982 break;
1983 } else if ( ch >= '0' && ch <= '9' ) {
1984 if ( scale < 100000 ) {
1985 scale *= 10;
1986 num = (num*10) + (ch-'0');
1988 } else {
1989 break;
1991 cp++;
1993 *cp_p = cp;
1995 /* user says num divided by scale and we say internally that
1996 * is MAX_SCORE * num / scale.
1998 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
2001 int diff_scoreopt_parse(const char *opt)
2003 int opt1, opt2, cmd;
2005 if (*opt++ != '-')
2006 return -1;
2007 cmd = *opt++;
2008 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2009 return -1; /* that is not a -M, -C nor -B option */
2011 opt1 = parse_num(&opt);
2012 if (cmd != 'B')
2013 opt2 = 0;
2014 else {
2015 if (*opt == 0)
2016 opt2 = 0;
2017 else if (*opt != '/')
2018 return -1; /* we expect -B80/99 or -B80 */
2019 else {
2020 opt++;
2021 opt2 = parse_num(&opt);
2024 if (*opt != 0)
2025 return -1;
2026 return opt1 | (opt2 << 16);
2029 struct diff_queue_struct diff_queued_diff;
2031 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2033 if (queue->alloc <= queue->nr) {
2034 queue->alloc = alloc_nr(queue->alloc);
2035 queue->queue = xrealloc(queue->queue,
2036 sizeof(dp) * queue->alloc);
2038 queue->queue[queue->nr++] = dp;
2041 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2042 struct diff_filespec *one,
2043 struct diff_filespec *two)
2045 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2046 dp->one = one;
2047 dp->two = two;
2048 if (queue)
2049 diff_q(queue, dp);
2050 return dp;
2053 void diff_free_filepair(struct diff_filepair *p)
2055 diff_free_filespec_data(p->one);
2056 diff_free_filespec_data(p->two);
2057 free(p->one);
2058 free(p->two);
2059 free(p);
2062 /* This is different from find_unique_abbrev() in that
2063 * it stuffs the result with dots for alignment.
2065 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2067 int abblen;
2068 const char *abbrev;
2069 if (len == 40)
2070 return sha1_to_hex(sha1);
2072 abbrev = find_unique_abbrev(sha1, len);
2073 if (!abbrev)
2074 return sha1_to_hex(sha1);
2075 abblen = strlen(abbrev);
2076 if (abblen < 37) {
2077 static char hex[41];
2078 if (len < abblen && abblen <= len + 2)
2079 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2080 else
2081 sprintf(hex, "%s...", abbrev);
2082 return hex;
2084 return sha1_to_hex(sha1);
2087 static void diff_flush_raw(struct diff_filepair *p,
2088 struct diff_options *options)
2090 int two_paths;
2091 char status[10];
2092 int abbrev = options->abbrev;
2093 const char *path_one, *path_two;
2094 int inter_name_termination = '\t';
2095 int line_termination = options->line_termination;
2097 if (!line_termination)
2098 inter_name_termination = 0;
2100 path_one = p->one->path;
2101 path_two = p->two->path;
2102 if (line_termination) {
2103 path_one = quote_one(path_one);
2104 path_two = quote_one(path_two);
2107 if (p->score)
2108 sprintf(status, "%c%03d", p->status,
2109 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2110 else {
2111 status[0] = p->status;
2112 status[1] = 0;
2114 switch (p->status) {
2115 case DIFF_STATUS_COPIED:
2116 case DIFF_STATUS_RENAMED:
2117 two_paths = 1;
2118 break;
2119 case DIFF_STATUS_ADDED:
2120 case DIFF_STATUS_DELETED:
2121 two_paths = 0;
2122 break;
2123 default:
2124 two_paths = 0;
2125 break;
2127 if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2128 printf(":%06o %06o %s ",
2129 p->one->mode, p->two->mode,
2130 diff_unique_abbrev(p->one->sha1, abbrev));
2131 printf("%s ",
2132 diff_unique_abbrev(p->two->sha1, abbrev));
2134 printf("%s%c%s", status, inter_name_termination, path_one);
2135 if (two_paths)
2136 printf("%c%s", inter_name_termination, path_two);
2137 putchar(line_termination);
2138 if (path_one != p->one->path)
2139 free((void*)path_one);
2140 if (path_two != p->two->path)
2141 free((void*)path_two);
2144 static void diff_flush_name(struct diff_filepair *p, int line_termination)
2146 char *path = p->two->path;
2148 if (line_termination)
2149 path = quote_one(p->two->path);
2150 printf("%s%c", path, line_termination);
2151 if (p->two->path != path)
2152 free(path);
2155 int diff_unmodified_pair(struct diff_filepair *p)
2157 /* This function is written stricter than necessary to support
2158 * the currently implemented transformers, but the idea is to
2159 * let transformers to produce diff_filepairs any way they want,
2160 * and filter and clean them up here before producing the output.
2162 struct diff_filespec *one, *two;
2164 if (DIFF_PAIR_UNMERGED(p))
2165 return 0; /* unmerged is interesting */
2167 one = p->one;
2168 two = p->two;
2170 /* deletion, addition, mode or type change
2171 * and rename are all interesting.
2173 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2174 DIFF_PAIR_MODE_CHANGED(p) ||
2175 strcmp(one->path, two->path))
2176 return 0;
2178 /* both are valid and point at the same path. that is, we are
2179 * dealing with a change.
2181 if (one->sha1_valid && two->sha1_valid &&
2182 !hashcmp(one->sha1, two->sha1))
2183 return 1; /* no change */
2184 if (!one->sha1_valid && !two->sha1_valid)
2185 return 1; /* both look at the same file on the filesystem. */
2186 return 0;
2189 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2191 if (diff_unmodified_pair(p))
2192 return;
2194 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2195 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2196 return; /* no tree diffs in patch format */
2198 run_diff(p, o);
2201 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2202 struct diffstat_t *diffstat)
2204 if (diff_unmodified_pair(p))
2205 return;
2207 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2208 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2209 return; /* no tree diffs in patch format */
2211 run_diffstat(p, o, diffstat);
2214 static void diff_flush_checkdiff(struct diff_filepair *p,
2215 struct diff_options *o)
2217 if (diff_unmodified_pair(p))
2218 return;
2220 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2221 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2222 return; /* no tree diffs in patch format */
2224 run_checkdiff(p, o);
2227 int diff_queue_is_empty(void)
2229 struct diff_queue_struct *q = &diff_queued_diff;
2230 int i;
2231 for (i = 0; i < q->nr; i++)
2232 if (!diff_unmodified_pair(q->queue[i]))
2233 return 0;
2234 return 1;
2237 #if DIFF_DEBUG
2238 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2240 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2241 x, one ? one : "",
2242 s->path,
2243 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2244 s->mode,
2245 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2246 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2247 x, one ? one : "",
2248 s->size, s->xfrm_flags);
2251 void diff_debug_filepair(const struct diff_filepair *p, int i)
2253 diff_debug_filespec(p->one, i, "one");
2254 diff_debug_filespec(p->two, i, "two");
2255 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2256 p->score, p->status ? p->status : '?',
2257 p->source_stays, p->broken_pair);
2260 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2262 int i;
2263 if (msg)
2264 fprintf(stderr, "%s\n", msg);
2265 fprintf(stderr, "q->nr = %d\n", q->nr);
2266 for (i = 0; i < q->nr; i++) {
2267 struct diff_filepair *p = q->queue[i];
2268 diff_debug_filepair(p, i);
2271 #endif
2273 static void diff_resolve_rename_copy(void)
2275 int i, j;
2276 struct diff_filepair *p, *pp;
2277 struct diff_queue_struct *q = &diff_queued_diff;
2279 diff_debug_queue("resolve-rename-copy", q);
2281 for (i = 0; i < q->nr; i++) {
2282 p = q->queue[i];
2283 p->status = 0; /* undecided */
2284 if (DIFF_PAIR_UNMERGED(p))
2285 p->status = DIFF_STATUS_UNMERGED;
2286 else if (!DIFF_FILE_VALID(p->one))
2287 p->status = DIFF_STATUS_ADDED;
2288 else if (!DIFF_FILE_VALID(p->two))
2289 p->status = DIFF_STATUS_DELETED;
2290 else if (DIFF_PAIR_TYPE_CHANGED(p))
2291 p->status = DIFF_STATUS_TYPE_CHANGED;
2293 /* from this point on, we are dealing with a pair
2294 * whose both sides are valid and of the same type, i.e.
2295 * either in-place edit or rename/copy edit.
2297 else if (DIFF_PAIR_RENAME(p)) {
2298 if (p->source_stays) {
2299 p->status = DIFF_STATUS_COPIED;
2300 continue;
2302 /* See if there is some other filepair that
2303 * copies from the same source as us. If so
2304 * we are a copy. Otherwise we are either a
2305 * copy if the path stays, or a rename if it
2306 * does not, but we already handled "stays" case.
2308 for (j = i + 1; j < q->nr; j++) {
2309 pp = q->queue[j];
2310 if (strcmp(pp->one->path, p->one->path))
2311 continue; /* not us */
2312 if (!DIFF_PAIR_RENAME(pp))
2313 continue; /* not a rename/copy */
2314 /* pp is a rename/copy from the same source */
2315 p->status = DIFF_STATUS_COPIED;
2316 break;
2318 if (!p->status)
2319 p->status = DIFF_STATUS_RENAMED;
2321 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2322 p->one->mode != p->two->mode)
2323 p->status = DIFF_STATUS_MODIFIED;
2324 else {
2325 /* This is a "no-change" entry and should not
2326 * happen anymore, but prepare for broken callers.
2328 error("feeding unmodified %s to diffcore",
2329 p->one->path);
2330 p->status = DIFF_STATUS_UNKNOWN;
2333 diff_debug_queue("resolve-rename-copy done", q);
2336 static int check_pair_status(struct diff_filepair *p)
2338 switch (p->status) {
2339 case DIFF_STATUS_UNKNOWN:
2340 return 0;
2341 case 0:
2342 die("internal error in diff-resolve-rename-copy");
2343 default:
2344 return 1;
2348 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2350 int fmt = opt->output_format;
2352 if (fmt & DIFF_FORMAT_CHECKDIFF)
2353 diff_flush_checkdiff(p, opt);
2354 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2355 diff_flush_raw(p, opt);
2356 else if (fmt & DIFF_FORMAT_NAME)
2357 diff_flush_name(p, opt->line_termination);
2360 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2362 if (fs->mode)
2363 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2364 else
2365 printf(" %s %s\n", newdelete, fs->path);
2369 static void show_mode_change(struct diff_filepair *p, int show_name)
2371 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2372 if (show_name)
2373 printf(" mode change %06o => %06o %s\n",
2374 p->one->mode, p->two->mode, p->two->path);
2375 else
2376 printf(" mode change %06o => %06o\n",
2377 p->one->mode, p->two->mode);
2381 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2383 const char *old, *new;
2385 /* Find common prefix */
2386 old = p->one->path;
2387 new = p->two->path;
2388 while (1) {
2389 const char *slash_old, *slash_new;
2390 slash_old = strchr(old, '/');
2391 slash_new = strchr(new, '/');
2392 if (!slash_old ||
2393 !slash_new ||
2394 slash_old - old != slash_new - new ||
2395 memcmp(old, new, slash_new - new))
2396 break;
2397 old = slash_old + 1;
2398 new = slash_new + 1;
2400 /* p->one->path thru old is the common prefix, and old and new
2401 * through the end of names are renames
2403 if (old != p->one->path)
2404 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2405 (int)(old - p->one->path), p->one->path,
2406 old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2407 else
2408 printf(" %s %s => %s (%d%%)\n", renamecopy,
2409 p->one->path, p->two->path,
2410 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2411 show_mode_change(p, 0);
2414 static void diff_summary(struct diff_filepair *p)
2416 switch(p->status) {
2417 case DIFF_STATUS_DELETED:
2418 show_file_mode_name("delete", p->one);
2419 break;
2420 case DIFF_STATUS_ADDED:
2421 show_file_mode_name("create", p->two);
2422 break;
2423 case DIFF_STATUS_COPIED:
2424 show_rename_copy("copy", p);
2425 break;
2426 case DIFF_STATUS_RENAMED:
2427 show_rename_copy("rename", p);
2428 break;
2429 default:
2430 if (p->score) {
2431 printf(" rewrite %s (%d%%)\n", p->two->path,
2432 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2433 show_mode_change(p, 0);
2434 } else show_mode_change(p, 1);
2435 break;
2439 struct patch_id_t {
2440 struct xdiff_emit_state xm;
2441 SHA_CTX *ctx;
2442 int patchlen;
2445 static int remove_space(char *line, int len)
2447 int i;
2448 char *dst = line;
2449 unsigned char c;
2451 for (i = 0; i < len; i++)
2452 if (!isspace((c = line[i])))
2453 *dst++ = c;
2455 return dst - line;
2458 static void patch_id_consume(void *priv, char *line, unsigned long len)
2460 struct patch_id_t *data = priv;
2461 int new_len;
2463 /* Ignore line numbers when computing the SHA1 of the patch */
2464 if (!strncmp(line, "@@ -", 4))
2465 return;
2467 new_len = remove_space(line, len);
2469 SHA1_Update(data->ctx, line, new_len);
2470 data->patchlen += new_len;
2473 /* returns 0 upon success, and writes result into sha1 */
2474 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2476 struct diff_queue_struct *q = &diff_queued_diff;
2477 int i;
2478 SHA_CTX ctx;
2479 struct patch_id_t data;
2480 char buffer[PATH_MAX * 4 + 20];
2482 SHA1_Init(&ctx);
2483 memset(&data, 0, sizeof(struct patch_id_t));
2484 data.ctx = &ctx;
2485 data.xm.consume = patch_id_consume;
2487 for (i = 0; i < q->nr; i++) {
2488 xpparam_t xpp;
2489 xdemitconf_t xecfg;
2490 xdemitcb_t ecb;
2491 mmfile_t mf1, mf2;
2492 struct diff_filepair *p = q->queue[i];
2493 int len1, len2;
2495 if (p->status == 0)
2496 return error("internal diff status error");
2497 if (p->status == DIFF_STATUS_UNKNOWN)
2498 continue;
2499 if (diff_unmodified_pair(p))
2500 continue;
2501 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2502 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2503 continue;
2504 if (DIFF_PAIR_UNMERGED(p))
2505 continue;
2507 diff_fill_sha1_info(p->one);
2508 diff_fill_sha1_info(p->two);
2509 if (fill_mmfile(&mf1, p->one) < 0 ||
2510 fill_mmfile(&mf2, p->two) < 0)
2511 return error("unable to read files to diff");
2513 /* Maybe hash p->two? into the patch id? */
2514 if (mmfile_is_binary(&mf2))
2515 continue;
2517 len1 = remove_space(p->one->path, strlen(p->one->path));
2518 len2 = remove_space(p->two->path, strlen(p->two->path));
2519 if (p->one->mode == 0)
2520 len1 = snprintf(buffer, sizeof(buffer),
2521 "diff--gita/%.*sb/%.*s"
2522 "newfilemode%06o"
2523 "---/dev/null"
2524 "+++b/%.*s",
2525 len1, p->one->path,
2526 len2, p->two->path,
2527 p->two->mode,
2528 len2, p->two->path);
2529 else if (p->two->mode == 0)
2530 len1 = snprintf(buffer, sizeof(buffer),
2531 "diff--gita/%.*sb/%.*s"
2532 "deletedfilemode%06o"
2533 "---a/%.*s"
2534 "+++/dev/null",
2535 len1, p->one->path,
2536 len2, p->two->path,
2537 p->one->mode,
2538 len1, p->one->path);
2539 else
2540 len1 = snprintf(buffer, sizeof(buffer),
2541 "diff--gita/%.*sb/%.*s"
2542 "---a/%.*s"
2543 "+++b/%.*s",
2544 len1, p->one->path,
2545 len2, p->two->path,
2546 len1, p->one->path,
2547 len2, p->two->path);
2548 SHA1_Update(&ctx, buffer, len1);
2550 xpp.flags = XDF_NEED_MINIMAL;
2551 xecfg.ctxlen = 3;
2552 xecfg.flags = XDL_EMIT_FUNCNAMES;
2553 ecb.outf = xdiff_outf;
2554 ecb.priv = &data;
2555 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2558 SHA1_Final(sha1, &ctx);
2559 return 0;
2562 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2564 struct diff_queue_struct *q = &diff_queued_diff;
2565 int i;
2566 int result = diff_get_patch_id(options, sha1);
2568 for (i = 0; i < q->nr; i++)
2569 diff_free_filepair(q->queue[i]);
2571 free(q->queue);
2572 q->queue = NULL;
2573 q->nr = q->alloc = 0;
2575 return result;
2578 static int is_summary_empty(const struct diff_queue_struct *q)
2580 int i;
2582 for (i = 0; i < q->nr; i++) {
2583 const struct diff_filepair *p = q->queue[i];
2585 switch (p->status) {
2586 case DIFF_STATUS_DELETED:
2587 case DIFF_STATUS_ADDED:
2588 case DIFF_STATUS_COPIED:
2589 case DIFF_STATUS_RENAMED:
2590 return 0;
2591 default:
2592 if (p->score)
2593 return 0;
2594 if (p->one->mode && p->two->mode &&
2595 p->one->mode != p->two->mode)
2596 return 0;
2597 break;
2600 return 1;
2603 void diff_flush(struct diff_options *options)
2605 struct diff_queue_struct *q = &diff_queued_diff;
2606 int i, output_format = options->output_format;
2607 int separator = 0;
2610 * Order: raw, stat, summary, patch
2611 * or: name/name-status/checkdiff (other bits clear)
2613 if (!q->nr)
2614 goto free_queue;
2616 if (output_format & (DIFF_FORMAT_RAW |
2617 DIFF_FORMAT_NAME |
2618 DIFF_FORMAT_NAME_STATUS |
2619 DIFF_FORMAT_CHECKDIFF)) {
2620 for (i = 0; i < q->nr; i++) {
2621 struct diff_filepair *p = q->queue[i];
2622 if (check_pair_status(p))
2623 flush_one_pair(p, options);
2625 separator++;
2628 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_NUMSTAT)) {
2629 struct diffstat_t diffstat;
2631 memset(&diffstat, 0, sizeof(struct diffstat_t));
2632 diffstat.xm.consume = diffstat_consume;
2633 for (i = 0; i < q->nr; i++) {
2634 struct diff_filepair *p = q->queue[i];
2635 if (check_pair_status(p))
2636 diff_flush_stat(p, options, &diffstat);
2638 if (output_format & DIFF_FORMAT_NUMSTAT)
2639 show_numstat(&diffstat, options);
2640 if (output_format & DIFF_FORMAT_DIFFSTAT)
2641 show_stats(&diffstat, options);
2642 separator++;
2645 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2646 for (i = 0; i < q->nr; i++)
2647 diff_summary(q->queue[i]);
2648 separator++;
2651 if (output_format & DIFF_FORMAT_PATCH) {
2652 if (separator) {
2653 if (options->stat_sep) {
2654 /* attach patch instead of inline */
2655 fputs(options->stat_sep, stdout);
2656 } else {
2657 putchar(options->line_termination);
2661 for (i = 0; i < q->nr; i++) {
2662 struct diff_filepair *p = q->queue[i];
2663 if (check_pair_status(p))
2664 diff_flush_patch(p, options);
2668 if (output_format & DIFF_FORMAT_CALLBACK)
2669 options->format_callback(q, options, options->format_callback_data);
2671 for (i = 0; i < q->nr; i++)
2672 diff_free_filepair(q->queue[i]);
2673 free_queue:
2674 free(q->queue);
2675 q->queue = NULL;
2676 q->nr = q->alloc = 0;
2679 static void diffcore_apply_filter(const char *filter)
2681 int i;
2682 struct diff_queue_struct *q = &diff_queued_diff;
2683 struct diff_queue_struct outq;
2684 outq.queue = NULL;
2685 outq.nr = outq.alloc = 0;
2687 if (!filter)
2688 return;
2690 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2691 int found;
2692 for (i = found = 0; !found && i < q->nr; i++) {
2693 struct diff_filepair *p = q->queue[i];
2694 if (((p->status == DIFF_STATUS_MODIFIED) &&
2695 ((p->score &&
2696 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2697 (!p->score &&
2698 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2699 ((p->status != DIFF_STATUS_MODIFIED) &&
2700 strchr(filter, p->status)))
2701 found++;
2703 if (found)
2704 return;
2706 /* otherwise we will clear the whole queue
2707 * by copying the empty outq at the end of this
2708 * function, but first clear the current entries
2709 * in the queue.
2711 for (i = 0; i < q->nr; i++)
2712 diff_free_filepair(q->queue[i]);
2714 else {
2715 /* Only the matching ones */
2716 for (i = 0; i < q->nr; i++) {
2717 struct diff_filepair *p = q->queue[i];
2719 if (((p->status == DIFF_STATUS_MODIFIED) &&
2720 ((p->score &&
2721 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2722 (!p->score &&
2723 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2724 ((p->status != DIFF_STATUS_MODIFIED) &&
2725 strchr(filter, p->status)))
2726 diff_q(&outq, p);
2727 else
2728 diff_free_filepair(p);
2731 free(q->queue);
2732 *q = outq;
2735 void diffcore_std(struct diff_options *options)
2737 if (options->break_opt != -1)
2738 diffcore_break(options->break_opt);
2739 if (options->detect_rename)
2740 diffcore_rename(options);
2741 if (options->break_opt != -1)
2742 diffcore_merge_broken();
2743 if (options->pickaxe)
2744 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2745 if (options->orderfile)
2746 diffcore_order(options->orderfile);
2747 diff_resolve_rename_copy();
2748 diffcore_apply_filter(options->filter);
2752 void diffcore_std_no_resolve(struct diff_options *options)
2754 if (options->pickaxe)
2755 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2756 if (options->orderfile)
2757 diffcore_order(options->orderfile);
2758 diffcore_apply_filter(options->filter);
2761 void diff_addremove(struct diff_options *options,
2762 int addremove, unsigned mode,
2763 const unsigned char *sha1,
2764 const char *base, const char *path)
2766 char concatpath[PATH_MAX];
2767 struct diff_filespec *one, *two;
2769 /* This may look odd, but it is a preparation for
2770 * feeding "there are unchanged files which should
2771 * not produce diffs, but when you are doing copy
2772 * detection you would need them, so here they are"
2773 * entries to the diff-core. They will be prefixed
2774 * with something like '=' or '*' (I haven't decided
2775 * which but should not make any difference).
2776 * Feeding the same new and old to diff_change()
2777 * also has the same effect.
2778 * Before the final output happens, they are pruned after
2779 * merged into rename/copy pairs as appropriate.
2781 if (options->reverse_diff)
2782 addremove = (addremove == '+' ? '-' :
2783 addremove == '-' ? '+' : addremove);
2785 if (!path) path = "";
2786 sprintf(concatpath, "%s%s", base, path);
2787 one = alloc_filespec(concatpath);
2788 two = alloc_filespec(concatpath);
2790 if (addremove != '+')
2791 fill_filespec(one, sha1, mode);
2792 if (addremove != '-')
2793 fill_filespec(two, sha1, mode);
2795 diff_queue(&diff_queued_diff, one, two);
2798 void diff_change(struct diff_options *options,
2799 unsigned old_mode, unsigned new_mode,
2800 const unsigned char *old_sha1,
2801 const unsigned char *new_sha1,
2802 const char *base, const char *path)
2804 char concatpath[PATH_MAX];
2805 struct diff_filespec *one, *two;
2807 if (options->reverse_diff) {
2808 unsigned tmp;
2809 const unsigned char *tmp_c;
2810 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2811 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2813 if (!path) path = "";
2814 sprintf(concatpath, "%s%s", base, path);
2815 one = alloc_filespec(concatpath);
2816 two = alloc_filespec(concatpath);
2817 fill_filespec(one, old_sha1, old_mode);
2818 fill_filespec(two, new_sha1, new_mode);
2820 diff_queue(&diff_queued_diff, one, two);
2823 void diff_unmerge(struct diff_options *options,
2824 const char *path)
2826 struct diff_filespec *one, *two;
2827 one = alloc_filespec(path);
2828 two = alloc_filespec(path);
2829 diff_queue(&diff_queued_diff, one, two);