git-svn: remove svnadmin dependency from the tests
[git/jnareb-git.git] / diff.c
blobf14288bb8a100c43c6709f658b9a9ca44832fd7f
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"
12 #ifdef NO_FAST_WORKING_DIRECTORY
13 #define FAST_WORKING_DIRECTORY 0
14 #else
15 #define FAST_WORKING_DIRECTORY 1
16 #endif
18 static int use_size_cache;
20 static int diff_detect_rename_default;
21 static int diff_rename_limit_default = -1;
22 static int diff_use_color_default;
24 static char diff_colors[][COLOR_MAXLEN] = {
25 "\033[m", /* reset */
26 "", /* PLAIN (normal) */
27 "\033[1m", /* METAINFO (bold) */
28 "\033[36m", /* FRAGINFO (cyan) */
29 "\033[31m", /* OLD (red) */
30 "\033[32m", /* NEW (green) */
31 "\033[33m", /* COMMIT (yellow) */
32 "\033[41m", /* WHITESPACE (red background) */
35 static int parse_diff_color_slot(const char *var, int ofs)
37 if (!strcasecmp(var+ofs, "plain"))
38 return DIFF_PLAIN;
39 if (!strcasecmp(var+ofs, "meta"))
40 return DIFF_METAINFO;
41 if (!strcasecmp(var+ofs, "frag"))
42 return DIFF_FRAGINFO;
43 if (!strcasecmp(var+ofs, "old"))
44 return DIFF_FILE_OLD;
45 if (!strcasecmp(var+ofs, "new"))
46 return DIFF_FILE_NEW;
47 if (!strcasecmp(var+ofs, "commit"))
48 return DIFF_COMMIT;
49 if (!strcasecmp(var+ofs, "whitespace"))
50 return DIFF_WHITESPACE;
51 die("bad config variable '%s'", var);
55 * These are to give UI layer defaults.
56 * The core-level commands such as git-diff-files should
57 * never be affected by the setting of diff.renames
58 * the user happens to have in the configuration file.
60 int git_diff_ui_config(const char *var, const char *value)
62 if (!strcmp(var, "diff.renamelimit")) {
63 diff_rename_limit_default = git_config_int(var, value);
64 return 0;
66 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
67 diff_use_color_default = git_config_colorbool(var, value);
68 return 0;
70 if (!strcmp(var, "diff.renames")) {
71 if (!value)
72 diff_detect_rename_default = DIFF_DETECT_RENAME;
73 else if (!strcasecmp(value, "copies") ||
74 !strcasecmp(value, "copy"))
75 diff_detect_rename_default = DIFF_DETECT_COPY;
76 else if (git_config_bool(var,value))
77 diff_detect_rename_default = DIFF_DETECT_RENAME;
78 return 0;
80 if (!strncmp(var, "diff.color.", 11) || !strncmp(var, "color.diff.", 11)) {
81 int slot = parse_diff_color_slot(var, 11);
82 color_parse(value, var, diff_colors[slot]);
83 return 0;
85 return git_default_config(var, value);
88 static char *quote_one(const char *str)
90 int needlen;
91 char *xp;
93 if (!str)
94 return NULL;
95 needlen = quote_c_style(str, NULL, NULL, 0);
96 if (!needlen)
97 return xstrdup(str);
98 xp = xmalloc(needlen + 1);
99 quote_c_style(str, xp, NULL, 0);
100 return xp;
103 static char *quote_two(const char *one, const char *two)
105 int need_one = quote_c_style(one, NULL, NULL, 1);
106 int need_two = quote_c_style(two, NULL, NULL, 1);
107 char *xp;
109 if (need_one + need_two) {
110 if (!need_one) need_one = strlen(one);
111 if (!need_two) need_one = strlen(two);
113 xp = xmalloc(need_one + need_two + 3);
114 xp[0] = '"';
115 quote_c_style(one, xp + 1, NULL, 1);
116 quote_c_style(two, xp + need_one + 1, NULL, 1);
117 strcpy(xp + need_one + need_two + 1, "\"");
118 return xp;
120 need_one = strlen(one);
121 need_two = strlen(two);
122 xp = xmalloc(need_one + need_two + 1);
123 strcpy(xp, one);
124 strcpy(xp + need_one, two);
125 return xp;
128 static const char *external_diff(void)
130 static const char *external_diff_cmd = NULL;
131 static int done_preparing = 0;
133 if (done_preparing)
134 return external_diff_cmd;
135 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
136 done_preparing = 1;
137 return external_diff_cmd;
140 #define TEMPFILE_PATH_LEN 50
142 static struct diff_tempfile {
143 const char *name; /* filename external diff should read from */
144 char hex[41];
145 char mode[10];
146 char tmp_path[TEMPFILE_PATH_LEN];
147 } diff_temp[2];
149 static int count_lines(const char *data, int size)
151 int count, ch, completely_empty = 1, nl_just_seen = 0;
152 count = 0;
153 while (0 < size--) {
154 ch = *data++;
155 if (ch == '\n') {
156 count++;
157 nl_just_seen = 1;
158 completely_empty = 0;
160 else {
161 nl_just_seen = 0;
162 completely_empty = 0;
165 if (completely_empty)
166 return 0;
167 if (!nl_just_seen)
168 count++; /* no trailing newline */
169 return count;
172 static void print_line_count(int count)
174 switch (count) {
175 case 0:
176 printf("0,0");
177 break;
178 case 1:
179 printf("1");
180 break;
181 default:
182 printf("1,%d", count);
183 break;
187 static void copy_file(int prefix, const char *data, int size)
189 int ch, nl_just_seen = 1;
190 while (0 < size--) {
191 ch = *data++;
192 if (nl_just_seen)
193 putchar(prefix);
194 putchar(ch);
195 if (ch == '\n')
196 nl_just_seen = 1;
197 else
198 nl_just_seen = 0;
200 if (!nl_just_seen)
201 printf("\n\\ No newline at end of file\n");
204 static void emit_rewrite_diff(const char *name_a,
205 const char *name_b,
206 struct diff_filespec *one,
207 struct diff_filespec *two)
209 int lc_a, lc_b;
210 diff_populate_filespec(one, 0);
211 diff_populate_filespec(two, 0);
212 lc_a = count_lines(one->data, one->size);
213 lc_b = count_lines(two->data, two->size);
214 printf("--- a/%s\n+++ b/%s\n@@ -", name_a, name_b);
215 print_line_count(lc_a);
216 printf(" +");
217 print_line_count(lc_b);
218 printf(" @@\n");
219 if (lc_a)
220 copy_file('-', one->data, one->size);
221 if (lc_b)
222 copy_file('+', two->data, two->size);
225 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
227 if (!DIFF_FILE_VALID(one)) {
228 mf->ptr = (char *)""; /* does not matter */
229 mf->size = 0;
230 return 0;
232 else if (diff_populate_filespec(one, 0))
233 return -1;
234 mf->ptr = one->data;
235 mf->size = one->size;
236 return 0;
239 struct diff_words_buffer {
240 mmfile_t text;
241 long alloc;
242 long current; /* output pointer */
243 int suppressed_newline;
246 static void diff_words_append(char *line, unsigned long len,
247 struct diff_words_buffer *buffer)
249 if (buffer->text.size + len > buffer->alloc) {
250 buffer->alloc = (buffer->text.size + len) * 3 / 2;
251 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
253 line++;
254 len--;
255 memcpy(buffer->text.ptr + buffer->text.size, line, len);
256 buffer->text.size += len;
259 struct diff_words_data {
260 struct xdiff_emit_state xm;
261 struct diff_words_buffer minus, plus;
264 static void print_word(struct diff_words_buffer *buffer, int len, int color,
265 int suppress_newline)
267 const char *ptr;
268 int eol = 0;
270 if (len == 0)
271 return;
273 ptr = buffer->text.ptr + buffer->current;
274 buffer->current += len;
276 if (ptr[len - 1] == '\n') {
277 eol = 1;
278 len--;
281 fputs(diff_get_color(1, color), stdout);
282 fwrite(ptr, len, 1, stdout);
283 fputs(diff_get_color(1, DIFF_RESET), stdout);
285 if (eol) {
286 if (suppress_newline)
287 buffer->suppressed_newline = 1;
288 else
289 putchar('\n');
293 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
295 struct diff_words_data *diff_words = priv;
297 if (diff_words->minus.suppressed_newline) {
298 if (line[0] != '+')
299 putchar('\n');
300 diff_words->minus.suppressed_newline = 0;
303 len--;
304 switch (line[0]) {
305 case '-':
306 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
307 break;
308 case '+':
309 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
310 break;
311 case ' ':
312 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
313 diff_words->minus.current += len;
314 break;
318 /* this executes the word diff on the accumulated buffers */
319 static void diff_words_show(struct diff_words_data *diff_words)
321 xpparam_t xpp;
322 xdemitconf_t xecfg;
323 xdemitcb_t ecb;
324 mmfile_t minus, plus;
325 int i;
327 minus.size = diff_words->minus.text.size;
328 minus.ptr = xmalloc(minus.size);
329 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
330 for (i = 0; i < minus.size; i++)
331 if (isspace(minus.ptr[i]))
332 minus.ptr[i] = '\n';
333 diff_words->minus.current = 0;
335 plus.size = diff_words->plus.text.size;
336 plus.ptr = xmalloc(plus.size);
337 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
338 for (i = 0; i < plus.size; i++)
339 if (isspace(plus.ptr[i]))
340 plus.ptr[i] = '\n';
341 diff_words->plus.current = 0;
343 xpp.flags = XDF_NEED_MINIMAL;
344 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
345 xecfg.flags = 0;
346 ecb.outf = xdiff_outf;
347 ecb.priv = diff_words;
348 diff_words->xm.consume = fn_out_diff_words_aux;
349 xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
351 free(minus.ptr);
352 free(plus.ptr);
353 diff_words->minus.text.size = diff_words->plus.text.size = 0;
355 if (diff_words->minus.suppressed_newline) {
356 putchar('\n');
357 diff_words->minus.suppressed_newline = 0;
361 struct emit_callback {
362 struct xdiff_emit_state xm;
363 int nparents, color_diff;
364 const char **label_path;
365 struct diff_words_data *diff_words;
368 static void free_diff_words_data(struct emit_callback *ecbdata)
370 if (ecbdata->diff_words) {
371 /* flush buffers */
372 if (ecbdata->diff_words->minus.text.size ||
373 ecbdata->diff_words->plus.text.size)
374 diff_words_show(ecbdata->diff_words);
376 if (ecbdata->diff_words->minus.text.ptr)
377 free (ecbdata->diff_words->minus.text.ptr);
378 if (ecbdata->diff_words->plus.text.ptr)
379 free (ecbdata->diff_words->plus.text.ptr);
380 free(ecbdata->diff_words);
381 ecbdata->diff_words = NULL;
385 const char *diff_get_color(int diff_use_color, enum color_diff ix)
387 if (diff_use_color)
388 return diff_colors[ix];
389 return "";
392 static void emit_line(const char *set, const char *reset, const char *line, int len)
394 if (len > 0 && line[len-1] == '\n')
395 len--;
396 fputs(set, stdout);
397 fwrite(line, len, 1, stdout);
398 puts(reset);
401 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
403 int col0 = ecbdata->nparents;
404 int last_tab_in_indent = -1;
405 int last_space_in_indent = -1;
406 int i;
407 int tail = len;
408 int need_highlight_leading_space = 0;
409 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
410 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
412 if (!*ws) {
413 emit_line(set, reset, line, len);
414 return;
417 /* The line is a newly added line. Does it have funny leading
418 * whitespaces? In indent, SP should never precede a TAB.
420 for (i = col0; i < len; i++) {
421 if (line[i] == '\t') {
422 last_tab_in_indent = i;
423 if (0 <= last_space_in_indent)
424 need_highlight_leading_space = 1;
426 else if (line[i] == ' ')
427 last_space_in_indent = i;
428 else
429 break;
431 fputs(set, stdout);
432 fwrite(line, col0, 1, stdout);
433 fputs(reset, stdout);
434 if (((i == len) || line[i] == '\n') && i != col0) {
435 /* The whole line was indent */
436 emit_line(ws, reset, line + col0, len - col0);
437 return;
439 i = col0;
440 if (need_highlight_leading_space) {
441 while (i < last_tab_in_indent) {
442 if (line[i] == ' ') {
443 fputs(ws, stdout);
444 putchar(' ');
445 fputs(reset, stdout);
447 else
448 putchar(line[i]);
449 i++;
452 tail = len - 1;
453 if (line[tail] == '\n' && i < tail)
454 tail--;
455 while (i < tail) {
456 if (!isspace(line[tail]))
457 break;
458 tail--;
460 if ((i < tail && line[tail + 1] != '\n')) {
461 /* This has whitespace between tail+1..len */
462 fputs(set, stdout);
463 fwrite(line + i, tail - i + 1, 1, stdout);
464 fputs(reset, stdout);
465 emit_line(ws, reset, line + tail + 1, len - tail - 1);
467 else
468 emit_line(set, reset, line + i, len - i);
471 static void fn_out_consume(void *priv, char *line, unsigned long len)
473 int i;
474 int color;
475 struct emit_callback *ecbdata = priv;
476 const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
477 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
479 if (ecbdata->label_path[0]) {
480 printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
481 printf("%s+++ %s%s\n", set, ecbdata->label_path[1], reset);
482 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
485 /* This is not really necessary for now because
486 * this codepath only deals with two-way diffs.
488 for (i = 0; i < len && line[i] == '@'; i++)
490 if (2 <= i && i < len && line[i] == ' ') {
491 ecbdata->nparents = i - 1;
492 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
493 reset, line, len);
494 return;
497 if (len < ecbdata->nparents) {
498 set = reset;
499 emit_line(reset, reset, line, len);
500 return;
503 color = DIFF_PLAIN;
504 if (ecbdata->diff_words && ecbdata->nparents != 1)
505 /* fall back to normal diff */
506 free_diff_words_data(ecbdata);
507 if (ecbdata->diff_words) {
508 if (line[0] == '-') {
509 diff_words_append(line, len,
510 &ecbdata->diff_words->minus);
511 return;
512 } else if (line[0] == '+') {
513 diff_words_append(line, len,
514 &ecbdata->diff_words->plus);
515 return;
517 if (ecbdata->diff_words->minus.text.size ||
518 ecbdata->diff_words->plus.text.size)
519 diff_words_show(ecbdata->diff_words);
520 line++;
521 len--;
522 emit_line(set, reset, line, len);
523 return;
525 for (i = 0; i < ecbdata->nparents && len; i++) {
526 if (line[i] == '-')
527 color = DIFF_FILE_OLD;
528 else if (line[i] == '+')
529 color = DIFF_FILE_NEW;
532 if (color != DIFF_FILE_NEW) {
533 emit_line(diff_get_color(ecbdata->color_diff, color),
534 reset, line, len);
535 return;
537 emit_add_line(reset, ecbdata, line, len);
540 static char *pprint_rename(const char *a, const char *b)
542 const char *old = a;
543 const char *new = b;
544 char *name = NULL;
545 int pfx_length, sfx_length;
546 int len_a = strlen(a);
547 int len_b = strlen(b);
549 /* Find common prefix */
550 pfx_length = 0;
551 while (*old && *new && *old == *new) {
552 if (*old == '/')
553 pfx_length = old - a + 1;
554 old++;
555 new++;
558 /* Find common suffix */
559 old = a + len_a;
560 new = b + len_b;
561 sfx_length = 0;
562 while (a <= old && b <= new && *old == *new) {
563 if (*old == '/')
564 sfx_length = len_a - (old - a);
565 old--;
566 new--;
570 * pfx{mid-a => mid-b}sfx
571 * {pfx-a => pfx-b}sfx
572 * pfx{sfx-a => sfx-b}
573 * name-a => name-b
575 if (pfx_length + sfx_length) {
576 int a_midlen = len_a - pfx_length - sfx_length;
577 int b_midlen = len_b - pfx_length - sfx_length;
578 if (a_midlen < 0) a_midlen = 0;
579 if (b_midlen < 0) b_midlen = 0;
581 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
582 sprintf(name, "%.*s{%.*s => %.*s}%s",
583 pfx_length, a,
584 a_midlen, a + pfx_length,
585 b_midlen, b + pfx_length,
586 a + len_a - sfx_length);
588 else {
589 name = xmalloc(len_a + len_b + 5);
590 sprintf(name, "%s => %s", a, b);
592 return name;
595 struct diffstat_t {
596 struct xdiff_emit_state xm;
598 int nr;
599 int alloc;
600 struct diffstat_file {
601 char *name;
602 unsigned is_unmerged:1;
603 unsigned is_binary:1;
604 unsigned is_renamed:1;
605 unsigned int added, deleted;
606 } **files;
609 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
610 const char *name_a,
611 const char *name_b)
613 struct diffstat_file *x;
614 x = xcalloc(sizeof (*x), 1);
615 if (diffstat->nr == diffstat->alloc) {
616 diffstat->alloc = alloc_nr(diffstat->alloc);
617 diffstat->files = xrealloc(diffstat->files,
618 diffstat->alloc * sizeof(x));
620 diffstat->files[diffstat->nr++] = x;
621 if (name_b) {
622 x->name = pprint_rename(name_a, name_b);
623 x->is_renamed = 1;
625 else
626 x->name = xstrdup(name_a);
627 return x;
630 static void diffstat_consume(void *priv, char *line, unsigned long len)
632 struct diffstat_t *diffstat = priv;
633 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
635 if (line[0] == '+')
636 x->added++;
637 else if (line[0] == '-')
638 x->deleted++;
641 const char mime_boundary_leader[] = "------------";
643 static int scale_linear(int it, int width, int max_change)
646 * make sure that at least one '-' is printed if there were deletions,
647 * and likewise for '+'.
649 if (max_change < 2)
650 return it;
651 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
654 static void show_name(const char *prefix, const char *name, int len,
655 const char *reset, const char *set)
657 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
660 static void show_graph(char ch, int cnt, const char *set, const char *reset)
662 if (cnt <= 0)
663 return;
664 printf("%s", set);
665 while (cnt--)
666 putchar(ch);
667 printf("%s", reset);
670 static void show_stats(struct diffstat_t* data, struct diff_options *options)
672 int i, len, add, del, total, adds = 0, dels = 0;
673 int max_change = 0, max_len = 0;
674 int total_files = data->nr;
675 int width, name_width;
676 const char *reset, *set, *add_c, *del_c;
678 if (data->nr == 0)
679 return;
681 width = options->stat_width ? options->stat_width : 80;
682 name_width = options->stat_name_width ? options->stat_name_width : 50;
684 /* Sanity: give at least 5 columns to the graph,
685 * but leave at least 10 columns for the name.
687 if (width < name_width + 15) {
688 if (name_width <= 25)
689 width = name_width + 15;
690 else
691 name_width = width - 15;
694 /* Find the longest filename and max number of changes */
695 reset = diff_get_color(options->color_diff, DIFF_RESET);
696 set = diff_get_color(options->color_diff, DIFF_PLAIN);
697 add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
698 del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
700 for (i = 0; i < data->nr; i++) {
701 struct diffstat_file *file = data->files[i];
702 int change = file->added + file->deleted;
704 len = quote_c_style(file->name, NULL, NULL, 0);
705 if (len) {
706 char *qname = xmalloc(len + 1);
707 quote_c_style(file->name, qname, NULL, 0);
708 free(file->name);
709 file->name = qname;
712 len = strlen(file->name);
713 if (max_len < len)
714 max_len = len;
716 if (file->is_binary || file->is_unmerged)
717 continue;
718 if (max_change < change)
719 max_change = change;
722 /* Compute the width of the graph part;
723 * 10 is for one blank at the beginning of the line plus
724 * " | count " between the name and the graph.
726 * From here on, name_width is the width of the name area,
727 * and width is the width of the graph area.
729 name_width = (name_width < max_len) ? name_width : max_len;
730 if (width < (name_width + 10) + max_change)
731 width = width - (name_width + 10);
732 else
733 width = max_change;
735 for (i = 0; i < data->nr; i++) {
736 const char *prefix = "";
737 char *name = data->files[i]->name;
738 int added = data->files[i]->added;
739 int deleted = data->files[i]->deleted;
740 int name_len;
743 * "scale" the filename
745 len = name_width;
746 name_len = strlen(name);
747 if (name_width < name_len) {
748 char *slash;
749 prefix = "...";
750 len -= 3;
751 name += name_len - len;
752 slash = strchr(name, '/');
753 if (slash)
754 name = slash;
757 if (data->files[i]->is_binary) {
758 show_name(prefix, name, len, reset, set);
759 printf(" Bin\n");
760 goto free_diffstat_file;
762 else if (data->files[i]->is_unmerged) {
763 show_name(prefix, name, len, reset, set);
764 printf(" Unmerged\n");
765 goto free_diffstat_file;
767 else if (!data->files[i]->is_renamed &&
768 (added + deleted == 0)) {
769 total_files--;
770 goto free_diffstat_file;
774 * scale the add/delete
776 add = added;
777 del = deleted;
778 total = add + del;
779 adds += add;
780 dels += del;
782 if (width <= max_change) {
783 add = scale_linear(add, width, max_change);
784 del = scale_linear(del, width, max_change);
785 total = add + del;
787 show_name(prefix, name, len, reset, set);
788 printf("%5d ", added + deleted);
789 show_graph('+', add, add_c, reset);
790 show_graph('-', del, del_c, reset);
791 putchar('\n');
792 free_diffstat_file:
793 free(data->files[i]->name);
794 free(data->files[i]);
796 free(data->files);
797 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
798 set, total_files, adds, dels, reset);
801 static void show_shortstats(struct diffstat_t* data)
803 int i, adds = 0, dels = 0, total_files = data->nr;
805 if (data->nr == 0)
806 return;
808 for (i = 0; i < data->nr; i++) {
809 if (!data->files[i]->is_binary &&
810 !data->files[i]->is_unmerged) {
811 int added = data->files[i]->added;
812 int deleted= data->files[i]->deleted;
813 if (!data->files[i]->is_renamed &&
814 (added + deleted == 0)) {
815 total_files--;
816 } else {
817 adds += added;
818 dels += deleted;
821 free(data->files[i]->name);
822 free(data->files[i]);
824 free(data->files);
826 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
827 total_files, adds, dels);
830 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
832 int i;
834 for (i = 0; i < data->nr; i++) {
835 struct diffstat_file *file = data->files[i];
837 if (file->is_binary)
838 printf("-\t-\t");
839 else
840 printf("%d\t%d\t", file->added, file->deleted);
841 if (options->line_termination &&
842 quote_c_style(file->name, NULL, NULL, 0))
843 quote_c_style(file->name, NULL, stdout, 0);
844 else
845 fputs(file->name, stdout);
846 putchar(options->line_termination);
850 struct checkdiff_t {
851 struct xdiff_emit_state xm;
852 const char *filename;
853 int lineno;
856 static void checkdiff_consume(void *priv, char *line, unsigned long len)
858 struct checkdiff_t *data = priv;
860 if (line[0] == '+') {
861 int i, spaces = 0;
863 /* check space before tab */
864 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
865 if (line[i] == ' ')
866 spaces++;
867 if (line[i - 1] == '\t' && spaces)
868 printf("%s:%d: space before tab:%.*s\n",
869 data->filename, data->lineno, (int)len, line);
871 /* check white space at line end */
872 if (line[len - 1] == '\n')
873 len--;
874 if (isspace(line[len - 1]))
875 printf("%s:%d: white space at end: %.*s\n",
876 data->filename, data->lineno, (int)len, line);
878 data->lineno++;
879 } else if (line[0] == ' ')
880 data->lineno++;
881 else if (line[0] == '@') {
882 char *plus = strchr(line, '+');
883 if (plus)
884 data->lineno = strtol(plus, NULL, 10);
885 else
886 die("invalid diff");
890 static unsigned char *deflate_it(char *data,
891 unsigned long size,
892 unsigned long *result_size)
894 int bound;
895 unsigned char *deflated;
896 z_stream stream;
898 memset(&stream, 0, sizeof(stream));
899 deflateInit(&stream, zlib_compression_level);
900 bound = deflateBound(&stream, size);
901 deflated = xmalloc(bound);
902 stream.next_out = deflated;
903 stream.avail_out = bound;
905 stream.next_in = (unsigned char *)data;
906 stream.avail_in = size;
907 while (deflate(&stream, Z_FINISH) == Z_OK)
908 ; /* nothing */
909 deflateEnd(&stream);
910 *result_size = stream.total_out;
911 return deflated;
914 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
916 void *cp;
917 void *delta;
918 void *deflated;
919 void *data;
920 unsigned long orig_size;
921 unsigned long delta_size;
922 unsigned long deflate_size;
923 unsigned long data_size;
925 /* We could do deflated delta, or we could do just deflated two,
926 * whichever is smaller.
928 delta = NULL;
929 deflated = deflate_it(two->ptr, two->size, &deflate_size);
930 if (one->size && two->size) {
931 delta = diff_delta(one->ptr, one->size,
932 two->ptr, two->size,
933 &delta_size, deflate_size);
934 if (delta) {
935 void *to_free = delta;
936 orig_size = delta_size;
937 delta = deflate_it(delta, delta_size, &delta_size);
938 free(to_free);
942 if (delta && delta_size < deflate_size) {
943 printf("delta %lu\n", orig_size);
944 free(deflated);
945 data = delta;
946 data_size = delta_size;
948 else {
949 printf("literal %lu\n", two->size);
950 free(delta);
951 data = deflated;
952 data_size = deflate_size;
955 /* emit data encoded in base85 */
956 cp = data;
957 while (data_size) {
958 int bytes = (52 < data_size) ? 52 : data_size;
959 char line[70];
960 data_size -= bytes;
961 if (bytes <= 26)
962 line[0] = bytes + 'A' - 1;
963 else
964 line[0] = bytes - 26 + 'a' - 1;
965 encode_85(line + 1, cp, bytes);
966 cp = (char *) cp + bytes;
967 puts(line);
969 printf("\n");
970 free(data);
973 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
975 printf("GIT binary patch\n");
976 emit_binary_diff_body(one, two);
977 emit_binary_diff_body(two, one);
980 #define FIRST_FEW_BYTES 8000
981 static int mmfile_is_binary(mmfile_t *mf)
983 long sz = mf->size;
984 if (FIRST_FEW_BYTES < sz)
985 sz = FIRST_FEW_BYTES;
986 return !!memchr(mf->ptr, 0, sz);
989 static void builtin_diff(const char *name_a,
990 const char *name_b,
991 struct diff_filespec *one,
992 struct diff_filespec *two,
993 const char *xfrm_msg,
994 struct diff_options *o,
995 int complete_rewrite)
997 mmfile_t mf1, mf2;
998 const char *lbl[2];
999 char *a_one, *b_two;
1000 const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1001 const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1003 a_one = quote_two("a/", name_a);
1004 b_two = quote_two("b/", name_b);
1005 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1006 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1007 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1008 if (lbl[0][0] == '/') {
1009 /* /dev/null */
1010 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1011 if (xfrm_msg && xfrm_msg[0])
1012 printf("%s%s%s\n", set, xfrm_msg, reset);
1014 else if (lbl[1][0] == '/') {
1015 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1016 if (xfrm_msg && xfrm_msg[0])
1017 printf("%s%s%s\n", set, xfrm_msg, reset);
1019 else {
1020 if (one->mode != two->mode) {
1021 printf("%sold mode %06o%s\n", set, one->mode, reset);
1022 printf("%snew mode %06o%s\n", set, two->mode, reset);
1024 if (xfrm_msg && xfrm_msg[0])
1025 printf("%s%s%s\n", set, xfrm_msg, reset);
1027 * we do not run diff between different kind
1028 * of objects.
1030 if ((one->mode ^ two->mode) & S_IFMT)
1031 goto free_ab_and_return;
1032 if (complete_rewrite) {
1033 emit_rewrite_diff(name_a, name_b, one, two);
1034 goto free_ab_and_return;
1038 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1039 die("unable to read files to diff");
1041 if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
1042 /* Quite common confusing case */
1043 if (mf1.size == mf2.size &&
1044 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1045 goto free_ab_and_return;
1046 if (o->binary)
1047 emit_binary_diff(&mf1, &mf2);
1048 else
1049 printf("Binary files %s and %s differ\n",
1050 lbl[0], lbl[1]);
1052 else {
1053 /* Crazy xdl interfaces.. */
1054 const char *diffopts = getenv("GIT_DIFF_OPTS");
1055 xpparam_t xpp;
1056 xdemitconf_t xecfg;
1057 xdemitcb_t ecb;
1058 struct emit_callback ecbdata;
1060 memset(&ecbdata, 0, sizeof(ecbdata));
1061 ecbdata.label_path = lbl;
1062 ecbdata.color_diff = o->color_diff;
1063 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1064 xecfg.ctxlen = o->context;
1065 xecfg.flags = XDL_EMIT_FUNCNAMES;
1066 if (!diffopts)
1068 else if (!strncmp(diffopts, "--unified=", 10))
1069 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1070 else if (!strncmp(diffopts, "-u", 2))
1071 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1072 ecb.outf = xdiff_outf;
1073 ecb.priv = &ecbdata;
1074 ecbdata.xm.consume = fn_out_consume;
1075 if (o->color_diff_words)
1076 ecbdata.diff_words =
1077 xcalloc(1, sizeof(struct diff_words_data));
1078 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1079 if (o->color_diff_words)
1080 free_diff_words_data(&ecbdata);
1083 free_ab_and_return:
1084 free(a_one);
1085 free(b_two);
1086 return;
1089 static void builtin_diffstat(const char *name_a, const char *name_b,
1090 struct diff_filespec *one,
1091 struct diff_filespec *two,
1092 struct diffstat_t *diffstat,
1093 struct diff_options *o,
1094 int complete_rewrite)
1096 mmfile_t mf1, mf2;
1097 struct diffstat_file *data;
1099 data = diffstat_add(diffstat, name_a, name_b);
1101 if (!one || !two) {
1102 data->is_unmerged = 1;
1103 return;
1105 if (complete_rewrite) {
1106 diff_populate_filespec(one, 0);
1107 diff_populate_filespec(two, 0);
1108 data->deleted = count_lines(one->data, one->size);
1109 data->added = count_lines(two->data, two->size);
1110 return;
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(&mf1) || mmfile_is_binary(&mf2))
1116 data->is_binary = 1;
1117 else {
1118 /* Crazy xdl interfaces.. */
1119 xpparam_t xpp;
1120 xdemitconf_t xecfg;
1121 xdemitcb_t ecb;
1123 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1124 xecfg.ctxlen = 0;
1125 xecfg.flags = 0;
1126 ecb.outf = xdiff_outf;
1127 ecb.priv = diffstat;
1128 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1132 static void builtin_checkdiff(const char *name_a, const char *name_b,
1133 struct diff_filespec *one,
1134 struct diff_filespec *two)
1136 mmfile_t mf1, mf2;
1137 struct checkdiff_t data;
1139 if (!two)
1140 return;
1142 memset(&data, 0, sizeof(data));
1143 data.xm.consume = checkdiff_consume;
1144 data.filename = name_b ? name_b : name_a;
1145 data.lineno = 0;
1147 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1148 die("unable to read files to diff");
1150 if (mmfile_is_binary(&mf2))
1151 return;
1152 else {
1153 /* Crazy xdl interfaces.. */
1154 xpparam_t xpp;
1155 xdemitconf_t xecfg;
1156 xdemitcb_t ecb;
1158 xpp.flags = XDF_NEED_MINIMAL;
1159 xecfg.ctxlen = 0;
1160 xecfg.flags = 0;
1161 ecb.outf = xdiff_outf;
1162 ecb.priv = &data;
1163 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1167 struct diff_filespec *alloc_filespec(const char *path)
1169 int namelen = strlen(path);
1170 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1172 memset(spec, 0, sizeof(*spec));
1173 spec->path = (char *)(spec + 1);
1174 memcpy(spec->path, path, namelen+1);
1175 return spec;
1178 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1179 unsigned short mode)
1181 if (mode) {
1182 spec->mode = canon_mode(mode);
1183 hashcpy(spec->sha1, sha1);
1184 spec->sha1_valid = !is_null_sha1(sha1);
1189 * Given a name and sha1 pair, if the dircache tells us the file in
1190 * the work tree has that object contents, return true, so that
1191 * prepare_temp_file() does not have to inflate and extract.
1193 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1195 struct cache_entry *ce;
1196 struct stat st;
1197 int pos, len;
1199 /* We do not read the cache ourselves here, because the
1200 * benchmark with my previous version that always reads cache
1201 * shows that it makes things worse for diff-tree comparing
1202 * two linux-2.6 kernel trees in an already checked out work
1203 * tree. This is because most diff-tree comparisons deal with
1204 * only a small number of files, while reading the cache is
1205 * expensive for a large project, and its cost outweighs the
1206 * savings we get by not inflating the object to a temporary
1207 * file. Practically, this code only helps when we are used
1208 * by diff-cache --cached, which does read the cache before
1209 * calling us.
1211 if (!active_cache)
1212 return 0;
1214 /* We want to avoid the working directory if our caller
1215 * doesn't need the data in a normal file, this system
1216 * is rather slow with its stat/open/mmap/close syscalls,
1217 * and the object is contained in a pack file. The pack
1218 * is probably already open and will be faster to obtain
1219 * the data through than the working directory. Loose
1220 * objects however would tend to be slower as they need
1221 * to be individually opened and inflated.
1223 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1224 return 0;
1226 len = strlen(name);
1227 pos = cache_name_pos(name, len);
1228 if (pos < 0)
1229 return 0;
1230 ce = active_cache[pos];
1231 if ((lstat(name, &st) < 0) ||
1232 !S_ISREG(st.st_mode) || /* careful! */
1233 ce_match_stat(ce, &st, 0) ||
1234 hashcmp(sha1, ce->sha1))
1235 return 0;
1236 /* we return 1 only when we can stat, it is a regular file,
1237 * stat information matches, and sha1 recorded in the cache
1238 * matches. I.e. we know the file in the work tree really is
1239 * the same as the <name, sha1> pair.
1241 return 1;
1244 static struct sha1_size_cache {
1245 unsigned char sha1[20];
1246 unsigned long size;
1247 } **sha1_size_cache;
1248 static int sha1_size_cache_nr, sha1_size_cache_alloc;
1250 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
1251 int find_only,
1252 unsigned long size)
1254 int first, last;
1255 struct sha1_size_cache *e;
1257 first = 0;
1258 last = sha1_size_cache_nr;
1259 while (last > first) {
1260 int cmp, next = (last + first) >> 1;
1261 e = sha1_size_cache[next];
1262 cmp = hashcmp(e->sha1, sha1);
1263 if (!cmp)
1264 return e;
1265 if (cmp < 0) {
1266 last = next;
1267 continue;
1269 first = next+1;
1271 /* not found */
1272 if (find_only)
1273 return NULL;
1274 /* insert to make it at "first" */
1275 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1276 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1277 sha1_size_cache = xrealloc(sha1_size_cache,
1278 sha1_size_cache_alloc *
1279 sizeof(*sha1_size_cache));
1281 sha1_size_cache_nr++;
1282 if (first < sha1_size_cache_nr)
1283 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1284 (sha1_size_cache_nr - first - 1) *
1285 sizeof(*sha1_size_cache));
1286 e = xmalloc(sizeof(struct sha1_size_cache));
1287 sha1_size_cache[first] = e;
1288 hashcpy(e->sha1, sha1);
1289 e->size = size;
1290 return e;
1294 * While doing rename detection and pickaxe operation, we may need to
1295 * grab the data for the blob (or file) for our own in-core comparison.
1296 * diff_filespec has data and size fields for this purpose.
1298 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1300 int err = 0;
1301 if (!DIFF_FILE_VALID(s))
1302 die("internal error: asking to populate invalid file.");
1303 if (S_ISDIR(s->mode))
1304 return -1;
1306 if (!use_size_cache)
1307 size_only = 0;
1309 if (s->data)
1310 return err;
1311 if (!s->sha1_valid ||
1312 reuse_worktree_file(s->path, s->sha1, 0)) {
1313 struct stat st;
1314 int fd;
1315 if (lstat(s->path, &st) < 0) {
1316 if (errno == ENOENT) {
1317 err_empty:
1318 err = -1;
1319 empty:
1320 s->data = (char *)"";
1321 s->size = 0;
1322 return err;
1325 s->size = st.st_size;
1326 if (!s->size)
1327 goto empty;
1328 if (size_only)
1329 return 0;
1330 if (S_ISLNK(st.st_mode)) {
1331 int ret;
1332 s->data = xmalloc(s->size);
1333 s->should_free = 1;
1334 ret = readlink(s->path, s->data, s->size);
1335 if (ret < 0) {
1336 free(s->data);
1337 goto err_empty;
1339 return 0;
1341 fd = open(s->path, O_RDONLY);
1342 if (fd < 0)
1343 goto err_empty;
1344 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1345 close(fd);
1346 if (s->data == MAP_FAILED)
1347 goto err_empty;
1348 s->should_munmap = 1;
1350 else {
1351 char type[20];
1352 struct sha1_size_cache *e;
1354 if (size_only) {
1355 e = locate_size_cache(s->sha1, 1, 0);
1356 if (e) {
1357 s->size = e->size;
1358 return 0;
1360 if (!sha1_object_info(s->sha1, type, &s->size))
1361 locate_size_cache(s->sha1, 0, s->size);
1363 else {
1364 s->data = read_sha1_file(s->sha1, type, &s->size);
1365 s->should_free = 1;
1368 return 0;
1371 void diff_free_filespec_data(struct diff_filespec *s)
1373 if (s->should_free)
1374 free(s->data);
1375 else if (s->should_munmap)
1376 munmap(s->data, s->size);
1377 s->should_free = s->should_munmap = 0;
1378 s->data = NULL;
1379 free(s->cnt_data);
1380 s->cnt_data = NULL;
1383 static void prep_temp_blob(struct diff_tempfile *temp,
1384 void *blob,
1385 unsigned long size,
1386 const unsigned char *sha1,
1387 int mode)
1389 int fd;
1391 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1392 if (fd < 0)
1393 die("unable to create temp-file");
1394 if (write(fd, blob, size) != size)
1395 die("unable to write temp-file");
1396 close(fd);
1397 temp->name = temp->tmp_path;
1398 strcpy(temp->hex, sha1_to_hex(sha1));
1399 temp->hex[40] = 0;
1400 sprintf(temp->mode, "%06o", mode);
1403 static void prepare_temp_file(const char *name,
1404 struct diff_tempfile *temp,
1405 struct diff_filespec *one)
1407 if (!DIFF_FILE_VALID(one)) {
1408 not_a_valid_file:
1409 /* A '-' entry produces this for file-2, and
1410 * a '+' entry produces this for file-1.
1412 temp->name = "/dev/null";
1413 strcpy(temp->hex, ".");
1414 strcpy(temp->mode, ".");
1415 return;
1418 if (!one->sha1_valid ||
1419 reuse_worktree_file(name, one->sha1, 1)) {
1420 struct stat st;
1421 if (lstat(name, &st) < 0) {
1422 if (errno == ENOENT)
1423 goto not_a_valid_file;
1424 die("stat(%s): %s", name, strerror(errno));
1426 if (S_ISLNK(st.st_mode)) {
1427 int ret;
1428 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1429 if (sizeof(buf) <= st.st_size)
1430 die("symlink too long: %s", name);
1431 ret = readlink(name, buf, st.st_size);
1432 if (ret < 0)
1433 die("readlink(%s)", name);
1434 prep_temp_blob(temp, buf, st.st_size,
1435 (one->sha1_valid ?
1436 one->sha1 : null_sha1),
1437 (one->sha1_valid ?
1438 one->mode : S_IFLNK));
1440 else {
1441 /* we can borrow from the file in the work tree */
1442 temp->name = name;
1443 if (!one->sha1_valid)
1444 strcpy(temp->hex, sha1_to_hex(null_sha1));
1445 else
1446 strcpy(temp->hex, sha1_to_hex(one->sha1));
1447 /* Even though we may sometimes borrow the
1448 * contents from the work tree, we always want
1449 * one->mode. mode is trustworthy even when
1450 * !(one->sha1_valid), as long as
1451 * DIFF_FILE_VALID(one).
1453 sprintf(temp->mode, "%06o", one->mode);
1455 return;
1457 else {
1458 if (diff_populate_filespec(one, 0))
1459 die("cannot read data blob for %s", one->path);
1460 prep_temp_blob(temp, one->data, one->size,
1461 one->sha1, one->mode);
1465 static void remove_tempfile(void)
1467 int i;
1469 for (i = 0; i < 2; i++)
1470 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1471 unlink(diff_temp[i].name);
1472 diff_temp[i].name = NULL;
1476 static void remove_tempfile_on_signal(int signo)
1478 remove_tempfile();
1479 signal(SIGINT, SIG_DFL);
1480 raise(signo);
1483 static int spawn_prog(const char *pgm, const char **arg)
1485 pid_t pid;
1486 int status;
1488 fflush(NULL);
1489 pid = fork();
1490 if (pid < 0)
1491 die("unable to fork");
1492 if (!pid) {
1493 execvp(pgm, (char *const*) arg);
1494 exit(255);
1497 while (waitpid(pid, &status, 0) < 0) {
1498 if (errno == EINTR)
1499 continue;
1500 return -1;
1503 /* Earlier we did not check the exit status because
1504 * diff exits non-zero if files are different, and
1505 * we are not interested in knowing that. It was a
1506 * mistake which made it harder to quit a diff-*
1507 * session that uses the git-apply-patch-script as
1508 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1509 * should also exit non-zero only when it wants to
1510 * abort the entire diff-* session.
1512 if (WIFEXITED(status) && !WEXITSTATUS(status))
1513 return 0;
1514 return -1;
1517 /* An external diff command takes:
1519 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1520 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1523 static void run_external_diff(const char *pgm,
1524 const char *name,
1525 const char *other,
1526 struct diff_filespec *one,
1527 struct diff_filespec *two,
1528 const char *xfrm_msg,
1529 int complete_rewrite)
1531 const char *spawn_arg[10];
1532 struct diff_tempfile *temp = diff_temp;
1533 int retval;
1534 static int atexit_asked = 0;
1535 const char *othername;
1536 const char **arg = &spawn_arg[0];
1538 othername = (other? other : name);
1539 if (one && two) {
1540 prepare_temp_file(name, &temp[0], one);
1541 prepare_temp_file(othername, &temp[1], two);
1542 if (! atexit_asked &&
1543 (temp[0].name == temp[0].tmp_path ||
1544 temp[1].name == temp[1].tmp_path)) {
1545 atexit_asked = 1;
1546 atexit(remove_tempfile);
1548 signal(SIGINT, remove_tempfile_on_signal);
1551 if (one && two) {
1552 *arg++ = pgm;
1553 *arg++ = name;
1554 *arg++ = temp[0].name;
1555 *arg++ = temp[0].hex;
1556 *arg++ = temp[0].mode;
1557 *arg++ = temp[1].name;
1558 *arg++ = temp[1].hex;
1559 *arg++ = temp[1].mode;
1560 if (other) {
1561 *arg++ = other;
1562 *arg++ = xfrm_msg;
1564 } else {
1565 *arg++ = pgm;
1566 *arg++ = name;
1568 *arg = NULL;
1569 retval = spawn_prog(pgm, spawn_arg);
1570 remove_tempfile();
1571 if (retval) {
1572 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1573 exit(1);
1577 static void run_diff_cmd(const char *pgm,
1578 const char *name,
1579 const char *other,
1580 struct diff_filespec *one,
1581 struct diff_filespec *two,
1582 const char *xfrm_msg,
1583 struct diff_options *o,
1584 int complete_rewrite)
1586 if (pgm) {
1587 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1588 complete_rewrite);
1589 return;
1591 if (one && two)
1592 builtin_diff(name, other ? other : name,
1593 one, two, xfrm_msg, o, complete_rewrite);
1594 else
1595 printf("* Unmerged path %s\n", name);
1598 static void diff_fill_sha1_info(struct diff_filespec *one)
1600 if (DIFF_FILE_VALID(one)) {
1601 if (!one->sha1_valid) {
1602 struct stat st;
1603 if (lstat(one->path, &st) < 0)
1604 die("stat %s", one->path);
1605 if (index_path(one->sha1, one->path, &st, 0))
1606 die("cannot hash %s\n", one->path);
1609 else
1610 hashclr(one->sha1);
1613 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1615 const char *pgm = external_diff();
1616 char msg[PATH_MAX*2+300], *xfrm_msg;
1617 struct diff_filespec *one;
1618 struct diff_filespec *two;
1619 const char *name;
1620 const char *other;
1621 char *name_munged, *other_munged;
1622 int complete_rewrite = 0;
1623 int len;
1625 if (DIFF_PAIR_UNMERGED(p)) {
1626 /* unmerged */
1627 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1628 return;
1631 name = p->one->path;
1632 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1633 name_munged = quote_one(name);
1634 other_munged = quote_one(other);
1635 one = p->one; two = p->two;
1637 diff_fill_sha1_info(one);
1638 diff_fill_sha1_info(two);
1640 len = 0;
1641 switch (p->status) {
1642 case DIFF_STATUS_COPIED:
1643 len += snprintf(msg + len, sizeof(msg) - len,
1644 "similarity index %d%%\n"
1645 "copy from %s\n"
1646 "copy to %s\n",
1647 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1648 name_munged, other_munged);
1649 break;
1650 case DIFF_STATUS_RENAMED:
1651 len += snprintf(msg + len, sizeof(msg) - len,
1652 "similarity index %d%%\n"
1653 "rename from %s\n"
1654 "rename to %s\n",
1655 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1656 name_munged, other_munged);
1657 break;
1658 case DIFF_STATUS_MODIFIED:
1659 if (p->score) {
1660 len += snprintf(msg + len, sizeof(msg) - len,
1661 "dissimilarity index %d%%\n",
1662 (int)(0.5 + p->score *
1663 100.0/MAX_SCORE));
1664 complete_rewrite = 1;
1665 break;
1667 /* fallthru */
1668 default:
1669 /* nothing */
1673 if (hashcmp(one->sha1, two->sha1)) {
1674 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1676 if (o->binary) {
1677 mmfile_t mf;
1678 if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
1679 (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
1680 abbrev = 40;
1682 len += snprintf(msg + len, sizeof(msg) - len,
1683 "index %.*s..%.*s",
1684 abbrev, sha1_to_hex(one->sha1),
1685 abbrev, sha1_to_hex(two->sha1));
1686 if (one->mode == two->mode)
1687 len += snprintf(msg + len, sizeof(msg) - len,
1688 " %06o", one->mode);
1689 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1692 if (len)
1693 msg[--len] = 0;
1694 xfrm_msg = len ? msg : NULL;
1696 if (!pgm &&
1697 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1698 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1699 /* a filepair that changes between file and symlink
1700 * needs to be split into deletion and creation.
1702 struct diff_filespec *null = alloc_filespec(two->path);
1703 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1704 free(null);
1705 null = alloc_filespec(one->path);
1706 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1707 free(null);
1709 else
1710 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1711 complete_rewrite);
1713 free(name_munged);
1714 free(other_munged);
1717 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1718 struct diffstat_t *diffstat)
1720 const char *name;
1721 const char *other;
1722 int complete_rewrite = 0;
1724 if (DIFF_PAIR_UNMERGED(p)) {
1725 /* unmerged */
1726 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1727 return;
1730 name = p->one->path;
1731 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1733 diff_fill_sha1_info(p->one);
1734 diff_fill_sha1_info(p->two);
1736 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1737 complete_rewrite = 1;
1738 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1741 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1743 const char *name;
1744 const char *other;
1746 if (DIFF_PAIR_UNMERGED(p)) {
1747 /* unmerged */
1748 return;
1751 name = p->one->path;
1752 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1754 diff_fill_sha1_info(p->one);
1755 diff_fill_sha1_info(p->two);
1757 builtin_checkdiff(name, other, p->one, p->two);
1760 void diff_setup(struct diff_options *options)
1762 memset(options, 0, sizeof(*options));
1763 options->line_termination = '\n';
1764 options->break_opt = -1;
1765 options->rename_limit = -1;
1766 options->context = 3;
1767 options->msg_sep = "";
1769 options->change = diff_change;
1770 options->add_remove = diff_addremove;
1771 options->color_diff = diff_use_color_default;
1772 options->detect_rename = diff_detect_rename_default;
1775 int diff_setup_done(struct diff_options *options)
1777 int count = 0;
1779 if (options->output_format & DIFF_FORMAT_NAME)
1780 count++;
1781 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
1782 count++;
1783 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
1784 count++;
1785 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
1786 count++;
1787 if (count > 1)
1788 die("--name-only, --name-status, --check and -s are mutually exclusive");
1790 if (options->find_copies_harder)
1791 options->detect_rename = DIFF_DETECT_COPY;
1793 if (options->output_format & (DIFF_FORMAT_NAME |
1794 DIFF_FORMAT_NAME_STATUS |
1795 DIFF_FORMAT_CHECKDIFF |
1796 DIFF_FORMAT_NO_OUTPUT))
1797 options->output_format &= ~(DIFF_FORMAT_RAW |
1798 DIFF_FORMAT_NUMSTAT |
1799 DIFF_FORMAT_DIFFSTAT |
1800 DIFF_FORMAT_SHORTSTAT |
1801 DIFF_FORMAT_SUMMARY |
1802 DIFF_FORMAT_PATCH);
1805 * These cases always need recursive; we do not drop caller-supplied
1806 * recursive bits for other formats here.
1808 if (options->output_format & (DIFF_FORMAT_PATCH |
1809 DIFF_FORMAT_NUMSTAT |
1810 DIFF_FORMAT_DIFFSTAT |
1811 DIFF_FORMAT_SHORTSTAT |
1812 DIFF_FORMAT_SUMMARY |
1813 DIFF_FORMAT_CHECKDIFF))
1814 options->recursive = 1;
1816 * Also pickaxe would not work very well if you do not say recursive
1818 if (options->pickaxe)
1819 options->recursive = 1;
1821 if (options->detect_rename && options->rename_limit < 0)
1822 options->rename_limit = diff_rename_limit_default;
1823 if (options->setup & DIFF_SETUP_USE_CACHE) {
1824 if (!active_cache)
1825 /* read-cache does not die even when it fails
1826 * so it is safe for us to do this here. Also
1827 * it does not smudge active_cache or active_nr
1828 * when it fails, so we do not have to worry about
1829 * cleaning it up ourselves either.
1831 read_cache();
1833 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1834 use_size_cache = 1;
1835 if (options->abbrev <= 0 || 40 < options->abbrev)
1836 options->abbrev = 40; /* full */
1838 return 0;
1841 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1843 char c, *eq;
1844 int len;
1846 if (*arg != '-')
1847 return 0;
1848 c = *++arg;
1849 if (!c)
1850 return 0;
1851 if (c == arg_short) {
1852 c = *++arg;
1853 if (!c)
1854 return 1;
1855 if (val && isdigit(c)) {
1856 char *end;
1857 int n = strtoul(arg, &end, 10);
1858 if (*end)
1859 return 0;
1860 *val = n;
1861 return 1;
1863 return 0;
1865 if (c != '-')
1866 return 0;
1867 arg++;
1868 eq = strchr(arg, '=');
1869 if (eq)
1870 len = eq - arg;
1871 else
1872 len = strlen(arg);
1873 if (!len || strncmp(arg, arg_long, len))
1874 return 0;
1875 if (eq) {
1876 int n;
1877 char *end;
1878 if (!isdigit(*++eq))
1879 return 0;
1880 n = strtoul(eq, &end, 10);
1881 if (*end)
1882 return 0;
1883 *val = n;
1885 return 1;
1888 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1890 const char *arg = av[0];
1891 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1892 options->output_format |= DIFF_FORMAT_PATCH;
1893 else if (opt_arg(arg, 'U', "unified", &options->context))
1894 options->output_format |= DIFF_FORMAT_PATCH;
1895 else if (!strcmp(arg, "--raw"))
1896 options->output_format |= DIFF_FORMAT_RAW;
1897 else if (!strcmp(arg, "--patch-with-raw")) {
1898 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1900 else if (!strcmp(arg, "--numstat")) {
1901 options->output_format |= DIFF_FORMAT_NUMSTAT;
1903 else if (!strcmp(arg, "--shortstat")) {
1904 options->output_format |= DIFF_FORMAT_SHORTSTAT;
1906 else if (!strncmp(arg, "--stat", 6)) {
1907 char *end;
1908 int width = options->stat_width;
1909 int name_width = options->stat_name_width;
1910 arg += 6;
1911 end = (char *)arg;
1913 switch (*arg) {
1914 case '-':
1915 if (!strncmp(arg, "-width=", 7))
1916 width = strtoul(arg + 7, &end, 10);
1917 else if (!strncmp(arg, "-name-width=", 12))
1918 name_width = strtoul(arg + 12, &end, 10);
1919 break;
1920 case '=':
1921 width = strtoul(arg+1, &end, 10);
1922 if (*end == ',')
1923 name_width = strtoul(end+1, &end, 10);
1926 /* Important! This checks all the error cases! */
1927 if (*end)
1928 return 0;
1929 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1930 options->stat_name_width = name_width;
1931 options->stat_width = width;
1933 else if (!strcmp(arg, "--check"))
1934 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1935 else if (!strcmp(arg, "--summary"))
1936 options->output_format |= DIFF_FORMAT_SUMMARY;
1937 else if (!strcmp(arg, "--patch-with-stat")) {
1938 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1940 else if (!strcmp(arg, "-z"))
1941 options->line_termination = 0;
1942 else if (!strncmp(arg, "-l", 2))
1943 options->rename_limit = strtoul(arg+2, NULL, 10);
1944 else if (!strcmp(arg, "--full-index"))
1945 options->full_index = 1;
1946 else if (!strcmp(arg, "--binary")) {
1947 options->output_format |= DIFF_FORMAT_PATCH;
1948 options->binary = 1;
1950 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
1951 options->text = 1;
1953 else if (!strcmp(arg, "--name-only"))
1954 options->output_format |= DIFF_FORMAT_NAME;
1955 else if (!strcmp(arg, "--name-status"))
1956 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1957 else if (!strcmp(arg, "-R"))
1958 options->reverse_diff = 1;
1959 else if (!strncmp(arg, "-S", 2))
1960 options->pickaxe = arg + 2;
1961 else if (!strcmp(arg, "-s")) {
1962 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1964 else if (!strncmp(arg, "-O", 2))
1965 options->orderfile = arg + 2;
1966 else if (!strncmp(arg, "--diff-filter=", 14))
1967 options->filter = arg + 14;
1968 else if (!strcmp(arg, "--pickaxe-all"))
1969 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1970 else if (!strcmp(arg, "--pickaxe-regex"))
1971 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1972 else if (!strncmp(arg, "-B", 2)) {
1973 if ((options->break_opt =
1974 diff_scoreopt_parse(arg)) == -1)
1975 return -1;
1977 else if (!strncmp(arg, "-M", 2)) {
1978 if ((options->rename_score =
1979 diff_scoreopt_parse(arg)) == -1)
1980 return -1;
1981 options->detect_rename = DIFF_DETECT_RENAME;
1983 else if (!strncmp(arg, "-C", 2)) {
1984 if ((options->rename_score =
1985 diff_scoreopt_parse(arg)) == -1)
1986 return -1;
1987 options->detect_rename = DIFF_DETECT_COPY;
1989 else if (!strcmp(arg, "--find-copies-harder"))
1990 options->find_copies_harder = 1;
1991 else if (!strcmp(arg, "--abbrev"))
1992 options->abbrev = DEFAULT_ABBREV;
1993 else if (!strncmp(arg, "--abbrev=", 9)) {
1994 options->abbrev = strtoul(arg + 9, NULL, 10);
1995 if (options->abbrev < MINIMUM_ABBREV)
1996 options->abbrev = MINIMUM_ABBREV;
1997 else if (40 < options->abbrev)
1998 options->abbrev = 40;
2000 else if (!strcmp(arg, "--color"))
2001 options->color_diff = 1;
2002 else if (!strcmp(arg, "--no-color"))
2003 options->color_diff = 0;
2004 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2005 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2006 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2007 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2008 else if (!strcmp(arg, "--color-words"))
2009 options->color_diff = options->color_diff_words = 1;
2010 else if (!strcmp(arg, "--no-renames"))
2011 options->detect_rename = 0;
2012 else
2013 return 0;
2014 return 1;
2017 static int parse_num(const char **cp_p)
2019 unsigned long num, scale;
2020 int ch, dot;
2021 const char *cp = *cp_p;
2023 num = 0;
2024 scale = 1;
2025 dot = 0;
2026 for(;;) {
2027 ch = *cp;
2028 if ( !dot && ch == '.' ) {
2029 scale = 1;
2030 dot = 1;
2031 } else if ( ch == '%' ) {
2032 scale = dot ? scale*100 : 100;
2033 cp++; /* % is always at the end */
2034 break;
2035 } else if ( ch >= '0' && ch <= '9' ) {
2036 if ( scale < 100000 ) {
2037 scale *= 10;
2038 num = (num*10) + (ch-'0');
2040 } else {
2041 break;
2043 cp++;
2045 *cp_p = cp;
2047 /* user says num divided by scale and we say internally that
2048 * is MAX_SCORE * num / scale.
2050 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
2053 int diff_scoreopt_parse(const char *opt)
2055 int opt1, opt2, cmd;
2057 if (*opt++ != '-')
2058 return -1;
2059 cmd = *opt++;
2060 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2061 return -1; /* that is not a -M, -C nor -B option */
2063 opt1 = parse_num(&opt);
2064 if (cmd != 'B')
2065 opt2 = 0;
2066 else {
2067 if (*opt == 0)
2068 opt2 = 0;
2069 else if (*opt != '/')
2070 return -1; /* we expect -B80/99 or -B80 */
2071 else {
2072 opt++;
2073 opt2 = parse_num(&opt);
2076 if (*opt != 0)
2077 return -1;
2078 return opt1 | (opt2 << 16);
2081 struct diff_queue_struct diff_queued_diff;
2083 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2085 if (queue->alloc <= queue->nr) {
2086 queue->alloc = alloc_nr(queue->alloc);
2087 queue->queue = xrealloc(queue->queue,
2088 sizeof(dp) * queue->alloc);
2090 queue->queue[queue->nr++] = dp;
2093 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2094 struct diff_filespec *one,
2095 struct diff_filespec *two)
2097 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2098 dp->one = one;
2099 dp->two = two;
2100 if (queue)
2101 diff_q(queue, dp);
2102 return dp;
2105 void diff_free_filepair(struct diff_filepair *p)
2107 diff_free_filespec_data(p->one);
2108 diff_free_filespec_data(p->two);
2109 free(p->one);
2110 free(p->two);
2111 free(p);
2114 /* This is different from find_unique_abbrev() in that
2115 * it stuffs the result with dots for alignment.
2117 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2119 int abblen;
2120 const char *abbrev;
2121 if (len == 40)
2122 return sha1_to_hex(sha1);
2124 abbrev = find_unique_abbrev(sha1, len);
2125 if (!abbrev)
2126 return sha1_to_hex(sha1);
2127 abblen = strlen(abbrev);
2128 if (abblen < 37) {
2129 static char hex[41];
2130 if (len < abblen && abblen <= len + 2)
2131 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2132 else
2133 sprintf(hex, "%s...", abbrev);
2134 return hex;
2136 return sha1_to_hex(sha1);
2139 static void diff_flush_raw(struct diff_filepair *p,
2140 struct diff_options *options)
2142 int two_paths;
2143 char status[10];
2144 int abbrev = options->abbrev;
2145 const char *path_one, *path_two;
2146 int inter_name_termination = '\t';
2147 int line_termination = options->line_termination;
2149 if (!line_termination)
2150 inter_name_termination = 0;
2152 path_one = p->one->path;
2153 path_two = p->two->path;
2154 if (line_termination) {
2155 path_one = quote_one(path_one);
2156 path_two = quote_one(path_two);
2159 if (p->score)
2160 sprintf(status, "%c%03d", p->status,
2161 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2162 else {
2163 status[0] = p->status;
2164 status[1] = 0;
2166 switch (p->status) {
2167 case DIFF_STATUS_COPIED:
2168 case DIFF_STATUS_RENAMED:
2169 two_paths = 1;
2170 break;
2171 case DIFF_STATUS_ADDED:
2172 case DIFF_STATUS_DELETED:
2173 two_paths = 0;
2174 break;
2175 default:
2176 two_paths = 0;
2177 break;
2179 if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2180 printf(":%06o %06o %s ",
2181 p->one->mode, p->two->mode,
2182 diff_unique_abbrev(p->one->sha1, abbrev));
2183 printf("%s ",
2184 diff_unique_abbrev(p->two->sha1, abbrev));
2186 printf("%s%c%s", status, inter_name_termination, path_one);
2187 if (two_paths)
2188 printf("%c%s", inter_name_termination, path_two);
2189 putchar(line_termination);
2190 if (path_one != p->one->path)
2191 free((void*)path_one);
2192 if (path_two != p->two->path)
2193 free((void*)path_two);
2196 static void diff_flush_name(struct diff_filepair *p, int line_termination)
2198 char *path = p->two->path;
2200 if (line_termination)
2201 path = quote_one(p->two->path);
2202 printf("%s%c", path, line_termination);
2203 if (p->two->path != path)
2204 free(path);
2207 int diff_unmodified_pair(struct diff_filepair *p)
2209 /* This function is written stricter than necessary to support
2210 * the currently implemented transformers, but the idea is to
2211 * let transformers to produce diff_filepairs any way they want,
2212 * and filter and clean them up here before producing the output.
2214 struct diff_filespec *one, *two;
2216 if (DIFF_PAIR_UNMERGED(p))
2217 return 0; /* unmerged is interesting */
2219 one = p->one;
2220 two = p->two;
2222 /* deletion, addition, mode or type change
2223 * and rename are all interesting.
2225 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2226 DIFF_PAIR_MODE_CHANGED(p) ||
2227 strcmp(one->path, two->path))
2228 return 0;
2230 /* both are valid and point at the same path. that is, we are
2231 * dealing with a change.
2233 if (one->sha1_valid && two->sha1_valid &&
2234 !hashcmp(one->sha1, two->sha1))
2235 return 1; /* no change */
2236 if (!one->sha1_valid && !two->sha1_valid)
2237 return 1; /* both look at the same file on the filesystem. */
2238 return 0;
2241 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2243 if (diff_unmodified_pair(p))
2244 return;
2246 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2247 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2248 return; /* no tree diffs in patch format */
2250 run_diff(p, o);
2253 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2254 struct diffstat_t *diffstat)
2256 if (diff_unmodified_pair(p))
2257 return;
2259 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2260 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2261 return; /* no tree diffs in patch format */
2263 run_diffstat(p, o, diffstat);
2266 static void diff_flush_checkdiff(struct diff_filepair *p,
2267 struct diff_options *o)
2269 if (diff_unmodified_pair(p))
2270 return;
2272 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2273 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2274 return; /* no tree diffs in patch format */
2276 run_checkdiff(p, o);
2279 int diff_queue_is_empty(void)
2281 struct diff_queue_struct *q = &diff_queued_diff;
2282 int i;
2283 for (i = 0; i < q->nr; i++)
2284 if (!diff_unmodified_pair(q->queue[i]))
2285 return 0;
2286 return 1;
2289 #if DIFF_DEBUG
2290 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2292 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2293 x, one ? one : "",
2294 s->path,
2295 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2296 s->mode,
2297 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2298 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2299 x, one ? one : "",
2300 s->size, s->xfrm_flags);
2303 void diff_debug_filepair(const struct diff_filepair *p, int i)
2305 diff_debug_filespec(p->one, i, "one");
2306 diff_debug_filespec(p->two, i, "two");
2307 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2308 p->score, p->status ? p->status : '?',
2309 p->source_stays, p->broken_pair);
2312 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2314 int i;
2315 if (msg)
2316 fprintf(stderr, "%s\n", msg);
2317 fprintf(stderr, "q->nr = %d\n", q->nr);
2318 for (i = 0; i < q->nr; i++) {
2319 struct diff_filepair *p = q->queue[i];
2320 diff_debug_filepair(p, i);
2323 #endif
2325 static void diff_resolve_rename_copy(void)
2327 int i, j;
2328 struct diff_filepair *p, *pp;
2329 struct diff_queue_struct *q = &diff_queued_diff;
2331 diff_debug_queue("resolve-rename-copy", q);
2333 for (i = 0; i < q->nr; i++) {
2334 p = q->queue[i];
2335 p->status = 0; /* undecided */
2336 if (DIFF_PAIR_UNMERGED(p))
2337 p->status = DIFF_STATUS_UNMERGED;
2338 else if (!DIFF_FILE_VALID(p->one))
2339 p->status = DIFF_STATUS_ADDED;
2340 else if (!DIFF_FILE_VALID(p->two))
2341 p->status = DIFF_STATUS_DELETED;
2342 else if (DIFF_PAIR_TYPE_CHANGED(p))
2343 p->status = DIFF_STATUS_TYPE_CHANGED;
2345 /* from this point on, we are dealing with a pair
2346 * whose both sides are valid and of the same type, i.e.
2347 * either in-place edit or rename/copy edit.
2349 else if (DIFF_PAIR_RENAME(p)) {
2350 if (p->source_stays) {
2351 p->status = DIFF_STATUS_COPIED;
2352 continue;
2354 /* See if there is some other filepair that
2355 * copies from the same source as us. If so
2356 * we are a copy. Otherwise we are either a
2357 * copy if the path stays, or a rename if it
2358 * does not, but we already handled "stays" case.
2360 for (j = i + 1; j < q->nr; j++) {
2361 pp = q->queue[j];
2362 if (strcmp(pp->one->path, p->one->path))
2363 continue; /* not us */
2364 if (!DIFF_PAIR_RENAME(pp))
2365 continue; /* not a rename/copy */
2366 /* pp is a rename/copy from the same source */
2367 p->status = DIFF_STATUS_COPIED;
2368 break;
2370 if (!p->status)
2371 p->status = DIFF_STATUS_RENAMED;
2373 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2374 p->one->mode != p->two->mode)
2375 p->status = DIFF_STATUS_MODIFIED;
2376 else {
2377 /* This is a "no-change" entry and should not
2378 * happen anymore, but prepare for broken callers.
2380 error("feeding unmodified %s to diffcore",
2381 p->one->path);
2382 p->status = DIFF_STATUS_UNKNOWN;
2385 diff_debug_queue("resolve-rename-copy done", q);
2388 static int check_pair_status(struct diff_filepair *p)
2390 switch (p->status) {
2391 case DIFF_STATUS_UNKNOWN:
2392 return 0;
2393 case 0:
2394 die("internal error in diff-resolve-rename-copy");
2395 default:
2396 return 1;
2400 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2402 int fmt = opt->output_format;
2404 if (fmt & DIFF_FORMAT_CHECKDIFF)
2405 diff_flush_checkdiff(p, opt);
2406 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2407 diff_flush_raw(p, opt);
2408 else if (fmt & DIFF_FORMAT_NAME)
2409 diff_flush_name(p, opt->line_termination);
2412 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2414 if (fs->mode)
2415 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2416 else
2417 printf(" %s %s\n", newdelete, fs->path);
2421 static void show_mode_change(struct diff_filepair *p, int show_name)
2423 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2424 if (show_name)
2425 printf(" mode change %06o => %06o %s\n",
2426 p->one->mode, p->two->mode, p->two->path);
2427 else
2428 printf(" mode change %06o => %06o\n",
2429 p->one->mode, p->two->mode);
2433 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2435 const char *old, *new;
2437 /* Find common prefix */
2438 old = p->one->path;
2439 new = p->two->path;
2440 while (1) {
2441 const char *slash_old, *slash_new;
2442 slash_old = strchr(old, '/');
2443 slash_new = strchr(new, '/');
2444 if (!slash_old ||
2445 !slash_new ||
2446 slash_old - old != slash_new - new ||
2447 memcmp(old, new, slash_new - new))
2448 break;
2449 old = slash_old + 1;
2450 new = slash_new + 1;
2452 /* p->one->path thru old is the common prefix, and old and new
2453 * through the end of names are renames
2455 if (old != p->one->path)
2456 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2457 (int)(old - p->one->path), p->one->path,
2458 old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2459 else
2460 printf(" %s %s => %s (%d%%)\n", renamecopy,
2461 p->one->path, p->two->path,
2462 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2463 show_mode_change(p, 0);
2466 static void diff_summary(struct diff_filepair *p)
2468 switch(p->status) {
2469 case DIFF_STATUS_DELETED:
2470 show_file_mode_name("delete", p->one);
2471 break;
2472 case DIFF_STATUS_ADDED:
2473 show_file_mode_name("create", p->two);
2474 break;
2475 case DIFF_STATUS_COPIED:
2476 show_rename_copy("copy", p);
2477 break;
2478 case DIFF_STATUS_RENAMED:
2479 show_rename_copy("rename", p);
2480 break;
2481 default:
2482 if (p->score) {
2483 printf(" rewrite %s (%d%%)\n", p->two->path,
2484 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2485 show_mode_change(p, 0);
2486 } else show_mode_change(p, 1);
2487 break;
2491 struct patch_id_t {
2492 struct xdiff_emit_state xm;
2493 SHA_CTX *ctx;
2494 int patchlen;
2497 static int remove_space(char *line, int len)
2499 int i;
2500 char *dst = line;
2501 unsigned char c;
2503 for (i = 0; i < len; i++)
2504 if (!isspace((c = line[i])))
2505 *dst++ = c;
2507 return dst - line;
2510 static void patch_id_consume(void *priv, char *line, unsigned long len)
2512 struct patch_id_t *data = priv;
2513 int new_len;
2515 /* Ignore line numbers when computing the SHA1 of the patch */
2516 if (!strncmp(line, "@@ -", 4))
2517 return;
2519 new_len = remove_space(line, len);
2521 SHA1_Update(data->ctx, line, new_len);
2522 data->patchlen += new_len;
2525 /* returns 0 upon success, and writes result into sha1 */
2526 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2528 struct diff_queue_struct *q = &diff_queued_diff;
2529 int i;
2530 SHA_CTX ctx;
2531 struct patch_id_t data;
2532 char buffer[PATH_MAX * 4 + 20];
2534 SHA1_Init(&ctx);
2535 memset(&data, 0, sizeof(struct patch_id_t));
2536 data.ctx = &ctx;
2537 data.xm.consume = patch_id_consume;
2539 for (i = 0; i < q->nr; i++) {
2540 xpparam_t xpp;
2541 xdemitconf_t xecfg;
2542 xdemitcb_t ecb;
2543 mmfile_t mf1, mf2;
2544 struct diff_filepair *p = q->queue[i];
2545 int len1, len2;
2547 if (p->status == 0)
2548 return error("internal diff status error");
2549 if (p->status == DIFF_STATUS_UNKNOWN)
2550 continue;
2551 if (diff_unmodified_pair(p))
2552 continue;
2553 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2554 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2555 continue;
2556 if (DIFF_PAIR_UNMERGED(p))
2557 continue;
2559 diff_fill_sha1_info(p->one);
2560 diff_fill_sha1_info(p->two);
2561 if (fill_mmfile(&mf1, p->one) < 0 ||
2562 fill_mmfile(&mf2, p->two) < 0)
2563 return error("unable to read files to diff");
2565 /* Maybe hash p->two? into the patch id? */
2566 if (mmfile_is_binary(&mf2))
2567 continue;
2569 len1 = remove_space(p->one->path, strlen(p->one->path));
2570 len2 = remove_space(p->two->path, strlen(p->two->path));
2571 if (p->one->mode == 0)
2572 len1 = snprintf(buffer, sizeof(buffer),
2573 "diff--gita/%.*sb/%.*s"
2574 "newfilemode%06o"
2575 "---/dev/null"
2576 "+++b/%.*s",
2577 len1, p->one->path,
2578 len2, p->two->path,
2579 p->two->mode,
2580 len2, p->two->path);
2581 else if (p->two->mode == 0)
2582 len1 = snprintf(buffer, sizeof(buffer),
2583 "diff--gita/%.*sb/%.*s"
2584 "deletedfilemode%06o"
2585 "---a/%.*s"
2586 "+++/dev/null",
2587 len1, p->one->path,
2588 len2, p->two->path,
2589 p->one->mode,
2590 len1, p->one->path);
2591 else
2592 len1 = snprintf(buffer, sizeof(buffer),
2593 "diff--gita/%.*sb/%.*s"
2594 "---a/%.*s"
2595 "+++b/%.*s",
2596 len1, p->one->path,
2597 len2, p->two->path,
2598 len1, p->one->path,
2599 len2, p->two->path);
2600 SHA1_Update(&ctx, buffer, len1);
2602 xpp.flags = XDF_NEED_MINIMAL;
2603 xecfg.ctxlen = 3;
2604 xecfg.flags = XDL_EMIT_FUNCNAMES;
2605 ecb.outf = xdiff_outf;
2606 ecb.priv = &data;
2607 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2610 SHA1_Final(sha1, &ctx);
2611 return 0;
2614 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2616 struct diff_queue_struct *q = &diff_queued_diff;
2617 int i;
2618 int result = diff_get_patch_id(options, sha1);
2620 for (i = 0; i < q->nr; i++)
2621 diff_free_filepair(q->queue[i]);
2623 free(q->queue);
2624 q->queue = NULL;
2625 q->nr = q->alloc = 0;
2627 return result;
2630 static int is_summary_empty(const struct diff_queue_struct *q)
2632 int i;
2634 for (i = 0; i < q->nr; i++) {
2635 const struct diff_filepair *p = q->queue[i];
2637 switch (p->status) {
2638 case DIFF_STATUS_DELETED:
2639 case DIFF_STATUS_ADDED:
2640 case DIFF_STATUS_COPIED:
2641 case DIFF_STATUS_RENAMED:
2642 return 0;
2643 default:
2644 if (p->score)
2645 return 0;
2646 if (p->one->mode && p->two->mode &&
2647 p->one->mode != p->two->mode)
2648 return 0;
2649 break;
2652 return 1;
2655 void diff_flush(struct diff_options *options)
2657 struct diff_queue_struct *q = &diff_queued_diff;
2658 int i, output_format = options->output_format;
2659 int separator = 0;
2662 * Order: raw, stat, summary, patch
2663 * or: name/name-status/checkdiff (other bits clear)
2665 if (!q->nr)
2666 goto free_queue;
2668 if (output_format & (DIFF_FORMAT_RAW |
2669 DIFF_FORMAT_NAME |
2670 DIFF_FORMAT_NAME_STATUS |
2671 DIFF_FORMAT_CHECKDIFF)) {
2672 for (i = 0; i < q->nr; i++) {
2673 struct diff_filepair *p = q->queue[i];
2674 if (check_pair_status(p))
2675 flush_one_pair(p, options);
2677 separator++;
2680 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2681 struct diffstat_t diffstat;
2683 memset(&diffstat, 0, sizeof(struct diffstat_t));
2684 diffstat.xm.consume = diffstat_consume;
2685 for (i = 0; i < q->nr; i++) {
2686 struct diff_filepair *p = q->queue[i];
2687 if (check_pair_status(p))
2688 diff_flush_stat(p, options, &diffstat);
2690 if (output_format & DIFF_FORMAT_NUMSTAT)
2691 show_numstat(&diffstat, options);
2692 if (output_format & DIFF_FORMAT_DIFFSTAT)
2693 show_stats(&diffstat, options);
2694 else if (output_format & DIFF_FORMAT_SHORTSTAT)
2695 show_shortstats(&diffstat);
2696 separator++;
2699 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2700 for (i = 0; i < q->nr; i++)
2701 diff_summary(q->queue[i]);
2702 separator++;
2705 if (output_format & DIFF_FORMAT_PATCH) {
2706 if (separator) {
2707 if (options->stat_sep) {
2708 /* attach patch instead of inline */
2709 fputs(options->stat_sep, stdout);
2710 } else {
2711 putchar(options->line_termination);
2715 for (i = 0; i < q->nr; i++) {
2716 struct diff_filepair *p = q->queue[i];
2717 if (check_pair_status(p))
2718 diff_flush_patch(p, options);
2722 if (output_format & DIFF_FORMAT_CALLBACK)
2723 options->format_callback(q, options, options->format_callback_data);
2725 for (i = 0; i < q->nr; i++)
2726 diff_free_filepair(q->queue[i]);
2727 free_queue:
2728 free(q->queue);
2729 q->queue = NULL;
2730 q->nr = q->alloc = 0;
2733 static void diffcore_apply_filter(const char *filter)
2735 int i;
2736 struct diff_queue_struct *q = &diff_queued_diff;
2737 struct diff_queue_struct outq;
2738 outq.queue = NULL;
2739 outq.nr = outq.alloc = 0;
2741 if (!filter)
2742 return;
2744 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2745 int found;
2746 for (i = found = 0; !found && i < q->nr; i++) {
2747 struct diff_filepair *p = q->queue[i];
2748 if (((p->status == DIFF_STATUS_MODIFIED) &&
2749 ((p->score &&
2750 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2751 (!p->score &&
2752 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2753 ((p->status != DIFF_STATUS_MODIFIED) &&
2754 strchr(filter, p->status)))
2755 found++;
2757 if (found)
2758 return;
2760 /* otherwise we will clear the whole queue
2761 * by copying the empty outq at the end of this
2762 * function, but first clear the current entries
2763 * in the queue.
2765 for (i = 0; i < q->nr; i++)
2766 diff_free_filepair(q->queue[i]);
2768 else {
2769 /* Only the matching ones */
2770 for (i = 0; i < q->nr; i++) {
2771 struct diff_filepair *p = q->queue[i];
2773 if (((p->status == DIFF_STATUS_MODIFIED) &&
2774 ((p->score &&
2775 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2776 (!p->score &&
2777 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2778 ((p->status != DIFF_STATUS_MODIFIED) &&
2779 strchr(filter, p->status)))
2780 diff_q(&outq, p);
2781 else
2782 diff_free_filepair(p);
2785 free(q->queue);
2786 *q = outq;
2789 void diffcore_std(struct diff_options *options)
2791 if (options->break_opt != -1)
2792 diffcore_break(options->break_opt);
2793 if (options->detect_rename)
2794 diffcore_rename(options);
2795 if (options->break_opt != -1)
2796 diffcore_merge_broken();
2797 if (options->pickaxe)
2798 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2799 if (options->orderfile)
2800 diffcore_order(options->orderfile);
2801 diff_resolve_rename_copy();
2802 diffcore_apply_filter(options->filter);
2806 void diffcore_std_no_resolve(struct diff_options *options)
2808 if (options->pickaxe)
2809 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2810 if (options->orderfile)
2811 diffcore_order(options->orderfile);
2812 diffcore_apply_filter(options->filter);
2815 void diff_addremove(struct diff_options *options,
2816 int addremove, unsigned mode,
2817 const unsigned char *sha1,
2818 const char *base, const char *path)
2820 char concatpath[PATH_MAX];
2821 struct diff_filespec *one, *two;
2823 /* This may look odd, but it is a preparation for
2824 * feeding "there are unchanged files which should
2825 * not produce diffs, but when you are doing copy
2826 * detection you would need them, so here they are"
2827 * entries to the diff-core. They will be prefixed
2828 * with something like '=' or '*' (I haven't decided
2829 * which but should not make any difference).
2830 * Feeding the same new and old to diff_change()
2831 * also has the same effect.
2832 * Before the final output happens, they are pruned after
2833 * merged into rename/copy pairs as appropriate.
2835 if (options->reverse_diff)
2836 addremove = (addremove == '+' ? '-' :
2837 addremove == '-' ? '+' : addremove);
2839 if (!path) path = "";
2840 sprintf(concatpath, "%s%s", base, path);
2841 one = alloc_filespec(concatpath);
2842 two = alloc_filespec(concatpath);
2844 if (addremove != '+')
2845 fill_filespec(one, sha1, mode);
2846 if (addremove != '-')
2847 fill_filespec(two, sha1, mode);
2849 diff_queue(&diff_queued_diff, one, two);
2852 void diff_change(struct diff_options *options,
2853 unsigned old_mode, unsigned new_mode,
2854 const unsigned char *old_sha1,
2855 const unsigned char *new_sha1,
2856 const char *base, const char *path)
2858 char concatpath[PATH_MAX];
2859 struct diff_filespec *one, *two;
2861 if (options->reverse_diff) {
2862 unsigned tmp;
2863 const unsigned char *tmp_c;
2864 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2865 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2867 if (!path) path = "";
2868 sprintf(concatpath, "%s%s", base, path);
2869 one = alloc_filespec(concatpath);
2870 two = alloc_filespec(concatpath);
2871 fill_filespec(one, old_sha1, old_mode);
2872 fill_filespec(two, new_sha1, new_mode);
2874 diff_queue(&diff_queued_diff, one, two);
2877 void diff_unmerge(struct diff_options *options,
2878 const char *path)
2880 struct diff_filespec *one, *two;
2881 one = alloc_filespec(path);
2882 two = alloc_filespec(path);
2883 diff_queue(&diff_queued_diff, one, two);