blame: -b (blame.blankboundary) and --root (blame.showroot)
[git/gitweb.git] / diff.c
blob99744354a77418ab462b6b797c6ab334cc364960
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 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
21 static int use_size_cache;
23 static int diff_detect_rename_default;
24 static int diff_rename_limit_default = -1;
25 static int diff_use_color_default;
27 static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
38 static int parse_diff_color_slot(const char *var, int ofs)
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
58 * These are to give UI layer defaults.
59 * The core-level commands such as git-diff-files should
60 * never be affected by the setting of diff.renames
61 * the user happens to have in the configuration file.
63 int git_diff_ui_config(const char *var, const char *value)
65 if (!strcmp(var, "diff.renamelimit")) {
66 diff_rename_limit_default = git_config_int(var, value);
67 return 0;
69 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
70 diff_use_color_default = git_config_colorbool(var, value);
71 return 0;
73 if (!strcmp(var, "diff.renames")) {
74 if (!value)
75 diff_detect_rename_default = DIFF_DETECT_RENAME;
76 else if (!strcasecmp(value, "copies") ||
77 !strcasecmp(value, "copy"))
78 diff_detect_rename_default = DIFF_DETECT_COPY;
79 else if (git_config_bool(var,value))
80 diff_detect_rename_default = DIFF_DETECT_RENAME;
81 return 0;
83 if (!strncmp(var, "diff.color.", 11) || !strncmp(var, "color.diff.", 11)) {
84 int slot = parse_diff_color_slot(var, 11);
85 color_parse(value, var, diff_colors[slot]);
86 return 0;
88 return git_default_config(var, value);
91 static char *quote_one(const char *str)
93 int needlen;
94 char *xp;
96 if (!str)
97 return NULL;
98 needlen = quote_c_style(str, NULL, NULL, 0);
99 if (!needlen)
100 return xstrdup(str);
101 xp = xmalloc(needlen + 1);
102 quote_c_style(str, xp, NULL, 0);
103 return xp;
106 static char *quote_two(const char *one, const char *two)
108 int need_one = quote_c_style(one, NULL, NULL, 1);
109 int need_two = quote_c_style(two, NULL, NULL, 1);
110 char *xp;
112 if (need_one + need_two) {
113 if (!need_one) need_one = strlen(one);
114 if (!need_two) need_one = strlen(two);
116 xp = xmalloc(need_one + need_two + 3);
117 xp[0] = '"';
118 quote_c_style(one, xp + 1, NULL, 1);
119 quote_c_style(two, xp + need_one + 1, NULL, 1);
120 strcpy(xp + need_one + need_two + 1, "\"");
121 return xp;
123 need_one = strlen(one);
124 need_two = strlen(two);
125 xp = xmalloc(need_one + need_two + 1);
126 strcpy(xp, one);
127 strcpy(xp + need_one, two);
128 return xp;
131 static const char *external_diff(void)
133 static const char *external_diff_cmd = NULL;
134 static int done_preparing = 0;
136 if (done_preparing)
137 return external_diff_cmd;
138 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
139 done_preparing = 1;
140 return external_diff_cmd;
143 #define TEMPFILE_PATH_LEN 50
145 static struct diff_tempfile {
146 const char *name; /* filename external diff should read from */
147 char hex[41];
148 char mode[10];
149 char tmp_path[TEMPFILE_PATH_LEN];
150 } diff_temp[2];
152 static int count_lines(const char *data, int size)
154 int count, ch, completely_empty = 1, nl_just_seen = 0;
155 count = 0;
156 while (0 < size--) {
157 ch = *data++;
158 if (ch == '\n') {
159 count++;
160 nl_just_seen = 1;
161 completely_empty = 0;
163 else {
164 nl_just_seen = 0;
165 completely_empty = 0;
168 if (completely_empty)
169 return 0;
170 if (!nl_just_seen)
171 count++; /* no trailing newline */
172 return count;
175 static void print_line_count(int count)
177 switch (count) {
178 case 0:
179 printf("0,0");
180 break;
181 case 1:
182 printf("1");
183 break;
184 default:
185 printf("1,%d", count);
186 break;
190 static void copy_file(int prefix, const char *data, int size)
192 int ch, nl_just_seen = 1;
193 while (0 < size--) {
194 ch = *data++;
195 if (nl_just_seen)
196 putchar(prefix);
197 putchar(ch);
198 if (ch == '\n')
199 nl_just_seen = 1;
200 else
201 nl_just_seen = 0;
203 if (!nl_just_seen)
204 printf("\n\\ No newline at end of file\n");
207 static void emit_rewrite_diff(const char *name_a,
208 const char *name_b,
209 struct diff_filespec *one,
210 struct diff_filespec *two)
212 int lc_a, lc_b;
213 diff_populate_filespec(one, 0);
214 diff_populate_filespec(two, 0);
215 lc_a = count_lines(one->data, one->size);
216 lc_b = count_lines(two->data, two->size);
217 printf("--- a/%s\n+++ b/%s\n@@ -", name_a, name_b);
218 print_line_count(lc_a);
219 printf(" +");
220 print_line_count(lc_b);
221 printf(" @@\n");
222 if (lc_a)
223 copy_file('-', one->data, one->size);
224 if (lc_b)
225 copy_file('+', two->data, two->size);
228 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
230 if (!DIFF_FILE_VALID(one)) {
231 mf->ptr = (char *)""; /* does not matter */
232 mf->size = 0;
233 return 0;
235 else if (diff_populate_filespec(one, 0))
236 return -1;
237 mf->ptr = one->data;
238 mf->size = one->size;
239 return 0;
242 struct diff_words_buffer {
243 mmfile_t text;
244 long alloc;
245 long current; /* output pointer */
246 int suppressed_newline;
249 static void diff_words_append(char *line, unsigned long len,
250 struct diff_words_buffer *buffer)
252 if (buffer->text.size + len > buffer->alloc) {
253 buffer->alloc = (buffer->text.size + len) * 3 / 2;
254 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
256 line++;
257 len--;
258 memcpy(buffer->text.ptr + buffer->text.size, line, len);
259 buffer->text.size += len;
262 struct diff_words_data {
263 struct xdiff_emit_state xm;
264 struct diff_words_buffer minus, plus;
267 static void print_word(struct diff_words_buffer *buffer, int len, int color,
268 int suppress_newline)
270 const char *ptr;
271 int eol = 0;
273 if (len == 0)
274 return;
276 ptr = buffer->text.ptr + buffer->current;
277 buffer->current += len;
279 if (ptr[len - 1] == '\n') {
280 eol = 1;
281 len--;
284 fputs(diff_get_color(1, color), stdout);
285 fwrite(ptr, len, 1, stdout);
286 fputs(diff_get_color(1, DIFF_RESET), stdout);
288 if (eol) {
289 if (suppress_newline)
290 buffer->suppressed_newline = 1;
291 else
292 putchar('\n');
296 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
298 struct diff_words_data *diff_words = priv;
300 if (diff_words->minus.suppressed_newline) {
301 if (line[0] != '+')
302 putchar('\n');
303 diff_words->minus.suppressed_newline = 0;
306 len--;
307 switch (line[0]) {
308 case '-':
309 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
310 break;
311 case '+':
312 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
313 break;
314 case ' ':
315 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
316 diff_words->minus.current += len;
317 break;
321 /* this executes the word diff on the accumulated buffers */
322 static void diff_words_show(struct diff_words_data *diff_words)
324 xpparam_t xpp;
325 xdemitconf_t xecfg;
326 xdemitcb_t ecb;
327 mmfile_t minus, plus;
328 int i;
330 minus.size = diff_words->minus.text.size;
331 minus.ptr = xmalloc(minus.size);
332 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
333 for (i = 0; i < minus.size; i++)
334 if (isspace(minus.ptr[i]))
335 minus.ptr[i] = '\n';
336 diff_words->minus.current = 0;
338 plus.size = diff_words->plus.text.size;
339 plus.ptr = xmalloc(plus.size);
340 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
341 for (i = 0; i < plus.size; i++)
342 if (isspace(plus.ptr[i]))
343 plus.ptr[i] = '\n';
344 diff_words->plus.current = 0;
346 xpp.flags = XDF_NEED_MINIMAL;
347 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
348 xecfg.flags = 0;
349 ecb.outf = xdiff_outf;
350 ecb.priv = diff_words;
351 diff_words->xm.consume = fn_out_diff_words_aux;
352 xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
354 free(minus.ptr);
355 free(plus.ptr);
356 diff_words->minus.text.size = diff_words->plus.text.size = 0;
358 if (diff_words->minus.suppressed_newline) {
359 putchar('\n');
360 diff_words->minus.suppressed_newline = 0;
364 struct emit_callback {
365 struct xdiff_emit_state xm;
366 int nparents, color_diff;
367 const char **label_path;
368 struct diff_words_data *diff_words;
371 static void free_diff_words_data(struct emit_callback *ecbdata)
373 if (ecbdata->diff_words) {
374 /* flush buffers */
375 if (ecbdata->diff_words->minus.text.size ||
376 ecbdata->diff_words->plus.text.size)
377 diff_words_show(ecbdata->diff_words);
379 if (ecbdata->diff_words->minus.text.ptr)
380 free (ecbdata->diff_words->minus.text.ptr);
381 if (ecbdata->diff_words->plus.text.ptr)
382 free (ecbdata->diff_words->plus.text.ptr);
383 free(ecbdata->diff_words);
384 ecbdata->diff_words = NULL;
388 const char *diff_get_color(int diff_use_color, enum color_diff ix)
390 if (diff_use_color)
391 return diff_colors[ix];
392 return "";
395 static void emit_line(const char *set, const char *reset, const char *line, int len)
397 if (len > 0 && line[len-1] == '\n')
398 len--;
399 fputs(set, stdout);
400 fwrite(line, len, 1, stdout);
401 puts(reset);
404 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
406 int col0 = ecbdata->nparents;
407 int last_tab_in_indent = -1;
408 int last_space_in_indent = -1;
409 int i;
410 int tail = len;
411 int need_highlight_leading_space = 0;
412 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
413 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
415 if (!*ws) {
416 emit_line(set, reset, line, len);
417 return;
420 /* The line is a newly added line. Does it have funny leading
421 * whitespaces? In indent, SP should never precede a TAB.
423 for (i = col0; i < len; i++) {
424 if (line[i] == '\t') {
425 last_tab_in_indent = i;
426 if (0 <= last_space_in_indent)
427 need_highlight_leading_space = 1;
429 else if (line[i] == ' ')
430 last_space_in_indent = i;
431 else
432 break;
434 fputs(set, stdout);
435 fwrite(line, col0, 1, stdout);
436 fputs(reset, stdout);
437 if (((i == len) || line[i] == '\n') && i != col0) {
438 /* The whole line was indent */
439 emit_line(ws, reset, line + col0, len - col0);
440 return;
442 i = col0;
443 if (need_highlight_leading_space) {
444 while (i < last_tab_in_indent) {
445 if (line[i] == ' ') {
446 fputs(ws, stdout);
447 putchar(' ');
448 fputs(reset, stdout);
450 else
451 putchar(line[i]);
452 i++;
455 tail = len - 1;
456 if (line[tail] == '\n' && i < tail)
457 tail--;
458 while (i < tail) {
459 if (!isspace(line[tail]))
460 break;
461 tail--;
463 if ((i < tail && line[tail + 1] != '\n')) {
464 /* This has whitespace between tail+1..len */
465 fputs(set, stdout);
466 fwrite(line + i, tail - i + 1, 1, stdout);
467 fputs(reset, stdout);
468 emit_line(ws, reset, line + tail + 1, len - tail - 1);
470 else
471 emit_line(set, reset, line + i, len - i);
474 static void fn_out_consume(void *priv, char *line, unsigned long len)
476 int i;
477 int color;
478 struct emit_callback *ecbdata = priv;
479 const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
480 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
482 if (ecbdata->label_path[0]) {
483 printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
484 printf("%s+++ %s%s\n", set, ecbdata->label_path[1], reset);
485 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
488 /* This is not really necessary for now because
489 * this codepath only deals with two-way diffs.
491 for (i = 0; i < len && line[i] == '@'; i++)
493 if (2 <= i && i < len && line[i] == ' ') {
494 ecbdata->nparents = i - 1;
495 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
496 reset, line, len);
497 return;
500 if (len < ecbdata->nparents) {
501 set = reset;
502 emit_line(reset, reset, line, len);
503 return;
506 color = DIFF_PLAIN;
507 if (ecbdata->diff_words && ecbdata->nparents != 1)
508 /* fall back to normal diff */
509 free_diff_words_data(ecbdata);
510 if (ecbdata->diff_words) {
511 if (line[0] == '-') {
512 diff_words_append(line, len,
513 &ecbdata->diff_words->minus);
514 return;
515 } else if (line[0] == '+') {
516 diff_words_append(line, len,
517 &ecbdata->diff_words->plus);
518 return;
520 if (ecbdata->diff_words->minus.text.size ||
521 ecbdata->diff_words->plus.text.size)
522 diff_words_show(ecbdata->diff_words);
523 line++;
524 len--;
525 emit_line(set, reset, line, len);
526 return;
528 for (i = 0; i < ecbdata->nparents && len; i++) {
529 if (line[i] == '-')
530 color = DIFF_FILE_OLD;
531 else if (line[i] == '+')
532 color = DIFF_FILE_NEW;
535 if (color != DIFF_FILE_NEW) {
536 emit_line(diff_get_color(ecbdata->color_diff, color),
537 reset, line, len);
538 return;
540 emit_add_line(reset, ecbdata, line, len);
543 static char *pprint_rename(const char *a, const char *b)
545 const char *old = a;
546 const char *new = b;
547 char *name = NULL;
548 int pfx_length, sfx_length;
549 int len_a = strlen(a);
550 int len_b = strlen(b);
552 /* Find common prefix */
553 pfx_length = 0;
554 while (*old && *new && *old == *new) {
555 if (*old == '/')
556 pfx_length = old - a + 1;
557 old++;
558 new++;
561 /* Find common suffix */
562 old = a + len_a;
563 new = b + len_b;
564 sfx_length = 0;
565 while (a <= old && b <= new && *old == *new) {
566 if (*old == '/')
567 sfx_length = len_a - (old - a);
568 old--;
569 new--;
573 * pfx{mid-a => mid-b}sfx
574 * {pfx-a => pfx-b}sfx
575 * pfx{sfx-a => sfx-b}
576 * name-a => name-b
578 if (pfx_length + sfx_length) {
579 int a_midlen = len_a - pfx_length - sfx_length;
580 int b_midlen = len_b - pfx_length - sfx_length;
581 if (a_midlen < 0) a_midlen = 0;
582 if (b_midlen < 0) b_midlen = 0;
584 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
585 sprintf(name, "%.*s{%.*s => %.*s}%s",
586 pfx_length, a,
587 a_midlen, a + pfx_length,
588 b_midlen, b + pfx_length,
589 a + len_a - sfx_length);
591 else {
592 name = xmalloc(len_a + len_b + 5);
593 sprintf(name, "%s => %s", a, b);
595 return name;
598 struct diffstat_t {
599 struct xdiff_emit_state xm;
601 int nr;
602 int alloc;
603 struct diffstat_file {
604 char *name;
605 unsigned is_unmerged:1;
606 unsigned is_binary:1;
607 unsigned is_renamed:1;
608 unsigned int added, deleted;
609 } **files;
612 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
613 const char *name_a,
614 const char *name_b)
616 struct diffstat_file *x;
617 x = xcalloc(sizeof (*x), 1);
618 if (diffstat->nr == diffstat->alloc) {
619 diffstat->alloc = alloc_nr(diffstat->alloc);
620 diffstat->files = xrealloc(diffstat->files,
621 diffstat->alloc * sizeof(x));
623 diffstat->files[diffstat->nr++] = x;
624 if (name_b) {
625 x->name = pprint_rename(name_a, name_b);
626 x->is_renamed = 1;
628 else
629 x->name = xstrdup(name_a);
630 return x;
633 static void diffstat_consume(void *priv, char *line, unsigned long len)
635 struct diffstat_t *diffstat = priv;
636 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
638 if (line[0] == '+')
639 x->added++;
640 else if (line[0] == '-')
641 x->deleted++;
644 const char mime_boundary_leader[] = "------------";
646 static int scale_linear(int it, int width, int max_change)
649 * make sure that at least one '-' is printed if there were deletions,
650 * and likewise for '+'.
652 if (max_change < 2)
653 return it;
654 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
657 static void show_name(const char *prefix, const char *name, int len,
658 const char *reset, const char *set)
660 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
663 static void show_graph(char ch, int cnt, const char *set, const char *reset)
665 if (cnt <= 0)
666 return;
667 printf("%s", set);
668 while (cnt--)
669 putchar(ch);
670 printf("%s", reset);
673 static void show_stats(struct diffstat_t* data, struct diff_options *options)
675 int i, len, add, del, total, adds = 0, dels = 0;
676 int max_change = 0, max_len = 0;
677 int total_files = data->nr;
678 int width, name_width;
679 const char *reset, *set, *add_c, *del_c;
681 if (data->nr == 0)
682 return;
684 width = options->stat_width ? options->stat_width : 80;
685 name_width = options->stat_name_width ? options->stat_name_width : 50;
687 /* Sanity: give at least 5 columns to the graph,
688 * but leave at least 10 columns for the name.
690 if (width < name_width + 15) {
691 if (name_width <= 25)
692 width = name_width + 15;
693 else
694 name_width = width - 15;
697 /* Find the longest filename and max number of changes */
698 reset = diff_get_color(options->color_diff, DIFF_RESET);
699 set = diff_get_color(options->color_diff, DIFF_PLAIN);
700 add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
701 del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
703 for (i = 0; i < data->nr; i++) {
704 struct diffstat_file *file = data->files[i];
705 int change = file->added + file->deleted;
707 len = quote_c_style(file->name, NULL, NULL, 0);
708 if (len) {
709 char *qname = xmalloc(len + 1);
710 quote_c_style(file->name, qname, NULL, 0);
711 free(file->name);
712 file->name = qname;
715 len = strlen(file->name);
716 if (max_len < len)
717 max_len = len;
719 if (file->is_binary || file->is_unmerged)
720 continue;
721 if (max_change < change)
722 max_change = change;
725 /* Compute the width of the graph part;
726 * 10 is for one blank at the beginning of the line plus
727 * " | count " between the name and the graph.
729 * From here on, name_width is the width of the name area,
730 * and width is the width of the graph area.
732 name_width = (name_width < max_len) ? name_width : max_len;
733 if (width < (name_width + 10) + max_change)
734 width = width - (name_width + 10);
735 else
736 width = max_change;
738 for (i = 0; i < data->nr; i++) {
739 const char *prefix = "";
740 char *name = data->files[i]->name;
741 int added = data->files[i]->added;
742 int deleted = data->files[i]->deleted;
743 int name_len;
746 * "scale" the filename
748 len = name_width;
749 name_len = strlen(name);
750 if (name_width < name_len) {
751 char *slash;
752 prefix = "...";
753 len -= 3;
754 name += name_len - len;
755 slash = strchr(name, '/');
756 if (slash)
757 name = slash;
760 if (data->files[i]->is_binary) {
761 show_name(prefix, name, len, reset, set);
762 printf(" Bin\n");
763 goto free_diffstat_file;
765 else if (data->files[i]->is_unmerged) {
766 show_name(prefix, name, len, reset, set);
767 printf(" Unmerged\n");
768 goto free_diffstat_file;
770 else if (!data->files[i]->is_renamed &&
771 (added + deleted == 0)) {
772 total_files--;
773 goto free_diffstat_file;
777 * scale the add/delete
779 add = added;
780 del = deleted;
781 total = add + del;
782 adds += add;
783 dels += del;
785 if (width <= max_change) {
786 add = scale_linear(add, width, max_change);
787 del = scale_linear(del, width, max_change);
788 total = add + del;
790 show_name(prefix, name, len, reset, set);
791 printf("%5d ", added + deleted);
792 show_graph('+', add, add_c, reset);
793 show_graph('-', del, del_c, reset);
794 putchar('\n');
795 free_diffstat_file:
796 free(data->files[i]->name);
797 free(data->files[i]);
799 free(data->files);
800 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
801 set, total_files, adds, dels, reset);
804 static void show_shortstats(struct diffstat_t* data)
806 int i, adds = 0, dels = 0, total_files = data->nr;
808 if (data->nr == 0)
809 return;
811 for (i = 0; i < data->nr; i++) {
812 if (!data->files[i]->is_binary &&
813 !data->files[i]->is_unmerged) {
814 int added = data->files[i]->added;
815 int deleted= data->files[i]->deleted;
816 if (!data->files[i]->is_renamed &&
817 (added + deleted == 0)) {
818 total_files--;
819 } else {
820 adds += added;
821 dels += deleted;
824 free(data->files[i]->name);
825 free(data->files[i]);
827 free(data->files);
829 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
830 total_files, adds, dels);
833 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
835 int i;
837 for (i = 0; i < data->nr; i++) {
838 struct diffstat_file *file = data->files[i];
840 if (file->is_binary)
841 printf("-\t-\t");
842 else
843 printf("%d\t%d\t", file->added, file->deleted);
844 if (options->line_termination &&
845 quote_c_style(file->name, NULL, NULL, 0))
846 quote_c_style(file->name, NULL, stdout, 0);
847 else
848 fputs(file->name, stdout);
849 putchar(options->line_termination);
853 struct checkdiff_t {
854 struct xdiff_emit_state xm;
855 const char *filename;
856 int lineno;
859 static void checkdiff_consume(void *priv, char *line, unsigned long len)
861 struct checkdiff_t *data = priv;
863 if (line[0] == '+') {
864 int i, spaces = 0;
866 data->lineno++;
868 /* check space before tab */
869 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
870 if (line[i] == ' ')
871 spaces++;
872 if (line[i - 1] == '\t' && spaces)
873 printf("%s:%d: space before tab:%.*s\n",
874 data->filename, data->lineno, (int)len, line);
876 /* check white space at line end */
877 if (line[len - 1] == '\n')
878 len--;
879 if (isspace(line[len - 1]))
880 printf("%s:%d: white space at end: %.*s\n",
881 data->filename, data->lineno, (int)len, line);
882 } else if (line[0] == ' ')
883 data->lineno++;
884 else if (line[0] == '@') {
885 char *plus = strchr(line, '+');
886 if (plus)
887 data->lineno = strtol(plus, NULL, 10);
888 else
889 die("invalid diff");
893 static unsigned char *deflate_it(char *data,
894 unsigned long size,
895 unsigned long *result_size)
897 int bound;
898 unsigned char *deflated;
899 z_stream stream;
901 memset(&stream, 0, sizeof(stream));
902 deflateInit(&stream, zlib_compression_level);
903 bound = deflateBound(&stream, size);
904 deflated = xmalloc(bound);
905 stream.next_out = deflated;
906 stream.avail_out = bound;
908 stream.next_in = (unsigned char *)data;
909 stream.avail_in = size;
910 while (deflate(&stream, Z_FINISH) == Z_OK)
911 ; /* nothing */
912 deflateEnd(&stream);
913 *result_size = stream.total_out;
914 return deflated;
917 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
919 void *cp;
920 void *delta;
921 void *deflated;
922 void *data;
923 unsigned long orig_size;
924 unsigned long delta_size;
925 unsigned long deflate_size;
926 unsigned long data_size;
928 /* We could do deflated delta, or we could do just deflated two,
929 * whichever is smaller.
931 delta = NULL;
932 deflated = deflate_it(two->ptr, two->size, &deflate_size);
933 if (one->size && two->size) {
934 delta = diff_delta(one->ptr, one->size,
935 two->ptr, two->size,
936 &delta_size, deflate_size);
937 if (delta) {
938 void *to_free = delta;
939 orig_size = delta_size;
940 delta = deflate_it(delta, delta_size, &delta_size);
941 free(to_free);
945 if (delta && delta_size < deflate_size) {
946 printf("delta %lu\n", orig_size);
947 free(deflated);
948 data = delta;
949 data_size = delta_size;
951 else {
952 printf("literal %lu\n", two->size);
953 free(delta);
954 data = deflated;
955 data_size = deflate_size;
958 /* emit data encoded in base85 */
959 cp = data;
960 while (data_size) {
961 int bytes = (52 < data_size) ? 52 : data_size;
962 char line[70];
963 data_size -= bytes;
964 if (bytes <= 26)
965 line[0] = bytes + 'A' - 1;
966 else
967 line[0] = bytes - 26 + 'a' - 1;
968 encode_85(line + 1, cp, bytes);
969 cp = (char *) cp + bytes;
970 puts(line);
972 printf("\n");
973 free(data);
976 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
978 printf("GIT binary patch\n");
979 emit_binary_diff_body(one, two);
980 emit_binary_diff_body(two, one);
983 #define FIRST_FEW_BYTES 8000
984 static int mmfile_is_binary(mmfile_t *mf)
986 long sz = mf->size;
987 if (FIRST_FEW_BYTES < sz)
988 sz = FIRST_FEW_BYTES;
989 return !!memchr(mf->ptr, 0, sz);
992 static void builtin_diff(const char *name_a,
993 const char *name_b,
994 struct diff_filespec *one,
995 struct diff_filespec *two,
996 const char *xfrm_msg,
997 struct diff_options *o,
998 int complete_rewrite)
1000 mmfile_t mf1, mf2;
1001 const char *lbl[2];
1002 char *a_one, *b_two;
1003 const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1004 const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1006 a_one = quote_two("a/", name_a);
1007 b_two = quote_two("b/", name_b);
1008 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1009 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1010 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1011 if (lbl[0][0] == '/') {
1012 /* /dev/null */
1013 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1014 if (xfrm_msg && xfrm_msg[0])
1015 printf("%s%s%s\n", set, xfrm_msg, reset);
1017 else if (lbl[1][0] == '/') {
1018 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1019 if (xfrm_msg && xfrm_msg[0])
1020 printf("%s%s%s\n", set, xfrm_msg, reset);
1022 else {
1023 if (one->mode != two->mode) {
1024 printf("%sold mode %06o%s\n", set, one->mode, reset);
1025 printf("%snew mode %06o%s\n", set, two->mode, reset);
1027 if (xfrm_msg && xfrm_msg[0])
1028 printf("%s%s%s\n", set, xfrm_msg, reset);
1030 * we do not run diff between different kind
1031 * of objects.
1033 if ((one->mode ^ two->mode) & S_IFMT)
1034 goto free_ab_and_return;
1035 if (complete_rewrite) {
1036 emit_rewrite_diff(name_a, name_b, one, two);
1037 goto free_ab_and_return;
1041 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1042 die("unable to read files to diff");
1044 if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
1045 /* Quite common confusing case */
1046 if (mf1.size == mf2.size &&
1047 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1048 goto free_ab_and_return;
1049 if (o->binary)
1050 emit_binary_diff(&mf1, &mf2);
1051 else
1052 printf("Binary files %s and %s differ\n",
1053 lbl[0], lbl[1]);
1055 else {
1056 /* Crazy xdl interfaces.. */
1057 const char *diffopts = getenv("GIT_DIFF_OPTS");
1058 xpparam_t xpp;
1059 xdemitconf_t xecfg;
1060 xdemitcb_t ecb;
1061 struct emit_callback ecbdata;
1063 memset(&ecbdata, 0, sizeof(ecbdata));
1064 ecbdata.label_path = lbl;
1065 ecbdata.color_diff = o->color_diff;
1066 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1067 xecfg.ctxlen = o->context;
1068 xecfg.flags = XDL_EMIT_FUNCNAMES;
1069 if (!diffopts)
1071 else if (!strncmp(diffopts, "--unified=", 10))
1072 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1073 else if (!strncmp(diffopts, "-u", 2))
1074 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1075 ecb.outf = xdiff_outf;
1076 ecb.priv = &ecbdata;
1077 ecbdata.xm.consume = fn_out_consume;
1078 if (o->color_diff_words)
1079 ecbdata.diff_words =
1080 xcalloc(1, sizeof(struct diff_words_data));
1081 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1082 if (o->color_diff_words)
1083 free_diff_words_data(&ecbdata);
1086 free_ab_and_return:
1087 free(a_one);
1088 free(b_two);
1089 return;
1092 static void builtin_diffstat(const char *name_a, const char *name_b,
1093 struct diff_filespec *one,
1094 struct diff_filespec *two,
1095 struct diffstat_t *diffstat,
1096 struct diff_options *o,
1097 int complete_rewrite)
1099 mmfile_t mf1, mf2;
1100 struct diffstat_file *data;
1102 data = diffstat_add(diffstat, name_a, name_b);
1104 if (!one || !two) {
1105 data->is_unmerged = 1;
1106 return;
1108 if (complete_rewrite) {
1109 diff_populate_filespec(one, 0);
1110 diff_populate_filespec(two, 0);
1111 data->deleted = count_lines(one->data, one->size);
1112 data->added = count_lines(two->data, two->size);
1113 return;
1115 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1116 die("unable to read files to diff");
1118 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
1119 data->is_binary = 1;
1120 else {
1121 /* Crazy xdl interfaces.. */
1122 xpparam_t xpp;
1123 xdemitconf_t xecfg;
1124 xdemitcb_t ecb;
1126 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1127 xecfg.ctxlen = 0;
1128 xecfg.flags = 0;
1129 ecb.outf = xdiff_outf;
1130 ecb.priv = diffstat;
1131 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1135 static void builtin_checkdiff(const char *name_a, const char *name_b,
1136 struct diff_filespec *one,
1137 struct diff_filespec *two)
1139 mmfile_t mf1, mf2;
1140 struct checkdiff_t data;
1142 if (!two)
1143 return;
1145 memset(&data, 0, sizeof(data));
1146 data.xm.consume = checkdiff_consume;
1147 data.filename = name_b ? name_b : name_a;
1148 data.lineno = 0;
1150 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1151 die("unable to read files to diff");
1153 if (mmfile_is_binary(&mf2))
1154 return;
1155 else {
1156 /* Crazy xdl interfaces.. */
1157 xpparam_t xpp;
1158 xdemitconf_t xecfg;
1159 xdemitcb_t ecb;
1161 xpp.flags = XDF_NEED_MINIMAL;
1162 xecfg.ctxlen = 0;
1163 xecfg.flags = 0;
1164 ecb.outf = xdiff_outf;
1165 ecb.priv = &data;
1166 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1170 struct diff_filespec *alloc_filespec(const char *path)
1172 int namelen = strlen(path);
1173 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1175 memset(spec, 0, sizeof(*spec));
1176 spec->path = (char *)(spec + 1);
1177 memcpy(spec->path, path, namelen+1);
1178 return spec;
1181 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1182 unsigned short mode)
1184 if (mode) {
1185 spec->mode = canon_mode(mode);
1186 hashcpy(spec->sha1, sha1);
1187 spec->sha1_valid = !is_null_sha1(sha1);
1192 * Given a name and sha1 pair, if the dircache tells us the file in
1193 * the work tree has that object contents, return true, so that
1194 * prepare_temp_file() does not have to inflate and extract.
1196 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1198 struct cache_entry *ce;
1199 struct stat st;
1200 int pos, len;
1202 /* We do not read the cache ourselves here, because the
1203 * benchmark with my previous version that always reads cache
1204 * shows that it makes things worse for diff-tree comparing
1205 * two linux-2.6 kernel trees in an already checked out work
1206 * tree. This is because most diff-tree comparisons deal with
1207 * only a small number of files, while reading the cache is
1208 * expensive for a large project, and its cost outweighs the
1209 * savings we get by not inflating the object to a temporary
1210 * file. Practically, this code only helps when we are used
1211 * by diff-cache --cached, which does read the cache before
1212 * calling us.
1214 if (!active_cache)
1215 return 0;
1217 /* We want to avoid the working directory if our caller
1218 * doesn't need the data in a normal file, this system
1219 * is rather slow with its stat/open/mmap/close syscalls,
1220 * and the object is contained in a pack file. The pack
1221 * is probably already open and will be faster to obtain
1222 * the data through than the working directory. Loose
1223 * objects however would tend to be slower as they need
1224 * to be individually opened and inflated.
1226 if (FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1227 return 0;
1229 len = strlen(name);
1230 pos = cache_name_pos(name, len);
1231 if (pos < 0)
1232 return 0;
1233 ce = active_cache[pos];
1234 if ((lstat(name, &st) < 0) ||
1235 !S_ISREG(st.st_mode) || /* careful! */
1236 ce_match_stat(ce, &st, 0) ||
1237 hashcmp(sha1, ce->sha1))
1238 return 0;
1239 /* we return 1 only when we can stat, it is a regular file,
1240 * stat information matches, and sha1 recorded in the cache
1241 * matches. I.e. we know the file in the work tree really is
1242 * the same as the <name, sha1> pair.
1244 return 1;
1247 static struct sha1_size_cache {
1248 unsigned char sha1[20];
1249 unsigned long size;
1250 } **sha1_size_cache;
1251 static int sha1_size_cache_nr, sha1_size_cache_alloc;
1253 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
1254 int find_only,
1255 unsigned long size)
1257 int first, last;
1258 struct sha1_size_cache *e;
1260 first = 0;
1261 last = sha1_size_cache_nr;
1262 while (last > first) {
1263 int cmp, next = (last + first) >> 1;
1264 e = sha1_size_cache[next];
1265 cmp = hashcmp(e->sha1, sha1);
1266 if (!cmp)
1267 return e;
1268 if (cmp < 0) {
1269 last = next;
1270 continue;
1272 first = next+1;
1274 /* not found */
1275 if (find_only)
1276 return NULL;
1277 /* insert to make it at "first" */
1278 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1279 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1280 sha1_size_cache = xrealloc(sha1_size_cache,
1281 sha1_size_cache_alloc *
1282 sizeof(*sha1_size_cache));
1284 sha1_size_cache_nr++;
1285 if (first < sha1_size_cache_nr)
1286 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1287 (sha1_size_cache_nr - first - 1) *
1288 sizeof(*sha1_size_cache));
1289 e = xmalloc(sizeof(struct sha1_size_cache));
1290 sha1_size_cache[first] = e;
1291 hashcpy(e->sha1, sha1);
1292 e->size = size;
1293 return e;
1297 * While doing rename detection and pickaxe operation, we may need to
1298 * grab the data for the blob (or file) for our own in-core comparison.
1299 * diff_filespec has data and size fields for this purpose.
1301 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1303 int err = 0;
1304 if (!DIFF_FILE_VALID(s))
1305 die("internal error: asking to populate invalid file.");
1306 if (S_ISDIR(s->mode))
1307 return -1;
1309 if (!use_size_cache)
1310 size_only = 0;
1312 if (s->data)
1313 return err;
1314 if (!s->sha1_valid ||
1315 reuse_worktree_file(s->path, s->sha1, 0)) {
1316 struct stat st;
1317 int fd;
1318 if (lstat(s->path, &st) < 0) {
1319 if (errno == ENOENT) {
1320 err_empty:
1321 err = -1;
1322 empty:
1323 s->data = (char *)"";
1324 s->size = 0;
1325 return err;
1328 s->size = st.st_size;
1329 if (!s->size)
1330 goto empty;
1331 if (size_only)
1332 return 0;
1333 if (S_ISLNK(st.st_mode)) {
1334 int ret;
1335 s->data = xmalloc(s->size);
1336 s->should_free = 1;
1337 ret = readlink(s->path, s->data, s->size);
1338 if (ret < 0) {
1339 free(s->data);
1340 goto err_empty;
1342 return 0;
1344 fd = open(s->path, O_RDONLY);
1345 if (fd < 0)
1346 goto err_empty;
1347 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1348 close(fd);
1349 if (s->data == MAP_FAILED)
1350 goto err_empty;
1351 s->should_munmap = 1;
1353 else {
1354 char type[20];
1355 struct sha1_size_cache *e;
1357 if (size_only) {
1358 e = locate_size_cache(s->sha1, 1, 0);
1359 if (e) {
1360 s->size = e->size;
1361 return 0;
1363 if (!sha1_object_info(s->sha1, type, &s->size))
1364 locate_size_cache(s->sha1, 0, s->size);
1366 else {
1367 s->data = read_sha1_file(s->sha1, type, &s->size);
1368 s->should_free = 1;
1371 return 0;
1374 void diff_free_filespec_data(struct diff_filespec *s)
1376 if (s->should_free)
1377 free(s->data);
1378 else if (s->should_munmap)
1379 munmap(s->data, s->size);
1380 s->should_free = s->should_munmap = 0;
1381 s->data = NULL;
1382 free(s->cnt_data);
1383 s->cnt_data = NULL;
1386 static void prep_temp_blob(struct diff_tempfile *temp,
1387 void *blob,
1388 unsigned long size,
1389 const unsigned char *sha1,
1390 int mode)
1392 int fd;
1394 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1395 if (fd < 0)
1396 die("unable to create temp-file");
1397 if (write(fd, blob, size) != size)
1398 die("unable to write temp-file");
1399 close(fd);
1400 temp->name = temp->tmp_path;
1401 strcpy(temp->hex, sha1_to_hex(sha1));
1402 temp->hex[40] = 0;
1403 sprintf(temp->mode, "%06o", mode);
1406 static void prepare_temp_file(const char *name,
1407 struct diff_tempfile *temp,
1408 struct diff_filespec *one)
1410 if (!DIFF_FILE_VALID(one)) {
1411 not_a_valid_file:
1412 /* A '-' entry produces this for file-2, and
1413 * a '+' entry produces this for file-1.
1415 temp->name = "/dev/null";
1416 strcpy(temp->hex, ".");
1417 strcpy(temp->mode, ".");
1418 return;
1421 if (!one->sha1_valid ||
1422 reuse_worktree_file(name, one->sha1, 1)) {
1423 struct stat st;
1424 if (lstat(name, &st) < 0) {
1425 if (errno == ENOENT)
1426 goto not_a_valid_file;
1427 die("stat(%s): %s", name, strerror(errno));
1429 if (S_ISLNK(st.st_mode)) {
1430 int ret;
1431 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1432 if (sizeof(buf) <= st.st_size)
1433 die("symlink too long: %s", name);
1434 ret = readlink(name, buf, st.st_size);
1435 if (ret < 0)
1436 die("readlink(%s)", name);
1437 prep_temp_blob(temp, buf, st.st_size,
1438 (one->sha1_valid ?
1439 one->sha1 : null_sha1),
1440 (one->sha1_valid ?
1441 one->mode : S_IFLNK));
1443 else {
1444 /* we can borrow from the file in the work tree */
1445 temp->name = name;
1446 if (!one->sha1_valid)
1447 strcpy(temp->hex, sha1_to_hex(null_sha1));
1448 else
1449 strcpy(temp->hex, sha1_to_hex(one->sha1));
1450 /* Even though we may sometimes borrow the
1451 * contents from the work tree, we always want
1452 * one->mode. mode is trustworthy even when
1453 * !(one->sha1_valid), as long as
1454 * DIFF_FILE_VALID(one).
1456 sprintf(temp->mode, "%06o", one->mode);
1458 return;
1460 else {
1461 if (diff_populate_filespec(one, 0))
1462 die("cannot read data blob for %s", one->path);
1463 prep_temp_blob(temp, one->data, one->size,
1464 one->sha1, one->mode);
1468 static void remove_tempfile(void)
1470 int i;
1472 for (i = 0; i < 2; i++)
1473 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1474 unlink(diff_temp[i].name);
1475 diff_temp[i].name = NULL;
1479 static void remove_tempfile_on_signal(int signo)
1481 remove_tempfile();
1482 signal(SIGINT, SIG_DFL);
1483 raise(signo);
1486 static int spawn_prog(const char *pgm, const char **arg)
1488 pid_t pid;
1489 int status;
1491 fflush(NULL);
1492 pid = fork();
1493 if (pid < 0)
1494 die("unable to fork");
1495 if (!pid) {
1496 execvp(pgm, (char *const*) arg);
1497 exit(255);
1500 while (waitpid(pid, &status, 0) < 0) {
1501 if (errno == EINTR)
1502 continue;
1503 return -1;
1506 /* Earlier we did not check the exit status because
1507 * diff exits non-zero if files are different, and
1508 * we are not interested in knowing that. It was a
1509 * mistake which made it harder to quit a diff-*
1510 * session that uses the git-apply-patch-script as
1511 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1512 * should also exit non-zero only when it wants to
1513 * abort the entire diff-* session.
1515 if (WIFEXITED(status) && !WEXITSTATUS(status))
1516 return 0;
1517 return -1;
1520 /* An external diff command takes:
1522 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1523 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1526 static void run_external_diff(const char *pgm,
1527 const char *name,
1528 const char *other,
1529 struct diff_filespec *one,
1530 struct diff_filespec *two,
1531 const char *xfrm_msg,
1532 int complete_rewrite)
1534 const char *spawn_arg[10];
1535 struct diff_tempfile *temp = diff_temp;
1536 int retval;
1537 static int atexit_asked = 0;
1538 const char *othername;
1539 const char **arg = &spawn_arg[0];
1541 othername = (other? other : name);
1542 if (one && two) {
1543 prepare_temp_file(name, &temp[0], one);
1544 prepare_temp_file(othername, &temp[1], two);
1545 if (! atexit_asked &&
1546 (temp[0].name == temp[0].tmp_path ||
1547 temp[1].name == temp[1].tmp_path)) {
1548 atexit_asked = 1;
1549 atexit(remove_tempfile);
1551 signal(SIGINT, remove_tempfile_on_signal);
1554 if (one && two) {
1555 *arg++ = pgm;
1556 *arg++ = name;
1557 *arg++ = temp[0].name;
1558 *arg++ = temp[0].hex;
1559 *arg++ = temp[0].mode;
1560 *arg++ = temp[1].name;
1561 *arg++ = temp[1].hex;
1562 *arg++ = temp[1].mode;
1563 if (other) {
1564 *arg++ = other;
1565 *arg++ = xfrm_msg;
1567 } else {
1568 *arg++ = pgm;
1569 *arg++ = name;
1571 *arg = NULL;
1572 retval = spawn_prog(pgm, spawn_arg);
1573 remove_tempfile();
1574 if (retval) {
1575 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1576 exit(1);
1580 static void run_diff_cmd(const char *pgm,
1581 const char *name,
1582 const char *other,
1583 struct diff_filespec *one,
1584 struct diff_filespec *two,
1585 const char *xfrm_msg,
1586 struct diff_options *o,
1587 int complete_rewrite)
1589 if (pgm) {
1590 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1591 complete_rewrite);
1592 return;
1594 if (one && two)
1595 builtin_diff(name, other ? other : name,
1596 one, two, xfrm_msg, o, complete_rewrite);
1597 else
1598 printf("* Unmerged path %s\n", name);
1601 static void diff_fill_sha1_info(struct diff_filespec *one)
1603 if (DIFF_FILE_VALID(one)) {
1604 if (!one->sha1_valid) {
1605 struct stat st;
1606 if (lstat(one->path, &st) < 0)
1607 die("stat %s", one->path);
1608 if (index_path(one->sha1, one->path, &st, 0))
1609 die("cannot hash %s\n", one->path);
1612 else
1613 hashclr(one->sha1);
1616 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1618 const char *pgm = external_diff();
1619 char msg[PATH_MAX*2+300], *xfrm_msg;
1620 struct diff_filespec *one;
1621 struct diff_filespec *two;
1622 const char *name;
1623 const char *other;
1624 char *name_munged, *other_munged;
1625 int complete_rewrite = 0;
1626 int len;
1628 if (DIFF_PAIR_UNMERGED(p)) {
1629 /* unmerged */
1630 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1631 return;
1634 name = p->one->path;
1635 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1636 name_munged = quote_one(name);
1637 other_munged = quote_one(other);
1638 one = p->one; two = p->two;
1640 diff_fill_sha1_info(one);
1641 diff_fill_sha1_info(two);
1643 len = 0;
1644 switch (p->status) {
1645 case DIFF_STATUS_COPIED:
1646 len += snprintf(msg + len, sizeof(msg) - len,
1647 "similarity index %d%%\n"
1648 "copy from %s\n"
1649 "copy to %s\n",
1650 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1651 name_munged, other_munged);
1652 break;
1653 case DIFF_STATUS_RENAMED:
1654 len += snprintf(msg + len, sizeof(msg) - len,
1655 "similarity index %d%%\n"
1656 "rename from %s\n"
1657 "rename to %s\n",
1658 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1659 name_munged, other_munged);
1660 break;
1661 case DIFF_STATUS_MODIFIED:
1662 if (p->score) {
1663 len += snprintf(msg + len, sizeof(msg) - len,
1664 "dissimilarity index %d%%\n",
1665 (int)(0.5 + p->score *
1666 100.0/MAX_SCORE));
1667 complete_rewrite = 1;
1668 break;
1670 /* fallthru */
1671 default:
1672 /* nothing */
1676 if (hashcmp(one->sha1, two->sha1)) {
1677 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1679 if (o->binary) {
1680 mmfile_t mf;
1681 if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
1682 (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
1683 abbrev = 40;
1685 len += snprintf(msg + len, sizeof(msg) - len,
1686 "index %.*s..%.*s",
1687 abbrev, sha1_to_hex(one->sha1),
1688 abbrev, sha1_to_hex(two->sha1));
1689 if (one->mode == two->mode)
1690 len += snprintf(msg + len, sizeof(msg) - len,
1691 " %06o", one->mode);
1692 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1695 if (len)
1696 msg[--len] = 0;
1697 xfrm_msg = len ? msg : NULL;
1699 if (!pgm &&
1700 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1701 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1702 /* a filepair that changes between file and symlink
1703 * needs to be split into deletion and creation.
1705 struct diff_filespec *null = alloc_filespec(two->path);
1706 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1707 free(null);
1708 null = alloc_filespec(one->path);
1709 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1710 free(null);
1712 else
1713 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1714 complete_rewrite);
1716 free(name_munged);
1717 free(other_munged);
1720 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1721 struct diffstat_t *diffstat)
1723 const char *name;
1724 const char *other;
1725 int complete_rewrite = 0;
1727 if (DIFF_PAIR_UNMERGED(p)) {
1728 /* unmerged */
1729 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1730 return;
1733 name = p->one->path;
1734 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1736 diff_fill_sha1_info(p->one);
1737 diff_fill_sha1_info(p->two);
1739 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1740 complete_rewrite = 1;
1741 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1744 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1746 const char *name;
1747 const char *other;
1749 if (DIFF_PAIR_UNMERGED(p)) {
1750 /* unmerged */
1751 return;
1754 name = p->one->path;
1755 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1757 diff_fill_sha1_info(p->one);
1758 diff_fill_sha1_info(p->two);
1760 builtin_checkdiff(name, other, p->one, p->two);
1763 void diff_setup(struct diff_options *options)
1765 memset(options, 0, sizeof(*options));
1766 options->line_termination = '\n';
1767 options->break_opt = -1;
1768 options->rename_limit = -1;
1769 options->context = 3;
1770 options->msg_sep = "";
1772 options->change = diff_change;
1773 options->add_remove = diff_addremove;
1774 options->color_diff = diff_use_color_default;
1775 options->detect_rename = diff_detect_rename_default;
1778 int diff_setup_done(struct diff_options *options)
1780 int count = 0;
1782 if (options->output_format & DIFF_FORMAT_NAME)
1783 count++;
1784 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
1785 count++;
1786 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
1787 count++;
1788 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
1789 count++;
1790 if (count > 1)
1791 die("--name-only, --name-status, --check and -s are mutually exclusive");
1793 if (options->find_copies_harder)
1794 options->detect_rename = DIFF_DETECT_COPY;
1796 if (options->output_format & (DIFF_FORMAT_NAME |
1797 DIFF_FORMAT_NAME_STATUS |
1798 DIFF_FORMAT_CHECKDIFF |
1799 DIFF_FORMAT_NO_OUTPUT))
1800 options->output_format &= ~(DIFF_FORMAT_RAW |
1801 DIFF_FORMAT_NUMSTAT |
1802 DIFF_FORMAT_DIFFSTAT |
1803 DIFF_FORMAT_SHORTSTAT |
1804 DIFF_FORMAT_SUMMARY |
1805 DIFF_FORMAT_PATCH);
1808 * These cases always need recursive; we do not drop caller-supplied
1809 * recursive bits for other formats here.
1811 if (options->output_format & (DIFF_FORMAT_PATCH |
1812 DIFF_FORMAT_NUMSTAT |
1813 DIFF_FORMAT_DIFFSTAT |
1814 DIFF_FORMAT_SHORTSTAT |
1815 DIFF_FORMAT_SUMMARY |
1816 DIFF_FORMAT_CHECKDIFF))
1817 options->recursive = 1;
1819 * Also pickaxe would not work very well if you do not say recursive
1821 if (options->pickaxe)
1822 options->recursive = 1;
1824 if (options->detect_rename && options->rename_limit < 0)
1825 options->rename_limit = diff_rename_limit_default;
1826 if (options->setup & DIFF_SETUP_USE_CACHE) {
1827 if (!active_cache)
1828 /* read-cache does not die even when it fails
1829 * so it is safe for us to do this here. Also
1830 * it does not smudge active_cache or active_nr
1831 * when it fails, so we do not have to worry about
1832 * cleaning it up ourselves either.
1834 read_cache();
1836 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1837 use_size_cache = 1;
1838 if (options->abbrev <= 0 || 40 < options->abbrev)
1839 options->abbrev = 40; /* full */
1841 return 0;
1844 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1846 char c, *eq;
1847 int len;
1849 if (*arg != '-')
1850 return 0;
1851 c = *++arg;
1852 if (!c)
1853 return 0;
1854 if (c == arg_short) {
1855 c = *++arg;
1856 if (!c)
1857 return 1;
1858 if (val && isdigit(c)) {
1859 char *end;
1860 int n = strtoul(arg, &end, 10);
1861 if (*end)
1862 return 0;
1863 *val = n;
1864 return 1;
1866 return 0;
1868 if (c != '-')
1869 return 0;
1870 arg++;
1871 eq = strchr(arg, '=');
1872 if (eq)
1873 len = eq - arg;
1874 else
1875 len = strlen(arg);
1876 if (!len || strncmp(arg, arg_long, len))
1877 return 0;
1878 if (eq) {
1879 int n;
1880 char *end;
1881 if (!isdigit(*++eq))
1882 return 0;
1883 n = strtoul(eq, &end, 10);
1884 if (*end)
1885 return 0;
1886 *val = n;
1888 return 1;
1891 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1893 const char *arg = av[0];
1894 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1895 options->output_format |= DIFF_FORMAT_PATCH;
1896 else if (opt_arg(arg, 'U', "unified", &options->context))
1897 options->output_format |= DIFF_FORMAT_PATCH;
1898 else if (!strcmp(arg, "--raw"))
1899 options->output_format |= DIFF_FORMAT_RAW;
1900 else if (!strcmp(arg, "--patch-with-raw")) {
1901 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1903 else if (!strcmp(arg, "--numstat")) {
1904 options->output_format |= DIFF_FORMAT_NUMSTAT;
1906 else if (!strcmp(arg, "--shortstat")) {
1907 options->output_format |= DIFF_FORMAT_SHORTSTAT;
1909 else if (!strncmp(arg, "--stat", 6)) {
1910 char *end;
1911 int width = options->stat_width;
1912 int name_width = options->stat_name_width;
1913 arg += 6;
1914 end = (char *)arg;
1916 switch (*arg) {
1917 case '-':
1918 if (!strncmp(arg, "-width=", 7))
1919 width = strtoul(arg + 7, &end, 10);
1920 else if (!strncmp(arg, "-name-width=", 12))
1921 name_width = strtoul(arg + 12, &end, 10);
1922 break;
1923 case '=':
1924 width = strtoul(arg+1, &end, 10);
1925 if (*end == ',')
1926 name_width = strtoul(end+1, &end, 10);
1929 /* Important! This checks all the error cases! */
1930 if (*end)
1931 return 0;
1932 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1933 options->stat_name_width = name_width;
1934 options->stat_width = width;
1936 else if (!strcmp(arg, "--check"))
1937 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1938 else if (!strcmp(arg, "--summary"))
1939 options->output_format |= DIFF_FORMAT_SUMMARY;
1940 else if (!strcmp(arg, "--patch-with-stat")) {
1941 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1943 else if (!strcmp(arg, "-z"))
1944 options->line_termination = 0;
1945 else if (!strncmp(arg, "-l", 2))
1946 options->rename_limit = strtoul(arg+2, NULL, 10);
1947 else if (!strcmp(arg, "--full-index"))
1948 options->full_index = 1;
1949 else if (!strcmp(arg, "--binary")) {
1950 options->output_format |= DIFF_FORMAT_PATCH;
1951 options->binary = 1;
1953 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
1954 options->text = 1;
1956 else if (!strcmp(arg, "--name-only"))
1957 options->output_format |= DIFF_FORMAT_NAME;
1958 else if (!strcmp(arg, "--name-status"))
1959 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1960 else if (!strcmp(arg, "-R"))
1961 options->reverse_diff = 1;
1962 else if (!strncmp(arg, "-S", 2))
1963 options->pickaxe = arg + 2;
1964 else if (!strcmp(arg, "-s")) {
1965 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1967 else if (!strncmp(arg, "-O", 2))
1968 options->orderfile = arg + 2;
1969 else if (!strncmp(arg, "--diff-filter=", 14))
1970 options->filter = arg + 14;
1971 else if (!strcmp(arg, "--pickaxe-all"))
1972 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1973 else if (!strcmp(arg, "--pickaxe-regex"))
1974 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1975 else if (!strncmp(arg, "-B", 2)) {
1976 if ((options->break_opt =
1977 diff_scoreopt_parse(arg)) == -1)
1978 return -1;
1980 else if (!strncmp(arg, "-M", 2)) {
1981 if ((options->rename_score =
1982 diff_scoreopt_parse(arg)) == -1)
1983 return -1;
1984 options->detect_rename = DIFF_DETECT_RENAME;
1986 else if (!strncmp(arg, "-C", 2)) {
1987 if ((options->rename_score =
1988 diff_scoreopt_parse(arg)) == -1)
1989 return -1;
1990 options->detect_rename = DIFF_DETECT_COPY;
1992 else if (!strcmp(arg, "--find-copies-harder"))
1993 options->find_copies_harder = 1;
1994 else if (!strcmp(arg, "--abbrev"))
1995 options->abbrev = DEFAULT_ABBREV;
1996 else if (!strncmp(arg, "--abbrev=", 9)) {
1997 options->abbrev = strtoul(arg + 9, NULL, 10);
1998 if (options->abbrev < MINIMUM_ABBREV)
1999 options->abbrev = MINIMUM_ABBREV;
2000 else if (40 < options->abbrev)
2001 options->abbrev = 40;
2003 else if (!strcmp(arg, "--color"))
2004 options->color_diff = 1;
2005 else if (!strcmp(arg, "--no-color"))
2006 options->color_diff = 0;
2007 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2008 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2009 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2010 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2011 else if (!strcmp(arg, "--color-words"))
2012 options->color_diff = options->color_diff_words = 1;
2013 else if (!strcmp(arg, "--no-renames"))
2014 options->detect_rename = 0;
2015 else
2016 return 0;
2017 return 1;
2020 static int parse_num(const char **cp_p)
2022 unsigned long num, scale;
2023 int ch, dot;
2024 const char *cp = *cp_p;
2026 num = 0;
2027 scale = 1;
2028 dot = 0;
2029 for(;;) {
2030 ch = *cp;
2031 if ( !dot && ch == '.' ) {
2032 scale = 1;
2033 dot = 1;
2034 } else if ( ch == '%' ) {
2035 scale = dot ? scale*100 : 100;
2036 cp++; /* % is always at the end */
2037 break;
2038 } else if ( ch >= '0' && ch <= '9' ) {
2039 if ( scale < 100000 ) {
2040 scale *= 10;
2041 num = (num*10) + (ch-'0');
2043 } else {
2044 break;
2046 cp++;
2048 *cp_p = cp;
2050 /* user says num divided by scale and we say internally that
2051 * is MAX_SCORE * num / scale.
2053 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
2056 int diff_scoreopt_parse(const char *opt)
2058 int opt1, opt2, cmd;
2060 if (*opt++ != '-')
2061 return -1;
2062 cmd = *opt++;
2063 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2064 return -1; /* that is not a -M, -C nor -B option */
2066 opt1 = parse_num(&opt);
2067 if (cmd != 'B')
2068 opt2 = 0;
2069 else {
2070 if (*opt == 0)
2071 opt2 = 0;
2072 else if (*opt != '/')
2073 return -1; /* we expect -B80/99 or -B80 */
2074 else {
2075 opt++;
2076 opt2 = parse_num(&opt);
2079 if (*opt != 0)
2080 return -1;
2081 return opt1 | (opt2 << 16);
2084 struct diff_queue_struct diff_queued_diff;
2086 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2088 if (queue->alloc <= queue->nr) {
2089 queue->alloc = alloc_nr(queue->alloc);
2090 queue->queue = xrealloc(queue->queue,
2091 sizeof(dp) * queue->alloc);
2093 queue->queue[queue->nr++] = dp;
2096 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2097 struct diff_filespec *one,
2098 struct diff_filespec *two)
2100 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2101 dp->one = one;
2102 dp->two = two;
2103 if (queue)
2104 diff_q(queue, dp);
2105 return dp;
2108 void diff_free_filepair(struct diff_filepair *p)
2110 diff_free_filespec_data(p->one);
2111 diff_free_filespec_data(p->two);
2112 free(p->one);
2113 free(p->two);
2114 free(p);
2117 /* This is different from find_unique_abbrev() in that
2118 * it stuffs the result with dots for alignment.
2120 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2122 int abblen;
2123 const char *abbrev;
2124 if (len == 40)
2125 return sha1_to_hex(sha1);
2127 abbrev = find_unique_abbrev(sha1, len);
2128 if (!abbrev)
2129 return sha1_to_hex(sha1);
2130 abblen = strlen(abbrev);
2131 if (abblen < 37) {
2132 static char hex[41];
2133 if (len < abblen && abblen <= len + 2)
2134 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2135 else
2136 sprintf(hex, "%s...", abbrev);
2137 return hex;
2139 return sha1_to_hex(sha1);
2142 static void diff_flush_raw(struct diff_filepair *p,
2143 struct diff_options *options)
2145 int two_paths;
2146 char status[10];
2147 int abbrev = options->abbrev;
2148 const char *path_one, *path_two;
2149 int inter_name_termination = '\t';
2150 int line_termination = options->line_termination;
2152 if (!line_termination)
2153 inter_name_termination = 0;
2155 path_one = p->one->path;
2156 path_two = p->two->path;
2157 if (line_termination) {
2158 path_one = quote_one(path_one);
2159 path_two = quote_one(path_two);
2162 if (p->score)
2163 sprintf(status, "%c%03d", p->status,
2164 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2165 else {
2166 status[0] = p->status;
2167 status[1] = 0;
2169 switch (p->status) {
2170 case DIFF_STATUS_COPIED:
2171 case DIFF_STATUS_RENAMED:
2172 two_paths = 1;
2173 break;
2174 case DIFF_STATUS_ADDED:
2175 case DIFF_STATUS_DELETED:
2176 two_paths = 0;
2177 break;
2178 default:
2179 two_paths = 0;
2180 break;
2182 if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2183 printf(":%06o %06o %s ",
2184 p->one->mode, p->two->mode,
2185 diff_unique_abbrev(p->one->sha1, abbrev));
2186 printf("%s ",
2187 diff_unique_abbrev(p->two->sha1, abbrev));
2189 printf("%s%c%s", status, inter_name_termination, path_one);
2190 if (two_paths)
2191 printf("%c%s", inter_name_termination, path_two);
2192 putchar(line_termination);
2193 if (path_one != p->one->path)
2194 free((void*)path_one);
2195 if (path_two != p->two->path)
2196 free((void*)path_two);
2199 static void diff_flush_name(struct diff_filepair *p, int line_termination)
2201 char *path = p->two->path;
2203 if (line_termination)
2204 path = quote_one(p->two->path);
2205 printf("%s%c", path, line_termination);
2206 if (p->two->path != path)
2207 free(path);
2210 int diff_unmodified_pair(struct diff_filepair *p)
2212 /* This function is written stricter than necessary to support
2213 * the currently implemented transformers, but the idea is to
2214 * let transformers to produce diff_filepairs any way they want,
2215 * and filter and clean them up here before producing the output.
2217 struct diff_filespec *one, *two;
2219 if (DIFF_PAIR_UNMERGED(p))
2220 return 0; /* unmerged is interesting */
2222 one = p->one;
2223 two = p->two;
2225 /* deletion, addition, mode or type change
2226 * and rename are all interesting.
2228 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2229 DIFF_PAIR_MODE_CHANGED(p) ||
2230 strcmp(one->path, two->path))
2231 return 0;
2233 /* both are valid and point at the same path. that is, we are
2234 * dealing with a change.
2236 if (one->sha1_valid && two->sha1_valid &&
2237 !hashcmp(one->sha1, two->sha1))
2238 return 1; /* no change */
2239 if (!one->sha1_valid && !two->sha1_valid)
2240 return 1; /* both look at the same file on the filesystem. */
2241 return 0;
2244 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2246 if (diff_unmodified_pair(p))
2247 return;
2249 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2250 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2251 return; /* no tree diffs in patch format */
2253 run_diff(p, o);
2256 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2257 struct diffstat_t *diffstat)
2259 if (diff_unmodified_pair(p))
2260 return;
2262 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2263 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2264 return; /* no tree diffs in patch format */
2266 run_diffstat(p, o, diffstat);
2269 static void diff_flush_checkdiff(struct diff_filepair *p,
2270 struct diff_options *o)
2272 if (diff_unmodified_pair(p))
2273 return;
2275 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2276 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2277 return; /* no tree diffs in patch format */
2279 run_checkdiff(p, o);
2282 int diff_queue_is_empty(void)
2284 struct diff_queue_struct *q = &diff_queued_diff;
2285 int i;
2286 for (i = 0; i < q->nr; i++)
2287 if (!diff_unmodified_pair(q->queue[i]))
2288 return 0;
2289 return 1;
2292 #if DIFF_DEBUG
2293 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2295 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2296 x, one ? one : "",
2297 s->path,
2298 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2299 s->mode,
2300 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2301 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2302 x, one ? one : "",
2303 s->size, s->xfrm_flags);
2306 void diff_debug_filepair(const struct diff_filepair *p, int i)
2308 diff_debug_filespec(p->one, i, "one");
2309 diff_debug_filespec(p->two, i, "two");
2310 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2311 p->score, p->status ? p->status : '?',
2312 p->source_stays, p->broken_pair);
2315 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2317 int i;
2318 if (msg)
2319 fprintf(stderr, "%s\n", msg);
2320 fprintf(stderr, "q->nr = %d\n", q->nr);
2321 for (i = 0; i < q->nr; i++) {
2322 struct diff_filepair *p = q->queue[i];
2323 diff_debug_filepair(p, i);
2326 #endif
2328 static void diff_resolve_rename_copy(void)
2330 int i, j;
2331 struct diff_filepair *p, *pp;
2332 struct diff_queue_struct *q = &diff_queued_diff;
2334 diff_debug_queue("resolve-rename-copy", q);
2336 for (i = 0; i < q->nr; i++) {
2337 p = q->queue[i];
2338 p->status = 0; /* undecided */
2339 if (DIFF_PAIR_UNMERGED(p))
2340 p->status = DIFF_STATUS_UNMERGED;
2341 else if (!DIFF_FILE_VALID(p->one))
2342 p->status = DIFF_STATUS_ADDED;
2343 else if (!DIFF_FILE_VALID(p->two))
2344 p->status = DIFF_STATUS_DELETED;
2345 else if (DIFF_PAIR_TYPE_CHANGED(p))
2346 p->status = DIFF_STATUS_TYPE_CHANGED;
2348 /* from this point on, we are dealing with a pair
2349 * whose both sides are valid and of the same type, i.e.
2350 * either in-place edit or rename/copy edit.
2352 else if (DIFF_PAIR_RENAME(p)) {
2353 if (p->source_stays) {
2354 p->status = DIFF_STATUS_COPIED;
2355 continue;
2357 /* See if there is some other filepair that
2358 * copies from the same source as us. If so
2359 * we are a copy. Otherwise we are either a
2360 * copy if the path stays, or a rename if it
2361 * does not, but we already handled "stays" case.
2363 for (j = i + 1; j < q->nr; j++) {
2364 pp = q->queue[j];
2365 if (strcmp(pp->one->path, p->one->path))
2366 continue; /* not us */
2367 if (!DIFF_PAIR_RENAME(pp))
2368 continue; /* not a rename/copy */
2369 /* pp is a rename/copy from the same source */
2370 p->status = DIFF_STATUS_COPIED;
2371 break;
2373 if (!p->status)
2374 p->status = DIFF_STATUS_RENAMED;
2376 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2377 p->one->mode != p->two->mode)
2378 p->status = DIFF_STATUS_MODIFIED;
2379 else {
2380 /* This is a "no-change" entry and should not
2381 * happen anymore, but prepare for broken callers.
2383 error("feeding unmodified %s to diffcore",
2384 p->one->path);
2385 p->status = DIFF_STATUS_UNKNOWN;
2388 diff_debug_queue("resolve-rename-copy done", q);
2391 static int check_pair_status(struct diff_filepair *p)
2393 switch (p->status) {
2394 case DIFF_STATUS_UNKNOWN:
2395 return 0;
2396 case 0:
2397 die("internal error in diff-resolve-rename-copy");
2398 default:
2399 return 1;
2403 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2405 int fmt = opt->output_format;
2407 if (fmt & DIFF_FORMAT_CHECKDIFF)
2408 diff_flush_checkdiff(p, opt);
2409 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2410 diff_flush_raw(p, opt);
2411 else if (fmt & DIFF_FORMAT_NAME)
2412 diff_flush_name(p, opt->line_termination);
2415 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2417 if (fs->mode)
2418 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2419 else
2420 printf(" %s %s\n", newdelete, fs->path);
2424 static void show_mode_change(struct diff_filepair *p, int show_name)
2426 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2427 if (show_name)
2428 printf(" mode change %06o => %06o %s\n",
2429 p->one->mode, p->two->mode, p->two->path);
2430 else
2431 printf(" mode change %06o => %06o\n",
2432 p->one->mode, p->two->mode);
2436 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2438 const char *old, *new;
2440 /* Find common prefix */
2441 old = p->one->path;
2442 new = p->two->path;
2443 while (1) {
2444 const char *slash_old, *slash_new;
2445 slash_old = strchr(old, '/');
2446 slash_new = strchr(new, '/');
2447 if (!slash_old ||
2448 !slash_new ||
2449 slash_old - old != slash_new - new ||
2450 memcmp(old, new, slash_new - new))
2451 break;
2452 old = slash_old + 1;
2453 new = slash_new + 1;
2455 /* p->one->path thru old is the common prefix, and old and new
2456 * through the end of names are renames
2458 if (old != p->one->path)
2459 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2460 (int)(old - p->one->path), p->one->path,
2461 old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2462 else
2463 printf(" %s %s => %s (%d%%)\n", renamecopy,
2464 p->one->path, p->two->path,
2465 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2466 show_mode_change(p, 0);
2469 static void diff_summary(struct diff_filepair *p)
2471 switch(p->status) {
2472 case DIFF_STATUS_DELETED:
2473 show_file_mode_name("delete", p->one);
2474 break;
2475 case DIFF_STATUS_ADDED:
2476 show_file_mode_name("create", p->two);
2477 break;
2478 case DIFF_STATUS_COPIED:
2479 show_rename_copy("copy", p);
2480 break;
2481 case DIFF_STATUS_RENAMED:
2482 show_rename_copy("rename", p);
2483 break;
2484 default:
2485 if (p->score) {
2486 printf(" rewrite %s (%d%%)\n", p->two->path,
2487 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2488 show_mode_change(p, 0);
2489 } else show_mode_change(p, 1);
2490 break;
2494 struct patch_id_t {
2495 struct xdiff_emit_state xm;
2496 SHA_CTX *ctx;
2497 int patchlen;
2500 static int remove_space(char *line, int len)
2502 int i;
2503 char *dst = line;
2504 unsigned char c;
2506 for (i = 0; i < len; i++)
2507 if (!isspace((c = line[i])))
2508 *dst++ = c;
2510 return dst - line;
2513 static void patch_id_consume(void *priv, char *line, unsigned long len)
2515 struct patch_id_t *data = priv;
2516 int new_len;
2518 /* Ignore line numbers when computing the SHA1 of the patch */
2519 if (!strncmp(line, "@@ -", 4))
2520 return;
2522 new_len = remove_space(line, len);
2524 SHA1_Update(data->ctx, line, new_len);
2525 data->patchlen += new_len;
2528 /* returns 0 upon success, and writes result into sha1 */
2529 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2531 struct diff_queue_struct *q = &diff_queued_diff;
2532 int i;
2533 SHA_CTX ctx;
2534 struct patch_id_t data;
2535 char buffer[PATH_MAX * 4 + 20];
2537 SHA1_Init(&ctx);
2538 memset(&data, 0, sizeof(struct patch_id_t));
2539 data.ctx = &ctx;
2540 data.xm.consume = patch_id_consume;
2542 for (i = 0; i < q->nr; i++) {
2543 xpparam_t xpp;
2544 xdemitconf_t xecfg;
2545 xdemitcb_t ecb;
2546 mmfile_t mf1, mf2;
2547 struct diff_filepair *p = q->queue[i];
2548 int len1, len2;
2550 if (p->status == 0)
2551 return error("internal diff status error");
2552 if (p->status == DIFF_STATUS_UNKNOWN)
2553 continue;
2554 if (diff_unmodified_pair(p))
2555 continue;
2556 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2557 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2558 continue;
2559 if (DIFF_PAIR_UNMERGED(p))
2560 continue;
2562 diff_fill_sha1_info(p->one);
2563 diff_fill_sha1_info(p->two);
2564 if (fill_mmfile(&mf1, p->one) < 0 ||
2565 fill_mmfile(&mf2, p->two) < 0)
2566 return error("unable to read files to diff");
2568 /* Maybe hash p->two? into the patch id? */
2569 if (mmfile_is_binary(&mf2))
2570 continue;
2572 len1 = remove_space(p->one->path, strlen(p->one->path));
2573 len2 = remove_space(p->two->path, strlen(p->two->path));
2574 if (p->one->mode == 0)
2575 len1 = snprintf(buffer, sizeof(buffer),
2576 "diff--gita/%.*sb/%.*s"
2577 "newfilemode%06o"
2578 "---/dev/null"
2579 "+++b/%.*s",
2580 len1, p->one->path,
2581 len2, p->two->path,
2582 p->two->mode,
2583 len2, p->two->path);
2584 else if (p->two->mode == 0)
2585 len1 = snprintf(buffer, sizeof(buffer),
2586 "diff--gita/%.*sb/%.*s"
2587 "deletedfilemode%06o"
2588 "---a/%.*s"
2589 "+++/dev/null",
2590 len1, p->one->path,
2591 len2, p->two->path,
2592 p->one->mode,
2593 len1, p->one->path);
2594 else
2595 len1 = snprintf(buffer, sizeof(buffer),
2596 "diff--gita/%.*sb/%.*s"
2597 "---a/%.*s"
2598 "+++b/%.*s",
2599 len1, p->one->path,
2600 len2, p->two->path,
2601 len1, p->one->path,
2602 len2, p->two->path);
2603 SHA1_Update(&ctx, buffer, len1);
2605 xpp.flags = XDF_NEED_MINIMAL;
2606 xecfg.ctxlen = 3;
2607 xecfg.flags = XDL_EMIT_FUNCNAMES;
2608 ecb.outf = xdiff_outf;
2609 ecb.priv = &data;
2610 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2613 SHA1_Final(sha1, &ctx);
2614 return 0;
2617 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2619 struct diff_queue_struct *q = &diff_queued_diff;
2620 int i;
2621 int result = diff_get_patch_id(options, sha1);
2623 for (i = 0; i < q->nr; i++)
2624 diff_free_filepair(q->queue[i]);
2626 free(q->queue);
2627 q->queue = NULL;
2628 q->nr = q->alloc = 0;
2630 return result;
2633 static int is_summary_empty(const struct diff_queue_struct *q)
2635 int i;
2637 for (i = 0; i < q->nr; i++) {
2638 const struct diff_filepair *p = q->queue[i];
2640 switch (p->status) {
2641 case DIFF_STATUS_DELETED:
2642 case DIFF_STATUS_ADDED:
2643 case DIFF_STATUS_COPIED:
2644 case DIFF_STATUS_RENAMED:
2645 return 0;
2646 default:
2647 if (p->score)
2648 return 0;
2649 if (p->one->mode && p->two->mode &&
2650 p->one->mode != p->two->mode)
2651 return 0;
2652 break;
2655 return 1;
2658 void diff_flush(struct diff_options *options)
2660 struct diff_queue_struct *q = &diff_queued_diff;
2661 int i, output_format = options->output_format;
2662 int separator = 0;
2665 * Order: raw, stat, summary, patch
2666 * or: name/name-status/checkdiff (other bits clear)
2668 if (!q->nr)
2669 goto free_queue;
2671 if (output_format & (DIFF_FORMAT_RAW |
2672 DIFF_FORMAT_NAME |
2673 DIFF_FORMAT_NAME_STATUS |
2674 DIFF_FORMAT_CHECKDIFF)) {
2675 for (i = 0; i < q->nr; i++) {
2676 struct diff_filepair *p = q->queue[i];
2677 if (check_pair_status(p))
2678 flush_one_pair(p, options);
2680 separator++;
2683 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2684 struct diffstat_t diffstat;
2686 memset(&diffstat, 0, sizeof(struct diffstat_t));
2687 diffstat.xm.consume = diffstat_consume;
2688 for (i = 0; i < q->nr; i++) {
2689 struct diff_filepair *p = q->queue[i];
2690 if (check_pair_status(p))
2691 diff_flush_stat(p, options, &diffstat);
2693 if (output_format & DIFF_FORMAT_NUMSTAT)
2694 show_numstat(&diffstat, options);
2695 if (output_format & DIFF_FORMAT_DIFFSTAT)
2696 show_stats(&diffstat, options);
2697 else if (output_format & DIFF_FORMAT_SHORTSTAT)
2698 show_shortstats(&diffstat);
2699 separator++;
2702 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2703 for (i = 0; i < q->nr; i++)
2704 diff_summary(q->queue[i]);
2705 separator++;
2708 if (output_format & DIFF_FORMAT_PATCH) {
2709 if (separator) {
2710 if (options->stat_sep) {
2711 /* attach patch instead of inline */
2712 fputs(options->stat_sep, stdout);
2713 } else {
2714 putchar(options->line_termination);
2718 for (i = 0; i < q->nr; i++) {
2719 struct diff_filepair *p = q->queue[i];
2720 if (check_pair_status(p))
2721 diff_flush_patch(p, options);
2725 if (output_format & DIFF_FORMAT_CALLBACK)
2726 options->format_callback(q, options, options->format_callback_data);
2728 for (i = 0; i < q->nr; i++)
2729 diff_free_filepair(q->queue[i]);
2730 free_queue:
2731 free(q->queue);
2732 q->queue = NULL;
2733 q->nr = q->alloc = 0;
2736 static void diffcore_apply_filter(const char *filter)
2738 int i;
2739 struct diff_queue_struct *q = &diff_queued_diff;
2740 struct diff_queue_struct outq;
2741 outq.queue = NULL;
2742 outq.nr = outq.alloc = 0;
2744 if (!filter)
2745 return;
2747 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2748 int found;
2749 for (i = found = 0; !found && i < q->nr; i++) {
2750 struct diff_filepair *p = q->queue[i];
2751 if (((p->status == DIFF_STATUS_MODIFIED) &&
2752 ((p->score &&
2753 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2754 (!p->score &&
2755 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2756 ((p->status != DIFF_STATUS_MODIFIED) &&
2757 strchr(filter, p->status)))
2758 found++;
2760 if (found)
2761 return;
2763 /* otherwise we will clear the whole queue
2764 * by copying the empty outq at the end of this
2765 * function, but first clear the current entries
2766 * in the queue.
2768 for (i = 0; i < q->nr; i++)
2769 diff_free_filepair(q->queue[i]);
2771 else {
2772 /* Only the matching ones */
2773 for (i = 0; i < q->nr; i++) {
2774 struct diff_filepair *p = q->queue[i];
2776 if (((p->status == DIFF_STATUS_MODIFIED) &&
2777 ((p->score &&
2778 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2779 (!p->score &&
2780 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2781 ((p->status != DIFF_STATUS_MODIFIED) &&
2782 strchr(filter, p->status)))
2783 diff_q(&outq, p);
2784 else
2785 diff_free_filepair(p);
2788 free(q->queue);
2789 *q = outq;
2792 void diffcore_std(struct diff_options *options)
2794 if (options->break_opt != -1)
2795 diffcore_break(options->break_opt);
2796 if (options->detect_rename)
2797 diffcore_rename(options);
2798 if (options->break_opt != -1)
2799 diffcore_merge_broken();
2800 if (options->pickaxe)
2801 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2802 if (options->orderfile)
2803 diffcore_order(options->orderfile);
2804 diff_resolve_rename_copy();
2805 diffcore_apply_filter(options->filter);
2809 void diffcore_std_no_resolve(struct diff_options *options)
2811 if (options->pickaxe)
2812 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2813 if (options->orderfile)
2814 diffcore_order(options->orderfile);
2815 diffcore_apply_filter(options->filter);
2818 void diff_addremove(struct diff_options *options,
2819 int addremove, unsigned mode,
2820 const unsigned char *sha1,
2821 const char *base, const char *path)
2823 char concatpath[PATH_MAX];
2824 struct diff_filespec *one, *two;
2826 /* This may look odd, but it is a preparation for
2827 * feeding "there are unchanged files which should
2828 * not produce diffs, but when you are doing copy
2829 * detection you would need them, so here they are"
2830 * entries to the diff-core. They will be prefixed
2831 * with something like '=' or '*' (I haven't decided
2832 * which but should not make any difference).
2833 * Feeding the same new and old to diff_change()
2834 * also has the same effect.
2835 * Before the final output happens, they are pruned after
2836 * merged into rename/copy pairs as appropriate.
2838 if (options->reverse_diff)
2839 addremove = (addremove == '+' ? '-' :
2840 addremove == '-' ? '+' : addremove);
2842 if (!path) path = "";
2843 sprintf(concatpath, "%s%s", base, path);
2844 one = alloc_filespec(concatpath);
2845 two = alloc_filespec(concatpath);
2847 if (addremove != '+')
2848 fill_filespec(one, sha1, mode);
2849 if (addremove != '-')
2850 fill_filespec(two, sha1, mode);
2852 diff_queue(&diff_queued_diff, one, two);
2855 void diff_change(struct diff_options *options,
2856 unsigned old_mode, unsigned new_mode,
2857 const unsigned char *old_sha1,
2858 const unsigned char *new_sha1,
2859 const char *base, const char *path)
2861 char concatpath[PATH_MAX];
2862 struct diff_filespec *one, *two;
2864 if (options->reverse_diff) {
2865 unsigned tmp;
2866 const unsigned char *tmp_c;
2867 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2868 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2870 if (!path) path = "";
2871 sprintf(concatpath, "%s%s", base, path);
2872 one = alloc_filespec(concatpath);
2873 two = alloc_filespec(concatpath);
2874 fill_filespec(one, old_sha1, old_mode);
2875 fill_filespec(two, new_sha1, new_mode);
2877 diff_queue(&diff_queued_diff, one, two);
2880 void diff_unmerge(struct diff_options *options,
2881 const char *path)
2883 struct diff_filespec *one, *two;
2884 one = alloc_filespec(path);
2885 two = alloc_filespec(path);
2886 diff_queue(&diff_queued_diff, one, two);