user-manual: fix clone and fetch typos
[git/vmiklos.git] / diff.c
blobb28933f870d7dd11596fd3715280e1488ac2430f
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 (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
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,
188 const char *set, const char *reset)
190 int ch, nl_just_seen = 1;
191 while (0 < size--) {
192 ch = *data++;
193 if (nl_just_seen) {
194 fputs(set, stdout);
195 putchar(prefix);
197 if (ch == '\n') {
198 nl_just_seen = 1;
199 fputs(reset, stdout);
200 } else
201 nl_just_seen = 0;
202 putchar(ch);
204 if (!nl_just_seen)
205 printf("%s\n\\ No newline at end of file\n", reset);
208 static void emit_rewrite_diff(const char *name_a,
209 const char *name_b,
210 struct diff_filespec *one,
211 struct diff_filespec *two,
212 int color_diff)
214 int lc_a, lc_b;
215 const char *name_a_tab, *name_b_tab;
216 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
217 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
218 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
219 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
220 const char *reset = diff_get_color(color_diff, DIFF_RESET);
222 name_a += (*name_a == '/');
223 name_b += (*name_b == '/');
224 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
225 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
227 diff_populate_filespec(one, 0);
228 diff_populate_filespec(two, 0);
229 lc_a = count_lines(one->data, one->size);
230 lc_b = count_lines(two->data, two->size);
231 printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
232 metainfo, name_a, name_a_tab, reset,
233 metainfo, name_b, name_b_tab, reset, fraginfo);
234 print_line_count(lc_a);
235 printf(" +");
236 print_line_count(lc_b);
237 printf(" @@%s\n", reset);
238 if (lc_a)
239 copy_file('-', one->data, one->size, old, reset);
240 if (lc_b)
241 copy_file('+', two->data, two->size, new, reset);
244 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
246 if (!DIFF_FILE_VALID(one)) {
247 mf->ptr = (char *)""; /* does not matter */
248 mf->size = 0;
249 return 0;
251 else if (diff_populate_filespec(one, 0))
252 return -1;
253 mf->ptr = one->data;
254 mf->size = one->size;
255 return 0;
258 struct diff_words_buffer {
259 mmfile_t text;
260 long alloc;
261 long current; /* output pointer */
262 int suppressed_newline;
265 static void diff_words_append(char *line, unsigned long len,
266 struct diff_words_buffer *buffer)
268 if (buffer->text.size + len > buffer->alloc) {
269 buffer->alloc = (buffer->text.size + len) * 3 / 2;
270 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
272 line++;
273 len--;
274 memcpy(buffer->text.ptr + buffer->text.size, line, len);
275 buffer->text.size += len;
278 struct diff_words_data {
279 struct xdiff_emit_state xm;
280 struct diff_words_buffer minus, plus;
283 static void print_word(struct diff_words_buffer *buffer, int len, int color,
284 int suppress_newline)
286 const char *ptr;
287 int eol = 0;
289 if (len == 0)
290 return;
292 ptr = buffer->text.ptr + buffer->current;
293 buffer->current += len;
295 if (ptr[len - 1] == '\n') {
296 eol = 1;
297 len--;
300 fputs(diff_get_color(1, color), stdout);
301 fwrite(ptr, len, 1, stdout);
302 fputs(diff_get_color(1, DIFF_RESET), stdout);
304 if (eol) {
305 if (suppress_newline)
306 buffer->suppressed_newline = 1;
307 else
308 putchar('\n');
312 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
314 struct diff_words_data *diff_words = priv;
316 if (diff_words->minus.suppressed_newline) {
317 if (line[0] != '+')
318 putchar('\n');
319 diff_words->minus.suppressed_newline = 0;
322 len--;
323 switch (line[0]) {
324 case '-':
325 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
326 break;
327 case '+':
328 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
329 break;
330 case ' ':
331 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
332 diff_words->minus.current += len;
333 break;
337 /* this executes the word diff on the accumulated buffers */
338 static void diff_words_show(struct diff_words_data *diff_words)
340 xpparam_t xpp;
341 xdemitconf_t xecfg;
342 xdemitcb_t ecb;
343 mmfile_t minus, plus;
344 int i;
346 minus.size = diff_words->minus.text.size;
347 minus.ptr = xmalloc(minus.size);
348 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
349 for (i = 0; i < minus.size; i++)
350 if (isspace(minus.ptr[i]))
351 minus.ptr[i] = '\n';
352 diff_words->minus.current = 0;
354 plus.size = diff_words->plus.text.size;
355 plus.ptr = xmalloc(plus.size);
356 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
357 for (i = 0; i < plus.size; i++)
358 if (isspace(plus.ptr[i]))
359 plus.ptr[i] = '\n';
360 diff_words->plus.current = 0;
362 xpp.flags = XDF_NEED_MINIMAL;
363 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
364 xecfg.flags = 0;
365 ecb.outf = xdiff_outf;
366 ecb.priv = diff_words;
367 diff_words->xm.consume = fn_out_diff_words_aux;
368 xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
370 free(minus.ptr);
371 free(plus.ptr);
372 diff_words->minus.text.size = diff_words->plus.text.size = 0;
374 if (diff_words->minus.suppressed_newline) {
375 putchar('\n');
376 diff_words->minus.suppressed_newline = 0;
380 struct emit_callback {
381 struct xdiff_emit_state xm;
382 int nparents, color_diff;
383 const char **label_path;
384 struct diff_words_data *diff_words;
385 int *found_changesp;
388 static void free_diff_words_data(struct emit_callback *ecbdata)
390 if (ecbdata->diff_words) {
391 /* flush buffers */
392 if (ecbdata->diff_words->minus.text.size ||
393 ecbdata->diff_words->plus.text.size)
394 diff_words_show(ecbdata->diff_words);
396 if (ecbdata->diff_words->minus.text.ptr)
397 free (ecbdata->diff_words->minus.text.ptr);
398 if (ecbdata->diff_words->plus.text.ptr)
399 free (ecbdata->diff_words->plus.text.ptr);
400 free(ecbdata->diff_words);
401 ecbdata->diff_words = NULL;
405 const char *diff_get_color(int diff_use_color, enum color_diff ix)
407 if (diff_use_color)
408 return diff_colors[ix];
409 return "";
412 static void emit_line(const char *set, const char *reset, const char *line, int len)
414 if (len > 0 && line[len-1] == '\n')
415 len--;
416 fputs(set, stdout);
417 fwrite(line, len, 1, stdout);
418 puts(reset);
421 static void emit_line_with_ws(int nparents,
422 const char *set, const char *reset, const char *ws,
423 const char *line, int len)
425 int col0 = nparents;
426 int last_tab_in_indent = -1;
427 int last_space_in_indent = -1;
428 int i;
429 int tail = len;
430 int need_highlight_leading_space = 0;
431 /* The line is a newly added line. Does it have funny leading
432 * whitespaces? In indent, SP should never precede a TAB.
434 for (i = col0; i < len; i++) {
435 if (line[i] == '\t') {
436 last_tab_in_indent = i;
437 if (0 <= last_space_in_indent)
438 need_highlight_leading_space = 1;
440 else if (line[i] == ' ')
441 last_space_in_indent = i;
442 else
443 break;
445 fputs(set, stdout);
446 fwrite(line, col0, 1, stdout);
447 fputs(reset, stdout);
448 if (((i == len) || line[i] == '\n') && i != col0) {
449 /* The whole line was indent */
450 emit_line(ws, reset, line + col0, len - col0);
451 return;
453 i = col0;
454 if (need_highlight_leading_space) {
455 while (i < last_tab_in_indent) {
456 if (line[i] == ' ') {
457 fputs(ws, stdout);
458 putchar(' ');
459 fputs(reset, stdout);
461 else
462 putchar(line[i]);
463 i++;
466 tail = len - 1;
467 if (line[tail] == '\n' && i < tail)
468 tail--;
469 while (i < tail) {
470 if (!isspace(line[tail]))
471 break;
472 tail--;
474 if ((i < tail && line[tail + 1] != '\n')) {
475 /* This has whitespace between tail+1..len */
476 fputs(set, stdout);
477 fwrite(line + i, tail - i + 1, 1, stdout);
478 fputs(reset, stdout);
479 emit_line(ws, reset, line + tail + 1, len - tail - 1);
481 else
482 emit_line(set, reset, line + i, len - i);
485 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
487 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
488 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
490 if (!*ws)
491 emit_line(set, reset, line, len);
492 else
493 emit_line_with_ws(ecbdata->nparents, set, reset, ws,
494 line, len);
497 static void fn_out_consume(void *priv, char *line, unsigned long len)
499 int i;
500 int color;
501 struct emit_callback *ecbdata = priv;
502 const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
503 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
505 *(ecbdata->found_changesp) = 1;
507 if (ecbdata->label_path[0]) {
508 const char *name_a_tab, *name_b_tab;
510 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
511 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
513 printf("%s--- %s%s%s\n",
514 set, ecbdata->label_path[0], reset, name_a_tab);
515 printf("%s+++ %s%s%s\n",
516 set, ecbdata->label_path[1], reset, name_b_tab);
517 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
520 /* This is not really necessary for now because
521 * this codepath only deals with two-way diffs.
523 for (i = 0; i < len && line[i] == '@'; i++)
525 if (2 <= i && i < len && line[i] == ' ') {
526 ecbdata->nparents = i - 1;
527 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
528 reset, line, len);
529 return;
532 if (len < ecbdata->nparents) {
533 set = reset;
534 emit_line(reset, reset, line, len);
535 return;
538 color = DIFF_PLAIN;
539 if (ecbdata->diff_words && ecbdata->nparents != 1)
540 /* fall back to normal diff */
541 free_diff_words_data(ecbdata);
542 if (ecbdata->diff_words) {
543 if (line[0] == '-') {
544 diff_words_append(line, len,
545 &ecbdata->diff_words->minus);
546 return;
547 } else if (line[0] == '+') {
548 diff_words_append(line, len,
549 &ecbdata->diff_words->plus);
550 return;
552 if (ecbdata->diff_words->minus.text.size ||
553 ecbdata->diff_words->plus.text.size)
554 diff_words_show(ecbdata->diff_words);
555 line++;
556 len--;
557 emit_line(set, reset, line, len);
558 return;
560 for (i = 0; i < ecbdata->nparents && len; i++) {
561 if (line[i] == '-')
562 color = DIFF_FILE_OLD;
563 else if (line[i] == '+')
564 color = DIFF_FILE_NEW;
567 if (color != DIFF_FILE_NEW) {
568 emit_line(diff_get_color(ecbdata->color_diff, color),
569 reset, line, len);
570 return;
572 emit_add_line(reset, ecbdata, line, len);
575 static char *pprint_rename(const char *a, const char *b)
577 const char *old = a;
578 const char *new = b;
579 char *name = NULL;
580 int pfx_length, sfx_length;
581 int len_a = strlen(a);
582 int len_b = strlen(b);
583 int qlen_a = quote_c_style(a, NULL, NULL, 0);
584 int qlen_b = quote_c_style(b, NULL, NULL, 0);
586 if (qlen_a || qlen_b) {
587 if (qlen_a) len_a = qlen_a;
588 if (qlen_b) len_b = qlen_b;
589 name = xmalloc( len_a + len_b + 5 );
590 if (qlen_a)
591 quote_c_style(a, name, NULL, 0);
592 else
593 memcpy(name, a, len_a);
594 memcpy(name + len_a, " => ", 4);
595 if (qlen_b)
596 quote_c_style(b, name + len_a + 4, NULL, 0);
597 else
598 memcpy(name + len_a + 4, b, len_b + 1);
599 return name;
602 /* Find common prefix */
603 pfx_length = 0;
604 while (*old && *new && *old == *new) {
605 if (*old == '/')
606 pfx_length = old - a + 1;
607 old++;
608 new++;
611 /* Find common suffix */
612 old = a + len_a;
613 new = b + len_b;
614 sfx_length = 0;
615 while (a <= old && b <= new && *old == *new) {
616 if (*old == '/')
617 sfx_length = len_a - (old - a);
618 old--;
619 new--;
623 * pfx{mid-a => mid-b}sfx
624 * {pfx-a => pfx-b}sfx
625 * pfx{sfx-a => sfx-b}
626 * name-a => name-b
628 if (pfx_length + sfx_length) {
629 int a_midlen = len_a - pfx_length - sfx_length;
630 int b_midlen = len_b - pfx_length - sfx_length;
631 if (a_midlen < 0) a_midlen = 0;
632 if (b_midlen < 0) b_midlen = 0;
634 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
635 sprintf(name, "%.*s{%.*s => %.*s}%s",
636 pfx_length, a,
637 a_midlen, a + pfx_length,
638 b_midlen, b + pfx_length,
639 a + len_a - sfx_length);
641 else {
642 name = xmalloc(len_a + len_b + 5);
643 sprintf(name, "%s => %s", a, b);
645 return name;
648 struct diffstat_t {
649 struct xdiff_emit_state xm;
651 int nr;
652 int alloc;
653 struct diffstat_file {
654 char *name;
655 unsigned is_unmerged:1;
656 unsigned is_binary:1;
657 unsigned is_renamed:1;
658 unsigned int added, deleted;
659 } **files;
662 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
663 const char *name_a,
664 const char *name_b)
666 struct diffstat_file *x;
667 x = xcalloc(sizeof (*x), 1);
668 if (diffstat->nr == diffstat->alloc) {
669 diffstat->alloc = alloc_nr(diffstat->alloc);
670 diffstat->files = xrealloc(diffstat->files,
671 diffstat->alloc * sizeof(x));
673 diffstat->files[diffstat->nr++] = x;
674 if (name_b) {
675 x->name = pprint_rename(name_a, name_b);
676 x->is_renamed = 1;
678 else
679 x->name = xstrdup(name_a);
680 return x;
683 static void diffstat_consume(void *priv, char *line, unsigned long len)
685 struct diffstat_t *diffstat = priv;
686 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
688 if (line[0] == '+')
689 x->added++;
690 else if (line[0] == '-')
691 x->deleted++;
694 const char mime_boundary_leader[] = "------------";
696 static int scale_linear(int it, int width, int max_change)
699 * make sure that at least one '-' is printed if there were deletions,
700 * and likewise for '+'.
702 if (max_change < 2)
703 return it;
704 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
707 static void show_name(const char *prefix, const char *name, int len,
708 const char *reset, const char *set)
710 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
713 static void show_graph(char ch, int cnt, const char *set, const char *reset)
715 if (cnt <= 0)
716 return;
717 printf("%s", set);
718 while (cnt--)
719 putchar(ch);
720 printf("%s", reset);
723 static void show_stats(struct diffstat_t* data, struct diff_options *options)
725 int i, len, add, del, total, adds = 0, dels = 0;
726 int max_change = 0, max_len = 0;
727 int total_files = data->nr;
728 int width, name_width;
729 const char *reset, *set, *add_c, *del_c;
731 if (data->nr == 0)
732 return;
734 width = options->stat_width ? options->stat_width : 80;
735 name_width = options->stat_name_width ? options->stat_name_width : 50;
737 /* Sanity: give at least 5 columns to the graph,
738 * but leave at least 10 columns for the name.
740 if (width < name_width + 15) {
741 if (name_width <= 25)
742 width = name_width + 15;
743 else
744 name_width = width - 15;
747 /* Find the longest filename and max number of changes */
748 reset = diff_get_color(options->color_diff, DIFF_RESET);
749 set = diff_get_color(options->color_diff, DIFF_PLAIN);
750 add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
751 del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
753 for (i = 0; i < data->nr; i++) {
754 struct diffstat_file *file = data->files[i];
755 int change = file->added + file->deleted;
757 if (!file->is_renamed) { /* renames are already quoted by pprint_rename */
758 len = quote_c_style(file->name, NULL, NULL, 0);
759 if (len) {
760 char *qname = xmalloc(len + 1);
761 quote_c_style(file->name, qname, NULL, 0);
762 free(file->name);
763 file->name = qname;
767 len = strlen(file->name);
768 if (max_len < len)
769 max_len = len;
771 if (file->is_binary || file->is_unmerged)
772 continue;
773 if (max_change < change)
774 max_change = change;
777 /* Compute the width of the graph part;
778 * 10 is for one blank at the beginning of the line plus
779 * " | count " between the name and the graph.
781 * From here on, name_width is the width of the name area,
782 * and width is the width of the graph area.
784 name_width = (name_width < max_len) ? name_width : max_len;
785 if (width < (name_width + 10) + max_change)
786 width = width - (name_width + 10);
787 else
788 width = max_change;
790 for (i = 0; i < data->nr; i++) {
791 const char *prefix = "";
792 char *name = data->files[i]->name;
793 int added = data->files[i]->added;
794 int deleted = data->files[i]->deleted;
795 int name_len;
798 * "scale" the filename
800 len = name_width;
801 name_len = strlen(name);
802 if (name_width < name_len) {
803 char *slash;
804 prefix = "...";
805 len -= 3;
806 name += name_len - len;
807 slash = strchr(name, '/');
808 if (slash)
809 name = slash;
812 if (data->files[i]->is_binary) {
813 show_name(prefix, name, len, reset, set);
814 printf(" Bin\n");
815 goto free_diffstat_file;
817 else if (data->files[i]->is_unmerged) {
818 show_name(prefix, name, len, reset, set);
819 printf(" Unmerged\n");
820 goto free_diffstat_file;
822 else if (!data->files[i]->is_renamed &&
823 (added + deleted == 0)) {
824 total_files--;
825 goto free_diffstat_file;
829 * scale the add/delete
831 add = added;
832 del = deleted;
833 total = add + del;
834 adds += add;
835 dels += del;
837 if (width <= max_change) {
838 add = scale_linear(add, width, max_change);
839 del = scale_linear(del, width, max_change);
840 total = add + del;
842 show_name(prefix, name, len, reset, set);
843 printf("%5d ", added + deleted);
844 show_graph('+', add, add_c, reset);
845 show_graph('-', del, del_c, reset);
846 putchar('\n');
847 free_diffstat_file:
848 free(data->files[i]->name);
849 free(data->files[i]);
851 free(data->files);
852 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
853 set, total_files, adds, dels, reset);
856 static void show_shortstats(struct diffstat_t* data)
858 int i, adds = 0, dels = 0, total_files = data->nr;
860 if (data->nr == 0)
861 return;
863 for (i = 0; i < data->nr; i++) {
864 if (!data->files[i]->is_binary &&
865 !data->files[i]->is_unmerged) {
866 int added = data->files[i]->added;
867 int deleted= data->files[i]->deleted;
868 if (!data->files[i]->is_renamed &&
869 (added + deleted == 0)) {
870 total_files--;
871 } else {
872 adds += added;
873 dels += deleted;
876 free(data->files[i]->name);
877 free(data->files[i]);
879 free(data->files);
881 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
882 total_files, adds, dels);
885 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
887 int i;
889 for (i = 0; i < data->nr; i++) {
890 struct diffstat_file *file = data->files[i];
892 if (file->is_binary)
893 printf("-\t-\t");
894 else
895 printf("%d\t%d\t", file->added, file->deleted);
896 if (options->line_termination && !file->is_renamed &&
897 quote_c_style(file->name, NULL, NULL, 0))
898 quote_c_style(file->name, NULL, stdout, 0);
899 else
900 fputs(file->name, stdout);
901 putchar(options->line_termination);
905 struct checkdiff_t {
906 struct xdiff_emit_state xm;
907 const char *filename;
908 int lineno, color_diff;
911 static void checkdiff_consume(void *priv, char *line, unsigned long len)
913 struct checkdiff_t *data = priv;
914 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
915 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
916 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
918 if (line[0] == '+') {
919 int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
921 /* check space before tab */
922 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
923 if (line[i] == ' ')
924 spaces++;
925 if (line[i - 1] == '\t' && spaces)
926 space_before_tab = 1;
928 /* check white space at line end */
929 if (line[len - 1] == '\n')
930 len--;
931 if (isspace(line[len - 1]))
932 white_space_at_end = 1;
934 if (space_before_tab || white_space_at_end) {
935 printf("%s:%d: %s", data->filename, data->lineno, ws);
936 if (space_before_tab) {
937 printf("space before tab");
938 if (white_space_at_end)
939 putchar(',');
941 if (white_space_at_end)
942 printf("white space at end");
943 printf(":%s ", reset);
944 emit_line_with_ws(1, set, reset, ws, line, len);
947 data->lineno++;
948 } else if (line[0] == ' ')
949 data->lineno++;
950 else if (line[0] == '@') {
951 char *plus = strchr(line, '+');
952 if (plus)
953 data->lineno = strtol(plus, NULL, 10);
954 else
955 die("invalid diff");
959 static unsigned char *deflate_it(char *data,
960 unsigned long size,
961 unsigned long *result_size)
963 int bound;
964 unsigned char *deflated;
965 z_stream stream;
967 memset(&stream, 0, sizeof(stream));
968 deflateInit(&stream, zlib_compression_level);
969 bound = deflateBound(&stream, size);
970 deflated = xmalloc(bound);
971 stream.next_out = deflated;
972 stream.avail_out = bound;
974 stream.next_in = (unsigned char *)data;
975 stream.avail_in = size;
976 while (deflate(&stream, Z_FINISH) == Z_OK)
977 ; /* nothing */
978 deflateEnd(&stream);
979 *result_size = stream.total_out;
980 return deflated;
983 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
985 void *cp;
986 void *delta;
987 void *deflated;
988 void *data;
989 unsigned long orig_size;
990 unsigned long delta_size;
991 unsigned long deflate_size;
992 unsigned long data_size;
994 /* We could do deflated delta, or we could do just deflated two,
995 * whichever is smaller.
997 delta = NULL;
998 deflated = deflate_it(two->ptr, two->size, &deflate_size);
999 if (one->size && two->size) {
1000 delta = diff_delta(one->ptr, one->size,
1001 two->ptr, two->size,
1002 &delta_size, deflate_size);
1003 if (delta) {
1004 void *to_free = delta;
1005 orig_size = delta_size;
1006 delta = deflate_it(delta, delta_size, &delta_size);
1007 free(to_free);
1011 if (delta && delta_size < deflate_size) {
1012 printf("delta %lu\n", orig_size);
1013 free(deflated);
1014 data = delta;
1015 data_size = delta_size;
1017 else {
1018 printf("literal %lu\n", two->size);
1019 free(delta);
1020 data = deflated;
1021 data_size = deflate_size;
1024 /* emit data encoded in base85 */
1025 cp = data;
1026 while (data_size) {
1027 int bytes = (52 < data_size) ? 52 : data_size;
1028 char line[70];
1029 data_size -= bytes;
1030 if (bytes <= 26)
1031 line[0] = bytes + 'A' - 1;
1032 else
1033 line[0] = bytes - 26 + 'a' - 1;
1034 encode_85(line + 1, cp, bytes);
1035 cp = (char *) cp + bytes;
1036 puts(line);
1038 printf("\n");
1039 free(data);
1042 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1044 printf("GIT binary patch\n");
1045 emit_binary_diff_body(one, two);
1046 emit_binary_diff_body(two, one);
1049 #define FIRST_FEW_BYTES 8000
1050 static int mmfile_is_binary(mmfile_t *mf)
1052 long sz = mf->size;
1053 if (FIRST_FEW_BYTES < sz)
1054 sz = FIRST_FEW_BYTES;
1055 return !!memchr(mf->ptr, 0, sz);
1058 static void builtin_diff(const char *name_a,
1059 const char *name_b,
1060 struct diff_filespec *one,
1061 struct diff_filespec *two,
1062 const char *xfrm_msg,
1063 struct diff_options *o,
1064 int complete_rewrite)
1066 mmfile_t mf1, mf2;
1067 const char *lbl[2];
1068 char *a_one, *b_two;
1069 const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1070 const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1072 a_one = quote_two("a/", name_a + (*name_a == '/'));
1073 b_two = quote_two("b/", name_b + (*name_b == '/'));
1074 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1075 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1076 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1077 if (lbl[0][0] == '/') {
1078 /* /dev/null */
1079 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1080 if (xfrm_msg && xfrm_msg[0])
1081 printf("%s%s%s\n", set, xfrm_msg, reset);
1083 else if (lbl[1][0] == '/') {
1084 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1085 if (xfrm_msg && xfrm_msg[0])
1086 printf("%s%s%s\n", set, xfrm_msg, reset);
1088 else {
1089 if (one->mode != two->mode) {
1090 printf("%sold mode %06o%s\n", set, one->mode, reset);
1091 printf("%snew mode %06o%s\n", set, two->mode, reset);
1093 if (xfrm_msg && xfrm_msg[0])
1094 printf("%s%s%s\n", set, xfrm_msg, reset);
1096 * we do not run diff between different kind
1097 * of objects.
1099 if ((one->mode ^ two->mode) & S_IFMT)
1100 goto free_ab_and_return;
1101 if (complete_rewrite) {
1102 emit_rewrite_diff(name_a, name_b, one, two,
1103 o->color_diff);
1104 o->found_changes = 1;
1105 goto free_ab_and_return;
1109 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1110 die("unable to read files to diff");
1112 if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
1113 /* Quite common confusing case */
1114 if (mf1.size == mf2.size &&
1115 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1116 goto free_ab_and_return;
1117 if (o->binary)
1118 emit_binary_diff(&mf1, &mf2);
1119 else
1120 printf("Binary files %s and %s differ\n",
1121 lbl[0], lbl[1]);
1122 o->found_changes = 1;
1124 else {
1125 /* Crazy xdl interfaces.. */
1126 const char *diffopts = getenv("GIT_DIFF_OPTS");
1127 xpparam_t xpp;
1128 xdemitconf_t xecfg;
1129 xdemitcb_t ecb;
1130 struct emit_callback ecbdata;
1132 memset(&ecbdata, 0, sizeof(ecbdata));
1133 ecbdata.label_path = lbl;
1134 ecbdata.color_diff = o->color_diff;
1135 ecbdata.found_changesp = &o->found_changes;
1136 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1137 xecfg.ctxlen = o->context;
1138 xecfg.flags = XDL_EMIT_FUNCNAMES;
1139 if (!diffopts)
1141 else if (!prefixcmp(diffopts, "--unified="))
1142 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1143 else if (!prefixcmp(diffopts, "-u"))
1144 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1145 ecb.outf = xdiff_outf;
1146 ecb.priv = &ecbdata;
1147 ecbdata.xm.consume = fn_out_consume;
1148 if (o->color_diff_words)
1149 ecbdata.diff_words =
1150 xcalloc(1, sizeof(struct diff_words_data));
1151 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1152 if (o->color_diff_words)
1153 free_diff_words_data(&ecbdata);
1156 free_ab_and_return:
1157 free(a_one);
1158 free(b_two);
1159 return;
1162 static void builtin_diffstat(const char *name_a, const char *name_b,
1163 struct diff_filespec *one,
1164 struct diff_filespec *two,
1165 struct diffstat_t *diffstat,
1166 struct diff_options *o,
1167 int complete_rewrite)
1169 mmfile_t mf1, mf2;
1170 struct diffstat_file *data;
1172 data = diffstat_add(diffstat, name_a, name_b);
1174 if (!one || !two) {
1175 data->is_unmerged = 1;
1176 return;
1178 if (complete_rewrite) {
1179 diff_populate_filespec(one, 0);
1180 diff_populate_filespec(two, 0);
1181 data->deleted = count_lines(one->data, one->size);
1182 data->added = count_lines(two->data, two->size);
1183 return;
1185 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1186 die("unable to read files to diff");
1188 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
1189 data->is_binary = 1;
1190 else {
1191 /* Crazy xdl interfaces.. */
1192 xpparam_t xpp;
1193 xdemitconf_t xecfg;
1194 xdemitcb_t ecb;
1196 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1197 xecfg.ctxlen = 0;
1198 xecfg.flags = 0;
1199 ecb.outf = xdiff_outf;
1200 ecb.priv = diffstat;
1201 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1205 static void builtin_checkdiff(const char *name_a, const char *name_b,
1206 struct diff_filespec *one,
1207 struct diff_filespec *two, struct diff_options *o)
1209 mmfile_t mf1, mf2;
1210 struct checkdiff_t data;
1212 if (!two)
1213 return;
1215 memset(&data, 0, sizeof(data));
1216 data.xm.consume = checkdiff_consume;
1217 data.filename = name_b ? name_b : name_a;
1218 data.lineno = 0;
1219 data.color_diff = o->color_diff;
1221 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1222 die("unable to read files to diff");
1224 if (mmfile_is_binary(&mf2))
1225 return;
1226 else {
1227 /* Crazy xdl interfaces.. */
1228 xpparam_t xpp;
1229 xdemitconf_t xecfg;
1230 xdemitcb_t ecb;
1232 xpp.flags = XDF_NEED_MINIMAL;
1233 xecfg.ctxlen = 0;
1234 xecfg.flags = 0;
1235 ecb.outf = xdiff_outf;
1236 ecb.priv = &data;
1237 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1241 struct diff_filespec *alloc_filespec(const char *path)
1243 int namelen = strlen(path);
1244 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1246 memset(spec, 0, sizeof(*spec));
1247 spec->path = (char *)(spec + 1);
1248 memcpy(spec->path, path, namelen+1);
1249 return spec;
1252 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1253 unsigned short mode)
1255 if (mode) {
1256 spec->mode = canon_mode(mode);
1257 hashcpy(spec->sha1, sha1);
1258 spec->sha1_valid = !is_null_sha1(sha1);
1263 * Given a name and sha1 pair, if the dircache tells us the file in
1264 * the work tree has that object contents, return true, so that
1265 * prepare_temp_file() does not have to inflate and extract.
1267 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1269 struct cache_entry *ce;
1270 struct stat st;
1271 int pos, len;
1273 /* We do not read the cache ourselves here, because the
1274 * benchmark with my previous version that always reads cache
1275 * shows that it makes things worse for diff-tree comparing
1276 * two linux-2.6 kernel trees in an already checked out work
1277 * tree. This is because most diff-tree comparisons deal with
1278 * only a small number of files, while reading the cache is
1279 * expensive for a large project, and its cost outweighs the
1280 * savings we get by not inflating the object to a temporary
1281 * file. Practically, this code only helps when we are used
1282 * by diff-cache --cached, which does read the cache before
1283 * calling us.
1285 if (!active_cache)
1286 return 0;
1288 /* We want to avoid the working directory if our caller
1289 * doesn't need the data in a normal file, this system
1290 * is rather slow with its stat/open/mmap/close syscalls,
1291 * and the object is contained in a pack file. The pack
1292 * is probably already open and will be faster to obtain
1293 * the data through than the working directory. Loose
1294 * objects however would tend to be slower as they need
1295 * to be individually opened and inflated.
1297 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1298 return 0;
1300 len = strlen(name);
1301 pos = cache_name_pos(name, len);
1302 if (pos < 0)
1303 return 0;
1304 ce = active_cache[pos];
1305 if ((lstat(name, &st) < 0) ||
1306 !S_ISREG(st.st_mode) || /* careful! */
1307 ce_match_stat(ce, &st, 0) ||
1308 hashcmp(sha1, ce->sha1))
1309 return 0;
1310 /* we return 1 only when we can stat, it is a regular file,
1311 * stat information matches, and sha1 recorded in the cache
1312 * matches. I.e. we know the file in the work tree really is
1313 * the same as the <name, sha1> pair.
1315 return 1;
1318 static struct sha1_size_cache {
1319 unsigned char sha1[20];
1320 unsigned long size;
1321 } **sha1_size_cache;
1322 static int sha1_size_cache_nr, sha1_size_cache_alloc;
1324 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
1325 int find_only,
1326 unsigned long size)
1328 int first, last;
1329 struct sha1_size_cache *e;
1331 first = 0;
1332 last = sha1_size_cache_nr;
1333 while (last > first) {
1334 int cmp, next = (last + first) >> 1;
1335 e = sha1_size_cache[next];
1336 cmp = hashcmp(e->sha1, sha1);
1337 if (!cmp)
1338 return e;
1339 if (cmp < 0) {
1340 last = next;
1341 continue;
1343 first = next+1;
1345 /* not found */
1346 if (find_only)
1347 return NULL;
1348 /* insert to make it at "first" */
1349 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1350 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1351 sha1_size_cache = xrealloc(sha1_size_cache,
1352 sha1_size_cache_alloc *
1353 sizeof(*sha1_size_cache));
1355 sha1_size_cache_nr++;
1356 if (first < sha1_size_cache_nr)
1357 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1358 (sha1_size_cache_nr - first - 1) *
1359 sizeof(*sha1_size_cache));
1360 e = xmalloc(sizeof(struct sha1_size_cache));
1361 sha1_size_cache[first] = e;
1362 hashcpy(e->sha1, sha1);
1363 e->size = size;
1364 return e;
1367 static int populate_from_stdin(struct diff_filespec *s)
1369 #define INCREMENT 1024
1370 char *buf;
1371 unsigned long size;
1372 int got;
1374 size = 0;
1375 buf = NULL;
1376 while (1) {
1377 buf = xrealloc(buf, size + INCREMENT);
1378 got = xread(0, buf + size, INCREMENT);
1379 if (!got)
1380 break; /* EOF */
1381 if (got < 0)
1382 return error("error while reading from stdin %s",
1383 strerror(errno));
1384 size += got;
1386 s->should_munmap = 0;
1387 s->data = buf;
1388 s->size = size;
1389 s->should_free = 1;
1390 return 0;
1394 * While doing rename detection and pickaxe operation, we may need to
1395 * grab the data for the blob (or file) for our own in-core comparison.
1396 * diff_filespec has data and size fields for this purpose.
1398 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1400 int err = 0;
1401 if (!DIFF_FILE_VALID(s))
1402 die("internal error: asking to populate invalid file.");
1403 if (S_ISDIR(s->mode))
1404 return -1;
1406 if (!use_size_cache)
1407 size_only = 0;
1409 if (s->data)
1410 return err;
1411 if (!s->sha1_valid ||
1412 reuse_worktree_file(s->path, s->sha1, 0)) {
1413 struct stat st;
1414 int fd;
1415 char *buf;
1416 unsigned long size;
1418 if (!strcmp(s->path, "-"))
1419 return populate_from_stdin(s);
1421 if (lstat(s->path, &st) < 0) {
1422 if (errno == ENOENT) {
1423 err_empty:
1424 err = -1;
1425 empty:
1426 s->data = (char *)"";
1427 s->size = 0;
1428 return err;
1431 s->size = xsize_t(st.st_size);
1432 if (!s->size)
1433 goto empty;
1434 if (size_only)
1435 return 0;
1436 if (S_ISLNK(st.st_mode)) {
1437 int ret;
1438 s->data = xmalloc(s->size);
1439 s->should_free = 1;
1440 ret = readlink(s->path, s->data, s->size);
1441 if (ret < 0) {
1442 free(s->data);
1443 goto err_empty;
1445 return 0;
1447 fd = open(s->path, O_RDONLY);
1448 if (fd < 0)
1449 goto err_empty;
1450 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1451 close(fd);
1452 s->should_munmap = 1;
1455 * Convert from working tree format to canonical git format
1457 buf = s->data;
1458 size = s->size;
1459 if (convert_to_git(s->path, &buf, &size)) {
1460 munmap(s->data, s->size);
1461 s->should_munmap = 0;
1462 s->data = buf;
1463 s->size = size;
1464 s->should_free = 1;
1467 else {
1468 enum object_type type;
1469 struct sha1_size_cache *e;
1471 if (size_only && use_size_cache &&
1472 (e = locate_size_cache(s->sha1, 1, 0)) != NULL) {
1473 s->size = e->size;
1474 return 0;
1477 if (size_only) {
1478 type = sha1_object_info(s->sha1, &s->size);
1479 if (use_size_cache && 0 < type)
1480 locate_size_cache(s->sha1, 0, s->size);
1482 else {
1483 s->data = read_sha1_file(s->sha1, &type, &s->size);
1484 s->should_free = 1;
1487 return 0;
1490 void diff_free_filespec_data(struct diff_filespec *s)
1492 if (s->should_free)
1493 free(s->data);
1494 else if (s->should_munmap)
1495 munmap(s->data, s->size);
1496 s->should_free = s->should_munmap = 0;
1497 s->data = NULL;
1498 free(s->cnt_data);
1499 s->cnt_data = NULL;
1502 static void prep_temp_blob(struct diff_tempfile *temp,
1503 void *blob,
1504 unsigned long size,
1505 const unsigned char *sha1,
1506 int mode)
1508 int fd;
1510 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1511 if (fd < 0)
1512 die("unable to create temp-file");
1513 if (write_in_full(fd, blob, size) != size)
1514 die("unable to write temp-file");
1515 close(fd);
1516 temp->name = temp->tmp_path;
1517 strcpy(temp->hex, sha1_to_hex(sha1));
1518 temp->hex[40] = 0;
1519 sprintf(temp->mode, "%06o", mode);
1522 static void prepare_temp_file(const char *name,
1523 struct diff_tempfile *temp,
1524 struct diff_filespec *one)
1526 if (!DIFF_FILE_VALID(one)) {
1527 not_a_valid_file:
1528 /* A '-' entry produces this for file-2, and
1529 * a '+' entry produces this for file-1.
1531 temp->name = "/dev/null";
1532 strcpy(temp->hex, ".");
1533 strcpy(temp->mode, ".");
1534 return;
1537 if (!one->sha1_valid ||
1538 reuse_worktree_file(name, one->sha1, 1)) {
1539 struct stat st;
1540 if (lstat(name, &st) < 0) {
1541 if (errno == ENOENT)
1542 goto not_a_valid_file;
1543 die("stat(%s): %s", name, strerror(errno));
1545 if (S_ISLNK(st.st_mode)) {
1546 int ret;
1547 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1548 size_t sz = xsize_t(st.st_size);
1549 if (sizeof(buf) <= st.st_size)
1550 die("symlink too long: %s", name);
1551 ret = readlink(name, buf, sz);
1552 if (ret < 0)
1553 die("readlink(%s)", name);
1554 prep_temp_blob(temp, buf, sz,
1555 (one->sha1_valid ?
1556 one->sha1 : null_sha1),
1557 (one->sha1_valid ?
1558 one->mode : S_IFLNK));
1560 else {
1561 /* we can borrow from the file in the work tree */
1562 temp->name = name;
1563 if (!one->sha1_valid)
1564 strcpy(temp->hex, sha1_to_hex(null_sha1));
1565 else
1566 strcpy(temp->hex, sha1_to_hex(one->sha1));
1567 /* Even though we may sometimes borrow the
1568 * contents from the work tree, we always want
1569 * one->mode. mode is trustworthy even when
1570 * !(one->sha1_valid), as long as
1571 * DIFF_FILE_VALID(one).
1573 sprintf(temp->mode, "%06o", one->mode);
1575 return;
1577 else {
1578 if (diff_populate_filespec(one, 0))
1579 die("cannot read data blob for %s", one->path);
1580 prep_temp_blob(temp, one->data, one->size,
1581 one->sha1, one->mode);
1585 static void remove_tempfile(void)
1587 int i;
1589 for (i = 0; i < 2; i++)
1590 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1591 unlink(diff_temp[i].name);
1592 diff_temp[i].name = NULL;
1596 static void remove_tempfile_on_signal(int signo)
1598 remove_tempfile();
1599 signal(SIGINT, SIG_DFL);
1600 raise(signo);
1603 static int spawn_prog(const char *pgm, const char **arg)
1605 pid_t pid;
1606 int status;
1608 fflush(NULL);
1609 pid = fork();
1610 if (pid < 0)
1611 die("unable to fork");
1612 if (!pid) {
1613 execvp(pgm, (char *const*) arg);
1614 exit(255);
1617 while (waitpid(pid, &status, 0) < 0) {
1618 if (errno == EINTR)
1619 continue;
1620 return -1;
1623 /* Earlier we did not check the exit status because
1624 * diff exits non-zero if files are different, and
1625 * we are not interested in knowing that. It was a
1626 * mistake which made it harder to quit a diff-*
1627 * session that uses the git-apply-patch-script as
1628 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1629 * should also exit non-zero only when it wants to
1630 * abort the entire diff-* session.
1632 if (WIFEXITED(status) && !WEXITSTATUS(status))
1633 return 0;
1634 return -1;
1637 /* An external diff command takes:
1639 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1640 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1643 static void run_external_diff(const char *pgm,
1644 const char *name,
1645 const char *other,
1646 struct diff_filespec *one,
1647 struct diff_filespec *two,
1648 const char *xfrm_msg,
1649 int complete_rewrite)
1651 const char *spawn_arg[10];
1652 struct diff_tempfile *temp = diff_temp;
1653 int retval;
1654 static int atexit_asked = 0;
1655 const char *othername;
1656 const char **arg = &spawn_arg[0];
1658 othername = (other? other : name);
1659 if (one && two) {
1660 prepare_temp_file(name, &temp[0], one);
1661 prepare_temp_file(othername, &temp[1], two);
1662 if (! atexit_asked &&
1663 (temp[0].name == temp[0].tmp_path ||
1664 temp[1].name == temp[1].tmp_path)) {
1665 atexit_asked = 1;
1666 atexit(remove_tempfile);
1668 signal(SIGINT, remove_tempfile_on_signal);
1671 if (one && two) {
1672 *arg++ = pgm;
1673 *arg++ = name;
1674 *arg++ = temp[0].name;
1675 *arg++ = temp[0].hex;
1676 *arg++ = temp[0].mode;
1677 *arg++ = temp[1].name;
1678 *arg++ = temp[1].hex;
1679 *arg++ = temp[1].mode;
1680 if (other) {
1681 *arg++ = other;
1682 *arg++ = xfrm_msg;
1684 } else {
1685 *arg++ = pgm;
1686 *arg++ = name;
1688 *arg = NULL;
1689 retval = spawn_prog(pgm, spawn_arg);
1690 remove_tempfile();
1691 if (retval) {
1692 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1693 exit(1);
1697 static void run_diff_cmd(const char *pgm,
1698 const char *name,
1699 const char *other,
1700 struct diff_filespec *one,
1701 struct diff_filespec *two,
1702 const char *xfrm_msg,
1703 struct diff_options *o,
1704 int complete_rewrite)
1706 if (pgm) {
1707 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1708 complete_rewrite);
1709 return;
1711 if (one && two)
1712 builtin_diff(name, other ? other : name,
1713 one, two, xfrm_msg, o, complete_rewrite);
1714 else
1715 printf("* Unmerged path %s\n", name);
1718 static void diff_fill_sha1_info(struct diff_filespec *one)
1720 if (DIFF_FILE_VALID(one)) {
1721 if (!one->sha1_valid) {
1722 struct stat st;
1723 if (!strcmp(one->path, "-")) {
1724 hashcpy(one->sha1, null_sha1);
1725 return;
1727 if (lstat(one->path, &st) < 0)
1728 die("stat %s", one->path);
1729 if (index_path(one->sha1, one->path, &st, 0))
1730 die("cannot hash %s\n", one->path);
1733 else
1734 hashclr(one->sha1);
1737 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1739 const char *pgm = external_diff();
1740 char msg[PATH_MAX*2+300], *xfrm_msg;
1741 struct diff_filespec *one;
1742 struct diff_filespec *two;
1743 const char *name;
1744 const char *other;
1745 char *name_munged, *other_munged;
1746 int complete_rewrite = 0;
1747 int len;
1749 if (DIFF_PAIR_UNMERGED(p)) {
1750 /* unmerged */
1751 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1752 return;
1755 name = p->one->path;
1756 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1757 name_munged = quote_one(name);
1758 other_munged = quote_one(other);
1759 one = p->one; two = p->two;
1761 diff_fill_sha1_info(one);
1762 diff_fill_sha1_info(two);
1764 len = 0;
1765 switch (p->status) {
1766 case DIFF_STATUS_COPIED:
1767 len += snprintf(msg + len, sizeof(msg) - len,
1768 "similarity index %d%%\n"
1769 "copy from %s\n"
1770 "copy to %s\n",
1771 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1772 name_munged, other_munged);
1773 break;
1774 case DIFF_STATUS_RENAMED:
1775 len += snprintf(msg + len, sizeof(msg) - len,
1776 "similarity index %d%%\n"
1777 "rename from %s\n"
1778 "rename to %s\n",
1779 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1780 name_munged, other_munged);
1781 break;
1782 case DIFF_STATUS_MODIFIED:
1783 if (p->score) {
1784 len += snprintf(msg + len, sizeof(msg) - len,
1785 "dissimilarity index %d%%\n",
1786 (int)(0.5 + p->score *
1787 100.0/MAX_SCORE));
1788 complete_rewrite = 1;
1789 break;
1791 /* fallthru */
1792 default:
1793 /* nothing */
1797 if (hashcmp(one->sha1, two->sha1)) {
1798 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1800 if (o->binary) {
1801 mmfile_t mf;
1802 if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
1803 (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
1804 abbrev = 40;
1806 len += snprintf(msg + len, sizeof(msg) - len,
1807 "index %.*s..%.*s",
1808 abbrev, sha1_to_hex(one->sha1),
1809 abbrev, sha1_to_hex(two->sha1));
1810 if (one->mode == two->mode)
1811 len += snprintf(msg + len, sizeof(msg) - len,
1812 " %06o", one->mode);
1813 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1816 if (len)
1817 msg[--len] = 0;
1818 xfrm_msg = len ? msg : NULL;
1820 if (!pgm &&
1821 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1822 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1823 /* a filepair that changes between file and symlink
1824 * needs to be split into deletion and creation.
1826 struct diff_filespec *null = alloc_filespec(two->path);
1827 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1828 free(null);
1829 null = alloc_filespec(one->path);
1830 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1831 free(null);
1833 else
1834 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1835 complete_rewrite);
1837 free(name_munged);
1838 free(other_munged);
1841 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1842 struct diffstat_t *diffstat)
1844 const char *name;
1845 const char *other;
1846 int complete_rewrite = 0;
1848 if (DIFF_PAIR_UNMERGED(p)) {
1849 /* unmerged */
1850 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1851 return;
1854 name = p->one->path;
1855 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1857 diff_fill_sha1_info(p->one);
1858 diff_fill_sha1_info(p->two);
1860 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1861 complete_rewrite = 1;
1862 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1865 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1867 const char *name;
1868 const char *other;
1870 if (DIFF_PAIR_UNMERGED(p)) {
1871 /* unmerged */
1872 return;
1875 name = p->one->path;
1876 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1878 diff_fill_sha1_info(p->one);
1879 diff_fill_sha1_info(p->two);
1881 builtin_checkdiff(name, other, p->one, p->two, o);
1884 void diff_setup(struct diff_options *options)
1886 memset(options, 0, sizeof(*options));
1887 options->line_termination = '\n';
1888 options->break_opt = -1;
1889 options->rename_limit = -1;
1890 options->context = 3;
1891 options->msg_sep = "";
1893 options->change = diff_change;
1894 options->add_remove = diff_addremove;
1895 options->color_diff = diff_use_color_default;
1896 options->detect_rename = diff_detect_rename_default;
1899 int diff_setup_done(struct diff_options *options)
1901 int count = 0;
1903 if (options->output_format & DIFF_FORMAT_NAME)
1904 count++;
1905 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
1906 count++;
1907 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
1908 count++;
1909 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
1910 count++;
1911 if (count > 1)
1912 die("--name-only, --name-status, --check and -s are mutually exclusive");
1914 if (options->find_copies_harder)
1915 options->detect_rename = DIFF_DETECT_COPY;
1917 if (options->output_format & (DIFF_FORMAT_NAME |
1918 DIFF_FORMAT_NAME_STATUS |
1919 DIFF_FORMAT_CHECKDIFF |
1920 DIFF_FORMAT_NO_OUTPUT))
1921 options->output_format &= ~(DIFF_FORMAT_RAW |
1922 DIFF_FORMAT_NUMSTAT |
1923 DIFF_FORMAT_DIFFSTAT |
1924 DIFF_FORMAT_SHORTSTAT |
1925 DIFF_FORMAT_SUMMARY |
1926 DIFF_FORMAT_PATCH);
1929 * These cases always need recursive; we do not drop caller-supplied
1930 * recursive bits for other formats here.
1932 if (options->output_format & (DIFF_FORMAT_PATCH |
1933 DIFF_FORMAT_NUMSTAT |
1934 DIFF_FORMAT_DIFFSTAT |
1935 DIFF_FORMAT_SHORTSTAT |
1936 DIFF_FORMAT_SUMMARY |
1937 DIFF_FORMAT_CHECKDIFF))
1938 options->recursive = 1;
1940 * Also pickaxe would not work very well if you do not say recursive
1942 if (options->pickaxe)
1943 options->recursive = 1;
1945 if (options->detect_rename && options->rename_limit < 0)
1946 options->rename_limit = diff_rename_limit_default;
1947 if (options->setup & DIFF_SETUP_USE_CACHE) {
1948 if (!active_cache)
1949 /* read-cache does not die even when it fails
1950 * so it is safe for us to do this here. Also
1951 * it does not smudge active_cache or active_nr
1952 * when it fails, so we do not have to worry about
1953 * cleaning it up ourselves either.
1955 read_cache();
1957 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1958 use_size_cache = 1;
1959 if (options->abbrev <= 0 || 40 < options->abbrev)
1960 options->abbrev = 40; /* full */
1963 * It does not make sense to show the first hit we happened
1964 * to have found. It does not make sense not to return with
1965 * exit code in such a case either.
1967 if (options->quiet) {
1968 options->output_format = DIFF_FORMAT_NO_OUTPUT;
1969 options->exit_with_status = 1;
1973 * If we postprocess in diffcore, we cannot simply return
1974 * upon the first hit. We need to run diff as usual.
1976 if (options->pickaxe || options->filter)
1977 options->quiet = 0;
1979 return 0;
1982 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1984 char c, *eq;
1985 int len;
1987 if (*arg != '-')
1988 return 0;
1989 c = *++arg;
1990 if (!c)
1991 return 0;
1992 if (c == arg_short) {
1993 c = *++arg;
1994 if (!c)
1995 return 1;
1996 if (val && isdigit(c)) {
1997 char *end;
1998 int n = strtoul(arg, &end, 10);
1999 if (*end)
2000 return 0;
2001 *val = n;
2002 return 1;
2004 return 0;
2006 if (c != '-')
2007 return 0;
2008 arg++;
2009 eq = strchr(arg, '=');
2010 if (eq)
2011 len = eq - arg;
2012 else
2013 len = strlen(arg);
2014 if (!len || strncmp(arg, arg_long, len))
2015 return 0;
2016 if (eq) {
2017 int n;
2018 char *end;
2019 if (!isdigit(*++eq))
2020 return 0;
2021 n = strtoul(eq, &end, 10);
2022 if (*end)
2023 return 0;
2024 *val = n;
2026 return 1;
2029 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2031 const char *arg = av[0];
2032 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2033 options->output_format |= DIFF_FORMAT_PATCH;
2034 else if (opt_arg(arg, 'U', "unified", &options->context))
2035 options->output_format |= DIFF_FORMAT_PATCH;
2036 else if (!strcmp(arg, "--raw"))
2037 options->output_format |= DIFF_FORMAT_RAW;
2038 else if (!strcmp(arg, "--patch-with-raw")) {
2039 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2041 else if (!strcmp(arg, "--numstat")) {
2042 options->output_format |= DIFF_FORMAT_NUMSTAT;
2044 else if (!strcmp(arg, "--shortstat")) {
2045 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2047 else if (!prefixcmp(arg, "--stat")) {
2048 char *end;
2049 int width = options->stat_width;
2050 int name_width = options->stat_name_width;
2051 arg += 6;
2052 end = (char *)arg;
2054 switch (*arg) {
2055 case '-':
2056 if (!prefixcmp(arg, "-width="))
2057 width = strtoul(arg + 7, &end, 10);
2058 else if (!prefixcmp(arg, "-name-width="))
2059 name_width = strtoul(arg + 12, &end, 10);
2060 break;
2061 case '=':
2062 width = strtoul(arg+1, &end, 10);
2063 if (*end == ',')
2064 name_width = strtoul(end+1, &end, 10);
2067 /* Important! This checks all the error cases! */
2068 if (*end)
2069 return 0;
2070 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2071 options->stat_name_width = name_width;
2072 options->stat_width = width;
2074 else if (!strcmp(arg, "--check"))
2075 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2076 else if (!strcmp(arg, "--summary"))
2077 options->output_format |= DIFF_FORMAT_SUMMARY;
2078 else if (!strcmp(arg, "--patch-with-stat")) {
2079 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2081 else if (!strcmp(arg, "-z"))
2082 options->line_termination = 0;
2083 else if (!prefixcmp(arg, "-l"))
2084 options->rename_limit = strtoul(arg+2, NULL, 10);
2085 else if (!strcmp(arg, "--full-index"))
2086 options->full_index = 1;
2087 else if (!strcmp(arg, "--binary")) {
2088 options->output_format |= DIFF_FORMAT_PATCH;
2089 options->binary = 1;
2091 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
2092 options->text = 1;
2094 else if (!strcmp(arg, "--name-only"))
2095 options->output_format |= DIFF_FORMAT_NAME;
2096 else if (!strcmp(arg, "--name-status"))
2097 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2098 else if (!strcmp(arg, "-R"))
2099 options->reverse_diff = 1;
2100 else if (!prefixcmp(arg, "-S"))
2101 options->pickaxe = arg + 2;
2102 else if (!strcmp(arg, "-s")) {
2103 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2105 else if (!prefixcmp(arg, "-O"))
2106 options->orderfile = arg + 2;
2107 else if (!prefixcmp(arg, "--diff-filter="))
2108 options->filter = arg + 14;
2109 else if (!strcmp(arg, "--pickaxe-all"))
2110 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2111 else if (!strcmp(arg, "--pickaxe-regex"))
2112 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2113 else if (!prefixcmp(arg, "-B")) {
2114 if ((options->break_opt =
2115 diff_scoreopt_parse(arg)) == -1)
2116 return -1;
2118 else if (!prefixcmp(arg, "-M")) {
2119 if ((options->rename_score =
2120 diff_scoreopt_parse(arg)) == -1)
2121 return -1;
2122 options->detect_rename = DIFF_DETECT_RENAME;
2124 else if (!prefixcmp(arg, "-C")) {
2125 if ((options->rename_score =
2126 diff_scoreopt_parse(arg)) == -1)
2127 return -1;
2128 options->detect_rename = DIFF_DETECT_COPY;
2130 else if (!strcmp(arg, "--find-copies-harder"))
2131 options->find_copies_harder = 1;
2132 else if (!strcmp(arg, "--abbrev"))
2133 options->abbrev = DEFAULT_ABBREV;
2134 else if (!prefixcmp(arg, "--abbrev=")) {
2135 options->abbrev = strtoul(arg + 9, NULL, 10);
2136 if (options->abbrev < MINIMUM_ABBREV)
2137 options->abbrev = MINIMUM_ABBREV;
2138 else if (40 < options->abbrev)
2139 options->abbrev = 40;
2141 else if (!strcmp(arg, "--color"))
2142 options->color_diff = 1;
2143 else if (!strcmp(arg, "--no-color"))
2144 options->color_diff = 0;
2145 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2146 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2147 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2148 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2149 else if (!strcmp(arg, "--ignore-space-at-eol"))
2150 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2151 else if (!strcmp(arg, "--color-words"))
2152 options->color_diff = options->color_diff_words = 1;
2153 else if (!strcmp(arg, "--no-renames"))
2154 options->detect_rename = 0;
2155 else if (!strcmp(arg, "--exit-code"))
2156 options->exit_with_status = 1;
2157 else if (!strcmp(arg, "--quiet"))
2158 options->quiet = 1;
2159 else
2160 return 0;
2161 return 1;
2164 static int parse_num(const char **cp_p)
2166 unsigned long num, scale;
2167 int ch, dot;
2168 const char *cp = *cp_p;
2170 num = 0;
2171 scale = 1;
2172 dot = 0;
2173 for(;;) {
2174 ch = *cp;
2175 if ( !dot && ch == '.' ) {
2176 scale = 1;
2177 dot = 1;
2178 } else if ( ch == '%' ) {
2179 scale = dot ? scale*100 : 100;
2180 cp++; /* % is always at the end */
2181 break;
2182 } else if ( ch >= '0' && ch <= '9' ) {
2183 if ( scale < 100000 ) {
2184 scale *= 10;
2185 num = (num*10) + (ch-'0');
2187 } else {
2188 break;
2190 cp++;
2192 *cp_p = cp;
2194 /* user says num divided by scale and we say internally that
2195 * is MAX_SCORE * num / scale.
2197 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2200 int diff_scoreopt_parse(const char *opt)
2202 int opt1, opt2, cmd;
2204 if (*opt++ != '-')
2205 return -1;
2206 cmd = *opt++;
2207 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2208 return -1; /* that is not a -M, -C nor -B option */
2210 opt1 = parse_num(&opt);
2211 if (cmd != 'B')
2212 opt2 = 0;
2213 else {
2214 if (*opt == 0)
2215 opt2 = 0;
2216 else if (*opt != '/')
2217 return -1; /* we expect -B80/99 or -B80 */
2218 else {
2219 opt++;
2220 opt2 = parse_num(&opt);
2223 if (*opt != 0)
2224 return -1;
2225 return opt1 | (opt2 << 16);
2228 struct diff_queue_struct diff_queued_diff;
2230 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2232 if (queue->alloc <= queue->nr) {
2233 queue->alloc = alloc_nr(queue->alloc);
2234 queue->queue = xrealloc(queue->queue,
2235 sizeof(dp) * queue->alloc);
2237 queue->queue[queue->nr++] = dp;
2240 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2241 struct diff_filespec *one,
2242 struct diff_filespec *two)
2244 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2245 dp->one = one;
2246 dp->two = two;
2247 if (queue)
2248 diff_q(queue, dp);
2249 return dp;
2252 void diff_free_filepair(struct diff_filepair *p)
2254 diff_free_filespec_data(p->one);
2255 diff_free_filespec_data(p->two);
2256 free(p->one);
2257 free(p->two);
2258 free(p);
2261 /* This is different from find_unique_abbrev() in that
2262 * it stuffs the result with dots for alignment.
2264 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2266 int abblen;
2267 const char *abbrev;
2268 if (len == 40)
2269 return sha1_to_hex(sha1);
2271 abbrev = find_unique_abbrev(sha1, len);
2272 if (!abbrev)
2273 return sha1_to_hex(sha1);
2274 abblen = strlen(abbrev);
2275 if (abblen < 37) {
2276 static char hex[41];
2277 if (len < abblen && abblen <= len + 2)
2278 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2279 else
2280 sprintf(hex, "%s...", abbrev);
2281 return hex;
2283 return sha1_to_hex(sha1);
2286 static void diff_flush_raw(struct diff_filepair *p,
2287 struct diff_options *options)
2289 int two_paths;
2290 char status[10];
2291 int abbrev = options->abbrev;
2292 const char *path_one, *path_two;
2293 int inter_name_termination = '\t';
2294 int line_termination = options->line_termination;
2296 if (!line_termination)
2297 inter_name_termination = 0;
2299 path_one = p->one->path;
2300 path_two = p->two->path;
2301 if (line_termination) {
2302 path_one = quote_one(path_one);
2303 path_two = quote_one(path_two);
2306 if (p->score)
2307 sprintf(status, "%c%03d", p->status,
2308 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2309 else {
2310 status[0] = p->status;
2311 status[1] = 0;
2313 switch (p->status) {
2314 case DIFF_STATUS_COPIED:
2315 case DIFF_STATUS_RENAMED:
2316 two_paths = 1;
2317 break;
2318 case DIFF_STATUS_ADDED:
2319 case DIFF_STATUS_DELETED:
2320 two_paths = 0;
2321 break;
2322 default:
2323 two_paths = 0;
2324 break;
2326 if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2327 printf(":%06o %06o %s ",
2328 p->one->mode, p->two->mode,
2329 diff_unique_abbrev(p->one->sha1, abbrev));
2330 printf("%s ",
2331 diff_unique_abbrev(p->two->sha1, abbrev));
2333 printf("%s%c%s", status, inter_name_termination, path_one);
2334 if (two_paths)
2335 printf("%c%s", inter_name_termination, path_two);
2336 putchar(line_termination);
2337 if (path_one != p->one->path)
2338 free((void*)path_one);
2339 if (path_two != p->two->path)
2340 free((void*)path_two);
2343 static void diff_flush_name(struct diff_filepair *p, struct diff_options *opt)
2345 char *path = p->two->path;
2347 if (opt->line_termination)
2348 path = quote_one(p->two->path);
2349 printf("%s%c", path, opt->line_termination);
2350 if (p->two->path != path)
2351 free(path);
2354 int diff_unmodified_pair(struct diff_filepair *p)
2356 /* This function is written stricter than necessary to support
2357 * the currently implemented transformers, but the idea is to
2358 * let transformers to produce diff_filepairs any way they want,
2359 * and filter and clean them up here before producing the output.
2361 struct diff_filespec *one, *two;
2363 if (DIFF_PAIR_UNMERGED(p))
2364 return 0; /* unmerged is interesting */
2366 one = p->one;
2367 two = p->two;
2369 /* deletion, addition, mode or type change
2370 * and rename are all interesting.
2372 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2373 DIFF_PAIR_MODE_CHANGED(p) ||
2374 strcmp(one->path, two->path))
2375 return 0;
2377 /* both are valid and point at the same path. that is, we are
2378 * dealing with a change.
2380 if (one->sha1_valid && two->sha1_valid &&
2381 !hashcmp(one->sha1, two->sha1))
2382 return 1; /* no change */
2383 if (!one->sha1_valid && !two->sha1_valid)
2384 return 1; /* both look at the same file on the filesystem. */
2385 return 0;
2388 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2390 if (diff_unmodified_pair(p))
2391 return;
2393 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2394 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2395 return; /* no tree diffs in patch format */
2397 run_diff(p, o);
2400 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2401 struct diffstat_t *diffstat)
2403 if (diff_unmodified_pair(p))
2404 return;
2406 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2407 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2408 return; /* no tree diffs in patch format */
2410 run_diffstat(p, o, diffstat);
2413 static void diff_flush_checkdiff(struct diff_filepair *p,
2414 struct diff_options *o)
2416 if (diff_unmodified_pair(p))
2417 return;
2419 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2420 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2421 return; /* no tree diffs in patch format */
2423 run_checkdiff(p, o);
2426 int diff_queue_is_empty(void)
2428 struct diff_queue_struct *q = &diff_queued_diff;
2429 int i;
2430 for (i = 0; i < q->nr; i++)
2431 if (!diff_unmodified_pair(q->queue[i]))
2432 return 0;
2433 return 1;
2436 #if DIFF_DEBUG
2437 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2439 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2440 x, one ? one : "",
2441 s->path,
2442 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2443 s->mode,
2444 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2445 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2446 x, one ? one : "",
2447 s->size, s->xfrm_flags);
2450 void diff_debug_filepair(const struct diff_filepair *p, int i)
2452 diff_debug_filespec(p->one, i, "one");
2453 diff_debug_filespec(p->two, i, "two");
2454 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2455 p->score, p->status ? p->status : '?',
2456 p->source_stays, p->broken_pair);
2459 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2461 int i;
2462 if (msg)
2463 fprintf(stderr, "%s\n", msg);
2464 fprintf(stderr, "q->nr = %d\n", q->nr);
2465 for (i = 0; i < q->nr; i++) {
2466 struct diff_filepair *p = q->queue[i];
2467 diff_debug_filepair(p, i);
2470 #endif
2472 static void diff_resolve_rename_copy(void)
2474 int i, j;
2475 struct diff_filepair *p, *pp;
2476 struct diff_queue_struct *q = &diff_queued_diff;
2478 diff_debug_queue("resolve-rename-copy", q);
2480 for (i = 0; i < q->nr; i++) {
2481 p = q->queue[i];
2482 p->status = 0; /* undecided */
2483 if (DIFF_PAIR_UNMERGED(p))
2484 p->status = DIFF_STATUS_UNMERGED;
2485 else if (!DIFF_FILE_VALID(p->one))
2486 p->status = DIFF_STATUS_ADDED;
2487 else if (!DIFF_FILE_VALID(p->two))
2488 p->status = DIFF_STATUS_DELETED;
2489 else if (DIFF_PAIR_TYPE_CHANGED(p))
2490 p->status = DIFF_STATUS_TYPE_CHANGED;
2492 /* from this point on, we are dealing with a pair
2493 * whose both sides are valid and of the same type, i.e.
2494 * either in-place edit or rename/copy edit.
2496 else if (DIFF_PAIR_RENAME(p)) {
2497 if (p->source_stays) {
2498 p->status = DIFF_STATUS_COPIED;
2499 continue;
2501 /* See if there is some other filepair that
2502 * copies from the same source as us. If so
2503 * we are a copy. Otherwise we are either a
2504 * copy if the path stays, or a rename if it
2505 * does not, but we already handled "stays" case.
2507 for (j = i + 1; j < q->nr; j++) {
2508 pp = q->queue[j];
2509 if (strcmp(pp->one->path, p->one->path))
2510 continue; /* not us */
2511 if (!DIFF_PAIR_RENAME(pp))
2512 continue; /* not a rename/copy */
2513 /* pp is a rename/copy from the same source */
2514 p->status = DIFF_STATUS_COPIED;
2515 break;
2517 if (!p->status)
2518 p->status = DIFF_STATUS_RENAMED;
2520 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2521 p->one->mode != p->two->mode ||
2522 is_null_sha1(p->one->sha1))
2523 p->status = DIFF_STATUS_MODIFIED;
2524 else {
2525 /* This is a "no-change" entry and should not
2526 * happen anymore, but prepare for broken callers.
2528 error("feeding unmodified %s to diffcore",
2529 p->one->path);
2530 p->status = DIFF_STATUS_UNKNOWN;
2533 diff_debug_queue("resolve-rename-copy done", q);
2536 static int check_pair_status(struct diff_filepair *p)
2538 switch (p->status) {
2539 case DIFF_STATUS_UNKNOWN:
2540 return 0;
2541 case 0:
2542 die("internal error in diff-resolve-rename-copy");
2543 default:
2544 return 1;
2548 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2550 int fmt = opt->output_format;
2552 if (fmt & DIFF_FORMAT_CHECKDIFF)
2553 diff_flush_checkdiff(p, opt);
2554 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2555 diff_flush_raw(p, opt);
2556 else if (fmt & DIFF_FORMAT_NAME)
2557 diff_flush_name(p, opt);
2560 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2562 char *name = quote_one(fs->path);
2563 if (fs->mode)
2564 printf(" %s mode %06o %s\n", newdelete, fs->mode, name);
2565 else
2566 printf(" %s %s\n", newdelete, name);
2567 free(name);
2571 static void show_mode_change(struct diff_filepair *p, int show_name)
2573 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2574 if (show_name) {
2575 char *name = quote_one(p->two->path);
2576 printf(" mode change %06o => %06o %s\n",
2577 p->one->mode, p->two->mode, name);
2578 free(name);
2580 else
2581 printf(" mode change %06o => %06o\n",
2582 p->one->mode, p->two->mode);
2586 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2588 char *names = pprint_rename(p->one->path, p->two->path);
2590 printf(" %s %s (%d%%)\n", renamecopy, names,
2591 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2592 free(names);
2593 show_mode_change(p, 0);
2596 static void diff_summary(struct diff_filepair *p)
2598 switch(p->status) {
2599 case DIFF_STATUS_DELETED:
2600 show_file_mode_name("delete", p->one);
2601 break;
2602 case DIFF_STATUS_ADDED:
2603 show_file_mode_name("create", p->two);
2604 break;
2605 case DIFF_STATUS_COPIED:
2606 show_rename_copy("copy", p);
2607 break;
2608 case DIFF_STATUS_RENAMED:
2609 show_rename_copy("rename", p);
2610 break;
2611 default:
2612 if (p->score) {
2613 char *name = quote_one(p->two->path);
2614 printf(" rewrite %s (%d%%)\n", name,
2615 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2616 free(name);
2617 show_mode_change(p, 0);
2618 } else show_mode_change(p, 1);
2619 break;
2623 struct patch_id_t {
2624 struct xdiff_emit_state xm;
2625 SHA_CTX *ctx;
2626 int patchlen;
2629 static int remove_space(char *line, int len)
2631 int i;
2632 char *dst = line;
2633 unsigned char c;
2635 for (i = 0; i < len; i++)
2636 if (!isspace((c = line[i])))
2637 *dst++ = c;
2639 return dst - line;
2642 static void patch_id_consume(void *priv, char *line, unsigned long len)
2644 struct patch_id_t *data = priv;
2645 int new_len;
2647 /* Ignore line numbers when computing the SHA1 of the patch */
2648 if (!prefixcmp(line, "@@ -"))
2649 return;
2651 new_len = remove_space(line, len);
2653 SHA1_Update(data->ctx, line, new_len);
2654 data->patchlen += new_len;
2657 /* returns 0 upon success, and writes result into sha1 */
2658 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2660 struct diff_queue_struct *q = &diff_queued_diff;
2661 int i;
2662 SHA_CTX ctx;
2663 struct patch_id_t data;
2664 char buffer[PATH_MAX * 4 + 20];
2666 SHA1_Init(&ctx);
2667 memset(&data, 0, sizeof(struct patch_id_t));
2668 data.ctx = &ctx;
2669 data.xm.consume = patch_id_consume;
2671 for (i = 0; i < q->nr; i++) {
2672 xpparam_t xpp;
2673 xdemitconf_t xecfg;
2674 xdemitcb_t ecb;
2675 mmfile_t mf1, mf2;
2676 struct diff_filepair *p = q->queue[i];
2677 int len1, len2;
2679 if (p->status == 0)
2680 return error("internal diff status error");
2681 if (p->status == DIFF_STATUS_UNKNOWN)
2682 continue;
2683 if (diff_unmodified_pair(p))
2684 continue;
2685 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2686 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2687 continue;
2688 if (DIFF_PAIR_UNMERGED(p))
2689 continue;
2691 diff_fill_sha1_info(p->one);
2692 diff_fill_sha1_info(p->two);
2693 if (fill_mmfile(&mf1, p->one) < 0 ||
2694 fill_mmfile(&mf2, p->two) < 0)
2695 return error("unable to read files to diff");
2697 /* Maybe hash p->two? into the patch id? */
2698 if (mmfile_is_binary(&mf2))
2699 continue;
2701 len1 = remove_space(p->one->path, strlen(p->one->path));
2702 len2 = remove_space(p->two->path, strlen(p->two->path));
2703 if (p->one->mode == 0)
2704 len1 = snprintf(buffer, sizeof(buffer),
2705 "diff--gita/%.*sb/%.*s"
2706 "newfilemode%06o"
2707 "---/dev/null"
2708 "+++b/%.*s",
2709 len1, p->one->path,
2710 len2, p->two->path,
2711 p->two->mode,
2712 len2, p->two->path);
2713 else if (p->two->mode == 0)
2714 len1 = snprintf(buffer, sizeof(buffer),
2715 "diff--gita/%.*sb/%.*s"
2716 "deletedfilemode%06o"
2717 "---a/%.*s"
2718 "+++/dev/null",
2719 len1, p->one->path,
2720 len2, p->two->path,
2721 p->one->mode,
2722 len1, p->one->path);
2723 else
2724 len1 = snprintf(buffer, sizeof(buffer),
2725 "diff--gita/%.*sb/%.*s"
2726 "---a/%.*s"
2727 "+++b/%.*s",
2728 len1, p->one->path,
2729 len2, p->two->path,
2730 len1, p->one->path,
2731 len2, p->two->path);
2732 SHA1_Update(&ctx, buffer, len1);
2734 xpp.flags = XDF_NEED_MINIMAL;
2735 xecfg.ctxlen = 3;
2736 xecfg.flags = XDL_EMIT_FUNCNAMES;
2737 ecb.outf = xdiff_outf;
2738 ecb.priv = &data;
2739 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2742 SHA1_Final(sha1, &ctx);
2743 return 0;
2746 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2748 struct diff_queue_struct *q = &diff_queued_diff;
2749 int i;
2750 int result = diff_get_patch_id(options, sha1);
2752 for (i = 0; i < q->nr; i++)
2753 diff_free_filepair(q->queue[i]);
2755 free(q->queue);
2756 q->queue = NULL;
2757 q->nr = q->alloc = 0;
2759 return result;
2762 static int is_summary_empty(const struct diff_queue_struct *q)
2764 int i;
2766 for (i = 0; i < q->nr; i++) {
2767 const struct diff_filepair *p = q->queue[i];
2769 switch (p->status) {
2770 case DIFF_STATUS_DELETED:
2771 case DIFF_STATUS_ADDED:
2772 case DIFF_STATUS_COPIED:
2773 case DIFF_STATUS_RENAMED:
2774 return 0;
2775 default:
2776 if (p->score)
2777 return 0;
2778 if (p->one->mode && p->two->mode &&
2779 p->one->mode != p->two->mode)
2780 return 0;
2781 break;
2784 return 1;
2787 void diff_flush(struct diff_options *options)
2789 struct diff_queue_struct *q = &diff_queued_diff;
2790 int i, output_format = options->output_format;
2791 int separator = 0;
2794 * Order: raw, stat, summary, patch
2795 * or: name/name-status/checkdiff (other bits clear)
2797 if (!q->nr)
2798 goto free_queue;
2800 if (output_format & (DIFF_FORMAT_RAW |
2801 DIFF_FORMAT_NAME |
2802 DIFF_FORMAT_NAME_STATUS |
2803 DIFF_FORMAT_CHECKDIFF)) {
2804 for (i = 0; i < q->nr; i++) {
2805 struct diff_filepair *p = q->queue[i];
2806 if (check_pair_status(p))
2807 flush_one_pair(p, options);
2809 separator++;
2812 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2813 struct diffstat_t diffstat;
2815 memset(&diffstat, 0, sizeof(struct diffstat_t));
2816 diffstat.xm.consume = diffstat_consume;
2817 for (i = 0; i < q->nr; i++) {
2818 struct diff_filepair *p = q->queue[i];
2819 if (check_pair_status(p))
2820 diff_flush_stat(p, options, &diffstat);
2822 if (output_format & DIFF_FORMAT_NUMSTAT)
2823 show_numstat(&diffstat, options);
2824 if (output_format & DIFF_FORMAT_DIFFSTAT)
2825 show_stats(&diffstat, options);
2826 else if (output_format & DIFF_FORMAT_SHORTSTAT)
2827 show_shortstats(&diffstat);
2828 separator++;
2831 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2832 for (i = 0; i < q->nr; i++)
2833 diff_summary(q->queue[i]);
2834 separator++;
2837 if (output_format & DIFF_FORMAT_PATCH) {
2838 if (separator) {
2839 if (options->stat_sep) {
2840 /* attach patch instead of inline */
2841 fputs(options->stat_sep, stdout);
2842 } else {
2843 putchar(options->line_termination);
2847 for (i = 0; i < q->nr; i++) {
2848 struct diff_filepair *p = q->queue[i];
2849 if (check_pair_status(p))
2850 diff_flush_patch(p, options);
2854 if (output_format & DIFF_FORMAT_CALLBACK)
2855 options->format_callback(q, options, options->format_callback_data);
2857 for (i = 0; i < q->nr; i++)
2858 diff_free_filepair(q->queue[i]);
2859 free_queue:
2860 free(q->queue);
2861 q->queue = NULL;
2862 q->nr = q->alloc = 0;
2865 static void diffcore_apply_filter(const char *filter)
2867 int i;
2868 struct diff_queue_struct *q = &diff_queued_diff;
2869 struct diff_queue_struct outq;
2870 outq.queue = NULL;
2871 outq.nr = outq.alloc = 0;
2873 if (!filter)
2874 return;
2876 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2877 int found;
2878 for (i = found = 0; !found && i < q->nr; i++) {
2879 struct diff_filepair *p = q->queue[i];
2880 if (((p->status == DIFF_STATUS_MODIFIED) &&
2881 ((p->score &&
2882 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2883 (!p->score &&
2884 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2885 ((p->status != DIFF_STATUS_MODIFIED) &&
2886 strchr(filter, p->status)))
2887 found++;
2889 if (found)
2890 return;
2892 /* otherwise we will clear the whole queue
2893 * by copying the empty outq at the end of this
2894 * function, but first clear the current entries
2895 * in the queue.
2897 for (i = 0; i < q->nr; i++)
2898 diff_free_filepair(q->queue[i]);
2900 else {
2901 /* Only the matching ones */
2902 for (i = 0; i < q->nr; i++) {
2903 struct diff_filepair *p = q->queue[i];
2905 if (((p->status == DIFF_STATUS_MODIFIED) &&
2906 ((p->score &&
2907 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2908 (!p->score &&
2909 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2910 ((p->status != DIFF_STATUS_MODIFIED) &&
2911 strchr(filter, p->status)))
2912 diff_q(&outq, p);
2913 else
2914 diff_free_filepair(p);
2917 free(q->queue);
2918 *q = outq;
2921 void diffcore_std(struct diff_options *options)
2923 if (options->quiet)
2924 return;
2925 if (options->break_opt != -1)
2926 diffcore_break(options->break_opt);
2927 if (options->detect_rename)
2928 diffcore_rename(options);
2929 if (options->break_opt != -1)
2930 diffcore_merge_broken();
2931 if (options->pickaxe)
2932 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2933 if (options->orderfile)
2934 diffcore_order(options->orderfile);
2935 diff_resolve_rename_copy();
2936 diffcore_apply_filter(options->filter);
2938 options->has_changes = !!diff_queued_diff.nr;
2942 void diff_addremove(struct diff_options *options,
2943 int addremove, unsigned mode,
2944 const unsigned char *sha1,
2945 const char *base, const char *path)
2947 char concatpath[PATH_MAX];
2948 struct diff_filespec *one, *two;
2950 /* This may look odd, but it is a preparation for
2951 * feeding "there are unchanged files which should
2952 * not produce diffs, but when you are doing copy
2953 * detection you would need them, so here they are"
2954 * entries to the diff-core. They will be prefixed
2955 * with something like '=' or '*' (I haven't decided
2956 * which but should not make any difference).
2957 * Feeding the same new and old to diff_change()
2958 * also has the same effect.
2959 * Before the final output happens, they are pruned after
2960 * merged into rename/copy pairs as appropriate.
2962 if (options->reverse_diff)
2963 addremove = (addremove == '+' ? '-' :
2964 addremove == '-' ? '+' : addremove);
2966 if (!path) path = "";
2967 sprintf(concatpath, "%s%s", base, path);
2968 one = alloc_filespec(concatpath);
2969 two = alloc_filespec(concatpath);
2971 if (addremove != '+')
2972 fill_filespec(one, sha1, mode);
2973 if (addremove != '-')
2974 fill_filespec(two, sha1, mode);
2976 diff_queue(&diff_queued_diff, one, two);
2977 options->has_changes = 1;
2980 void diff_change(struct diff_options *options,
2981 unsigned old_mode, unsigned new_mode,
2982 const unsigned char *old_sha1,
2983 const unsigned char *new_sha1,
2984 const char *base, const char *path)
2986 char concatpath[PATH_MAX];
2987 struct diff_filespec *one, *two;
2989 if (options->reverse_diff) {
2990 unsigned tmp;
2991 const unsigned char *tmp_c;
2992 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2993 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2995 if (!path) path = "";
2996 sprintf(concatpath, "%s%s", base, path);
2997 one = alloc_filespec(concatpath);
2998 two = alloc_filespec(concatpath);
2999 fill_filespec(one, old_sha1, old_mode);
3000 fill_filespec(two, new_sha1, new_mode);
3002 diff_queue(&diff_queued_diff, one, two);
3003 options->has_changes = 1;
3006 void diff_unmerge(struct diff_options *options,
3007 const char *path,
3008 unsigned mode, const unsigned char *sha1)
3010 struct diff_filespec *one, *two;
3011 one = alloc_filespec(path);
3012 two = alloc_filespec(path);
3013 fill_filespec(one, sha1, mode);
3014 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;