2df14b2469362bfd8a30a45457953d993baae66d
[git/git.git] / diff.c
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"
14
15 static int use_size_cache;
16
17 static int diff_detect_rename_default;
18 static int diff_rename_limit_default = -1;
19 static int diff_use_color_default;
20
21 static char diff_colors[][COLOR_MAXLEN] = {
22 "\033[m", /* reset */
23 "", /* PLAIN (normal) */
24 "\033[1m", /* METAINFO (bold) */
25 "\033[36m", /* FRAGINFO (cyan) */
26 "\033[31m", /* OLD (red) */
27 "\033[32m", /* NEW (green) */
28 "\033[33m", /* COMMIT (yellow) */
29 "\033[41m", /* WHITESPACE (red background) */
30 };
31
32 static int parse_diff_color_slot(const char *var, int ofs)
33 {
34 if (!strcasecmp(var+ofs, "plain"))
35 return DIFF_PLAIN;
36 if (!strcasecmp(var+ofs, "meta"))
37 return DIFF_METAINFO;
38 if (!strcasecmp(var+ofs, "frag"))
39 return DIFF_FRAGINFO;
40 if (!strcasecmp(var+ofs, "old"))
41 return DIFF_FILE_OLD;
42 if (!strcasecmp(var+ofs, "new"))
43 return DIFF_FILE_NEW;
44 if (!strcasecmp(var+ofs, "commit"))
45 return DIFF_COMMIT;
46 if (!strcasecmp(var+ofs, "whitespace"))
47 return DIFF_WHITESPACE;
48 die("bad config variable '%s'", var);
49 }
50
51 /*
52 * These are to give UI layer defaults.
53 * The core-level commands such as git-diff-files should
54 * never be affected by the setting of diff.renames
55 * the user happens to have in the configuration file.
56 */
57 int git_diff_ui_config(const char *var, const char *value)
58 {
59 if (!strcmp(var, "diff.renamelimit")) {
60 diff_rename_limit_default = git_config_int(var, value);
61 return 0;
62 }
63 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
64 diff_use_color_default = git_config_colorbool(var, value);
65 return 0;
66 }
67 if (!strcmp(var, "diff.renames")) {
68 if (!value)
69 diff_detect_rename_default = DIFF_DETECT_RENAME;
70 else if (!strcasecmp(value, "copies") ||
71 !strcasecmp(value, "copy"))
72 diff_detect_rename_default = DIFF_DETECT_COPY;
73 else if (git_config_bool(var,value))
74 diff_detect_rename_default = DIFF_DETECT_RENAME;
75 return 0;
76 }
77 if (!strncmp(var, "diff.color.", 11) || !strncmp(var, "color.diff.", 11)) {
78 int slot = parse_diff_color_slot(var, 11);
79 color_parse(value, var, diff_colors[slot]);
80 return 0;
81 }
82 return git_default_config(var, value);
83 }
84
85 static char *quote_one(const char *str)
86 {
87 int needlen;
88 char *xp;
89
90 if (!str)
91 return NULL;
92 needlen = quote_c_style(str, NULL, NULL, 0);
93 if (!needlen)
94 return xstrdup(str);
95 xp = xmalloc(needlen + 1);
96 quote_c_style(str, xp, NULL, 0);
97 return xp;
98 }
99
100 static char *quote_two(const char *one, const char *two)
101 {
102 int need_one = quote_c_style(one, NULL, NULL, 1);
103 int need_two = quote_c_style(two, NULL, NULL, 1);
104 char *xp;
105
106 if (need_one + need_two) {
107 if (!need_one) need_one = strlen(one);
108 if (!need_two) need_one = strlen(two);
109
110 xp = xmalloc(need_one + need_two + 3);
111 xp[0] = '"';
112 quote_c_style(one, xp + 1, NULL, 1);
113 quote_c_style(two, xp + need_one + 1, NULL, 1);
114 strcpy(xp + need_one + need_two + 1, "\"");
115 return xp;
116 }
117 need_one = strlen(one);
118 need_two = strlen(two);
119 xp = xmalloc(need_one + need_two + 1);
120 strcpy(xp, one);
121 strcpy(xp + need_one, two);
122 return xp;
123 }
124
125 static const char *external_diff(void)
126 {
127 static const char *external_diff_cmd = NULL;
128 static int done_preparing = 0;
129
130 if (done_preparing)
131 return external_diff_cmd;
132 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
133 done_preparing = 1;
134 return external_diff_cmd;
135 }
136
137 #define TEMPFILE_PATH_LEN 50
138
139 static struct diff_tempfile {
140 const char *name; /* filename external diff should read from */
141 char hex[41];
142 char mode[10];
143 char tmp_path[TEMPFILE_PATH_LEN];
144 } diff_temp[2];
145
146 static int count_lines(const char *data, int size)
147 {
148 int count, ch, completely_empty = 1, nl_just_seen = 0;
149 count = 0;
150 while (0 < size--) {
151 ch = *data++;
152 if (ch == '\n') {
153 count++;
154 nl_just_seen = 1;
155 completely_empty = 0;
156 }
157 else {
158 nl_just_seen = 0;
159 completely_empty = 0;
160 }
161 }
162 if (completely_empty)
163 return 0;
164 if (!nl_just_seen)
165 count++; /* no trailing newline */
166 return count;
167 }
168
169 static void print_line_count(int count)
170 {
171 switch (count) {
172 case 0:
173 printf("0,0");
174 break;
175 case 1:
176 printf("1");
177 break;
178 default:
179 printf("1,%d", count);
180 break;
181 }
182 }
183
184 static void copy_file(int prefix, const char *data, int size)
185 {
186 int ch, nl_just_seen = 1;
187 while (0 < size--) {
188 ch = *data++;
189 if (nl_just_seen)
190 putchar(prefix);
191 putchar(ch);
192 if (ch == '\n')
193 nl_just_seen = 1;
194 else
195 nl_just_seen = 0;
196 }
197 if (!nl_just_seen)
198 printf("\n\\ No newline at end of file\n");
199 }
200
201 static void emit_rewrite_diff(const char *name_a,
202 const char *name_b,
203 struct diff_filespec *one,
204 struct diff_filespec *two)
205 {
206 int lc_a, lc_b;
207 diff_populate_filespec(one, 0);
208 diff_populate_filespec(two, 0);
209 lc_a = count_lines(one->data, one->size);
210 lc_b = count_lines(two->data, two->size);
211 printf("--- a/%s\n+++ b/%s\n@@ -", name_a, name_b);
212 print_line_count(lc_a);
213 printf(" +");
214 print_line_count(lc_b);
215 printf(" @@\n");
216 if (lc_a)
217 copy_file('-', one->data, one->size);
218 if (lc_b)
219 copy_file('+', two->data, two->size);
220 }
221
222 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
223 {
224 if (!DIFF_FILE_VALID(one)) {
225 mf->ptr = (char *)""; /* does not matter */
226 mf->size = 0;
227 return 0;
228 }
229 else if (diff_populate_filespec(one, 0))
230 return -1;
231 mf->ptr = one->data;
232 mf->size = one->size;
233 return 0;
234 }
235
236 struct diff_words_buffer {
237 mmfile_t text;
238 long alloc;
239 long current; /* output pointer */
240 int suppressed_newline;
241 };
242
243 static void diff_words_append(char *line, unsigned long len,
244 struct diff_words_buffer *buffer)
245 {
246 if (buffer->text.size + len > buffer->alloc) {
247 buffer->alloc = (buffer->text.size + len) * 3 / 2;
248 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
249 }
250 line++;
251 len--;
252 memcpy(buffer->text.ptr + buffer->text.size, line, len);
253 buffer->text.size += len;
254 }
255
256 struct diff_words_data {
257 struct xdiff_emit_state xm;
258 struct diff_words_buffer minus, plus;
259 };
260
261 static void print_word(struct diff_words_buffer *buffer, int len, int color,
262 int suppress_newline)
263 {
264 const char *ptr;
265 int eol = 0;
266
267 if (len == 0)
268 return;
269
270 ptr = buffer->text.ptr + buffer->current;
271 buffer->current += len;
272
273 if (ptr[len - 1] == '\n') {
274 eol = 1;
275 len--;
276 }
277
278 fputs(diff_get_color(1, color), stdout);
279 fwrite(ptr, len, 1, stdout);
280 fputs(diff_get_color(1, DIFF_RESET), stdout);
281
282 if (eol) {
283 if (suppress_newline)
284 buffer->suppressed_newline = 1;
285 else
286 putchar('\n');
287 }
288 }
289
290 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
291 {
292 struct diff_words_data *diff_words = priv;
293
294 if (diff_words->minus.suppressed_newline) {
295 if (line[0] != '+')
296 putchar('\n');
297 diff_words->minus.suppressed_newline = 0;
298 }
299
300 len--;
301 switch (line[0]) {
302 case '-':
303 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
304 break;
305 case '+':
306 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
307 break;
308 case ' ':
309 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
310 diff_words->minus.current += len;
311 break;
312 }
313 }
314
315 /* this executes the word diff on the accumulated buffers */
316 static void diff_words_show(struct diff_words_data *diff_words)
317 {
318 xpparam_t xpp;
319 xdemitconf_t xecfg;
320 xdemitcb_t ecb;
321 mmfile_t minus, plus;
322 int i;
323
324 minus.size = diff_words->minus.text.size;
325 minus.ptr = xmalloc(minus.size);
326 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
327 for (i = 0; i < minus.size; i++)
328 if (isspace(minus.ptr[i]))
329 minus.ptr[i] = '\n';
330 diff_words->minus.current = 0;
331
332 plus.size = diff_words->plus.text.size;
333 plus.ptr = xmalloc(plus.size);
334 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
335 for (i = 0; i < plus.size; i++)
336 if (isspace(plus.ptr[i]))
337 plus.ptr[i] = '\n';
338 diff_words->plus.current = 0;
339
340 xpp.flags = XDF_NEED_MINIMAL;
341 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
342 xecfg.flags = 0;
343 ecb.outf = xdiff_outf;
344 ecb.priv = diff_words;
345 diff_words->xm.consume = fn_out_diff_words_aux;
346 xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
347
348 free(minus.ptr);
349 free(plus.ptr);
350 diff_words->minus.text.size = diff_words->plus.text.size = 0;
351
352 if (diff_words->minus.suppressed_newline) {
353 putchar('\n');
354 diff_words->minus.suppressed_newline = 0;
355 }
356 }
357
358 struct emit_callback {
359 struct xdiff_emit_state xm;
360 int nparents, color_diff;
361 const char **label_path;
362 struct diff_words_data *diff_words;
363 };
364
365 static void free_diff_words_data(struct emit_callback *ecbdata)
366 {
367 if (ecbdata->diff_words) {
368 /* flush buffers */
369 if (ecbdata->diff_words->minus.text.size ||
370 ecbdata->diff_words->plus.text.size)
371 diff_words_show(ecbdata->diff_words);
372
373 if (ecbdata->diff_words->minus.text.ptr)
374 free (ecbdata->diff_words->minus.text.ptr);
375 if (ecbdata->diff_words->plus.text.ptr)
376 free (ecbdata->diff_words->plus.text.ptr);
377 free(ecbdata->diff_words);
378 ecbdata->diff_words = NULL;
379 }
380 }
381
382 const char *diff_get_color(int diff_use_color, enum color_diff ix)
383 {
384 if (diff_use_color)
385 return diff_colors[ix];
386 return "";
387 }
388
389 static void emit_line(const char *set, const char *reset, const char *line, int len)
390 {
391 if (len > 0 && line[len-1] == '\n')
392 len--;
393 fputs(set, stdout);
394 fwrite(line, len, 1, stdout);
395 puts(reset);
396 }
397
398 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
399 {
400 int col0 = ecbdata->nparents;
401 int last_tab_in_indent = -1;
402 int last_space_in_indent = -1;
403 int i;
404 int tail = len;
405 int need_highlight_leading_space = 0;
406 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
407 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
408
409 if (!*ws) {
410 emit_line(set, reset, line, len);
411 return;
412 }
413
414 /* The line is a newly added line. Does it have funny leading
415 * whitespaces? In indent, SP should never precede a TAB.
416 */
417 for (i = col0; i < len; i++) {
418 if (line[i] == '\t') {
419 last_tab_in_indent = i;
420 if (0 <= last_space_in_indent)
421 need_highlight_leading_space = 1;
422 }
423 else if (line[i] == ' ')
424 last_space_in_indent = i;
425 else
426 break;
427 }
428 fputs(set, stdout);
429 fwrite(line, col0, 1, stdout);
430 fputs(reset, stdout);
431 if (((i == len) || line[i] == '\n') && i != col0) {
432 /* The whole line was indent */
433 emit_line(ws, reset, line + col0, len - col0);
434 return;
435 }
436 i = col0;
437 if (need_highlight_leading_space) {
438 while (i < last_tab_in_indent) {
439 if (line[i] == ' ') {
440 fputs(ws, stdout);
441 putchar(' ');
442 fputs(reset, stdout);
443 }
444 else
445 putchar(line[i]);
446 i++;
447 }
448 }
449 tail = len - 1;
450 if (line[tail] == '\n' && i < tail)
451 tail--;
452 while (i < tail) {
453 if (!isspace(line[tail]))
454 break;
455 tail--;
456 }
457 if ((i < tail && line[tail + 1] != '\n')) {
458 /* This has whitespace between tail+1..len */
459 fputs(set, stdout);
460 fwrite(line + i, tail - i + 1, 1, stdout);
461 fputs(reset, stdout);
462 emit_line(ws, reset, line + tail + 1, len - tail - 1);
463 }
464 else
465 emit_line(set, reset, line + i, len - i);
466 }
467
468 static void fn_out_consume(void *priv, char *line, unsigned long len)
469 {
470 int i;
471 int color;
472 struct emit_callback *ecbdata = priv;
473 const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
474 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
475
476 if (ecbdata->label_path[0]) {
477 printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
478 printf("%s+++ %s%s\n", set, ecbdata->label_path[1], reset);
479 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
480 }
481
482 /* This is not really necessary for now because
483 * this codepath only deals with two-way diffs.
484 */
485 for (i = 0; i < len && line[i] == '@'; i++)
486 ;
487 if (2 <= i && i < len && line[i] == ' ') {
488 ecbdata->nparents = i - 1;
489 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
490 reset, line, len);
491 return;
492 }
493
494 if (len < ecbdata->nparents) {
495 set = reset;
496 emit_line(reset, reset, line, len);
497 return;
498 }
499
500 color = DIFF_PLAIN;
501 if (ecbdata->diff_words && ecbdata->nparents != 1)
502 /* fall back to normal diff */
503 free_diff_words_data(ecbdata);
504 if (ecbdata->diff_words) {
505 if (line[0] == '-') {
506 diff_words_append(line, len,
507 &ecbdata->diff_words->minus);
508 return;
509 } else if (line[0] == '+') {
510 diff_words_append(line, len,
511 &ecbdata->diff_words->plus);
512 return;
513 }
514 if (ecbdata->diff_words->minus.text.size ||
515 ecbdata->diff_words->plus.text.size)
516 diff_words_show(ecbdata->diff_words);
517 line++;
518 len--;
519 emit_line(set, reset, line, len);
520 return;
521 }
522 for (i = 0; i < ecbdata->nparents && len; i++) {
523 if (line[i] == '-')
524 color = DIFF_FILE_OLD;
525 else if (line[i] == '+')
526 color = DIFF_FILE_NEW;
527 }
528
529 if (color != DIFF_FILE_NEW) {
530 emit_line(diff_get_color(ecbdata->color_diff, color),
531 reset, line, len);
532 return;
533 }
534 emit_add_line(reset, ecbdata, line, len);
535 }
536
537 static char *pprint_rename(const char *a, const char *b)
538 {
539 const char *old = a;
540 const char *new = b;
541 char *name = NULL;
542 int pfx_length, sfx_length;
543 int len_a = strlen(a);
544 int len_b = strlen(b);
545
546 /* Find common prefix */
547 pfx_length = 0;
548 while (*old && *new && *old == *new) {
549 if (*old == '/')
550 pfx_length = old - a + 1;
551 old++;
552 new++;
553 }
554
555 /* Find common suffix */
556 old = a + len_a;
557 new = b + len_b;
558 sfx_length = 0;
559 while (a <= old && b <= new && *old == *new) {
560 if (*old == '/')
561 sfx_length = len_a - (old - a);
562 old--;
563 new--;
564 }
565
566 /*
567 * pfx{mid-a => mid-b}sfx
568 * {pfx-a => pfx-b}sfx
569 * pfx{sfx-a => sfx-b}
570 * name-a => name-b
571 */
572 if (pfx_length + sfx_length) {
573 int a_midlen = len_a - pfx_length - sfx_length;
574 int b_midlen = len_b - pfx_length - sfx_length;
575 if (a_midlen < 0) a_midlen = 0;
576 if (b_midlen < 0) b_midlen = 0;
577
578 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
579 sprintf(name, "%.*s{%.*s => %.*s}%s",
580 pfx_length, a,
581 a_midlen, a + pfx_length,
582 b_midlen, b + pfx_length,
583 a + len_a - sfx_length);
584 }
585 else {
586 name = xmalloc(len_a + len_b + 5);
587 sprintf(name, "%s => %s", a, b);
588 }
589 return name;
590 }
591
592 struct diffstat_t {
593 struct xdiff_emit_state xm;
594
595 int nr;
596 int alloc;
597 struct diffstat_file {
598 char *name;
599 unsigned is_unmerged:1;
600 unsigned is_binary:1;
601 unsigned is_renamed:1;
602 unsigned int added, deleted;
603 } **files;
604 };
605
606 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
607 const char *name_a,
608 const char *name_b)
609 {
610 struct diffstat_file *x;
611 x = xcalloc(sizeof (*x), 1);
612 if (diffstat->nr == diffstat->alloc) {
613 diffstat->alloc = alloc_nr(diffstat->alloc);
614 diffstat->files = xrealloc(diffstat->files,
615 diffstat->alloc * sizeof(x));
616 }
617 diffstat->files[diffstat->nr++] = x;
618 if (name_b) {
619 x->name = pprint_rename(name_a, name_b);
620 x->is_renamed = 1;
621 }
622 else
623 x->name = xstrdup(name_a);
624 return x;
625 }
626
627 static void diffstat_consume(void *priv, char *line, unsigned long len)
628 {
629 struct diffstat_t *diffstat = priv;
630 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
631
632 if (line[0] == '+')
633 x->added++;
634 else if (line[0] == '-')
635 x->deleted++;
636 }
637
638 const char mime_boundary_leader[] = "------------";
639
640 static int scale_linear(int it, int width, int max_change)
641 {
642 /*
643 * make sure that at least one '-' is printed if there were deletions,
644 * and likewise for '+'.
645 */
646 if (max_change < 2)
647 return it;
648 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
649 }
650
651 static void show_name(const char *prefix, const char *name, int len,
652 const char *reset, const char *set)
653 {
654 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
655 }
656
657 static void show_graph(char ch, int cnt, const char *set, const char *reset)
658 {
659 if (cnt <= 0)
660 return;
661 printf("%s", set);
662 while (cnt--)
663 putchar(ch);
664 printf("%s", reset);
665 }
666
667 static void show_stats(struct diffstat_t* data, struct diff_options *options)
668 {
669 int i, len, add, del, total, adds = 0, dels = 0;
670 int max_change = 0, max_len = 0;
671 int total_files = data->nr;
672 int width, name_width;
673 const char *reset, *set, *add_c, *del_c;
674
675 if (data->nr == 0)
676 return;
677
678 width = options->stat_width ? options->stat_width : 80;
679 name_width = options->stat_name_width ? options->stat_name_width : 50;
680
681 /* Sanity: give at least 5 columns to the graph,
682 * but leave at least 10 columns for the name.
683 */
684 if (width < name_width + 15) {
685 if (name_width <= 25)
686 width = name_width + 15;
687 else
688 name_width = width - 15;
689 }
690
691 /* Find the longest filename and max number of changes */
692 reset = diff_get_color(options->color_diff, DIFF_RESET);
693 set = diff_get_color(options->color_diff, DIFF_PLAIN);
694 add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
695 del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
696
697 for (i = 0; i < data->nr; i++) {
698 struct diffstat_file *file = data->files[i];
699 int change = file->added + file->deleted;
700
701 len = quote_c_style(file->name, NULL, NULL, 0);
702 if (len) {
703 char *qname = xmalloc(len + 1);
704 quote_c_style(file->name, qname, NULL, 0);
705 free(file->name);
706 file->name = qname;
707 }
708
709 len = strlen(file->name);
710 if (max_len < len)
711 max_len = len;
712
713 if (file->is_binary || file->is_unmerged)
714 continue;
715 if (max_change < change)
716 max_change = change;
717 }
718
719 /* Compute the width of the graph part;
720 * 10 is for one blank at the beginning of the line plus
721 * " | count " between the name and the graph.
722 *
723 * From here on, name_width is the width of the name area,
724 * and width is the width of the graph area.
725 */
726 name_width = (name_width < max_len) ? name_width : max_len;
727 if (width < (name_width + 10) + max_change)
728 width = width - (name_width + 10);
729 else
730 width = max_change;
731
732 for (i = 0; i < data->nr; i++) {
733 const char *prefix = "";
734 char *name = data->files[i]->name;
735 int added = data->files[i]->added;
736 int deleted = data->files[i]->deleted;
737 int name_len;
738
739 /*
740 * "scale" the filename
741 */
742 len = name_width;
743 name_len = strlen(name);
744 if (name_width < name_len) {
745 char *slash;
746 prefix = "...";
747 len -= 3;
748 name += name_len - len;
749 slash = strchr(name, '/');
750 if (slash)
751 name = slash;
752 }
753
754 if (data->files[i]->is_binary) {
755 show_name(prefix, name, len, reset, set);
756 printf(" Bin\n");
757 goto free_diffstat_file;
758 }
759 else if (data->files[i]->is_unmerged) {
760 show_name(prefix, name, len, reset, set);
761 printf(" Unmerged\n");
762 goto free_diffstat_file;
763 }
764 else if (!data->files[i]->is_renamed &&
765 (added + deleted == 0)) {
766 total_files--;
767 goto free_diffstat_file;
768 }
769
770 /*
771 * scale the add/delete
772 */
773 add = added;
774 del = deleted;
775 total = add + del;
776 adds += add;
777 dels += del;
778
779 if (width <= max_change) {
780 add = scale_linear(add, width, max_change);
781 del = scale_linear(del, width, max_change);
782 total = add + del;
783 }
784 show_name(prefix, name, len, reset, set);
785 printf("%5d ", added + deleted);
786 show_graph('+', add, add_c, reset);
787 show_graph('-', del, del_c, reset);
788 putchar('\n');
789 free_diffstat_file:
790 free(data->files[i]->name);
791 free(data->files[i]);
792 }
793 free(data->files);
794 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
795 set, total_files, adds, dels, reset);
796 }
797
798 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
799 {
800 int i;
801
802 for (i = 0; i < data->nr; i++) {
803 struct diffstat_file *file = data->files[i];
804
805 if (file->is_binary)
806 printf("-\t-\t");
807 else
808 printf("%d\t%d\t", file->added, file->deleted);
809 if (options->line_termination &&
810 quote_c_style(file->name, NULL, NULL, 0))
811 quote_c_style(file->name, NULL, stdout, 0);
812 else
813 fputs(file->name, stdout);
814 putchar(options->line_termination);
815 }
816 }
817
818 struct checkdiff_t {
819 struct xdiff_emit_state xm;
820 const char *filename;
821 int lineno;
822 };
823
824 static void checkdiff_consume(void *priv, char *line, unsigned long len)
825 {
826 struct checkdiff_t *data = priv;
827
828 if (line[0] == '+') {
829 int i, spaces = 0;
830
831 data->lineno++;
832
833 /* check space before tab */
834 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
835 if (line[i] == ' ')
836 spaces++;
837 if (line[i - 1] == '\t' && spaces)
838 printf("%s:%d: space before tab:%.*s\n",
839 data->filename, data->lineno, (int)len, line);
840
841 /* check white space at line end */
842 if (line[len - 1] == '\n')
843 len--;
844 if (isspace(line[len - 1]))
845 printf("%s:%d: white space at end: %.*s\n",
846 data->filename, data->lineno, (int)len, line);
847 } else if (line[0] == ' ')
848 data->lineno++;
849 else if (line[0] == '@') {
850 char *plus = strchr(line, '+');
851 if (plus)
852 data->lineno = strtol(plus, NULL, 10);
853 else
854 die("invalid diff");
855 }
856 }
857
858 static unsigned char *deflate_it(char *data,
859 unsigned long size,
860 unsigned long *result_size)
861 {
862 int bound;
863 unsigned char *deflated;
864 z_stream stream;
865
866 memset(&stream, 0, sizeof(stream));
867 deflateInit(&stream, zlib_compression_level);
868 bound = deflateBound(&stream, size);
869 deflated = xmalloc(bound);
870 stream.next_out = deflated;
871 stream.avail_out = bound;
872
873 stream.next_in = (unsigned char *)data;
874 stream.avail_in = size;
875 while (deflate(&stream, Z_FINISH) == Z_OK)
876 ; /* nothing */
877 deflateEnd(&stream);
878 *result_size = stream.total_out;
879 return deflated;
880 }
881
882 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
883 {
884 void *cp;
885 void *delta;
886 void *deflated;
887 void *data;
888 unsigned long orig_size;
889 unsigned long delta_size;
890 unsigned long deflate_size;
891 unsigned long data_size;
892
893 /* We could do deflated delta, or we could do just deflated two,
894 * whichever is smaller.
895 */
896 delta = NULL;
897 deflated = deflate_it(two->ptr, two->size, &deflate_size);
898 if (one->size && two->size) {
899 delta = diff_delta(one->ptr, one->size,
900 two->ptr, two->size,
901 &delta_size, deflate_size);
902 if (delta) {
903 void *to_free = delta;
904 orig_size = delta_size;
905 delta = deflate_it(delta, delta_size, &delta_size);
906 free(to_free);
907 }
908 }
909
910 if (delta && delta_size < deflate_size) {
911 printf("delta %lu\n", orig_size);
912 free(deflated);
913 data = delta;
914 data_size = delta_size;
915 }
916 else {
917 printf("literal %lu\n", two->size);
918 free(delta);
919 data = deflated;
920 data_size = deflate_size;
921 }
922
923 /* emit data encoded in base85 */
924 cp = data;
925 while (data_size) {
926 int bytes = (52 < data_size) ? 52 : data_size;
927 char line[70];
928 data_size -= bytes;
929 if (bytes <= 26)
930 line[0] = bytes + 'A' - 1;
931 else
932 line[0] = bytes - 26 + 'a' - 1;
933 encode_85(line + 1, cp, bytes);
934 cp = (char *) cp + bytes;
935 puts(line);
936 }
937 printf("\n");
938 free(data);
939 }
940
941 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
942 {
943 printf("GIT binary patch\n");
944 emit_binary_diff_body(one, two);
945 emit_binary_diff_body(two, one);
946 }
947
948 #define FIRST_FEW_BYTES 8000
949 static int mmfile_is_binary(mmfile_t *mf)
950 {
951 long sz = mf->size;
952 if (FIRST_FEW_BYTES < sz)
953 sz = FIRST_FEW_BYTES;
954 return !!memchr(mf->ptr, 0, sz);
955 }
956
957 static void builtin_diff(const char *name_a,
958 const char *name_b,
959 struct diff_filespec *one,
960 struct diff_filespec *two,
961 const char *xfrm_msg,
962 struct diff_options *o,
963 int complete_rewrite)
964 {
965 mmfile_t mf1, mf2;
966 const char *lbl[2];
967 char *a_one, *b_two;
968 const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
969 const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
970
971 a_one = quote_two("a/", name_a);
972 b_two = quote_two("b/", name_b);
973 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
974 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
975 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
976 if (lbl[0][0] == '/') {
977 /* /dev/null */
978 printf("%snew file mode %06o%s\n", set, two->mode, reset);
979 if (xfrm_msg && xfrm_msg[0])
980 printf("%s%s%s\n", set, xfrm_msg, reset);
981 }
982 else if (lbl[1][0] == '/') {
983 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
984 if (xfrm_msg && xfrm_msg[0])
985 printf("%s%s%s\n", set, xfrm_msg, reset);
986 }
987 else {
988 if (one->mode != two->mode) {
989 printf("%sold mode %06o%s\n", set, one->mode, reset);
990 printf("%snew mode %06o%s\n", set, two->mode, reset);
991 }
992 if (xfrm_msg && xfrm_msg[0])
993 printf("%s%s%s\n", set, xfrm_msg, reset);
994 /*
995 * we do not run diff between different kind
996 * of objects.
997 */
998 if ((one->mode ^ two->mode) & S_IFMT)
999 goto free_ab_and_return;
1000 if (complete_rewrite) {
1001 emit_rewrite_diff(name_a, name_b, one, two);
1002 goto free_ab_and_return;
1003 }
1004 }
1005
1006 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1007 die("unable to read files to diff");
1008
1009 if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
1010 /* Quite common confusing case */
1011 if (mf1.size == mf2.size &&
1012 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1013 goto free_ab_and_return;
1014 if (o->binary)
1015 emit_binary_diff(&mf1, &mf2);
1016 else
1017 printf("Binary files %s and %s differ\n",
1018 lbl[0], lbl[1]);
1019 }
1020 else {
1021 /* Crazy xdl interfaces.. */
1022 const char *diffopts = getenv("GIT_DIFF_OPTS");
1023 xpparam_t xpp;
1024 xdemitconf_t xecfg;
1025 xdemitcb_t ecb;
1026 struct emit_callback ecbdata;
1027
1028 memset(&ecbdata, 0, sizeof(ecbdata));
1029 ecbdata.label_path = lbl;
1030 ecbdata.color_diff = o->color_diff;
1031 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1032 xecfg.ctxlen = o->context;
1033 xecfg.flags = XDL_EMIT_FUNCNAMES;
1034 if (!diffopts)
1035 ;
1036 else if (!strncmp(diffopts, "--unified=", 10))
1037 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1038 else if (!strncmp(diffopts, "-u", 2))
1039 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1040 ecb.outf = xdiff_outf;
1041 ecb.priv = &ecbdata;
1042 ecbdata.xm.consume = fn_out_consume;
1043 if (o->color_diff_words)
1044 ecbdata.diff_words =
1045 xcalloc(1, sizeof(struct diff_words_data));
1046 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1047 if (o->color_diff_words)
1048 free_diff_words_data(&ecbdata);
1049 }
1050
1051 free_ab_and_return:
1052 free(a_one);
1053 free(b_two);
1054 return;
1055 }
1056
1057 static void builtin_diffstat(const char *name_a, const char *name_b,
1058 struct diff_filespec *one,
1059 struct diff_filespec *two,
1060 struct diffstat_t *diffstat,
1061 struct diff_options *o,
1062 int complete_rewrite)
1063 {
1064 mmfile_t mf1, mf2;
1065 struct diffstat_file *data;
1066
1067 data = diffstat_add(diffstat, name_a, name_b);
1068
1069 if (!one || !two) {
1070 data->is_unmerged = 1;
1071 return;
1072 }
1073 if (complete_rewrite) {
1074 diff_populate_filespec(one, 0);
1075 diff_populate_filespec(two, 0);
1076 data->deleted = count_lines(one->data, one->size);
1077 data->added = count_lines(two->data, two->size);
1078 return;
1079 }
1080 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1081 die("unable to read files to diff");
1082
1083 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
1084 data->is_binary = 1;
1085 else {
1086 /* Crazy xdl interfaces.. */
1087 xpparam_t xpp;
1088 xdemitconf_t xecfg;
1089 xdemitcb_t ecb;
1090
1091 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1092 xecfg.ctxlen = 0;
1093 xecfg.flags = 0;
1094 ecb.outf = xdiff_outf;
1095 ecb.priv = diffstat;
1096 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1097 }
1098 }
1099
1100 static void builtin_checkdiff(const char *name_a, const char *name_b,
1101 struct diff_filespec *one,
1102 struct diff_filespec *two)
1103 {
1104 mmfile_t mf1, mf2;
1105 struct checkdiff_t data;
1106
1107 if (!two)
1108 return;
1109
1110 memset(&data, 0, sizeof(data));
1111 data.xm.consume = checkdiff_consume;
1112 data.filename = name_b ? name_b : name_a;
1113 data.lineno = 0;
1114
1115 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1116 die("unable to read files to diff");
1117
1118 if (mmfile_is_binary(&mf2))
1119 return;
1120 else {
1121 /* Crazy xdl interfaces.. */
1122 xpparam_t xpp;
1123 xdemitconf_t xecfg;
1124 xdemitcb_t ecb;
1125
1126 xpp.flags = XDF_NEED_MINIMAL;
1127 xecfg.ctxlen = 0;
1128 xecfg.flags = 0;
1129 ecb.outf = xdiff_outf;
1130 ecb.priv = &data;
1131 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1132 }
1133 }
1134
1135 struct diff_filespec *alloc_filespec(const char *path)
1136 {
1137 int namelen = strlen(path);
1138 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1139
1140 memset(spec, 0, sizeof(*spec));
1141 spec->path = (char *)(spec + 1);
1142 memcpy(spec->path, path, namelen+1);
1143 return spec;
1144 }
1145
1146 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1147 unsigned short mode)
1148 {
1149 if (mode) {
1150 spec->mode = canon_mode(mode);
1151 hashcpy(spec->sha1, sha1);
1152 spec->sha1_valid = !is_null_sha1(sha1);
1153 }
1154 }
1155
1156 /*
1157 * Given a name and sha1 pair, if the dircache tells us the file in
1158 * the work tree has that object contents, return true, so that
1159 * prepare_temp_file() does not have to inflate and extract.
1160 */
1161 static int work_tree_matches(const char *name, const unsigned char *sha1)
1162 {
1163 struct cache_entry *ce;
1164 struct stat st;
1165 int pos, len;
1166
1167 /* We do not read the cache ourselves here, because the
1168 * benchmark with my previous version that always reads cache
1169 * shows that it makes things worse for diff-tree comparing
1170 * two linux-2.6 kernel trees in an already checked out work
1171 * tree. This is because most diff-tree comparisons deal with
1172 * only a small number of files, while reading the cache is
1173 * expensive for a large project, and its cost outweighs the
1174 * savings we get by not inflating the object to a temporary
1175 * file. Practically, this code only helps when we are used
1176 * by diff-cache --cached, which does read the cache before
1177 * calling us.
1178 */
1179 if (!active_cache)
1180 return 0;
1181
1182 len = strlen(name);
1183 pos = cache_name_pos(name, len);
1184 if (pos < 0)
1185 return 0;
1186 ce = active_cache[pos];
1187 if ((lstat(name, &st) < 0) ||
1188 !S_ISREG(st.st_mode) || /* careful! */
1189 ce_match_stat(ce, &st, 0) ||
1190 hashcmp(sha1, ce->sha1))
1191 return 0;
1192 /* we return 1 only when we can stat, it is a regular file,
1193 * stat information matches, and sha1 recorded in the cache
1194 * matches. I.e. we know the file in the work tree really is
1195 * the same as the <name, sha1> pair.
1196 */
1197 return 1;
1198 }
1199
1200 static struct sha1_size_cache {
1201 unsigned char sha1[20];
1202 unsigned long size;
1203 } **sha1_size_cache;
1204 static int sha1_size_cache_nr, sha1_size_cache_alloc;
1205
1206 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
1207 int find_only,
1208 unsigned long size)
1209 {
1210 int first, last;
1211 struct sha1_size_cache *e;
1212
1213 first = 0;
1214 last = sha1_size_cache_nr;
1215 while (last > first) {
1216 int cmp, next = (last + first) >> 1;
1217 e = sha1_size_cache[next];
1218 cmp = hashcmp(e->sha1, sha1);
1219 if (!cmp)
1220 return e;
1221 if (cmp < 0) {
1222 last = next;
1223 continue;
1224 }
1225 first = next+1;
1226 }
1227 /* not found */
1228 if (find_only)
1229 return NULL;
1230 /* insert to make it at "first" */
1231 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1232 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1233 sha1_size_cache = xrealloc(sha1_size_cache,
1234 sha1_size_cache_alloc *
1235 sizeof(*sha1_size_cache));
1236 }
1237 sha1_size_cache_nr++;
1238 if (first < sha1_size_cache_nr)
1239 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1240 (sha1_size_cache_nr - first - 1) *
1241 sizeof(*sha1_size_cache));
1242 e = xmalloc(sizeof(struct sha1_size_cache));
1243 sha1_size_cache[first] = e;
1244 hashcpy(e->sha1, sha1);
1245 e->size = size;
1246 return e;
1247 }
1248
1249 /*
1250 * While doing rename detection and pickaxe operation, we may need to
1251 * grab the data for the blob (or file) for our own in-core comparison.
1252 * diff_filespec has data and size fields for this purpose.
1253 */
1254 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1255 {
1256 int err = 0;
1257 if (!DIFF_FILE_VALID(s))
1258 die("internal error: asking to populate invalid file.");
1259 if (S_ISDIR(s->mode))
1260 return -1;
1261
1262 if (!use_size_cache)
1263 size_only = 0;
1264
1265 if (s->data)
1266 return err;
1267 if (!s->sha1_valid ||
1268 work_tree_matches(s->path, s->sha1)) {
1269 struct stat st;
1270 int fd;
1271 if (lstat(s->path, &st) < 0) {
1272 if (errno == ENOENT) {
1273 err_empty:
1274 err = -1;
1275 empty:
1276 s->data = (char *)"";
1277 s->size = 0;
1278 return err;
1279 }
1280 }
1281 s->size = st.st_size;
1282 if (!s->size)
1283 goto empty;
1284 if (size_only)
1285 return 0;
1286 if (S_ISLNK(st.st_mode)) {
1287 int ret;
1288 s->data = xmalloc(s->size);
1289 s->should_free = 1;
1290 ret = readlink(s->path, s->data, s->size);
1291 if (ret < 0) {
1292 free(s->data);
1293 goto err_empty;
1294 }
1295 return 0;
1296 }
1297 fd = open(s->path, O_RDONLY);
1298 if (fd < 0)
1299 goto err_empty;
1300 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1301 close(fd);
1302 if (s->data == MAP_FAILED)
1303 goto err_empty;
1304 s->should_munmap = 1;
1305 }
1306 else {
1307 char type[20];
1308 struct sha1_size_cache *e;
1309
1310 if (size_only) {
1311 e = locate_size_cache(s->sha1, 1, 0);
1312 if (e) {
1313 s->size = e->size;
1314 return 0;
1315 }
1316 if (!sha1_object_info(s->sha1, type, &s->size))
1317 locate_size_cache(s->sha1, 0, s->size);
1318 }
1319 else {
1320 s->data = read_sha1_file(s->sha1, type, &s->size);
1321 s->should_free = 1;
1322 }
1323 }
1324 return 0;
1325 }
1326
1327 void diff_free_filespec_data(struct diff_filespec *s)
1328 {
1329 if (s->should_free)
1330 free(s->data);
1331 else if (s->should_munmap)
1332 munmap(s->data, s->size);
1333 s->should_free = s->should_munmap = 0;
1334 s->data = NULL;
1335 free(s->cnt_data);
1336 s->cnt_data = NULL;
1337 }
1338
1339 static void prep_temp_blob(struct diff_tempfile *temp,
1340 void *blob,
1341 unsigned long size,
1342 const unsigned char *sha1,
1343 int mode)
1344 {
1345 int fd;
1346
1347 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1348 if (fd < 0)
1349 die("unable to create temp-file");
1350 if (write(fd, blob, size) != size)
1351 die("unable to write temp-file");
1352 close(fd);
1353 temp->name = temp->tmp_path;
1354 strcpy(temp->hex, sha1_to_hex(sha1));
1355 temp->hex[40] = 0;
1356 sprintf(temp->mode, "%06o", mode);
1357 }
1358
1359 static void prepare_temp_file(const char *name,
1360 struct diff_tempfile *temp,
1361 struct diff_filespec *one)
1362 {
1363 if (!DIFF_FILE_VALID(one)) {
1364 not_a_valid_file:
1365 /* A '-' entry produces this for file-2, and
1366 * a '+' entry produces this for file-1.
1367 */
1368 temp->name = "/dev/null";
1369 strcpy(temp->hex, ".");
1370 strcpy(temp->mode, ".");
1371 return;
1372 }
1373
1374 if (!one->sha1_valid ||
1375 work_tree_matches(name, one->sha1)) {
1376 struct stat st;
1377 if (lstat(name, &st) < 0) {
1378 if (errno == ENOENT)
1379 goto not_a_valid_file;
1380 die("stat(%s): %s", name, strerror(errno));
1381 }
1382 if (S_ISLNK(st.st_mode)) {
1383 int ret;
1384 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1385 if (sizeof(buf) <= st.st_size)
1386 die("symlink too long: %s", name);
1387 ret = readlink(name, buf, st.st_size);
1388 if (ret < 0)
1389 die("readlink(%s)", name);
1390 prep_temp_blob(temp, buf, st.st_size,
1391 (one->sha1_valid ?
1392 one->sha1 : null_sha1),
1393 (one->sha1_valid ?
1394 one->mode : S_IFLNK));
1395 }
1396 else {
1397 /* we can borrow from the file in the work tree */
1398 temp->name = name;
1399 if (!one->sha1_valid)
1400 strcpy(temp->hex, sha1_to_hex(null_sha1));
1401 else
1402 strcpy(temp->hex, sha1_to_hex(one->sha1));
1403 /* Even though we may sometimes borrow the
1404 * contents from the work tree, we always want
1405 * one->mode. mode is trustworthy even when
1406 * !(one->sha1_valid), as long as
1407 * DIFF_FILE_VALID(one).
1408 */
1409 sprintf(temp->mode, "%06o", one->mode);
1410 }
1411 return;
1412 }
1413 else {
1414 if (diff_populate_filespec(one, 0))
1415 die("cannot read data blob for %s", one->path);
1416 prep_temp_blob(temp, one->data, one->size,
1417 one->sha1, one->mode);
1418 }
1419 }
1420
1421 static void remove_tempfile(void)
1422 {
1423 int i;
1424
1425 for (i = 0; i < 2; i++)
1426 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1427 unlink(diff_temp[i].name);
1428 diff_temp[i].name = NULL;
1429 }
1430 }
1431
1432 static void remove_tempfile_on_signal(int signo)
1433 {
1434 remove_tempfile();
1435 signal(SIGINT, SIG_DFL);
1436 raise(signo);
1437 }
1438
1439 static int spawn_prog(const char *pgm, const char **arg)
1440 {
1441 pid_t pid;
1442 int status;
1443
1444 fflush(NULL);
1445 pid = fork();
1446 if (pid < 0)
1447 die("unable to fork");
1448 if (!pid) {
1449 execvp(pgm, (char *const*) arg);
1450 exit(255);
1451 }
1452
1453 while (waitpid(pid, &status, 0) < 0) {
1454 if (errno == EINTR)
1455 continue;
1456 return -1;
1457 }
1458
1459 /* Earlier we did not check the exit status because
1460 * diff exits non-zero if files are different, and
1461 * we are not interested in knowing that. It was a
1462 * mistake which made it harder to quit a diff-*
1463 * session that uses the git-apply-patch-script as
1464 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1465 * should also exit non-zero only when it wants to
1466 * abort the entire diff-* session.
1467 */
1468 if (WIFEXITED(status) && !WEXITSTATUS(status))
1469 return 0;
1470 return -1;
1471 }
1472
1473 /* An external diff command takes:
1474 *
1475 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1476 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1477 *
1478 */
1479 static void run_external_diff(const char *pgm,
1480 const char *name,
1481 const char *other,
1482 struct diff_filespec *one,
1483 struct diff_filespec *two,
1484 const char *xfrm_msg,
1485 int complete_rewrite)
1486 {
1487 const char *spawn_arg[10];
1488 struct diff_tempfile *temp = diff_temp;
1489 int retval;
1490 static int atexit_asked = 0;
1491 const char *othername;
1492 const char **arg = &spawn_arg[0];
1493
1494 othername = (other? other : name);
1495 if (one && two) {
1496 prepare_temp_file(name, &temp[0], one);
1497 prepare_temp_file(othername, &temp[1], two);
1498 if (! atexit_asked &&
1499 (temp[0].name == temp[0].tmp_path ||
1500 temp[1].name == temp[1].tmp_path)) {
1501 atexit_asked = 1;
1502 atexit(remove_tempfile);
1503 }
1504 signal(SIGINT, remove_tempfile_on_signal);
1505 }
1506
1507 if (one && two) {
1508 *arg++ = pgm;
1509 *arg++ = name;
1510 *arg++ = temp[0].name;
1511 *arg++ = temp[0].hex;
1512 *arg++ = temp[0].mode;
1513 *arg++ = temp[1].name;
1514 *arg++ = temp[1].hex;
1515 *arg++ = temp[1].mode;
1516 if (other) {
1517 *arg++ = other;
1518 *arg++ = xfrm_msg;
1519 }
1520 } else {
1521 *arg++ = pgm;
1522 *arg++ = name;
1523 }
1524 *arg = NULL;
1525 retval = spawn_prog(pgm, spawn_arg);
1526 remove_tempfile();
1527 if (retval) {
1528 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1529 exit(1);
1530 }
1531 }
1532
1533 static void run_diff_cmd(const char *pgm,
1534 const char *name,
1535 const char *other,
1536 struct diff_filespec *one,
1537 struct diff_filespec *two,
1538 const char *xfrm_msg,
1539 struct diff_options *o,
1540 int complete_rewrite)
1541 {
1542 if (pgm) {
1543 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1544 complete_rewrite);
1545 return;
1546 }
1547 if (one && two)
1548 builtin_diff(name, other ? other : name,
1549 one, two, xfrm_msg, o, complete_rewrite);
1550 else
1551 printf("* Unmerged path %s\n", name);
1552 }
1553
1554 static void diff_fill_sha1_info(struct diff_filespec *one)
1555 {
1556 if (DIFF_FILE_VALID(one)) {
1557 if (!one->sha1_valid) {
1558 struct stat st;
1559 if (lstat(one->path, &st) < 0)
1560 die("stat %s", one->path);
1561 if (index_path(one->sha1, one->path, &st, 0))
1562 die("cannot hash %s\n", one->path);
1563 }
1564 }
1565 else
1566 hashclr(one->sha1);
1567 }
1568
1569 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1570 {
1571 const char *pgm = external_diff();
1572 char msg[PATH_MAX*2+300], *xfrm_msg;
1573 struct diff_filespec *one;
1574 struct diff_filespec *two;
1575 const char *name;
1576 const char *other;
1577 char *name_munged, *other_munged;
1578 int complete_rewrite = 0;
1579 int len;
1580
1581 if (DIFF_PAIR_UNMERGED(p)) {
1582 /* unmerged */
1583 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1584 return;
1585 }
1586
1587 name = p->one->path;
1588 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1589 name_munged = quote_one(name);
1590 other_munged = quote_one(other);
1591 one = p->one; two = p->two;
1592
1593 diff_fill_sha1_info(one);
1594 diff_fill_sha1_info(two);
1595
1596 len = 0;
1597 switch (p->status) {
1598 case DIFF_STATUS_COPIED:
1599 len += snprintf(msg + len, sizeof(msg) - len,
1600 "similarity index %d%%\n"
1601 "copy from %s\n"
1602 "copy to %s\n",
1603 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1604 name_munged, other_munged);
1605 break;
1606 case DIFF_STATUS_RENAMED:
1607 len += snprintf(msg + len, sizeof(msg) - len,
1608 "similarity index %d%%\n"
1609 "rename from %s\n"
1610 "rename to %s\n",
1611 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1612 name_munged, other_munged);
1613 break;
1614 case DIFF_STATUS_MODIFIED:
1615 if (p->score) {
1616 len += snprintf(msg + len, sizeof(msg) - len,
1617 "dissimilarity index %d%%\n",
1618 (int)(0.5 + p->score *
1619 100.0/MAX_SCORE));
1620 complete_rewrite = 1;
1621 break;
1622 }
1623 /* fallthru */
1624 default:
1625 /* nothing */
1626 ;
1627 }
1628
1629 if (hashcmp(one->sha1, two->sha1)) {
1630 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1631
1632 if (o->binary) {
1633 mmfile_t mf;
1634 if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
1635 (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
1636 abbrev = 40;
1637 }
1638 len += snprintf(msg + len, sizeof(msg) - len,
1639 "index %.*s..%.*s",
1640 abbrev, sha1_to_hex(one->sha1),
1641 abbrev, sha1_to_hex(two->sha1));
1642 if (one->mode == two->mode)
1643 len += snprintf(msg + len, sizeof(msg) - len,
1644 " %06o", one->mode);
1645 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1646 }
1647
1648 if (len)
1649 msg[--len] = 0;
1650 xfrm_msg = len ? msg : NULL;
1651
1652 if (!pgm &&
1653 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1654 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1655 /* a filepair that changes between file and symlink
1656 * needs to be split into deletion and creation.
1657 */
1658 struct diff_filespec *null = alloc_filespec(two->path);
1659 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1660 free(null);
1661 null = alloc_filespec(one->path);
1662 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1663 free(null);
1664 }
1665 else
1666 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1667 complete_rewrite);
1668
1669 free(name_munged);
1670 free(other_munged);
1671 }
1672
1673 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1674 struct diffstat_t *diffstat)
1675 {
1676 const char *name;
1677 const char *other;
1678 int complete_rewrite = 0;
1679
1680 if (DIFF_PAIR_UNMERGED(p)) {
1681 /* unmerged */
1682 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1683 return;
1684 }
1685
1686 name = p->one->path;
1687 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1688
1689 diff_fill_sha1_info(p->one);
1690 diff_fill_sha1_info(p->two);
1691
1692 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1693 complete_rewrite = 1;
1694 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1695 }
1696
1697 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1698 {
1699 const char *name;
1700 const char *other;
1701
1702 if (DIFF_PAIR_UNMERGED(p)) {
1703 /* unmerged */
1704 return;
1705 }
1706
1707 name = p->one->path;
1708 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1709
1710 diff_fill_sha1_info(p->one);
1711 diff_fill_sha1_info(p->two);
1712
1713 builtin_checkdiff(name, other, p->one, p->two);
1714 }
1715
1716 void diff_setup(struct diff_options *options)
1717 {
1718 memset(options, 0, sizeof(*options));
1719 options->line_termination = '\n';
1720 options->break_opt = -1;
1721 options->rename_limit = -1;
1722 options->context = 3;
1723 options->msg_sep = "";
1724
1725 options->change = diff_change;
1726 options->add_remove = diff_addremove;
1727 options->color_diff = diff_use_color_default;
1728 options->detect_rename = diff_detect_rename_default;
1729 }
1730
1731 int diff_setup_done(struct diff_options *options)
1732 {
1733 int count = 0;
1734
1735 if (options->output_format & DIFF_FORMAT_NAME)
1736 count++;
1737 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
1738 count++;
1739 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
1740 count++;
1741 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
1742 count++;
1743 if (count > 1)
1744 die("--name-only, --name-status, --check and -s are mutually exclusive");
1745
1746 if (options->find_copies_harder)
1747 options->detect_rename = DIFF_DETECT_COPY;
1748
1749 if (options->output_format & (DIFF_FORMAT_NAME |
1750 DIFF_FORMAT_NAME_STATUS |
1751 DIFF_FORMAT_CHECKDIFF |
1752 DIFF_FORMAT_NO_OUTPUT))
1753 options->output_format &= ~(DIFF_FORMAT_RAW |
1754 DIFF_FORMAT_NUMSTAT |
1755 DIFF_FORMAT_DIFFSTAT |
1756 DIFF_FORMAT_SUMMARY |
1757 DIFF_FORMAT_PATCH);
1758
1759 /*
1760 * These cases always need recursive; we do not drop caller-supplied
1761 * recursive bits for other formats here.
1762 */
1763 if (options->output_format & (DIFF_FORMAT_PATCH |
1764 DIFF_FORMAT_NUMSTAT |
1765 DIFF_FORMAT_DIFFSTAT |
1766 DIFF_FORMAT_SUMMARY |
1767 DIFF_FORMAT_CHECKDIFF))
1768 options->recursive = 1;
1769 /*
1770 * Also pickaxe would not work very well if you do not say recursive
1771 */
1772 if (options->pickaxe)
1773 options->recursive = 1;
1774
1775 if (options->detect_rename && options->rename_limit < 0)
1776 options->rename_limit = diff_rename_limit_default;
1777 if (options->setup & DIFF_SETUP_USE_CACHE) {
1778 if (!active_cache)
1779 /* read-cache does not die even when it fails
1780 * so it is safe for us to do this here. Also
1781 * it does not smudge active_cache or active_nr
1782 * when it fails, so we do not have to worry about
1783 * cleaning it up ourselves either.
1784 */
1785 read_cache();
1786 }
1787 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1788 use_size_cache = 1;
1789 if (options->abbrev <= 0 || 40 < options->abbrev)
1790 options->abbrev = 40; /* full */
1791
1792 return 0;
1793 }
1794
1795 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1796 {
1797 char c, *eq;
1798 int len;
1799
1800 if (*arg != '-')
1801 return 0;
1802 c = *++arg;
1803 if (!c)
1804 return 0;
1805 if (c == arg_short) {
1806 c = *++arg;
1807 if (!c)
1808 return 1;
1809 if (val && isdigit(c)) {
1810 char *end;
1811 int n = strtoul(arg, &end, 10);
1812 if (*end)
1813 return 0;
1814 *val = n;
1815 return 1;
1816 }
1817 return 0;
1818 }
1819 if (c != '-')
1820 return 0;
1821 arg++;
1822 eq = strchr(arg, '=');
1823 if (eq)
1824 len = eq - arg;
1825 else
1826 len = strlen(arg);
1827 if (!len || strncmp(arg, arg_long, len))
1828 return 0;
1829 if (eq) {
1830 int n;
1831 char *end;
1832 if (!isdigit(*++eq))
1833 return 0;
1834 n = strtoul(eq, &end, 10);
1835 if (*end)
1836 return 0;
1837 *val = n;
1838 }
1839 return 1;
1840 }
1841
1842 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1843 {
1844 const char *arg = av[0];
1845 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1846 options->output_format |= DIFF_FORMAT_PATCH;
1847 else if (opt_arg(arg, 'U', "unified", &options->context))
1848 options->output_format |= DIFF_FORMAT_PATCH;
1849 else if (!strcmp(arg, "--raw"))
1850 options->output_format |= DIFF_FORMAT_RAW;
1851 else if (!strcmp(arg, "--patch-with-raw")) {
1852 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1853 }
1854 else if (!strcmp(arg, "--numstat")) {
1855 options->output_format |= DIFF_FORMAT_NUMSTAT;
1856 }
1857 else if (!strncmp(arg, "--stat", 6)) {
1858 char *end;
1859 int width = options->stat_width;
1860 int name_width = options->stat_name_width;
1861 arg += 6;
1862 end = (char *)arg;
1863
1864 switch (*arg) {
1865 case '-':
1866 if (!strncmp(arg, "-width=", 7))
1867 width = strtoul(arg + 7, &end, 10);
1868 else if (!strncmp(arg, "-name-width=", 12))
1869 name_width = strtoul(arg + 12, &end, 10);
1870 break;
1871 case '=':
1872 width = strtoul(arg+1, &end, 10);
1873 if (*end == ',')
1874 name_width = strtoul(end+1, &end, 10);
1875 }
1876
1877 /* Important! This checks all the error cases! */
1878 if (*end)
1879 return 0;
1880 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1881 options->stat_name_width = name_width;
1882 options->stat_width = width;
1883 }
1884 else if (!strcmp(arg, "--check"))
1885 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1886 else if (!strcmp(arg, "--summary"))
1887 options->output_format |= DIFF_FORMAT_SUMMARY;
1888 else if (!strcmp(arg, "--patch-with-stat")) {
1889 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1890 }
1891 else if (!strcmp(arg, "-z"))
1892 options->line_termination = 0;
1893 else if (!strncmp(arg, "-l", 2))
1894 options->rename_limit = strtoul(arg+2, NULL, 10);
1895 else if (!strcmp(arg, "--full-index"))
1896 options->full_index = 1;
1897 else if (!strcmp(arg, "--binary")) {
1898 options->output_format |= DIFF_FORMAT_PATCH;
1899 options->binary = 1;
1900 }
1901 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
1902 options->text = 1;
1903 }
1904 else if (!strcmp(arg, "--name-only"))
1905 options->output_format |= DIFF_FORMAT_NAME;
1906 else if (!strcmp(arg, "--name-status"))
1907 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1908 else if (!strcmp(arg, "-R"))
1909 options->reverse_diff = 1;
1910 else if (!strncmp(arg, "-S", 2))
1911 options->pickaxe = arg + 2;
1912 else if (!strcmp(arg, "-s")) {
1913 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1914 }
1915 else if (!strncmp(arg, "-O", 2))
1916 options->orderfile = arg + 2;
1917 else if (!strncmp(arg, "--diff-filter=", 14))
1918 options->filter = arg + 14;
1919 else if (!strcmp(arg, "--pickaxe-all"))
1920 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1921 else if (!strcmp(arg, "--pickaxe-regex"))
1922 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1923 else if (!strncmp(arg, "-B", 2)) {
1924 if ((options->break_opt =
1925 diff_scoreopt_parse(arg)) == -1)
1926 return -1;
1927 }
1928 else if (!strncmp(arg, "-M", 2)) {
1929 if ((options->rename_score =
1930 diff_scoreopt_parse(arg)) == -1)
1931 return -1;
1932 options->detect_rename = DIFF_DETECT_RENAME;
1933 }
1934 else if (!strncmp(arg, "-C", 2)) {
1935 if ((options->rename_score =
1936 diff_scoreopt_parse(arg)) == -1)
1937 return -1;
1938 options->detect_rename = DIFF_DETECT_COPY;
1939 }
1940 else if (!strcmp(arg, "--find-copies-harder"))
1941 options->find_copies_harder = 1;
1942 else if (!strcmp(arg, "--abbrev"))
1943 options->abbrev = DEFAULT_ABBREV;
1944 else if (!strncmp(arg, "--abbrev=", 9)) {
1945 options->abbrev = strtoul(arg + 9, NULL, 10);
1946 if (options->abbrev < MINIMUM_ABBREV)
1947 options->abbrev = MINIMUM_ABBREV;
1948 else if (40 < options->abbrev)
1949 options->abbrev = 40;
1950 }
1951 else if (!strcmp(arg, "--color"))
1952 options->color_diff = 1;
1953 else if (!strcmp(arg, "--no-color"))
1954 options->color_diff = 0;
1955 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
1956 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
1957 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
1958 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
1959 else if (!strcmp(arg, "--color-words"))
1960 options->color_diff = options->color_diff_words = 1;
1961 else if (!strcmp(arg, "--no-renames"))
1962 options->detect_rename = 0;
1963 else
1964 return 0;
1965 return 1;
1966 }
1967
1968 static int parse_num(const char **cp_p)
1969 {
1970 unsigned long num, scale;
1971 int ch, dot;
1972 const char *cp = *cp_p;
1973
1974 num = 0;
1975 scale = 1;
1976 dot = 0;
1977 for(;;) {
1978 ch = *cp;
1979 if ( !dot && ch == '.' ) {
1980 scale = 1;
1981 dot = 1;
1982 } else if ( ch == '%' ) {
1983 scale = dot ? scale*100 : 100;
1984 cp++; /* % is always at the end */
1985 break;
1986 } else if ( ch >= '0' && ch <= '9' ) {
1987 if ( scale < 100000 ) {
1988 scale *= 10;
1989 num = (num*10) + (ch-'0');
1990 }
1991 } else {
1992 break;
1993 }
1994 cp++;
1995 }
1996 *cp_p = cp;
1997
1998 /* user says num divided by scale and we say internally that
1999 * is MAX_SCORE * num / scale.
2000 */
2001 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
2002 }
2003
2004 int diff_scoreopt_parse(const char *opt)
2005 {
2006 int opt1, opt2, cmd;
2007
2008 if (*opt++ != '-')
2009 return -1;
2010 cmd = *opt++;
2011 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2012 return -1; /* that is not a -M, -C nor -B option */
2013
2014 opt1 = parse_num(&opt);
2015 if (cmd != 'B')
2016 opt2 = 0;
2017 else {
2018 if (*opt == 0)
2019 opt2 = 0;
2020 else if (*opt != '/')
2021 return -1; /* we expect -B80/99 or -B80 */
2022 else {
2023 opt++;
2024 opt2 = parse_num(&opt);
2025 }
2026 }
2027 if (*opt != 0)
2028 return -1;
2029 return opt1 | (opt2 << 16);
2030 }
2031
2032 struct diff_queue_struct diff_queued_diff;
2033
2034 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2035 {
2036 if (queue->alloc <= queue->nr) {
2037 queue->alloc = alloc_nr(queue->alloc);
2038 queue->queue = xrealloc(queue->queue,
2039 sizeof(dp) * queue->alloc);
2040 }
2041 queue->queue[queue->nr++] = dp;
2042 }
2043
2044 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2045 struct diff_filespec *one,
2046 struct diff_filespec *two)
2047 {
2048 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2049 dp->one = one;
2050 dp->two = two;
2051 if (queue)
2052 diff_q(queue, dp);
2053 return dp;
2054 }
2055
2056 void diff_free_filepair(struct diff_filepair *p)
2057 {
2058 diff_free_filespec_data(p->one);
2059 diff_free_filespec_data(p->two);
2060 free(p->one);
2061 free(p->two);
2062 free(p);
2063 }
2064
2065 /* This is different from find_unique_abbrev() in that
2066 * it stuffs the result with dots for alignment.
2067 */
2068 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2069 {
2070 int abblen;
2071 const char *abbrev;
2072 if (len == 40)
2073 return sha1_to_hex(sha1);
2074
2075 abbrev = find_unique_abbrev(sha1, len);
2076 if (!abbrev)
2077 return sha1_to_hex(sha1);
2078 abblen = strlen(abbrev);
2079 if (abblen < 37) {
2080 static char hex[41];
2081 if (len < abblen && abblen <= len + 2)
2082 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2083 else
2084 sprintf(hex, "%s...", abbrev);
2085 return hex;
2086 }
2087 return sha1_to_hex(sha1);
2088 }
2089
2090 static void diff_flush_raw(struct diff_filepair *p,
2091 struct diff_options *options)
2092 {
2093 int two_paths;
2094 char status[10];
2095 int abbrev = options->abbrev;
2096 const char *path_one, *path_two;
2097 int inter_name_termination = '\t';
2098 int line_termination = options->line_termination;
2099
2100 if (!line_termination)
2101 inter_name_termination = 0;
2102
2103 path_one = p->one->path;
2104 path_two = p->two->path;
2105 if (line_termination) {
2106 path_one = quote_one(path_one);
2107 path_two = quote_one(path_two);
2108 }
2109
2110 if (p->score)
2111 sprintf(status, "%c%03d", p->status,
2112 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2113 else {
2114 status[0] = p->status;
2115 status[1] = 0;
2116 }
2117 switch (p->status) {
2118 case DIFF_STATUS_COPIED:
2119 case DIFF_STATUS_RENAMED:
2120 two_paths = 1;
2121 break;
2122 case DIFF_STATUS_ADDED:
2123 case DIFF_STATUS_DELETED:
2124 two_paths = 0;
2125 break;
2126 default:
2127 two_paths = 0;
2128 break;
2129 }
2130 if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2131 printf(":%06o %06o %s ",
2132 p->one->mode, p->two->mode,
2133 diff_unique_abbrev(p->one->sha1, abbrev));
2134 printf("%s ",
2135 diff_unique_abbrev(p->two->sha1, abbrev));
2136 }
2137 printf("%s%c%s", status, inter_name_termination, path_one);
2138 if (two_paths)
2139 printf("%c%s", inter_name_termination, path_two);
2140 putchar(line_termination);
2141 if (path_one != p->one->path)
2142 free((void*)path_one);
2143 if (path_two != p->two->path)
2144 free((void*)path_two);
2145 }
2146
2147 static void diff_flush_name(struct diff_filepair *p, int line_termination)
2148 {
2149 char *path = p->two->path;
2150
2151 if (line_termination)
2152 path = quote_one(p->two->path);
2153 printf("%s%c", path, line_termination);
2154 if (p->two->path != path)
2155 free(path);
2156 }
2157
2158 int diff_unmodified_pair(struct diff_filepair *p)
2159 {
2160 /* This function is written stricter than necessary to support
2161 * the currently implemented transformers, but the idea is to
2162 * let transformers to produce diff_filepairs any way they want,
2163 * and filter and clean them up here before producing the output.
2164 */
2165 struct diff_filespec *one, *two;
2166
2167 if (DIFF_PAIR_UNMERGED(p))
2168 return 0; /* unmerged is interesting */
2169
2170 one = p->one;
2171 two = p->two;
2172
2173 /* deletion, addition, mode or type change
2174 * and rename are all interesting.
2175 */
2176 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2177 DIFF_PAIR_MODE_CHANGED(p) ||
2178 strcmp(one->path, two->path))
2179 return 0;
2180
2181 /* both are valid and point at the same path. that is, we are
2182 * dealing with a change.
2183 */
2184 if (one->sha1_valid && two->sha1_valid &&
2185 !hashcmp(one->sha1, two->sha1))
2186 return 1; /* no change */
2187 if (!one->sha1_valid && !two->sha1_valid)
2188 return 1; /* both look at the same file on the filesystem. */
2189 return 0;
2190 }
2191
2192 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2193 {
2194 if (diff_unmodified_pair(p))
2195 return;
2196
2197 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2198 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2199 return; /* no tree diffs in patch format */
2200
2201 run_diff(p, o);
2202 }
2203
2204 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2205 struct diffstat_t *diffstat)
2206 {
2207 if (diff_unmodified_pair(p))
2208 return;
2209
2210 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2211 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2212 return; /* no tree diffs in patch format */
2213
2214 run_diffstat(p, o, diffstat);
2215 }
2216
2217 static void diff_flush_checkdiff(struct diff_filepair *p,
2218 struct diff_options *o)
2219 {
2220 if (diff_unmodified_pair(p))
2221 return;
2222
2223 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2224 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2225 return; /* no tree diffs in patch format */
2226
2227 run_checkdiff(p, o);
2228 }
2229
2230 int diff_queue_is_empty(void)
2231 {
2232 struct diff_queue_struct *q = &diff_queued_diff;
2233 int i;
2234 for (i = 0; i < q->nr; i++)
2235 if (!diff_unmodified_pair(q->queue[i]))
2236 return 0;
2237 return 1;
2238 }
2239
2240 #if DIFF_DEBUG
2241 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2242 {
2243 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2244 x, one ? one : "",
2245 s->path,
2246 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2247 s->mode,
2248 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2249 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2250 x, one ? one : "",
2251 s->size, s->xfrm_flags);
2252 }
2253
2254 void diff_debug_filepair(const struct diff_filepair *p, int i)
2255 {
2256 diff_debug_filespec(p->one, i, "one");
2257 diff_debug_filespec(p->two, i, "two");
2258 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2259 p->score, p->status ? p->status : '?',
2260 p->source_stays, p->broken_pair);
2261 }
2262
2263 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2264 {
2265 int i;
2266 if (msg)
2267 fprintf(stderr, "%s\n", msg);
2268 fprintf(stderr, "q->nr = %d\n", q->nr);
2269 for (i = 0; i < q->nr; i++) {
2270 struct diff_filepair *p = q->queue[i];
2271 diff_debug_filepair(p, i);
2272 }
2273 }
2274 #endif
2275
2276 static void diff_resolve_rename_copy(void)
2277 {
2278 int i, j;
2279 struct diff_filepair *p, *pp;
2280 struct diff_queue_struct *q = &diff_queued_diff;
2281
2282 diff_debug_queue("resolve-rename-copy", q);
2283
2284 for (i = 0; i < q->nr; i++) {
2285 p = q->queue[i];
2286 p->status = 0; /* undecided */
2287 if (DIFF_PAIR_UNMERGED(p))
2288 p->status = DIFF_STATUS_UNMERGED;
2289 else if (!DIFF_FILE_VALID(p->one))
2290 p->status = DIFF_STATUS_ADDED;
2291 else if (!DIFF_FILE_VALID(p->two))
2292 p->status = DIFF_STATUS_DELETED;
2293 else if (DIFF_PAIR_TYPE_CHANGED(p))
2294 p->status = DIFF_STATUS_TYPE_CHANGED;
2295
2296 /* from this point on, we are dealing with a pair
2297 * whose both sides are valid and of the same type, i.e.
2298 * either in-place edit or rename/copy edit.
2299 */
2300 else if (DIFF_PAIR_RENAME(p)) {
2301 if (p->source_stays) {
2302 p->status = DIFF_STATUS_COPIED;
2303 continue;
2304 }
2305 /* See if there is some other filepair that
2306 * copies from the same source as us. If so
2307 * we are a copy. Otherwise we are either a
2308 * copy if the path stays, or a rename if it
2309 * does not, but we already handled "stays" case.
2310 */
2311 for (j = i + 1; j < q->nr; j++) {
2312 pp = q->queue[j];
2313 if (strcmp(pp->one->path, p->one->path))
2314 continue; /* not us */
2315 if (!DIFF_PAIR_RENAME(pp))
2316 continue; /* not a rename/copy */
2317 /* pp is a rename/copy from the same source */
2318 p->status = DIFF_STATUS_COPIED;
2319 break;
2320 }
2321 if (!p->status)
2322 p->status = DIFF_STATUS_RENAMED;
2323 }
2324 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2325 p->one->mode != p->two->mode)
2326 p->status = DIFF_STATUS_MODIFIED;
2327 else {
2328 /* This is a "no-change" entry and should not
2329 * happen anymore, but prepare for broken callers.
2330 */
2331 error("feeding unmodified %s to diffcore",
2332 p->one->path);
2333 p->status = DIFF_STATUS_UNKNOWN;
2334 }
2335 }
2336 diff_debug_queue("resolve-rename-copy done", q);
2337 }
2338
2339 static int check_pair_status(struct diff_filepair *p)
2340 {
2341 switch (p->status) {
2342 case DIFF_STATUS_UNKNOWN:
2343 return 0;
2344 case 0:
2345 die("internal error in diff-resolve-rename-copy");
2346 default:
2347 return 1;
2348 }
2349 }
2350
2351 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2352 {
2353 int fmt = opt->output_format;
2354
2355 if (fmt & DIFF_FORMAT_CHECKDIFF)
2356 diff_flush_checkdiff(p, opt);
2357 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2358 diff_flush_raw(p, opt);
2359 else if (fmt & DIFF_FORMAT_NAME)
2360 diff_flush_name(p, opt->line_termination);
2361 }
2362
2363 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2364 {
2365 if (fs->mode)
2366 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2367 else
2368 printf(" %s %s\n", newdelete, fs->path);
2369 }
2370
2371
2372 static void show_mode_change(struct diff_filepair *p, int show_name)
2373 {
2374 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2375 if (show_name)
2376 printf(" mode change %06o => %06o %s\n",
2377 p->one->mode, p->two->mode, p->two->path);
2378 else
2379 printf(" mode change %06o => %06o\n",
2380 p->one->mode, p->two->mode);
2381 }
2382 }
2383
2384 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2385 {
2386 const char *old, *new;
2387
2388 /* Find common prefix */
2389 old = p->one->path;
2390 new = p->two->path;
2391 while (1) {
2392 const char *slash_old, *slash_new;
2393 slash_old = strchr(old, '/');
2394 slash_new = strchr(new, '/');
2395 if (!slash_old ||
2396 !slash_new ||
2397 slash_old - old != slash_new - new ||
2398 memcmp(old, new, slash_new - new))
2399 break;
2400 old = slash_old + 1;
2401 new = slash_new + 1;
2402 }
2403 /* p->one->path thru old is the common prefix, and old and new
2404 * through the end of names are renames
2405 */
2406 if (old != p->one->path)
2407 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2408 (int)(old - p->one->path), p->one->path,
2409 old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2410 else
2411 printf(" %s %s => %s (%d%%)\n", renamecopy,
2412 p->one->path, p->two->path,
2413 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2414 show_mode_change(p, 0);
2415 }
2416
2417 static void diff_summary(struct diff_filepair *p)
2418 {
2419 switch(p->status) {
2420 case DIFF_STATUS_DELETED:
2421 show_file_mode_name("delete", p->one);
2422 break;
2423 case DIFF_STATUS_ADDED:
2424 show_file_mode_name("create", p->two);
2425 break;
2426 case DIFF_STATUS_COPIED:
2427 show_rename_copy("copy", p);
2428 break;
2429 case DIFF_STATUS_RENAMED:
2430 show_rename_copy("rename", p);
2431 break;
2432 default:
2433 if (p->score) {
2434 printf(" rewrite %s (%d%%)\n", p->two->path,
2435 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2436 show_mode_change(p, 0);
2437 } else show_mode_change(p, 1);
2438 break;
2439 }
2440 }
2441
2442 struct patch_id_t {
2443 struct xdiff_emit_state xm;
2444 SHA_CTX *ctx;
2445 int patchlen;
2446 };
2447
2448 static int remove_space(char *line, int len)
2449 {
2450 int i;
2451 char *dst = line;
2452 unsigned char c;
2453
2454 for (i = 0; i < len; i++)
2455 if (!isspace((c = line[i])))
2456 *dst++ = c;
2457
2458 return dst - line;
2459 }
2460
2461 static void patch_id_consume(void *priv, char *line, unsigned long len)
2462 {
2463 struct patch_id_t *data = priv;
2464 int new_len;
2465
2466 /* Ignore line numbers when computing the SHA1 of the patch */
2467 if (!strncmp(line, "@@ -", 4))
2468 return;
2469
2470 new_len = remove_space(line, len);
2471
2472 SHA1_Update(data->ctx, line, new_len);
2473 data->patchlen += new_len;
2474 }
2475
2476 /* returns 0 upon success, and writes result into sha1 */
2477 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2478 {
2479 struct diff_queue_struct *q = &diff_queued_diff;
2480 int i;
2481 SHA_CTX ctx;
2482 struct patch_id_t data;
2483 char buffer[PATH_MAX * 4 + 20];
2484
2485 SHA1_Init(&ctx);
2486 memset(&data, 0, sizeof(struct patch_id_t));
2487 data.ctx = &ctx;
2488 data.xm.consume = patch_id_consume;
2489
2490 for (i = 0; i < q->nr; i++) {
2491 xpparam_t xpp;
2492 xdemitconf_t xecfg;
2493 xdemitcb_t ecb;
2494 mmfile_t mf1, mf2;
2495 struct diff_filepair *p = q->queue[i];
2496 int len1, len2;
2497
2498 if (p->status == 0)
2499 return error("internal diff status error");
2500 if (p->status == DIFF_STATUS_UNKNOWN)
2501 continue;
2502 if (diff_unmodified_pair(p))
2503 continue;
2504 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2505 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2506 continue;
2507 if (DIFF_PAIR_UNMERGED(p))
2508 continue;
2509
2510 diff_fill_sha1_info(p->one);
2511 diff_fill_sha1_info(p->two);
2512 if (fill_mmfile(&mf1, p->one) < 0 ||
2513 fill_mmfile(&mf2, p->two) < 0)
2514 return error("unable to read files to diff");
2515
2516 /* Maybe hash p->two? into the patch id? */
2517 if (mmfile_is_binary(&mf2))
2518 continue;
2519
2520 len1 = remove_space(p->one->path, strlen(p->one->path));
2521 len2 = remove_space(p->two->path, strlen(p->two->path));
2522 if (p->one->mode == 0)
2523 len1 = snprintf(buffer, sizeof(buffer),
2524 "diff--gita/%.*sb/%.*s"
2525 "newfilemode%06o"
2526 "---/dev/null"
2527 "+++b/%.*s",
2528 len1, p->one->path,
2529 len2, p->two->path,
2530 p->two->mode,
2531 len2, p->two->path);
2532 else if (p->two->mode == 0)
2533 len1 = snprintf(buffer, sizeof(buffer),
2534 "diff--gita/%.*sb/%.*s"
2535 "deletedfilemode%06o"
2536 "---a/%.*s"
2537 "+++/dev/null",
2538 len1, p->one->path,
2539 len2, p->two->path,
2540 p->one->mode,
2541 len1, p->one->path);
2542 else
2543 len1 = snprintf(buffer, sizeof(buffer),
2544 "diff--gita/%.*sb/%.*s"
2545 "---a/%.*s"
2546 "+++b/%.*s",
2547 len1, p->one->path,
2548 len2, p->two->path,
2549 len1, p->one->path,
2550 len2, p->two->path);
2551 SHA1_Update(&ctx, buffer, len1);
2552
2553 xpp.flags = XDF_NEED_MINIMAL;
2554 xecfg.ctxlen = 3;
2555 xecfg.flags = XDL_EMIT_FUNCNAMES;
2556 ecb.outf = xdiff_outf;
2557 ecb.priv = &data;
2558 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2559 }
2560
2561 SHA1_Final(sha1, &ctx);
2562 return 0;
2563 }
2564
2565 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2566 {
2567 struct diff_queue_struct *q = &diff_queued_diff;
2568 int i;
2569 int result = diff_get_patch_id(options, sha1);
2570
2571 for (i = 0; i < q->nr; i++)
2572 diff_free_filepair(q->queue[i]);
2573
2574 free(q->queue);
2575 q->queue = NULL;
2576 q->nr = q->alloc = 0;
2577
2578 return result;
2579 }
2580
2581 static int is_summary_empty(const struct diff_queue_struct *q)
2582 {
2583 int i;
2584
2585 for (i = 0; i < q->nr; i++) {
2586 const struct diff_filepair *p = q->queue[i];
2587
2588 switch (p->status) {
2589 case DIFF_STATUS_DELETED:
2590 case DIFF_STATUS_ADDED:
2591 case DIFF_STATUS_COPIED:
2592 case DIFF_STATUS_RENAMED:
2593 return 0;
2594 default:
2595 if (p->score)
2596 return 0;
2597 if (p->one->mode && p->two->mode &&
2598 p->one->mode != p->two->mode)
2599 return 0;
2600 break;
2601 }
2602 }
2603 return 1;
2604 }
2605
2606 void diff_flush(struct diff_options *options)
2607 {
2608 struct diff_queue_struct *q = &diff_queued_diff;
2609 int i, output_format = options->output_format;
2610 int separator = 0;
2611
2612 /*
2613 * Order: raw, stat, summary, patch
2614 * or: name/name-status/checkdiff (other bits clear)
2615 */
2616 if (!q->nr)
2617 goto free_queue;
2618
2619 if (output_format & (DIFF_FORMAT_RAW |
2620 DIFF_FORMAT_NAME |
2621 DIFF_FORMAT_NAME_STATUS |
2622 DIFF_FORMAT_CHECKDIFF)) {
2623 for (i = 0; i < q->nr; i++) {
2624 struct diff_filepair *p = q->queue[i];
2625 if (check_pair_status(p))
2626 flush_one_pair(p, options);
2627 }
2628 separator++;
2629 }
2630
2631 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_NUMSTAT)) {
2632 struct diffstat_t diffstat;
2633
2634 memset(&diffstat, 0, sizeof(struct diffstat_t));
2635 diffstat.xm.consume = diffstat_consume;
2636 for (i = 0; i < q->nr; i++) {
2637 struct diff_filepair *p = q->queue[i];
2638 if (check_pair_status(p))
2639 diff_flush_stat(p, options, &diffstat);
2640 }
2641 if (output_format & DIFF_FORMAT_NUMSTAT)
2642 show_numstat(&diffstat, options);
2643 if (output_format & DIFF_FORMAT_DIFFSTAT)
2644 show_stats(&diffstat, options);
2645 separator++;
2646 }
2647
2648 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2649 for (i = 0; i < q->nr; i++)
2650 diff_summary(q->queue[i]);
2651 separator++;
2652 }
2653
2654 if (output_format & DIFF_FORMAT_PATCH) {
2655 if (separator) {
2656 if (options->stat_sep) {
2657 /* attach patch instead of inline */
2658 fputs(options->stat_sep, stdout);
2659 } else {
2660 putchar(options->line_termination);
2661 }
2662 }
2663
2664 for (i = 0; i < q->nr; i++) {
2665 struct diff_filepair *p = q->queue[i];
2666 if (check_pair_status(p))
2667 diff_flush_patch(p, options);
2668 }
2669 }
2670
2671 if (output_format & DIFF_FORMAT_CALLBACK)
2672 options->format_callback(q, options, options->format_callback_data);
2673
2674 for (i = 0; i < q->nr; i++)
2675 diff_free_filepair(q->queue[i]);
2676 free_queue:
2677 free(q->queue);
2678 q->queue = NULL;
2679 q->nr = q->alloc = 0;
2680 }
2681
2682 static void diffcore_apply_filter(const char *filter)
2683 {
2684 int i;
2685 struct diff_queue_struct *q = &diff_queued_diff;
2686 struct diff_queue_struct outq;
2687 outq.queue = NULL;
2688 outq.nr = outq.alloc = 0;
2689
2690 if (!filter)
2691 return;
2692
2693 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2694 int found;
2695 for (i = found = 0; !found && i < q->nr; i++) {
2696 struct diff_filepair *p = q->queue[i];
2697 if (((p->status == DIFF_STATUS_MODIFIED) &&
2698 ((p->score &&
2699 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2700 (!p->score &&
2701 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2702 ((p->status != DIFF_STATUS_MODIFIED) &&
2703 strchr(filter, p->status)))
2704 found++;
2705 }
2706 if (found)
2707 return;
2708
2709 /* otherwise we will clear the whole queue
2710 * by copying the empty outq at the end of this
2711 * function, but first clear the current entries
2712 * in the queue.
2713 */
2714 for (i = 0; i < q->nr; i++)
2715 diff_free_filepair(q->queue[i]);
2716 }
2717 else {
2718 /* Only the matching ones */
2719 for (i = 0; i < q->nr; i++) {
2720 struct diff_filepair *p = q->queue[i];
2721
2722 if (((p->status == DIFF_STATUS_MODIFIED) &&
2723 ((p->score &&
2724 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2725 (!p->score &&
2726 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2727 ((p->status != DIFF_STATUS_MODIFIED) &&
2728 strchr(filter, p->status)))
2729 diff_q(&outq, p);
2730 else
2731 diff_free_filepair(p);
2732 }
2733 }
2734 free(q->queue);
2735 *q = outq;
2736 }
2737
2738 void diffcore_std(struct diff_options *options)
2739 {
2740 if (options->break_opt != -1)
2741 diffcore_break(options->break_opt);
2742 if (options->detect_rename)
2743 diffcore_rename(options);
2744 if (options->break_opt != -1)
2745 diffcore_merge_broken();
2746 if (options->pickaxe)
2747 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2748 if (options->orderfile)
2749 diffcore_order(options->orderfile);
2750 diff_resolve_rename_copy();
2751 diffcore_apply_filter(options->filter);
2752 }
2753
2754
2755 void diffcore_std_no_resolve(struct diff_options *options)
2756 {
2757 if (options->pickaxe)
2758 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2759 if (options->orderfile)
2760 diffcore_order(options->orderfile);
2761 diffcore_apply_filter(options->filter);
2762 }
2763
2764 void diff_addremove(struct diff_options *options,
2765 int addremove, unsigned mode,
2766 const unsigned char *sha1,
2767 const char *base, const char *path)
2768 {
2769 char concatpath[PATH_MAX];
2770 struct diff_filespec *one, *two;
2771
2772 /* This may look odd, but it is a preparation for
2773 * feeding "there are unchanged files which should
2774 * not produce diffs, but when you are doing copy
2775 * detection you would need them, so here they are"
2776 * entries to the diff-core. They will be prefixed
2777 * with something like '=' or '*' (I haven't decided
2778 * which but should not make any difference).
2779 * Feeding the same new and old to diff_change()
2780 * also has the same effect.
2781 * Before the final output happens, they are pruned after
2782 * merged into rename/copy pairs as appropriate.
2783 */
2784 if (options->reverse_diff)
2785 addremove = (addremove == '+' ? '-' :
2786 addremove == '-' ? '+' : addremove);
2787
2788 if (!path) path = "";
2789 sprintf(concatpath, "%s%s", base, path);
2790 one = alloc_filespec(concatpath);
2791 two = alloc_filespec(concatpath);
2792
2793 if (addremove != '+')
2794 fill_filespec(one, sha1, mode);
2795 if (addremove != '-')
2796 fill_filespec(two, sha1, mode);
2797
2798 diff_queue(&diff_queued_diff, one, two);
2799 }
2800
2801 void diff_change(struct diff_options *options,
2802 unsigned old_mode, unsigned new_mode,
2803 const unsigned char *old_sha1,
2804 const unsigned char *new_sha1,
2805 const char *base, const char *path)
2806 {
2807 char concatpath[PATH_MAX];
2808 struct diff_filespec *one, *two;
2809
2810 if (options->reverse_diff) {
2811 unsigned tmp;
2812 const unsigned char *tmp_c;
2813 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2814 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2815 }
2816 if (!path) path = "";
2817 sprintf(concatpath, "%s%s", base, path);
2818 one = alloc_filespec(concatpath);
2819 two = alloc_filespec(concatpath);
2820 fill_filespec(one, old_sha1, old_mode);
2821 fill_filespec(two, new_sha1, new_mode);
2822
2823 diff_queue(&diff_queued_diff, one, two);
2824 }
2825
2826 void diff_unmerge(struct diff_options *options,
2827 const char *path)
2828 {
2829 struct diff_filespec *one, *two;
2830 one = alloc_filespec(path);
2831 two = alloc_filespec(path);
2832 diff_queue(&diff_queued_diff, one, two);
2833 }