diffcore-rename: somewhat optimized.
[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
12 static const char *diff_opts = "-pu";
13
14 static int use_size_cache;
15
16 int diff_rename_limit_default = -1;
17
18 int git_diff_config(const char *var, const char *value)
19 {
20 if (!strcmp(var, "diff.renamelimit")) {
21 diff_rename_limit_default = git_config_int(var, value);
22 return 0;
23 }
24
25 return git_default_config(var, value);
26 }
27
28 static char *quote_one(const char *str)
29 {
30 int needlen;
31 char *xp;
32
33 if (!str)
34 return NULL;
35 needlen = quote_c_style(str, NULL, NULL, 0);
36 if (!needlen)
37 return strdup(str);
38 xp = xmalloc(needlen + 1);
39 quote_c_style(str, xp, NULL, 0);
40 return xp;
41 }
42
43 static char *quote_two(const char *one, const char *two)
44 {
45 int need_one = quote_c_style(one, NULL, NULL, 1);
46 int need_two = quote_c_style(two, NULL, NULL, 1);
47 char *xp;
48
49 if (need_one + need_two) {
50 if (!need_one) need_one = strlen(one);
51 if (!need_two) need_one = strlen(two);
52
53 xp = xmalloc(need_one + need_two + 3);
54 xp[0] = '"';
55 quote_c_style(one, xp + 1, NULL, 1);
56 quote_c_style(two, xp + need_one + 1, NULL, 1);
57 strcpy(xp + need_one + need_two + 1, "\"");
58 return xp;
59 }
60 need_one = strlen(one);
61 need_two = strlen(two);
62 xp = xmalloc(need_one + need_two + 1);
63 strcpy(xp, one);
64 strcpy(xp + need_one, two);
65 return xp;
66 }
67
68 static const char *external_diff(void)
69 {
70 static const char *external_diff_cmd = NULL;
71 static int done_preparing = 0;
72 const char *env_diff_opts;
73
74 if (done_preparing)
75 return external_diff_cmd;
76
77 /*
78 * Default values above are meant to match the
79 * Linux kernel development style. Examples of
80 * alternative styles you can specify via environment
81 * variables are:
82 *
83 * GIT_DIFF_OPTS="-c";
84 */
85 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
86
87 /* In case external diff fails... */
88 env_diff_opts = getenv("GIT_DIFF_OPTS");
89 if (env_diff_opts) diff_opts = env_diff_opts;
90
91 done_preparing = 1;
92 return external_diff_cmd;
93 }
94
95 #define TEMPFILE_PATH_LEN 50
96
97 static struct diff_tempfile {
98 const char *name; /* filename external diff should read from */
99 char hex[41];
100 char mode[10];
101 char tmp_path[TEMPFILE_PATH_LEN];
102 } diff_temp[2];
103
104 static int count_lines(const char *filename)
105 {
106 FILE *in;
107 int count, ch, completely_empty = 1, nl_just_seen = 0;
108 in = fopen(filename, "r");
109 count = 0;
110 while ((ch = fgetc(in)) != EOF)
111 if (ch == '\n') {
112 count++;
113 nl_just_seen = 1;
114 completely_empty = 0;
115 }
116 else {
117 nl_just_seen = 0;
118 completely_empty = 0;
119 }
120 fclose(in);
121 if (completely_empty)
122 return 0;
123 if (!nl_just_seen)
124 count++; /* no trailing newline */
125 return count;
126 }
127
128 static void print_line_count(int count)
129 {
130 switch (count) {
131 case 0:
132 printf("0,0");
133 break;
134 case 1:
135 printf("1");
136 break;
137 default:
138 printf("1,%d", count);
139 break;
140 }
141 }
142
143 static void copy_file(int prefix, const char *filename)
144 {
145 FILE *in;
146 int ch, nl_just_seen = 1;
147 in = fopen(filename, "r");
148 while ((ch = fgetc(in)) != EOF) {
149 if (nl_just_seen)
150 putchar(prefix);
151 putchar(ch);
152 if (ch == '\n')
153 nl_just_seen = 1;
154 else
155 nl_just_seen = 0;
156 }
157 fclose(in);
158 if (!nl_just_seen)
159 printf("\n\\ No newline at end of file\n");
160 }
161
162 static void emit_rewrite_diff(const char *name_a,
163 const char *name_b,
164 struct diff_tempfile *temp)
165 {
166 /* Use temp[i].name as input, name_a and name_b as labels */
167 int lc_a, lc_b;
168 lc_a = count_lines(temp[0].name);
169 lc_b = count_lines(temp[1].name);
170 printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
171 print_line_count(lc_a);
172 printf(" +");
173 print_line_count(lc_b);
174 printf(" @@\n");
175 if (lc_a)
176 copy_file('-', temp[0].name);
177 if (lc_b)
178 copy_file('+', temp[1].name);
179 }
180
181 static const char *builtin_diff(const char *name_a,
182 const char *name_b,
183 struct diff_tempfile *temp,
184 const char *xfrm_msg,
185 int complete_rewrite,
186 const char **args)
187 {
188 int i, next_at, cmd_size;
189 const char *const diff_cmd = "diff -L%s -L%s";
190 const char *const diff_arg = "-- %s %s||:"; /* "||:" is to return 0 */
191 const char *input_name_sq[2];
192 const char *label_path[2];
193 char *cmd;
194
195 /* diff_cmd and diff_arg have 4 %s in total which makes
196 * the sum of these strings 8 bytes larger than required.
197 * we use 2 spaces around diff-opts, and we need to count
198 * terminating NUL; we used to subtract 5 here, but we do not
199 * care about small leaks in this subprocess that is about
200 * to exec "diff" anymore.
201 */
202 cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
203 + 128);
204
205 for (i = 0; i < 2; i++) {
206 input_name_sq[i] = sq_quote(temp[i].name);
207 if (!strcmp(temp[i].name, "/dev/null"))
208 label_path[i] = "/dev/null";
209 else if (!i)
210 label_path[i] = sq_quote(quote_two("a/", name_a));
211 else
212 label_path[i] = sq_quote(quote_two("b/", name_b));
213 cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
214 }
215
216 cmd = xmalloc(cmd_size);
217
218 next_at = 0;
219 next_at += snprintf(cmd+next_at, cmd_size-next_at,
220 diff_cmd, label_path[0], label_path[1]);
221 next_at += snprintf(cmd+next_at, cmd_size-next_at,
222 " %s ", diff_opts);
223 next_at += snprintf(cmd+next_at, cmd_size-next_at,
224 diff_arg, input_name_sq[0], input_name_sq[1]);
225
226 printf("diff --git %s %s\n",
227 quote_two("a/", name_a), quote_two("b/", name_b));
228 if (label_path[0][0] == '/') {
229 /* dev/null */
230 printf("new file mode %s\n", temp[1].mode);
231 if (xfrm_msg && xfrm_msg[0])
232 puts(xfrm_msg);
233 }
234 else if (label_path[1][0] == '/') {
235 printf("deleted file mode %s\n", temp[0].mode);
236 if (xfrm_msg && xfrm_msg[0])
237 puts(xfrm_msg);
238 }
239 else {
240 if (strcmp(temp[0].mode, temp[1].mode)) {
241 printf("old mode %s\n", temp[0].mode);
242 printf("new mode %s\n", temp[1].mode);
243 }
244 if (xfrm_msg && xfrm_msg[0])
245 puts(xfrm_msg);
246 /*
247 * we do not run diff between different kind
248 * of objects.
249 */
250 if (strncmp(temp[0].mode, temp[1].mode, 3))
251 return NULL;
252 if (complete_rewrite) {
253 emit_rewrite_diff(name_a, name_b, temp);
254 return NULL;
255 }
256 }
257
258 /* This is disgusting */
259 *args++ = "sh";
260 *args++ = "-c";
261 *args++ = cmd;
262 *args = NULL;
263 return "/bin/sh";
264 }
265
266 struct diff_filespec *alloc_filespec(const char *path)
267 {
268 int namelen = strlen(path);
269 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
270
271 memset(spec, 0, sizeof(*spec));
272 spec->path = (char *)(spec + 1);
273 memcpy(spec->path, path, namelen+1);
274 return spec;
275 }
276
277 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
278 unsigned short mode)
279 {
280 if (mode) {
281 spec->mode = DIFF_FILE_CANON_MODE(mode);
282 memcpy(spec->sha1, sha1, 20);
283 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
284 }
285 }
286
287 /*
288 * Given a name and sha1 pair, if the dircache tells us the file in
289 * the work tree has that object contents, return true, so that
290 * prepare_temp_file() does not have to inflate and extract.
291 */
292 static int work_tree_matches(const char *name, const unsigned char *sha1)
293 {
294 struct cache_entry *ce;
295 struct stat st;
296 int pos, len;
297
298 /* We do not read the cache ourselves here, because the
299 * benchmark with my previous version that always reads cache
300 * shows that it makes things worse for diff-tree comparing
301 * two linux-2.6 kernel trees in an already checked out work
302 * tree. This is because most diff-tree comparisons deal with
303 * only a small number of files, while reading the cache is
304 * expensive for a large project, and its cost outweighs the
305 * savings we get by not inflating the object to a temporary
306 * file. Practically, this code only helps when we are used
307 * by diff-cache --cached, which does read the cache before
308 * calling us.
309 */
310 if (!active_cache)
311 return 0;
312
313 len = strlen(name);
314 pos = cache_name_pos(name, len);
315 if (pos < 0)
316 return 0;
317 ce = active_cache[pos];
318 if ((lstat(name, &st) < 0) ||
319 !S_ISREG(st.st_mode) || /* careful! */
320 ce_match_stat(ce, &st, 0) ||
321 memcmp(sha1, ce->sha1, 20))
322 return 0;
323 /* we return 1 only when we can stat, it is a regular file,
324 * stat information matches, and sha1 recorded in the cache
325 * matches. I.e. we know the file in the work tree really is
326 * the same as the <name, sha1> pair.
327 */
328 return 1;
329 }
330
331 static struct sha1_size_cache {
332 unsigned char sha1[20];
333 unsigned long size;
334 } **sha1_size_cache;
335 static int sha1_size_cache_nr, sha1_size_cache_alloc;
336
337 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
338 int find_only,
339 unsigned long size)
340 {
341 int first, last;
342 struct sha1_size_cache *e;
343
344 first = 0;
345 last = sha1_size_cache_nr;
346 while (last > first) {
347 int cmp, next = (last + first) >> 1;
348 e = sha1_size_cache[next];
349 cmp = memcmp(e->sha1, sha1, 20);
350 if (!cmp)
351 return e;
352 if (cmp < 0) {
353 last = next;
354 continue;
355 }
356 first = next+1;
357 }
358 /* not found */
359 if (find_only)
360 return NULL;
361 /* insert to make it at "first" */
362 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
363 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
364 sha1_size_cache = xrealloc(sha1_size_cache,
365 sha1_size_cache_alloc *
366 sizeof(*sha1_size_cache));
367 }
368 sha1_size_cache_nr++;
369 if (first < sha1_size_cache_nr)
370 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
371 (sha1_size_cache_nr - first - 1) *
372 sizeof(*sha1_size_cache));
373 e = xmalloc(sizeof(struct sha1_size_cache));
374 sha1_size_cache[first] = e;
375 memcpy(e->sha1, sha1, 20);
376 e->size = size;
377 return e;
378 }
379
380 /*
381 * While doing rename detection and pickaxe operation, we may need to
382 * grab the data for the blob (or file) for our own in-core comparison.
383 * diff_filespec has data and size fields for this purpose.
384 */
385 int diff_populate_filespec(struct diff_filespec *s, int size_only)
386 {
387 int err = 0;
388 if (!DIFF_FILE_VALID(s))
389 die("internal error: asking to populate invalid file.");
390 if (S_ISDIR(s->mode))
391 return -1;
392
393 if (!use_size_cache)
394 size_only = 0;
395
396 if (s->data)
397 return err;
398 if (!s->sha1_valid ||
399 work_tree_matches(s->path, s->sha1)) {
400 struct stat st;
401 int fd;
402 if (lstat(s->path, &st) < 0) {
403 if (errno == ENOENT) {
404 err_empty:
405 err = -1;
406 empty:
407 s->data = "";
408 s->size = 0;
409 return err;
410 }
411 }
412 s->size = st.st_size;
413 if (!s->size)
414 goto empty;
415 if (size_only)
416 return 0;
417 if (S_ISLNK(st.st_mode)) {
418 int ret;
419 s->data = xmalloc(s->size);
420 s->should_free = 1;
421 ret = readlink(s->path, s->data, s->size);
422 if (ret < 0) {
423 free(s->data);
424 goto err_empty;
425 }
426 return 0;
427 }
428 fd = open(s->path, O_RDONLY);
429 if (fd < 0)
430 goto err_empty;
431 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
432 close(fd);
433 if (s->data == MAP_FAILED)
434 goto err_empty;
435 s->should_munmap = 1;
436 }
437 else {
438 char type[20];
439 struct sha1_size_cache *e;
440
441 if (size_only) {
442 e = locate_size_cache(s->sha1, 1, 0);
443 if (e) {
444 s->size = e->size;
445 return 0;
446 }
447 if (!sha1_object_info(s->sha1, type, &s->size))
448 locate_size_cache(s->sha1, 0, s->size);
449 }
450 else {
451 s->data = read_sha1_file(s->sha1, type, &s->size);
452 s->should_free = 1;
453 }
454 }
455 return 0;
456 }
457
458 void diff_free_filespec_data(struct diff_filespec *s)
459 {
460 if (s->should_free)
461 free(s->data);
462 else if (s->should_munmap)
463 munmap(s->data, s->size);
464 s->should_free = s->should_munmap = 0;
465 s->data = NULL;
466 free(s->cnt_data);
467 s->cnt_data = NULL;
468 }
469
470 static void prep_temp_blob(struct diff_tempfile *temp,
471 void *blob,
472 unsigned long size,
473 const unsigned char *sha1,
474 int mode)
475 {
476 int fd;
477
478 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
479 if (fd < 0)
480 die("unable to create temp-file");
481 if (write(fd, blob, size) != size)
482 die("unable to write temp-file");
483 close(fd);
484 temp->name = temp->tmp_path;
485 strcpy(temp->hex, sha1_to_hex(sha1));
486 temp->hex[40] = 0;
487 sprintf(temp->mode, "%06o", mode);
488 }
489
490 static void prepare_temp_file(const char *name,
491 struct diff_tempfile *temp,
492 struct diff_filespec *one)
493 {
494 if (!DIFF_FILE_VALID(one)) {
495 not_a_valid_file:
496 /* A '-' entry produces this for file-2, and
497 * a '+' entry produces this for file-1.
498 */
499 temp->name = "/dev/null";
500 strcpy(temp->hex, ".");
501 strcpy(temp->mode, ".");
502 return;
503 }
504
505 if (!one->sha1_valid ||
506 work_tree_matches(name, one->sha1)) {
507 struct stat st;
508 if (lstat(name, &st) < 0) {
509 if (errno == ENOENT)
510 goto not_a_valid_file;
511 die("stat(%s): %s", name, strerror(errno));
512 }
513 if (S_ISLNK(st.st_mode)) {
514 int ret;
515 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
516 if (sizeof(buf) <= st.st_size)
517 die("symlink too long: %s", name);
518 ret = readlink(name, buf, st.st_size);
519 if (ret < 0)
520 die("readlink(%s)", name);
521 prep_temp_blob(temp, buf, st.st_size,
522 (one->sha1_valid ?
523 one->sha1 : null_sha1),
524 (one->sha1_valid ?
525 one->mode : S_IFLNK));
526 }
527 else {
528 /* we can borrow from the file in the work tree */
529 temp->name = name;
530 if (!one->sha1_valid)
531 strcpy(temp->hex, sha1_to_hex(null_sha1));
532 else
533 strcpy(temp->hex, sha1_to_hex(one->sha1));
534 /* Even though we may sometimes borrow the
535 * contents from the work tree, we always want
536 * one->mode. mode is trustworthy even when
537 * !(one->sha1_valid), as long as
538 * DIFF_FILE_VALID(one).
539 */
540 sprintf(temp->mode, "%06o", one->mode);
541 }
542 return;
543 }
544 else {
545 if (diff_populate_filespec(one, 0))
546 die("cannot read data blob for %s", one->path);
547 prep_temp_blob(temp, one->data, one->size,
548 one->sha1, one->mode);
549 }
550 }
551
552 static void remove_tempfile(void)
553 {
554 int i;
555
556 for (i = 0; i < 2; i++)
557 if (diff_temp[i].name == diff_temp[i].tmp_path) {
558 unlink(diff_temp[i].name);
559 diff_temp[i].name = NULL;
560 }
561 }
562
563 static void remove_tempfile_on_signal(int signo)
564 {
565 remove_tempfile();
566 signal(SIGINT, SIG_DFL);
567 raise(signo);
568 }
569
570 static int spawn_prog(const char *pgm, const char **arg)
571 {
572 pid_t pid;
573 int status;
574
575 fflush(NULL);
576 pid = fork();
577 if (pid < 0)
578 die("unable to fork");
579 if (!pid) {
580 execvp(pgm, (char *const*) arg);
581 exit(255);
582 }
583
584 while (waitpid(pid, &status, 0) < 0) {
585 if (errno == EINTR)
586 continue;
587 return -1;
588 }
589
590 /* Earlier we did not check the exit status because
591 * diff exits non-zero if files are different, and
592 * we are not interested in knowing that. It was a
593 * mistake which made it harder to quit a diff-*
594 * session that uses the git-apply-patch-script as
595 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
596 * should also exit non-zero only when it wants to
597 * abort the entire diff-* session.
598 */
599 if (WIFEXITED(status) && !WEXITSTATUS(status))
600 return 0;
601 return -1;
602 }
603
604 /* An external diff command takes:
605 *
606 * diff-cmd name infile1 infile1-sha1 infile1-mode \
607 * infile2 infile2-sha1 infile2-mode [ rename-to ]
608 *
609 */
610 static void run_external_diff(const char *pgm,
611 const char *name,
612 const char *other,
613 struct diff_filespec *one,
614 struct diff_filespec *two,
615 const char *xfrm_msg,
616 int complete_rewrite)
617 {
618 const char *spawn_arg[10];
619 struct diff_tempfile *temp = diff_temp;
620 int retval;
621 static int atexit_asked = 0;
622 const char *othername;
623
624 othername = (other? other : name);
625 if (one && two) {
626 prepare_temp_file(name, &temp[0], one);
627 prepare_temp_file(othername, &temp[1], two);
628 if (! atexit_asked &&
629 (temp[0].name == temp[0].tmp_path ||
630 temp[1].name == temp[1].tmp_path)) {
631 atexit_asked = 1;
632 atexit(remove_tempfile);
633 }
634 signal(SIGINT, remove_tempfile_on_signal);
635 }
636
637 if (pgm) {
638 const char **arg = &spawn_arg[0];
639 if (one && two) {
640 *arg++ = pgm;
641 *arg++ = name;
642 *arg++ = temp[0].name;
643 *arg++ = temp[0].hex;
644 *arg++ = temp[0].mode;
645 *arg++ = temp[1].name;
646 *arg++ = temp[1].hex;
647 *arg++ = temp[1].mode;
648 if (other) {
649 *arg++ = other;
650 *arg++ = xfrm_msg;
651 }
652 } else {
653 *arg++ = pgm;
654 *arg++ = name;
655 }
656 *arg = NULL;
657 } else {
658 if (one && two) {
659 pgm = builtin_diff(name, othername, temp, xfrm_msg, complete_rewrite, spawn_arg);
660 } else
661 printf("* Unmerged path %s\n", name);
662 }
663
664 retval = 0;
665 if (pgm)
666 retval = spawn_prog(pgm, spawn_arg);
667 remove_tempfile();
668 if (retval) {
669 fprintf(stderr, "external diff died, stopping at %s.\n", name);
670 exit(1);
671 }
672 }
673
674 static void diff_fill_sha1_info(struct diff_filespec *one)
675 {
676 if (DIFF_FILE_VALID(one)) {
677 if (!one->sha1_valid) {
678 struct stat st;
679 if (lstat(one->path, &st) < 0)
680 die("stat %s", one->path);
681 if (index_path(one->sha1, one->path, &st, 0))
682 die("cannot hash %s\n", one->path);
683 }
684 }
685 else
686 memset(one->sha1, 0, 20);
687 }
688
689 static void run_diff(struct diff_filepair *p, struct diff_options *o)
690 {
691 const char *pgm = external_diff();
692 char msg[PATH_MAX*2+300], *xfrm_msg;
693 struct diff_filespec *one;
694 struct diff_filespec *two;
695 const char *name;
696 const char *other;
697 char *name_munged, *other_munged;
698 int complete_rewrite = 0;
699 int len;
700
701 if (DIFF_PAIR_UNMERGED(p)) {
702 /* unmerged */
703 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
704 0);
705 return;
706 }
707
708 name = p->one->path;
709 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
710 name_munged = quote_one(name);
711 other_munged = quote_one(other);
712 one = p->one; two = p->two;
713
714 diff_fill_sha1_info(one);
715 diff_fill_sha1_info(two);
716
717 len = 0;
718 switch (p->status) {
719 case DIFF_STATUS_COPIED:
720 len += snprintf(msg + len, sizeof(msg) - len,
721 "similarity index %d%%\n"
722 "copy from %s\n"
723 "copy to %s\n",
724 (int)(0.5 + p->score * 100.0/MAX_SCORE),
725 name_munged, other_munged);
726 break;
727 case DIFF_STATUS_RENAMED:
728 len += snprintf(msg + len, sizeof(msg) - len,
729 "similarity index %d%%\n"
730 "rename from %s\n"
731 "rename to %s\n",
732 (int)(0.5 + p->score * 100.0/MAX_SCORE),
733 name_munged, other_munged);
734 break;
735 case DIFF_STATUS_MODIFIED:
736 if (p->score) {
737 len += snprintf(msg + len, sizeof(msg) - len,
738 "dissimilarity index %d%%\n",
739 (int)(0.5 + p->score *
740 100.0/MAX_SCORE));
741 complete_rewrite = 1;
742 break;
743 }
744 /* fallthru */
745 default:
746 /* nothing */
747 ;
748 }
749
750 if (memcmp(one->sha1, two->sha1, 20)) {
751 char one_sha1[41];
752 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
753 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
754
755 len += snprintf(msg + len, sizeof(msg) - len,
756 "index %.*s..%.*s",
757 abbrev, one_sha1, abbrev,
758 sha1_to_hex(two->sha1));
759 if (one->mode == two->mode)
760 len += snprintf(msg + len, sizeof(msg) - len,
761 " %06o", one->mode);
762 len += snprintf(msg + len, sizeof(msg) - len, "\n");
763 }
764
765 if (len)
766 msg[--len] = 0;
767 xfrm_msg = len ? msg : NULL;
768
769 if (!pgm &&
770 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
771 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
772 /* a filepair that changes between file and symlink
773 * needs to be split into deletion and creation.
774 */
775 struct diff_filespec *null = alloc_filespec(two->path);
776 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
777 free(null);
778 null = alloc_filespec(one->path);
779 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
780 free(null);
781 }
782 else
783 run_external_diff(pgm, name, other, one, two, xfrm_msg,
784 complete_rewrite);
785
786 free(name_munged);
787 free(other_munged);
788 }
789
790 void diff_setup(struct diff_options *options)
791 {
792 memset(options, 0, sizeof(*options));
793 options->output_format = DIFF_FORMAT_RAW;
794 options->line_termination = '\n';
795 options->break_opt = -1;
796 options->rename_limit = -1;
797
798 options->change = diff_change;
799 options->add_remove = diff_addremove;
800 }
801
802 int diff_setup_done(struct diff_options *options)
803 {
804 if ((options->find_copies_harder &&
805 options->detect_rename != DIFF_DETECT_COPY) ||
806 (0 <= options->rename_limit && !options->detect_rename))
807 return -1;
808 if (options->detect_rename && options->rename_limit < 0)
809 options->rename_limit = diff_rename_limit_default;
810 if (options->setup & DIFF_SETUP_USE_CACHE) {
811 if (!active_cache)
812 /* read-cache does not die even when it fails
813 * so it is safe for us to do this here. Also
814 * it does not smudge active_cache or active_nr
815 * when it fails, so we do not have to worry about
816 * cleaning it up ourselves either.
817 */
818 read_cache();
819 }
820 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
821 use_size_cache = 1;
822 if (options->abbrev <= 0 || 40 < options->abbrev)
823 options->abbrev = 40; /* full */
824
825 return 0;
826 }
827
828 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
829 {
830 const char *arg = av[0];
831 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
832 options->output_format = DIFF_FORMAT_PATCH;
833 else if (!strcmp(arg, "-z"))
834 options->line_termination = 0;
835 else if (!strncmp(arg, "-l", 2))
836 options->rename_limit = strtoul(arg+2, NULL, 10);
837 else if (!strcmp(arg, "--full-index"))
838 options->full_index = 1;
839 else if (!strcmp(arg, "--name-only"))
840 options->output_format = DIFF_FORMAT_NAME;
841 else if (!strcmp(arg, "--name-status"))
842 options->output_format = DIFF_FORMAT_NAME_STATUS;
843 else if (!strcmp(arg, "-R"))
844 options->reverse_diff = 1;
845 else if (!strncmp(arg, "-S", 2))
846 options->pickaxe = arg + 2;
847 else if (!strcmp(arg, "-s"))
848 options->output_format = DIFF_FORMAT_NO_OUTPUT;
849 else if (!strncmp(arg, "-O", 2))
850 options->orderfile = arg + 2;
851 else if (!strncmp(arg, "--diff-filter=", 14))
852 options->filter = arg + 14;
853 else if (!strcmp(arg, "--pickaxe-all"))
854 options->pickaxe_opts = DIFF_PICKAXE_ALL;
855 else if (!strncmp(arg, "-B", 2)) {
856 if ((options->break_opt =
857 diff_scoreopt_parse(arg)) == -1)
858 return -1;
859 }
860 else if (!strncmp(arg, "-M", 2)) {
861 if ((options->rename_score =
862 diff_scoreopt_parse(arg)) == -1)
863 return -1;
864 options->detect_rename = DIFF_DETECT_RENAME;
865 }
866 else if (!strncmp(arg, "-C", 2)) {
867 if ((options->rename_score =
868 diff_scoreopt_parse(arg)) == -1)
869 return -1;
870 options->detect_rename = DIFF_DETECT_COPY;
871 }
872 else if (!strcmp(arg, "--find-copies-harder"))
873 options->find_copies_harder = 1;
874 else if (!strcmp(arg, "--abbrev"))
875 options->abbrev = DEFAULT_ABBREV;
876 else if (!strncmp(arg, "--abbrev=", 9)) {
877 options->abbrev = strtoul(arg + 9, NULL, 10);
878 if (options->abbrev < MINIMUM_ABBREV)
879 options->abbrev = MINIMUM_ABBREV;
880 else if (40 < options->abbrev)
881 options->abbrev = 40;
882 }
883 else
884 return 0;
885 return 1;
886 }
887
888 static int parse_num(const char **cp_p)
889 {
890 unsigned long num, scale;
891 int ch, dot;
892 const char *cp = *cp_p;
893
894 num = 0;
895 scale = 1;
896 dot = 0;
897 for(;;) {
898 ch = *cp;
899 if ( !dot && ch == '.' ) {
900 scale = 1;
901 dot = 1;
902 } else if ( ch == '%' ) {
903 scale = dot ? scale*100 : 100;
904 cp++; /* % is always at the end */
905 break;
906 } else if ( ch >= '0' && ch <= '9' ) {
907 if ( scale < 100000 ) {
908 scale *= 10;
909 num = (num*10) + (ch-'0');
910 }
911 } else {
912 break;
913 }
914 cp++;
915 }
916 *cp_p = cp;
917
918 /* user says num divided by scale and we say internally that
919 * is MAX_SCORE * num / scale.
920 */
921 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
922 }
923
924 int diff_scoreopt_parse(const char *opt)
925 {
926 int opt1, opt2, cmd;
927
928 if (*opt++ != '-')
929 return -1;
930 cmd = *opt++;
931 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
932 return -1; /* that is not a -M, -C nor -B option */
933
934 opt1 = parse_num(&opt);
935 if (cmd != 'B')
936 opt2 = 0;
937 else {
938 if (*opt == 0)
939 opt2 = 0;
940 else if (*opt != '/')
941 return -1; /* we expect -B80/99 or -B80 */
942 else {
943 opt++;
944 opt2 = parse_num(&opt);
945 }
946 }
947 if (*opt != 0)
948 return -1;
949 return opt1 | (opt2 << 16);
950 }
951
952 struct diff_queue_struct diff_queued_diff;
953
954 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
955 {
956 if (queue->alloc <= queue->nr) {
957 queue->alloc = alloc_nr(queue->alloc);
958 queue->queue = xrealloc(queue->queue,
959 sizeof(dp) * queue->alloc);
960 }
961 queue->queue[queue->nr++] = dp;
962 }
963
964 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
965 struct diff_filespec *one,
966 struct diff_filespec *two)
967 {
968 struct diff_filepair *dp = xmalloc(sizeof(*dp));
969 dp->one = one;
970 dp->two = two;
971 dp->score = 0;
972 dp->status = 0;
973 dp->source_stays = 0;
974 dp->broken_pair = 0;
975 if (queue)
976 diff_q(queue, dp);
977 return dp;
978 }
979
980 void diff_free_filepair(struct diff_filepair *p)
981 {
982 diff_free_filespec_data(p->one);
983 diff_free_filespec_data(p->two);
984 free(p->one);
985 free(p->two);
986 free(p);
987 }
988
989 /* This is different from find_unique_abbrev() in that
990 * it stuffs the result with dots for alignment.
991 */
992 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
993 {
994 int abblen;
995 const char *abbrev;
996 if (len == 40)
997 return sha1_to_hex(sha1);
998
999 abbrev = find_unique_abbrev(sha1, len);
1000 if (!abbrev)
1001 return sha1_to_hex(sha1);
1002 abblen = strlen(abbrev);
1003 if (abblen < 37) {
1004 static char hex[41];
1005 if (len < abblen && abblen <= len + 2)
1006 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1007 else
1008 sprintf(hex, "%s...", abbrev);
1009 return hex;
1010 }
1011 return sha1_to_hex(sha1);
1012 }
1013
1014 static void diff_flush_raw(struct diff_filepair *p,
1015 int line_termination,
1016 int inter_name_termination,
1017 struct diff_options *options)
1018 {
1019 int two_paths;
1020 char status[10];
1021 int abbrev = options->abbrev;
1022 const char *path_one, *path_two;
1023 int output_format = options->output_format;
1024
1025 path_one = p->one->path;
1026 path_two = p->two->path;
1027 if (line_termination) {
1028 path_one = quote_one(path_one);
1029 path_two = quote_one(path_two);
1030 }
1031
1032 if (p->score)
1033 sprintf(status, "%c%03d", p->status,
1034 (int)(0.5 + p->score * 100.0/MAX_SCORE));
1035 else {
1036 status[0] = p->status;
1037 status[1] = 0;
1038 }
1039 switch (p->status) {
1040 case DIFF_STATUS_COPIED:
1041 case DIFF_STATUS_RENAMED:
1042 two_paths = 1;
1043 break;
1044 case DIFF_STATUS_ADDED:
1045 case DIFF_STATUS_DELETED:
1046 two_paths = 0;
1047 break;
1048 default:
1049 two_paths = 0;
1050 break;
1051 }
1052 if (output_format != DIFF_FORMAT_NAME_STATUS) {
1053 printf(":%06o %06o %s ",
1054 p->one->mode, p->two->mode,
1055 diff_unique_abbrev(p->one->sha1, abbrev));
1056 printf("%s ",
1057 diff_unique_abbrev(p->two->sha1, abbrev));
1058 }
1059 printf("%s%c%s", status, inter_name_termination, path_one);
1060 if (two_paths)
1061 printf("%c%s", inter_name_termination, path_two);
1062 putchar(line_termination);
1063 if (path_one != p->one->path)
1064 free((void*)path_one);
1065 if (path_two != p->two->path)
1066 free((void*)path_two);
1067 }
1068
1069 static void diff_flush_name(struct diff_filepair *p,
1070 int inter_name_termination,
1071 int line_termination)
1072 {
1073 char *path = p->two->path;
1074
1075 if (line_termination)
1076 path = quote_one(p->two->path);
1077 else
1078 path = p->two->path;
1079 printf("%s%c", path, line_termination);
1080 if (p->two->path != path)
1081 free(path);
1082 }
1083
1084 int diff_unmodified_pair(struct diff_filepair *p)
1085 {
1086 /* This function is written stricter than necessary to support
1087 * the currently implemented transformers, but the idea is to
1088 * let transformers to produce diff_filepairs any way they want,
1089 * and filter and clean them up here before producing the output.
1090 */
1091 struct diff_filespec *one, *two;
1092
1093 if (DIFF_PAIR_UNMERGED(p))
1094 return 0; /* unmerged is interesting */
1095
1096 one = p->one;
1097 two = p->two;
1098
1099 /* deletion, addition, mode or type change
1100 * and rename are all interesting.
1101 */
1102 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1103 DIFF_PAIR_MODE_CHANGED(p) ||
1104 strcmp(one->path, two->path))
1105 return 0;
1106
1107 /* both are valid and point at the same path. that is, we are
1108 * dealing with a change.
1109 */
1110 if (one->sha1_valid && two->sha1_valid &&
1111 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1112 return 1; /* no change */
1113 if (!one->sha1_valid && !two->sha1_valid)
1114 return 1; /* both look at the same file on the filesystem. */
1115 return 0;
1116 }
1117
1118 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1119 {
1120 if (diff_unmodified_pair(p))
1121 return;
1122
1123 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1124 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1125 return; /* no tree diffs in patch format */
1126
1127 run_diff(p, o);
1128 }
1129
1130 int diff_queue_is_empty(void)
1131 {
1132 struct diff_queue_struct *q = &diff_queued_diff;
1133 int i;
1134 for (i = 0; i < q->nr; i++)
1135 if (!diff_unmodified_pair(q->queue[i]))
1136 return 0;
1137 return 1;
1138 }
1139
1140 #if DIFF_DEBUG
1141 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1142 {
1143 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1144 x, one ? one : "",
1145 s->path,
1146 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1147 s->mode,
1148 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1149 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1150 x, one ? one : "",
1151 s->size, s->xfrm_flags);
1152 }
1153
1154 void diff_debug_filepair(const struct diff_filepair *p, int i)
1155 {
1156 diff_debug_filespec(p->one, i, "one");
1157 diff_debug_filespec(p->two, i, "two");
1158 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1159 p->score, p->status ? p->status : '?',
1160 p->source_stays, p->broken_pair);
1161 }
1162
1163 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1164 {
1165 int i;
1166 if (msg)
1167 fprintf(stderr, "%s\n", msg);
1168 fprintf(stderr, "q->nr = %d\n", q->nr);
1169 for (i = 0; i < q->nr; i++) {
1170 struct diff_filepair *p = q->queue[i];
1171 diff_debug_filepair(p, i);
1172 }
1173 }
1174 #endif
1175
1176 static void diff_resolve_rename_copy(void)
1177 {
1178 int i, j;
1179 struct diff_filepair *p, *pp;
1180 struct diff_queue_struct *q = &diff_queued_diff;
1181
1182 diff_debug_queue("resolve-rename-copy", q);
1183
1184 for (i = 0; i < q->nr; i++) {
1185 p = q->queue[i];
1186 p->status = 0; /* undecided */
1187 if (DIFF_PAIR_UNMERGED(p))
1188 p->status = DIFF_STATUS_UNMERGED;
1189 else if (!DIFF_FILE_VALID(p->one))
1190 p->status = DIFF_STATUS_ADDED;
1191 else if (!DIFF_FILE_VALID(p->two))
1192 p->status = DIFF_STATUS_DELETED;
1193 else if (DIFF_PAIR_TYPE_CHANGED(p))
1194 p->status = DIFF_STATUS_TYPE_CHANGED;
1195
1196 /* from this point on, we are dealing with a pair
1197 * whose both sides are valid and of the same type, i.e.
1198 * either in-place edit or rename/copy edit.
1199 */
1200 else if (DIFF_PAIR_RENAME(p)) {
1201 if (p->source_stays) {
1202 p->status = DIFF_STATUS_COPIED;
1203 continue;
1204 }
1205 /* See if there is some other filepair that
1206 * copies from the same source as us. If so
1207 * we are a copy. Otherwise we are either a
1208 * copy if the path stays, or a rename if it
1209 * does not, but we already handled "stays" case.
1210 */
1211 for (j = i + 1; j < q->nr; j++) {
1212 pp = q->queue[j];
1213 if (strcmp(pp->one->path, p->one->path))
1214 continue; /* not us */
1215 if (!DIFF_PAIR_RENAME(pp))
1216 continue; /* not a rename/copy */
1217 /* pp is a rename/copy from the same source */
1218 p->status = DIFF_STATUS_COPIED;
1219 break;
1220 }
1221 if (!p->status)
1222 p->status = DIFF_STATUS_RENAMED;
1223 }
1224 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1225 p->one->mode != p->two->mode)
1226 p->status = DIFF_STATUS_MODIFIED;
1227 else {
1228 /* This is a "no-change" entry and should not
1229 * happen anymore, but prepare for broken callers.
1230 */
1231 error("feeding unmodified %s to diffcore",
1232 p->one->path);
1233 p->status = DIFF_STATUS_UNKNOWN;
1234 }
1235 }
1236 diff_debug_queue("resolve-rename-copy done", q);
1237 }
1238
1239 void diff_flush(struct diff_options *options)
1240 {
1241 struct diff_queue_struct *q = &diff_queued_diff;
1242 int i;
1243 int inter_name_termination = '\t';
1244 int diff_output_format = options->output_format;
1245 int line_termination = options->line_termination;
1246
1247 if (!line_termination)
1248 inter_name_termination = 0;
1249
1250 for (i = 0; i < q->nr; i++) {
1251 struct diff_filepair *p = q->queue[i];
1252 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1253 (p->status == DIFF_STATUS_UNKNOWN))
1254 continue;
1255 if (p->status == 0)
1256 die("internal error in diff-resolve-rename-copy");
1257 switch (diff_output_format) {
1258 case DIFF_FORMAT_PATCH:
1259 diff_flush_patch(p, options);
1260 break;
1261 case DIFF_FORMAT_RAW:
1262 case DIFF_FORMAT_NAME_STATUS:
1263 diff_flush_raw(p, line_termination,
1264 inter_name_termination,
1265 options);
1266 break;
1267 case DIFF_FORMAT_NAME:
1268 diff_flush_name(p,
1269 inter_name_termination,
1270 line_termination);
1271 break;
1272 }
1273 diff_free_filepair(q->queue[i]);
1274 }
1275 free(q->queue);
1276 q->queue = NULL;
1277 q->nr = q->alloc = 0;
1278 }
1279
1280 static void diffcore_apply_filter(const char *filter)
1281 {
1282 int i;
1283 struct diff_queue_struct *q = &diff_queued_diff;
1284 struct diff_queue_struct outq;
1285 outq.queue = NULL;
1286 outq.nr = outq.alloc = 0;
1287
1288 if (!filter)
1289 return;
1290
1291 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1292 int found;
1293 for (i = found = 0; !found && i < q->nr; i++) {
1294 struct diff_filepair *p = q->queue[i];
1295 if (((p->status == DIFF_STATUS_MODIFIED) &&
1296 ((p->score &&
1297 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1298 (!p->score &&
1299 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1300 ((p->status != DIFF_STATUS_MODIFIED) &&
1301 strchr(filter, p->status)))
1302 found++;
1303 }
1304 if (found)
1305 return;
1306
1307 /* otherwise we will clear the whole queue
1308 * by copying the empty outq at the end of this
1309 * function, but first clear the current entries
1310 * in the queue.
1311 */
1312 for (i = 0; i < q->nr; i++)
1313 diff_free_filepair(q->queue[i]);
1314 }
1315 else {
1316 /* Only the matching ones */
1317 for (i = 0; i < q->nr; i++) {
1318 struct diff_filepair *p = q->queue[i];
1319
1320 if (((p->status == DIFF_STATUS_MODIFIED) &&
1321 ((p->score &&
1322 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1323 (!p->score &&
1324 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1325 ((p->status != DIFF_STATUS_MODIFIED) &&
1326 strchr(filter, p->status)))
1327 diff_q(&outq, p);
1328 else
1329 diff_free_filepair(p);
1330 }
1331 }
1332 free(q->queue);
1333 *q = outq;
1334 }
1335
1336 void diffcore_std(struct diff_options *options)
1337 {
1338 if (options->paths && options->paths[0])
1339 diffcore_pathspec(options->paths);
1340 if (options->break_opt != -1)
1341 diffcore_break(options->break_opt);
1342 if (options->detect_rename)
1343 diffcore_rename(options);
1344 if (options->break_opt != -1)
1345 diffcore_merge_broken();
1346 if (options->pickaxe)
1347 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1348 if (options->orderfile)
1349 diffcore_order(options->orderfile);
1350 diff_resolve_rename_copy();
1351 diffcore_apply_filter(options->filter);
1352 }
1353
1354
1355 void diffcore_std_no_resolve(struct diff_options *options)
1356 {
1357 if (options->pickaxe)
1358 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1359 if (options->orderfile)
1360 diffcore_order(options->orderfile);
1361 diffcore_apply_filter(options->filter);
1362 }
1363
1364 void diff_addremove(struct diff_options *options,
1365 int addremove, unsigned mode,
1366 const unsigned char *sha1,
1367 const char *base, const char *path)
1368 {
1369 char concatpath[PATH_MAX];
1370 struct diff_filespec *one, *two;
1371
1372 /* This may look odd, but it is a preparation for
1373 * feeding "there are unchanged files which should
1374 * not produce diffs, but when you are doing copy
1375 * detection you would need them, so here they are"
1376 * entries to the diff-core. They will be prefixed
1377 * with something like '=' or '*' (I haven't decided
1378 * which but should not make any difference).
1379 * Feeding the same new and old to diff_change()
1380 * also has the same effect.
1381 * Before the final output happens, they are pruned after
1382 * merged into rename/copy pairs as appropriate.
1383 */
1384 if (options->reverse_diff)
1385 addremove = (addremove == '+' ? '-' :
1386 addremove == '-' ? '+' : addremove);
1387
1388 if (!path) path = "";
1389 sprintf(concatpath, "%s%s", base, path);
1390 one = alloc_filespec(concatpath);
1391 two = alloc_filespec(concatpath);
1392
1393 if (addremove != '+')
1394 fill_filespec(one, sha1, mode);
1395 if (addremove != '-')
1396 fill_filespec(two, sha1, mode);
1397
1398 diff_queue(&diff_queued_diff, one, two);
1399 }
1400
1401 void diff_change(struct diff_options *options,
1402 unsigned old_mode, unsigned new_mode,
1403 const unsigned char *old_sha1,
1404 const unsigned char *new_sha1,
1405 const char *base, const char *path)
1406 {
1407 char concatpath[PATH_MAX];
1408 struct diff_filespec *one, *two;
1409
1410 if (options->reverse_diff) {
1411 unsigned tmp;
1412 const unsigned char *tmp_c;
1413 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1414 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1415 }
1416 if (!path) path = "";
1417 sprintf(concatpath, "%s%s", base, path);
1418 one = alloc_filespec(concatpath);
1419 two = alloc_filespec(concatpath);
1420 fill_filespec(one, old_sha1, old_mode);
1421 fill_filespec(two, new_sha1, new_mode);
1422
1423 diff_queue(&diff_queued_diff, one, two);
1424 }
1425
1426 void diff_unmerge(struct diff_options *options,
1427 const char *path)
1428 {
1429 struct diff_filespec *one, *two;
1430 one = alloc_filespec(path);
1431 two = alloc_filespec(path);
1432 diff_queue(&diff_queued_diff, one, two);
1433 }