]> git.ipfire.org Git - thirdparty/git.git/blob - diff.c
GIT 0.99.9k
[thirdparty/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 void 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 {
187 int i, next_at, cmd_size;
188 const char *const diff_cmd = "diff -L%s -L%s";
189 const char *const diff_arg = "-- %s %s||:"; /* "||:" is to return 0 */
190 const char *input_name_sq[2];
191 const char *label_path[2];
192 char *cmd;
193
194 /* diff_cmd and diff_arg have 4 %s in total which makes
195 * the sum of these strings 8 bytes larger than required.
196 * we use 2 spaces around diff-opts, and we need to count
197 * terminating NUL; we used to subtract 5 here, but we do not
198 * care about small leaks in this subprocess that is about
199 * to exec "diff" anymore.
200 */
201 cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
202 + 128);
203
204 for (i = 0; i < 2; i++) {
205 input_name_sq[i] = sq_quote(temp[i].name);
206 if (!strcmp(temp[i].name, "/dev/null"))
207 label_path[i] = "/dev/null";
208 else if (!i)
209 label_path[i] = sq_quote(quote_two("a/", name_a));
210 else
211 label_path[i] = sq_quote(quote_two("b/", name_b));
212 cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
213 }
214
215 cmd = xmalloc(cmd_size);
216
217 next_at = 0;
218 next_at += snprintf(cmd+next_at, cmd_size-next_at,
219 diff_cmd, label_path[0], label_path[1]);
220 next_at += snprintf(cmd+next_at, cmd_size-next_at,
221 " %s ", diff_opts);
222 next_at += snprintf(cmd+next_at, cmd_size-next_at,
223 diff_arg, input_name_sq[0], input_name_sq[1]);
224
225 printf("diff --git %s %s\n",
226 quote_two("a/", name_a), quote_two("b/", name_b));
227 if (label_path[0][0] == '/') {
228 /* dev/null */
229 printf("new file mode %s\n", temp[1].mode);
230 if (xfrm_msg && xfrm_msg[0])
231 puts(xfrm_msg);
232 }
233 else if (label_path[1][0] == '/') {
234 printf("deleted file mode %s\n", temp[0].mode);
235 if (xfrm_msg && xfrm_msg[0])
236 puts(xfrm_msg);
237 }
238 else {
239 if (strcmp(temp[0].mode, temp[1].mode)) {
240 printf("old mode %s\n", temp[0].mode);
241 printf("new mode %s\n", temp[1].mode);
242 }
243 if (xfrm_msg && xfrm_msg[0])
244 puts(xfrm_msg);
245 if (strncmp(temp[0].mode, temp[1].mode, 3))
246 /* we do not run diff between different kind
247 * of objects.
248 */
249 exit(0);
250 if (complete_rewrite) {
251 fflush(NULL);
252 emit_rewrite_diff(name_a, name_b, temp);
253 exit(0);
254 }
255 }
256 fflush(NULL);
257 execlp("/bin/sh","sh", "-c", cmd, NULL);
258 }
259
260 struct diff_filespec *alloc_filespec(const char *path)
261 {
262 int namelen = strlen(path);
263 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
264
265 memset(spec, 0, sizeof(*spec));
266 spec->path = (char *)(spec + 1);
267 memcpy(spec->path, path, namelen+1);
268 return spec;
269 }
270
271 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
272 unsigned short mode)
273 {
274 if (mode) {
275 spec->mode = DIFF_FILE_CANON_MODE(mode);
276 memcpy(spec->sha1, sha1, 20);
277 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
278 }
279 }
280
281 /*
282 * Given a name and sha1 pair, if the dircache tells us the file in
283 * the work tree has that object contents, return true, so that
284 * prepare_temp_file() does not have to inflate and extract.
285 */
286 static int work_tree_matches(const char *name, const unsigned char *sha1)
287 {
288 struct cache_entry *ce;
289 struct stat st;
290 int pos, len;
291
292 /* We do not read the cache ourselves here, because the
293 * benchmark with my previous version that always reads cache
294 * shows that it makes things worse for diff-tree comparing
295 * two linux-2.6 kernel trees in an already checked out work
296 * tree. This is because most diff-tree comparisons deal with
297 * only a small number of files, while reading the cache is
298 * expensive for a large project, and its cost outweighs the
299 * savings we get by not inflating the object to a temporary
300 * file. Practically, this code only helps when we are used
301 * by diff-cache --cached, which does read the cache before
302 * calling us.
303 */
304 if (!active_cache)
305 return 0;
306
307 len = strlen(name);
308 pos = cache_name_pos(name, len);
309 if (pos < 0)
310 return 0;
311 ce = active_cache[pos];
312 if ((lstat(name, &st) < 0) ||
313 !S_ISREG(st.st_mode) || /* careful! */
314 ce_match_stat(ce, &st) ||
315 memcmp(sha1, ce->sha1, 20))
316 return 0;
317 /* we return 1 only when we can stat, it is a regular file,
318 * stat information matches, and sha1 recorded in the cache
319 * matches. I.e. we know the file in the work tree really is
320 * the same as the <name, sha1> pair.
321 */
322 return 1;
323 }
324
325 static struct sha1_size_cache {
326 unsigned char sha1[20];
327 unsigned long size;
328 } **sha1_size_cache;
329 static int sha1_size_cache_nr, sha1_size_cache_alloc;
330
331 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
332 int find_only,
333 unsigned long size)
334 {
335 int first, last;
336 struct sha1_size_cache *e;
337
338 first = 0;
339 last = sha1_size_cache_nr;
340 while (last > first) {
341 int cmp, next = (last + first) >> 1;
342 e = sha1_size_cache[next];
343 cmp = memcmp(e->sha1, sha1, 20);
344 if (!cmp)
345 return e;
346 if (cmp < 0) {
347 last = next;
348 continue;
349 }
350 first = next+1;
351 }
352 /* not found */
353 if (find_only)
354 return NULL;
355 /* insert to make it at "first" */
356 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
357 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
358 sha1_size_cache = xrealloc(sha1_size_cache,
359 sha1_size_cache_alloc *
360 sizeof(*sha1_size_cache));
361 }
362 sha1_size_cache_nr++;
363 if (first < sha1_size_cache_nr)
364 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
365 (sha1_size_cache_nr - first - 1) *
366 sizeof(*sha1_size_cache));
367 e = xmalloc(sizeof(struct sha1_size_cache));
368 sha1_size_cache[first] = e;
369 memcpy(e->sha1, sha1, 20);
370 e->size = size;
371 return e;
372 }
373
374 /*
375 * While doing rename detection and pickaxe operation, we may need to
376 * grab the data for the blob (or file) for our own in-core comparison.
377 * diff_filespec has data and size fields for this purpose.
378 */
379 int diff_populate_filespec(struct diff_filespec *s, int size_only)
380 {
381 int err = 0;
382 if (!DIFF_FILE_VALID(s))
383 die("internal error: asking to populate invalid file.");
384 if (S_ISDIR(s->mode))
385 return -1;
386
387 if (!use_size_cache)
388 size_only = 0;
389
390 if (s->data)
391 return err;
392 if (!s->sha1_valid ||
393 work_tree_matches(s->path, s->sha1)) {
394 struct stat st;
395 int fd;
396 if (lstat(s->path, &st) < 0) {
397 if (errno == ENOENT) {
398 err_empty:
399 err = -1;
400 empty:
401 s->data = "";
402 s->size = 0;
403 return err;
404 }
405 }
406 s->size = st.st_size;
407 if (!s->size)
408 goto empty;
409 if (size_only)
410 return 0;
411 if (S_ISLNK(st.st_mode)) {
412 int ret;
413 s->data = xmalloc(s->size);
414 s->should_free = 1;
415 ret = readlink(s->path, s->data, s->size);
416 if (ret < 0) {
417 free(s->data);
418 goto err_empty;
419 }
420 return 0;
421 }
422 fd = open(s->path, O_RDONLY);
423 if (fd < 0)
424 goto err_empty;
425 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
426 close(fd);
427 if (s->data == MAP_FAILED)
428 goto err_empty;
429 s->should_munmap = 1;
430 }
431 else {
432 char type[20];
433 struct sha1_size_cache *e;
434
435 if (size_only) {
436 e = locate_size_cache(s->sha1, 1, 0);
437 if (e) {
438 s->size = e->size;
439 return 0;
440 }
441 if (!sha1_object_info(s->sha1, type, &s->size))
442 locate_size_cache(s->sha1, 0, s->size);
443 }
444 else {
445 s->data = read_sha1_file(s->sha1, type, &s->size);
446 s->should_free = 1;
447 }
448 }
449 return 0;
450 }
451
452 void diff_free_filespec_data(struct diff_filespec *s)
453 {
454 if (s->should_free)
455 free(s->data);
456 else if (s->should_munmap)
457 munmap(s->data, s->size);
458 s->should_free = s->should_munmap = 0;
459 s->data = NULL;
460 }
461
462 static void prep_temp_blob(struct diff_tempfile *temp,
463 void *blob,
464 unsigned long size,
465 const unsigned char *sha1,
466 int mode)
467 {
468 int fd;
469
470 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
471 if (fd < 0)
472 die("unable to create temp-file");
473 if (write(fd, blob, size) != size)
474 die("unable to write temp-file");
475 close(fd);
476 temp->name = temp->tmp_path;
477 strcpy(temp->hex, sha1_to_hex(sha1));
478 temp->hex[40] = 0;
479 sprintf(temp->mode, "%06o", mode);
480 }
481
482 static void prepare_temp_file(const char *name,
483 struct diff_tempfile *temp,
484 struct diff_filespec *one)
485 {
486 if (!DIFF_FILE_VALID(one)) {
487 not_a_valid_file:
488 /* A '-' entry produces this for file-2, and
489 * a '+' entry produces this for file-1.
490 */
491 temp->name = "/dev/null";
492 strcpy(temp->hex, ".");
493 strcpy(temp->mode, ".");
494 return;
495 }
496
497 if (!one->sha1_valid ||
498 work_tree_matches(name, one->sha1)) {
499 struct stat st;
500 if (lstat(name, &st) < 0) {
501 if (errno == ENOENT)
502 goto not_a_valid_file;
503 die("stat(%s): %s", name, strerror(errno));
504 }
505 if (S_ISLNK(st.st_mode)) {
506 int ret;
507 char *buf, buf_[1024];
508 buf = ((sizeof(buf_) < st.st_size) ?
509 xmalloc(st.st_size) : buf_);
510 ret = readlink(name, buf, st.st_size);
511 if (ret < 0)
512 die("readlink(%s)", name);
513 prep_temp_blob(temp, buf, st.st_size,
514 (one->sha1_valid ?
515 one->sha1 : null_sha1),
516 (one->sha1_valid ?
517 one->mode : S_IFLNK));
518 }
519 else {
520 /* we can borrow from the file in the work tree */
521 temp->name = name;
522 if (!one->sha1_valid)
523 strcpy(temp->hex, sha1_to_hex(null_sha1));
524 else
525 strcpy(temp->hex, sha1_to_hex(one->sha1));
526 /* Even though we may sometimes borrow the
527 * contents from the work tree, we always want
528 * one->mode. mode is trustworthy even when
529 * !(one->sha1_valid), as long as
530 * DIFF_FILE_VALID(one).
531 */
532 sprintf(temp->mode, "%06o", one->mode);
533 }
534 return;
535 }
536 else {
537 if (diff_populate_filespec(one, 0))
538 die("cannot read data blob for %s", one->path);
539 prep_temp_blob(temp, one->data, one->size,
540 one->sha1, one->mode);
541 }
542 }
543
544 static void remove_tempfile(void)
545 {
546 int i;
547
548 for (i = 0; i < 2; i++)
549 if (diff_temp[i].name == diff_temp[i].tmp_path) {
550 unlink(diff_temp[i].name);
551 diff_temp[i].name = NULL;
552 }
553 }
554
555 static void remove_tempfile_on_signal(int signo)
556 {
557 remove_tempfile();
558 }
559
560 /* An external diff command takes:
561 *
562 * diff-cmd name infile1 infile1-sha1 infile1-mode \
563 * infile2 infile2-sha1 infile2-mode [ rename-to ]
564 *
565 */
566 static void run_external_diff(const char *pgm,
567 const char *name,
568 const char *other,
569 struct diff_filespec *one,
570 struct diff_filespec *two,
571 const char *xfrm_msg,
572 int complete_rewrite)
573 {
574 struct diff_tempfile *temp = diff_temp;
575 pid_t pid;
576 int status;
577 static int atexit_asked = 0;
578 const char *othername;
579
580 othername = (other? other : name);
581 if (one && two) {
582 prepare_temp_file(name, &temp[0], one);
583 prepare_temp_file(othername, &temp[1], two);
584 if (! atexit_asked &&
585 (temp[0].name == temp[0].tmp_path ||
586 temp[1].name == temp[1].tmp_path)) {
587 atexit_asked = 1;
588 atexit(remove_tempfile);
589 }
590 signal(SIGINT, remove_tempfile_on_signal);
591 }
592
593 fflush(NULL);
594 pid = fork();
595 if (pid < 0)
596 die("unable to fork");
597 if (!pid) {
598 if (pgm) {
599 if (one && two) {
600 const char *exec_arg[10];
601 const char **arg = &exec_arg[0];
602 *arg++ = pgm;
603 *arg++ = name;
604 *arg++ = temp[0].name;
605 *arg++ = temp[0].hex;
606 *arg++ = temp[0].mode;
607 *arg++ = temp[1].name;
608 *arg++ = temp[1].hex;
609 *arg++ = temp[1].mode;
610 if (other) {
611 *arg++ = other;
612 *arg++ = xfrm_msg;
613 }
614 *arg = NULL;
615 execvp(pgm, (char *const*) exec_arg);
616 }
617 else
618 execlp(pgm, pgm, name, NULL);
619 }
620 /*
621 * otherwise we use the built-in one.
622 */
623 if (one && two)
624 builtin_diff(name, othername, temp, xfrm_msg,
625 complete_rewrite);
626 else
627 printf("* Unmerged path %s\n", name);
628 exit(0);
629 }
630 if (waitpid(pid, &status, 0) < 0 ||
631 !WIFEXITED(status) || WEXITSTATUS(status)) {
632 /* Earlier we did not check the exit status because
633 * diff exits non-zero if files are different, and
634 * we are not interested in knowing that. It was a
635 * mistake which made it harder to quit a diff-*
636 * session that uses the git-apply-patch-script as
637 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
638 * should also exit non-zero only when it wants to
639 * abort the entire diff-* session.
640 */
641 remove_tempfile();
642 fprintf(stderr, "external diff died, stopping at %s.\n", name);
643 exit(1);
644 }
645 remove_tempfile();
646 }
647
648 static void diff_fill_sha1_info(struct diff_filespec *one)
649 {
650 if (DIFF_FILE_VALID(one)) {
651 if (!one->sha1_valid) {
652 struct stat st;
653 if (stat(one->path, &st) < 0)
654 die("stat %s", one->path);
655 if (index_path(one->sha1, one->path, &st, 0))
656 die("cannot hash %s\n", one->path);
657 }
658 }
659 else
660 memset(one->sha1, 0, 20);
661 }
662
663 static void run_diff(struct diff_filepair *p, struct diff_options *o)
664 {
665 const char *pgm = external_diff();
666 char msg[PATH_MAX*2+300], *xfrm_msg;
667 struct diff_filespec *one;
668 struct diff_filespec *two;
669 const char *name;
670 const char *other;
671 char *name_munged, *other_munged;
672 int complete_rewrite = 0;
673 int len;
674
675 if (DIFF_PAIR_UNMERGED(p)) {
676 /* unmerged */
677 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
678 0);
679 return;
680 }
681
682 name = p->one->path;
683 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
684 name_munged = quote_one(name);
685 other_munged = quote_one(other);
686 one = p->one; two = p->two;
687
688 diff_fill_sha1_info(one);
689 diff_fill_sha1_info(two);
690
691 len = 0;
692 switch (p->status) {
693 case DIFF_STATUS_COPIED:
694 len += snprintf(msg + len, sizeof(msg) - len,
695 "similarity index %d%%\n"
696 "copy from %s\n"
697 "copy to %s\n",
698 (int)(0.5 + p->score * 100.0/MAX_SCORE),
699 name_munged, other_munged);
700 break;
701 case DIFF_STATUS_RENAMED:
702 len += snprintf(msg + len, sizeof(msg) - len,
703 "similarity index %d%%\n"
704 "rename from %s\n"
705 "rename to %s\n",
706 (int)(0.5 + p->score * 100.0/MAX_SCORE),
707 name_munged, other_munged);
708 break;
709 case DIFF_STATUS_MODIFIED:
710 if (p->score) {
711 len += snprintf(msg + len, sizeof(msg) - len,
712 "dissimilarity index %d%%\n",
713 (int)(0.5 + p->score *
714 100.0/MAX_SCORE));
715 complete_rewrite = 1;
716 break;
717 }
718 /* fallthru */
719 default:
720 /* nothing */
721 ;
722 }
723
724 if (memcmp(one->sha1, two->sha1, 20)) {
725 char one_sha1[41];
726 const char *index_fmt = o->full_index ? "index %s..%s" : "index %.7s..%.7s";
727 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
728
729 len += snprintf(msg + len, sizeof(msg) - len,
730 index_fmt, one_sha1, sha1_to_hex(two->sha1));
731 if (one->mode == two->mode)
732 len += snprintf(msg + len, sizeof(msg) - len,
733 " %06o", one->mode);
734 len += snprintf(msg + len, sizeof(msg) - len, "\n");
735 }
736
737 if (len)
738 msg[--len] = 0;
739 xfrm_msg = len ? msg : NULL;
740
741 if (!pgm &&
742 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
743 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
744 /* a filepair that changes between file and symlink
745 * needs to be split into deletion and creation.
746 */
747 struct diff_filespec *null = alloc_filespec(two->path);
748 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
749 free(null);
750 null = alloc_filespec(one->path);
751 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
752 free(null);
753 }
754 else
755 run_external_diff(pgm, name, other, one, two, xfrm_msg,
756 complete_rewrite);
757
758 free(name_munged);
759 free(other_munged);
760 }
761
762 void diff_setup(struct diff_options *options)
763 {
764 memset(options, 0, sizeof(*options));
765 options->output_format = DIFF_FORMAT_RAW;
766 options->line_termination = '\n';
767 options->break_opt = -1;
768 options->rename_limit = -1;
769
770 options->change = diff_change;
771 options->add_remove = diff_addremove;
772 }
773
774 int diff_setup_done(struct diff_options *options)
775 {
776 if ((options->find_copies_harder &&
777 options->detect_rename != DIFF_DETECT_COPY) ||
778 (0 <= options->rename_limit && !options->detect_rename))
779 return -1;
780 if (options->detect_rename && options->rename_limit < 0)
781 options->rename_limit = diff_rename_limit_default;
782 if (options->setup & DIFF_SETUP_USE_CACHE) {
783 if (!active_cache)
784 /* read-cache does not die even when it fails
785 * so it is safe for us to do this here. Also
786 * it does not smudge active_cache or active_nr
787 * when it fails, so we do not have to worry about
788 * cleaning it up oufselves either.
789 */
790 read_cache();
791 }
792 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
793 use_size_cache = 1;
794
795 return 0;
796 }
797
798 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
799 {
800 const char *arg = av[0];
801 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
802 options->output_format = DIFF_FORMAT_PATCH;
803 else if (!strcmp(arg, "-z"))
804 options->line_termination = 0;
805 else if (!strncmp(arg, "-l", 2))
806 options->rename_limit = strtoul(arg+2, NULL, 10);
807 else if (!strcmp(arg, "--full-index"))
808 options->full_index = 1;
809 else if (!strcmp(arg, "--name-only"))
810 options->output_format = DIFF_FORMAT_NAME;
811 else if (!strcmp(arg, "--name-status"))
812 options->output_format = DIFF_FORMAT_NAME_STATUS;
813 else if (!strcmp(arg, "-R"))
814 options->reverse_diff = 1;
815 else if (!strncmp(arg, "-S", 2))
816 options->pickaxe = arg + 2;
817 else if (!strcmp(arg, "-s"))
818 options->output_format = DIFF_FORMAT_NO_OUTPUT;
819 else if (!strncmp(arg, "-O", 2))
820 options->orderfile = arg + 2;
821 else if (!strncmp(arg, "--diff-filter=", 14))
822 options->filter = arg + 14;
823 else if (!strcmp(arg, "--pickaxe-all"))
824 options->pickaxe_opts = DIFF_PICKAXE_ALL;
825 else if (!strncmp(arg, "-B", 2)) {
826 if ((options->break_opt =
827 diff_scoreopt_parse(arg)) == -1)
828 return -1;
829 }
830 else if (!strncmp(arg, "-M", 2)) {
831 if ((options->rename_score =
832 diff_scoreopt_parse(arg)) == -1)
833 return -1;
834 options->detect_rename = DIFF_DETECT_RENAME;
835 }
836 else if (!strncmp(arg, "-C", 2)) {
837 if ((options->rename_score =
838 diff_scoreopt_parse(arg)) == -1)
839 return -1;
840 options->detect_rename = DIFF_DETECT_COPY;
841 }
842 else if (!strcmp(arg, "--find-copies-harder"))
843 options->find_copies_harder = 1;
844 else
845 return 0;
846 return 1;
847 }
848
849 static int parse_num(const char **cp_p)
850 {
851 unsigned long num, scale;
852 int ch, dot;
853 const char *cp = *cp_p;
854
855 num = 0;
856 scale = 1;
857 dot = 0;
858 for(;;) {
859 ch = *cp;
860 if ( !dot && ch == '.' ) {
861 scale = 1;
862 dot = 1;
863 } else if ( ch == '%' ) {
864 scale = dot ? scale*100 : 100;
865 cp++; /* % is always at the end */
866 break;
867 } else if ( ch >= '0' && ch <= '9' ) {
868 if ( scale < 100000 ) {
869 scale *= 10;
870 num = (num*10) + (ch-'0');
871 }
872 } else {
873 break;
874 }
875 cp++;
876 }
877 *cp_p = cp;
878
879 /* user says num divided by scale and we say internally that
880 * is MAX_SCORE * num / scale.
881 */
882 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
883 }
884
885 int diff_scoreopt_parse(const char *opt)
886 {
887 int opt1, opt2, cmd;
888
889 if (*opt++ != '-')
890 return -1;
891 cmd = *opt++;
892 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
893 return -1; /* that is not a -M, -C nor -B option */
894
895 opt1 = parse_num(&opt);
896 if (cmd != 'B')
897 opt2 = 0;
898 else {
899 if (*opt == 0)
900 opt2 = 0;
901 else if (*opt != '/')
902 return -1; /* we expect -B80/99 or -B80 */
903 else {
904 opt++;
905 opt2 = parse_num(&opt);
906 }
907 }
908 if (*opt != 0)
909 return -1;
910 return opt1 | (opt2 << 16);
911 }
912
913 struct diff_queue_struct diff_queued_diff;
914
915 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
916 {
917 if (queue->alloc <= queue->nr) {
918 queue->alloc = alloc_nr(queue->alloc);
919 queue->queue = xrealloc(queue->queue,
920 sizeof(dp) * queue->alloc);
921 }
922 queue->queue[queue->nr++] = dp;
923 }
924
925 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
926 struct diff_filespec *one,
927 struct diff_filespec *two)
928 {
929 struct diff_filepair *dp = xmalloc(sizeof(*dp));
930 dp->one = one;
931 dp->two = two;
932 dp->score = 0;
933 dp->status = 0;
934 dp->source_stays = 0;
935 dp->broken_pair = 0;
936 if (queue)
937 diff_q(queue, dp);
938 return dp;
939 }
940
941 void diff_free_filepair(struct diff_filepair *p)
942 {
943 diff_free_filespec_data(p->one);
944 diff_free_filespec_data(p->two);
945 free(p->one);
946 free(p->two);
947 free(p);
948 }
949
950 static void diff_flush_raw(struct diff_filepair *p,
951 int line_termination,
952 int inter_name_termination,
953 int output_format)
954 {
955 int two_paths;
956 char status[10];
957 const char *path_one, *path_two;
958
959 path_one = p->one->path;
960 path_two = p->two->path;
961 if (line_termination) {
962 path_one = quote_one(path_one);
963 path_two = quote_one(path_two);
964 }
965
966 if (p->score)
967 sprintf(status, "%c%03d", p->status,
968 (int)(0.5 + p->score * 100.0/MAX_SCORE));
969 else {
970 status[0] = p->status;
971 status[1] = 0;
972 }
973 switch (p->status) {
974 case DIFF_STATUS_COPIED:
975 case DIFF_STATUS_RENAMED:
976 two_paths = 1;
977 break;
978 case DIFF_STATUS_ADDED:
979 case DIFF_STATUS_DELETED:
980 two_paths = 0;
981 break;
982 default:
983 two_paths = 0;
984 break;
985 }
986 if (output_format != DIFF_FORMAT_NAME_STATUS) {
987 printf(":%06o %06o %s ",
988 p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
989 printf("%s ", sha1_to_hex(p->two->sha1));
990 }
991 printf("%s%c%s", status, inter_name_termination, path_one);
992 if (two_paths)
993 printf("%c%s", inter_name_termination, path_two);
994 putchar(line_termination);
995 if (path_one != p->one->path)
996 free((void*)path_one);
997 if (path_two != p->two->path)
998 free((void*)path_two);
999 }
1000
1001 static void diff_flush_name(struct diff_filepair *p,
1002 int inter_name_termination,
1003 int line_termination)
1004 {
1005 char *path = p->two->path;
1006
1007 if (line_termination)
1008 path = quote_one(p->two->path);
1009 else
1010 path = p->two->path;
1011 printf("%s%c", path, line_termination);
1012 if (p->two->path != path)
1013 free(path);
1014 }
1015
1016 int diff_unmodified_pair(struct diff_filepair *p)
1017 {
1018 /* This function is written stricter than necessary to support
1019 * the currently implemented transformers, but the idea is to
1020 * let transformers to produce diff_filepairs any way they want,
1021 * and filter and clean them up here before producing the output.
1022 */
1023 struct diff_filespec *one, *two;
1024
1025 if (DIFF_PAIR_UNMERGED(p))
1026 return 0; /* unmerged is interesting */
1027
1028 one = p->one;
1029 two = p->two;
1030
1031 /* deletion, addition, mode or type change
1032 * and rename are all interesting.
1033 */
1034 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1035 DIFF_PAIR_MODE_CHANGED(p) ||
1036 strcmp(one->path, two->path))
1037 return 0;
1038
1039 /* both are valid and point at the same path. that is, we are
1040 * dealing with a change.
1041 */
1042 if (one->sha1_valid && two->sha1_valid &&
1043 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1044 return 1; /* no change */
1045 if (!one->sha1_valid && !two->sha1_valid)
1046 return 1; /* both look at the same file on the filesystem. */
1047 return 0;
1048 }
1049
1050 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1051 {
1052 if (diff_unmodified_pair(p))
1053 return;
1054
1055 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1056 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1057 return; /* no tree diffs in patch format */
1058
1059 run_diff(p, o);
1060 }
1061
1062 int diff_queue_is_empty(void)
1063 {
1064 struct diff_queue_struct *q = &diff_queued_diff;
1065 int i;
1066 for (i = 0; i < q->nr; i++)
1067 if (!diff_unmodified_pair(q->queue[i]))
1068 return 0;
1069 return 1;
1070 }
1071
1072 #if DIFF_DEBUG
1073 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1074 {
1075 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1076 x, one ? one : "",
1077 s->path,
1078 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1079 s->mode,
1080 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1081 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1082 x, one ? one : "",
1083 s->size, s->xfrm_flags);
1084 }
1085
1086 void diff_debug_filepair(const struct diff_filepair *p, int i)
1087 {
1088 diff_debug_filespec(p->one, i, "one");
1089 diff_debug_filespec(p->two, i, "two");
1090 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1091 p->score, p->status ? p->status : '?',
1092 p->source_stays, p->broken_pair);
1093 }
1094
1095 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1096 {
1097 int i;
1098 if (msg)
1099 fprintf(stderr, "%s\n", msg);
1100 fprintf(stderr, "q->nr = %d\n", q->nr);
1101 for (i = 0; i < q->nr; i++) {
1102 struct diff_filepair *p = q->queue[i];
1103 diff_debug_filepair(p, i);
1104 }
1105 }
1106 #endif
1107
1108 static void diff_resolve_rename_copy(void)
1109 {
1110 int i, j;
1111 struct diff_filepair *p, *pp;
1112 struct diff_queue_struct *q = &diff_queued_diff;
1113
1114 diff_debug_queue("resolve-rename-copy", q);
1115
1116 for (i = 0; i < q->nr; i++) {
1117 p = q->queue[i];
1118 p->status = 0; /* undecided */
1119 if (DIFF_PAIR_UNMERGED(p))
1120 p->status = DIFF_STATUS_UNMERGED;
1121 else if (!DIFF_FILE_VALID(p->one))
1122 p->status = DIFF_STATUS_ADDED;
1123 else if (!DIFF_FILE_VALID(p->two))
1124 p->status = DIFF_STATUS_DELETED;
1125 else if (DIFF_PAIR_TYPE_CHANGED(p))
1126 p->status = DIFF_STATUS_TYPE_CHANGED;
1127
1128 /* from this point on, we are dealing with a pair
1129 * whose both sides are valid and of the same type, i.e.
1130 * either in-place edit or rename/copy edit.
1131 */
1132 else if (DIFF_PAIR_RENAME(p)) {
1133 if (p->source_stays) {
1134 p->status = DIFF_STATUS_COPIED;
1135 continue;
1136 }
1137 /* See if there is some other filepair that
1138 * copies from the same source as us. If so
1139 * we are a copy. Otherwise we are either a
1140 * copy if the path stays, or a rename if it
1141 * does not, but we already handled "stays" case.
1142 */
1143 for (j = i + 1; j < q->nr; j++) {
1144 pp = q->queue[j];
1145 if (strcmp(pp->one->path, p->one->path))
1146 continue; /* not us */
1147 if (!DIFF_PAIR_RENAME(pp))
1148 continue; /* not a rename/copy */
1149 /* pp is a rename/copy from the same source */
1150 p->status = DIFF_STATUS_COPIED;
1151 break;
1152 }
1153 if (!p->status)
1154 p->status = DIFF_STATUS_RENAMED;
1155 }
1156 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1157 p->one->mode != p->two->mode)
1158 p->status = DIFF_STATUS_MODIFIED;
1159 else {
1160 /* This is a "no-change" entry and should not
1161 * happen anymore, but prepare for broken callers.
1162 */
1163 error("feeding unmodified %s to diffcore",
1164 p->one->path);
1165 p->status = DIFF_STATUS_UNKNOWN;
1166 }
1167 }
1168 diff_debug_queue("resolve-rename-copy done", q);
1169 }
1170
1171 void diff_flush(struct diff_options *options)
1172 {
1173 struct diff_queue_struct *q = &diff_queued_diff;
1174 int i;
1175 int inter_name_termination = '\t';
1176 int diff_output_format = options->output_format;
1177 int line_termination = options->line_termination;
1178
1179 if (!line_termination)
1180 inter_name_termination = 0;
1181
1182 for (i = 0; i < q->nr; i++) {
1183 struct diff_filepair *p = q->queue[i];
1184 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1185 (p->status == DIFF_STATUS_UNKNOWN))
1186 continue;
1187 if (p->status == 0)
1188 die("internal error in diff-resolve-rename-copy");
1189 switch (diff_output_format) {
1190 case DIFF_FORMAT_PATCH:
1191 diff_flush_patch(p, options);
1192 break;
1193 case DIFF_FORMAT_RAW:
1194 case DIFF_FORMAT_NAME_STATUS:
1195 diff_flush_raw(p, line_termination,
1196 inter_name_termination,
1197 diff_output_format);
1198 break;
1199 case DIFF_FORMAT_NAME:
1200 diff_flush_name(p,
1201 inter_name_termination,
1202 line_termination);
1203 break;
1204 }
1205 diff_free_filepair(q->queue[i]);
1206 }
1207 free(q->queue);
1208 q->queue = NULL;
1209 q->nr = q->alloc = 0;
1210 }
1211
1212 static void diffcore_apply_filter(const char *filter)
1213 {
1214 int i;
1215 struct diff_queue_struct *q = &diff_queued_diff;
1216 struct diff_queue_struct outq;
1217 outq.queue = NULL;
1218 outq.nr = outq.alloc = 0;
1219
1220 if (!filter)
1221 return;
1222
1223 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1224 int found;
1225 for (i = found = 0; !found && i < q->nr; i++) {
1226 struct diff_filepair *p = q->queue[i];
1227 if (((p->status == DIFF_STATUS_MODIFIED) &&
1228 ((p->score &&
1229 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1230 (!p->score &&
1231 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1232 ((p->status != DIFF_STATUS_MODIFIED) &&
1233 strchr(filter, p->status)))
1234 found++;
1235 }
1236 if (found)
1237 return;
1238
1239 /* otherwise we will clear the whole queue
1240 * by copying the empty outq at the end of this
1241 * function, but first clear the current entries
1242 * in the queue.
1243 */
1244 for (i = 0; i < q->nr; i++)
1245 diff_free_filepair(q->queue[i]);
1246 }
1247 else {
1248 /* Only the matching ones */
1249 for (i = 0; i < q->nr; i++) {
1250 struct diff_filepair *p = q->queue[i];
1251
1252 if (((p->status == DIFF_STATUS_MODIFIED) &&
1253 ((p->score &&
1254 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1255 (!p->score &&
1256 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1257 ((p->status != DIFF_STATUS_MODIFIED) &&
1258 strchr(filter, p->status)))
1259 diff_q(&outq, p);
1260 else
1261 diff_free_filepair(p);
1262 }
1263 }
1264 free(q->queue);
1265 *q = outq;
1266 }
1267
1268 void diffcore_std(struct diff_options *options)
1269 {
1270 if (options->paths && options->paths[0])
1271 diffcore_pathspec(options->paths);
1272 if (options->break_opt != -1)
1273 diffcore_break(options->break_opt);
1274 if (options->detect_rename)
1275 diffcore_rename(options);
1276 if (options->break_opt != -1)
1277 diffcore_merge_broken();
1278 if (options->pickaxe)
1279 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1280 if (options->orderfile)
1281 diffcore_order(options->orderfile);
1282 diff_resolve_rename_copy();
1283 diffcore_apply_filter(options->filter);
1284 }
1285
1286
1287 void diffcore_std_no_resolve(struct diff_options *options)
1288 {
1289 if (options->pickaxe)
1290 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1291 if (options->orderfile)
1292 diffcore_order(options->orderfile);
1293 diffcore_apply_filter(options->filter);
1294 }
1295
1296 void diff_addremove(struct diff_options *options,
1297 int addremove, unsigned mode,
1298 const unsigned char *sha1,
1299 const char *base, const char *path)
1300 {
1301 char concatpath[PATH_MAX];
1302 struct diff_filespec *one, *two;
1303
1304 /* This may look odd, but it is a preparation for
1305 * feeding "there are unchanged files which should
1306 * not produce diffs, but when you are doing copy
1307 * detection you would need them, so here they are"
1308 * entries to the diff-core. They will be prefixed
1309 * with something like '=' or '*' (I haven't decided
1310 * which but should not make any difference).
1311 * Feeding the same new and old to diff_change()
1312 * also has the same effect.
1313 * Before the final output happens, they are pruned after
1314 * merged into rename/copy pairs as appropriate.
1315 */
1316 if (options->reverse_diff)
1317 addremove = (addremove == '+' ? '-' :
1318 addremove == '-' ? '+' : addremove);
1319
1320 if (!path) path = "";
1321 sprintf(concatpath, "%s%s", base, path);
1322 one = alloc_filespec(concatpath);
1323 two = alloc_filespec(concatpath);
1324
1325 if (addremove != '+')
1326 fill_filespec(one, sha1, mode);
1327 if (addremove != '-')
1328 fill_filespec(two, sha1, mode);
1329
1330 diff_queue(&diff_queued_diff, one, two);
1331 }
1332
1333 void diff_change(struct diff_options *options,
1334 unsigned old_mode, unsigned new_mode,
1335 const unsigned char *old_sha1,
1336 const unsigned char *new_sha1,
1337 const char *base, const char *path)
1338 {
1339 char concatpath[PATH_MAX];
1340 struct diff_filespec *one, *two;
1341
1342 if (options->reverse_diff) {
1343 unsigned tmp;
1344 const unsigned char *tmp_c;
1345 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1346 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1347 }
1348 if (!path) path = "";
1349 sprintf(concatpath, "%s%s", base, path);
1350 one = alloc_filespec(concatpath);
1351 two = alloc_filespec(concatpath);
1352 fill_filespec(one, old_sha1, old_mode);
1353 fill_filespec(two, new_sha1, new_mode);
1354
1355 diff_queue(&diff_queued_diff, one, two);
1356 }
1357
1358 void diff_unmerge(struct diff_options *options,
1359 const char *path)
1360 {
1361 struct diff_filespec *one, *two;
1362 one = alloc_filespec(path);
1363 two = alloc_filespec(path);
1364 diff_queue(&diff_queued_diff, one, two);
1365 }