[PATCH] Prepare diffcore interface for diff-tree header supression.
[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 <limits.h>
8 #include "cache.h"
9 #include "diff.h"
10 #include "diffcore.h"
11
12 static const char *diff_opts = "-pu";
13 static unsigned char null_sha1[20] = { 0, };
14
15 static int reverse_diff;
16 static int diff_raw_output = -1;
17 static const char **pathspec;
18 static int speccnt;
19
20 static const char *external_diff(void)
21 {
22 static const char *external_diff_cmd = NULL;
23 static int done_preparing = 0;
24
25 if (done_preparing)
26 return external_diff_cmd;
27
28 /*
29 * Default values above are meant to match the
30 * Linux kernel development style. Examples of
31 * alternative styles you can specify via environment
32 * variables are:
33 *
34 * GIT_DIFF_OPTS="-c";
35 */
36 if (gitenv("GIT_EXTERNAL_DIFF"))
37 external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
38
39 /* In case external diff fails... */
40 diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
41
42 done_preparing = 1;
43 return external_diff_cmd;
44 }
45
46 /* Help to copy the thing properly quoted for the shell safety.
47 * any single quote is replaced with '\'', and the caller is
48 * expected to enclose the result within a single quote pair.
49 *
50 * E.g.
51 * original sq_expand result
52 * name ==> name ==> 'name'
53 * a b ==> a b ==> 'a b'
54 * a'b ==> a'\''b ==> 'a'\''b'
55 */
56 static char *sq_expand(const char *src)
57 {
58 static char *buf = NULL;
59 int cnt, c;
60 const char *cp;
61 char *bp;
62
63 /* count bytes needed to store the quoted string. */
64 for (cnt = 1, cp = src; *cp; cnt++, cp++)
65 if (*cp == '\'')
66 cnt += 3;
67
68 buf = xmalloc(cnt);
69 bp = buf;
70 while ((c = *src++)) {
71 if (c != '\'')
72 *bp++ = c;
73 else {
74 bp = strcpy(bp, "'\\''");
75 bp += 4;
76 }
77 }
78 *bp = 0;
79 return buf;
80 }
81
82 static struct diff_tempfile {
83 const char *name; /* filename external diff should read from */
84 char hex[41];
85 char mode[10];
86 char tmp_path[50];
87 } diff_temp[2];
88
89 static void builtin_diff(const char *name_a,
90 const char *name_b,
91 struct diff_tempfile *temp,
92 const char *xfrm_msg)
93 {
94 int i, next_at, cmd_size;
95 const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
96 const char *diff_arg = "'%s' '%s'||:"; /* "||:" is to return 0 */
97 const char *input_name_sq[2];
98 const char *path0[2];
99 const char *path1[2];
100 const char *name_sq[2];
101 char *cmd;
102
103 name_sq[0] = sq_expand(name_a);
104 name_sq[1] = sq_expand(name_b);
105
106 /* diff_cmd and diff_arg have 6 %s in total which makes
107 * the sum of these strings 12 bytes larger than required.
108 * we use 2 spaces around diff-opts, and we need to count
109 * terminating NUL, so we subtract 9 here.
110 */
111 cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
112 strlen(diff_arg) - 9);
113 for (i = 0; i < 2; i++) {
114 input_name_sq[i] = sq_expand(temp[i].name);
115 if (!strcmp(temp[i].name, "/dev/null")) {
116 path0[i] = "/dev/null";
117 path1[i] = "";
118 } else {
119 path0[i] = i ? "b/" : "a/";
120 path1[i] = name_sq[i];
121 }
122 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
123 strlen(input_name_sq[i]));
124 }
125
126 cmd = xmalloc(cmd_size);
127
128 next_at = 0;
129 next_at += snprintf(cmd+next_at, cmd_size-next_at,
130 diff_cmd,
131 path0[0], path1[0], path0[1], path1[1]);
132 next_at += snprintf(cmd+next_at, cmd_size-next_at,
133 " %s ", diff_opts);
134 next_at += snprintf(cmd+next_at, cmd_size-next_at,
135 diff_arg, input_name_sq[0], input_name_sq[1]);
136
137 printf("diff --git a/%s b/%s\n", name_a, name_b);
138 if (!path1[0][0])
139 printf("new file mode %s\n", temp[1].mode);
140 else if (!path1[1][0])
141 printf("deleted file mode %s\n", temp[0].mode);
142 else {
143 if (strcmp(temp[0].mode, temp[1].mode)) {
144 printf("old mode %s\n", temp[0].mode);
145 printf("new mode %s\n", temp[1].mode);
146 }
147 if (xfrm_msg && xfrm_msg[0])
148 fputs(xfrm_msg, stdout);
149
150 if (strncmp(temp[0].mode, temp[1].mode, 3))
151 /* we do not run diff between different kind
152 * of objects.
153 */
154 exit(0);
155 }
156 fflush(NULL);
157 execlp("/bin/sh","sh", "-c", cmd, NULL);
158 }
159
160 struct diff_filespec *alloc_filespec(const char *path)
161 {
162 int namelen = strlen(path);
163 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
164 spec->path = (char *)(spec + 1);
165 strcpy(spec->path, path);
166 spec->should_free = spec->should_munmap = spec->file_valid = 0;
167 spec->xfrm_flags = 0;
168 spec->size = 0;
169 spec->data = 0;
170 return spec;
171 }
172
173 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
174 unsigned short mode)
175 {
176 spec->mode = mode;
177 memcpy(spec->sha1, sha1, 20);
178 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
179 spec->file_valid = 1;
180 }
181
182 /*
183 * Given a name and sha1 pair, if the dircache tells us the file in
184 * the work tree has that object contents, return true, so that
185 * prepare_temp_file() does not have to inflate and extract.
186 */
187 static int work_tree_matches(const char *name, const unsigned char *sha1)
188 {
189 struct cache_entry *ce;
190 struct stat st;
191 int pos, len;
192
193 /* We do not read the cache ourselves here, because the
194 * benchmark with my previous version that always reads cache
195 * shows that it makes things worse for diff-tree comparing
196 * two linux-2.6 kernel trees in an already checked out work
197 * tree. This is because most diff-tree comparisons deal with
198 * only a small number of files, while reading the cache is
199 * expensive for a large project, and its cost outweighs the
200 * savings we get by not inflating the object to a temporary
201 * file. Practically, this code only helps when we are used
202 * by diff-cache --cached, which does read the cache before
203 * calling us.
204 */
205 if (!active_cache)
206 return 0;
207
208 len = strlen(name);
209 pos = cache_name_pos(name, len);
210 if (pos < 0)
211 return 0;
212 ce = active_cache[pos];
213 if ((lstat(name, &st) < 0) ||
214 !S_ISREG(st.st_mode) || /* careful! */
215 ce_match_stat(ce, &st) ||
216 memcmp(sha1, ce->sha1, 20))
217 return 0;
218 /* we return 1 only when we can stat, it is a regular file,
219 * stat information matches, and sha1 recorded in the cache
220 * matches. I.e. we know the file in the work tree really is
221 * the same as the <name, sha1> pair.
222 */
223 return 1;
224 }
225
226 /*
227 * While doing rename detection and pickaxe operation, we may need to
228 * grab the data for the blob (or file) for our own in-core comparison.
229 * diff_filespec has data and size fields for this purpose.
230 */
231 int diff_populate_filespec(struct diff_filespec *s)
232 {
233 int err = 0;
234 if (!s->file_valid)
235 die("internal error: asking to populate invalid file.");
236 if (S_ISDIR(s->mode))
237 return -1;
238
239 if (s->data)
240 return err;
241 if (!s->sha1_valid ||
242 work_tree_matches(s->path, s->sha1)) {
243 struct stat st;
244 int fd;
245 if (lstat(s->path, &st) < 0) {
246 if (errno == ENOENT) {
247 err_empty:
248 err = -1;
249 empty:
250 s->data = "";
251 s->size = 0;
252 return err;
253 }
254 }
255 s->size = st.st_size;
256 if (!s->size)
257 goto empty;
258 if (S_ISLNK(st.st_mode)) {
259 int ret;
260 s->data = xmalloc(s->size);
261 s->should_free = 1;
262 ret = readlink(s->path, s->data, s->size);
263 if (ret < 0) {
264 free(s->data);
265 goto err_empty;
266 }
267 return 0;
268 }
269 fd = open(s->path, O_RDONLY);
270 if (fd < 0)
271 goto err_empty;
272 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
273 s->should_munmap = 1;
274 close(fd);
275 }
276 else {
277 char type[20];
278 s->data = read_sha1_file(s->sha1, type, &s->size);
279 s->should_free = 1;
280 }
281 return 0;
282 }
283
284 void diff_free_filespec_data(struct diff_filespec *s)
285 {
286 if (s->should_free)
287 free(s->data);
288 else if (s->should_munmap)
289 munmap(s->data, s->size);
290 s->should_free = s->should_munmap = 0;
291 s->data = 0;
292 }
293
294 static void prep_temp_blob(struct diff_tempfile *temp,
295 void *blob,
296 unsigned long size,
297 unsigned char *sha1,
298 int mode)
299 {
300 int fd;
301
302 strcpy(temp->tmp_path, ".diff_XXXXXX");
303 fd = mkstemp(temp->tmp_path);
304 if (fd < 0)
305 die("unable to create temp-file");
306 if (write(fd, blob, size) != size)
307 die("unable to write temp-file");
308 close(fd);
309 temp->name = temp->tmp_path;
310 strcpy(temp->hex, sha1_to_hex(sha1));
311 temp->hex[40] = 0;
312 sprintf(temp->mode, "%06o", mode);
313 }
314
315 static void prepare_temp_file(const char *name,
316 struct diff_tempfile *temp,
317 struct diff_filespec *one)
318 {
319 if (!one->file_valid) {
320 not_a_valid_file:
321 /* A '-' entry produces this for file-2, and
322 * a '+' entry produces this for file-1.
323 */
324 temp->name = "/dev/null";
325 strcpy(temp->hex, ".");
326 strcpy(temp->mode, ".");
327 return;
328 }
329
330 if (!one->sha1_valid ||
331 work_tree_matches(name, one->sha1)) {
332 struct stat st;
333 if (lstat(name, &st) < 0) {
334 if (errno == ENOENT)
335 goto not_a_valid_file;
336 die("stat(%s): %s", name, strerror(errno));
337 }
338 if (S_ISLNK(st.st_mode)) {
339 int ret;
340 char *buf, buf_[1024];
341 buf = ((sizeof(buf_) < st.st_size) ?
342 xmalloc(st.st_size) : buf_);
343 ret = readlink(name, buf, st.st_size);
344 if (ret < 0)
345 die("readlink(%s)", name);
346 prep_temp_blob(temp, buf, st.st_size,
347 (one->sha1_valid ?
348 one->sha1 : null_sha1),
349 (one->sha1_valid ?
350 one->mode : S_IFLNK));
351 }
352 else {
353 /* we can borrow from the file in the work tree */
354 temp->name = name;
355 if (!one->sha1_valid)
356 strcpy(temp->hex, sha1_to_hex(null_sha1));
357 else
358 strcpy(temp->hex, sha1_to_hex(one->sha1));
359 sprintf(temp->mode, "%06o",
360 S_IFREG |ce_permissions(st.st_mode));
361 }
362 return;
363 }
364 else {
365 if (diff_populate_filespec(one))
366 die("cannot read data blob for %s", one->path);
367 prep_temp_blob(temp, one->data, one->size,
368 one->sha1, one->mode);
369 }
370 }
371
372 static void remove_tempfile(void)
373 {
374 int i;
375
376 for (i = 0; i < 2; i++)
377 if (diff_temp[i].name == diff_temp[i].tmp_path) {
378 unlink(diff_temp[i].name);
379 diff_temp[i].name = NULL;
380 }
381 }
382
383 static void remove_tempfile_on_signal(int signo)
384 {
385 remove_tempfile();
386 }
387
388 static int matches_pathspec(const char *name)
389 {
390 int i;
391 int namelen;
392
393 if (speccnt == 0)
394 return 1;
395
396 namelen = strlen(name);
397 for (i = 0; i < speccnt; i++) {
398 int speclen = strlen(pathspec[i]);
399 if (! strncmp(pathspec[i], name, speclen) &&
400 speclen <= namelen &&
401 (name[speclen] == 0 || name[speclen] == '/'))
402 return 1;
403 }
404 return 0;
405 }
406
407 /* An external diff command takes:
408 *
409 * diff-cmd name infile1 infile1-sha1 infile1-mode \
410 * infile2 infile2-sha1 infile2-mode [ rename-to ]
411 *
412 */
413 static void run_external_diff(const char *name,
414 const char *other,
415 struct diff_filespec *one,
416 struct diff_filespec *two,
417 const char *xfrm_msg)
418 {
419 struct diff_tempfile *temp = diff_temp;
420 pid_t pid;
421 int status;
422 static int atexit_asked = 0;
423
424 if (!matches_pathspec(name) && (!other || !matches_pathspec(other)))
425 return;
426
427 if (one && two) {
428 prepare_temp_file(name, &temp[0], one);
429 prepare_temp_file(other ? : name, &temp[1], two);
430 if (! atexit_asked &&
431 (temp[0].name == temp[0].tmp_path ||
432 temp[1].name == temp[1].tmp_path)) {
433 atexit_asked = 1;
434 atexit(remove_tempfile);
435 }
436 signal(SIGINT, remove_tempfile_on_signal);
437 }
438
439 fflush(NULL);
440 pid = fork();
441 if (pid < 0)
442 die("unable to fork");
443 if (!pid) {
444 const char *pgm = external_diff();
445 if (pgm) {
446 if (one && two) {
447 const char *exec_arg[10];
448 const char **arg = &exec_arg[0];
449 *arg++ = pgm;
450 *arg++ = name;
451 *arg++ = temp[0].name;
452 *arg++ = temp[0].hex;
453 *arg++ = temp[0].mode;
454 *arg++ = temp[1].name;
455 *arg++ = temp[1].hex;
456 *arg++ = temp[1].mode;
457 if (other) {
458 *arg++ = other;
459 *arg++ = xfrm_msg;
460 }
461 *arg = 0;
462 execvp(pgm, (char *const*) exec_arg);
463 }
464 else
465 execlp(pgm, pgm, name, NULL);
466 }
467 /*
468 * otherwise we use the built-in one.
469 */
470 if (one && two)
471 builtin_diff(name, other ? : name, temp, xfrm_msg);
472 else
473 printf("* Unmerged path %s\n", name);
474 exit(0);
475 }
476 if (waitpid(pid, &status, 0) < 0 ||
477 !WIFEXITED(status) || WEXITSTATUS(status)) {
478 /* Earlier we did not check the exit status because
479 * diff exits non-zero if files are different, and
480 * we are not interested in knowing that. It was a
481 * mistake which made it harder to quit a diff-*
482 * session that uses the git-apply-patch-script as
483 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
484 * should also exit non-zero only when it wants to
485 * abort the entire diff-* session.
486 */
487 remove_tempfile();
488 fprintf(stderr, "external diff died, stopping at %s.\n", name);
489 exit(1);
490 }
491 remove_tempfile();
492 }
493
494 int diff_scoreopt_parse(const char *opt)
495 {
496 int diglen, num, scale, i;
497 if (opt[0] != '-' || (opt[1] != 'M' && opt[1] != 'C'))
498 return -1; /* that is not a -M nor -C option */
499 diglen = strspn(opt+2, "0123456789");
500 if (diglen == 0 || strlen(opt+2) != diglen)
501 return 0; /* use default */
502 sscanf(opt+2, "%d", &num);
503 for (i = 0, scale = 1; i < diglen; i++)
504 scale *= 10;
505
506 /* user says num divided by scale and we say internally that
507 * is MAX_SCORE * num / scale.
508 */
509 return MAX_SCORE * num / scale;
510 }
511
512 void diff_setup(int reverse_diff_, int diff_raw_output_)
513 {
514 reverse_diff = reverse_diff_;
515 diff_raw_output = diff_raw_output_;
516 }
517
518 struct diff_queue_struct diff_queued_diff;
519
520 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
521 struct diff_filespec *one,
522 struct diff_filespec *two)
523 {
524 struct diff_filepair *dp = xmalloc(sizeof(*dp));
525 dp->one = one;
526 dp->two = two;
527 dp->xfrm_msg = 0;
528 dp->orig_order = queue->nr;
529 dp->xfrm_work = 0;
530 if (queue->alloc <= queue->nr) {
531 queue->alloc = alloc_nr(queue->alloc);
532 queue->queue = xrealloc(queue->queue,
533 sizeof(dp) * queue->alloc);
534 }
535 queue->queue[queue->nr++] = dp;
536 return dp;
537 }
538
539 static const char *git_object_type(unsigned mode)
540 {
541 return S_ISDIR(mode) ? "tree" : "blob";
542 }
543
544 static void diff_flush_raw(struct diff_filepair *p)
545 {
546 struct diff_filespec *it;
547 int addremove;
548
549 /* raw output does not have a way to express rename nor copy */
550 if (strcmp(p->one->path, p->two->path))
551 return;
552
553 if (p->one->file_valid && p->two->file_valid) {
554 char hex[41];
555 strcpy(hex, sha1_to_hex(p->one->sha1));
556 printf("*%06o->%06o %s %s->%s %s%c",
557 p->one->mode, p->two->mode,
558 git_object_type(p->one->mode),
559 hex, sha1_to_hex(p->two->sha1),
560 p->one->path, diff_raw_output);
561 return;
562 }
563
564 if (p->one->file_valid) {
565 it = p->one;
566 addremove = '-';
567 } else {
568 it = p->two;
569 addremove = '+';
570 }
571
572 printf("%c%06o %s %s %s%c",
573 addremove,
574 it->mode, git_object_type(it->mode),
575 sha1_to_hex(it->sha1), it->path, diff_raw_output);
576 }
577
578 static void diff_flush_patch(struct diff_filepair *p)
579 {
580 const char *name, *other;
581
582 name = p->one->path;
583 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
584 if ((p->one->file_valid && S_ISDIR(p->one->mode)) ||
585 (p->two->file_valid && S_ISDIR(p->two->mode)))
586 return; /* no tree diffs in patch format */
587
588 run_external_diff(name, other, p->one, p->two, p->xfrm_msg);
589 }
590
591 static int identical(struct diff_filespec *one, struct diff_filespec *two)
592 {
593 /* This function is written stricter than necessary to support
594 * the currently implemented transformers, but the idea is to
595 * let transformers to produce diff_filepairs any way they want,
596 * and filter and clean them up here before producing the output.
597 */
598
599 if (!one->file_valid && !two->file_valid)
600 return 1; /* not interesting */
601
602 /* deletion, addition, mode change and renames are all interesting. */
603 if ((one->file_valid != two->file_valid) || (one->mode != two->mode) ||
604 strcmp(one->path, two->path))
605 return 0;
606
607 /* both are valid and point at the same path. that is, we are
608 * dealing with a change.
609 */
610 if (one->sha1_valid && two->sha1_valid &&
611 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
612 return 1; /* no change */
613 if (!one->sha1_valid && !two->sha1_valid)
614 return 1; /* both look at the same file on the filesystem. */
615 return 0;
616 }
617
618 static void diff_flush_one(struct diff_filepair *p)
619 {
620 if (identical(p->one, p->two))
621 return;
622 if (0 <= diff_raw_output)
623 diff_flush_raw(p);
624 else
625 diff_flush_patch(p);
626 }
627
628 int diff_queue_is_empty(void)
629 {
630 struct diff_queue_struct *q = &diff_queued_diff;
631 int i;
632
633 for (i = 0; i < q->nr; i++) {
634 struct diff_filepair *p = q->queue[i];
635 if (!identical(p->one, p->two))
636 return 0;
637 }
638 return 1;
639 }
640
641 void diff_flush(const char **pathspec_, int speccnt_)
642 {
643 struct diff_queue_struct *q = &diff_queued_diff;
644 int i;
645
646 pathspec = pathspec_;
647 speccnt = speccnt_;
648
649 for (i = 0; i < q->nr; i++)
650 diff_flush_one(q->queue[i]);
651
652 for (i = 0; i < q->nr; i++) {
653 struct diff_filepair *p = q->queue[i];
654 diff_free_filespec_data(p->one);
655 diff_free_filespec_data(p->two);
656 free(p->xfrm_msg);
657 free(p);
658 }
659 free(q->queue);
660 q->queue = NULL;
661 q->nr = q->alloc = 0;
662 }
663
664 void diff_addremove(int addremove, unsigned mode,
665 const unsigned char *sha1,
666 const char *base, const char *path)
667 {
668 char concatpath[PATH_MAX];
669 struct diff_filespec *one, *two;
670
671 /* This may look odd, but it is a preparation for
672 * feeding "there are unchanged files which should
673 * not produce diffs, but when you are doing copy
674 * detection you would need them, so here they are"
675 * entries to the diff-core. They will be prefixed
676 * with something like '=' or '*' (I haven't decided
677 * which but should not make any difference).
678 * Feeding the same new and old to diff_change() should
679 * also have the same effect. diff_flush() should
680 * filter the identical ones out at the final output
681 * stage.
682 */
683 if (reverse_diff)
684 addremove = (addremove == '+' ? '-' :
685 addremove == '-' ? '+' : addremove);
686
687 if (!path) path = "";
688 sprintf(concatpath, "%s%s", base, path);
689 one = alloc_filespec(concatpath);
690 two = alloc_filespec(concatpath);
691
692 if (addremove != '+')
693 fill_filespec(one, sha1, mode);
694 if (addremove != '-')
695 fill_filespec(two, sha1, mode);
696
697 diff_queue(&diff_queued_diff, one, two);
698 }
699
700 void diff_change(unsigned old_mode, unsigned new_mode,
701 const unsigned char *old_sha1,
702 const unsigned char *new_sha1,
703 const char *base, const char *path) {
704 char concatpath[PATH_MAX];
705 struct diff_filespec *one, *two;
706
707 if (reverse_diff) {
708 unsigned tmp;
709 const unsigned char *tmp_c;
710 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
711 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
712 }
713 if (!path) path = "";
714 sprintf(concatpath, "%s%s", base, path);
715 one = alloc_filespec(concatpath);
716 two = alloc_filespec(concatpath);
717 fill_filespec(one, old_sha1, old_mode);
718 fill_filespec(two, new_sha1, new_mode);
719
720 diff_queue(&diff_queued_diff, one, two);
721 }
722
723 void diff_unmerge(const char *path)
724 {
725 if (0 <= diff_raw_output) {
726 printf("U %s%c", path, diff_raw_output);
727 return;
728 }
729 run_external_diff(path, NULL, NULL, NULL, NULL);
730 }