log-tree: separate major part of diff-tree.
[git/jnareb-git/bp-gitweb.git] / diff.c
blob30e4d50b951257829d2727c6076f7baa6488a473
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "quote.h"
9 #include "diff.h"
10 #include "diffcore.h"
11 #include "xdiff/xdiff.h"
13 static int use_size_cache;
15 int diff_rename_limit_default = -1;
17 int git_diff_config(const char *var, const char *value)
19 if (!strcmp(var, "diff.renamelimit")) {
20 diff_rename_limit_default = git_config_int(var, value);
21 return 0;
24 return git_default_config(var, value);
27 static char *quote_one(const char *str)
29 int needlen;
30 char *xp;
32 if (!str)
33 return NULL;
34 needlen = quote_c_style(str, NULL, NULL, 0);
35 if (!needlen)
36 return strdup(str);
37 xp = xmalloc(needlen + 1);
38 quote_c_style(str, xp, NULL, 0);
39 return xp;
42 static char *quote_two(const char *one, const char *two)
44 int need_one = quote_c_style(one, NULL, NULL, 1);
45 int need_two = quote_c_style(two, NULL, NULL, 1);
46 char *xp;
48 if (need_one + need_two) {
49 if (!need_one) need_one = strlen(one);
50 if (!need_two) need_one = strlen(two);
52 xp = xmalloc(need_one + need_two + 3);
53 xp[0] = '"';
54 quote_c_style(one, xp + 1, NULL, 1);
55 quote_c_style(two, xp + need_one + 1, NULL, 1);
56 strcpy(xp + need_one + need_two + 1, "\"");
57 return xp;
59 need_one = strlen(one);
60 need_two = strlen(two);
61 xp = xmalloc(need_one + need_two + 1);
62 strcpy(xp, one);
63 strcpy(xp + need_one, two);
64 return xp;
67 static const char *external_diff(void)
69 static const char *external_diff_cmd = NULL;
70 static int done_preparing = 0;
72 if (done_preparing)
73 return external_diff_cmd;
74 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
75 done_preparing = 1;
76 return external_diff_cmd;
79 #define TEMPFILE_PATH_LEN 50
81 static struct diff_tempfile {
82 const char *name; /* filename external diff should read from */
83 char hex[41];
84 char mode[10];
85 char tmp_path[TEMPFILE_PATH_LEN];
86 } diff_temp[2];
88 static int count_lines(const char *data, int size)
90 int count, ch, completely_empty = 1, nl_just_seen = 0;
91 count = 0;
92 while (0 < size--) {
93 ch = *data++;
94 if (ch == '\n') {
95 count++;
96 nl_just_seen = 1;
97 completely_empty = 0;
99 else {
100 nl_just_seen = 0;
101 completely_empty = 0;
104 if (completely_empty)
105 return 0;
106 if (!nl_just_seen)
107 count++; /* no trailing newline */
108 return count;
111 static void print_line_count(int count)
113 switch (count) {
114 case 0:
115 printf("0,0");
116 break;
117 case 1:
118 printf("1");
119 break;
120 default:
121 printf("1,%d", count);
122 break;
126 static void copy_file(int prefix, const char *data, int size)
128 int ch, nl_just_seen = 1;
129 while (0 < size--) {
130 ch = *data++;
131 if (nl_just_seen)
132 putchar(prefix);
133 putchar(ch);
134 if (ch == '\n')
135 nl_just_seen = 1;
136 else
137 nl_just_seen = 0;
139 if (!nl_just_seen)
140 printf("\n\\ No newline at end of file\n");
143 static void emit_rewrite_diff(const char *name_a,
144 const char *name_b,
145 struct diff_filespec *one,
146 struct diff_filespec *two)
148 int lc_a, lc_b;
149 diff_populate_filespec(one, 0);
150 diff_populate_filespec(two, 0);
151 lc_a = count_lines(one->data, one->size);
152 lc_b = count_lines(two->data, two->size);
153 printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
154 print_line_count(lc_a);
155 printf(" +");
156 print_line_count(lc_b);
157 printf(" @@\n");
158 if (lc_a)
159 copy_file('-', one->data, one->size);
160 if (lc_b)
161 copy_file('+', two->data, two->size);
164 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
166 if (!DIFF_FILE_VALID(one)) {
167 mf->ptr = ""; /* does not matter */
168 mf->size = 0;
169 return 0;
171 else if (diff_populate_filespec(one, 0))
172 return -1;
173 mf->ptr = one->data;
174 mf->size = one->size;
175 return 0;
178 struct emit_callback {
179 const char **label_path;
182 static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
184 int i;
185 struct emit_callback *ecbdata = priv;
187 if (ecbdata->label_path[0]) {
188 printf("--- %s\n", ecbdata->label_path[0]);
189 printf("+++ %s\n", ecbdata->label_path[1]);
190 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
192 for (i = 0; i < nbuf; i++)
193 if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
194 return -1;
195 return 0;
198 #define FIRST_FEW_BYTES 8000
199 static int mmfile_is_binary(mmfile_t *mf)
201 long sz = mf->size;
202 if (FIRST_FEW_BYTES < sz)
203 sz = FIRST_FEW_BYTES;
204 if (memchr(mf->ptr, 0, sz))
205 return 1;
206 return 0;
209 static void builtin_diff(const char *name_a,
210 const char *name_b,
211 struct diff_filespec *one,
212 struct diff_filespec *two,
213 const char *xfrm_msg,
214 int complete_rewrite)
216 mmfile_t mf1, mf2;
217 const char *lbl[2];
218 char *a_one, *b_two;
220 a_one = quote_two("a/", name_a);
221 b_two = quote_two("b/", name_b);
222 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
223 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
224 printf("diff --git %s %s\n", a_one, b_two);
225 if (lbl[0][0] == '/') {
226 /* /dev/null */
227 printf("new file mode %06o\n", two->mode);
228 if (xfrm_msg && xfrm_msg[0])
229 puts(xfrm_msg);
231 else if (lbl[1][0] == '/') {
232 printf("deleted file mode %06o\n", one->mode);
233 if (xfrm_msg && xfrm_msg[0])
234 puts(xfrm_msg);
236 else {
237 if (one->mode != two->mode) {
238 printf("old mode %06o\n", one->mode);
239 printf("new mode %06o\n", two->mode);
241 if (xfrm_msg && xfrm_msg[0])
242 puts(xfrm_msg);
244 * we do not run diff between different kind
245 * of objects.
247 if ((one->mode ^ two->mode) & S_IFMT)
248 goto free_ab_and_return;
249 if (complete_rewrite) {
250 emit_rewrite_diff(name_a, name_b, one, two);
251 goto free_ab_and_return;
255 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
256 die("unable to read files to diff");
258 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
259 printf("Binary files %s and %s differ\n", lbl[0], lbl[1]);
260 else {
261 /* Crazy xdl interfaces.. */
262 const char *diffopts = getenv("GIT_DIFF_OPTS");
263 xpparam_t xpp;
264 xdemitconf_t xecfg;
265 xdemitcb_t ecb;
266 struct emit_callback ecbdata;
268 ecbdata.label_path = lbl;
269 xpp.flags = XDF_NEED_MINIMAL;
270 xecfg.ctxlen = 3;
271 xecfg.flags = XDL_EMIT_FUNCNAMES;
272 if (!diffopts)
274 else if (!strncmp(diffopts, "--unified=", 10))
275 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
276 else if (!strncmp(diffopts, "-u", 2))
277 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
278 ecb.outf = fn_out;
279 ecb.priv = &ecbdata;
280 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
283 free_ab_and_return:
284 free(a_one);
285 free(b_two);
286 return;
289 struct diff_filespec *alloc_filespec(const char *path)
291 int namelen = strlen(path);
292 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
294 memset(spec, 0, sizeof(*spec));
295 spec->path = (char *)(spec + 1);
296 memcpy(spec->path, path, namelen+1);
297 return spec;
300 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
301 unsigned short mode)
303 if (mode) {
304 spec->mode = canon_mode(mode);
305 memcpy(spec->sha1, sha1, 20);
306 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
311 * Given a name and sha1 pair, if the dircache tells us the file in
312 * the work tree has that object contents, return true, so that
313 * prepare_temp_file() does not have to inflate and extract.
315 static int work_tree_matches(const char *name, const unsigned char *sha1)
317 struct cache_entry *ce;
318 struct stat st;
319 int pos, len;
321 /* We do not read the cache ourselves here, because the
322 * benchmark with my previous version that always reads cache
323 * shows that it makes things worse for diff-tree comparing
324 * two linux-2.6 kernel trees in an already checked out work
325 * tree. This is because most diff-tree comparisons deal with
326 * only a small number of files, while reading the cache is
327 * expensive for a large project, and its cost outweighs the
328 * savings we get by not inflating the object to a temporary
329 * file. Practically, this code only helps when we are used
330 * by diff-cache --cached, which does read the cache before
331 * calling us.
333 if (!active_cache)
334 return 0;
336 len = strlen(name);
337 pos = cache_name_pos(name, len);
338 if (pos < 0)
339 return 0;
340 ce = active_cache[pos];
341 if ((lstat(name, &st) < 0) ||
342 !S_ISREG(st.st_mode) || /* careful! */
343 ce_match_stat(ce, &st, 0) ||
344 memcmp(sha1, ce->sha1, 20))
345 return 0;
346 /* we return 1 only when we can stat, it is a regular file,
347 * stat information matches, and sha1 recorded in the cache
348 * matches. I.e. we know the file in the work tree really is
349 * the same as the <name, sha1> pair.
351 return 1;
354 static struct sha1_size_cache {
355 unsigned char sha1[20];
356 unsigned long size;
357 } **sha1_size_cache;
358 static int sha1_size_cache_nr, sha1_size_cache_alloc;
360 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
361 int find_only,
362 unsigned long size)
364 int first, last;
365 struct sha1_size_cache *e;
367 first = 0;
368 last = sha1_size_cache_nr;
369 while (last > first) {
370 int cmp, next = (last + first) >> 1;
371 e = sha1_size_cache[next];
372 cmp = memcmp(e->sha1, sha1, 20);
373 if (!cmp)
374 return e;
375 if (cmp < 0) {
376 last = next;
377 continue;
379 first = next+1;
381 /* not found */
382 if (find_only)
383 return NULL;
384 /* insert to make it at "first" */
385 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
386 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
387 sha1_size_cache = xrealloc(sha1_size_cache,
388 sha1_size_cache_alloc *
389 sizeof(*sha1_size_cache));
391 sha1_size_cache_nr++;
392 if (first < sha1_size_cache_nr)
393 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
394 (sha1_size_cache_nr - first - 1) *
395 sizeof(*sha1_size_cache));
396 e = xmalloc(sizeof(struct sha1_size_cache));
397 sha1_size_cache[first] = e;
398 memcpy(e->sha1, sha1, 20);
399 e->size = size;
400 return e;
404 * While doing rename detection and pickaxe operation, we may need to
405 * grab the data for the blob (or file) for our own in-core comparison.
406 * diff_filespec has data and size fields for this purpose.
408 int diff_populate_filespec(struct diff_filespec *s, int size_only)
410 int err = 0;
411 if (!DIFF_FILE_VALID(s))
412 die("internal error: asking to populate invalid file.");
413 if (S_ISDIR(s->mode))
414 return -1;
416 if (!use_size_cache)
417 size_only = 0;
419 if (s->data)
420 return err;
421 if (!s->sha1_valid ||
422 work_tree_matches(s->path, s->sha1)) {
423 struct stat st;
424 int fd;
425 if (lstat(s->path, &st) < 0) {
426 if (errno == ENOENT) {
427 err_empty:
428 err = -1;
429 empty:
430 s->data = "";
431 s->size = 0;
432 return err;
435 s->size = st.st_size;
436 if (!s->size)
437 goto empty;
438 if (size_only)
439 return 0;
440 if (S_ISLNK(st.st_mode)) {
441 int ret;
442 s->data = xmalloc(s->size);
443 s->should_free = 1;
444 ret = readlink(s->path, s->data, s->size);
445 if (ret < 0) {
446 free(s->data);
447 goto err_empty;
449 return 0;
451 fd = open(s->path, O_RDONLY);
452 if (fd < 0)
453 goto err_empty;
454 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
455 close(fd);
456 if (s->data == MAP_FAILED)
457 goto err_empty;
458 s->should_munmap = 1;
460 else {
461 char type[20];
462 struct sha1_size_cache *e;
464 if (size_only) {
465 e = locate_size_cache(s->sha1, 1, 0);
466 if (e) {
467 s->size = e->size;
468 return 0;
470 if (!sha1_object_info(s->sha1, type, &s->size))
471 locate_size_cache(s->sha1, 0, s->size);
473 else {
474 s->data = read_sha1_file(s->sha1, type, &s->size);
475 s->should_free = 1;
478 return 0;
481 void diff_free_filespec_data(struct diff_filespec *s)
483 if (s->should_free)
484 free(s->data);
485 else if (s->should_munmap)
486 munmap(s->data, s->size);
487 s->should_free = s->should_munmap = 0;
488 s->data = NULL;
489 free(s->cnt_data);
490 s->cnt_data = NULL;
493 static void prep_temp_blob(struct diff_tempfile *temp,
494 void *blob,
495 unsigned long size,
496 const unsigned char *sha1,
497 int mode)
499 int fd;
501 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
502 if (fd < 0)
503 die("unable to create temp-file");
504 if (write(fd, blob, size) != size)
505 die("unable to write temp-file");
506 close(fd);
507 temp->name = temp->tmp_path;
508 strcpy(temp->hex, sha1_to_hex(sha1));
509 temp->hex[40] = 0;
510 sprintf(temp->mode, "%06o", mode);
513 static void prepare_temp_file(const char *name,
514 struct diff_tempfile *temp,
515 struct diff_filespec *one)
517 if (!DIFF_FILE_VALID(one)) {
518 not_a_valid_file:
519 /* A '-' entry produces this for file-2, and
520 * a '+' entry produces this for file-1.
522 temp->name = "/dev/null";
523 strcpy(temp->hex, ".");
524 strcpy(temp->mode, ".");
525 return;
528 if (!one->sha1_valid ||
529 work_tree_matches(name, one->sha1)) {
530 struct stat st;
531 if (lstat(name, &st) < 0) {
532 if (errno == ENOENT)
533 goto not_a_valid_file;
534 die("stat(%s): %s", name, strerror(errno));
536 if (S_ISLNK(st.st_mode)) {
537 int ret;
538 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
539 if (sizeof(buf) <= st.st_size)
540 die("symlink too long: %s", name);
541 ret = readlink(name, buf, st.st_size);
542 if (ret < 0)
543 die("readlink(%s)", name);
544 prep_temp_blob(temp, buf, st.st_size,
545 (one->sha1_valid ?
546 one->sha1 : null_sha1),
547 (one->sha1_valid ?
548 one->mode : S_IFLNK));
550 else {
551 /* we can borrow from the file in the work tree */
552 temp->name = name;
553 if (!one->sha1_valid)
554 strcpy(temp->hex, sha1_to_hex(null_sha1));
555 else
556 strcpy(temp->hex, sha1_to_hex(one->sha1));
557 /* Even though we may sometimes borrow the
558 * contents from the work tree, we always want
559 * one->mode. mode is trustworthy even when
560 * !(one->sha1_valid), as long as
561 * DIFF_FILE_VALID(one).
563 sprintf(temp->mode, "%06o", one->mode);
565 return;
567 else {
568 if (diff_populate_filespec(one, 0))
569 die("cannot read data blob for %s", one->path);
570 prep_temp_blob(temp, one->data, one->size,
571 one->sha1, one->mode);
575 static void remove_tempfile(void)
577 int i;
579 for (i = 0; i < 2; i++)
580 if (diff_temp[i].name == diff_temp[i].tmp_path) {
581 unlink(diff_temp[i].name);
582 diff_temp[i].name = NULL;
586 static void remove_tempfile_on_signal(int signo)
588 remove_tempfile();
589 signal(SIGINT, SIG_DFL);
590 raise(signo);
593 static int spawn_prog(const char *pgm, const char **arg)
595 pid_t pid;
596 int status;
598 fflush(NULL);
599 pid = fork();
600 if (pid < 0)
601 die("unable to fork");
602 if (!pid) {
603 execvp(pgm, (char *const*) arg);
604 exit(255);
607 while (waitpid(pid, &status, 0) < 0) {
608 if (errno == EINTR)
609 continue;
610 return -1;
613 /* Earlier we did not check the exit status because
614 * diff exits non-zero if files are different, and
615 * we are not interested in knowing that. It was a
616 * mistake which made it harder to quit a diff-*
617 * session that uses the git-apply-patch-script as
618 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
619 * should also exit non-zero only when it wants to
620 * abort the entire diff-* session.
622 if (WIFEXITED(status) && !WEXITSTATUS(status))
623 return 0;
624 return -1;
627 /* An external diff command takes:
629 * diff-cmd name infile1 infile1-sha1 infile1-mode \
630 * infile2 infile2-sha1 infile2-mode [ rename-to ]
633 static void run_external_diff(const char *pgm,
634 const char *name,
635 const char *other,
636 struct diff_filespec *one,
637 struct diff_filespec *two,
638 const char *xfrm_msg,
639 int complete_rewrite)
641 const char *spawn_arg[10];
642 struct diff_tempfile *temp = diff_temp;
643 int retval;
644 static int atexit_asked = 0;
645 const char *othername;
646 const char **arg = &spawn_arg[0];
648 othername = (other? other : name);
649 if (one && two) {
650 prepare_temp_file(name, &temp[0], one);
651 prepare_temp_file(othername, &temp[1], two);
652 if (! atexit_asked &&
653 (temp[0].name == temp[0].tmp_path ||
654 temp[1].name == temp[1].tmp_path)) {
655 atexit_asked = 1;
656 atexit(remove_tempfile);
658 signal(SIGINT, remove_tempfile_on_signal);
661 if (one && two) {
662 *arg++ = pgm;
663 *arg++ = name;
664 *arg++ = temp[0].name;
665 *arg++ = temp[0].hex;
666 *arg++ = temp[0].mode;
667 *arg++ = temp[1].name;
668 *arg++ = temp[1].hex;
669 *arg++ = temp[1].mode;
670 if (other) {
671 *arg++ = other;
672 *arg++ = xfrm_msg;
674 } else {
675 *arg++ = pgm;
676 *arg++ = name;
678 *arg = NULL;
679 retval = spawn_prog(pgm, spawn_arg);
680 remove_tempfile();
681 if (retval) {
682 fprintf(stderr, "external diff died, stopping at %s.\n", name);
683 exit(1);
687 static void run_diff_cmd(const char *pgm,
688 const char *name,
689 const char *other,
690 struct diff_filespec *one,
691 struct diff_filespec *two,
692 const char *xfrm_msg,
693 int complete_rewrite)
695 if (pgm) {
696 run_external_diff(pgm, name, other, one, two, xfrm_msg,
697 complete_rewrite);
698 return;
700 if (one && two)
701 builtin_diff(name, other ? other : name,
702 one, two, xfrm_msg, complete_rewrite);
703 else
704 printf("* Unmerged path %s\n", name);
707 static void diff_fill_sha1_info(struct diff_filespec *one)
709 if (DIFF_FILE_VALID(one)) {
710 if (!one->sha1_valid) {
711 struct stat st;
712 if (lstat(one->path, &st) < 0)
713 die("stat %s", one->path);
714 if (index_path(one->sha1, one->path, &st, 0))
715 die("cannot hash %s\n", one->path);
718 else
719 memset(one->sha1, 0, 20);
722 static void run_diff(struct diff_filepair *p, struct diff_options *o)
724 const char *pgm = external_diff();
725 char msg[PATH_MAX*2+300], *xfrm_msg;
726 struct diff_filespec *one;
727 struct diff_filespec *two;
728 const char *name;
729 const char *other;
730 char *name_munged, *other_munged;
731 int complete_rewrite = 0;
732 int len;
734 if (DIFF_PAIR_UNMERGED(p)) {
735 /* unmerged */
736 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, 0);
737 return;
740 name = p->one->path;
741 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
742 name_munged = quote_one(name);
743 other_munged = quote_one(other);
744 one = p->one; two = p->two;
746 diff_fill_sha1_info(one);
747 diff_fill_sha1_info(two);
749 len = 0;
750 switch (p->status) {
751 case DIFF_STATUS_COPIED:
752 len += snprintf(msg + len, sizeof(msg) - len,
753 "similarity index %d%%\n"
754 "copy from %s\n"
755 "copy to %s\n",
756 (int)(0.5 + p->score * 100.0/MAX_SCORE),
757 name_munged, other_munged);
758 break;
759 case DIFF_STATUS_RENAMED:
760 len += snprintf(msg + len, sizeof(msg) - len,
761 "similarity index %d%%\n"
762 "rename from %s\n"
763 "rename to %s\n",
764 (int)(0.5 + p->score * 100.0/MAX_SCORE),
765 name_munged, other_munged);
766 break;
767 case DIFF_STATUS_MODIFIED:
768 if (p->score) {
769 len += snprintf(msg + len, sizeof(msg) - len,
770 "dissimilarity index %d%%\n",
771 (int)(0.5 + p->score *
772 100.0/MAX_SCORE));
773 complete_rewrite = 1;
774 break;
776 /* fallthru */
777 default:
778 /* nothing */
782 if (memcmp(one->sha1, two->sha1, 20)) {
783 char one_sha1[41];
784 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
785 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
787 len += snprintf(msg + len, sizeof(msg) - len,
788 "index %.*s..%.*s",
789 abbrev, one_sha1, abbrev,
790 sha1_to_hex(two->sha1));
791 if (one->mode == two->mode)
792 len += snprintf(msg + len, sizeof(msg) - len,
793 " %06o", one->mode);
794 len += snprintf(msg + len, sizeof(msg) - len, "\n");
797 if (len)
798 msg[--len] = 0;
799 xfrm_msg = len ? msg : NULL;
801 if (!pgm &&
802 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
803 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
804 /* a filepair that changes between file and symlink
805 * needs to be split into deletion and creation.
807 struct diff_filespec *null = alloc_filespec(two->path);
808 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, 0);
809 free(null);
810 null = alloc_filespec(one->path);
811 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, 0);
812 free(null);
814 else
815 run_diff_cmd(pgm, name, other, one, two, xfrm_msg,
816 complete_rewrite);
818 free(name_munged);
819 free(other_munged);
822 void diff_setup(struct diff_options *options)
824 memset(options, 0, sizeof(*options));
825 options->output_format = DIFF_FORMAT_RAW;
826 options->line_termination = '\n';
827 options->break_opt = -1;
828 options->rename_limit = -1;
830 options->change = diff_change;
831 options->add_remove = diff_addremove;
834 int diff_setup_done(struct diff_options *options)
836 if ((options->find_copies_harder &&
837 options->detect_rename != DIFF_DETECT_COPY) ||
838 (0 <= options->rename_limit && !options->detect_rename))
839 return -1;
840 if (options->detect_rename && options->rename_limit < 0)
841 options->rename_limit = diff_rename_limit_default;
842 if (options->setup & DIFF_SETUP_USE_CACHE) {
843 if (!active_cache)
844 /* read-cache does not die even when it fails
845 * so it is safe for us to do this here. Also
846 * it does not smudge active_cache or active_nr
847 * when it fails, so we do not have to worry about
848 * cleaning it up ourselves either.
850 read_cache();
852 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
853 use_size_cache = 1;
854 if (options->abbrev <= 0 || 40 < options->abbrev)
855 options->abbrev = 40; /* full */
857 return 0;
860 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
862 const char *arg = av[0];
863 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
864 options->output_format = DIFF_FORMAT_PATCH;
865 else if (!strcmp(arg, "-z"))
866 options->line_termination = 0;
867 else if (!strncmp(arg, "-l", 2))
868 options->rename_limit = strtoul(arg+2, NULL, 10);
869 else if (!strcmp(arg, "--full-index"))
870 options->full_index = 1;
871 else if (!strcmp(arg, "--name-only"))
872 options->output_format = DIFF_FORMAT_NAME;
873 else if (!strcmp(arg, "--name-status"))
874 options->output_format = DIFF_FORMAT_NAME_STATUS;
875 else if (!strcmp(arg, "-R"))
876 options->reverse_diff = 1;
877 else if (!strncmp(arg, "-S", 2))
878 options->pickaxe = arg + 2;
879 else if (!strcmp(arg, "-s"))
880 options->output_format = DIFF_FORMAT_NO_OUTPUT;
881 else if (!strncmp(arg, "-O", 2))
882 options->orderfile = arg + 2;
883 else if (!strncmp(arg, "--diff-filter=", 14))
884 options->filter = arg + 14;
885 else if (!strcmp(arg, "--pickaxe-all"))
886 options->pickaxe_opts = DIFF_PICKAXE_ALL;
887 else if (!strcmp(arg, "--pickaxe-regex"))
888 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
889 else if (!strncmp(arg, "-B", 2)) {
890 if ((options->break_opt =
891 diff_scoreopt_parse(arg)) == -1)
892 return -1;
894 else if (!strncmp(arg, "-M", 2)) {
895 if ((options->rename_score =
896 diff_scoreopt_parse(arg)) == -1)
897 return -1;
898 options->detect_rename = DIFF_DETECT_RENAME;
900 else if (!strncmp(arg, "-C", 2)) {
901 if ((options->rename_score =
902 diff_scoreopt_parse(arg)) == -1)
903 return -1;
904 options->detect_rename = DIFF_DETECT_COPY;
906 else if (!strcmp(arg, "--find-copies-harder"))
907 options->find_copies_harder = 1;
908 else if (!strcmp(arg, "--abbrev"))
909 options->abbrev = DEFAULT_ABBREV;
910 else if (!strncmp(arg, "--abbrev=", 9)) {
911 options->abbrev = strtoul(arg + 9, NULL, 10);
912 if (options->abbrev < MINIMUM_ABBREV)
913 options->abbrev = MINIMUM_ABBREV;
914 else if (40 < options->abbrev)
915 options->abbrev = 40;
917 else
918 return 0;
919 return 1;
922 static int parse_num(const char **cp_p)
924 unsigned long num, scale;
925 int ch, dot;
926 const char *cp = *cp_p;
928 num = 0;
929 scale = 1;
930 dot = 0;
931 for(;;) {
932 ch = *cp;
933 if ( !dot && ch == '.' ) {
934 scale = 1;
935 dot = 1;
936 } else if ( ch == '%' ) {
937 scale = dot ? scale*100 : 100;
938 cp++; /* % is always at the end */
939 break;
940 } else if ( ch >= '0' && ch <= '9' ) {
941 if ( scale < 100000 ) {
942 scale *= 10;
943 num = (num*10) + (ch-'0');
945 } else {
946 break;
948 cp++;
950 *cp_p = cp;
952 /* user says num divided by scale and we say internally that
953 * is MAX_SCORE * num / scale.
955 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
958 int diff_scoreopt_parse(const char *opt)
960 int opt1, opt2, cmd;
962 if (*opt++ != '-')
963 return -1;
964 cmd = *opt++;
965 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
966 return -1; /* that is not a -M, -C nor -B option */
968 opt1 = parse_num(&opt);
969 if (cmd != 'B')
970 opt2 = 0;
971 else {
972 if (*opt == 0)
973 opt2 = 0;
974 else if (*opt != '/')
975 return -1; /* we expect -B80/99 or -B80 */
976 else {
977 opt++;
978 opt2 = parse_num(&opt);
981 if (*opt != 0)
982 return -1;
983 return opt1 | (opt2 << 16);
986 struct diff_queue_struct diff_queued_diff;
988 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
990 if (queue->alloc <= queue->nr) {
991 queue->alloc = alloc_nr(queue->alloc);
992 queue->queue = xrealloc(queue->queue,
993 sizeof(dp) * queue->alloc);
995 queue->queue[queue->nr++] = dp;
998 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
999 struct diff_filespec *one,
1000 struct diff_filespec *two)
1002 struct diff_filepair *dp = xmalloc(sizeof(*dp));
1003 dp->one = one;
1004 dp->two = two;
1005 dp->score = 0;
1006 dp->status = 0;
1007 dp->source_stays = 0;
1008 dp->broken_pair = 0;
1009 if (queue)
1010 diff_q(queue, dp);
1011 return dp;
1014 void diff_free_filepair(struct diff_filepair *p)
1016 diff_free_filespec_data(p->one);
1017 diff_free_filespec_data(p->two);
1018 free(p->one);
1019 free(p->two);
1020 free(p);
1023 /* This is different from find_unique_abbrev() in that
1024 * it stuffs the result with dots for alignment.
1026 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1028 int abblen;
1029 const char *abbrev;
1030 if (len == 40)
1031 return sha1_to_hex(sha1);
1033 abbrev = find_unique_abbrev(sha1, len);
1034 if (!abbrev)
1035 return sha1_to_hex(sha1);
1036 abblen = strlen(abbrev);
1037 if (abblen < 37) {
1038 static char hex[41];
1039 if (len < abblen && abblen <= len + 2)
1040 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1041 else
1042 sprintf(hex, "%s...", abbrev);
1043 return hex;
1045 return sha1_to_hex(sha1);
1048 static void diff_flush_raw(struct diff_filepair *p,
1049 int line_termination,
1050 int inter_name_termination,
1051 struct diff_options *options)
1053 int two_paths;
1054 char status[10];
1055 int abbrev = options->abbrev;
1056 const char *path_one, *path_two;
1057 int output_format = options->output_format;
1059 path_one = p->one->path;
1060 path_two = p->two->path;
1061 if (line_termination) {
1062 path_one = quote_one(path_one);
1063 path_two = quote_one(path_two);
1066 if (p->score)
1067 sprintf(status, "%c%03d", p->status,
1068 (int)(0.5 + p->score * 100.0/MAX_SCORE));
1069 else {
1070 status[0] = p->status;
1071 status[1] = 0;
1073 switch (p->status) {
1074 case DIFF_STATUS_COPIED:
1075 case DIFF_STATUS_RENAMED:
1076 two_paths = 1;
1077 break;
1078 case DIFF_STATUS_ADDED:
1079 case DIFF_STATUS_DELETED:
1080 two_paths = 0;
1081 break;
1082 default:
1083 two_paths = 0;
1084 break;
1086 if (output_format != DIFF_FORMAT_NAME_STATUS) {
1087 printf(":%06o %06o %s ",
1088 p->one->mode, p->two->mode,
1089 diff_unique_abbrev(p->one->sha1, abbrev));
1090 printf("%s ",
1091 diff_unique_abbrev(p->two->sha1, abbrev));
1093 printf("%s%c%s", status, inter_name_termination, path_one);
1094 if (two_paths)
1095 printf("%c%s", inter_name_termination, path_two);
1096 putchar(line_termination);
1097 if (path_one != p->one->path)
1098 free((void*)path_one);
1099 if (path_two != p->two->path)
1100 free((void*)path_two);
1103 static void diff_flush_name(struct diff_filepair *p,
1104 int inter_name_termination,
1105 int line_termination)
1107 char *path = p->two->path;
1109 if (line_termination)
1110 path = quote_one(p->two->path);
1111 else
1112 path = p->two->path;
1113 printf("%s%c", path, line_termination);
1114 if (p->two->path != path)
1115 free(path);
1118 int diff_unmodified_pair(struct diff_filepair *p)
1120 /* This function is written stricter than necessary to support
1121 * the currently implemented transformers, but the idea is to
1122 * let transformers to produce diff_filepairs any way they want,
1123 * and filter and clean them up here before producing the output.
1125 struct diff_filespec *one, *two;
1127 if (DIFF_PAIR_UNMERGED(p))
1128 return 0; /* unmerged is interesting */
1130 one = p->one;
1131 two = p->two;
1133 /* deletion, addition, mode or type change
1134 * and rename are all interesting.
1136 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1137 DIFF_PAIR_MODE_CHANGED(p) ||
1138 strcmp(one->path, two->path))
1139 return 0;
1141 /* both are valid and point at the same path. that is, we are
1142 * dealing with a change.
1144 if (one->sha1_valid && two->sha1_valid &&
1145 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1146 return 1; /* no change */
1147 if (!one->sha1_valid && !two->sha1_valid)
1148 return 1; /* both look at the same file on the filesystem. */
1149 return 0;
1152 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1154 if (diff_unmodified_pair(p))
1155 return;
1157 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1158 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1159 return; /* no tree diffs in patch format */
1161 run_diff(p, o);
1164 int diff_queue_is_empty(void)
1166 struct diff_queue_struct *q = &diff_queued_diff;
1167 int i;
1168 for (i = 0; i < q->nr; i++)
1169 if (!diff_unmodified_pair(q->queue[i]))
1170 return 0;
1171 return 1;
1174 #if DIFF_DEBUG
1175 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1177 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1178 x, one ? one : "",
1179 s->path,
1180 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1181 s->mode,
1182 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1183 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1184 x, one ? one : "",
1185 s->size, s->xfrm_flags);
1188 void diff_debug_filepair(const struct diff_filepair *p, int i)
1190 diff_debug_filespec(p->one, i, "one");
1191 diff_debug_filespec(p->two, i, "two");
1192 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1193 p->score, p->status ? p->status : '?',
1194 p->source_stays, p->broken_pair);
1197 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1199 int i;
1200 if (msg)
1201 fprintf(stderr, "%s\n", msg);
1202 fprintf(stderr, "q->nr = %d\n", q->nr);
1203 for (i = 0; i < q->nr; i++) {
1204 struct diff_filepair *p = q->queue[i];
1205 diff_debug_filepair(p, i);
1208 #endif
1210 static void diff_resolve_rename_copy(void)
1212 int i, j;
1213 struct diff_filepair *p, *pp;
1214 struct diff_queue_struct *q = &diff_queued_diff;
1216 diff_debug_queue("resolve-rename-copy", q);
1218 for (i = 0; i < q->nr; i++) {
1219 p = q->queue[i];
1220 p->status = 0; /* undecided */
1221 if (DIFF_PAIR_UNMERGED(p))
1222 p->status = DIFF_STATUS_UNMERGED;
1223 else if (!DIFF_FILE_VALID(p->one))
1224 p->status = DIFF_STATUS_ADDED;
1225 else if (!DIFF_FILE_VALID(p->two))
1226 p->status = DIFF_STATUS_DELETED;
1227 else if (DIFF_PAIR_TYPE_CHANGED(p))
1228 p->status = DIFF_STATUS_TYPE_CHANGED;
1230 /* from this point on, we are dealing with a pair
1231 * whose both sides are valid and of the same type, i.e.
1232 * either in-place edit or rename/copy edit.
1234 else if (DIFF_PAIR_RENAME(p)) {
1235 if (p->source_stays) {
1236 p->status = DIFF_STATUS_COPIED;
1237 continue;
1239 /* See if there is some other filepair that
1240 * copies from the same source as us. If so
1241 * we are a copy. Otherwise we are either a
1242 * copy if the path stays, or a rename if it
1243 * does not, but we already handled "stays" case.
1245 for (j = i + 1; j < q->nr; j++) {
1246 pp = q->queue[j];
1247 if (strcmp(pp->one->path, p->one->path))
1248 continue; /* not us */
1249 if (!DIFF_PAIR_RENAME(pp))
1250 continue; /* not a rename/copy */
1251 /* pp is a rename/copy from the same source */
1252 p->status = DIFF_STATUS_COPIED;
1253 break;
1255 if (!p->status)
1256 p->status = DIFF_STATUS_RENAMED;
1258 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1259 p->one->mode != p->two->mode)
1260 p->status = DIFF_STATUS_MODIFIED;
1261 else {
1262 /* This is a "no-change" entry and should not
1263 * happen anymore, but prepare for broken callers.
1265 error("feeding unmodified %s to diffcore",
1266 p->one->path);
1267 p->status = DIFF_STATUS_UNKNOWN;
1270 diff_debug_queue("resolve-rename-copy done", q);
1273 void diff_flush(struct diff_options *options)
1275 struct diff_queue_struct *q = &diff_queued_diff;
1276 int i;
1277 int inter_name_termination = '\t';
1278 int diff_output_format = options->output_format;
1279 int line_termination = options->line_termination;
1281 if (!line_termination)
1282 inter_name_termination = 0;
1284 for (i = 0; i < q->nr; i++) {
1285 struct diff_filepair *p = q->queue[i];
1287 switch (p->status) {
1288 case DIFF_STATUS_UNKNOWN:
1289 break;
1290 case 0:
1291 die("internal error in diff-resolve-rename-copy");
1292 break;
1293 default:
1294 switch (diff_output_format) {
1295 case DIFF_FORMAT_PATCH:
1296 diff_flush_patch(p, options);
1297 break;
1298 case DIFF_FORMAT_RAW:
1299 case DIFF_FORMAT_NAME_STATUS:
1300 diff_flush_raw(p, line_termination,
1301 inter_name_termination,
1302 options);
1303 break;
1304 case DIFF_FORMAT_NAME:
1305 diff_flush_name(p,
1306 inter_name_termination,
1307 line_termination);
1308 break;
1309 case DIFF_FORMAT_NO_OUTPUT:
1310 break;
1313 diff_free_filepair(p);
1315 free(q->queue);
1316 q->queue = NULL;
1317 q->nr = q->alloc = 0;
1320 static void diffcore_apply_filter(const char *filter)
1322 int i;
1323 struct diff_queue_struct *q = &diff_queued_diff;
1324 struct diff_queue_struct outq;
1325 outq.queue = NULL;
1326 outq.nr = outq.alloc = 0;
1328 if (!filter)
1329 return;
1331 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1332 int found;
1333 for (i = found = 0; !found && i < q->nr; i++) {
1334 struct diff_filepair *p = q->queue[i];
1335 if (((p->status == DIFF_STATUS_MODIFIED) &&
1336 ((p->score &&
1337 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1338 (!p->score &&
1339 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1340 ((p->status != DIFF_STATUS_MODIFIED) &&
1341 strchr(filter, p->status)))
1342 found++;
1344 if (found)
1345 return;
1347 /* otherwise we will clear the whole queue
1348 * by copying the empty outq at the end of this
1349 * function, but first clear the current entries
1350 * in the queue.
1352 for (i = 0; i < q->nr; i++)
1353 diff_free_filepair(q->queue[i]);
1355 else {
1356 /* Only the matching ones */
1357 for (i = 0; i < q->nr; i++) {
1358 struct diff_filepair *p = q->queue[i];
1360 if (((p->status == DIFF_STATUS_MODIFIED) &&
1361 ((p->score &&
1362 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1363 (!p->score &&
1364 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1365 ((p->status != DIFF_STATUS_MODIFIED) &&
1366 strchr(filter, p->status)))
1367 diff_q(&outq, p);
1368 else
1369 diff_free_filepair(p);
1372 free(q->queue);
1373 *q = outq;
1376 void diffcore_std(struct diff_options *options)
1378 if (options->paths && options->paths[0])
1379 diffcore_pathspec(options->paths);
1380 if (options->break_opt != -1)
1381 diffcore_break(options->break_opt);
1382 if (options->detect_rename)
1383 diffcore_rename(options);
1384 if (options->break_opt != -1)
1385 diffcore_merge_broken();
1386 if (options->pickaxe)
1387 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1388 if (options->orderfile)
1389 diffcore_order(options->orderfile);
1390 diff_resolve_rename_copy();
1391 diffcore_apply_filter(options->filter);
1395 void diffcore_std_no_resolve(struct diff_options *options)
1397 if (options->pickaxe)
1398 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1399 if (options->orderfile)
1400 diffcore_order(options->orderfile);
1401 diffcore_apply_filter(options->filter);
1404 void diff_addremove(struct diff_options *options,
1405 int addremove, unsigned mode,
1406 const unsigned char *sha1,
1407 const char *base, const char *path)
1409 char concatpath[PATH_MAX];
1410 struct diff_filespec *one, *two;
1412 /* This may look odd, but it is a preparation for
1413 * feeding "there are unchanged files which should
1414 * not produce diffs, but when you are doing copy
1415 * detection you would need them, so here they are"
1416 * entries to the diff-core. They will be prefixed
1417 * with something like '=' or '*' (I haven't decided
1418 * which but should not make any difference).
1419 * Feeding the same new and old to diff_change()
1420 * also has the same effect.
1421 * Before the final output happens, they are pruned after
1422 * merged into rename/copy pairs as appropriate.
1424 if (options->reverse_diff)
1425 addremove = (addremove == '+' ? '-' :
1426 addremove == '-' ? '+' : addremove);
1428 if (!path) path = "";
1429 sprintf(concatpath, "%s%s", base, path);
1430 one = alloc_filespec(concatpath);
1431 two = alloc_filespec(concatpath);
1433 if (addremove != '+')
1434 fill_filespec(one, sha1, mode);
1435 if (addremove != '-')
1436 fill_filespec(two, sha1, mode);
1438 diff_queue(&diff_queued_diff, one, two);
1441 void diff_change(struct diff_options *options,
1442 unsigned old_mode, unsigned new_mode,
1443 const unsigned char *old_sha1,
1444 const unsigned char *new_sha1,
1445 const char *base, const char *path)
1447 char concatpath[PATH_MAX];
1448 struct diff_filespec *one, *two;
1450 if (options->reverse_diff) {
1451 unsigned tmp;
1452 const unsigned char *tmp_c;
1453 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1454 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1456 if (!path) path = "";
1457 sprintf(concatpath, "%s%s", base, path);
1458 one = alloc_filespec(concatpath);
1459 two = alloc_filespec(concatpath);
1460 fill_filespec(one, old_sha1, old_mode);
1461 fill_filespec(two, new_sha1, new_mode);
1463 diff_queue(&diff_queued_diff, one, two);
1466 void diff_unmerge(struct diff_options *options,
1467 const char *path)
1469 struct diff_filespec *one, *two;
1470 one = alloc_filespec(path);
1471 two = alloc_filespec(path);
1472 diff_queue(&diff_queued_diff, one, two);