Merge branch 'lt/diffgen' into next
[git/jnareb-git.git] / diff.c
blob4440465b4781cb62bdd6a0ea3d04617287cc5ca1
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 const char *diff_opts = "-pu";
15 static int use_size_cache;
17 int diff_rename_limit_default = -1;
19 int git_diff_config(const char *var, const char *value)
21 if (!strcmp(var, "diff.renamelimit")) {
22 diff_rename_limit_default = git_config_int(var, value);
23 return 0;
26 return git_default_config(var, value);
29 static char *quote_one(const char *str)
31 int needlen;
32 char *xp;
34 if (!str)
35 return NULL;
36 needlen = quote_c_style(str, NULL, NULL, 0);
37 if (!needlen)
38 return strdup(str);
39 xp = xmalloc(needlen + 1);
40 quote_c_style(str, xp, NULL, 0);
41 return xp;
44 static char *quote_two(const char *one, const char *two)
46 int need_one = quote_c_style(one, NULL, NULL, 1);
47 int need_two = quote_c_style(two, NULL, NULL, 1);
48 char *xp;
50 if (need_one + need_two) {
51 if (!need_one) need_one = strlen(one);
52 if (!need_two) need_one = strlen(two);
54 xp = xmalloc(need_one + need_two + 3);
55 xp[0] = '"';
56 quote_c_style(one, xp + 1, NULL, 1);
57 quote_c_style(two, xp + need_one + 1, NULL, 1);
58 strcpy(xp + need_one + need_two + 1, "\"");
59 return xp;
61 need_one = strlen(one);
62 need_two = strlen(two);
63 xp = xmalloc(need_one + need_two + 1);
64 strcpy(xp, one);
65 strcpy(xp + need_one, two);
66 return xp;
69 static const char *external_diff(void)
71 static const char *external_diff_cmd = NULL;
72 static int done_preparing = 0;
73 const char *env_diff_opts;
75 if (done_preparing)
76 return external_diff_cmd;
79 * Default values above are meant to match the
80 * Linux kernel development style. Examples of
81 * alternative styles you can specify via environment
82 * variables are:
84 * GIT_DIFF_OPTS="-c";
86 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
88 /* In case external diff fails... */
89 env_diff_opts = getenv("GIT_DIFF_OPTS");
90 if (env_diff_opts) diff_opts = env_diff_opts;
92 done_preparing = 1;
93 return external_diff_cmd;
96 #define TEMPFILE_PATH_LEN 50
98 static struct diff_tempfile {
99 const char *name; /* filename external diff should read from */
100 char hex[41];
101 char mode[10];
102 char tmp_path[TEMPFILE_PATH_LEN];
103 } diff_temp[2];
105 static int count_lines(const char *filename)
107 FILE *in;
108 int count, ch, completely_empty = 1, nl_just_seen = 0;
109 in = fopen(filename, "r");
110 count = 0;
111 while ((ch = fgetc(in)) != EOF)
112 if (ch == '\n') {
113 count++;
114 nl_just_seen = 1;
115 completely_empty = 0;
117 else {
118 nl_just_seen = 0;
119 completely_empty = 0;
121 fclose(in);
122 if (completely_empty)
123 return 0;
124 if (!nl_just_seen)
125 count++; /* no trailing newline */
126 return count;
129 static void print_line_count(int count)
131 switch (count) {
132 case 0:
133 printf("0,0");
134 break;
135 case 1:
136 printf("1");
137 break;
138 default:
139 printf("1,%d", count);
140 break;
144 static void copy_file(int prefix, const char *filename)
146 FILE *in;
147 int ch, nl_just_seen = 1;
148 in = fopen(filename, "r");
149 while ((ch = fgetc(in)) != EOF) {
150 if (nl_just_seen)
151 putchar(prefix);
152 putchar(ch);
153 if (ch == '\n')
154 nl_just_seen = 1;
155 else
156 nl_just_seen = 0;
158 fclose(in);
159 if (!nl_just_seen)
160 printf("\n\\ No newline at end of file\n");
163 static void emit_rewrite_diff(const char *name_a,
164 const char *name_b,
165 struct diff_tempfile *temp)
167 /* Use temp[i].name as input, name_a and name_b as labels */
168 int lc_a, lc_b;
169 lc_a = count_lines(temp[0].name);
170 lc_b = count_lines(temp[1].name);
171 printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
172 print_line_count(lc_a);
173 printf(" +");
174 print_line_count(lc_b);
175 printf(" @@\n");
176 if (lc_a)
177 copy_file('-', temp[0].name);
178 if (lc_b)
179 copy_file('+', temp[1].name);
182 static int fill_mmfile(mmfile_t *mf, const char *file)
184 int fd = open(file, O_RDONLY);
185 struct stat st;
186 char *buf;
187 unsigned long size;
189 mf->ptr = NULL;
190 mf->size = 0;
191 if (fd < 0)
192 return 0;
193 fstat(fd, &st);
194 size = st.st_size;
195 buf = xmalloc(size);
196 mf->ptr = buf;
197 mf->size = size;
198 while (size) {
199 int retval = read(fd, buf, size);
200 if (retval < 0) {
201 if (errno == EINTR || errno == EAGAIN)
202 continue;
203 break;
205 if (!retval)
206 break;
207 buf += retval;
208 size -= retval;
210 mf->size -= size;
211 close(fd);
212 return 0;
215 struct emit_callback {
216 const char **label_path;
219 static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
221 int i;
222 struct emit_callback *ecbdata = priv;
224 if (ecbdata->label_path[0]) {
225 printf("--- %s\n", ecbdata->label_path[0]);
226 printf("+++ %s\n", ecbdata->label_path[1]);
227 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
229 for (i = 0; i < nbuf; i++)
230 if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
231 return -1;
232 return 0;
235 #define FIRST_FEW_BYTES 8000
236 static int mmfile_is_binary(mmfile_t *mf)
238 long sz = mf->size;
239 if (FIRST_FEW_BYTES < sz)
240 sz = FIRST_FEW_BYTES;
241 if (memchr(mf->ptr, 0, sz))
242 return 1;
243 return 0;
246 static const char *builtin_diff(const char *name_a,
247 const char *name_b,
248 struct diff_tempfile *temp,
249 const char *xfrm_msg,
250 int complete_rewrite,
251 const char **args)
253 int i, next_at, cmd_size;
254 mmfile_t mf1, mf2;
255 const char *const diff_cmd = "diff -L%s -L%s";
256 const char *const diff_arg = "-- %s %s||:"; /* "||:" is to return 0 */
257 const char *input_name_sq[2];
258 const char *label_path[2];
259 char *cmd;
261 /* diff_cmd and diff_arg have 4 %s in total which makes
262 * the sum of these strings 8 bytes larger than required.
263 * we use 2 spaces around diff-opts, and we need to count
264 * terminating NUL; we used to subtract 5 here, but we do not
265 * care about small leaks in this subprocess that is about
266 * to exec "diff" anymore.
268 cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
269 + 128);
271 for (i = 0; i < 2; i++) {
272 input_name_sq[i] = sq_quote(temp[i].name);
273 if (!strcmp(temp[i].name, "/dev/null"))
274 label_path[i] = "/dev/null";
275 else if (!i)
276 label_path[i] = sq_quote(quote_two("a/", name_a));
277 else
278 label_path[i] = sq_quote(quote_two("b/", name_b));
279 cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
282 cmd = xmalloc(cmd_size);
284 next_at = 0;
285 next_at += snprintf(cmd+next_at, cmd_size-next_at,
286 diff_cmd, label_path[0], label_path[1]);
287 next_at += snprintf(cmd+next_at, cmd_size-next_at,
288 " %s ", diff_opts);
289 next_at += snprintf(cmd+next_at, cmd_size-next_at,
290 diff_arg, input_name_sq[0], input_name_sq[1]);
292 printf("diff --git %s %s\n",
293 quote_two("a/", name_a), quote_two("b/", name_b));
294 if (label_path[0][0] == '/') {
295 /* dev/null */
296 printf("new file mode %s\n", temp[1].mode);
297 if (xfrm_msg && xfrm_msg[0])
298 puts(xfrm_msg);
300 else if (label_path[1][0] == '/') {
301 printf("deleted file mode %s\n", temp[0].mode);
302 if (xfrm_msg && xfrm_msg[0])
303 puts(xfrm_msg);
305 else {
306 if (strcmp(temp[0].mode, temp[1].mode)) {
307 printf("old mode %s\n", temp[0].mode);
308 printf("new mode %s\n", temp[1].mode);
310 if (xfrm_msg && xfrm_msg[0])
311 puts(xfrm_msg);
313 * we do not run diff between different kind
314 * of objects.
316 if (strncmp(temp[0].mode, temp[1].mode, 3))
317 return NULL;
318 if (complete_rewrite) {
319 emit_rewrite_diff(name_a, name_b, temp);
320 return NULL;
324 /* Un-quote the paths */
325 if (label_path[0][0] != '/')
326 label_path[0] = quote_two("a/", name_a);
327 if (label_path[1][0] != '/')
328 label_path[1] = quote_two("b/", name_b);
330 if (fill_mmfile(&mf1, temp[0].name) < 0 ||
331 fill_mmfile(&mf2, temp[1].name) < 0)
332 die("unable to read files to diff");
334 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
335 printf("Binary files %s and %s differ\n",
336 label_path[0], label_path[1]);
337 else {
338 /* Crazy xdl interfaces.. */
339 const char *diffopts = getenv("GIT_DIFF_OPTS");
340 xpparam_t xpp;
341 xdemitconf_t xecfg;
342 xdemitcb_t ecb;
343 struct emit_callback ecbdata;
345 ecbdata.label_path = label_path;
346 xpp.flags = XDF_NEED_MINIMAL;
347 xecfg.ctxlen = 3;
348 if (!diffopts)
350 else if (!strncmp(diffopts, "--unified=", 10))
351 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
352 else if (!strncmp(diffopts, "-u", 2))
353 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
354 ecb.outf = fn_out;
355 ecb.priv = &ecbdata;
356 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
359 free(mf1.ptr);
360 free(mf2.ptr);
361 return NULL;
364 struct diff_filespec *alloc_filespec(const char *path)
366 int namelen = strlen(path);
367 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
369 memset(spec, 0, sizeof(*spec));
370 spec->path = (char *)(spec + 1);
371 memcpy(spec->path, path, namelen+1);
372 return spec;
375 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
376 unsigned short mode)
378 if (mode) {
379 spec->mode = DIFF_FILE_CANON_MODE(mode);
380 memcpy(spec->sha1, sha1, 20);
381 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
386 * Given a name and sha1 pair, if the dircache tells us the file in
387 * the work tree has that object contents, return true, so that
388 * prepare_temp_file() does not have to inflate and extract.
390 static int work_tree_matches(const char *name, const unsigned char *sha1)
392 struct cache_entry *ce;
393 struct stat st;
394 int pos, len;
396 /* We do not read the cache ourselves here, because the
397 * benchmark with my previous version that always reads cache
398 * shows that it makes things worse for diff-tree comparing
399 * two linux-2.6 kernel trees in an already checked out work
400 * tree. This is because most diff-tree comparisons deal with
401 * only a small number of files, while reading the cache is
402 * expensive for a large project, and its cost outweighs the
403 * savings we get by not inflating the object to a temporary
404 * file. Practically, this code only helps when we are used
405 * by diff-cache --cached, which does read the cache before
406 * calling us.
408 if (!active_cache)
409 return 0;
411 len = strlen(name);
412 pos = cache_name_pos(name, len);
413 if (pos < 0)
414 return 0;
415 ce = active_cache[pos];
416 if ((lstat(name, &st) < 0) ||
417 !S_ISREG(st.st_mode) || /* careful! */
418 ce_match_stat(ce, &st, 0) ||
419 memcmp(sha1, ce->sha1, 20))
420 return 0;
421 /* we return 1 only when we can stat, it is a regular file,
422 * stat information matches, and sha1 recorded in the cache
423 * matches. I.e. we know the file in the work tree really is
424 * the same as the <name, sha1> pair.
426 return 1;
429 static struct sha1_size_cache {
430 unsigned char sha1[20];
431 unsigned long size;
432 } **sha1_size_cache;
433 static int sha1_size_cache_nr, sha1_size_cache_alloc;
435 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
436 int find_only,
437 unsigned long size)
439 int first, last;
440 struct sha1_size_cache *e;
442 first = 0;
443 last = sha1_size_cache_nr;
444 while (last > first) {
445 int cmp, next = (last + first) >> 1;
446 e = sha1_size_cache[next];
447 cmp = memcmp(e->sha1, sha1, 20);
448 if (!cmp)
449 return e;
450 if (cmp < 0) {
451 last = next;
452 continue;
454 first = next+1;
456 /* not found */
457 if (find_only)
458 return NULL;
459 /* insert to make it at "first" */
460 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
461 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
462 sha1_size_cache = xrealloc(sha1_size_cache,
463 sha1_size_cache_alloc *
464 sizeof(*sha1_size_cache));
466 sha1_size_cache_nr++;
467 if (first < sha1_size_cache_nr)
468 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
469 (sha1_size_cache_nr - first - 1) *
470 sizeof(*sha1_size_cache));
471 e = xmalloc(sizeof(struct sha1_size_cache));
472 sha1_size_cache[first] = e;
473 memcpy(e->sha1, sha1, 20);
474 e->size = size;
475 return e;
479 * While doing rename detection and pickaxe operation, we may need to
480 * grab the data for the blob (or file) for our own in-core comparison.
481 * diff_filespec has data and size fields for this purpose.
483 int diff_populate_filespec(struct diff_filespec *s, int size_only)
485 int err = 0;
486 if (!DIFF_FILE_VALID(s))
487 die("internal error: asking to populate invalid file.");
488 if (S_ISDIR(s->mode))
489 return -1;
491 if (!use_size_cache)
492 size_only = 0;
494 if (s->data)
495 return err;
496 if (!s->sha1_valid ||
497 work_tree_matches(s->path, s->sha1)) {
498 struct stat st;
499 int fd;
500 if (lstat(s->path, &st) < 0) {
501 if (errno == ENOENT) {
502 err_empty:
503 err = -1;
504 empty:
505 s->data = "";
506 s->size = 0;
507 return err;
510 s->size = st.st_size;
511 if (!s->size)
512 goto empty;
513 if (size_only)
514 return 0;
515 if (S_ISLNK(st.st_mode)) {
516 int ret;
517 s->data = xmalloc(s->size);
518 s->should_free = 1;
519 ret = readlink(s->path, s->data, s->size);
520 if (ret < 0) {
521 free(s->data);
522 goto err_empty;
524 return 0;
526 fd = open(s->path, O_RDONLY);
527 if (fd < 0)
528 goto err_empty;
529 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
530 close(fd);
531 if (s->data == MAP_FAILED)
532 goto err_empty;
533 s->should_munmap = 1;
535 else {
536 char type[20];
537 struct sha1_size_cache *e;
539 if (size_only) {
540 e = locate_size_cache(s->sha1, 1, 0);
541 if (e) {
542 s->size = e->size;
543 return 0;
545 if (!sha1_object_info(s->sha1, type, &s->size))
546 locate_size_cache(s->sha1, 0, s->size);
548 else {
549 s->data = read_sha1_file(s->sha1, type, &s->size);
550 s->should_free = 1;
553 return 0;
556 void diff_free_filespec_data(struct diff_filespec *s)
558 if (s->should_free)
559 free(s->data);
560 else if (s->should_munmap)
561 munmap(s->data, s->size);
562 s->should_free = s->should_munmap = 0;
563 s->data = NULL;
564 free(s->cnt_data);
565 s->cnt_data = NULL;
568 static void prep_temp_blob(struct diff_tempfile *temp,
569 void *blob,
570 unsigned long size,
571 const unsigned char *sha1,
572 int mode)
574 int fd;
576 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
577 if (fd < 0)
578 die("unable to create temp-file");
579 if (write(fd, blob, size) != size)
580 die("unable to write temp-file");
581 close(fd);
582 temp->name = temp->tmp_path;
583 strcpy(temp->hex, sha1_to_hex(sha1));
584 temp->hex[40] = 0;
585 sprintf(temp->mode, "%06o", mode);
588 static void prepare_temp_file(const char *name,
589 struct diff_tempfile *temp,
590 struct diff_filespec *one)
592 if (!DIFF_FILE_VALID(one)) {
593 not_a_valid_file:
594 /* A '-' entry produces this for file-2, and
595 * a '+' entry produces this for file-1.
597 temp->name = "/dev/null";
598 strcpy(temp->hex, ".");
599 strcpy(temp->mode, ".");
600 return;
603 if (!one->sha1_valid ||
604 work_tree_matches(name, one->sha1)) {
605 struct stat st;
606 if (lstat(name, &st) < 0) {
607 if (errno == ENOENT)
608 goto not_a_valid_file;
609 die("stat(%s): %s", name, strerror(errno));
611 if (S_ISLNK(st.st_mode)) {
612 int ret;
613 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
614 if (sizeof(buf) <= st.st_size)
615 die("symlink too long: %s", name);
616 ret = readlink(name, buf, st.st_size);
617 if (ret < 0)
618 die("readlink(%s)", name);
619 prep_temp_blob(temp, buf, st.st_size,
620 (one->sha1_valid ?
621 one->sha1 : null_sha1),
622 (one->sha1_valid ?
623 one->mode : S_IFLNK));
625 else {
626 /* we can borrow from the file in the work tree */
627 temp->name = name;
628 if (!one->sha1_valid)
629 strcpy(temp->hex, sha1_to_hex(null_sha1));
630 else
631 strcpy(temp->hex, sha1_to_hex(one->sha1));
632 /* Even though we may sometimes borrow the
633 * contents from the work tree, we always want
634 * one->mode. mode is trustworthy even when
635 * !(one->sha1_valid), as long as
636 * DIFF_FILE_VALID(one).
638 sprintf(temp->mode, "%06o", one->mode);
640 return;
642 else {
643 if (diff_populate_filespec(one, 0))
644 die("cannot read data blob for %s", one->path);
645 prep_temp_blob(temp, one->data, one->size,
646 one->sha1, one->mode);
650 static void remove_tempfile(void)
652 int i;
654 for (i = 0; i < 2; i++)
655 if (diff_temp[i].name == diff_temp[i].tmp_path) {
656 unlink(diff_temp[i].name);
657 diff_temp[i].name = NULL;
661 static void remove_tempfile_on_signal(int signo)
663 remove_tempfile();
664 signal(SIGINT, SIG_DFL);
665 raise(signo);
668 static int spawn_prog(const char *pgm, const char **arg)
670 pid_t pid;
671 int status;
673 fflush(NULL);
674 pid = fork();
675 if (pid < 0)
676 die("unable to fork");
677 if (!pid) {
678 execvp(pgm, (char *const*) arg);
679 exit(255);
682 while (waitpid(pid, &status, 0) < 0) {
683 if (errno == EINTR)
684 continue;
685 return -1;
688 /* Earlier we did not check the exit status because
689 * diff exits non-zero if files are different, and
690 * we are not interested in knowing that. It was a
691 * mistake which made it harder to quit a diff-*
692 * session that uses the git-apply-patch-script as
693 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
694 * should also exit non-zero only when it wants to
695 * abort the entire diff-* session.
697 if (WIFEXITED(status) && !WEXITSTATUS(status))
698 return 0;
699 return -1;
702 /* An external diff command takes:
704 * diff-cmd name infile1 infile1-sha1 infile1-mode \
705 * infile2 infile2-sha1 infile2-mode [ rename-to ]
708 static void run_external_diff(const char *pgm,
709 const char *name,
710 const char *other,
711 struct diff_filespec *one,
712 struct diff_filespec *two,
713 const char *xfrm_msg,
714 int complete_rewrite)
716 const char *spawn_arg[10];
717 struct diff_tempfile *temp = diff_temp;
718 int retval;
719 static int atexit_asked = 0;
720 const char *othername;
722 othername = (other? other : name);
723 if (one && two) {
724 prepare_temp_file(name, &temp[0], one);
725 prepare_temp_file(othername, &temp[1], two);
726 if (! atexit_asked &&
727 (temp[0].name == temp[0].tmp_path ||
728 temp[1].name == temp[1].tmp_path)) {
729 atexit_asked = 1;
730 atexit(remove_tempfile);
732 signal(SIGINT, remove_tempfile_on_signal);
735 if (pgm) {
736 const char **arg = &spawn_arg[0];
737 if (one && two) {
738 *arg++ = pgm;
739 *arg++ = name;
740 *arg++ = temp[0].name;
741 *arg++ = temp[0].hex;
742 *arg++ = temp[0].mode;
743 *arg++ = temp[1].name;
744 *arg++ = temp[1].hex;
745 *arg++ = temp[1].mode;
746 if (other) {
747 *arg++ = other;
748 *arg++ = xfrm_msg;
750 } else {
751 *arg++ = pgm;
752 *arg++ = name;
754 *arg = NULL;
755 } else {
756 if (one && two) {
757 pgm = builtin_diff(name, othername, temp, xfrm_msg, complete_rewrite, spawn_arg);
758 } else
759 printf("* Unmerged path %s\n", name);
762 retval = 0;
763 if (pgm)
764 retval = spawn_prog(pgm, spawn_arg);
765 remove_tempfile();
766 if (retval) {
767 fprintf(stderr, "external diff died, stopping at %s.\n", name);
768 exit(1);
772 static void diff_fill_sha1_info(struct diff_filespec *one)
774 if (DIFF_FILE_VALID(one)) {
775 if (!one->sha1_valid) {
776 struct stat st;
777 if (lstat(one->path, &st) < 0)
778 die("stat %s", one->path);
779 if (index_path(one->sha1, one->path, &st, 0))
780 die("cannot hash %s\n", one->path);
783 else
784 memset(one->sha1, 0, 20);
787 static void run_diff(struct diff_filepair *p, struct diff_options *o)
789 const char *pgm = external_diff();
790 char msg[PATH_MAX*2+300], *xfrm_msg;
791 struct diff_filespec *one;
792 struct diff_filespec *two;
793 const char *name;
794 const char *other;
795 char *name_munged, *other_munged;
796 int complete_rewrite = 0;
797 int len;
799 if (DIFF_PAIR_UNMERGED(p)) {
800 /* unmerged */
801 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
803 return;
806 name = p->one->path;
807 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
808 name_munged = quote_one(name);
809 other_munged = quote_one(other);
810 one = p->one; two = p->two;
812 diff_fill_sha1_info(one);
813 diff_fill_sha1_info(two);
815 len = 0;
816 switch (p->status) {
817 case DIFF_STATUS_COPIED:
818 len += snprintf(msg + len, sizeof(msg) - len,
819 "similarity index %d%%\n"
820 "copy from %s\n"
821 "copy to %s\n",
822 (int)(0.5 + p->score * 100.0/MAX_SCORE),
823 name_munged, other_munged);
824 break;
825 case DIFF_STATUS_RENAMED:
826 len += snprintf(msg + len, sizeof(msg) - len,
827 "similarity index %d%%\n"
828 "rename from %s\n"
829 "rename to %s\n",
830 (int)(0.5 + p->score * 100.0/MAX_SCORE),
831 name_munged, other_munged);
832 break;
833 case DIFF_STATUS_MODIFIED:
834 if (p->score) {
835 len += snprintf(msg + len, sizeof(msg) - len,
836 "dissimilarity index %d%%\n",
837 (int)(0.5 + p->score *
838 100.0/MAX_SCORE));
839 complete_rewrite = 1;
840 break;
842 /* fallthru */
843 default:
844 /* nothing */
848 if (memcmp(one->sha1, two->sha1, 20)) {
849 char one_sha1[41];
850 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
851 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
853 len += snprintf(msg + len, sizeof(msg) - len,
854 "index %.*s..%.*s",
855 abbrev, one_sha1, abbrev,
856 sha1_to_hex(two->sha1));
857 if (one->mode == two->mode)
858 len += snprintf(msg + len, sizeof(msg) - len,
859 " %06o", one->mode);
860 len += snprintf(msg + len, sizeof(msg) - len, "\n");
863 if (len)
864 msg[--len] = 0;
865 xfrm_msg = len ? msg : NULL;
867 if (!pgm &&
868 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
869 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
870 /* a filepair that changes between file and symlink
871 * needs to be split into deletion and creation.
873 struct diff_filespec *null = alloc_filespec(two->path);
874 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
875 free(null);
876 null = alloc_filespec(one->path);
877 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
878 free(null);
880 else
881 run_external_diff(pgm, name, other, one, two, xfrm_msg,
882 complete_rewrite);
884 free(name_munged);
885 free(other_munged);
888 void diff_setup(struct diff_options *options)
890 memset(options, 0, sizeof(*options));
891 options->output_format = DIFF_FORMAT_RAW;
892 options->line_termination = '\n';
893 options->break_opt = -1;
894 options->rename_limit = -1;
896 options->change = diff_change;
897 options->add_remove = diff_addremove;
900 int diff_setup_done(struct diff_options *options)
902 if ((options->find_copies_harder &&
903 options->detect_rename != DIFF_DETECT_COPY) ||
904 (0 <= options->rename_limit && !options->detect_rename))
905 return -1;
906 if (options->detect_rename && options->rename_limit < 0)
907 options->rename_limit = diff_rename_limit_default;
908 if (options->setup & DIFF_SETUP_USE_CACHE) {
909 if (!active_cache)
910 /* read-cache does not die even when it fails
911 * so it is safe for us to do this here. Also
912 * it does not smudge active_cache or active_nr
913 * when it fails, so we do not have to worry about
914 * cleaning it up ourselves either.
916 read_cache();
918 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
919 use_size_cache = 1;
920 if (options->abbrev <= 0 || 40 < options->abbrev)
921 options->abbrev = 40; /* full */
923 return 0;
926 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
928 const char *arg = av[0];
929 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
930 options->output_format = DIFF_FORMAT_PATCH;
931 else if (!strcmp(arg, "-z"))
932 options->line_termination = 0;
933 else if (!strncmp(arg, "-l", 2))
934 options->rename_limit = strtoul(arg+2, NULL, 10);
935 else if (!strcmp(arg, "--full-index"))
936 options->full_index = 1;
937 else if (!strcmp(arg, "--name-only"))
938 options->output_format = DIFF_FORMAT_NAME;
939 else if (!strcmp(arg, "--name-status"))
940 options->output_format = DIFF_FORMAT_NAME_STATUS;
941 else if (!strcmp(arg, "-R"))
942 options->reverse_diff = 1;
943 else if (!strncmp(arg, "-S", 2))
944 options->pickaxe = arg + 2;
945 else if (!strcmp(arg, "-s"))
946 options->output_format = DIFF_FORMAT_NO_OUTPUT;
947 else if (!strncmp(arg, "-O", 2))
948 options->orderfile = arg + 2;
949 else if (!strncmp(arg, "--diff-filter=", 14))
950 options->filter = arg + 14;
951 else if (!strcmp(arg, "--pickaxe-all"))
952 options->pickaxe_opts = DIFF_PICKAXE_ALL;
953 else if (!strncmp(arg, "-B", 2)) {
954 if ((options->break_opt =
955 diff_scoreopt_parse(arg)) == -1)
956 return -1;
958 else if (!strncmp(arg, "-M", 2)) {
959 if ((options->rename_score =
960 diff_scoreopt_parse(arg)) == -1)
961 return -1;
962 options->detect_rename = DIFF_DETECT_RENAME;
964 else if (!strncmp(arg, "-C", 2)) {
965 if ((options->rename_score =
966 diff_scoreopt_parse(arg)) == -1)
967 return -1;
968 options->detect_rename = DIFF_DETECT_COPY;
970 else if (!strcmp(arg, "--find-copies-harder"))
971 options->find_copies_harder = 1;
972 else if (!strcmp(arg, "--abbrev"))
973 options->abbrev = DEFAULT_ABBREV;
974 else if (!strncmp(arg, "--abbrev=", 9)) {
975 options->abbrev = strtoul(arg + 9, NULL, 10);
976 if (options->abbrev < MINIMUM_ABBREV)
977 options->abbrev = MINIMUM_ABBREV;
978 else if (40 < options->abbrev)
979 options->abbrev = 40;
981 else
982 return 0;
983 return 1;
986 static int parse_num(const char **cp_p)
988 unsigned long num, scale;
989 int ch, dot;
990 const char *cp = *cp_p;
992 num = 0;
993 scale = 1;
994 dot = 0;
995 for(;;) {
996 ch = *cp;
997 if ( !dot && ch == '.' ) {
998 scale = 1;
999 dot = 1;
1000 } else if ( ch == '%' ) {
1001 scale = dot ? scale*100 : 100;
1002 cp++; /* % is always at the end */
1003 break;
1004 } else if ( ch >= '0' && ch <= '9' ) {
1005 if ( scale < 100000 ) {
1006 scale *= 10;
1007 num = (num*10) + (ch-'0');
1009 } else {
1010 break;
1012 cp++;
1014 *cp_p = cp;
1016 /* user says num divided by scale and we say internally that
1017 * is MAX_SCORE * num / scale.
1019 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1022 int diff_scoreopt_parse(const char *opt)
1024 int opt1, opt2, cmd;
1026 if (*opt++ != '-')
1027 return -1;
1028 cmd = *opt++;
1029 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1030 return -1; /* that is not a -M, -C nor -B option */
1032 opt1 = parse_num(&opt);
1033 if (cmd != 'B')
1034 opt2 = 0;
1035 else {
1036 if (*opt == 0)
1037 opt2 = 0;
1038 else if (*opt != '/')
1039 return -1; /* we expect -B80/99 or -B80 */
1040 else {
1041 opt++;
1042 opt2 = parse_num(&opt);
1045 if (*opt != 0)
1046 return -1;
1047 return opt1 | (opt2 << 16);
1050 struct diff_queue_struct diff_queued_diff;
1052 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1054 if (queue->alloc <= queue->nr) {
1055 queue->alloc = alloc_nr(queue->alloc);
1056 queue->queue = xrealloc(queue->queue,
1057 sizeof(dp) * queue->alloc);
1059 queue->queue[queue->nr++] = dp;
1062 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1063 struct diff_filespec *one,
1064 struct diff_filespec *two)
1066 struct diff_filepair *dp = xmalloc(sizeof(*dp));
1067 dp->one = one;
1068 dp->two = two;
1069 dp->score = 0;
1070 dp->status = 0;
1071 dp->source_stays = 0;
1072 dp->broken_pair = 0;
1073 if (queue)
1074 diff_q(queue, dp);
1075 return dp;
1078 void diff_free_filepair(struct diff_filepair *p)
1080 diff_free_filespec_data(p->one);
1081 diff_free_filespec_data(p->two);
1082 free(p->one);
1083 free(p->two);
1084 free(p);
1087 /* This is different from find_unique_abbrev() in that
1088 * it stuffs the result with dots for alignment.
1090 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1092 int abblen;
1093 const char *abbrev;
1094 if (len == 40)
1095 return sha1_to_hex(sha1);
1097 abbrev = find_unique_abbrev(sha1, len);
1098 if (!abbrev)
1099 return sha1_to_hex(sha1);
1100 abblen = strlen(abbrev);
1101 if (abblen < 37) {
1102 static char hex[41];
1103 if (len < abblen && abblen <= len + 2)
1104 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1105 else
1106 sprintf(hex, "%s...", abbrev);
1107 return hex;
1109 return sha1_to_hex(sha1);
1112 static void diff_flush_raw(struct diff_filepair *p,
1113 int line_termination,
1114 int inter_name_termination,
1115 struct diff_options *options)
1117 int two_paths;
1118 char status[10];
1119 int abbrev = options->abbrev;
1120 const char *path_one, *path_two;
1121 int output_format = options->output_format;
1123 path_one = p->one->path;
1124 path_two = p->two->path;
1125 if (line_termination) {
1126 path_one = quote_one(path_one);
1127 path_two = quote_one(path_two);
1130 if (p->score)
1131 sprintf(status, "%c%03d", p->status,
1132 (int)(0.5 + p->score * 100.0/MAX_SCORE));
1133 else {
1134 status[0] = p->status;
1135 status[1] = 0;
1137 switch (p->status) {
1138 case DIFF_STATUS_COPIED:
1139 case DIFF_STATUS_RENAMED:
1140 two_paths = 1;
1141 break;
1142 case DIFF_STATUS_ADDED:
1143 case DIFF_STATUS_DELETED:
1144 two_paths = 0;
1145 break;
1146 default:
1147 two_paths = 0;
1148 break;
1150 if (output_format != DIFF_FORMAT_NAME_STATUS) {
1151 printf(":%06o %06o %s ",
1152 p->one->mode, p->two->mode,
1153 diff_unique_abbrev(p->one->sha1, abbrev));
1154 printf("%s ",
1155 diff_unique_abbrev(p->two->sha1, abbrev));
1157 printf("%s%c%s", status, inter_name_termination, path_one);
1158 if (two_paths)
1159 printf("%c%s", inter_name_termination, path_two);
1160 putchar(line_termination);
1161 if (path_one != p->one->path)
1162 free((void*)path_one);
1163 if (path_two != p->two->path)
1164 free((void*)path_two);
1167 static void diff_flush_name(struct diff_filepair *p,
1168 int inter_name_termination,
1169 int line_termination)
1171 char *path = p->two->path;
1173 if (line_termination)
1174 path = quote_one(p->two->path);
1175 else
1176 path = p->two->path;
1177 printf("%s%c", path, line_termination);
1178 if (p->two->path != path)
1179 free(path);
1182 int diff_unmodified_pair(struct diff_filepair *p)
1184 /* This function is written stricter than necessary to support
1185 * the currently implemented transformers, but the idea is to
1186 * let transformers to produce diff_filepairs any way they want,
1187 * and filter and clean them up here before producing the output.
1189 struct diff_filespec *one, *two;
1191 if (DIFF_PAIR_UNMERGED(p))
1192 return 0; /* unmerged is interesting */
1194 one = p->one;
1195 two = p->two;
1197 /* deletion, addition, mode or type change
1198 * and rename are all interesting.
1200 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1201 DIFF_PAIR_MODE_CHANGED(p) ||
1202 strcmp(one->path, two->path))
1203 return 0;
1205 /* both are valid and point at the same path. that is, we are
1206 * dealing with a change.
1208 if (one->sha1_valid && two->sha1_valid &&
1209 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1210 return 1; /* no change */
1211 if (!one->sha1_valid && !two->sha1_valid)
1212 return 1; /* both look at the same file on the filesystem. */
1213 return 0;
1216 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1218 if (diff_unmodified_pair(p))
1219 return;
1221 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1222 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1223 return; /* no tree diffs in patch format */
1225 run_diff(p, o);
1228 int diff_queue_is_empty(void)
1230 struct diff_queue_struct *q = &diff_queued_diff;
1231 int i;
1232 for (i = 0; i < q->nr; i++)
1233 if (!diff_unmodified_pair(q->queue[i]))
1234 return 0;
1235 return 1;
1238 #if DIFF_DEBUG
1239 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1241 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1242 x, one ? one : "",
1243 s->path,
1244 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1245 s->mode,
1246 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1247 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1248 x, one ? one : "",
1249 s->size, s->xfrm_flags);
1252 void diff_debug_filepair(const struct diff_filepair *p, int i)
1254 diff_debug_filespec(p->one, i, "one");
1255 diff_debug_filespec(p->two, i, "two");
1256 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1257 p->score, p->status ? p->status : '?',
1258 p->source_stays, p->broken_pair);
1261 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1263 int i;
1264 if (msg)
1265 fprintf(stderr, "%s\n", msg);
1266 fprintf(stderr, "q->nr = %d\n", q->nr);
1267 for (i = 0; i < q->nr; i++) {
1268 struct diff_filepair *p = q->queue[i];
1269 diff_debug_filepair(p, i);
1272 #endif
1274 static void diff_resolve_rename_copy(void)
1276 int i, j;
1277 struct diff_filepair *p, *pp;
1278 struct diff_queue_struct *q = &diff_queued_diff;
1280 diff_debug_queue("resolve-rename-copy", q);
1282 for (i = 0; i < q->nr; i++) {
1283 p = q->queue[i];
1284 p->status = 0; /* undecided */
1285 if (DIFF_PAIR_UNMERGED(p))
1286 p->status = DIFF_STATUS_UNMERGED;
1287 else if (!DIFF_FILE_VALID(p->one))
1288 p->status = DIFF_STATUS_ADDED;
1289 else if (!DIFF_FILE_VALID(p->two))
1290 p->status = DIFF_STATUS_DELETED;
1291 else if (DIFF_PAIR_TYPE_CHANGED(p))
1292 p->status = DIFF_STATUS_TYPE_CHANGED;
1294 /* from this point on, we are dealing with a pair
1295 * whose both sides are valid and of the same type, i.e.
1296 * either in-place edit or rename/copy edit.
1298 else if (DIFF_PAIR_RENAME(p)) {
1299 if (p->source_stays) {
1300 p->status = DIFF_STATUS_COPIED;
1301 continue;
1303 /* See if there is some other filepair that
1304 * copies from the same source as us. If so
1305 * we are a copy. Otherwise we are either a
1306 * copy if the path stays, or a rename if it
1307 * does not, but we already handled "stays" case.
1309 for (j = i + 1; j < q->nr; j++) {
1310 pp = q->queue[j];
1311 if (strcmp(pp->one->path, p->one->path))
1312 continue; /* not us */
1313 if (!DIFF_PAIR_RENAME(pp))
1314 continue; /* not a rename/copy */
1315 /* pp is a rename/copy from the same source */
1316 p->status = DIFF_STATUS_COPIED;
1317 break;
1319 if (!p->status)
1320 p->status = DIFF_STATUS_RENAMED;
1322 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1323 p->one->mode != p->two->mode)
1324 p->status = DIFF_STATUS_MODIFIED;
1325 else {
1326 /* This is a "no-change" entry and should not
1327 * happen anymore, but prepare for broken callers.
1329 error("feeding unmodified %s to diffcore",
1330 p->one->path);
1331 p->status = DIFF_STATUS_UNKNOWN;
1334 diff_debug_queue("resolve-rename-copy done", q);
1337 void diff_flush(struct diff_options *options)
1339 struct diff_queue_struct *q = &diff_queued_diff;
1340 int i;
1341 int inter_name_termination = '\t';
1342 int diff_output_format = options->output_format;
1343 int line_termination = options->line_termination;
1345 if (!line_termination)
1346 inter_name_termination = 0;
1348 for (i = 0; i < q->nr; i++) {
1349 struct diff_filepair *p = q->queue[i];
1350 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1351 (p->status == DIFF_STATUS_UNKNOWN))
1352 continue;
1353 if (p->status == 0)
1354 die("internal error in diff-resolve-rename-copy");
1355 switch (diff_output_format) {
1356 case DIFF_FORMAT_PATCH:
1357 diff_flush_patch(p, options);
1358 break;
1359 case DIFF_FORMAT_RAW:
1360 case DIFF_FORMAT_NAME_STATUS:
1361 diff_flush_raw(p, line_termination,
1362 inter_name_termination,
1363 options);
1364 break;
1365 case DIFF_FORMAT_NAME:
1366 diff_flush_name(p,
1367 inter_name_termination,
1368 line_termination);
1369 break;
1371 diff_free_filepair(q->queue[i]);
1373 free(q->queue);
1374 q->queue = NULL;
1375 q->nr = q->alloc = 0;
1378 static void diffcore_apply_filter(const char *filter)
1380 int i;
1381 struct diff_queue_struct *q = &diff_queued_diff;
1382 struct diff_queue_struct outq;
1383 outq.queue = NULL;
1384 outq.nr = outq.alloc = 0;
1386 if (!filter)
1387 return;
1389 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1390 int found;
1391 for (i = found = 0; !found && i < q->nr; i++) {
1392 struct diff_filepair *p = q->queue[i];
1393 if (((p->status == DIFF_STATUS_MODIFIED) &&
1394 ((p->score &&
1395 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1396 (!p->score &&
1397 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1398 ((p->status != DIFF_STATUS_MODIFIED) &&
1399 strchr(filter, p->status)))
1400 found++;
1402 if (found)
1403 return;
1405 /* otherwise we will clear the whole queue
1406 * by copying the empty outq at the end of this
1407 * function, but first clear the current entries
1408 * in the queue.
1410 for (i = 0; i < q->nr; i++)
1411 diff_free_filepair(q->queue[i]);
1413 else {
1414 /* Only the matching ones */
1415 for (i = 0; i < q->nr; i++) {
1416 struct diff_filepair *p = q->queue[i];
1418 if (((p->status == DIFF_STATUS_MODIFIED) &&
1419 ((p->score &&
1420 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1421 (!p->score &&
1422 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1423 ((p->status != DIFF_STATUS_MODIFIED) &&
1424 strchr(filter, p->status)))
1425 diff_q(&outq, p);
1426 else
1427 diff_free_filepair(p);
1430 free(q->queue);
1431 *q = outq;
1434 void diffcore_std(struct diff_options *options)
1436 if (options->paths && options->paths[0])
1437 diffcore_pathspec(options->paths);
1438 if (options->break_opt != -1)
1439 diffcore_break(options->break_opt);
1440 if (options->detect_rename)
1441 diffcore_rename(options);
1442 if (options->break_opt != -1)
1443 diffcore_merge_broken();
1444 if (options->pickaxe)
1445 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1446 if (options->orderfile)
1447 diffcore_order(options->orderfile);
1448 diff_resolve_rename_copy();
1449 diffcore_apply_filter(options->filter);
1453 void diffcore_std_no_resolve(struct diff_options *options)
1455 if (options->pickaxe)
1456 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1457 if (options->orderfile)
1458 diffcore_order(options->orderfile);
1459 diffcore_apply_filter(options->filter);
1462 void diff_addremove(struct diff_options *options,
1463 int addremove, unsigned mode,
1464 const unsigned char *sha1,
1465 const char *base, const char *path)
1467 char concatpath[PATH_MAX];
1468 struct diff_filespec *one, *two;
1470 /* This may look odd, but it is a preparation for
1471 * feeding "there are unchanged files which should
1472 * not produce diffs, but when you are doing copy
1473 * detection you would need them, so here they are"
1474 * entries to the diff-core. They will be prefixed
1475 * with something like '=' or '*' (I haven't decided
1476 * which but should not make any difference).
1477 * Feeding the same new and old to diff_change()
1478 * also has the same effect.
1479 * Before the final output happens, they are pruned after
1480 * merged into rename/copy pairs as appropriate.
1482 if (options->reverse_diff)
1483 addremove = (addremove == '+' ? '-' :
1484 addremove == '-' ? '+' : addremove);
1486 if (!path) path = "";
1487 sprintf(concatpath, "%s%s", base, path);
1488 one = alloc_filespec(concatpath);
1489 two = alloc_filespec(concatpath);
1491 if (addremove != '+')
1492 fill_filespec(one, sha1, mode);
1493 if (addremove != '-')
1494 fill_filespec(two, sha1, mode);
1496 diff_queue(&diff_queued_diff, one, two);
1499 void diff_change(struct diff_options *options,
1500 unsigned old_mode, unsigned new_mode,
1501 const unsigned char *old_sha1,
1502 const unsigned char *new_sha1,
1503 const char *base, const char *path)
1505 char concatpath[PATH_MAX];
1506 struct diff_filespec *one, *two;
1508 if (options->reverse_diff) {
1509 unsigned tmp;
1510 const unsigned char *tmp_c;
1511 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1512 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1514 if (!path) path = "";
1515 sprintf(concatpath, "%s%s", base, path);
1516 one = alloc_filespec(concatpath);
1517 two = alloc_filespec(concatpath);
1518 fill_filespec(one, old_sha1, old_mode);
1519 fill_filespec(two, new_sha1, new_mode);
1521 diff_queue(&diff_queued_diff, one, two);
1524 void diff_unmerge(struct diff_options *options,
1525 const char *path)
1527 struct diff_filespec *one, *two;
1528 one = alloc_filespec(path);
1529 two = alloc_filespec(path);
1530 diff_queue(&diff_queued_diff, one, two);