Set the parallel HTTP request limit via an environment variable
[git/jrn.git] / diff.c
blobcbb86320a6b98ea57712a78c558c7dc84613087b
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"
12 static const char *diff_opts = "-pu";
14 static int use_size_cache;
16 static const char *external_diff(void)
18 static const char *external_diff_cmd = NULL;
19 static int done_preparing = 0;
20 const char *env_diff_opts;
22 if (done_preparing)
23 return external_diff_cmd;
26 * Default values above are meant to match the
27 * Linux kernel development style. Examples of
28 * alternative styles you can specify via environment
29 * variables are:
31 * GIT_DIFF_OPTS="-c";
33 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
35 /* In case external diff fails... */
36 env_diff_opts = getenv("GIT_DIFF_OPTS");
37 if (env_diff_opts) diff_opts = env_diff_opts;
39 done_preparing = 1;
40 return external_diff_cmd;
43 #define TEMPFILE_PATH_LEN 50
45 static struct diff_tempfile {
46 const char *name; /* filename external diff should read from */
47 char hex[41];
48 char mode[10];
49 char tmp_path[TEMPFILE_PATH_LEN];
50 } diff_temp[2];
52 static int count_lines(const char *filename)
54 FILE *in;
55 int count, ch, completely_empty = 1, nl_just_seen = 0;
56 in = fopen(filename, "r");
57 count = 0;
58 while ((ch = fgetc(in)) != EOF)
59 if (ch == '\n') {
60 count++;
61 nl_just_seen = 1;
62 completely_empty = 0;
64 else {
65 nl_just_seen = 0;
66 completely_empty = 0;
68 fclose(in);
69 if (completely_empty)
70 return 0;
71 if (!nl_just_seen)
72 count++; /* no trailing newline */
73 return count;
76 static void print_line_count(int count)
78 switch (count) {
79 case 0:
80 printf("0,0");
81 break;
82 case 1:
83 printf("1");
84 break;
85 default:
86 printf("1,%d", count);
87 break;
91 static void copy_file(int prefix, const char *filename)
93 FILE *in;
94 int ch, nl_just_seen = 1;
95 in = fopen(filename, "r");
96 while ((ch = fgetc(in)) != EOF) {
97 if (nl_just_seen)
98 putchar(prefix);
99 putchar(ch);
100 if (ch == '\n')
101 nl_just_seen = 1;
102 else
103 nl_just_seen = 0;
105 fclose(in);
106 if (!nl_just_seen)
107 printf("\n\\ No newline at end of file\n");
110 static void emit_rewrite_diff(const char *name_a,
111 const char *name_b,
112 struct diff_tempfile *temp)
114 /* Use temp[i].name as input, name_a and name_b as labels */
115 int lc_a, lc_b;
116 lc_a = count_lines(temp[0].name);
117 lc_b = count_lines(temp[1].name);
118 printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
119 print_line_count(lc_a);
120 printf(" +");
121 print_line_count(lc_b);
122 printf(" @@\n");
123 if (lc_a)
124 copy_file('-', temp[0].name);
125 if (lc_b)
126 copy_file('+', temp[1].name);
129 static void builtin_diff(const char *name_a,
130 const char *name_b,
131 struct diff_tempfile *temp,
132 const char *xfrm_msg,
133 int complete_rewrite)
135 int i, next_at, cmd_size;
136 const char *const diff_cmd = "diff -L%s%s -L%s%s";
137 const char *const diff_arg = "%s %s||:"; /* "||:" is to return 0 */
138 const char *input_name_sq[2];
139 const char *path0[2];
140 const char *path1[2];
141 const char *name_sq[2];
142 char *cmd;
144 name_sq[0] = sq_quote(name_a);
145 name_sq[1] = sq_quote(name_b);
147 /* diff_cmd and diff_arg have 6 %s in total which makes
148 * the sum of these strings 12 bytes larger than required.
149 * we use 2 spaces around diff-opts, and we need to count
150 * terminating NUL, so we subtract 9 here.
152 cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
153 strlen(diff_arg) - 9);
154 for (i = 0; i < 2; i++) {
155 input_name_sq[i] = sq_quote(temp[i].name);
156 if (!strcmp(temp[i].name, "/dev/null")) {
157 path0[i] = "/dev/null";
158 path1[i] = "";
159 } else {
160 path0[i] = i ? "b/" : "a/";
161 path1[i] = name_sq[i];
163 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
164 strlen(input_name_sq[i]));
167 cmd = xmalloc(cmd_size);
169 next_at = 0;
170 next_at += snprintf(cmd+next_at, cmd_size-next_at,
171 diff_cmd,
172 path0[0], path1[0], path0[1], path1[1]);
173 next_at += snprintf(cmd+next_at, cmd_size-next_at,
174 " %s ", diff_opts);
175 next_at += snprintf(cmd+next_at, cmd_size-next_at,
176 diff_arg, input_name_sq[0], input_name_sq[1]);
178 printf("diff --git a/%s b/%s\n", name_a, name_b);
179 if (!path1[0][0]) {
180 printf("new file mode %s\n", temp[1].mode);
181 if (xfrm_msg && xfrm_msg[0])
182 puts(xfrm_msg);
184 else if (!path1[1][0]) {
185 printf("deleted file mode %s\n", temp[0].mode);
186 if (xfrm_msg && xfrm_msg[0])
187 puts(xfrm_msg);
189 else {
190 if (strcmp(temp[0].mode, temp[1].mode)) {
191 printf("old mode %s\n", temp[0].mode);
192 printf("new mode %s\n", temp[1].mode);
194 if (xfrm_msg && xfrm_msg[0])
195 puts(xfrm_msg);
196 if (strncmp(temp[0].mode, temp[1].mode, 3))
197 /* we do not run diff between different kind
198 * of objects.
200 exit(0);
201 if (complete_rewrite) {
202 fflush(NULL);
203 emit_rewrite_diff(name_a, name_b, temp);
204 exit(0);
207 fflush(NULL);
208 execlp("/bin/sh","sh", "-c", cmd, NULL);
211 struct diff_filespec *alloc_filespec(const char *path)
213 int namelen = strlen(path);
214 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
216 memset(spec, 0, sizeof(*spec));
217 spec->path = (char *)(spec + 1);
218 memcpy(spec->path, path, namelen+1);
219 return spec;
222 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
223 unsigned short mode)
225 if (mode) {
226 spec->mode = DIFF_FILE_CANON_MODE(mode);
227 memcpy(spec->sha1, sha1, 20);
228 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
233 * Given a name and sha1 pair, if the dircache tells us the file in
234 * the work tree has that object contents, return true, so that
235 * prepare_temp_file() does not have to inflate and extract.
237 static int work_tree_matches(const char *name, const unsigned char *sha1)
239 struct cache_entry *ce;
240 struct stat st;
241 int pos, len;
243 /* We do not read the cache ourselves here, because the
244 * benchmark with my previous version that always reads cache
245 * shows that it makes things worse for diff-tree comparing
246 * two linux-2.6 kernel trees in an already checked out work
247 * tree. This is because most diff-tree comparisons deal with
248 * only a small number of files, while reading the cache is
249 * expensive for a large project, and its cost outweighs the
250 * savings we get by not inflating the object to a temporary
251 * file. Practically, this code only helps when we are used
252 * by diff-cache --cached, which does read the cache before
253 * calling us.
255 if (!active_cache)
256 return 0;
258 len = strlen(name);
259 pos = cache_name_pos(name, len);
260 if (pos < 0)
261 return 0;
262 ce = active_cache[pos];
263 if ((lstat(name, &st) < 0) ||
264 !S_ISREG(st.st_mode) || /* careful! */
265 ce_match_stat(ce, &st) ||
266 memcmp(sha1, ce->sha1, 20))
267 return 0;
268 /* we return 1 only when we can stat, it is a regular file,
269 * stat information matches, and sha1 recorded in the cache
270 * matches. I.e. we know the file in the work tree really is
271 * the same as the <name, sha1> pair.
273 return 1;
276 static struct sha1_size_cache {
277 unsigned char sha1[20];
278 unsigned long size;
279 } **sha1_size_cache;
280 static int sha1_size_cache_nr, sha1_size_cache_alloc;
282 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
283 int find_only,
284 unsigned long size)
286 int first, last;
287 struct sha1_size_cache *e;
289 first = 0;
290 last = sha1_size_cache_nr;
291 while (last > first) {
292 int cmp, next = (last + first) >> 1;
293 e = sha1_size_cache[next];
294 cmp = memcmp(e->sha1, sha1, 20);
295 if (!cmp)
296 return e;
297 if (cmp < 0) {
298 last = next;
299 continue;
301 first = next+1;
303 /* not found */
304 if (find_only)
305 return NULL;
306 /* insert to make it at "first" */
307 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
308 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
309 sha1_size_cache = xrealloc(sha1_size_cache,
310 sha1_size_cache_alloc *
311 sizeof(*sha1_size_cache));
313 sha1_size_cache_nr++;
314 if (first < sha1_size_cache_nr)
315 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
316 (sha1_size_cache_nr - first - 1) *
317 sizeof(*sha1_size_cache));
318 e = xmalloc(sizeof(struct sha1_size_cache));
319 sha1_size_cache[first] = e;
320 memcpy(e->sha1, sha1, 20);
321 e->size = size;
322 return e;
326 * While doing rename detection and pickaxe operation, we may need to
327 * grab the data for the blob (or file) for our own in-core comparison.
328 * diff_filespec has data and size fields for this purpose.
330 int diff_populate_filespec(struct diff_filespec *s, int size_only)
332 int err = 0;
333 if (!DIFF_FILE_VALID(s))
334 die("internal error: asking to populate invalid file.");
335 if (S_ISDIR(s->mode))
336 return -1;
338 if (!use_size_cache)
339 size_only = 0;
341 if (s->data)
342 return err;
343 if (!s->sha1_valid ||
344 work_tree_matches(s->path, s->sha1)) {
345 struct stat st;
346 int fd;
347 if (lstat(s->path, &st) < 0) {
348 if (errno == ENOENT) {
349 err_empty:
350 err = -1;
351 empty:
352 s->data = "";
353 s->size = 0;
354 return err;
357 s->size = st.st_size;
358 if (!s->size)
359 goto empty;
360 if (size_only)
361 return 0;
362 if (S_ISLNK(st.st_mode)) {
363 int ret;
364 s->data = xmalloc(s->size);
365 s->should_free = 1;
366 ret = readlink(s->path, s->data, s->size);
367 if (ret < 0) {
368 free(s->data);
369 goto err_empty;
371 return 0;
373 fd = open(s->path, O_RDONLY);
374 if (fd < 0)
375 goto err_empty;
376 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
377 close(fd);
378 if (s->data == MAP_FAILED)
379 goto err_empty;
380 s->should_munmap = 1;
382 else {
383 char type[20];
384 struct sha1_size_cache *e;
386 if (size_only) {
387 e = locate_size_cache(s->sha1, 1, 0);
388 if (e) {
389 s->size = e->size;
390 return 0;
392 if (!sha1_object_info(s->sha1, type, &s->size))
393 locate_size_cache(s->sha1, 0, s->size);
395 else {
396 s->data = read_sha1_file(s->sha1, type, &s->size);
397 s->should_free = 1;
400 return 0;
403 void diff_free_filespec_data(struct diff_filespec *s)
405 if (s->should_free)
406 free(s->data);
407 else if (s->should_munmap)
408 munmap(s->data, s->size);
409 s->should_free = s->should_munmap = 0;
410 s->data = NULL;
413 static void prep_temp_blob(struct diff_tempfile *temp,
414 void *blob,
415 unsigned long size,
416 const unsigned char *sha1,
417 int mode)
419 int fd;
421 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
422 if (fd < 0)
423 die("unable to create temp-file");
424 if (write(fd, blob, size) != size)
425 die("unable to write temp-file");
426 close(fd);
427 temp->name = temp->tmp_path;
428 strcpy(temp->hex, sha1_to_hex(sha1));
429 temp->hex[40] = 0;
430 sprintf(temp->mode, "%06o", mode);
433 static void prepare_temp_file(const char *name,
434 struct diff_tempfile *temp,
435 struct diff_filespec *one)
437 if (!DIFF_FILE_VALID(one)) {
438 not_a_valid_file:
439 /* A '-' entry produces this for file-2, and
440 * a '+' entry produces this for file-1.
442 temp->name = "/dev/null";
443 strcpy(temp->hex, ".");
444 strcpy(temp->mode, ".");
445 return;
448 if (!one->sha1_valid ||
449 work_tree_matches(name, one->sha1)) {
450 struct stat st;
451 if (lstat(name, &st) < 0) {
452 if (errno == ENOENT)
453 goto not_a_valid_file;
454 die("stat(%s): %s", name, strerror(errno));
456 if (S_ISLNK(st.st_mode)) {
457 int ret;
458 char *buf, buf_[1024];
459 buf = ((sizeof(buf_) < st.st_size) ?
460 xmalloc(st.st_size) : buf_);
461 ret = readlink(name, buf, st.st_size);
462 if (ret < 0)
463 die("readlink(%s)", name);
464 prep_temp_blob(temp, buf, st.st_size,
465 (one->sha1_valid ?
466 one->sha1 : null_sha1),
467 (one->sha1_valid ?
468 one->mode : S_IFLNK));
470 else {
471 /* we can borrow from the file in the work tree */
472 temp->name = name;
473 if (!one->sha1_valid)
474 strcpy(temp->hex, sha1_to_hex(null_sha1));
475 else
476 strcpy(temp->hex, sha1_to_hex(one->sha1));
477 /* Even though we may sometimes borrow the
478 * contents from the work tree, we always want
479 * one->mode. mode is trustworthy even when
480 * !(one->sha1_valid), as long as
481 * DIFF_FILE_VALID(one).
483 sprintf(temp->mode, "%06o", one->mode);
485 return;
487 else {
488 if (diff_populate_filespec(one, 0))
489 die("cannot read data blob for %s", one->path);
490 prep_temp_blob(temp, one->data, one->size,
491 one->sha1, one->mode);
495 static void remove_tempfile(void)
497 int i;
499 for (i = 0; i < 2; i++)
500 if (diff_temp[i].name == diff_temp[i].tmp_path) {
501 unlink(diff_temp[i].name);
502 diff_temp[i].name = NULL;
506 static void remove_tempfile_on_signal(int signo)
508 remove_tempfile();
511 /* An external diff command takes:
513 * diff-cmd name infile1 infile1-sha1 infile1-mode \
514 * infile2 infile2-sha1 infile2-mode [ rename-to ]
517 static void run_external_diff(const char *pgm,
518 const char *name,
519 const char *other,
520 struct diff_filespec *one,
521 struct diff_filespec *two,
522 const char *xfrm_msg,
523 int complete_rewrite)
525 struct diff_tempfile *temp = diff_temp;
526 pid_t pid;
527 int status;
528 static int atexit_asked = 0;
529 const char *othername;
531 othername = (other? other : name);
532 if (one && two) {
533 prepare_temp_file(name, &temp[0], one);
534 prepare_temp_file(othername, &temp[1], two);
535 if (! atexit_asked &&
536 (temp[0].name == temp[0].tmp_path ||
537 temp[1].name == temp[1].tmp_path)) {
538 atexit_asked = 1;
539 atexit(remove_tempfile);
541 signal(SIGINT, remove_tempfile_on_signal);
544 fflush(NULL);
545 pid = fork();
546 if (pid < 0)
547 die("unable to fork");
548 if (!pid) {
549 if (pgm) {
550 if (one && two) {
551 const char *exec_arg[10];
552 const char **arg = &exec_arg[0];
553 *arg++ = pgm;
554 *arg++ = name;
555 *arg++ = temp[0].name;
556 *arg++ = temp[0].hex;
557 *arg++ = temp[0].mode;
558 *arg++ = temp[1].name;
559 *arg++ = temp[1].hex;
560 *arg++ = temp[1].mode;
561 if (other) {
562 *arg++ = other;
563 *arg++ = xfrm_msg;
565 *arg = NULL;
566 execvp(pgm, (char *const*) exec_arg);
568 else
569 execlp(pgm, pgm, name, NULL);
572 * otherwise we use the built-in one.
574 if (one && two)
575 builtin_diff(name, othername, temp, xfrm_msg,
576 complete_rewrite);
577 else
578 printf("* Unmerged path %s\n", name);
579 exit(0);
581 if (waitpid(pid, &status, 0) < 0 ||
582 !WIFEXITED(status) || WEXITSTATUS(status)) {
583 /* Earlier we did not check the exit status because
584 * diff exits non-zero if files are different, and
585 * we are not interested in knowing that. It was a
586 * mistake which made it harder to quit a diff-*
587 * session that uses the git-apply-patch-script as
588 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
589 * should also exit non-zero only when it wants to
590 * abort the entire diff-* session.
592 remove_tempfile();
593 fprintf(stderr, "external diff died, stopping at %s.\n", name);
594 exit(1);
596 remove_tempfile();
599 static void diff_fill_sha1_info(struct diff_filespec *one)
601 if (DIFF_FILE_VALID(one)) {
602 if (!one->sha1_valid) {
603 struct stat st;
604 if (stat(one->path, &st) < 0)
605 die("stat %s", one->path);
606 if (index_path(one->sha1, one->path, &st, 0))
607 die("cannot hash %s\n", one->path);
610 else
611 memset(one->sha1, 0, 20);
614 static void run_diff(struct diff_filepair *p)
616 const char *pgm = external_diff();
617 char msg[PATH_MAX*2+300], *xfrm_msg;
618 struct diff_filespec *one;
619 struct diff_filespec *two;
620 const char *name;
621 const char *other;
622 int complete_rewrite = 0;
623 int len;
625 if (DIFF_PAIR_UNMERGED(p)) {
626 /* unmerged */
627 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
629 return;
632 name = p->one->path;
633 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
634 one = p->one; two = p->two;
636 diff_fill_sha1_info(one);
637 diff_fill_sha1_info(two);
639 len = 0;
640 switch (p->status) {
641 case DIFF_STATUS_COPIED:
642 len += snprintf(msg + len, sizeof(msg) - len,
643 "similarity index %d%%\n"
644 "copy from %s\n"
645 "copy to %s\n",
646 (int)(0.5 + p->score * 100.0/MAX_SCORE),
647 name, other);
648 break;
649 case DIFF_STATUS_RENAMED:
650 len += snprintf(msg + len, sizeof(msg) - len,
651 "similarity index %d%%\n"
652 "rename from %s\n"
653 "rename to %s\n",
654 (int)(0.5 + p->score * 100.0/MAX_SCORE),
655 name, other);
656 break;
657 case DIFF_STATUS_MODIFIED:
658 if (p->score) {
659 len += snprintf(msg + len, sizeof(msg) - len,
660 "dissimilarity index %d%%\n",
661 (int)(0.5 + p->score *
662 100.0/MAX_SCORE));
663 complete_rewrite = 1;
664 break;
666 /* fallthru */
667 default:
668 /* nothing */
672 if (memcmp(one->sha1, two->sha1, 20)) {
673 char one_sha1[41];
674 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
676 len += snprintf(msg + len, sizeof(msg) - len,
677 "index %.7s..%.7s", one_sha1,
678 sha1_to_hex(two->sha1));
679 if (one->mode == two->mode)
680 len += snprintf(msg + len, sizeof(msg) - len,
681 " %06o", one->mode);
682 len += snprintf(msg + len, sizeof(msg) - len, "\n");
685 if (len)
686 msg[--len] = 0;
687 xfrm_msg = len ? msg : NULL;
689 if (!pgm &&
690 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
691 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
692 /* a filepair that changes between file and symlink
693 * needs to be split into deletion and creation.
695 struct diff_filespec *null = alloc_filespec(two->path);
696 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
697 free(null);
698 null = alloc_filespec(one->path);
699 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
700 free(null);
702 else
703 run_external_diff(pgm, name, other, one, two, xfrm_msg,
704 complete_rewrite);
707 void diff_setup(struct diff_options *options)
709 memset(options, 0, sizeof(*options));
710 options->output_format = DIFF_FORMAT_RAW;
711 options->line_termination = '\n';
712 options->break_opt = -1;
713 options->rename_limit = -1;
716 int diff_setup_done(struct diff_options *options)
718 if ((options->find_copies_harder || 0 <= options->rename_limit) &&
719 options->detect_rename != DIFF_DETECT_COPY)
720 return -1;
721 if (options->setup & DIFF_SETUP_USE_CACHE) {
722 if (!active_cache)
723 /* read-cache does not die even when it fails
724 * so it is safe for us to do this here. Also
725 * it does not smudge active_cache or active_nr
726 * when it fails, so we do not have to worry about
727 * cleaning it up oufselves either.
729 read_cache();
731 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
732 use_size_cache = 1;
734 return 0;
737 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
739 const char *arg = av[0];
740 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
741 options->output_format = DIFF_FORMAT_PATCH;
742 else if (!strcmp(arg, "-z"))
743 options->line_termination = 0;
744 else if (!strncmp(arg, "-l", 2))
745 options->rename_limit = strtoul(arg+2, NULL, 10);
746 else if (!strcmp(arg, "--name-only"))
747 options->output_format = DIFF_FORMAT_NAME;
748 else if (!strcmp(arg, "--name-status"))
749 options->output_format = DIFF_FORMAT_NAME_STATUS;
750 else if (!strcmp(arg, "-R"))
751 options->reverse_diff = 1;
752 else if (!strncmp(arg, "-S", 2))
753 options->pickaxe = arg + 2;
754 else if (!strcmp(arg, "-s"))
755 options->output_format = DIFF_FORMAT_NO_OUTPUT;
756 else if (!strncmp(arg, "-O", 2))
757 options->orderfile = arg + 2;
758 else if (!strncmp(arg, "--diff-filter=", 14))
759 options->filter = arg + 14;
760 else if (!strcmp(arg, "--pickaxe-all"))
761 options->pickaxe_opts = DIFF_PICKAXE_ALL;
762 else if (!strncmp(arg, "-B", 2)) {
763 if ((options->break_opt =
764 diff_scoreopt_parse(arg)) == -1)
765 return -1;
767 else if (!strncmp(arg, "-M", 2)) {
768 if ((options->rename_score =
769 diff_scoreopt_parse(arg)) == -1)
770 return -1;
771 options->detect_rename = DIFF_DETECT_RENAME;
773 else if (!strncmp(arg, "-C", 2)) {
774 if ((options->rename_score =
775 diff_scoreopt_parse(arg)) == -1)
776 return -1;
777 options->detect_rename = DIFF_DETECT_COPY;
779 else if (!strcmp(arg, "--find-copies-harder"))
780 options->find_copies_harder = 1;
781 else
782 return 0;
783 return 1;
786 static int parse_num(const char **cp_p)
788 int num, scale, ch, cnt;
789 const char *cp = *cp_p;
791 cnt = num = 0;
792 scale = 1;
793 while ('0' <= (ch = *cp) && ch <= '9') {
794 if (cnt++ < 5) {
795 /* We simply ignore more than 5 digits precision. */
796 scale *= 10;
797 num = num * 10 + ch - '0';
799 cp++;
801 *cp_p = cp;
803 /* user says num divided by scale and we say internally that
804 * is MAX_SCORE * num / scale.
806 return (MAX_SCORE * num / scale);
809 int diff_scoreopt_parse(const char *opt)
811 int opt1, opt2, cmd;
813 if (*opt++ != '-')
814 return -1;
815 cmd = *opt++;
816 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
817 return -1; /* that is not a -M, -C nor -B option */
819 opt1 = parse_num(&opt);
820 if (cmd != 'B')
821 opt2 = 0;
822 else {
823 if (*opt == 0)
824 opt2 = 0;
825 else if (*opt != '/')
826 return -1; /* we expect -B80/99 or -B80 */
827 else {
828 opt++;
829 opt2 = parse_num(&opt);
832 if (*opt != 0)
833 return -1;
834 return opt1 | (opt2 << 16);
837 struct diff_queue_struct diff_queued_diff;
839 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
841 if (queue->alloc <= queue->nr) {
842 queue->alloc = alloc_nr(queue->alloc);
843 queue->queue = xrealloc(queue->queue,
844 sizeof(dp) * queue->alloc);
846 queue->queue[queue->nr++] = dp;
849 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
850 struct diff_filespec *one,
851 struct diff_filespec *two)
853 struct diff_filepair *dp = xmalloc(sizeof(*dp));
854 dp->one = one;
855 dp->two = two;
856 dp->score = 0;
857 dp->status = 0;
858 dp->source_stays = 0;
859 dp->broken_pair = 0;
860 if (queue)
861 diff_q(queue, dp);
862 return dp;
865 void diff_free_filepair(struct diff_filepair *p)
867 diff_free_filespec_data(p->one);
868 diff_free_filespec_data(p->two);
869 free(p->one);
870 free(p->two);
871 free(p);
874 static void diff_flush_raw(struct diff_filepair *p,
875 int line_termination,
876 int inter_name_termination,
877 int output_format)
879 int two_paths;
880 char status[10];
882 if (line_termination) {
883 const char *const err =
884 "path %s cannot be expressed without -z";
885 if (strchr(p->one->path, line_termination) ||
886 strchr(p->one->path, inter_name_termination))
887 die(err, p->one->path);
888 if (strchr(p->two->path, line_termination) ||
889 strchr(p->two->path, inter_name_termination))
890 die(err, p->two->path);
893 if (p->score)
894 sprintf(status, "%c%03d", p->status,
895 (int)(0.5 + p->score * 100.0/MAX_SCORE));
896 else {
897 status[0] = p->status;
898 status[1] = 0;
900 switch (p->status) {
901 case DIFF_STATUS_COPIED:
902 case DIFF_STATUS_RENAMED:
903 two_paths = 1;
904 break;
905 case DIFF_STATUS_ADDED:
906 case DIFF_STATUS_DELETED:
907 two_paths = 0;
908 break;
909 default:
910 two_paths = 0;
911 break;
913 if (output_format != DIFF_FORMAT_NAME_STATUS) {
914 printf(":%06o %06o %s ",
915 p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
916 printf("%s ", sha1_to_hex(p->two->sha1));
918 printf("%s%c%s",status, inter_name_termination, p->one->path);
919 if (two_paths)
920 printf("%c%s", inter_name_termination, p->two->path);
921 putchar(line_termination);
924 static void diff_flush_name(struct diff_filepair *p,
925 int line_termination)
927 printf("%s%c", p->two->path, line_termination);
930 int diff_unmodified_pair(struct diff_filepair *p)
932 /* This function is written stricter than necessary to support
933 * the currently implemented transformers, but the idea is to
934 * let transformers to produce diff_filepairs any way they want,
935 * and filter and clean them up here before producing the output.
937 struct diff_filespec *one, *two;
939 if (DIFF_PAIR_UNMERGED(p))
940 return 0; /* unmerged is interesting */
942 one = p->one;
943 two = p->two;
945 /* deletion, addition, mode or type change
946 * and rename are all interesting.
948 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
949 DIFF_PAIR_MODE_CHANGED(p) ||
950 strcmp(one->path, two->path))
951 return 0;
953 /* both are valid and point at the same path. that is, we are
954 * dealing with a change.
956 if (one->sha1_valid && two->sha1_valid &&
957 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
958 return 1; /* no change */
959 if (!one->sha1_valid && !two->sha1_valid)
960 return 1; /* both look at the same file on the filesystem. */
961 return 0;
964 static void diff_flush_patch(struct diff_filepair *p)
966 if (diff_unmodified_pair(p))
967 return;
969 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
970 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
971 return; /* no tree diffs in patch format */
973 run_diff(p);
976 int diff_queue_is_empty(void)
978 struct diff_queue_struct *q = &diff_queued_diff;
979 int i;
980 for (i = 0; i < q->nr; i++)
981 if (!diff_unmodified_pair(q->queue[i]))
982 return 0;
983 return 1;
986 #if DIFF_DEBUG
987 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
989 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
990 x, one ? one : "",
991 s->path,
992 DIFF_FILE_VALID(s) ? "valid" : "invalid",
993 s->mode,
994 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
995 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
996 x, one ? one : "",
997 s->size, s->xfrm_flags);
1000 void diff_debug_filepair(const struct diff_filepair *p, int i)
1002 diff_debug_filespec(p->one, i, "one");
1003 diff_debug_filespec(p->two, i, "two");
1004 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1005 p->score, p->status ? p->status : '?',
1006 p->source_stays, p->broken_pair);
1009 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1011 int i;
1012 if (msg)
1013 fprintf(stderr, "%s\n", msg);
1014 fprintf(stderr, "q->nr = %d\n", q->nr);
1015 for (i = 0; i < q->nr; i++) {
1016 struct diff_filepair *p = q->queue[i];
1017 diff_debug_filepair(p, i);
1020 #endif
1022 static void diff_resolve_rename_copy(void)
1024 int i, j;
1025 struct diff_filepair *p, *pp;
1026 struct diff_queue_struct *q = &diff_queued_diff;
1028 diff_debug_queue("resolve-rename-copy", q);
1030 for (i = 0; i < q->nr; i++) {
1031 p = q->queue[i];
1032 p->status = 0; /* undecided */
1033 if (DIFF_PAIR_UNMERGED(p))
1034 p->status = DIFF_STATUS_UNMERGED;
1035 else if (!DIFF_FILE_VALID(p->one))
1036 p->status = DIFF_STATUS_ADDED;
1037 else if (!DIFF_FILE_VALID(p->two))
1038 p->status = DIFF_STATUS_DELETED;
1039 else if (DIFF_PAIR_TYPE_CHANGED(p))
1040 p->status = DIFF_STATUS_TYPE_CHANGED;
1042 /* from this point on, we are dealing with a pair
1043 * whose both sides are valid and of the same type, i.e.
1044 * either in-place edit or rename/copy edit.
1046 else if (DIFF_PAIR_RENAME(p)) {
1047 if (p->source_stays) {
1048 p->status = DIFF_STATUS_COPIED;
1049 continue;
1051 /* See if there is some other filepair that
1052 * copies from the same source as us. If so
1053 * we are a copy. Otherwise we are either a
1054 * copy if the path stays, or a rename if it
1055 * does not, but we already handled "stays" case.
1057 for (j = i + 1; j < q->nr; j++) {
1058 pp = q->queue[j];
1059 if (strcmp(pp->one->path, p->one->path))
1060 continue; /* not us */
1061 if (!DIFF_PAIR_RENAME(pp))
1062 continue; /* not a rename/copy */
1063 /* pp is a rename/copy from the same source */
1064 p->status = DIFF_STATUS_COPIED;
1065 break;
1067 if (!p->status)
1068 p->status = DIFF_STATUS_RENAMED;
1070 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1071 p->one->mode != p->two->mode)
1072 p->status = DIFF_STATUS_MODIFIED;
1073 else {
1074 /* This is a "no-change" entry and should not
1075 * happen anymore, but prepare for broken callers.
1077 error("feeding unmodified %s to diffcore",
1078 p->one->path);
1079 p->status = DIFF_STATUS_UNKNOWN;
1082 diff_debug_queue("resolve-rename-copy done", q);
1085 void diff_flush(struct diff_options *options)
1087 struct diff_queue_struct *q = &diff_queued_diff;
1088 int i;
1089 int inter_name_termination = '\t';
1090 int diff_output_format = options->output_format;
1091 int line_termination = options->line_termination;
1093 if (!line_termination)
1094 inter_name_termination = 0;
1096 for (i = 0; i < q->nr; i++) {
1097 struct diff_filepair *p = q->queue[i];
1098 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1099 (p->status == DIFF_STATUS_UNKNOWN))
1100 continue;
1101 if (p->status == 0)
1102 die("internal error in diff-resolve-rename-copy");
1103 switch (diff_output_format) {
1104 case DIFF_FORMAT_PATCH:
1105 diff_flush_patch(p);
1106 break;
1107 case DIFF_FORMAT_RAW:
1108 case DIFF_FORMAT_NAME_STATUS:
1109 diff_flush_raw(p, line_termination,
1110 inter_name_termination,
1111 diff_output_format);
1112 break;
1113 case DIFF_FORMAT_NAME:
1114 diff_flush_name(p, line_termination);
1115 break;
1117 diff_free_filepair(q->queue[i]);
1119 free(q->queue);
1120 q->queue = NULL;
1121 q->nr = q->alloc = 0;
1124 static void diffcore_apply_filter(const char *filter)
1126 int i;
1127 struct diff_queue_struct *q = &diff_queued_diff;
1128 struct diff_queue_struct outq;
1129 outq.queue = NULL;
1130 outq.nr = outq.alloc = 0;
1132 if (!filter)
1133 return;
1135 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1136 int found;
1137 for (i = found = 0; !found && i < q->nr; i++) {
1138 struct diff_filepair *p = q->queue[i];
1139 if (((p->status == DIFF_STATUS_MODIFIED) &&
1140 ((p->score &&
1141 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1142 (!p->score &&
1143 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1144 ((p->status != DIFF_STATUS_MODIFIED) &&
1145 strchr(filter, p->status)))
1146 found++;
1148 if (found)
1149 return;
1151 /* otherwise we will clear the whole queue
1152 * by copying the empty outq at the end of this
1153 * function, but first clear the current entries
1154 * in the queue.
1156 for (i = 0; i < q->nr; i++)
1157 diff_free_filepair(q->queue[i]);
1159 else {
1160 /* Only the matching ones */
1161 for (i = 0; i < q->nr; i++) {
1162 struct diff_filepair *p = q->queue[i];
1164 if (((p->status == DIFF_STATUS_MODIFIED) &&
1165 ((p->score &&
1166 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1167 (!p->score &&
1168 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1169 ((p->status != DIFF_STATUS_MODIFIED) &&
1170 strchr(filter, p->status)))
1171 diff_q(&outq, p);
1172 else
1173 diff_free_filepair(p);
1176 free(q->queue);
1177 *q = outq;
1180 void diffcore_std(struct diff_options *options)
1182 if (options->paths && options->paths[0])
1183 diffcore_pathspec(options->paths);
1184 if (options->break_opt != -1)
1185 diffcore_break(options->break_opt);
1186 if (options->detect_rename)
1187 diffcore_rename(options);
1188 if (options->break_opt != -1)
1189 diffcore_merge_broken();
1190 if (options->pickaxe)
1191 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1192 if (options->orderfile)
1193 diffcore_order(options->orderfile);
1194 diff_resolve_rename_copy();
1195 diffcore_apply_filter(options->filter);
1199 void diffcore_std_no_resolve(struct diff_options *options)
1201 if (options->pickaxe)
1202 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1203 if (options->orderfile)
1204 diffcore_order(options->orderfile);
1205 diffcore_apply_filter(options->filter);
1208 void diff_addremove(struct diff_options *options,
1209 int addremove, unsigned mode,
1210 const unsigned char *sha1,
1211 const char *base, const char *path)
1213 char concatpath[PATH_MAX];
1214 struct diff_filespec *one, *two;
1216 /* This may look odd, but it is a preparation for
1217 * feeding "there are unchanged files which should
1218 * not produce diffs, but when you are doing copy
1219 * detection you would need them, so here they are"
1220 * entries to the diff-core. They will be prefixed
1221 * with something like '=' or '*' (I haven't decided
1222 * which but should not make any difference).
1223 * Feeding the same new and old to diff_change()
1224 * also has the same effect.
1225 * Before the final output happens, they are pruned after
1226 * merged into rename/copy pairs as appropriate.
1228 if (options->reverse_diff)
1229 addremove = (addremove == '+' ? '-' :
1230 addremove == '-' ? '+' : addremove);
1232 if (!path) path = "";
1233 sprintf(concatpath, "%s%s", base, path);
1234 one = alloc_filespec(concatpath);
1235 two = alloc_filespec(concatpath);
1237 if (addremove != '+')
1238 fill_filespec(one, sha1, mode);
1239 if (addremove != '-')
1240 fill_filespec(two, sha1, mode);
1242 diff_queue(&diff_queued_diff, one, two);
1245 void diff_change(struct diff_options *options,
1246 unsigned old_mode, unsigned new_mode,
1247 const unsigned char *old_sha1,
1248 const unsigned char *new_sha1,
1249 const char *base, const char *path)
1251 char concatpath[PATH_MAX];
1252 struct diff_filespec *one, *two;
1254 if (options->reverse_diff) {
1255 unsigned tmp;
1256 const unsigned char *tmp_c;
1257 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1258 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1260 if (!path) path = "";
1261 sprintf(concatpath, "%s%s", base, path);
1262 one = alloc_filespec(concatpath);
1263 two = alloc_filespec(concatpath);
1264 fill_filespec(one, old_sha1, old_mode);
1265 fill_filespec(two, new_sha1, new_mode);
1267 diff_queue(&diff_queued_diff, one, two);
1270 void diff_unmerge(struct diff_options *options,
1271 const char *path)
1273 struct diff_filespec *one, *two;
1274 one = alloc_filespec(path);
1275 two = alloc_filespec(path);
1276 diff_queue(&diff_queued_diff, one, two);