2 * Copyright (C) 2005 Junio C Hamano
12 #include "xdiff-interface.h"
14 static int use_size_cache;
16 int diff_rename_limit_default = -1;
18 int git_diff_config(const char *var, const char *value)
20 if (!strcmp(var, "diff.renamelimit")) {
21 diff_rename_limit_default = git_config_int(var, value);
25 return git_default_config(var, value);
28 static char *quote_one(const char *str)
35 needlen = quote_c_style(str, NULL, NULL, 0);
38 xp = xmalloc(needlen + 1);
39 quote_c_style(str, xp, NULL, 0);
43 static char *quote_two(const char *one, const char *two)
45 int need_one = quote_c_style(one, NULL, NULL, 1);
46 int need_two = quote_c_style(two, NULL, NULL, 1);
49 if (need_one + need_two) {
50 if (!need_one) need_one = strlen(one);
51 if (!need_two) need_one = strlen(two);
53 xp = xmalloc(need_one + need_two + 3);
55 quote_c_style(one, xp + 1, NULL, 1);
56 quote_c_style(two, xp + need_one + 1, NULL, 1);
57 strcpy(xp + need_one + need_two + 1, "\"");
60 need_one = strlen(one);
61 need_two = strlen(two);
62 xp = xmalloc(need_one + need_two + 1);
64 strcpy(xp + need_one, two);
68 static const char *external_diff(void)
70 static const char *external_diff_cmd = NULL;
71 static int done_preparing = 0;
74 return external_diff_cmd;
75 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
77 return external_diff_cmd;
80 #define TEMPFILE_PATH_LEN 50
82 static struct diff_tempfile {
83 const char *name; /* filename external diff should read from */
86 char tmp_path[TEMPFILE_PATH_LEN];
89 static int count_lines(const char *data, int size)
91 int count, ch, completely_empty = 1, nl_just_seen = 0;
102 completely_empty = 0;
105 if (completely_empty)
108 count++; /* no trailing newline */
112 static void print_line_count(int count)
122 printf("1,%d", count);
127 static void copy_file(int prefix, const char *data, int size)
129 int ch, nl_just_seen = 1;
141 printf("\n\\ No newline at end of file\n");
144 static void emit_rewrite_diff(const char *name_a,
146 struct diff_filespec *one,
147 struct diff_filespec *two)
150 diff_populate_filespec(one, 0);
151 diff_populate_filespec(two, 0);
152 lc_a = count_lines(one->data, one->size);
153 lc_b = count_lines(two->data, two->size);
154 printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
155 print_line_count(lc_a);
157 print_line_count(lc_b);
160 copy_file('-', one->data, one->size);
162 copy_file('+', two->data, two->size);
165 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
167 if (!DIFF_FILE_VALID(one)) {
168 mf->ptr = ""; /* does not matter */
172 else if (diff_populate_filespec(one, 0))
175 mf->size = one->size;
179 struct emit_callback {
180 const char **label_path;
183 static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
186 struct emit_callback *ecbdata = priv;
188 if (ecbdata->label_path[0]) {
189 printf("--- %s\n", ecbdata->label_path[0]);
190 printf("+++ %s\n", ecbdata->label_path[1]);
191 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
193 for (i = 0; i < nbuf; i++)
194 if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
199 static char *pprint_rename(const char *a, const char *b)
204 int pfx_length, sfx_length;
205 int len_a = strlen(a);
206 int len_b = strlen(b);
208 /* Find common prefix */
210 while (*old && *new && *old == *new) {
212 pfx_length = old - a + 1;
217 /* Find common suffix */
221 while (a <= old && b <= new && *old == *new) {
223 sfx_length = len_a - (old - a);
229 * pfx{mid-a => mid-b}sfx
230 * {pfx-a => pfx-b}sfx
231 * pfx{sfx-a => sfx-b}
234 if (pfx_length + sfx_length) {
235 int a_midlen = len_a - pfx_length - sfx_length;
236 int b_midlen = len_b - pfx_length - sfx_length;
237 if (a_midlen < 0) a_midlen = 0;
238 if (b_midlen < 0) b_midlen = 0;
240 name = xmalloc(len_a + len_b - pfx_length - sfx_length + 7);
241 sprintf(name, "%.*s{%.*s => %.*s}%s",
243 a_midlen, a + pfx_length,
244 b_midlen, b + pfx_length,
245 a + len_a - sfx_length);
248 name = xmalloc(len_a + len_b + 5);
249 sprintf(name, "%s => %s", a, b);
255 struct xdiff_emit_state xm;
259 struct diffstat_file {
261 unsigned is_unmerged:1;
262 unsigned is_binary:1;
263 unsigned is_renamed:1;
264 unsigned int added, deleted;
268 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
272 struct diffstat_file *x;
273 x = xcalloc(sizeof (*x), 1);
274 if (diffstat->nr == diffstat->alloc) {
275 diffstat->alloc = alloc_nr(diffstat->alloc);
276 diffstat->files = xrealloc(diffstat->files,
277 diffstat->alloc * sizeof(x));
279 diffstat->files[diffstat->nr++] = x;
281 x->name = pprint_rename(name_a, name_b);
285 x->name = strdup(name_a);
289 static void diffstat_consume(void *priv, char *line, unsigned long len)
291 struct diffstat_t *diffstat = priv;
292 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
296 else if (line[0] == '-')
300 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
301 static const char minuses[]= "----------------------------------------------------------------------";
303 static void show_stats(struct diffstat_t* data)
305 int i, len, add, del, total, adds = 0, dels = 0;
306 int max, max_change = 0, max_len = 0;
307 int total_files = data->nr;
312 for (i = 0; i < data->nr; i++) {
313 struct diffstat_file *file = data->files[i];
315 len = strlen(file->name);
319 if (file->is_binary || file->is_unmerged)
321 if (max_change < file->added + file->deleted)
322 max_change = file->added + file->deleted;
325 for (i = 0; i < data->nr; i++) {
327 char *name = data->files[i]->name;
328 int added = data->files[i]->added;
329 int deleted = data->files[i]->deleted;
331 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
332 char *qname = xmalloc(len + 1);
333 quote_c_style(name, qname, NULL, 0);
335 data->files[i]->name = name = qname;
339 * "scale" the filename
350 slash = strchr(name, '/');
357 * scale the add/delete
363 if (data->files[i]->is_binary) {
364 printf(" %s%-*s | Bin\n", prefix, len, name);
365 goto free_diffstat_file;
367 else if (data->files[i]->is_unmerged) {
368 printf(" %s%-*s | Unmerged\n", prefix, len, name);
369 goto free_diffstat_file;
371 else if (!data->files[i]->is_renamed &&
372 (added + deleted == 0)) {
374 goto free_diffstat_file;
383 if (max_change > 0) {
384 total = (total * max + max_change / 2) / max_change;
385 add = (add * max + max_change / 2) / max_change;
388 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
389 len, name, added + deleted,
390 add, pluses, del, minuses);
392 free(data->files[i]->name);
393 free(data->files[i]);
396 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
397 total_files, adds, dels);
400 static unsigned char *deflate_it(char *data,
402 unsigned long *result_size)
405 unsigned char *deflated;
408 memset(&stream, 0, sizeof(stream));
409 deflateInit(&stream, Z_BEST_COMPRESSION);
410 bound = deflateBound(&stream, size);
411 deflated = xmalloc(bound);
412 stream.next_out = deflated;
413 stream.avail_out = bound;
415 stream.next_in = (unsigned char *)data;
416 stream.avail_in = size;
417 while (deflate(&stream, Z_FINISH) == Z_OK)
420 *result_size = stream.total_out;
424 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
430 unsigned long orig_size;
431 unsigned long delta_size;
432 unsigned long deflate_size;
433 unsigned long data_size;
435 printf("GIT binary patch\n");
436 /* We could do deflated delta, or we could do just deflated two,
437 * whichever is smaller.
440 deflated = deflate_it(two->ptr, two->size, &deflate_size);
441 if (one->size && two->size) {
442 delta = diff_delta(one->ptr, one->size,
444 &delta_size, deflate_size);
446 void *to_free = delta;
447 orig_size = delta_size;
448 delta = deflate_it(delta, delta_size, &delta_size);
453 if (delta && delta_size < deflate_size) {
454 printf("delta %lu\n", orig_size);
457 data_size = delta_size;
460 printf("literal %lu\n", two->size);
463 data_size = deflate_size;
466 /* emit data encoded in base85 */
469 int bytes = (52 < data_size) ? 52 : data_size;
473 line[0] = bytes + 'A' - 1;
475 line[0] = bytes - 26 + 'a' - 1;
476 encode_85(line + 1, cp, bytes);
484 #define FIRST_FEW_BYTES 8000
485 static int mmfile_is_binary(mmfile_t *mf)
488 if (FIRST_FEW_BYTES < sz)
489 sz = FIRST_FEW_BYTES;
490 if (memchr(mf->ptr, 0, sz))
495 static void builtin_diff(const char *name_a,
497 struct diff_filespec *one,
498 struct diff_filespec *two,
499 const char *xfrm_msg,
500 struct diff_options *o,
501 int complete_rewrite)
507 a_one = quote_two("a/", name_a);
508 b_two = quote_two("b/", name_b);
509 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
510 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
511 printf("diff --git %s %s\n", a_one, b_two);
512 if (lbl[0][0] == '/') {
514 printf("new file mode %06o\n", two->mode);
515 if (xfrm_msg && xfrm_msg[0])
518 else if (lbl[1][0] == '/') {
519 printf("deleted file mode %06o\n", one->mode);
520 if (xfrm_msg && xfrm_msg[0])
524 if (one->mode != two->mode) {
525 printf("old mode %06o\n", one->mode);
526 printf("new mode %06o\n", two->mode);
528 if (xfrm_msg && xfrm_msg[0])
531 * we do not run diff between different kind
534 if ((one->mode ^ two->mode) & S_IFMT)
535 goto free_ab_and_return;
536 if (complete_rewrite) {
537 emit_rewrite_diff(name_a, name_b, one, two);
538 goto free_ab_and_return;
542 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
543 die("unable to read files to diff");
545 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2)) {
546 /* Quite common confusing case */
547 if (mf1.size == mf2.size &&
548 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
549 goto free_ab_and_return;
551 emit_binary_diff(&mf1, &mf2);
553 printf("Binary files %s and %s differ\n",
557 /* Crazy xdl interfaces.. */
558 const char *diffopts = getenv("GIT_DIFF_OPTS");
562 struct emit_callback ecbdata;
564 ecbdata.label_path = lbl;
565 xpp.flags = XDF_NEED_MINIMAL;
567 xecfg.flags = XDL_EMIT_FUNCNAMES;
570 else if (!strncmp(diffopts, "--unified=", 10))
571 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
572 else if (!strncmp(diffopts, "-u", 2))
573 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
576 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
585 static void builtin_diffstat(const char *name_a, const char *name_b,
586 struct diff_filespec *one,
587 struct diff_filespec *two,
588 struct diffstat_t *diffstat,
589 int complete_rewrite)
592 struct diffstat_file *data;
594 data = diffstat_add(diffstat, name_a, name_b);
597 data->is_unmerged = 1;
600 if (complete_rewrite) {
601 diff_populate_filespec(one, 0);
602 diff_populate_filespec(two, 0);
603 data->deleted = count_lines(one->data, one->size);
604 data->added = count_lines(two->data, two->size);
607 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
608 die("unable to read files to diff");
610 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
613 /* Crazy xdl interfaces.. */
618 xpp.flags = XDF_NEED_MINIMAL;
621 ecb.outf = xdiff_outf;
623 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
627 struct diff_filespec *alloc_filespec(const char *path)
629 int namelen = strlen(path);
630 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
632 memset(spec, 0, sizeof(*spec));
633 spec->path = (char *)(spec + 1);
634 memcpy(spec->path, path, namelen+1);
638 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
642 spec->mode = canon_mode(mode);
643 memcpy(spec->sha1, sha1, 20);
644 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
649 * Given a name and sha1 pair, if the dircache tells us the file in
650 * the work tree has that object contents, return true, so that
651 * prepare_temp_file() does not have to inflate and extract.
653 static int work_tree_matches(const char *name, const unsigned char *sha1)
655 struct cache_entry *ce;
659 /* We do not read the cache ourselves here, because the
660 * benchmark with my previous version that always reads cache
661 * shows that it makes things worse for diff-tree comparing
662 * two linux-2.6 kernel trees in an already checked out work
663 * tree. This is because most diff-tree comparisons deal with
664 * only a small number of files, while reading the cache is
665 * expensive for a large project, and its cost outweighs the
666 * savings we get by not inflating the object to a temporary
667 * file. Practically, this code only helps when we are used
668 * by diff-cache --cached, which does read the cache before
675 pos = cache_name_pos(name, len);
678 ce = active_cache[pos];
679 if ((lstat(name, &st) < 0) ||
680 !S_ISREG(st.st_mode) || /* careful! */
681 ce_match_stat(ce, &st, 0) ||
682 memcmp(sha1, ce->sha1, 20))
684 /* we return 1 only when we can stat, it is a regular file,
685 * stat information matches, and sha1 recorded in the cache
686 * matches. I.e. we know the file in the work tree really is
687 * the same as the <name, sha1> pair.
692 static struct sha1_size_cache {
693 unsigned char sha1[20];
696 static int sha1_size_cache_nr, sha1_size_cache_alloc;
698 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
703 struct sha1_size_cache *e;
706 last = sha1_size_cache_nr;
707 while (last > first) {
708 int cmp, next = (last + first) >> 1;
709 e = sha1_size_cache[next];
710 cmp = memcmp(e->sha1, sha1, 20);
722 /* insert to make it at "first" */
723 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
724 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
725 sha1_size_cache = xrealloc(sha1_size_cache,
726 sha1_size_cache_alloc *
727 sizeof(*sha1_size_cache));
729 sha1_size_cache_nr++;
730 if (first < sha1_size_cache_nr)
731 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
732 (sha1_size_cache_nr - first - 1) *
733 sizeof(*sha1_size_cache));
734 e = xmalloc(sizeof(struct sha1_size_cache));
735 sha1_size_cache[first] = e;
736 memcpy(e->sha1, sha1, 20);
742 * While doing rename detection and pickaxe operation, we may need to
743 * grab the data for the blob (or file) for our own in-core comparison.
744 * diff_filespec has data and size fields for this purpose.
746 int diff_populate_filespec(struct diff_filespec *s, int size_only)
749 if (!DIFF_FILE_VALID(s))
750 die("internal error: asking to populate invalid file.");
751 if (S_ISDIR(s->mode))
759 if (!s->sha1_valid ||
760 work_tree_matches(s->path, s->sha1)) {
763 if (lstat(s->path, &st) < 0) {
764 if (errno == ENOENT) {
773 s->size = st.st_size;
778 if (S_ISLNK(st.st_mode)) {
780 s->data = xmalloc(s->size);
782 ret = readlink(s->path, s->data, s->size);
789 fd = open(s->path, O_RDONLY);
792 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
794 if (s->data == MAP_FAILED)
796 s->should_munmap = 1;
800 struct sha1_size_cache *e;
803 e = locate_size_cache(s->sha1, 1, 0);
808 if (!sha1_object_info(s->sha1, type, &s->size))
809 locate_size_cache(s->sha1, 0, s->size);
812 s->data = read_sha1_file(s->sha1, type, &s->size);
819 void diff_free_filespec_data(struct diff_filespec *s)
823 else if (s->should_munmap)
824 munmap(s->data, s->size);
825 s->should_free = s->should_munmap = 0;
831 static void prep_temp_blob(struct diff_tempfile *temp,
834 const unsigned char *sha1,
839 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
841 die("unable to create temp-file");
842 if (write(fd, blob, size) != size)
843 die("unable to write temp-file");
845 temp->name = temp->tmp_path;
846 strcpy(temp->hex, sha1_to_hex(sha1));
848 sprintf(temp->mode, "%06o", mode);
851 static void prepare_temp_file(const char *name,
852 struct diff_tempfile *temp,
853 struct diff_filespec *one)
855 if (!DIFF_FILE_VALID(one)) {
857 /* A '-' entry produces this for file-2, and
858 * a '+' entry produces this for file-1.
860 temp->name = "/dev/null";
861 strcpy(temp->hex, ".");
862 strcpy(temp->mode, ".");
866 if (!one->sha1_valid ||
867 work_tree_matches(name, one->sha1)) {
869 if (lstat(name, &st) < 0) {
871 goto not_a_valid_file;
872 die("stat(%s): %s", name, strerror(errno));
874 if (S_ISLNK(st.st_mode)) {
876 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
877 if (sizeof(buf) <= st.st_size)
878 die("symlink too long: %s", name);
879 ret = readlink(name, buf, st.st_size);
881 die("readlink(%s)", name);
882 prep_temp_blob(temp, buf, st.st_size,
884 one->sha1 : null_sha1),
886 one->mode : S_IFLNK));
889 /* we can borrow from the file in the work tree */
891 if (!one->sha1_valid)
892 strcpy(temp->hex, sha1_to_hex(null_sha1));
894 strcpy(temp->hex, sha1_to_hex(one->sha1));
895 /* Even though we may sometimes borrow the
896 * contents from the work tree, we always want
897 * one->mode. mode is trustworthy even when
898 * !(one->sha1_valid), as long as
899 * DIFF_FILE_VALID(one).
901 sprintf(temp->mode, "%06o", one->mode);
906 if (diff_populate_filespec(one, 0))
907 die("cannot read data blob for %s", one->path);
908 prep_temp_blob(temp, one->data, one->size,
909 one->sha1, one->mode);
913 static void remove_tempfile(void)
917 for (i = 0; i < 2; i++)
918 if (diff_temp[i].name == diff_temp[i].tmp_path) {
919 unlink(diff_temp[i].name);
920 diff_temp[i].name = NULL;
924 static void remove_tempfile_on_signal(int signo)
927 signal(SIGINT, SIG_DFL);
931 static int spawn_prog(const char *pgm, const char **arg)
939 die("unable to fork");
941 execvp(pgm, (char *const*) arg);
945 while (waitpid(pid, &status, 0) < 0) {
951 /* Earlier we did not check the exit status because
952 * diff exits non-zero if files are different, and
953 * we are not interested in knowing that. It was a
954 * mistake which made it harder to quit a diff-*
955 * session that uses the git-apply-patch-script as
956 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
957 * should also exit non-zero only when it wants to
958 * abort the entire diff-* session.
960 if (WIFEXITED(status) && !WEXITSTATUS(status))
965 /* An external diff command takes:
967 * diff-cmd name infile1 infile1-sha1 infile1-mode \
968 * infile2 infile2-sha1 infile2-mode [ rename-to ]
971 static void run_external_diff(const char *pgm,
974 struct diff_filespec *one,
975 struct diff_filespec *two,
976 const char *xfrm_msg,
977 int complete_rewrite)
979 const char *spawn_arg[10];
980 struct diff_tempfile *temp = diff_temp;
982 static int atexit_asked = 0;
983 const char *othername;
984 const char **arg = &spawn_arg[0];
986 othername = (other? other : name);
988 prepare_temp_file(name, &temp[0], one);
989 prepare_temp_file(othername, &temp[1], two);
990 if (! atexit_asked &&
991 (temp[0].name == temp[0].tmp_path ||
992 temp[1].name == temp[1].tmp_path)) {
994 atexit(remove_tempfile);
996 signal(SIGINT, remove_tempfile_on_signal);
1002 *arg++ = temp[0].name;
1003 *arg++ = temp[0].hex;
1004 *arg++ = temp[0].mode;
1005 *arg++ = temp[1].name;
1006 *arg++ = temp[1].hex;
1007 *arg++ = temp[1].mode;
1017 retval = spawn_prog(pgm, spawn_arg);
1020 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1025 static void run_diff_cmd(const char *pgm,
1028 struct diff_filespec *one,
1029 struct diff_filespec *two,
1030 const char *xfrm_msg,
1031 struct diff_options *o,
1032 int complete_rewrite)
1035 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1040 builtin_diff(name, other ? other : name,
1041 one, two, xfrm_msg, o, complete_rewrite);
1043 printf("* Unmerged path %s\n", name);
1046 static void diff_fill_sha1_info(struct diff_filespec *one)
1048 if (DIFF_FILE_VALID(one)) {
1049 if (!one->sha1_valid) {
1051 if (lstat(one->path, &st) < 0)
1052 die("stat %s", one->path);
1053 if (index_path(one->sha1, one->path, &st, 0))
1054 die("cannot hash %s\n", one->path);
1058 memset(one->sha1, 0, 20);
1061 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1063 const char *pgm = external_diff();
1064 char msg[PATH_MAX*2+300], *xfrm_msg;
1065 struct diff_filespec *one;
1066 struct diff_filespec *two;
1069 char *name_munged, *other_munged;
1070 int complete_rewrite = 0;
1073 if (DIFF_PAIR_UNMERGED(p)) {
1075 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1079 name = p->one->path;
1080 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1081 name_munged = quote_one(name);
1082 other_munged = quote_one(other);
1083 one = p->one; two = p->two;
1085 diff_fill_sha1_info(one);
1086 diff_fill_sha1_info(two);
1089 switch (p->status) {
1090 case DIFF_STATUS_COPIED:
1091 len += snprintf(msg + len, sizeof(msg) - len,
1092 "similarity index %d%%\n"
1095 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1096 name_munged, other_munged);
1098 case DIFF_STATUS_RENAMED:
1099 len += snprintf(msg + len, sizeof(msg) - len,
1100 "similarity index %d%%\n"
1103 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1104 name_munged, other_munged);
1106 case DIFF_STATUS_MODIFIED:
1108 len += snprintf(msg + len, sizeof(msg) - len,
1109 "dissimilarity index %d%%\n",
1110 (int)(0.5 + p->score *
1112 complete_rewrite = 1;
1121 if (memcmp(one->sha1, two->sha1, 20)) {
1122 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1124 len += snprintf(msg + len, sizeof(msg) - len,
1126 abbrev, sha1_to_hex(one->sha1),
1127 abbrev, sha1_to_hex(two->sha1));
1128 if (one->mode == two->mode)
1129 len += snprintf(msg + len, sizeof(msg) - len,
1130 " %06o", one->mode);
1131 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1136 xfrm_msg = len ? msg : NULL;
1139 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1140 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1141 /* a filepair that changes between file and symlink
1142 * needs to be split into deletion and creation.
1144 struct diff_filespec *null = alloc_filespec(two->path);
1145 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1147 null = alloc_filespec(one->path);
1148 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1152 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1159 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1160 struct diffstat_t *diffstat)
1164 int complete_rewrite = 0;
1166 if (DIFF_PAIR_UNMERGED(p)) {
1168 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, 0);
1172 name = p->one->path;
1173 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1175 diff_fill_sha1_info(p->one);
1176 diff_fill_sha1_info(p->two);
1178 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1179 complete_rewrite = 1;
1180 builtin_diffstat(name, other, p->one, p->two, diffstat, complete_rewrite);
1183 void diff_setup(struct diff_options *options)
1185 memset(options, 0, sizeof(*options));
1186 options->output_format = DIFF_FORMAT_RAW;
1187 options->line_termination = '\n';
1188 options->break_opt = -1;
1189 options->rename_limit = -1;
1191 options->change = diff_change;
1192 options->add_remove = diff_addremove;
1195 int diff_setup_done(struct diff_options *options)
1197 if ((options->find_copies_harder &&
1198 options->detect_rename != DIFF_DETECT_COPY) ||
1199 (0 <= options->rename_limit && !options->detect_rename))
1203 * These cases always need recursive; we do not drop caller-supplied
1204 * recursive bits for other formats here.
1206 if ((options->output_format == DIFF_FORMAT_PATCH) ||
1207 (options->output_format == DIFF_FORMAT_DIFFSTAT))
1208 options->recursive = 1;
1210 if (options->detect_rename && options->rename_limit < 0)
1211 options->rename_limit = diff_rename_limit_default;
1212 if (options->setup & DIFF_SETUP_USE_CACHE) {
1214 /* read-cache does not die even when it fails
1215 * so it is safe for us to do this here. Also
1216 * it does not smudge active_cache or active_nr
1217 * when it fails, so we do not have to worry about
1218 * cleaning it up ourselves either.
1222 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1224 if (options->abbrev <= 0 || 40 < options->abbrev)
1225 options->abbrev = 40; /* full */
1230 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1232 const char *arg = av[0];
1233 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1234 options->output_format = DIFF_FORMAT_PATCH;
1235 else if (!strcmp(arg, "--patch-with-raw")) {
1236 options->output_format = DIFF_FORMAT_PATCH;
1237 options->with_raw = 1;
1239 else if (!strcmp(arg, "--stat"))
1240 options->output_format = DIFF_FORMAT_DIFFSTAT;
1241 else if (!strcmp(arg, "--patch-with-stat")) {
1242 options->output_format = DIFF_FORMAT_PATCH;
1243 options->with_stat = 1;
1245 else if (!strcmp(arg, "-z"))
1246 options->line_termination = 0;
1247 else if (!strncmp(arg, "-l", 2))
1248 options->rename_limit = strtoul(arg+2, NULL, 10);
1249 else if (!strcmp(arg, "--full-index"))
1250 options->full_index = 1;
1251 else if (!strcmp(arg, "--binary")) {
1252 options->output_format = DIFF_FORMAT_PATCH;
1253 options->full_index = options->binary = 1;
1255 else if (!strcmp(arg, "--name-only"))
1256 options->output_format = DIFF_FORMAT_NAME;
1257 else if (!strcmp(arg, "--name-status"))
1258 options->output_format = DIFF_FORMAT_NAME_STATUS;
1259 else if (!strcmp(arg, "-R"))
1260 options->reverse_diff = 1;
1261 else if (!strncmp(arg, "-S", 2))
1262 options->pickaxe = arg + 2;
1263 else if (!strcmp(arg, "-s"))
1264 options->output_format = DIFF_FORMAT_NO_OUTPUT;
1265 else if (!strncmp(arg, "-O", 2))
1266 options->orderfile = arg + 2;
1267 else if (!strncmp(arg, "--diff-filter=", 14))
1268 options->filter = arg + 14;
1269 else if (!strcmp(arg, "--pickaxe-all"))
1270 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1271 else if (!strcmp(arg, "--pickaxe-regex"))
1272 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1273 else if (!strncmp(arg, "-B", 2)) {
1274 if ((options->break_opt =
1275 diff_scoreopt_parse(arg)) == -1)
1278 else if (!strncmp(arg, "-M", 2)) {
1279 if ((options->rename_score =
1280 diff_scoreopt_parse(arg)) == -1)
1282 options->detect_rename = DIFF_DETECT_RENAME;
1284 else if (!strncmp(arg, "-C", 2)) {
1285 if ((options->rename_score =
1286 diff_scoreopt_parse(arg)) == -1)
1288 options->detect_rename = DIFF_DETECT_COPY;
1290 else if (!strcmp(arg, "--find-copies-harder"))
1291 options->find_copies_harder = 1;
1292 else if (!strcmp(arg, "--abbrev"))
1293 options->abbrev = DEFAULT_ABBREV;
1294 else if (!strncmp(arg, "--abbrev=", 9)) {
1295 options->abbrev = strtoul(arg + 9, NULL, 10);
1296 if (options->abbrev < MINIMUM_ABBREV)
1297 options->abbrev = MINIMUM_ABBREV;
1298 else if (40 < options->abbrev)
1299 options->abbrev = 40;
1306 static int parse_num(const char **cp_p)
1308 unsigned long num, scale;
1310 const char *cp = *cp_p;
1317 if ( !dot && ch == '.' ) {
1320 } else if ( ch == '%' ) {
1321 scale = dot ? scale*100 : 100;
1322 cp++; /* % is always at the end */
1324 } else if ( ch >= '0' && ch <= '9' ) {
1325 if ( scale < 100000 ) {
1327 num = (num*10) + (ch-'0');
1336 /* user says num divided by scale and we say internally that
1337 * is MAX_SCORE * num / scale.
1339 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1342 int diff_scoreopt_parse(const char *opt)
1344 int opt1, opt2, cmd;
1349 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1350 return -1; /* that is not a -M, -C nor -B option */
1352 opt1 = parse_num(&opt);
1358 else if (*opt != '/')
1359 return -1; /* we expect -B80/99 or -B80 */
1362 opt2 = parse_num(&opt);
1367 return opt1 | (opt2 << 16);
1370 struct diff_queue_struct diff_queued_diff;
1372 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1374 if (queue->alloc <= queue->nr) {
1375 queue->alloc = alloc_nr(queue->alloc);
1376 queue->queue = xrealloc(queue->queue,
1377 sizeof(dp) * queue->alloc);
1379 queue->queue[queue->nr++] = dp;
1382 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1383 struct diff_filespec *one,
1384 struct diff_filespec *two)
1386 struct diff_filepair *dp = xmalloc(sizeof(*dp));
1391 dp->source_stays = 0;
1392 dp->broken_pair = 0;
1398 void diff_free_filepair(struct diff_filepair *p)
1400 diff_free_filespec_data(p->one);
1401 diff_free_filespec_data(p->two);
1407 /* This is different from find_unique_abbrev() in that
1408 * it stuffs the result with dots for alignment.
1410 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1415 return sha1_to_hex(sha1);
1417 abbrev = find_unique_abbrev(sha1, len);
1419 return sha1_to_hex(sha1);
1420 abblen = strlen(abbrev);
1422 static char hex[41];
1423 if (len < abblen && abblen <= len + 2)
1424 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1426 sprintf(hex, "%s...", abbrev);
1429 return sha1_to_hex(sha1);
1432 static void diff_flush_raw(struct diff_filepair *p,
1433 int line_termination,
1434 int inter_name_termination,
1435 struct diff_options *options,
1440 int abbrev = options->abbrev;
1441 const char *path_one, *path_two;
1443 path_one = p->one->path;
1444 path_two = p->two->path;
1445 if (line_termination) {
1446 path_one = quote_one(path_one);
1447 path_two = quote_one(path_two);
1451 sprintf(status, "%c%03d", p->status,
1452 (int)(0.5 + p->score * 100.0/MAX_SCORE));
1454 status[0] = p->status;
1457 switch (p->status) {
1458 case DIFF_STATUS_COPIED:
1459 case DIFF_STATUS_RENAMED:
1462 case DIFF_STATUS_ADDED:
1463 case DIFF_STATUS_DELETED:
1470 if (output_format != DIFF_FORMAT_NAME_STATUS) {
1471 printf(":%06o %06o %s ",
1472 p->one->mode, p->two->mode,
1473 diff_unique_abbrev(p->one->sha1, abbrev));
1475 diff_unique_abbrev(p->two->sha1, abbrev));
1477 printf("%s%c%s", status, inter_name_termination, path_one);
1479 printf("%c%s", inter_name_termination, path_two);
1480 putchar(line_termination);
1481 if (path_one != p->one->path)
1482 free((void*)path_one);
1483 if (path_two != p->two->path)
1484 free((void*)path_two);
1487 static void diff_flush_name(struct diff_filepair *p,
1488 int inter_name_termination,
1489 int line_termination)
1491 char *path = p->two->path;
1493 if (line_termination)
1494 path = quote_one(p->two->path);
1496 path = p->two->path;
1497 printf("%s%c", path, line_termination);
1498 if (p->two->path != path)
1502 int diff_unmodified_pair(struct diff_filepair *p)
1504 /* This function is written stricter than necessary to support
1505 * the currently implemented transformers, but the idea is to
1506 * let transformers to produce diff_filepairs any way they want,
1507 * and filter and clean them up here before producing the output.
1509 struct diff_filespec *one, *two;
1511 if (DIFF_PAIR_UNMERGED(p))
1512 return 0; /* unmerged is interesting */
1517 /* deletion, addition, mode or type change
1518 * and rename are all interesting.
1520 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1521 DIFF_PAIR_MODE_CHANGED(p) ||
1522 strcmp(one->path, two->path))
1525 /* both are valid and point at the same path. that is, we are
1526 * dealing with a change.
1528 if (one->sha1_valid && two->sha1_valid &&
1529 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1530 return 1; /* no change */
1531 if (!one->sha1_valid && !two->sha1_valid)
1532 return 1; /* both look at the same file on the filesystem. */
1536 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1538 if (diff_unmodified_pair(p))
1541 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1542 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1543 return; /* no tree diffs in patch format */
1548 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1549 struct diffstat_t *diffstat)
1551 if (diff_unmodified_pair(p))
1554 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1555 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1556 return; /* no tree diffs in patch format */
1558 run_diffstat(p, o, diffstat);
1561 int diff_queue_is_empty(void)
1563 struct diff_queue_struct *q = &diff_queued_diff;
1565 for (i = 0; i < q->nr; i++)
1566 if (!diff_unmodified_pair(q->queue[i]))
1572 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1574 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1577 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1579 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1580 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1582 s->size, s->xfrm_flags);
1585 void diff_debug_filepair(const struct diff_filepair *p, int i)
1587 diff_debug_filespec(p->one, i, "one");
1588 diff_debug_filespec(p->two, i, "two");
1589 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1590 p->score, p->status ? p->status : '?',
1591 p->source_stays, p->broken_pair);
1594 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1598 fprintf(stderr, "%s\n", msg);
1599 fprintf(stderr, "q->nr = %d\n", q->nr);
1600 for (i = 0; i < q->nr; i++) {
1601 struct diff_filepair *p = q->queue[i];
1602 diff_debug_filepair(p, i);
1607 static void diff_resolve_rename_copy(void)
1610 struct diff_filepair *p, *pp;
1611 struct diff_queue_struct *q = &diff_queued_diff;
1613 diff_debug_queue("resolve-rename-copy", q);
1615 for (i = 0; i < q->nr; i++) {
1617 p->status = 0; /* undecided */
1618 if (DIFF_PAIR_UNMERGED(p))
1619 p->status = DIFF_STATUS_UNMERGED;
1620 else if (!DIFF_FILE_VALID(p->one))
1621 p->status = DIFF_STATUS_ADDED;
1622 else if (!DIFF_FILE_VALID(p->two))
1623 p->status = DIFF_STATUS_DELETED;
1624 else if (DIFF_PAIR_TYPE_CHANGED(p))
1625 p->status = DIFF_STATUS_TYPE_CHANGED;
1627 /* from this point on, we are dealing with a pair
1628 * whose both sides are valid and of the same type, i.e.
1629 * either in-place edit or rename/copy edit.
1631 else if (DIFF_PAIR_RENAME(p)) {
1632 if (p->source_stays) {
1633 p->status = DIFF_STATUS_COPIED;
1636 /* See if there is some other filepair that
1637 * copies from the same source as us. If so
1638 * we are a copy. Otherwise we are either a
1639 * copy if the path stays, or a rename if it
1640 * does not, but we already handled "stays" case.
1642 for (j = i + 1; j < q->nr; j++) {
1644 if (strcmp(pp->one->path, p->one->path))
1645 continue; /* not us */
1646 if (!DIFF_PAIR_RENAME(pp))
1647 continue; /* not a rename/copy */
1648 /* pp is a rename/copy from the same source */
1649 p->status = DIFF_STATUS_COPIED;
1653 p->status = DIFF_STATUS_RENAMED;
1655 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1656 p->one->mode != p->two->mode)
1657 p->status = DIFF_STATUS_MODIFIED;
1659 /* This is a "no-change" entry and should not
1660 * happen anymore, but prepare for broken callers.
1662 error("feeding unmodified %s to diffcore",
1664 p->status = DIFF_STATUS_UNKNOWN;
1667 diff_debug_queue("resolve-rename-copy done", q);
1670 static void flush_one_pair(struct diff_filepair *p,
1671 int diff_output_format,
1672 struct diff_options *options,
1673 struct diffstat_t *diffstat)
1675 int inter_name_termination = '\t';
1676 int line_termination = options->line_termination;
1677 if (!line_termination)
1678 inter_name_termination = 0;
1680 switch (p->status) {
1681 case DIFF_STATUS_UNKNOWN:
1684 die("internal error in diff-resolve-rename-copy");
1687 switch (diff_output_format) {
1688 case DIFF_FORMAT_DIFFSTAT:
1689 diff_flush_stat(p, options, diffstat);
1691 case DIFF_FORMAT_PATCH:
1692 diff_flush_patch(p, options);
1694 case DIFF_FORMAT_RAW:
1695 case DIFF_FORMAT_NAME_STATUS:
1696 diff_flush_raw(p, line_termination,
1697 inter_name_termination,
1698 options, diff_output_format);
1700 case DIFF_FORMAT_NAME:
1702 inter_name_termination,
1705 case DIFF_FORMAT_NO_OUTPUT:
1711 void diff_flush(struct diff_options *options)
1713 struct diff_queue_struct *q = &diff_queued_diff;
1715 int diff_output_format = options->output_format;
1716 struct diffstat_t *diffstat = NULL;
1718 if (diff_output_format == DIFF_FORMAT_DIFFSTAT || options->with_stat) {
1719 diffstat = xcalloc(sizeof (struct diffstat_t), 1);
1720 diffstat->xm.consume = diffstat_consume;
1723 if (options->with_raw) {
1724 for (i = 0; i < q->nr; i++) {
1725 struct diff_filepair *p = q->queue[i];
1726 flush_one_pair(p, DIFF_FORMAT_RAW, options, NULL);
1728 putchar(options->line_termination);
1730 if (options->with_stat) {
1731 for (i = 0; i < q->nr; i++) {
1732 struct diff_filepair *p = q->queue[i];
1733 flush_one_pair(p, DIFF_FORMAT_DIFFSTAT, options,
1736 show_stats(diffstat);
1739 putchar(options->line_termination);
1741 for (i = 0; i < q->nr; i++) {
1742 struct diff_filepair *p = q->queue[i];
1743 flush_one_pair(p, diff_output_format, options, diffstat);
1744 diff_free_filepair(p);
1748 show_stats(diffstat);
1754 q->nr = q->alloc = 0;
1757 static void diffcore_apply_filter(const char *filter)
1760 struct diff_queue_struct *q = &diff_queued_diff;
1761 struct diff_queue_struct outq;
1763 outq.nr = outq.alloc = 0;
1768 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1770 for (i = found = 0; !found && i < q->nr; i++) {
1771 struct diff_filepair *p = q->queue[i];
1772 if (((p->status == DIFF_STATUS_MODIFIED) &&
1774 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1776 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1777 ((p->status != DIFF_STATUS_MODIFIED) &&
1778 strchr(filter, p->status)))
1784 /* otherwise we will clear the whole queue
1785 * by copying the empty outq at the end of this
1786 * function, but first clear the current entries
1789 for (i = 0; i < q->nr; i++)
1790 diff_free_filepair(q->queue[i]);
1793 /* Only the matching ones */
1794 for (i = 0; i < q->nr; i++) {
1795 struct diff_filepair *p = q->queue[i];
1797 if (((p->status == DIFF_STATUS_MODIFIED) &&
1799 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1801 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1802 ((p->status != DIFF_STATUS_MODIFIED) &&
1803 strchr(filter, p->status)))
1806 diff_free_filepair(p);
1813 void diffcore_std(struct diff_options *options)
1815 if (options->break_opt != -1)
1816 diffcore_break(options->break_opt);
1817 if (options->detect_rename)
1818 diffcore_rename(options);
1819 if (options->break_opt != -1)
1820 diffcore_merge_broken();
1821 if (options->pickaxe)
1822 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1823 if (options->orderfile)
1824 diffcore_order(options->orderfile);
1825 diff_resolve_rename_copy();
1826 diffcore_apply_filter(options->filter);
1830 void diffcore_std_no_resolve(struct diff_options *options)
1832 if (options->pickaxe)
1833 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1834 if (options->orderfile)
1835 diffcore_order(options->orderfile);
1836 diffcore_apply_filter(options->filter);
1839 void diff_addremove(struct diff_options *options,
1840 int addremove, unsigned mode,
1841 const unsigned char *sha1,
1842 const char *base, const char *path)
1844 char concatpath[PATH_MAX];
1845 struct diff_filespec *one, *two;
1847 /* This may look odd, but it is a preparation for
1848 * feeding "there are unchanged files which should
1849 * not produce diffs, but when you are doing copy
1850 * detection you would need them, so here they are"
1851 * entries to the diff-core. They will be prefixed
1852 * with something like '=' or '*' (I haven't decided
1853 * which but should not make any difference).
1854 * Feeding the same new and old to diff_change()
1855 * also has the same effect.
1856 * Before the final output happens, they are pruned after
1857 * merged into rename/copy pairs as appropriate.
1859 if (options->reverse_diff)
1860 addremove = (addremove == '+' ? '-' :
1861 addremove == '-' ? '+' : addremove);
1863 if (!path) path = "";
1864 sprintf(concatpath, "%s%s", base, path);
1865 one = alloc_filespec(concatpath);
1866 two = alloc_filespec(concatpath);
1868 if (addremove != '+')
1869 fill_filespec(one, sha1, mode);
1870 if (addremove != '-')
1871 fill_filespec(two, sha1, mode);
1873 diff_queue(&diff_queued_diff, one, two);
1876 void diff_change(struct diff_options *options,
1877 unsigned old_mode, unsigned new_mode,
1878 const unsigned char *old_sha1,
1879 const unsigned char *new_sha1,
1880 const char *base, const char *path)
1882 char concatpath[PATH_MAX];
1883 struct diff_filespec *one, *two;
1885 if (options->reverse_diff) {
1887 const unsigned char *tmp_c;
1888 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1889 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1891 if (!path) path = "";
1892 sprintf(concatpath, "%s%s", base, path);
1893 one = alloc_filespec(concatpath);
1894 two = alloc_filespec(concatpath);
1895 fill_filespec(one, old_sha1, old_mode);
1896 fill_filespec(two, new_sha1, new_mode);
1898 diff_queue(&diff_queued_diff, one, two);
1901 void diff_unmerge(struct diff_options *options,
1904 struct diff_filespec *one, *two;
1905 one = alloc_filespec(path);
1906 two = alloc_filespec(path);
1907 diff_queue(&diff_queued_diff, one, two);