Use a hashtable for objects instead of a sorted list
[git.git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "quote.h"
9 #include "diff.h"
10 #include "diffcore.h"
11
12 static const char *diff_opts = "-pu";
13
14 static int use_size_cache;
15
16 int diff_rename_limit_default = -1;
17
18 int git_diff_config(const char *var, const char *value)
19 {
20         if (!strcmp(var, "diff.renamelimit")) {
21                 diff_rename_limit_default = git_config_int(var, value);
22                 return 0;
23         }
24
25         return git_default_config(var, value);
26 }
27
28 static char *quote_one(const char *str)
29 {
30         int needlen;
31         char *xp;
32
33         if (!str)
34                 return NULL;
35         needlen = quote_c_style(str, NULL, NULL, 0);
36         if (!needlen)
37                 return strdup(str);
38         xp = xmalloc(needlen + 1);
39         quote_c_style(str, xp, NULL, 0);
40         return xp;
41 }
42
43 static char *quote_two(const char *one, const char *two)
44 {
45         int need_one = quote_c_style(one, NULL, NULL, 1);
46         int need_two = quote_c_style(two, NULL, NULL, 1);
47         char *xp;
48
49         if (need_one + need_two) {
50                 if (!need_one) need_one = strlen(one);
51                 if (!need_two) need_one = strlen(two);
52
53                 xp = xmalloc(need_one + need_two + 3);
54                 xp[0] = '"';
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, "\"");
58                 return xp;
59         }
60         need_one = strlen(one);
61         need_two = strlen(two);
62         xp = xmalloc(need_one + need_two + 1);
63         strcpy(xp, one);
64         strcpy(xp + need_one, two);
65         return xp;
66 }
67
68 static const char *external_diff(void)
69 {
70         static const char *external_diff_cmd = NULL;
71         static int done_preparing = 0;
72         const char *env_diff_opts;
73
74         if (done_preparing)
75                 return external_diff_cmd;
76
77         /*
78          * Default values above are meant to match the
79          * Linux kernel development style.  Examples of
80          * alternative styles you can specify via environment
81          * variables are:
82          *
83          * GIT_DIFF_OPTS="-c";
84          */
85         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
86
87         /* In case external diff fails... */
88         env_diff_opts = getenv("GIT_DIFF_OPTS");
89         if (env_diff_opts) diff_opts = env_diff_opts;
90
91         done_preparing = 1;
92         return external_diff_cmd;
93 }
94
95 #define TEMPFILE_PATH_LEN               50
96
97 static struct diff_tempfile {
98         const char *name; /* filename external diff should read from */
99         char hex[41];
100         char mode[10];
101         char tmp_path[TEMPFILE_PATH_LEN];
102 } diff_temp[2];
103
104 static int count_lines(const char *filename)
105 {
106         FILE *in;
107         int count, ch, completely_empty = 1, nl_just_seen = 0;
108         in = fopen(filename, "r");
109         count = 0;
110         while ((ch = fgetc(in)) != EOF)
111                 if (ch == '\n') {
112                         count++;
113                         nl_just_seen = 1;
114                         completely_empty = 0;
115                 }
116                 else {
117                         nl_just_seen = 0;
118                         completely_empty = 0;
119                 }
120         fclose(in);
121         if (completely_empty)
122                 return 0;
123         if (!nl_just_seen)
124                 count++; /* no trailing newline */
125         return count;
126 }
127
128 static void print_line_count(int count)
129 {
130         switch (count) {
131         case 0:
132                 printf("0,0");
133                 break;
134         case 1:
135                 printf("1");
136                 break;
137         default:
138                 printf("1,%d", count);
139                 break;
140         }
141 }
142
143 static void copy_file(int prefix, const char *filename)
144 {
145         FILE *in;
146         int ch, nl_just_seen = 1;
147         in = fopen(filename, "r");
148         while ((ch = fgetc(in)) != EOF) {
149                 if (nl_just_seen)
150                         putchar(prefix);
151                 putchar(ch);
152                 if (ch == '\n')
153                         nl_just_seen = 1;
154                 else
155                         nl_just_seen = 0;
156         }
157         fclose(in);
158         if (!nl_just_seen)
159                 printf("\n\\ No newline at end of file\n");
160 }
161
162 static void emit_rewrite_diff(const char *name_a,
163                               const char *name_b,
164                               struct diff_tempfile *temp)
165 {
166         /* Use temp[i].name as input, name_a and name_b as labels */
167         int lc_a, lc_b;
168         lc_a = count_lines(temp[0].name);
169         lc_b = count_lines(temp[1].name);
170         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
171         print_line_count(lc_a);
172         printf(" +");
173         print_line_count(lc_b);
174         printf(" @@\n");
175         if (lc_a)
176                 copy_file('-', temp[0].name);
177         if (lc_b)
178                 copy_file('+', temp[1].name);
179 }
180
181 static void builtin_diff(const char *name_a,
182                          const char *name_b,
183                          struct diff_tempfile *temp,
184                          const char *xfrm_msg,
185                          int complete_rewrite)
186 {
187         int i, next_at, cmd_size;
188         const char *const diff_cmd = "diff -L%s -L%s";
189         const char *const diff_arg  = "-- %s %s||:"; /* "||:" is to return 0 */
190         const char *input_name_sq[2];
191         const char *label_path[2];
192         char *cmd;
193
194         /* diff_cmd and diff_arg have 4 %s in total which makes
195          * the sum of these strings 8 bytes larger than required.
196          * we use 2 spaces around diff-opts, and we need to count
197          * terminating NUL; we used to subtract 5 here, but we do not
198          * care about small leaks in this subprocess that is about
199          * to exec "diff" anymore.
200          */
201         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
202                     + 128);
203
204         for (i = 0; i < 2; i++) {
205                 input_name_sq[i] = sq_quote(temp[i].name);
206                 if (!strcmp(temp[i].name, "/dev/null"))
207                         label_path[i] = "/dev/null";
208                 else if (!i)
209                         label_path[i] = sq_quote(quote_two("a/", name_a));
210                 else
211                         label_path[i] = sq_quote(quote_two("b/", name_b));
212                 cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
213         }
214
215         cmd = xmalloc(cmd_size);
216
217         next_at = 0;
218         next_at += snprintf(cmd+next_at, cmd_size-next_at,
219                             diff_cmd, label_path[0], label_path[1]);
220         next_at += snprintf(cmd+next_at, cmd_size-next_at,
221                             " %s ", diff_opts);
222         next_at += snprintf(cmd+next_at, cmd_size-next_at,
223                             diff_arg, input_name_sq[0], input_name_sq[1]);
224
225         printf("diff --git %s %s\n",
226                quote_two("a/", name_a), quote_two("b/", name_b));
227         if (label_path[0][0] == '/') {
228                 /* dev/null */
229                 printf("new file mode %s\n", temp[1].mode);
230                 if (xfrm_msg && xfrm_msg[0])
231                         puts(xfrm_msg);
232         }
233         else if (label_path[1][0] == '/') {
234                 printf("deleted file mode %s\n", temp[0].mode);
235                 if (xfrm_msg && xfrm_msg[0])
236                         puts(xfrm_msg);
237         }
238         else {
239                 if (strcmp(temp[0].mode, temp[1].mode)) {
240                         printf("old mode %s\n", temp[0].mode);
241                         printf("new mode %s\n", temp[1].mode);
242                 }
243                 if (xfrm_msg && xfrm_msg[0])
244                         puts(xfrm_msg);
245                 if (strncmp(temp[0].mode, temp[1].mode, 3))
246                         /* we do not run diff between different kind
247                          * of objects.
248                          */
249                         exit(0);
250                 if (complete_rewrite) {
251                         fflush(NULL);
252                         emit_rewrite_diff(name_a, name_b, temp);
253                         exit(0);
254                 }
255         }
256         fflush(NULL);
257         execlp("/bin/sh","sh", "-c", cmd, NULL);
258 }
259
260 struct diff_filespec *alloc_filespec(const char *path)
261 {
262         int namelen = strlen(path);
263         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
264
265         memset(spec, 0, sizeof(*spec));
266         spec->path = (char *)(spec + 1);
267         memcpy(spec->path, path, namelen+1);
268         return spec;
269 }
270
271 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
272                    unsigned short mode)
273 {
274         if (mode) {
275                 spec->mode = DIFF_FILE_CANON_MODE(mode);
276                 memcpy(spec->sha1, sha1, 20);
277                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
278         }
279 }
280
281 /*
282  * Given a name and sha1 pair, if the dircache tells us the file in
283  * the work tree has that object contents, return true, so that
284  * prepare_temp_file() does not have to inflate and extract.
285  */
286 static int work_tree_matches(const char *name, const unsigned char *sha1)
287 {
288         struct cache_entry *ce;
289         struct stat st;
290         int pos, len;
291
292         /* We do not read the cache ourselves here, because the
293          * benchmark with my previous version that always reads cache
294          * shows that it makes things worse for diff-tree comparing
295          * two linux-2.6 kernel trees in an already checked out work
296          * tree.  This is because most diff-tree comparisons deal with
297          * only a small number of files, while reading the cache is
298          * expensive for a large project, and its cost outweighs the
299          * savings we get by not inflating the object to a temporary
300          * file.  Practically, this code only helps when we are used
301          * by diff-cache --cached, which does read the cache before
302          * calling us.
303          */
304         if (!active_cache)
305                 return 0;
306
307         len = strlen(name);
308         pos = cache_name_pos(name, len);
309         if (pos < 0)
310                 return 0;
311         ce = active_cache[pos];
312         if ((lstat(name, &st) < 0) ||
313             !S_ISREG(st.st_mode) || /* careful! */
314             ce_match_stat(ce, &st) ||
315             memcmp(sha1, ce->sha1, 20))
316                 return 0;
317         /* we return 1 only when we can stat, it is a regular file,
318          * stat information matches, and sha1 recorded in the cache
319          * matches.  I.e. we know the file in the work tree really is
320          * the same as the <name, sha1> pair.
321          */
322         return 1;
323 }
324
325 static struct sha1_size_cache {
326         unsigned char sha1[20];
327         unsigned long size;
328 } **sha1_size_cache;
329 static int sha1_size_cache_nr, sha1_size_cache_alloc;
330
331 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
332                                                  int find_only,
333                                                  unsigned long size)
334 {
335         int first, last;
336         struct sha1_size_cache *e;
337
338         first = 0;
339         last = sha1_size_cache_nr;
340         while (last > first) {
341                 int cmp, next = (last + first) >> 1;
342                 e = sha1_size_cache[next];
343                 cmp = memcmp(e->sha1, sha1, 20);
344                 if (!cmp)
345                         return e;
346                 if (cmp < 0) {
347                         last = next;
348                         continue;
349                 }
350                 first = next+1;
351         }
352         /* not found */
353         if (find_only)
354                 return NULL;
355         /* insert to make it at "first" */
356         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
357                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
358                 sha1_size_cache = xrealloc(sha1_size_cache,
359                                            sha1_size_cache_alloc *
360                                            sizeof(*sha1_size_cache));
361         }
362         sha1_size_cache_nr++;
363         if (first < sha1_size_cache_nr)
364                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
365                         (sha1_size_cache_nr - first - 1) *
366                         sizeof(*sha1_size_cache));
367         e = xmalloc(sizeof(struct sha1_size_cache));
368         sha1_size_cache[first] = e;
369         memcpy(e->sha1, sha1, 20);
370         e->size = size;
371         return e;
372 }
373
374 /*
375  * While doing rename detection and pickaxe operation, we may need to
376  * grab the data for the blob (or file) for our own in-core comparison.
377  * diff_filespec has data and size fields for this purpose.
378  */
379 int diff_populate_filespec(struct diff_filespec *s, int size_only)
380 {
381         int err = 0;
382         if (!DIFF_FILE_VALID(s))
383                 die("internal error: asking to populate invalid file.");
384         if (S_ISDIR(s->mode))
385                 return -1;
386
387         if (!use_size_cache)
388                 size_only = 0;
389
390         if (s->data)
391                 return err;
392         if (!s->sha1_valid ||
393             work_tree_matches(s->path, s->sha1)) {
394                 struct stat st;
395                 int fd;
396                 if (lstat(s->path, &st) < 0) {
397                         if (errno == ENOENT) {
398                         err_empty:
399                                 err = -1;
400                         empty:
401                                 s->data = "";
402                                 s->size = 0;
403                                 return err;
404                         }
405                 }
406                 s->size = st.st_size;
407                 if (!s->size)
408                         goto empty;
409                 if (size_only)
410                         return 0;
411                 if (S_ISLNK(st.st_mode)) {
412                         int ret;
413                         s->data = xmalloc(s->size);
414                         s->should_free = 1;
415                         ret = readlink(s->path, s->data, s->size);
416                         if (ret < 0) {
417                                 free(s->data);
418                                 goto err_empty;
419                         }
420                         return 0;
421                 }
422                 fd = open(s->path, O_RDONLY);
423                 if (fd < 0)
424                         goto err_empty;
425                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
426                 close(fd);
427                 if (s->data == MAP_FAILED)
428                         goto err_empty;
429                 s->should_munmap = 1;
430         }
431         else {
432                 char type[20];
433                 struct sha1_size_cache *e;
434
435                 if (size_only) {
436                         e = locate_size_cache(s->sha1, 1, 0);
437                         if (e) {
438                                 s->size = e->size;
439                                 return 0;
440                         }
441                         if (!sha1_object_info(s->sha1, type, &s->size))
442                                 locate_size_cache(s->sha1, 0, s->size);
443                 }
444                 else {
445                         s->data = read_sha1_file(s->sha1, type, &s->size);
446                         s->should_free = 1;
447                 }
448         }
449         return 0;
450 }
451
452 void diff_free_filespec_data(struct diff_filespec *s)
453 {
454         if (s->should_free)
455                 free(s->data);
456         else if (s->should_munmap)
457                 munmap(s->data, s->size);
458         s->should_free = s->should_munmap = 0;
459         s->data = NULL;
460 }
461
462 static void prep_temp_blob(struct diff_tempfile *temp,
463                            void *blob,
464                            unsigned long size,
465                            const unsigned char *sha1,
466                            int mode)
467 {
468         int fd;
469
470         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
471         if (fd < 0)
472                 die("unable to create temp-file");
473         if (write(fd, blob, size) != size)
474                 die("unable to write temp-file");
475         close(fd);
476         temp->name = temp->tmp_path;
477         strcpy(temp->hex, sha1_to_hex(sha1));
478         temp->hex[40] = 0;
479         sprintf(temp->mode, "%06o", mode);
480 }
481
482 static void prepare_temp_file(const char *name,
483                               struct diff_tempfile *temp,
484                               struct diff_filespec *one)
485 {
486         if (!DIFF_FILE_VALID(one)) {
487         not_a_valid_file:
488                 /* A '-' entry produces this for file-2, and
489                  * a '+' entry produces this for file-1.
490                  */
491                 temp->name = "/dev/null";
492                 strcpy(temp->hex, ".");
493                 strcpy(temp->mode, ".");
494                 return;
495         }
496
497         if (!one->sha1_valid ||
498             work_tree_matches(name, one->sha1)) {
499                 struct stat st;
500                 if (lstat(name, &st) < 0) {
501                         if (errno == ENOENT)
502                                 goto not_a_valid_file;
503                         die("stat(%s): %s", name, strerror(errno));
504                 }
505                 if (S_ISLNK(st.st_mode)) {
506                         int ret;
507                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
508                         if (sizeof(buf) <= st.st_size)
509                                 die("symlink too long: %s", name);
510                         ret = readlink(name, buf, st.st_size);
511                         if (ret < 0)
512                                 die("readlink(%s)", name);
513                         prep_temp_blob(temp, buf, st.st_size,
514                                        (one->sha1_valid ?
515                                         one->sha1 : null_sha1),
516                                        (one->sha1_valid ?
517                                         one->mode : S_IFLNK));
518                 }
519                 else {
520                         /* we can borrow from the file in the work tree */
521                         temp->name = name;
522                         if (!one->sha1_valid)
523                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
524                         else
525                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
526                         /* Even though we may sometimes borrow the
527                          * contents from the work tree, we always want
528                          * one->mode.  mode is trustworthy even when
529                          * !(one->sha1_valid), as long as
530                          * DIFF_FILE_VALID(one).
531                          */
532                         sprintf(temp->mode, "%06o", one->mode);
533                 }
534                 return;
535         }
536         else {
537                 if (diff_populate_filespec(one, 0))
538                         die("cannot read data blob for %s", one->path);
539                 prep_temp_blob(temp, one->data, one->size,
540                                one->sha1, one->mode);
541         }
542 }
543
544 static void remove_tempfile(void)
545 {
546         int i;
547
548         for (i = 0; i < 2; i++)
549                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
550                         unlink(diff_temp[i].name);
551                         diff_temp[i].name = NULL;
552                 }
553 }
554
555 static void remove_tempfile_on_signal(int signo)
556 {
557         remove_tempfile();
558         signal(SIGINT, SIG_DFL);
559         raise(signo);
560 }
561
562 /* An external diff command takes:
563  *
564  * diff-cmd name infile1 infile1-sha1 infile1-mode \
565  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
566  *
567  */
568 static void run_external_diff(const char *pgm,
569                               const char *name,
570                               const char *other,
571                               struct diff_filespec *one,
572                               struct diff_filespec *two,
573                               const char *xfrm_msg,
574                               int complete_rewrite)
575 {
576         struct diff_tempfile *temp = diff_temp;
577         pid_t pid;
578         int status;
579         static int atexit_asked = 0;
580         const char *othername;
581
582         othername = (other? other : name);
583         if (one && two) {
584                 prepare_temp_file(name, &temp[0], one);
585                 prepare_temp_file(othername, &temp[1], two);
586                 if (! atexit_asked &&
587                     (temp[0].name == temp[0].tmp_path ||
588                      temp[1].name == temp[1].tmp_path)) {
589                         atexit_asked = 1;
590                         atexit(remove_tempfile);
591                 }
592                 signal(SIGINT, remove_tempfile_on_signal);
593         }
594
595         fflush(NULL);
596         pid = fork();
597         if (pid < 0)
598                 die("unable to fork");
599         if (!pid) {
600                 if (pgm) {
601                         if (one && two) {
602                                 const char *exec_arg[10];
603                                 const char **arg = &exec_arg[0];
604                                 *arg++ = pgm;
605                                 *arg++ = name;
606                                 *arg++ = temp[0].name;
607                                 *arg++ = temp[0].hex;
608                                 *arg++ = temp[0].mode;
609                                 *arg++ = temp[1].name;
610                                 *arg++ = temp[1].hex;
611                                 *arg++ = temp[1].mode;
612                                 if (other) {
613                                         *arg++ = other;
614                                         *arg++ = xfrm_msg;
615                                 }
616                                 *arg = NULL;
617                                 execvp(pgm, (char *const*) exec_arg);
618                         }
619                         else
620                                 execlp(pgm, pgm, name, NULL);
621                 }
622                 /*
623                  * otherwise we use the built-in one.
624                  */
625                 if (one && two)
626                         builtin_diff(name, othername, temp, xfrm_msg,
627                                      complete_rewrite);
628                 else
629                         printf("* Unmerged path %s\n", name);
630                 exit(0);
631         }
632         if (waitpid(pid, &status, 0) < 0 ||
633             !WIFEXITED(status) || WEXITSTATUS(status)) {
634                 /* Earlier we did not check the exit status because
635                  * diff exits non-zero if files are different, and
636                  * we are not interested in knowing that.  It was a
637                  * mistake which made it harder to quit a diff-*
638                  * session that uses the git-apply-patch-script as
639                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
640                  * should also exit non-zero only when it wants to
641                  * abort the entire diff-* session.
642                  */
643                 remove_tempfile();
644                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
645                 exit(1);
646         }
647         remove_tempfile();
648 }
649
650 static void diff_fill_sha1_info(struct diff_filespec *one)
651 {
652         if (DIFF_FILE_VALID(one)) {
653                 if (!one->sha1_valid) {
654                         struct stat st;
655                         if (lstat(one->path, &st) < 0)
656                                 die("stat %s", one->path);
657                         if (index_path(one->sha1, one->path, &st, 0))
658                                 die("cannot hash %s\n", one->path);
659                 }
660         }
661         else
662                 memset(one->sha1, 0, 20);
663 }
664
665 static void run_diff(struct diff_filepair *p, struct diff_options *o)
666 {
667         const char *pgm = external_diff();
668         char msg[PATH_MAX*2+300], *xfrm_msg;
669         struct diff_filespec *one;
670         struct diff_filespec *two;
671         const char *name;
672         const char *other;
673         char *name_munged, *other_munged;
674         int complete_rewrite = 0;
675         int len;
676
677         if (DIFF_PAIR_UNMERGED(p)) {
678                 /* unmerged */
679                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
680                                   0);
681                 return;
682         }
683
684         name = p->one->path;
685         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
686         name_munged = quote_one(name);
687         other_munged = quote_one(other);
688         one = p->one; two = p->two;
689
690         diff_fill_sha1_info(one);
691         diff_fill_sha1_info(two);
692
693         len = 0;
694         switch (p->status) {
695         case DIFF_STATUS_COPIED:
696                 len += snprintf(msg + len, sizeof(msg) - len,
697                                 "similarity index %d%%\n"
698                                 "copy from %s\n"
699                                 "copy to %s\n",
700                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
701                                 name_munged, other_munged);
702                 break;
703         case DIFF_STATUS_RENAMED:
704                 len += snprintf(msg + len, sizeof(msg) - len,
705                                 "similarity index %d%%\n"
706                                 "rename from %s\n"
707                                 "rename to %s\n",
708                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
709                                 name_munged, other_munged);
710                 break;
711         case DIFF_STATUS_MODIFIED:
712                 if (p->score) {
713                         len += snprintf(msg + len, sizeof(msg) - len,
714                                         "dissimilarity index %d%%\n",
715                                         (int)(0.5 + p->score *
716                                               100.0/MAX_SCORE));
717                         complete_rewrite = 1;
718                         break;
719                 }
720                 /* fallthru */
721         default:
722                 /* nothing */
723                 ;
724         }
725
726         if (memcmp(one->sha1, two->sha1, 20)) {
727                 char one_sha1[41];
728                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
729                 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
730
731                 len += snprintf(msg + len, sizeof(msg) - len,
732                                 "index %.*s..%.*s",
733                                 abbrev, one_sha1, abbrev,
734                                 sha1_to_hex(two->sha1));
735                 if (one->mode == two->mode)
736                         len += snprintf(msg + len, sizeof(msg) - len,
737                                         " %06o", one->mode);
738                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
739         }
740
741         if (len)
742                 msg[--len] = 0;
743         xfrm_msg = len ? msg : NULL;
744
745         if (!pgm &&
746             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
747             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
748                 /* a filepair that changes between file and symlink
749                  * needs to be split into deletion and creation.
750                  */
751                 struct diff_filespec *null = alloc_filespec(two->path);
752                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
753                 free(null);
754                 null = alloc_filespec(one->path);
755                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
756                 free(null);
757         }
758         else
759                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
760                                   complete_rewrite);
761
762         free(name_munged);
763         free(other_munged);
764 }
765
766 void diff_setup(struct diff_options *options)
767 {
768         memset(options, 0, sizeof(*options));
769         options->output_format = DIFF_FORMAT_RAW;
770         options->line_termination = '\n';
771         options->break_opt = -1;
772         options->rename_limit = -1;
773
774         options->change = diff_change;
775         options->add_remove = diff_addremove;
776 }
777
778 int diff_setup_done(struct diff_options *options)
779 {
780         if ((options->find_copies_harder &&
781              options->detect_rename != DIFF_DETECT_COPY) ||
782             (0 <= options->rename_limit && !options->detect_rename))
783                 return -1;
784         if (options->detect_rename && options->rename_limit < 0)
785                 options->rename_limit = diff_rename_limit_default;
786         if (options->setup & DIFF_SETUP_USE_CACHE) {
787                 if (!active_cache)
788                         /* read-cache does not die even when it fails
789                          * so it is safe for us to do this here.  Also
790                          * it does not smudge active_cache or active_nr
791                          * when it fails, so we do not have to worry about
792                          * cleaning it up ourselves either.
793                          */
794                         read_cache();
795         }
796         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
797                 use_size_cache = 1;
798         if (options->abbrev <= 0 || 40 < options->abbrev)
799                 options->abbrev = 40; /* full */
800
801         return 0;
802 }
803
804 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
805 {
806         const char *arg = av[0];
807         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
808                 options->output_format = DIFF_FORMAT_PATCH;
809         else if (!strcmp(arg, "-z"))
810                 options->line_termination = 0;
811         else if (!strncmp(arg, "-l", 2))
812                 options->rename_limit = strtoul(arg+2, NULL, 10);
813         else if (!strcmp(arg, "--full-index"))
814                 options->full_index = 1;
815         else if (!strcmp(arg, "--name-only"))
816                 options->output_format = DIFF_FORMAT_NAME;
817         else if (!strcmp(arg, "--name-status"))
818                 options->output_format = DIFF_FORMAT_NAME_STATUS;
819         else if (!strcmp(arg, "-R"))
820                 options->reverse_diff = 1;
821         else if (!strncmp(arg, "-S", 2))
822                 options->pickaxe = arg + 2;
823         else if (!strcmp(arg, "-s"))
824                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
825         else if (!strncmp(arg, "-O", 2))
826                 options->orderfile = arg + 2;
827         else if (!strncmp(arg, "--diff-filter=", 14))
828                 options->filter = arg + 14;
829         else if (!strcmp(arg, "--pickaxe-all"))
830                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
831         else if (!strncmp(arg, "-B", 2)) {
832                 if ((options->break_opt =
833                      diff_scoreopt_parse(arg)) == -1)
834                         return -1;
835         }
836         else if (!strncmp(arg, "-M", 2)) {
837                 if ((options->rename_score =
838                      diff_scoreopt_parse(arg)) == -1)
839                         return -1;
840                 options->detect_rename = DIFF_DETECT_RENAME;
841         }
842         else if (!strncmp(arg, "-C", 2)) {
843                 if ((options->rename_score =
844                      diff_scoreopt_parse(arg)) == -1)
845                         return -1;
846                 options->detect_rename = DIFF_DETECT_COPY;
847         }
848         else if (!strcmp(arg, "--find-copies-harder"))
849                 options->find_copies_harder = 1;
850         else if (!strcmp(arg, "--abbrev"))
851                 options->abbrev = DEFAULT_ABBREV;
852         else if (!strncmp(arg, "--abbrev=", 9)) {
853                 options->abbrev = strtoul(arg + 9, NULL, 10);
854                 if (options->abbrev < MINIMUM_ABBREV)
855                         options->abbrev = MINIMUM_ABBREV;
856                 else if (40 < options->abbrev)
857                         options->abbrev = 40;
858         }
859         else
860                 return 0;
861         return 1;
862 }
863
864 static int parse_num(const char **cp_p)
865 {
866         unsigned long num, scale;
867         int ch, dot;
868         const char *cp = *cp_p;
869
870         num = 0;
871         scale = 1;
872         dot = 0;
873         for(;;) {
874                 ch = *cp;
875                 if ( !dot && ch == '.' ) {
876                         scale = 1;
877                         dot = 1;
878                 } else if ( ch == '%' ) {
879                         scale = dot ? scale*100 : 100;
880                         cp++;   /* % is always at the end */
881                         break;
882                 } else if ( ch >= '0' && ch <= '9' ) {
883                         if ( scale < 100000 ) {
884                                 scale *= 10;
885                                 num = (num*10) + (ch-'0');
886                         }
887                 } else {
888                         break;
889                 }
890                 cp++;
891         }
892         *cp_p = cp;
893
894         /* user says num divided by scale and we say internally that
895          * is MAX_SCORE * num / scale.
896          */
897         return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
898 }
899
900 int diff_scoreopt_parse(const char *opt)
901 {
902         int opt1, opt2, cmd;
903
904         if (*opt++ != '-')
905                 return -1;
906         cmd = *opt++;
907         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
908                 return -1; /* that is not a -M, -C nor -B option */
909
910         opt1 = parse_num(&opt);
911         if (cmd != 'B')
912                 opt2 = 0;
913         else {
914                 if (*opt == 0)
915                         opt2 = 0;
916                 else if (*opt != '/')
917                         return -1; /* we expect -B80/99 or -B80 */
918                 else {
919                         opt++;
920                         opt2 = parse_num(&opt);
921                 }
922         }
923         if (*opt != 0)
924                 return -1;
925         return opt1 | (opt2 << 16);
926 }
927
928 struct diff_queue_struct diff_queued_diff;
929
930 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
931 {
932         if (queue->alloc <= queue->nr) {
933                 queue->alloc = alloc_nr(queue->alloc);
934                 queue->queue = xrealloc(queue->queue,
935                                         sizeof(dp) * queue->alloc);
936         }
937         queue->queue[queue->nr++] = dp;
938 }
939
940 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
941                                  struct diff_filespec *one,
942                                  struct diff_filespec *two)
943 {
944         struct diff_filepair *dp = xmalloc(sizeof(*dp));
945         dp->one = one;
946         dp->two = two;
947         dp->score = 0;
948         dp->status = 0;
949         dp->source_stays = 0;
950         dp->broken_pair = 0;
951         if (queue)
952                 diff_q(queue, dp);
953         return dp;
954 }
955
956 void diff_free_filepair(struct diff_filepair *p)
957 {
958         diff_free_filespec_data(p->one);
959         diff_free_filespec_data(p->two);
960         free(p->one);
961         free(p->two);
962         free(p);
963 }
964
965 /* This is different from find_unique_abbrev() in that
966  * it stuffs the result with dots for alignment.
967  */
968 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
969 {
970         int abblen;
971         const char *abbrev;
972         if (len == 40)
973                 return sha1_to_hex(sha1);
974
975         abbrev = find_unique_abbrev(sha1, len);
976         if (!abbrev)
977                 return sha1_to_hex(sha1);
978         abblen = strlen(abbrev);
979         if (abblen < 37) {
980                 static char hex[41];
981                 if (len < abblen && abblen <= len + 2)
982                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
983                 else
984                         sprintf(hex, "%s...", abbrev);
985                 return hex;
986         }
987         return sha1_to_hex(sha1);
988 }
989
990 static void diff_flush_raw(struct diff_filepair *p,
991                            int line_termination,
992                            int inter_name_termination,
993                            struct diff_options *options)
994 {
995         int two_paths;
996         char status[10];
997         int abbrev = options->abbrev;
998         const char *path_one, *path_two;
999         int output_format = options->output_format;
1000
1001         path_one = p->one->path;
1002         path_two = p->two->path;
1003         if (line_termination) {
1004                 path_one = quote_one(path_one);
1005                 path_two = quote_one(path_two);
1006         }
1007
1008         if (p->score)
1009                 sprintf(status, "%c%03d", p->status,
1010                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
1011         else {
1012                 status[0] = p->status;
1013                 status[1] = 0;
1014         }
1015         switch (p->status) {
1016         case DIFF_STATUS_COPIED:
1017         case DIFF_STATUS_RENAMED:
1018                 two_paths = 1;
1019                 break;
1020         case DIFF_STATUS_ADDED:
1021         case DIFF_STATUS_DELETED:
1022                 two_paths = 0;
1023                 break;
1024         default:
1025                 two_paths = 0;
1026                 break;
1027         }
1028         if (output_format != DIFF_FORMAT_NAME_STATUS) {
1029                 printf(":%06o %06o %s ",
1030                        p->one->mode, p->two->mode,
1031                        diff_unique_abbrev(p->one->sha1, abbrev));
1032                 printf("%s ",
1033                        diff_unique_abbrev(p->two->sha1, abbrev));
1034         }
1035         printf("%s%c%s", status, inter_name_termination, path_one);
1036         if (two_paths)
1037                 printf("%c%s", inter_name_termination, path_two);
1038         putchar(line_termination);
1039         if (path_one != p->one->path)
1040                 free((void*)path_one);
1041         if (path_two != p->two->path)
1042                 free((void*)path_two);
1043 }
1044
1045 static void diff_flush_name(struct diff_filepair *p,
1046                             int inter_name_termination,
1047                             int line_termination)
1048 {
1049         char *path = p->two->path;
1050
1051         if (line_termination)
1052                 path = quote_one(p->two->path);
1053         else
1054                 path = p->two->path;
1055         printf("%s%c", path, line_termination);
1056         if (p->two->path != path)
1057                 free(path);
1058 }
1059
1060 int diff_unmodified_pair(struct diff_filepair *p)
1061 {
1062         /* This function is written stricter than necessary to support
1063          * the currently implemented transformers, but the idea is to
1064          * let transformers to produce diff_filepairs any way they want,
1065          * and filter and clean them up here before producing the output.
1066          */
1067         struct diff_filespec *one, *two;
1068
1069         if (DIFF_PAIR_UNMERGED(p))
1070                 return 0; /* unmerged is interesting */
1071
1072         one = p->one;
1073         two = p->two;
1074
1075         /* deletion, addition, mode or type change
1076          * and rename are all interesting.
1077          */
1078         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1079             DIFF_PAIR_MODE_CHANGED(p) ||
1080             strcmp(one->path, two->path))
1081                 return 0;
1082
1083         /* both are valid and point at the same path.  that is, we are
1084          * dealing with a change.
1085          */
1086         if (one->sha1_valid && two->sha1_valid &&
1087             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1088                 return 1; /* no change */
1089         if (!one->sha1_valid && !two->sha1_valid)
1090                 return 1; /* both look at the same file on the filesystem. */
1091         return 0;
1092 }
1093
1094 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1095 {
1096         if (diff_unmodified_pair(p))
1097                 return;
1098
1099         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1100             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1101                 return; /* no tree diffs in patch format */ 
1102
1103         run_diff(p, o);
1104 }
1105
1106 int diff_queue_is_empty(void)
1107 {
1108         struct diff_queue_struct *q = &diff_queued_diff;
1109         int i;
1110         for (i = 0; i < q->nr; i++)
1111                 if (!diff_unmodified_pair(q->queue[i]))
1112                         return 0;
1113         return 1;
1114 }
1115
1116 #if DIFF_DEBUG
1117 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1118 {
1119         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1120                 x, one ? one : "",
1121                 s->path,
1122                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1123                 s->mode,
1124                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1125         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1126                 x, one ? one : "",
1127                 s->size, s->xfrm_flags);
1128 }
1129
1130 void diff_debug_filepair(const struct diff_filepair *p, int i)
1131 {
1132         diff_debug_filespec(p->one, i, "one");
1133         diff_debug_filespec(p->two, i, "two");
1134         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1135                 p->score, p->status ? p->status : '?',
1136                 p->source_stays, p->broken_pair);
1137 }
1138
1139 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1140 {
1141         int i;
1142         if (msg)
1143                 fprintf(stderr, "%s\n", msg);
1144         fprintf(stderr, "q->nr = %d\n", q->nr);
1145         for (i = 0; i < q->nr; i++) {
1146                 struct diff_filepair *p = q->queue[i];
1147                 diff_debug_filepair(p, i);
1148         }
1149 }
1150 #endif
1151
1152 static void diff_resolve_rename_copy(void)
1153 {
1154         int i, j;
1155         struct diff_filepair *p, *pp;
1156         struct diff_queue_struct *q = &diff_queued_diff;
1157
1158         diff_debug_queue("resolve-rename-copy", q);
1159
1160         for (i = 0; i < q->nr; i++) {
1161                 p = q->queue[i];
1162                 p->status = 0; /* undecided */
1163                 if (DIFF_PAIR_UNMERGED(p))
1164                         p->status = DIFF_STATUS_UNMERGED;
1165                 else if (!DIFF_FILE_VALID(p->one))
1166                         p->status = DIFF_STATUS_ADDED;
1167                 else if (!DIFF_FILE_VALID(p->two))
1168                         p->status = DIFF_STATUS_DELETED;
1169                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1170                         p->status = DIFF_STATUS_TYPE_CHANGED;
1171
1172                 /* from this point on, we are dealing with a pair
1173                  * whose both sides are valid and of the same type, i.e.
1174                  * either in-place edit or rename/copy edit.
1175                  */
1176                 else if (DIFF_PAIR_RENAME(p)) {
1177                         if (p->source_stays) {
1178                                 p->status = DIFF_STATUS_COPIED;
1179                                 continue;
1180                         }
1181                         /* See if there is some other filepair that
1182                          * copies from the same source as us.  If so
1183                          * we are a copy.  Otherwise we are either a
1184                          * copy if the path stays, or a rename if it
1185                          * does not, but we already handled "stays" case.
1186                          */
1187                         for (j = i + 1; j < q->nr; j++) {
1188                                 pp = q->queue[j];
1189                                 if (strcmp(pp->one->path, p->one->path))
1190                                         continue; /* not us */
1191                                 if (!DIFF_PAIR_RENAME(pp))
1192                                         continue; /* not a rename/copy */
1193                                 /* pp is a rename/copy from the same source */
1194                                 p->status = DIFF_STATUS_COPIED;
1195                                 break;
1196                         }
1197                         if (!p->status)
1198                                 p->status = DIFF_STATUS_RENAMED;
1199                 }
1200                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1201                          p->one->mode != p->two->mode)
1202                         p->status = DIFF_STATUS_MODIFIED;
1203                 else {
1204                         /* This is a "no-change" entry and should not
1205                          * happen anymore, but prepare for broken callers.
1206                          */
1207                         error("feeding unmodified %s to diffcore",
1208                               p->one->path);
1209                         p->status = DIFF_STATUS_UNKNOWN;
1210                 }
1211         }
1212         diff_debug_queue("resolve-rename-copy done", q);
1213 }
1214
1215 void diff_flush(struct diff_options *options)
1216 {
1217         struct diff_queue_struct *q = &diff_queued_diff;
1218         int i;
1219         int inter_name_termination = '\t';
1220         int diff_output_format = options->output_format;
1221         int line_termination = options->line_termination;
1222
1223         if (!line_termination)
1224                 inter_name_termination = 0;
1225
1226         for (i = 0; i < q->nr; i++) {
1227                 struct diff_filepair *p = q->queue[i];
1228                 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1229                     (p->status == DIFF_STATUS_UNKNOWN))
1230                         continue;
1231                 if (p->status == 0)
1232                         die("internal error in diff-resolve-rename-copy");
1233                 switch (diff_output_format) {
1234                 case DIFF_FORMAT_PATCH:
1235                         diff_flush_patch(p, options);
1236                         break;
1237                 case DIFF_FORMAT_RAW:
1238                 case DIFF_FORMAT_NAME_STATUS:
1239                         diff_flush_raw(p, line_termination,
1240                                        inter_name_termination,
1241                                        options);
1242                         break;
1243                 case DIFF_FORMAT_NAME:
1244                         diff_flush_name(p,
1245                                         inter_name_termination,
1246                                         line_termination);
1247                         break;
1248                 }
1249                 diff_free_filepair(q->queue[i]);
1250         }
1251         free(q->queue);
1252         q->queue = NULL;
1253         q->nr = q->alloc = 0;
1254 }
1255
1256 static void diffcore_apply_filter(const char *filter)
1257 {
1258         int i;
1259         struct diff_queue_struct *q = &diff_queued_diff;
1260         struct diff_queue_struct outq;
1261         outq.queue = NULL;
1262         outq.nr = outq.alloc = 0;
1263
1264         if (!filter)
1265                 return;
1266
1267         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1268                 int found;
1269                 for (i = found = 0; !found && i < q->nr; i++) {
1270                         struct diff_filepair *p = q->queue[i];
1271                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1272                              ((p->score &&
1273                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1274                               (!p->score &&
1275                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1276                             ((p->status != DIFF_STATUS_MODIFIED) &&
1277                              strchr(filter, p->status)))
1278                                 found++;
1279                 }
1280                 if (found)
1281                         return;
1282
1283                 /* otherwise we will clear the whole queue
1284                  * by copying the empty outq at the end of this
1285                  * function, but first clear the current entries
1286                  * in the queue.
1287                  */
1288                 for (i = 0; i < q->nr; i++)
1289                         diff_free_filepair(q->queue[i]);
1290         }
1291         else {
1292                 /* Only the matching ones */
1293                 for (i = 0; i < q->nr; i++) {
1294                         struct diff_filepair *p = q->queue[i];
1295
1296                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1297                              ((p->score &&
1298                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1299                               (!p->score &&
1300                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1301                             ((p->status != DIFF_STATUS_MODIFIED) &&
1302                              strchr(filter, p->status)))
1303                                 diff_q(&outq, p);
1304                         else
1305                                 diff_free_filepair(p);
1306                 }
1307         }
1308         free(q->queue);
1309         *q = outq;
1310 }
1311
1312 void diffcore_std(struct diff_options *options)
1313 {
1314         if (options->paths && options->paths[0])
1315                 diffcore_pathspec(options->paths);
1316         if (options->break_opt != -1)
1317                 diffcore_break(options->break_opt);
1318         if (options->detect_rename)
1319                 diffcore_rename(options);
1320         if (options->break_opt != -1)
1321                 diffcore_merge_broken();
1322         if (options->pickaxe)
1323                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1324         if (options->orderfile)
1325                 diffcore_order(options->orderfile);
1326         diff_resolve_rename_copy();
1327         diffcore_apply_filter(options->filter);
1328 }
1329
1330
1331 void diffcore_std_no_resolve(struct diff_options *options)
1332 {
1333         if (options->pickaxe)
1334                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1335         if (options->orderfile)
1336                 diffcore_order(options->orderfile);
1337         diffcore_apply_filter(options->filter);
1338 }
1339
1340 void diff_addremove(struct diff_options *options,
1341                     int addremove, unsigned mode,
1342                     const unsigned char *sha1,
1343                     const char *base, const char *path)
1344 {
1345         char concatpath[PATH_MAX];
1346         struct diff_filespec *one, *two;
1347
1348         /* This may look odd, but it is a preparation for
1349          * feeding "there are unchanged files which should
1350          * not produce diffs, but when you are doing copy
1351          * detection you would need them, so here they are"
1352          * entries to the diff-core.  They will be prefixed
1353          * with something like '=' or '*' (I haven't decided
1354          * which but should not make any difference).
1355          * Feeding the same new and old to diff_change() 
1356          * also has the same effect.
1357          * Before the final output happens, they are pruned after
1358          * merged into rename/copy pairs as appropriate.
1359          */
1360         if (options->reverse_diff)
1361                 addremove = (addremove == '+' ? '-' :
1362                              addremove == '-' ? '+' : addremove);
1363
1364         if (!path) path = "";
1365         sprintf(concatpath, "%s%s", base, path);
1366         one = alloc_filespec(concatpath);
1367         two = alloc_filespec(concatpath);
1368
1369         if (addremove != '+')
1370                 fill_filespec(one, sha1, mode);
1371         if (addremove != '-')
1372                 fill_filespec(two, sha1, mode);
1373
1374         diff_queue(&diff_queued_diff, one, two);
1375 }
1376
1377 void diff_change(struct diff_options *options,
1378                  unsigned old_mode, unsigned new_mode,
1379                  const unsigned char *old_sha1,
1380                  const unsigned char *new_sha1,
1381                  const char *base, const char *path) 
1382 {
1383         char concatpath[PATH_MAX];
1384         struct diff_filespec *one, *two;
1385
1386         if (options->reverse_diff) {
1387                 unsigned tmp;
1388                 const unsigned char *tmp_c;
1389                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1390                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1391         }
1392         if (!path) path = "";
1393         sprintf(concatpath, "%s%s", base, path);
1394         one = alloc_filespec(concatpath);
1395         two = alloc_filespec(concatpath);
1396         fill_filespec(one, old_sha1, old_mode);
1397         fill_filespec(two, new_sha1, new_mode);
1398
1399         diff_queue(&diff_queued_diff, one, two);
1400 }
1401
1402 void diff_unmerge(struct diff_options *options,
1403                   const char *path)
1404 {
1405         struct diff_filespec *one, *two;
1406         one = alloc_filespec(path);
1407         two = alloc_filespec(path);
1408         diff_queue(&diff_queued_diff, one, two);
1409 }