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 static char *quote_one(const char *str)
  17 {
  18         int needlen;
  19         char *xp;
  20
  21         if (!str)
  22                 return NULL;
  23         needlen = quote_c_style(str, NULL, NULL, 0);
  24         if (!needlen)
  25                 return strdup(str);
  26         xp = xmalloc(needlen + 1);
  27         quote_c_style(str, xp, NULL, 0);
  28         return xp;
  29 }
  30
  31 static char *quote_two(const char *one, const char *two)
  32 {
  33         int need_one = quote_c_style(one, NULL, NULL, 1);
  34         int need_two = quote_c_style(two, NULL, NULL, 1);
  35         char *xp;
  36
  37         if (need_one + need_two) {
  38                 if (!need_one) need_one = strlen(one);
  39                 if (!need_two) need_one = strlen(two);
  40
  41                 xp = xmalloc(need_one + need_two + 3);
  42                 xp[0] = '"';
  43                 quote_c_style(one, xp + 1, NULL, 1);
  44                 quote_c_style(two, xp + need_one + 1, NULL, 1);
  45                 strcpy(xp + need_one + need_two + 1, "\"");
  46                 return xp;
  47         }
  48         need_one = strlen(one);
  49         need_two = strlen(two);
  50         xp = xmalloc(need_one + need_two + 1);
  51         strcpy(xp, one);
  52         strcpy(xp + need_one, two);
  53         return xp;
  54 }
  55
  56 static const char *external_diff(void)
  57 {
  58         static const char *external_diff_cmd = NULL;
  59         static int done_preparing = 0;
  60         const char *env_diff_opts;
  61
  62         if (done_preparing)
  63                 return external_diff_cmd;
  64
  65         /*
  66          * Default values above are meant to match the
  67          * Linux kernel development style.  Examples of
  68          * alternative styles you can specify via environment
  69          * variables are:
  70          *
  71          * GIT_DIFF_OPTS="-c";
  72          */
  73         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
  74
  75         /* In case external diff fails... */
  76         env_diff_opts = getenv("GIT_DIFF_OPTS");
  77         if (env_diff_opts) diff_opts = env_diff_opts;
  78
  79         done_preparing = 1;
  80         return external_diff_cmd;
  81 }
  82
  83 #define TEMPFILE_PATH_LEN               50
  84
  85 static struct diff_tempfile {
  86         const char *name; /* filename external diff should read from */
  87         char hex[41];
  88         char mode[10];
  89         char tmp_path[TEMPFILE_PATH_LEN];
  90 } diff_temp[2];
  91
  92 static int count_lines(const char *filename)
  93 {
  94         FILE *in;
  95         int count, ch, completely_empty = 1, nl_just_seen = 0;
  96         in = fopen(filename, "r");
  97         count = 0;
  98         while ((ch = fgetc(in)) != EOF)
  99                 if (ch == '\n') {
 100                         count++;
 101                         nl_just_seen = 1;
 102                         completely_empty = 0;
 103                 }
 104                 else {
 105                         nl_just_seen = 0;
 106                         completely_empty = 0;
 107                 }
 108         fclose(in);
 109         if (completely_empty)
 110                 return 0;
 111         if (!nl_just_seen)
 112                 count++; /* no trailing newline */
 113         return count;
 114 }
 115
 116 static void print_line_count(int count)
 117 {
 118         switch (count) {
 119         case 0:
 120                 printf("0,0");
 121                 break;
 122         case 1:
 123                 printf("1");
 124                 break;
 125         default:
 126                 printf("1,%d", count);
 127                 break;
 128         }
 129 }
 130
 131 static void copy_file(int prefix, const char *filename)
 132 {
 133         FILE *in;
 134         int ch, nl_just_seen = 1;
 135         in = fopen(filename, "r");
 136         while ((ch = fgetc(in)) != EOF) {
 137                 if (nl_just_seen)
 138                         putchar(prefix);
 139                 putchar(ch);
 140                 if (ch == '\n')
 141                         nl_just_seen = 1;
 142                 else
 143                         nl_just_seen = 0;
 144         }
 145         fclose(in);
 146         if (!nl_just_seen)
 147                 printf("\n\\ No newline at end of file\n");
 148 }
 149
 150 static void emit_rewrite_diff(const char *name_a,
 151                               const char *name_b,
 152                               struct diff_tempfile *temp)
 153 {
 154         /* Use temp[i].name as input, name_a and name_b as labels */
 155         int lc_a, lc_b;
 156         lc_a = count_lines(temp[0].name);
 157         lc_b = count_lines(temp[1].name);
 158         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
 159         print_line_count(lc_a);
 160         printf(" +");
 161         print_line_count(lc_b);
 162         printf(" @@\n");
 163         if (lc_a)
 164                 copy_file('-', temp[0].name);
 165         if (lc_b)
 166                 copy_file('+', temp[1].name);
 167 }
 168
 169 static void builtin_diff(const char *name_a,
 170                          const char *name_b,
 171                          struct diff_tempfile *temp,
 172                          const char *xfrm_msg,
 173                          int complete_rewrite)
 174 {
 175         int i, next_at, cmd_size;
 176         const char *const diff_cmd = "diff -L%s -L%s";
 177         const char *const diff_arg  = "-- %s %s||:"; /* "||:" is to return 0 */
 178         const char *input_name_sq[2];
 179         const char *label_path[2];
 180         char *cmd;
 181
 182         /* diff_cmd and diff_arg have 4 %s in total which makes
 183          * the sum of these strings 8 bytes larger than required.
 184          * we use 2 spaces around diff-opts, and we need to count
 185          * terminating NUL; we used to subtract 5 here, but we do not
 186          * care about small leaks in this subprocess that is about
 187          * to exec "diff" anymore.
 188          */
 189         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
 190                     + 128);
 191
 192         for (i = 0; i < 2; i++) {
 193                 input_name_sq[i] = sq_quote(temp[i].name);
 194                 if (!strcmp(temp[i].name, "/dev/null"))
 195                         label_path[i] = "/dev/null";
 196                 else if (!i)
 197                         label_path[i] = sq_quote(quote_two("a/", name_a));
 198                 else
 199                         label_path[i] = sq_quote(quote_two("b/", name_b));
 200                 cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
 201         }
 202
 203         cmd = xmalloc(cmd_size);
 204
 205         next_at = 0;
 206         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 207                             diff_cmd, label_path[0], label_path[1]);
 208         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 209                             " %s ", diff_opts);
 210         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 211                             diff_arg, input_name_sq[0], input_name_sq[1]);
 212
 213         printf("diff --git %s %s\n",
 214                quote_two("a/", name_a), quote_two("b/", name_b));
 215         if (label_path[0][0] == '/') {
 216                 /* dev/null */
 217                 printf("new file mode %s\n", temp[1].mode);
 218                 if (xfrm_msg && xfrm_msg[0])
 219                         puts(xfrm_msg);
 220         }
 221         else if (label_path[1][0] == '/') {
 222                 printf("deleted file mode %s\n", temp[0].mode);
 223                 if (xfrm_msg && xfrm_msg[0])
 224                         puts(xfrm_msg);
 225         }
 226         else {
 227                 if (strcmp(temp[0].mode, temp[1].mode)) {
 228                         printf("old mode %s\n", temp[0].mode);
 229                         printf("new mode %s\n", temp[1].mode);
 230                 }
 231                 if (xfrm_msg && xfrm_msg[0])
 232                         puts(xfrm_msg);
 233                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 234                         /* we do not run diff between different kind
 235                          * of objects.
 236                          */
 237                         exit(0);
 238                 if (complete_rewrite) {
 239                         fflush(NULL);
 240                         emit_rewrite_diff(name_a, name_b, temp);
 241                         exit(0);
 242                 }
 243         }
 244         fflush(NULL);
 245         execlp("/bin/sh","sh", "-c", cmd, NULL);
 246 }
 247
 248 struct diff_filespec *alloc_filespec(const char *path)
 249 {
 250         int namelen = strlen(path);
 251         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 252
 253         memset(spec, 0, sizeof(*spec));
 254         spec->path = (char *)(spec + 1);
 255         memcpy(spec->path, path, namelen+1);
 256         return spec;
 257 }
 258
 259 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 260                    unsigned short mode)
 261 {
 262         if (mode) {
 263                 spec->mode = DIFF_FILE_CANON_MODE(mode);
 264                 memcpy(spec->sha1, sha1, 20);
 265                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 266         }
 267 }
 268
 269 /*
 270  * Given a name and sha1 pair, if the dircache tells us the file in
 271  * the work tree has that object contents, return true, so that
 272  * prepare_temp_file() does not have to inflate and extract.
 273  */
 274 static int work_tree_matches(const char *name, const unsigned char *sha1)
 275 {
 276         struct cache_entry *ce;
 277         struct stat st;
 278         int pos, len;
 279
 280         /* We do not read the cache ourselves here, because the
 281          * benchmark with my previous version that always reads cache
 282          * shows that it makes things worse for diff-tree comparing
 283          * two linux-2.6 kernel trees in an already checked out work
 284          * tree.  This is because most diff-tree comparisons deal with
 285          * only a small number of files, while reading the cache is
 286          * expensive for a large project, and its cost outweighs the
 287          * savings we get by not inflating the object to a temporary
 288          * file.  Practically, this code only helps when we are used
 289          * by diff-cache --cached, which does read the cache before
 290          * calling us.
 291          */
 292         if (!active_cache)
 293                 return 0;
 294
 295         len = strlen(name);
 296         pos = cache_name_pos(name, len);
 297         if (pos < 0)
 298                 return 0;
 299         ce = active_cache[pos];
 300         if ((lstat(name, &st) < 0) ||
 301             !S_ISREG(st.st_mode) || /* careful! */
 302             ce_match_stat(ce, &st) ||
 303             memcmp(sha1, ce->sha1, 20))
 304                 return 0;
 305         /* we return 1 only when we can stat, it is a regular file,
 306          * stat information matches, and sha1 recorded in the cache
 307          * matches.  I.e. we know the file in the work tree really is
 308          * the same as the <name, sha1> pair.
 309          */
 310         return 1;
 311 }
 312
 313 static struct sha1_size_cache {
 314         unsigned char sha1[20];
 315         unsigned long size;
 316 } **sha1_size_cache;
 317 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 318
 319 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 320                                                  int find_only,
 321                                                  unsigned long size)
 322 {
 323         int first, last;
 324         struct sha1_size_cache *e;
 325
 326         first = 0;
 327         last = sha1_size_cache_nr;
 328         while (last > first) {
 329                 int cmp, next = (last + first) >> 1;
 330                 e = sha1_size_cache[next];
 331                 cmp = memcmp(e->sha1, sha1, 20);
 332                 if (!cmp)
 333                         return e;
 334                 if (cmp < 0) {
 335                         last = next;
 336                         continue;
 337                 }
 338                 first = next+1;
 339         }
 340         /* not found */
 341         if (find_only)
 342                 return NULL;
 343         /* insert to make it at "first" */
 344         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 345                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 346                 sha1_size_cache = xrealloc(sha1_size_cache,
 347                                            sha1_size_cache_alloc *
 348                                            sizeof(*sha1_size_cache));
 349         }
 350         sha1_size_cache_nr++;
 351         if (first < sha1_size_cache_nr)
 352                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 353                         (sha1_size_cache_nr - first - 1) *
 354                         sizeof(*sha1_size_cache));
 355         e = xmalloc(sizeof(struct sha1_size_cache));
 356         sha1_size_cache[first] = e;
 357         memcpy(e->sha1, sha1, 20);
 358         e->size = size;
 359         return e;
 360 }
 361
 362 /*
 363  * While doing rename detection and pickaxe operation, we may need to
 364  * grab the data for the blob (or file) for our own in-core comparison.
 365  * diff_filespec has data and size fields for this purpose.
 366  */
 367 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 368 {
 369         int err = 0;
 370         if (!DIFF_FILE_VALID(s))
 371                 die("internal error: asking to populate invalid file.");
 372         if (S_ISDIR(s->mode))
 373                 return -1;
 374
 375         if (!use_size_cache)
 376                 size_only = 0;
 377
 378         if (s->data)
 379                 return err;
 380         if (!s->sha1_valid ||
 381             work_tree_matches(s->path, s->sha1)) {
 382                 struct stat st;
 383                 int fd;
 384                 if (lstat(s->path, &st) < 0) {
 385                         if (errno == ENOENT) {
 386                         err_empty:
 387                                 err = -1;
 388                         empty:
 389                                 s->data = "";
 390                                 s->size = 0;
 391                                 return err;
 392                         }
 393                 }
 394                 s->size = st.st_size;
 395                 if (!s->size)
 396                         goto empty;
 397                 if (size_only)
 398                         return 0;
 399                 if (S_ISLNK(st.st_mode)) {
 400                         int ret;
 401                         s->data = xmalloc(s->size);
 402                         s->should_free = 1;
 403                         ret = readlink(s->path, s->data, s->size);
 404                         if (ret < 0) {
 405                                 free(s->data);
 406                                 goto err_empty;
 407                         }
 408                         return 0;
 409                 }
 410                 fd = open(s->path, O_RDONLY);
 411                 if (fd < 0)
 412                         goto err_empty;
 413                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 414                 close(fd);
 415                 if (s->data == MAP_FAILED)
 416                         goto err_empty;
 417                 s->should_munmap = 1;
 418         }
 419         else {
 420                 char type[20];
 421                 struct sha1_size_cache *e;
 422
 423                 if (size_only) {
 424                         e = locate_size_cache(s->sha1, 1, 0);
 425                         if (e) {
 426                                 s->size = e->size;
 427                                 return 0;
 428                         }
 429                         if (!sha1_object_info(s->sha1, type, &s->size))
 430                                 locate_size_cache(s->sha1, 0, s->size);
 431                 }
 432                 else {
 433                         s->data = read_sha1_file(s->sha1, type, &s->size);
 434                         s->should_free = 1;
 435                 }
 436         }
 437         return 0;
 438 }
 439
 440 void diff_free_filespec_data(struct diff_filespec *s)
 441 {
 442         if (s->should_free)
 443                 free(s->data);
 444         else if (s->should_munmap)
 445                 munmap(s->data, s->size);
 446         s->should_free = s->should_munmap = 0;
 447         s->data = NULL;
 448 }
 449
 450 static void prep_temp_blob(struct diff_tempfile *temp,
 451                            void *blob,
 452                            unsigned long size,
 453                            const unsigned char *sha1,
 454                            int mode)
 455 {
 456         int fd;
 457
 458         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
 459         if (fd < 0)
 460                 die("unable to create temp-file");
 461         if (write(fd, blob, size) != size)
 462                 die("unable to write temp-file");
 463         close(fd);
 464         temp->name = temp->tmp_path;
 465         strcpy(temp->hex, sha1_to_hex(sha1));
 466         temp->hex[40] = 0;
 467         sprintf(temp->mode, "%06o", mode);
 468 }
 469
 470 static void prepare_temp_file(const char *name,
 471                               struct diff_tempfile *temp,
 472                               struct diff_filespec *one)
 473 {
 474         if (!DIFF_FILE_VALID(one)) {
 475         not_a_valid_file:
 476                 /* A '-' entry produces this for file-2, and
 477                  * a '+' entry produces this for file-1.
 478                  */
 479                 temp->name = "/dev/null";
 480                 strcpy(temp->hex, ".");
 481                 strcpy(temp->mode, ".");
 482                 return;
 483         }
 484
 485         if (!one->sha1_valid ||
 486             work_tree_matches(name, one->sha1)) {
 487                 struct stat st;
 488                 if (lstat(name, &st) < 0) {
 489                         if (errno == ENOENT)
 490                                 goto not_a_valid_file;
 491                         die("stat(%s): %s", name, strerror(errno));
 492                 }
 493                 if (S_ISLNK(st.st_mode)) {
 494                         int ret;
 495                         char *buf, buf_[1024];
 496                         buf = ((sizeof(buf_) < st.st_size) ?
 497                                xmalloc(st.st_size) : buf_);
 498                         ret = readlink(name, buf, st.st_size);
 499                         if (ret < 0)
 500                                 die("readlink(%s)", name);
 501                         prep_temp_blob(temp, buf, st.st_size,
 502                                        (one->sha1_valid ?
 503                                         one->sha1 : null_sha1),
 504                                        (one->sha1_valid ?
 505                                         one->mode : S_IFLNK));
 506                 }
 507                 else {
 508                         /* we can borrow from the file in the work tree */
 509                         temp->name = name;
 510                         if (!one->sha1_valid)
 511                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 512                         else
 513                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 514                         /* Even though we may sometimes borrow the
 515                          * contents from the work tree, we always want
 516                          * one->mode.  mode is trustworthy even when
 517                          * !(one->sha1_valid), as long as
 518                          * DIFF_FILE_VALID(one).
 519                          */
 520                         sprintf(temp->mode, "%06o", one->mode);
 521                 }
 522                 return;
 523         }
 524         else {
 525                 if (diff_populate_filespec(one, 0))
 526                         die("cannot read data blob for %s", one->path);
 527                 prep_temp_blob(temp, one->data, one->size,
 528                                one->sha1, one->mode);
 529         }
 530 }
 531
 532 static void remove_tempfile(void)
 533 {
 534         int i;
 535
 536         for (i = 0; i < 2; i++)
 537                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 538                         unlink(diff_temp[i].name);
 539                         diff_temp[i].name = NULL;
 540                 }
 541 }
 542
 543 static void remove_tempfile_on_signal(int signo)
 544 {
 545         remove_tempfile();
 546 }
 547
 548 /* An external diff command takes:
 549  *
 550  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 551  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 552  *
 553  */
 554 static void run_external_diff(const char *pgm,
 555                               const char *name,
 556                               const char *other,
 557                               struct diff_filespec *one,
 558                               struct diff_filespec *two,
 559                               const char *xfrm_msg,
 560                               int complete_rewrite)
 561 {
 562         struct diff_tempfile *temp = diff_temp;
 563         pid_t pid;
 564         int status;
 565         static int atexit_asked = 0;
 566         const char *othername;
 567
 568         othername = (other? other : name);
 569         if (one && two) {
 570                 prepare_temp_file(name, &temp[0], one);
 571                 prepare_temp_file(othername, &temp[1], two);
 572                 if (! atexit_asked &&
 573                     (temp[0].name == temp[0].tmp_path ||
 574                      temp[1].name == temp[1].tmp_path)) {
 575                         atexit_asked = 1;
 576                         atexit(remove_tempfile);
 577                 }
 578                 signal(SIGINT, remove_tempfile_on_signal);
 579         }
 580
 581         fflush(NULL);
 582         pid = fork();
 583         if (pid < 0)
 584                 die("unable to fork");
 585         if (!pid) {
 586                 if (pgm) {
 587                         if (one && two) {
 588                                 const char *exec_arg[10];
 589                                 const char **arg = &exec_arg[0];
 590                                 *arg++ = pgm;
 591                                 *arg++ = name;
 592                                 *arg++ = temp[0].name;
 593                                 *arg++ = temp[0].hex;
 594                                 *arg++ = temp[0].mode;
 595                                 *arg++ = temp[1].name;
 596                                 *arg++ = temp[1].hex;
 597                                 *arg++ = temp[1].mode;
 598                                 if (other) {
 599                                         *arg++ = other;
 600                                         *arg++ = xfrm_msg;
 601                                 }
 602                                 *arg = NULL;
 603                                 execvp(pgm, (char *const*) exec_arg);
 604                         }
 605                         else
 606                                 execlp(pgm, pgm, name, NULL);
 607                 }
 608                 /*
 609                  * otherwise we use the built-in one.
 610                  */
 611                 if (one && two)
 612                         builtin_diff(name, othername, temp, xfrm_msg,
 613                                      complete_rewrite);
 614                 else
 615                         printf("* Unmerged path %s\n", name);
 616                 exit(0);
 617         }
 618         if (waitpid(pid, &status, 0) < 0 ||
 619             !WIFEXITED(status) || WEXITSTATUS(status)) {
 620                 /* Earlier we did not check the exit status because
 621                  * diff exits non-zero if files are different, and
 622                  * we are not interested in knowing that.  It was a
 623                  * mistake which made it harder to quit a diff-*
 624                  * session that uses the git-apply-patch-script as
 625                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 626                  * should also exit non-zero only when it wants to
 627                  * abort the entire diff-* session.
 628                  */
 629                 remove_tempfile();
 630                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 631                 exit(1);
 632         }
 633         remove_tempfile();
 634 }
 635
 636 static void diff_fill_sha1_info(struct diff_filespec *one)
 637 {
 638         if (DIFF_FILE_VALID(one)) {
 639                 if (!one->sha1_valid) {
 640                         struct stat st;
 641                         if (stat(one->path, &st) < 0)
 642                                 die("stat %s", one->path);
 643                         if (index_path(one->sha1, one->path, &st, 0))
 644                                 die("cannot hash %s\n", one->path);
 645                 }
 646         }
 647         else
 648                 memset(one->sha1, 0, 20);
 649 }
 650
 651 static void run_diff(struct diff_filepair *p)
 652 {
 653         const char *pgm = external_diff();
 654         char msg[PATH_MAX*2+300], *xfrm_msg;
 655         struct diff_filespec *one;
 656         struct diff_filespec *two;
 657         const char *name;
 658         const char *other;
 659         char *name_munged, *other_munged;
 660         int complete_rewrite = 0;
 661         int len;
 662
 663         if (DIFF_PAIR_UNMERGED(p)) {
 664                 /* unmerged */
 665                 run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
 666                                   0);
 667                 return;
 668         }
 669
 670         name = p->one->path;
 671         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 672         name_munged = quote_one(name);
 673         other_munged = quote_one(other);
 674         one = p->one; two = p->two;
 675
 676         diff_fill_sha1_info(one);
 677         diff_fill_sha1_info(two);
 678
 679         len = 0;
 680         switch (p->status) {
 681         case DIFF_STATUS_COPIED:
 682                 len += snprintf(msg + len, sizeof(msg) - len,
 683                                 "similarity index %d%%\n"
 684                                 "copy from %s\n"
 685                                 "copy to %s\n",
 686                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 687                                 name_munged, other_munged);
 688                 break;
 689         case DIFF_STATUS_RENAMED:
 690                 len += snprintf(msg + len, sizeof(msg) - len,
 691                                 "similarity index %d%%\n"
 692                                 "rename from %s\n"
 693                                 "rename to %s\n",
 694                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
 695                                 name_munged, other_munged);
 696                 break;
 697         case DIFF_STATUS_MODIFIED:
 698                 if (p->score) {
 699                         len += snprintf(msg + len, sizeof(msg) - len,
 700                                         "dissimilarity index %d%%\n",
 701                                         (int)(0.5 + p->score *
 702                                               100.0/MAX_SCORE));
 703                         complete_rewrite = 1;
 704                         break;
 705                 }
 706                 /* fallthru */
 707         default:
 708                 /* nothing */
 709                 ;
 710         }
 711
 712         if (memcmp(one->sha1, two->sha1, 20)) {
 713                 char one_sha1[41];
 714                 memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
 715
 716                 len += snprintf(msg + len, sizeof(msg) - len,
 717                                 "index %.7s..%.7s", one_sha1,
 718                                 sha1_to_hex(two->sha1));
 719                 if (one->mode == two->mode)
 720                         len += snprintf(msg + len, sizeof(msg) - len,
 721                                         " %06o", one->mode);
 722                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
 723         }
 724
 725         if (len)
 726                 msg[--len] = 0;
 727         xfrm_msg = len ? msg : NULL;
 728
 729         if (!pgm &&
 730             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 731             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 732                 /* a filepair that changes between file and symlink
 733                  * needs to be split into deletion and creation.
 734                  */
 735                 struct diff_filespec *null = alloc_filespec(two->path);
 736                 run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
 737                 free(null);
 738                 null = alloc_filespec(one->path);
 739                 run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
 740                 free(null);
 741         }
 742         else
 743                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
 744                                   complete_rewrite);
 745
 746         free(name_munged);
 747         free(other_munged);
 748 }
 749
 750 void diff_setup(struct diff_options *options)
 751 {
 752         memset(options, 0, sizeof(*options));
 753         options->output_format = DIFF_FORMAT_RAW;
 754         options->line_termination = '\n';
 755         options->break_opt = -1;
 756         options->rename_limit = -1;
 757 }
 758
 759 int diff_setup_done(struct diff_options *options)
 760 {
 761         if ((options->find_copies_harder || 0 <= options->rename_limit) &&
 762             options->detect_rename != DIFF_DETECT_COPY)
 763                 return -1;
 764         if (options->setup & DIFF_SETUP_USE_CACHE) {
 765                 if (!active_cache)
 766                         /* read-cache does not die even when it fails
 767                          * so it is safe for us to do this here.  Also
 768                          * it does not smudge active_cache or active_nr
 769                          * when it fails, so we do not have to worry about
 770                          * cleaning it up oufselves either.
 771                          */
 772                         read_cache();
 773         }
 774         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
 775                 use_size_cache = 1;
 776
 777         return 0;
 778 }
 779
 780 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 781 {
 782         const char *arg = av[0];
 783         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
 784                 options->output_format = DIFF_FORMAT_PATCH;
 785         else if (!strcmp(arg, "-z"))
 786                 options->line_termination = 0;
 787         else if (!strncmp(arg, "-l", 2))
 788                 options->rename_limit = strtoul(arg+2, NULL, 10);
 789         else if (!strcmp(arg, "--name-only"))
 790                 options->output_format = DIFF_FORMAT_NAME;
 791         else if (!strcmp(arg, "--name-status"))
 792                 options->output_format = DIFF_FORMAT_NAME_STATUS;
 793         else if (!strcmp(arg, "-R"))
 794                 options->reverse_diff = 1;
 795         else if (!strncmp(arg, "-S", 2))
 796                 options->pickaxe = arg + 2;
 797         else if (!strcmp(arg, "-s"))
 798                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
 799         else if (!strncmp(arg, "-O", 2))
 800                 options->orderfile = arg + 2;
 801         else if (!strncmp(arg, "--diff-filter=", 14))
 802                 options->filter = arg + 14;
 803         else if (!strcmp(arg, "--pickaxe-all"))
 804                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
 805         else if (!strncmp(arg, "-B", 2)) {
 806                 if ((options->break_opt =
 807                      diff_scoreopt_parse(arg)) == -1)
 808                         return -1;
 809         }
 810         else if (!strncmp(arg, "-M", 2)) {
 811                 if ((options->rename_score =
 812                      diff_scoreopt_parse(arg)) == -1)
 813                         return -1;
 814                 options->detect_rename = DIFF_DETECT_RENAME;
 815         }
 816         else if (!strncmp(arg, "-C", 2)) {
 817                 if ((options->rename_score =
 818                      diff_scoreopt_parse(arg)) == -1)
 819                         return -1;
 820                 options->detect_rename = DIFF_DETECT_COPY;
 821         }
 822         else if (!strcmp(arg, "--find-copies-harder"))
 823                 options->find_copies_harder = 1;
 824         else
 825                 return 0;
 826         return 1;
 827 }
 828
 829 static int parse_num(const char **cp_p)
 830 {
 831         int num, scale, ch, cnt;
 832         const char *cp = *cp_p;
 833
 834         cnt = num = 0;
 835         scale = 1;
 836         while ('0' <= (ch = *cp) && ch <= '9') {
 837                 if (cnt++ < 5) {
 838                         /* We simply ignore more than 5 digits precision. */
 839                         scale *= 10;
 840                         num = num * 10 + ch - '0';
 841                 }
 842                 cp++;
 843         }
 844         *cp_p = cp;
 845
 846         /* user says num divided by scale and we say internally that
 847          * is MAX_SCORE * num / scale.
 848          */
 849         return (MAX_SCORE * num / scale);
 850 }
 851
 852 int diff_scoreopt_parse(const char *opt)
 853 {
 854         int opt1, opt2, cmd;
 855
 856         if (*opt++ != '-')
 857                 return -1;
 858         cmd = *opt++;
 859         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 860                 return -1; /* that is not a -M, -C nor -B option */
 861
 862         opt1 = parse_num(&opt);
 863         if (cmd != 'B')
 864                 opt2 = 0;
 865         else {
 866                 if (*opt == 0)
 867                         opt2 = 0;
 868                 else if (*opt != '/')
 869                         return -1; /* we expect -B80/99 or -B80 */
 870                 else {
 871                         opt++;
 872                         opt2 = parse_num(&opt);
 873                 }
 874         }
 875         if (*opt != 0)
 876                 return -1;
 877         return opt1 | (opt2 << 16);
 878 }
 879
 880 struct diff_queue_struct diff_queued_diff;
 881
 882 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 883 {
 884         if (queue->alloc <= queue->nr) {
 885                 queue->alloc = alloc_nr(queue->alloc);
 886                 queue->queue = xrealloc(queue->queue,
 887                                         sizeof(dp) * queue->alloc);
 888         }
 889         queue->queue[queue->nr++] = dp;
 890 }
 891
 892 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 893                                  struct diff_filespec *one,
 894                                  struct diff_filespec *two)
 895 {
 896         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 897         dp->one = one;
 898         dp->two = two;
 899         dp->score = 0;
 900         dp->status = 0;
 901         dp->source_stays = 0;
 902         dp->broken_pair = 0;
 903         if (queue)
 904                 diff_q(queue, dp);
 905         return dp;
 906 }
 907
 908 void diff_free_filepair(struct diff_filepair *p)
 909 {
 910         diff_free_filespec_data(p->one);
 911         diff_free_filespec_data(p->two);
 912         free(p->one);
 913         free(p->two);
 914         free(p);
 915 }
 916
 917 static void diff_flush_raw(struct diff_filepair *p,
 918                            int line_termination,
 919                            int inter_name_termination,
 920                            int output_format)
 921 {
 922         int two_paths;
 923         char status[10];
 924         const char *path_one, *path_two;
 925
 926         path_one = p->one->path;
 927         path_two = p->two->path;
 928         if (line_termination) {
 929                 path_one = quote_one(path_one);
 930                 path_two = quote_one(path_two);
 931         }
 932
 933         if (p->score)
 934                 sprintf(status, "%c%03d", p->status,
 935                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 936         else {
 937                 status[0] = p->status;
 938                 status[1] = 0;
 939         }
 940         switch (p->status) {
 941         case DIFF_STATUS_COPIED:
 942         case DIFF_STATUS_RENAMED:
 943                 two_paths = 1;
 944                 break;
 945         case DIFF_STATUS_ADDED:
 946         case DIFF_STATUS_DELETED:
 947                 two_paths = 0;
 948                 break;
 949         default:
 950                 two_paths = 0;
 951                 break;
 952         }
 953         if (output_format != DIFF_FORMAT_NAME_STATUS) {
 954                 printf(":%06o %06o %s ",
 955                        p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 956                 printf("%s ", sha1_to_hex(p->two->sha1));
 957         }
 958         printf("%s%c%s", status, inter_name_termination, path_one);
 959         if (two_paths)
 960                 printf("%c%s", inter_name_termination, path_two);
 961         putchar(line_termination);
 962         if (path_one != p->one->path)
 963                 free((void*)path_one);
 964         if (path_two != p->two->path)
 965                 free((void*)path_two);
 966 }
 967
 968 static void diff_flush_name(struct diff_filepair *p,
 969                             int inter_name_termination,
 970                             int line_termination)
 971 {
 972         char *path = p->two->path;
 973
 974         if (line_termination)
 975                 path = quote_one(p->two->path);
 976         else
 977                 path = p->two->path;
 978         printf("%s%c", path, line_termination);
 979         if (p->two->path != path)
 980                 free(path);
 981 }
 982
 983 int diff_unmodified_pair(struct diff_filepair *p)
 984 {
 985         /* This function is written stricter than necessary to support
 986          * the currently implemented transformers, but the idea is to
 987          * let transformers to produce diff_filepairs any way they want,
 988          * and filter and clean them up here before producing the output.
 989          */
 990         struct diff_filespec *one, *two;
 991
 992         if (DIFF_PAIR_UNMERGED(p))
 993                 return 0; /* unmerged is interesting */
 994
 995         one = p->one;
 996         two = p->two;
 997
 998         /* deletion, addition, mode or type change
 999          * and rename are all interesting.
1000          */
1001         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1002             DIFF_PAIR_MODE_CHANGED(p) ||
1003             strcmp(one->path, two->path))
1004                 return 0;
1005
1006         /* both are valid and point at the same path.  that is, we are
1007          * dealing with a change.
1008          */
1009         if (one->sha1_valid && two->sha1_valid &&
1010             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1011                 return 1; /* no change */
1012         if (!one->sha1_valid && !two->sha1_valid)
1013                 return 1; /* both look at the same file on the filesystem. */
1014         return 0;
1015 }
1016
1017 static void diff_flush_patch(struct diff_filepair *p)
1018 {
1019         if (diff_unmodified_pair(p))
1020                 return;
1021
1022         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1023             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1024                 return; /* no tree diffs in patch format */
1025
1026         run_diff(p);
1027 }
1028
1029 int diff_queue_is_empty(void)
1030 {
1031         struct diff_queue_struct *q = &diff_queued_diff;
1032         int i;
1033         for (i = 0; i < q->nr; i++)
1034                 if (!diff_unmodified_pair(q->queue[i]))
1035                         return 0;
1036         return 1;
1037 }
1038
1039 #if DIFF_DEBUG
1040 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1041 {
1042         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1043                 x, one ? one : "",
1044                 s->path,
1045                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1046                 s->mode,
1047                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1048         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1049                 x, one ? one : "",
1050                 s->size, s->xfrm_flags);
1051 }
1052
1053 void diff_debug_filepair(const struct diff_filepair *p, int i)
1054 {
1055         diff_debug_filespec(p->one, i, "one");
1056         diff_debug_filespec(p->two, i, "two");
1057         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1058                 p->score, p->status ? p->status : '?',
1059                 p->source_stays, p->broken_pair);
1060 }
1061
1062 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1063 {
1064         int i;
1065         if (msg)
1066                 fprintf(stderr, "%s\n", msg);
1067         fprintf(stderr, "q->nr = %d\n", q->nr);
1068         for (i = 0; i < q->nr; i++) {
1069                 struct diff_filepair *p = q->queue[i];
1070                 diff_debug_filepair(p, i);
1071         }
1072 }
1073 #endif
1074
1075 static void diff_resolve_rename_copy(void)
1076 {
1077         int i, j;
1078         struct diff_filepair *p, *pp;
1079         struct diff_queue_struct *q = &diff_queued_diff;
1080
1081         diff_debug_queue("resolve-rename-copy", q);
1082
1083         for (i = 0; i < q->nr; i++) {
1084                 p = q->queue[i];
1085                 p->status = 0; /* undecided */
1086                 if (DIFF_PAIR_UNMERGED(p))
1087                         p->status = DIFF_STATUS_UNMERGED;
1088                 else if (!DIFF_FILE_VALID(p->one))
1089                         p->status = DIFF_STATUS_ADDED;
1090                 else if (!DIFF_FILE_VALID(p->two))
1091                         p->status = DIFF_STATUS_DELETED;
1092                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1093                         p->status = DIFF_STATUS_TYPE_CHANGED;
1094
1095                 /* from this point on, we are dealing with a pair
1096                  * whose both sides are valid and of the same type, i.e.
1097                  * either in-place edit or rename/copy edit.
1098                  */
1099                 else if (DIFF_PAIR_RENAME(p)) {
1100                         if (p->source_stays) {
1101                                 p->status = DIFF_STATUS_COPIED;
1102                                 continue;
1103                         }
1104                         /* See if there is some other filepair that
1105                          * copies from the same source as us.  If so
1106                          * we are a copy.  Otherwise we are either a
1107                          * copy if the path stays, or a rename if it
1108                          * does not, but we already handled "stays" case.
1109                          */
1110                         for (j = i + 1; j < q->nr; j++) {
1111                                 pp = q->queue[j];
1112                                 if (strcmp(pp->one->path, p->one->path))
1113                                         continue; /* not us */
1114                                 if (!DIFF_PAIR_RENAME(pp))
1115                                         continue; /* not a rename/copy */
1116                                 /* pp is a rename/copy from the same source */
1117                                 p->status = DIFF_STATUS_COPIED;
1118                                 break;
1119                         }
1120                         if (!p->status)
1121                                 p->status = DIFF_STATUS_RENAMED;
1122                 }
1123                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1124                          p->one->mode != p->two->mode)
1125                         p->status = DIFF_STATUS_MODIFIED;
1126                 else {
1127                         /* This is a "no-change" entry and should not
1128                          * happen anymore, but prepare for broken callers.
1129                          */
1130                         error("feeding unmodified %s to diffcore",
1131                               p->one->path);
1132                         p->status = DIFF_STATUS_UNKNOWN;
1133                 }
1134         }
1135         diff_debug_queue("resolve-rename-copy done", q);
1136 }
1137
1138 void diff_flush(struct diff_options *options)
1139 {
1140         struct diff_queue_struct *q = &diff_queued_diff;
1141         int i;
1142         int inter_name_termination = '\t';
1143         int diff_output_format = options->output_format;
1144         int line_termination = options->line_termination;
1145
1146         if (!line_termination)
1147                 inter_name_termination = 0;
1148
1149         for (i = 0; i < q->nr; i++) {
1150                 struct diff_filepair *p = q->queue[i];
1151                 if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
1152                     (p->status == DIFF_STATUS_UNKNOWN))
1153                         continue;
1154                 if (p->status == 0)
1155                         die("internal error in diff-resolve-rename-copy");
1156                 switch (diff_output_format) {
1157                 case DIFF_FORMAT_PATCH:
1158                         diff_flush_patch(p);
1159                         break;
1160                 case DIFF_FORMAT_RAW:
1161                 case DIFF_FORMAT_NAME_STATUS:
1162                         diff_flush_raw(p, line_termination,
1163                                        inter_name_termination,
1164                                        diff_output_format);
1165                         break;
1166                 case DIFF_FORMAT_NAME:
1167                         diff_flush_name(p,
1168                                         inter_name_termination,
1169                                         line_termination);
1170                         break;
1171                 }
1172                 diff_free_filepair(q->queue[i]);
1173         }
1174         free(q->queue);
1175         q->queue = NULL;
1176         q->nr = q->alloc = 0;
1177 }
1178
1179 static void diffcore_apply_filter(const char *filter)
1180 {
1181         int i;
1182         struct diff_queue_struct *q = &diff_queued_diff;
1183         struct diff_queue_struct outq;
1184         outq.queue = NULL;
1185         outq.nr = outq.alloc = 0;
1186
1187         if (!filter)
1188                 return;
1189
1190         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
1191                 int found;
1192                 for (i = found = 0; !found && i < q->nr; i++) {
1193                         struct diff_filepair *p = q->queue[i];
1194                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1195                              ((p->score &&
1196                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1197                               (!p->score &&
1198                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1199                             ((p->status != DIFF_STATUS_MODIFIED) &&
1200                              strchr(filter, p->status)))
1201                                 found++;
1202                 }
1203                 if (found)
1204                         return;
1205
1206                 /* otherwise we will clear the whole queue
1207                  * by copying the empty outq at the end of this
1208                  * function, but first clear the current entries
1209                  * in the queue.
1210                  */
1211                 for (i = 0; i < q->nr; i++)
1212                         diff_free_filepair(q->queue[i]);
1213         }
1214         else {
1215                 /* Only the matching ones */
1216                 for (i = 0; i < q->nr; i++) {
1217                         struct diff_filepair *p = q->queue[i];
1218
1219                         if (((p->status == DIFF_STATUS_MODIFIED) &&
1220                              ((p->score &&
1221                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
1222                               (!p->score &&
1223                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
1224                             ((p->status != DIFF_STATUS_MODIFIED) &&
1225                              strchr(filter, p->status)))
1226                                 diff_q(&outq, p);
1227                         else
1228                                 diff_free_filepair(p);
1229                 }
1230         }
1231         free(q->queue);
1232         *q = outq;
1233 }
1234
1235 void diffcore_std(struct diff_options *options)
1236 {
1237         if (options->paths && options->paths[0])
1238                 diffcore_pathspec(options->paths);
1239         if (options->break_opt != -1)
1240                 diffcore_break(options->break_opt);
1241         if (options->detect_rename)
1242                 diffcore_rename(options);
1243         if (options->break_opt != -1)
1244                 diffcore_merge_broken();
1245         if (options->pickaxe)
1246                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1247         if (options->orderfile)
1248                 diffcore_order(options->orderfile);
1249         diff_resolve_rename_copy();
1250         diffcore_apply_filter(options->filter);
1251 }
1252
1253
1254 void diffcore_std_no_resolve(struct diff_options *options)
1255 {
1256         if (options->pickaxe)
1257                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
1258         if (options->orderfile)
1259                 diffcore_order(options->orderfile);
1260         diffcore_apply_filter(options->filter);
1261 }
1262
1263 void diff_addremove(struct diff_options *options,
1264                     int addremove, unsigned mode,
1265                     const unsigned char *sha1,
1266                     const char *base, const char *path)
1267 {
1268         char concatpath[PATH_MAX];
1269         struct diff_filespec *one, *two;
1270
1271         /* This may look odd, but it is a preparation for
1272          * feeding "there are unchanged files which should
1273          * not produce diffs, but when you are doing copy
1274          * detection you would need them, so here they are"
1275          * entries to the diff-core.  They will be prefixed
1276          * with something like '=' or '*' (I haven't decided
1277          * which but should not make any difference).
1278          * Feeding the same new and old to diff_change()
1279          * also has the same effect.
1280          * Before the final output happens, they are pruned after
1281          * merged into rename/copy pairs as appropriate.
1282          */
1283         if (options->reverse_diff)
1284                 addremove = (addremove == '+' ? '-' :
1285                              addremove == '-' ? '+' : addremove);
1286
1287         if (!path) path = "";
1288         sprintf(concatpath, "%s%s", base, path);
1289         one = alloc_filespec(concatpath);
1290         two = alloc_filespec(concatpath);
1291
1292         if (addremove != '+')
1293                 fill_filespec(one, sha1, mode);
1294         if (addremove != '-')
1295                 fill_filespec(two, sha1, mode);
1296
1297         diff_queue(&diff_queued_diff, one, two);
1298 }
1299
1300 void diff_change(struct diff_options *options,
1301                  unsigned old_mode, unsigned new_mode,
1302                  const unsigned char *old_sha1,
1303                  const unsigned char *new_sha1,
1304                  const char *base, const char *path)
1305 {
1306         char concatpath[PATH_MAX];
1307         struct diff_filespec *one, *two;
1308
1309         if (options->reverse_diff) {
1310                 unsigned tmp;
1311                 const unsigned char *tmp_c;
1312                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1313                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1314         }
1315         if (!path) path = "";
1316         sprintf(concatpath, "%s%s", base, path);
1317         one = alloc_filespec(concatpath);
1318         two = alloc_filespec(concatpath);
1319         fill_filespec(one, old_sha1, old_mode);
1320         fill_filespec(two, new_sha1, new_mode);
1321
1322         diff_queue(&diff_queued_diff, one, two);
1323 }
1324
1325 void diff_unmerge(struct diff_options *options,
1326                   const char *path)
1327 {
1328         struct diff_filespec *one, *two;
1329         one = alloc_filespec(path);
1330         two = alloc_filespec(path);
1331         diff_queue(&diff_queued_diff, one, two);
1332 }