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 "diff.h"
   9 #include "diffcore.h"
  10
  11 static const char *diff_opts = "-pu";
  12 static unsigned char null_sha1[20] = { 0, };
  13
  14 static int reverse_diff;
  15 static int use_size_cache;
  16
  17 static const char *external_diff(void)
  18 {
  19         static const char *external_diff_cmd = NULL;
  20         static int done_preparing = 0;
  21
  22         if (done_preparing)
  23                 return external_diff_cmd;
  24
  25         /*
  26          * Default values above are meant to match the
  27          * Linux kernel development style.  Examples of
  28          * alternative styles you can specify via environment
  29          * variables are:
  30          *
  31          * GIT_DIFF_OPTS="-c";
  32          */
  33         if (gitenv("GIT_EXTERNAL_DIFF"))
  34                 external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
  35
  36         /* In case external diff fails... */
  37         diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
  38
  39         done_preparing = 1;
  40         return external_diff_cmd;
  41 }
  42
  43 /* Help to copy the thing properly quoted for the shell safety.
  44  * any single quote is replaced with '\'', and the caller is
  45  * expected to enclose the result within a single quote pair.
  46  *
  47  * E.g.
  48  *  original     sq_expand     result
  49  *  name     ==> name      ==> 'name'
  50  *  a b      ==> a b       ==> 'a b'
  51  *  a'b      ==> a'\''b    ==> 'a'\''b'
  52  */
  53 static char *sq_expand(const char *src)
  54 {
  55         static char *buf = NULL;
  56         int cnt, c;
  57         const char *cp;
  58         char *bp;
  59
  60         /* count bytes needed to store the quoted string. */
  61         for (cnt = 1, cp = src; *cp; cnt++, cp++)
  62                 if (*cp == '\'')
  63                         cnt += 3;
  64
  65         buf = xmalloc(cnt);
  66         bp = buf;
  67         while ((c = *src++)) {
  68                 if (c != '\'')
  69                         *bp++ = c;
  70                 else {
  71                         bp = strcpy(bp, "'\\''");
  72                         bp += 4;
  73                 }
  74         }
  75         *bp = 0;
  76         return buf;
  77 }
  78
  79 static struct diff_tempfile {
  80         const char *name; /* filename external diff should read from */
  81         char hex[41];
  82         char mode[10];
  83         char tmp_path[50];
  84 } diff_temp[2];
  85
  86 static void builtin_diff(const char *name_a,
  87                          const char *name_b,
  88                          struct diff_tempfile *temp,
  89                          const char *xfrm_msg)
  90 {
  91         int i, next_at, cmd_size;
  92         const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
  93         const char *diff_arg  = "'%s' '%s'||:"; /* "||:" is to return 0 */
  94         const char *input_name_sq[2];
  95         const char *path0[2];
  96         const char *path1[2];
  97         const char *name_sq[2];
  98         char *cmd;
  99
 100         name_sq[0] = sq_expand(name_a);
 101         name_sq[1] = sq_expand(name_b);
 102
 103         /* diff_cmd and diff_arg have 6 %s in total which makes
 104          * the sum of these strings 12 bytes larger than required.
 105          * we use 2 spaces around diff-opts, and we need to count
 106          * terminating NUL, so we subtract 9 here.
 107          */
 108         cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 109                         strlen(diff_arg) - 9);
 110         for (i = 0; i < 2; i++) {
 111                 input_name_sq[i] = sq_expand(temp[i].name);
 112                 if (!strcmp(temp[i].name, "/dev/null")) {
 113                         path0[i] = "/dev/null";
 114                         path1[i] = "";
 115                 } else {
 116                         path0[i] = i ? "b/" : "a/";
 117                         path1[i] = name_sq[i];
 118                 }
 119                 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 120                              strlen(input_name_sq[i]));
 121         }
 122
 123         cmd = xmalloc(cmd_size);
 124
 125         next_at = 0;
 126         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 127                             diff_cmd,
 128                             path0[0], path1[0], path0[1], path1[1]);
 129         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 130                             " %s ", diff_opts);
 131         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 132                             diff_arg, input_name_sq[0], input_name_sq[1]);
 133
 134         printf("diff --git a/%s b/%s\n", name_a, name_b);
 135         if (!path1[0][0]) {
 136                 printf("new file mode %s\n", temp[1].mode);
 137                 if (xfrm_msg && xfrm_msg[0])
 138                         puts(xfrm_msg);
 139         }
 140         else if (!path1[1][0]) {
 141                 printf("deleted file mode %s\n", temp[0].mode);
 142                 if (xfrm_msg && xfrm_msg[0])
 143                         puts(xfrm_msg);
 144         }
 145         else {
 146                 if (strcmp(temp[0].mode, temp[1].mode)) {
 147                         printf("old mode %s\n", temp[0].mode);
 148                         printf("new mode %s\n", temp[1].mode);
 149                 }
 150                 if (xfrm_msg && xfrm_msg[0])
 151                         puts(xfrm_msg);
 152
 153                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 154                         /* we do not run diff between different kind
 155                          * of objects.
 156                          */
 157                         exit(0);
 158         }
 159         fflush(NULL);
 160         execlp("/bin/sh","sh", "-c", cmd, NULL);
 161 }
 162
 163 struct diff_filespec *alloc_filespec(const char *path)
 164 {
 165         int namelen = strlen(path);
 166         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
 167         spec->path = (char *)(spec + 1);
 168         strcpy(spec->path, path);
 169         spec->should_free = spec->should_munmap = 0;
 170         spec->xfrm_flags = 0;
 171         spec->size = 0;
 172         spec->data = NULL;
 173         spec->mode = 0;
 174         memset(spec->sha1, 0, 20);
 175         return spec;
 176 }
 177
 178 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 179                    unsigned short mode)
 180 {
 181         if (mode) {
 182                 spec->mode = DIFF_FILE_CANON_MODE(mode);
 183                 memcpy(spec->sha1, sha1, 20);
 184                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
 185         }
 186 }
 187
 188 /*
 189  * Given a name and sha1 pair, if the dircache tells us the file in
 190  * the work tree has that object contents, return true, so that
 191  * prepare_temp_file() does not have to inflate and extract.
 192  */
 193 static int work_tree_matches(const char *name, const unsigned char *sha1)
 194 {
 195         struct cache_entry *ce;
 196         struct stat st;
 197         int pos, len;
 198
 199         /* We do not read the cache ourselves here, because the
 200          * benchmark with my previous version that always reads cache
 201          * shows that it makes things worse for diff-tree comparing
 202          * two linux-2.6 kernel trees in an already checked out work
 203          * tree.  This is because most diff-tree comparisons deal with
 204          * only a small number of files, while reading the cache is
 205          * expensive for a large project, and its cost outweighs the
 206          * savings we get by not inflating the object to a temporary
 207          * file.  Practically, this code only helps when we are used
 208          * by diff-cache --cached, which does read the cache before
 209          * calling us.
 210          */
 211         if (!active_cache)
 212                 return 0;
 213
 214         len = strlen(name);
 215         pos = cache_name_pos(name, len);
 216         if (pos < 0)
 217                 return 0;
 218         ce = active_cache[pos];
 219         if ((lstat(name, &st) < 0) ||
 220             !S_ISREG(st.st_mode) || /* careful! */
 221             ce_match_stat(ce, &st) ||
 222             memcmp(sha1, ce->sha1, 20))
 223                 return 0;
 224         /* we return 1 only when we can stat, it is a regular file,
 225          * stat information matches, and sha1 recorded in the cache
 226          * matches.  I.e. we know the file in the work tree really is
 227          * the same as the <name, sha1> pair.
 228          */
 229         return 1;
 230 }
 231
 232 static struct sha1_size_cache {
 233         unsigned char sha1[20];
 234         unsigned long size;
 235 } **sha1_size_cache;
 236 static int sha1_size_cache_nr, sha1_size_cache_alloc;
 237
 238 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
 239                                                  unsigned long size)
 240 {
 241         int first, last;
 242         struct sha1_size_cache *e;
 243
 244         first = 0;
 245         last = sha1_size_cache_nr;
 246         while (last > first) {
 247                 int next = (last + first) >> 1;
 248                 e = sha1_size_cache[next];
 249                 int cmp = memcmp(e->sha1, sha1, 20);
 250                 if (!cmp)
 251                         return e;
 252                 if (cmp < 0) {
 253                         last = next;
 254                         continue;
 255                 }
 256                 first = next+1;
 257         }
 258         /* not found */
 259         if (size == UINT_MAX)
 260                 return NULL;
 261         /* insert to make it at "first" */
 262         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 263                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 264                 sha1_size_cache = xrealloc(sha1_size_cache,
 265                                            sha1_size_cache_alloc *
 266                                            sizeof(*sha1_size_cache));
 267         }
 268         sha1_size_cache_nr++;
 269         if (first < sha1_size_cache_nr)
 270                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 271                         (sha1_size_cache_nr - first - 1) *
 272                         sizeof(*sha1_size_cache));
 273         e = xmalloc(sizeof(struct sha1_size_cache));
 274         sha1_size_cache[first] = e;
 275         memcpy(e->sha1, sha1, 20);
 276         e->size = size;
 277         return e;
 278 }
 279
 280 /*
 281  * While doing rename detection and pickaxe operation, we may need to
 282  * grab the data for the blob (or file) for our own in-core comparison.
 283  * diff_filespec has data and size fields for this purpose.
 284  */
 285 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 286 {
 287         int err = 0;
 288         if (!DIFF_FILE_VALID(s))
 289                 die("internal error: asking to populate invalid file.");
 290         if (S_ISDIR(s->mode))
 291                 return -1;
 292
 293         if (!use_size_cache)
 294                 size_only = 0;
 295
 296         if (s->data)
 297                 return err;
 298         if (!s->sha1_valid ||
 299             work_tree_matches(s->path, s->sha1)) {
 300                 struct stat st;
 301                 int fd;
 302                 if (lstat(s->path, &st) < 0) {
 303                         if (errno == ENOENT) {
 304                         err_empty:
 305                                 err = -1;
 306                         empty:
 307                                 s->data = "";
 308                                 s->size = 0;
 309                                 return err;
 310                         }
 311                 }
 312                 s->size = st.st_size;
 313                 if (!s->size)
 314                         goto empty;
 315                 if (size_only)
 316                         return 0;
 317                 if (S_ISLNK(st.st_mode)) {
 318                         int ret;
 319                         s->data = xmalloc(s->size);
 320                         s->should_free = 1;
 321                         ret = readlink(s->path, s->data, s->size);
 322                         if (ret < 0) {
 323                                 free(s->data);
 324                                 goto err_empty;
 325                         }
 326                         return 0;
 327                 }
 328                 fd = open(s->path, O_RDONLY);
 329                 if (fd < 0)
 330                         goto err_empty;
 331                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 332                 s->should_munmap = 1;
 333                 close(fd);
 334         }
 335         else {
 336                 char type[20];
 337                 struct sha1_size_cache *e;
 338
 339                 if (size_only) {
 340                         e = locate_size_cache(s->sha1, UINT_MAX);
 341                         if (e) {
 342                                 s->size = e->size;
 343                                 return 0;
 344                         }
 345                         if (!sha1_file_size(s->sha1, &s->size))
 346                                 locate_size_cache(s->sha1, s->size);
 347                 }
 348                 else {
 349                         s->data = read_sha1_file(s->sha1, type, &s->size);
 350                         s->should_free = 1;
 351                 }
 352         }
 353         return 0;
 354 }
 355
 356 void diff_free_filespec_data(struct diff_filespec *s)
 357 {
 358         if (s->should_free)
 359                 free(s->data);
 360         else if (s->should_munmap)
 361                 munmap(s->data, s->size);
 362         s->should_free = s->should_munmap = 0;
 363         s->data = NULL;
 364 }
 365
 366 static void prep_temp_blob(struct diff_tempfile *temp,
 367                            void *blob,
 368                            unsigned long size,
 369                            unsigned char *sha1,
 370                            int mode)
 371 {
 372         int fd;
 373
 374         strcpy(temp->tmp_path, ".diff_XXXXXX");
 375         fd = mkstemp(temp->tmp_path);
 376         if (fd < 0)
 377                 die("unable to create temp-file");
 378         if (write(fd, blob, size) != size)
 379                 die("unable to write temp-file");
 380         close(fd);
 381         temp->name = temp->tmp_path;
 382         strcpy(temp->hex, sha1_to_hex(sha1));
 383         temp->hex[40] = 0;
 384         sprintf(temp->mode, "%06o", mode);
 385 }
 386
 387 static void prepare_temp_file(const char *name,
 388                               struct diff_tempfile *temp,
 389                               struct diff_filespec *one)
 390 {
 391         if (!DIFF_FILE_VALID(one)) {
 392         not_a_valid_file:
 393                 /* A '-' entry produces this for file-2, and
 394                  * a '+' entry produces this for file-1.
 395                  */
 396                 temp->name = "/dev/null";
 397                 strcpy(temp->hex, ".");
 398                 strcpy(temp->mode, ".");
 399                 return;
 400         }
 401
 402         if (!one->sha1_valid ||
 403             work_tree_matches(name, one->sha1)) {
 404                 struct stat st;
 405                 if (lstat(name, &st) < 0) {
 406                         if (errno == ENOENT)
 407                                 goto not_a_valid_file;
 408                         die("stat(%s): %s", name, strerror(errno));
 409                 }
 410                 if (S_ISLNK(st.st_mode)) {
 411                         int ret;
 412                         char *buf, buf_[1024];
 413                         buf = ((sizeof(buf_) < st.st_size) ?
 414                                xmalloc(st.st_size) : buf_);
 415                         ret = readlink(name, buf, st.st_size);
 416                         if (ret < 0)
 417                                 die("readlink(%s)", name);
 418                         prep_temp_blob(temp, buf, st.st_size,
 419                                        (one->sha1_valid ?
 420                                         one->sha1 : null_sha1),
 421                                        (one->sha1_valid ?
 422                                         one->mode : S_IFLNK));
 423                 }
 424                 else {
 425                         /* we can borrow from the file in the work tree */
 426                         temp->name = name;
 427                         if (!one->sha1_valid)
 428                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 429                         else
 430                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 431                         /* Even though we may sometimes borrow the
 432                          * contents from the work tree, we always want
 433                          * one->mode.  mode is trustworthy even when
 434                          * !(one->sha1_valid), as long as
 435                          * DIFF_FILE_VALID(one).
 436                          */
 437                         sprintf(temp->mode, "%06o", one->mode);
 438                 }
 439                 return;
 440         }
 441         else {
 442                 if (diff_populate_filespec(one, 0))
 443                         die("cannot read data blob for %s", one->path);
 444                 prep_temp_blob(temp, one->data, one->size,
 445                                one->sha1, one->mode);
 446         }
 447 }
 448
 449 static void remove_tempfile(void)
 450 {
 451         int i;
 452
 453         for (i = 0; i < 2; i++)
 454                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 455                         unlink(diff_temp[i].name);
 456                         diff_temp[i].name = NULL;
 457                 }
 458 }
 459
 460 static void remove_tempfile_on_signal(int signo)
 461 {
 462         remove_tempfile();
 463 }
 464
 465 /* An external diff command takes:
 466  *
 467  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 468  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 469  *
 470  */
 471 static void run_external_diff(const char *pgm,
 472                               const char *name,
 473                               const char *other,
 474                               struct diff_filespec *one,
 475                               struct diff_filespec *two,
 476                               const char *xfrm_msg)
 477 {
 478         struct diff_tempfile *temp = diff_temp;
 479         pid_t pid;
 480         int status;
 481         static int atexit_asked = 0;
 482
 483         if (one && two) {
 484                 prepare_temp_file(name, &temp[0], one);
 485                 prepare_temp_file(other ? : name, &temp[1], two);
 486                 if (! atexit_asked &&
 487                     (temp[0].name == temp[0].tmp_path ||
 488                      temp[1].name == temp[1].tmp_path)) {
 489                         atexit_asked = 1;
 490                         atexit(remove_tempfile);
 491                 }
 492                 signal(SIGINT, remove_tempfile_on_signal);
 493         }
 494
 495         fflush(NULL);
 496         pid = fork();
 497         if (pid < 0)
 498                 die("unable to fork");
 499         if (!pid) {
 500                 if (pgm) {
 501                         if (one && two) {
 502                                 const char *exec_arg[10];
 503                                 const char **arg = &exec_arg[0];
 504                                 *arg++ = pgm;
 505                                 *arg++ = name;
 506                                 *arg++ = temp[0].name;
 507                                 *arg++ = temp[0].hex;
 508                                 *arg++ = temp[0].mode;
 509                                 *arg++ = temp[1].name;
 510                                 *arg++ = temp[1].hex;
 511                                 *arg++ = temp[1].mode;
 512                                 if (other) {
 513                                         *arg++ = other;
 514                                         *arg++ = xfrm_msg;
 515                                 }
 516                                 *arg = NULL;
 517                                 execvp(pgm, (char *const*) exec_arg);
 518                         }
 519                         else
 520                                 execlp(pgm, pgm, name, NULL);
 521                 }
 522                 /*
 523                  * otherwise we use the built-in one.
 524                  */
 525                 if (one && two)
 526                         builtin_diff(name, other ? : name, temp, xfrm_msg);
 527                 else
 528                         printf("* Unmerged path %s\n", name);
 529                 exit(0);
 530         }
 531         if (waitpid(pid, &status, 0) < 0 ||
 532             !WIFEXITED(status) || WEXITSTATUS(status)) {
 533                 /* Earlier we did not check the exit status because
 534                  * diff exits non-zero if files are different, and
 535                  * we are not interested in knowing that.  It was a
 536                  * mistake which made it harder to quit a diff-*
 537                  * session that uses the git-apply-patch-script as
 538                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 539                  * should also exit non-zero only when it wants to
 540                  * abort the entire diff-* session.
 541                  */
 542                 remove_tempfile();
 543                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 544                 exit(1);
 545         }
 546         remove_tempfile();
 547 }
 548
 549 static void run_diff(const char *name,
 550                      const char *other,
 551                      struct diff_filespec *one,
 552                      struct diff_filespec *two,
 553                      const char *xfrm_msg)
 554 {
 555         const char *pgm = external_diff();
 556         if (!pgm &&
 557             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 558             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 559                 /* a filepair that changes between file and symlink
 560                  * needs to be split into deletion and creation.
 561                  */
 562                 struct diff_filespec *null = alloc_filespec(two->path);
 563                 run_external_diff(NULL, name, other, one, null, xfrm_msg);
 564                 free(null);
 565                 null = alloc_filespec(one->path);
 566                 run_external_diff(NULL, name, other, null, two, xfrm_msg);
 567                 free(null);
 568         }
 569         else
 570                 run_external_diff(pgm, name, other, one, two, xfrm_msg);
 571 }
 572
 573 void diff_setup(int flags)
 574 {
 575         if (flags & DIFF_SETUP_REVERSE)
 576                 reverse_diff = 1;
 577         if (flags & DIFF_SETUP_USE_CACHE) {
 578                 if (!active_cache)
 579                         /* read-cache does not die even when it fails
 580                          * so it is safe for us to do this here.  Also
 581                          * it does not smudge active_cache or active_nr
 582                          * when it fails, so we do not have to worry about
 583                          * cleaning it up oufselves either.
 584                          */
 585                         read_cache();
 586         }
 587         if (flags & DIFF_SETUP_USE_SIZE_CACHE)
 588                 use_size_cache = 1;
 589
 590 }
 591
 592 static int parse_num(const char **cp_p)
 593 {
 594         int num, scale, ch, cnt;
 595         const char *cp = *cp_p;
 596
 597         cnt = num = 0;
 598         scale = 1;
 599         while ('0' <= (ch = *cp) && ch <= '9') {
 600                 if (cnt++ < 5) {
 601                         /* We simply ignore more than 5 digits precision. */
 602                         scale *= 10;
 603                         num = num * 10 + ch - '0';
 604                 }
 605                 *cp++;
 606         }
 607         *cp_p = cp;
 608
 609         /* user says num divided by scale and we say internally that
 610          * is MAX_SCORE * num / scale.
 611          */
 612         return (MAX_SCORE * num / scale);
 613 }
 614
 615 int diff_scoreopt_parse(const char *opt)
 616 {
 617         int opt1, opt2, cmd;
 618
 619         if (*opt++ != '-')
 620                 return -1;
 621         cmd = *opt++;
 622         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 623                 return -1; /* that is not a -M, -C nor -B option */
 624
 625         opt1 = parse_num(&opt);
 626         if (cmd != 'B')
 627                 opt2 = 0;
 628         else {
 629                 if (*opt == 0)
 630                         opt2 = 0;
 631                 else if (*opt != '/')
 632                         return -1; /* we expect -B80/99 or -B80 */
 633                 else {
 634                         opt++;
 635                         opt2 = parse_num(&opt);
 636                 }
 637         }
 638         if (*opt != 0)
 639                 return -1;
 640         return opt1 | (opt2 << 16);
 641 }
 642
 643 struct diff_queue_struct diff_queued_diff;
 644
 645 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 646 {
 647         if (queue->alloc <= queue->nr) {
 648                 queue->alloc = alloc_nr(queue->alloc);
 649                 queue->queue = xrealloc(queue->queue,
 650                                         sizeof(dp) * queue->alloc);
 651         }
 652         queue->queue[queue->nr++] = dp;
 653 }
 654
 655 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 656                                  struct diff_filespec *one,
 657                                  struct diff_filespec *two)
 658 {
 659         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 660         dp->one = one;
 661         dp->two = two;
 662         dp->score = 0;
 663         dp->source_stays = 0;
 664         dp->broken_pair = 0;
 665         diff_q(queue, dp);
 666         return dp;
 667 }
 668
 669 void diff_free_filepair(struct diff_filepair *p)
 670 {
 671         diff_free_filespec_data(p->one);
 672         diff_free_filespec_data(p->two);
 673         free(p);
 674 }
 675
 676 static void diff_flush_raw(struct diff_filepair *p,
 677                            int line_termination,
 678                            int inter_name_termination)
 679 {
 680         int two_paths;
 681         char status[10];
 682
 683         if (line_termination) {
 684                 const char *err = "path %s cannot be expressed without -z";
 685                 if (strchr(p->one->path, line_termination) ||
 686                     strchr(p->one->path, inter_name_termination))
 687                         die(err, p->one->path);
 688                 if (strchr(p->two->path, line_termination) ||
 689                     strchr(p->two->path, inter_name_termination))
 690                         die(err, p->two->path);
 691         }
 692
 693         switch (p->status) {
 694         case 'C': case 'R':
 695                 two_paths = 1;
 696                 sprintf(status, "%c%03d", p->status,
 697                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 698                 break;
 699         case 'N': case 'D':
 700                 two_paths = 0;
 701                 if (p->score)
 702                         sprintf(status, "%c%03d", p->status,
 703                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 704                 else {
 705                         status[0] = p->status;
 706                         status[1] = 0;
 707                 }
 708                 break;
 709         default:
 710                 two_paths = 0;
 711                 status[0] = p->status;
 712                 status[1] = 0;
 713                 break;
 714         }
 715         printf(":%06o %06o %s ",
 716                p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 717         printf("%s %s%c%s",
 718                sha1_to_hex(p->two->sha1),
 719                status,
 720                inter_name_termination,
 721                p->one->path);
 722         if (two_paths)
 723                 printf("%c%s", inter_name_termination, p->two->path);
 724         putchar(line_termination);
 725 }
 726
 727 int diff_unmodified_pair(struct diff_filepair *p)
 728 {
 729         /* This function is written stricter than necessary to support
 730          * the currently implemented transformers, but the idea is to
 731          * let transformers to produce diff_filepairs any way they want,
 732          * and filter and clean them up here before producing the output.
 733          */
 734         struct diff_filespec *one, *two;
 735
 736         if (DIFF_PAIR_UNMERGED(p))
 737                 return 0; /* unmerged is interesting */
 738
 739         one = p->one;
 740         two = p->two;
 741
 742         /* deletion, addition, mode or type change
 743          * and rename are all interesting.
 744          */
 745         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 746             DIFF_PAIR_MODE_CHANGED(p) ||
 747             strcmp(one->path, two->path))
 748                 return 0;
 749
 750         /* both are valid and point at the same path.  that is, we are
 751          * dealing with a change.
 752          */
 753         if (one->sha1_valid && two->sha1_valid &&
 754             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 755                 return 1; /* no change */
 756         if (!one->sha1_valid && !two->sha1_valid)
 757                 return 1; /* both look at the same file on the filesystem. */
 758         return 0;
 759 }
 760
 761 static void diff_flush_patch(struct diff_filepair *p)
 762 {
 763         const char *name, *other;
 764         char msg_[PATH_MAX*2+200], *msg;
 765
 766         if (diff_unmodified_pair(p))
 767                 return;
 768
 769         name = p->one->path;
 770         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 771         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 772             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 773                 return; /* no tree diffs in patch format */
 774
 775         switch (p->status) {
 776         case 'C':
 777                 sprintf(msg_,
 778                         "similarity index %d%%\n"
 779                         "copy from %s\n"
 780                         "copy to %s",
 781                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 782                         p->one->path, p->two->path);
 783                 msg = msg_;
 784                 break;
 785         case 'R':
 786                 sprintf(msg_,
 787                         "similarity index %d%%\n"
 788                         "rename old %s\n"
 789                         "rename new %s",
 790                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 791                         p->one->path, p->two->path);
 792                 msg = msg_;
 793                 break;
 794         case 'D': case 'N':
 795                 if (DIFF_PAIR_BROKEN(p)) {
 796                         sprintf(msg_,
 797                                 "dissimilarity index %d%%",
 798                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 799                         msg = msg_;
 800                 }
 801                 else
 802                         msg = NULL;
 803                 break;
 804         default:
 805                 msg = NULL;
 806         }
 807
 808         if (DIFF_PAIR_UNMERGED(p))
 809                 run_diff(name, NULL, NULL, NULL, NULL);
 810         else
 811                 run_diff(name, other, p->one, p->two, msg);
 812 }
 813
 814 int diff_queue_is_empty(void)
 815 {
 816         struct diff_queue_struct *q = &diff_queued_diff;
 817         int i;
 818         for (i = 0; i < q->nr; i++)
 819                 if (!diff_unmodified_pair(q->queue[i]))
 820                         return 0;
 821         return 1;
 822 }
 823
 824 #if DIFF_DEBUG
 825 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 826 {
 827         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 828                 x, one ? : "",
 829                 s->path,
 830                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
 831                 s->mode,
 832                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 833         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 834                 x, one ? : "",
 835                 s->size, s->xfrm_flags);
 836 }
 837
 838 void diff_debug_filepair(const struct diff_filepair *p, int i)
 839 {
 840         diff_debug_filespec(p->one, i, "one");
 841         diff_debug_filespec(p->two, i, "two");
 842         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
 843                 p->score, p->status ? : '?',
 844                 p->source_stays, p->broken_pair);
 845 }
 846
 847 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 848 {
 849         int i;
 850         if (msg)
 851                 fprintf(stderr, "%s\n", msg);
 852         fprintf(stderr, "q->nr = %d\n", q->nr);
 853         for (i = 0; i < q->nr; i++) {
 854                 struct diff_filepair *p = q->queue[i];
 855                 diff_debug_filepair(p, i);
 856         }
 857 }
 858 #endif
 859
 860 static void diff_resolve_rename_copy(void)
 861 {
 862         int i, j;
 863         struct diff_filepair *p, *pp;
 864         struct diff_queue_struct *q = &diff_queued_diff;
 865
 866         diff_debug_queue("resolve-rename-copy", q);
 867
 868         for (i = 0; i < q->nr; i++) {
 869                 p = q->queue[i];
 870                 p->status = 0; /* undecided */
 871                 if (DIFF_PAIR_UNMERGED(p))
 872                         p->status = 'U';
 873                 else if (!DIFF_FILE_VALID(p->one))
 874                         p->status = 'N';
 875                 else if (!DIFF_FILE_VALID(p->two))
 876                         p->status = 'D';
 877                 else if (DIFF_PAIR_TYPE_CHANGED(p))
 878                         p->status = 'T';
 879
 880                 /* from this point on, we are dealing with a pair
 881                  * whose both sides are valid and of the same type, i.e.
 882                  * either in-place edit or rename/copy edit.
 883                  */
 884                 else if (DIFF_PAIR_RENAME(p)) {
 885                         if (p->source_stays) {
 886                                 p->status = 'C';
 887                                 continue;
 888                         }
 889                         /* See if there is some other filepair that
 890                          * copies from the same source as us.  If so
 891                          * we are a copy.  Otherwise we are a rename.
 892                          */
 893                         for (j = i + 1; j < q->nr; j++) {
 894                                 pp = q->queue[j];
 895                                 if (strcmp(pp->one->path, p->one->path))
 896                                         continue; /* not us */
 897                                 if (!DIFF_PAIR_RENAME(pp))
 898                                         continue; /* not a rename/copy */
 899                                 /* pp is a rename/copy from the same source */
 900                                 p->status = 'C';
 901                                 break;
 902                         }
 903                         if (!p->status)
 904                                 p->status = 'R';
 905                 }
 906                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
 907                          p->one->mode != p->two->mode)
 908                         p->status = 'M';
 909                 else {
 910                         /* This is a "no-change" entry and should not
 911                          * happen anymore, but prepare for broken callers.
 912                          */
 913                         error("feeding unmodified %s to diffcore",
 914                               p->one->path);
 915                         p->status = 'X';
 916                 }
 917         }
 918         diff_debug_queue("resolve-rename-copy done", q);
 919 }
 920
 921 void diff_flush(int diff_output_style, int resolve_rename_copy)
 922 {
 923         struct diff_queue_struct *q = &diff_queued_diff;
 924         int i;
 925         int line_termination = '\n';
 926         int inter_name_termination = '\t';
 927
 928         if (diff_output_style == DIFF_FORMAT_MACHINE)
 929                 line_termination = inter_name_termination = 0;
 930         if (resolve_rename_copy)
 931                 diff_resolve_rename_copy();
 932
 933         for (i = 0; i < q->nr; i++) {
 934                 struct diff_filepair *p = q->queue[i];
 935                 if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
 936                     (p->status == 'X'))
 937                         continue;
 938                 if (p->status == 0)
 939                         die("internal error in diff-resolve-rename-copy");
 940                 switch (diff_output_style) {
 941                 case DIFF_FORMAT_PATCH:
 942                         diff_flush_patch(p);
 943                         break;
 944                 case DIFF_FORMAT_HUMAN:
 945                 case DIFF_FORMAT_MACHINE:
 946                         diff_flush_raw(p, line_termination,
 947                                        inter_name_termination);
 948                         break;
 949                 }
 950         }
 951         for (i = 0; i < q->nr; i++)
 952                 diff_free_filepair(q->queue[i]);
 953         free(q->queue);
 954         q->queue = NULL;
 955         q->nr = q->alloc = 0;
 956 }
 957
 958 void diffcore_std(const char **paths,
 959                   int detect_rename, int rename_score,
 960                   const char *pickaxe, int pickaxe_opts,
 961                   int break_opt,
 962                   const char *orderfile)
 963 {
 964         if (paths && paths[0])
 965                 diffcore_pathspec(paths);
 966         if (0 <= break_opt)
 967                 diffcore_break(break_opt);
 968         if (detect_rename)
 969                 diffcore_rename(detect_rename, rename_score);
 970         if (0 <= break_opt)
 971                 diffcore_merge_broken();
 972         if (pickaxe)
 973                 diffcore_pickaxe(pickaxe, pickaxe_opts);
 974         if (orderfile)
 975                 diffcore_order(orderfile);
 976 }
 977
 978 void diff_addremove(int addremove, unsigned mode,
 979                     const unsigned char *sha1,
 980                     const char *base, const char *path)
 981 {
 982         char concatpath[PATH_MAX];
 983         struct diff_filespec *one, *two;
 984
 985         /* This may look odd, but it is a preparation for
 986          * feeding "there are unchanged files which should
 987          * not produce diffs, but when you are doing copy
 988          * detection you would need them, so here they are"
 989          * entries to the diff-core.  They will be prefixed
 990          * with something like '=' or '*' (I haven't decided
 991          * which but should not make any difference).
 992          * Feeding the same new and old to diff_change()
 993          * also has the same effect.
 994          * Before the final output happens, they are pruned after
 995          * merged into rename/copy pairs as appropriate.
 996          */
 997         if (reverse_diff)
 998                 addremove = (addremove == '+' ? '-' :
 999                              addremove == '-' ? '+' : addremove);
1000
1001         if (!path) path = "";
1002         sprintf(concatpath, "%s%s", base, path);
1003         one = alloc_filespec(concatpath);
1004         two = alloc_filespec(concatpath);
1005
1006         if (addremove != '+')
1007                 fill_filespec(one, sha1, mode);
1008         if (addremove != '-')
1009                 fill_filespec(two, sha1, mode);
1010
1011         diff_queue(&diff_queued_diff, one, two);
1012 }
1013
1014 void diff_helper_input(unsigned old_mode,
1015                        unsigned new_mode,
1016                        const unsigned char *old_sha1,
1017                        const unsigned char *new_sha1,
1018                        const char *old_path,
1019                        int status,
1020                        int score,
1021                        const char *new_path)
1022 {
1023         struct diff_filespec *one, *two;
1024         struct diff_filepair *dp;
1025
1026         one = alloc_filespec(old_path);
1027         two = alloc_filespec(new_path);
1028         if (old_mode)
1029                 fill_filespec(one, old_sha1, old_mode);
1030         if (new_mode)
1031                 fill_filespec(two, new_sha1, new_mode);
1032         dp = diff_queue(&diff_queued_diff, one, two);
1033         dp->score = score * MAX_SCORE / 100;
1034         dp->status = status;
1035 }
1036
1037 void diff_change(unsigned old_mode, unsigned new_mode,
1038                  const unsigned char *old_sha1,
1039                  const unsigned char *new_sha1,
1040                  const char *base, const char *path)
1041 {
1042         char concatpath[PATH_MAX];
1043         struct diff_filespec *one, *two;
1044
1045         if (reverse_diff) {
1046                 unsigned tmp;
1047                 const unsigned char *tmp_c;
1048                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1049                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1050         }
1051         if (!path) path = "";
1052         sprintf(concatpath, "%s%s", base, path);
1053         one = alloc_filespec(concatpath);
1054         two = alloc_filespec(concatpath);
1055         fill_filespec(one, old_sha1, old_mode);
1056         fill_filespec(two, new_sha1, new_mode);
1057
1058         diff_queue(&diff_queued_diff, one, two);
1059 }
1060
1061 void diff_unmerge(const char *path)
1062 {
1063         struct diff_filespec *one, *two;
1064         one = alloc_filespec(path);
1065         two = alloc_filespec(path);
1066         diff_queue(&diff_queued_diff, one, two);
1067 }