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                                                  int find_only,
 240                                                  unsigned long size)
 241 {
 242         int first, last;
 243         struct sha1_size_cache *e;
 244
 245         first = 0;
 246         last = sha1_size_cache_nr;
 247         while (last > first) {
 248                 int cmp, next = (last + first) >> 1;
 249                 e = sha1_size_cache[next];
 250                 cmp = memcmp(e->sha1, sha1, 20);
 251                 if (!cmp)
 252                         return e;
 253                 if (cmp < 0) {
 254                         last = next;
 255                         continue;
 256                 }
 257                 first = next+1;
 258         }
 259         /* not found */
 260         if (find_only)
 261                 return NULL;
 262         /* insert to make it at "first" */
 263         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
 264                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
 265                 sha1_size_cache = xrealloc(sha1_size_cache,
 266                                            sha1_size_cache_alloc *
 267                                            sizeof(*sha1_size_cache));
 268         }
 269         sha1_size_cache_nr++;
 270         if (first < sha1_size_cache_nr)
 271                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
 272                         (sha1_size_cache_nr - first - 1) *
 273                         sizeof(*sha1_size_cache));
 274         e = xmalloc(sizeof(struct sha1_size_cache));
 275         sha1_size_cache[first] = e;
 276         memcpy(e->sha1, sha1, 20);
 277         e->size = size;
 278         return e;
 279 }
 280
 281 /*
 282  * While doing rename detection and pickaxe operation, we may need to
 283  * grab the data for the blob (or file) for our own in-core comparison.
 284  * diff_filespec has data and size fields for this purpose.
 285  */
 286 int diff_populate_filespec(struct diff_filespec *s, int size_only)
 287 {
 288         int err = 0;
 289         if (!DIFF_FILE_VALID(s))
 290                 die("internal error: asking to populate invalid file.");
 291         if (S_ISDIR(s->mode))
 292                 return -1;
 293
 294         if (!use_size_cache)
 295                 size_only = 0;
 296
 297         if (s->data)
 298                 return err;
 299         if (!s->sha1_valid ||
 300             work_tree_matches(s->path, s->sha1)) {
 301                 struct stat st;
 302                 int fd;
 303                 if (lstat(s->path, &st) < 0) {
 304                         if (errno == ENOENT) {
 305                         err_empty:
 306                                 err = -1;
 307                         empty:
 308                                 s->data = "";
 309                                 s->size = 0;
 310                                 return err;
 311                         }
 312                 }
 313                 s->size = st.st_size;
 314                 if (!s->size)
 315                         goto empty;
 316                 if (size_only)
 317                         return 0;
 318                 if (S_ISLNK(st.st_mode)) {
 319                         int ret;
 320                         s->data = xmalloc(s->size);
 321                         s->should_free = 1;
 322                         ret = readlink(s->path, s->data, s->size);
 323                         if (ret < 0) {
 324                                 free(s->data);
 325                                 goto err_empty;
 326                         }
 327                         return 0;
 328                 }
 329                 fd = open(s->path, O_RDONLY);
 330                 if (fd < 0)
 331                         goto err_empty;
 332                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 333                 s->should_munmap = 1;
 334                 close(fd);
 335         }
 336         else {
 337                 char type[20];
 338                 struct sha1_size_cache *e;
 339
 340                 if (size_only) {
 341                         e = locate_size_cache(s->sha1, 1, 0);
 342                         if (e) {
 343                                 s->size = e->size;
 344                                 return 0;
 345                         }
 346                         if (!sha1_file_size(s->sha1, &s->size))
 347                                 locate_size_cache(s->sha1, 0, s->size);
 348                 }
 349                 else {
 350                         s->data = read_sha1_file(s->sha1, type, &s->size);
 351                         s->should_free = 1;
 352                 }
 353         }
 354         return 0;
 355 }
 356
 357 void diff_free_filespec_data(struct diff_filespec *s)
 358 {
 359         if (s->should_free)
 360                 free(s->data);
 361         else if (s->should_munmap)
 362                 munmap(s->data, s->size);
 363         s->should_free = s->should_munmap = 0;
 364         s->data = NULL;
 365 }
 366
 367 static void prep_temp_blob(struct diff_tempfile *temp,
 368                            void *blob,
 369                            unsigned long size,
 370                            unsigned char *sha1,
 371                            int mode)
 372 {
 373         int fd;
 374
 375         strcpy(temp->tmp_path, ".diff_XXXXXX");
 376         fd = mkstemp(temp->tmp_path);
 377         if (fd < 0)
 378                 die("unable to create temp-file");
 379         if (write(fd, blob, size) != size)
 380                 die("unable to write temp-file");
 381         close(fd);
 382         temp->name = temp->tmp_path;
 383         strcpy(temp->hex, sha1_to_hex(sha1));
 384         temp->hex[40] = 0;
 385         sprintf(temp->mode, "%06o", mode);
 386 }
 387
 388 static void prepare_temp_file(const char *name,
 389                               struct diff_tempfile *temp,
 390                               struct diff_filespec *one)
 391 {
 392         if (!DIFF_FILE_VALID(one)) {
 393         not_a_valid_file:
 394                 /* A '-' entry produces this for file-2, and
 395                  * a '+' entry produces this for file-1.
 396                  */
 397                 temp->name = "/dev/null";
 398                 strcpy(temp->hex, ".");
 399                 strcpy(temp->mode, ".");
 400                 return;
 401         }
 402
 403         if (!one->sha1_valid ||
 404             work_tree_matches(name, one->sha1)) {
 405                 struct stat st;
 406                 if (lstat(name, &st) < 0) {
 407                         if (errno == ENOENT)
 408                                 goto not_a_valid_file;
 409                         die("stat(%s): %s", name, strerror(errno));
 410                 }
 411                 if (S_ISLNK(st.st_mode)) {
 412                         int ret;
 413                         char *buf, buf_[1024];
 414                         buf = ((sizeof(buf_) < st.st_size) ?
 415                                xmalloc(st.st_size) : buf_);
 416                         ret = readlink(name, buf, st.st_size);
 417                         if (ret < 0)
 418                                 die("readlink(%s)", name);
 419                         prep_temp_blob(temp, buf, st.st_size,
 420                                        (one->sha1_valid ?
 421                                         one->sha1 : null_sha1),
 422                                        (one->sha1_valid ?
 423                                         one->mode : S_IFLNK));
 424                 }
 425                 else {
 426                         /* we can borrow from the file in the work tree */
 427                         temp->name = name;
 428                         if (!one->sha1_valid)
 429                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 430                         else
 431                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
 432                         /* Even though we may sometimes borrow the
 433                          * contents from the work tree, we always want
 434                          * one->mode.  mode is trustworthy even when
 435                          * !(one->sha1_valid), as long as
 436                          * DIFF_FILE_VALID(one).
 437                          */
 438                         sprintf(temp->mode, "%06o", one->mode);
 439                 }
 440                 return;
 441         }
 442         else {
 443                 if (diff_populate_filespec(one, 0))
 444                         die("cannot read data blob for %s", one->path);
 445                 prep_temp_blob(temp, one->data, one->size,
 446                                one->sha1, one->mode);
 447         }
 448 }
 449
 450 static void remove_tempfile(void)
 451 {
 452         int i;
 453
 454         for (i = 0; i < 2; i++)
 455                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 456                         unlink(diff_temp[i].name);
 457                         diff_temp[i].name = NULL;
 458                 }
 459 }
 460
 461 static void remove_tempfile_on_signal(int signo)
 462 {
 463         remove_tempfile();
 464 }
 465
 466 /* An external diff command takes:
 467  *
 468  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 469  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
 470  *
 471  */
 472 static void run_external_diff(const char *pgm,
 473                               const char *name,
 474                               const char *other,
 475                               struct diff_filespec *one,
 476                               struct diff_filespec *two,
 477                               const char *xfrm_msg)
 478 {
 479         struct diff_tempfile *temp = diff_temp;
 480         pid_t pid;
 481         int status;
 482         static int atexit_asked = 0;
 483
 484         if (one && two) {
 485                 prepare_temp_file(name, &temp[0], one);
 486                 prepare_temp_file(other ? : name, &temp[1], two);
 487                 if (! atexit_asked &&
 488                     (temp[0].name == temp[0].tmp_path ||
 489                      temp[1].name == temp[1].tmp_path)) {
 490                         atexit_asked = 1;
 491                         atexit(remove_tempfile);
 492                 }
 493                 signal(SIGINT, remove_tempfile_on_signal);
 494         }
 495
 496         fflush(NULL);
 497         pid = fork();
 498         if (pid < 0)
 499                 die("unable to fork");
 500         if (!pid) {
 501                 if (pgm) {
 502                         if (one && two) {
 503                                 const char *exec_arg[10];
 504                                 const char **arg = &exec_arg[0];
 505                                 *arg++ = pgm;
 506                                 *arg++ = name;
 507                                 *arg++ = temp[0].name;
 508                                 *arg++ = temp[0].hex;
 509                                 *arg++ = temp[0].mode;
 510                                 *arg++ = temp[1].name;
 511                                 *arg++ = temp[1].hex;
 512                                 *arg++ = temp[1].mode;
 513                                 if (other) {
 514                                         *arg++ = other;
 515                                         *arg++ = xfrm_msg;
 516                                 }
 517                                 *arg = NULL;
 518                                 execvp(pgm, (char *const*) exec_arg);
 519                         }
 520                         else
 521                                 execlp(pgm, pgm, name, NULL);
 522                 }
 523                 /*
 524                  * otherwise we use the built-in one.
 525                  */
 526                 if (one && two)
 527                         builtin_diff(name, other ? : name, temp, xfrm_msg);
 528                 else
 529                         printf("* Unmerged path %s\n", name);
 530                 exit(0);
 531         }
 532         if (waitpid(pid, &status, 0) < 0 ||
 533             !WIFEXITED(status) || WEXITSTATUS(status)) {
 534                 /* Earlier we did not check the exit status because
 535                  * diff exits non-zero if files are different, and
 536                  * we are not interested in knowing that.  It was a
 537                  * mistake which made it harder to quit a diff-*
 538                  * session that uses the git-apply-patch-script as
 539                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 540                  * should also exit non-zero only when it wants to
 541                  * abort the entire diff-* session.
 542                  */
 543                 remove_tempfile();
 544                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 545                 exit(1);
 546         }
 547         remove_tempfile();
 548 }
 549
 550 static void run_diff(const char *name,
 551                      const char *other,
 552                      struct diff_filespec *one,
 553                      struct diff_filespec *two,
 554                      const char *xfrm_msg)
 555 {
 556         const char *pgm = external_diff();
 557         if (!pgm &&
 558             one && two &&
 559             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
 560             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
 561                 /* a filepair that changes between file and symlink
 562                  * needs to be split into deletion and creation.
 563                  */
 564                 struct diff_filespec *null = alloc_filespec(two->path);
 565                 run_external_diff(NULL, name, other, one, null, xfrm_msg);
 566                 free(null);
 567                 null = alloc_filespec(one->path);
 568                 run_external_diff(NULL, name, other, null, two, xfrm_msg);
 569                 free(null);
 570         }
 571         else
 572                 run_external_diff(pgm, name, other, one, two, xfrm_msg);
 573 }
 574
 575 void diff_setup(int flags)
 576 {
 577         if (flags & DIFF_SETUP_REVERSE)
 578                 reverse_diff = 1;
 579         if (flags & DIFF_SETUP_USE_CACHE) {
 580                 if (!active_cache)
 581                         /* read-cache does not die even when it fails
 582                          * so it is safe for us to do this here.  Also
 583                          * it does not smudge active_cache or active_nr
 584                          * when it fails, so we do not have to worry about
 585                          * cleaning it up oufselves either.
 586                          */
 587                         read_cache();
 588         }
 589         if (flags & DIFF_SETUP_USE_SIZE_CACHE)
 590                 use_size_cache = 1;
 591
 592 }
 593
 594 static int parse_num(const char **cp_p)
 595 {
 596         int num, scale, ch, cnt;
 597         const char *cp = *cp_p;
 598
 599         cnt = num = 0;
 600         scale = 1;
 601         while ('0' <= (ch = *cp) && ch <= '9') {
 602                 if (cnt++ < 5) {
 603                         /* We simply ignore more than 5 digits precision. */
 604                         scale *= 10;
 605                         num = num * 10 + ch - '0';
 606                 }
 607                 *cp++;
 608         }
 609         *cp_p = cp;
 610
 611         /* user says num divided by scale and we say internally that
 612          * is MAX_SCORE * num / scale.
 613          */
 614         return (MAX_SCORE * num / scale);
 615 }
 616
 617 int diff_scoreopt_parse(const char *opt)
 618 {
 619         int opt1, opt2, cmd;
 620
 621         if (*opt++ != '-')
 622                 return -1;
 623         cmd = *opt++;
 624         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
 625                 return -1; /* that is not a -M, -C nor -B option */
 626
 627         opt1 = parse_num(&opt);
 628         if (cmd != 'B')
 629                 opt2 = 0;
 630         else {
 631                 if (*opt == 0)
 632                         opt2 = 0;
 633                 else if (*opt != '/')
 634                         return -1; /* we expect -B80/99 or -B80 */
 635                 else {
 636                         opt++;
 637                         opt2 = parse_num(&opt);
 638                 }
 639         }
 640         if (*opt != 0)
 641                 return -1;
 642         return opt1 | (opt2 << 16);
 643 }
 644
 645 struct diff_queue_struct diff_queued_diff;
 646
 647 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
 648 {
 649         if (queue->alloc <= queue->nr) {
 650                 queue->alloc = alloc_nr(queue->alloc);
 651                 queue->queue = xrealloc(queue->queue,
 652                                         sizeof(dp) * queue->alloc);
 653         }
 654         queue->queue[queue->nr++] = dp;
 655 }
 656
 657 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
 658                                  struct diff_filespec *one,
 659                                  struct diff_filespec *two)
 660 {
 661         struct diff_filepair *dp = xmalloc(sizeof(*dp));
 662         dp->one = one;
 663         dp->two = two;
 664         dp->score = 0;
 665         dp->source_stays = 0;
 666         dp->broken_pair = 0;
 667         diff_q(queue, dp);
 668         return dp;
 669 }
 670
 671 void diff_free_filepair(struct diff_filepair *p)
 672 {
 673         diff_free_filespec_data(p->one);
 674         diff_free_filespec_data(p->two);
 675         free(p);
 676 }
 677
 678 static void diff_flush_raw(struct diff_filepair *p,
 679                            int line_termination,
 680                            int inter_name_termination)
 681 {
 682         int two_paths;
 683         char status[10];
 684
 685         if (line_termination) {
 686                 const char *err = "path %s cannot be expressed without -z";
 687                 if (strchr(p->one->path, line_termination) ||
 688                     strchr(p->one->path, inter_name_termination))
 689                         die(err, p->one->path);
 690                 if (strchr(p->two->path, line_termination) ||
 691                     strchr(p->two->path, inter_name_termination))
 692                         die(err, p->two->path);
 693         }
 694
 695         switch (p->status) {
 696         case 'C': case 'R':
 697                 two_paths = 1;
 698                 sprintf(status, "%c%03d", p->status,
 699                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
 700                 break;
 701         case 'N': case 'D':
 702                 two_paths = 0;
 703                 if (p->score)
 704                         sprintf(status, "%c%03d", p->status,
 705                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 706                 else {
 707                         status[0] = p->status;
 708                         status[1] = 0;
 709                 }
 710                 break;
 711         default:
 712                 two_paths = 0;
 713                 status[0] = p->status;
 714                 status[1] = 0;
 715                 break;
 716         }
 717         printf(":%06o %06o %s ",
 718                p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
 719         printf("%s %s%c%s",
 720                sha1_to_hex(p->two->sha1),
 721                status,
 722                inter_name_termination,
 723                p->one->path);
 724         if (two_paths)
 725                 printf("%c%s", inter_name_termination, p->two->path);
 726         putchar(line_termination);
 727 }
 728
 729 int diff_unmodified_pair(struct diff_filepair *p)
 730 {
 731         /* This function is written stricter than necessary to support
 732          * the currently implemented transformers, but the idea is to
 733          * let transformers to produce diff_filepairs any way they want,
 734          * and filter and clean them up here before producing the output.
 735          */
 736         struct diff_filespec *one, *two;
 737
 738         if (DIFF_PAIR_UNMERGED(p))
 739                 return 0; /* unmerged is interesting */
 740
 741         one = p->one;
 742         two = p->two;
 743
 744         /* deletion, addition, mode or type change
 745          * and rename are all interesting.
 746          */
 747         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
 748             DIFF_PAIR_MODE_CHANGED(p) ||
 749             strcmp(one->path, two->path))
 750                 return 0;
 751
 752         /* both are valid and point at the same path.  that is, we are
 753          * dealing with a change.
 754          */
 755         if (one->sha1_valid && two->sha1_valid &&
 756             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
 757                 return 1; /* no change */
 758         if (!one->sha1_valid && !two->sha1_valid)
 759                 return 1; /* both look at the same file on the filesystem. */
 760         return 0;
 761 }
 762
 763 static void diff_flush_patch(struct diff_filepair *p)
 764 {
 765         const char *name, *other;
 766         char msg_[PATH_MAX*2+200], *msg;
 767
 768         if (diff_unmodified_pair(p))
 769                 return;
 770
 771         name = p->one->path;
 772         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
 773         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
 774             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
 775                 return; /* no tree diffs in patch format */
 776
 777         switch (p->status) {
 778         case 'C':
 779                 sprintf(msg_,
 780                         "similarity index %d%%\n"
 781                         "copy from %s\n"
 782                         "copy to %s",
 783                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 784                         p->one->path, p->two->path);
 785                 msg = msg_;
 786                 break;
 787         case 'R':
 788                 sprintf(msg_,
 789                         "similarity index %d%%\n"
 790                         "rename from %s\n"
 791                         "rename to %s",
 792                         (int)(0.5 + p->score * 100.0/MAX_SCORE),
 793                         p->one->path, p->two->path);
 794                 msg = msg_;
 795                 break;
 796         case 'D': case 'N':
 797                 if (DIFF_PAIR_BROKEN(p)) {
 798                         sprintf(msg_,
 799                                 "dissimilarity index %d%%",
 800                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
 801                         msg = msg_;
 802                 }
 803                 else
 804                         msg = NULL;
 805                 break;
 806         default:
 807                 msg = NULL;
 808         }
 809
 810         if (DIFF_PAIR_UNMERGED(p))
 811                 run_diff(name, NULL, NULL, NULL, NULL);
 812         else
 813                 run_diff(name, other, p->one, p->two, msg);
 814 }
 815
 816 int diff_queue_is_empty(void)
 817 {
 818         struct diff_queue_struct *q = &diff_queued_diff;
 819         int i;
 820         for (i = 0; i < q->nr; i++)
 821                 if (!diff_unmodified_pair(q->queue[i]))
 822                         return 0;
 823         return 1;
 824 }
 825
 826 #if DIFF_DEBUG
 827 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 828 {
 829         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
 830                 x, one ? : "",
 831                 s->path,
 832                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
 833                 s->mode,
 834                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
 835         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
 836                 x, one ? : "",
 837                 s->size, s->xfrm_flags);
 838 }
 839
 840 void diff_debug_filepair(const struct diff_filepair *p, int i)
 841 {
 842         diff_debug_filespec(p->one, i, "one");
 843         diff_debug_filespec(p->two, i, "two");
 844         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
 845                 p->score, p->status ? : '?',
 846                 p->source_stays, p->broken_pair);
 847 }
 848
 849 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
 850 {
 851         int i;
 852         if (msg)
 853                 fprintf(stderr, "%s\n", msg);
 854         fprintf(stderr, "q->nr = %d\n", q->nr);
 855         for (i = 0; i < q->nr; i++) {
 856                 struct diff_filepair *p = q->queue[i];
 857                 diff_debug_filepair(p, i);
 858         }
 859 }
 860 #endif
 861
 862 static void diff_resolve_rename_copy(void)
 863 {
 864         int i, j;
 865         struct diff_filepair *p, *pp;
 866         struct diff_queue_struct *q = &diff_queued_diff;
 867
 868         diff_debug_queue("resolve-rename-copy", q);
 869
 870         for (i = 0; i < q->nr; i++) {
 871                 p = q->queue[i];
 872                 p->status = 0; /* undecided */
 873                 if (DIFF_PAIR_UNMERGED(p))
 874                         p->status = 'U';
 875                 else if (!DIFF_FILE_VALID(p->one))
 876                         p->status = 'N';
 877                 else if (!DIFF_FILE_VALID(p->two))
 878                         p->status = 'D';
 879                 else if (DIFF_PAIR_TYPE_CHANGED(p))
 880                         p->status = 'T';
 881
 882                 /* from this point on, we are dealing with a pair
 883                  * whose both sides are valid and of the same type, i.e.
 884                  * either in-place edit or rename/copy edit.
 885                  */
 886                 else if (DIFF_PAIR_RENAME(p)) {
 887                         if (p->source_stays) {
 888                                 p->status = 'C';
 889                                 continue;
 890                         }
 891                         /* See if there is some other filepair that
 892                          * copies from the same source as us.  If so
 893                          * we are a copy.  Otherwise we are a rename.
 894                          */
 895                         for (j = i + 1; j < q->nr; j++) {
 896                                 pp = q->queue[j];
 897                                 if (strcmp(pp->one->path, p->one->path))
 898                                         continue; /* not us */
 899                                 if (!DIFF_PAIR_RENAME(pp))
 900                                         continue; /* not a rename/copy */
 901                                 /* pp is a rename/copy from the same source */
 902                                 p->status = 'C';
 903                                 break;
 904                         }
 905                         if (!p->status)
 906                                 p->status = 'R';
 907                 }
 908                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
 909                          p->one->mode != p->two->mode)
 910                         p->status = 'M';
 911                 else {
 912                         /* This is a "no-change" entry and should not
 913                          * happen anymore, but prepare for broken callers.
 914                          */
 915                         error("feeding unmodified %s to diffcore",
 916                               p->one->path);
 917                         p->status = 'X';
 918                 }
 919         }
 920         diff_debug_queue("resolve-rename-copy done", q);
 921 }
 922
 923 void diff_flush(int diff_output_style, int resolve_rename_copy)
 924 {
 925         struct diff_queue_struct *q = &diff_queued_diff;
 926         int i;
 927         int line_termination = '\n';
 928         int inter_name_termination = '\t';
 929
 930         if (diff_output_style == DIFF_FORMAT_MACHINE)
 931                 line_termination = inter_name_termination = 0;
 932         if (resolve_rename_copy)
 933                 diff_resolve_rename_copy();
 934
 935         for (i = 0; i < q->nr; i++) {
 936                 struct diff_filepair *p = q->queue[i];
 937                 if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
 938                     (p->status == 'X'))
 939                         continue;
 940                 if (p->status == 0)
 941                         die("internal error in diff-resolve-rename-copy");
 942                 switch (diff_output_style) {
 943                 case DIFF_FORMAT_PATCH:
 944                         diff_flush_patch(p);
 945                         break;
 946                 case DIFF_FORMAT_HUMAN:
 947                 case DIFF_FORMAT_MACHINE:
 948                         diff_flush_raw(p, line_termination,
 949                                        inter_name_termination);
 950                         break;
 951                 }
 952         }
 953         for (i = 0; i < q->nr; i++)
 954                 diff_free_filepair(q->queue[i]);
 955         free(q->queue);
 956         q->queue = NULL;
 957         q->nr = q->alloc = 0;
 958 }
 959
 960 void diffcore_std(const char **paths,
 961                   int detect_rename, int rename_score,
 962                   const char *pickaxe, int pickaxe_opts,
 963                   int break_opt,
 964                   const char *orderfile)
 965 {
 966         if (paths && paths[0])
 967                 diffcore_pathspec(paths);
 968         if (break_opt != -1)
 969                 diffcore_break(break_opt);
 970         if (detect_rename)
 971                 diffcore_rename(detect_rename, rename_score);
 972         if (break_opt != -1)
 973                 diffcore_merge_broken();
 974         if (pickaxe)
 975                 diffcore_pickaxe(pickaxe, pickaxe_opts);
 976         if (orderfile)
 977                 diffcore_order(orderfile);
 978 }
 979
 980 void diff_addremove(int addremove, unsigned mode,
 981                     const unsigned char *sha1,
 982                     const char *base, const char *path)
 983 {
 984         char concatpath[PATH_MAX];
 985         struct diff_filespec *one, *two;
 986
 987         /* This may look odd, but it is a preparation for
 988          * feeding "there are unchanged files which should
 989          * not produce diffs, but when you are doing copy
 990          * detection you would need them, so here they are"
 991          * entries to the diff-core.  They will be prefixed
 992          * with something like '=' or '*' (I haven't decided
 993          * which but should not make any difference).
 994          * Feeding the same new and old to diff_change()
 995          * also has the same effect.
 996          * Before the final output happens, they are pruned after
 997          * merged into rename/copy pairs as appropriate.
 998          */
 999         if (reverse_diff)
1000                 addremove = (addremove == '+' ? '-' :
1001                              addremove == '-' ? '+' : addremove);
1002
1003         if (!path) path = "";
1004         sprintf(concatpath, "%s%s", base, path);
1005         one = alloc_filespec(concatpath);
1006         two = alloc_filespec(concatpath);
1007
1008         if (addremove != '+')
1009                 fill_filespec(one, sha1, mode);
1010         if (addremove != '-')
1011                 fill_filespec(two, sha1, mode);
1012
1013         diff_queue(&diff_queued_diff, one, two);
1014 }
1015
1016 void diff_helper_input(unsigned old_mode,
1017                        unsigned new_mode,
1018                        const unsigned char *old_sha1,
1019                        const unsigned char *new_sha1,
1020                        const char *old_path,
1021                        int status,
1022                        int score,
1023                        const char *new_path)
1024 {
1025         struct diff_filespec *one, *two;
1026         struct diff_filepair *dp;
1027
1028         one = alloc_filespec(old_path);
1029         two = alloc_filespec(new_path);
1030         if (old_mode)
1031                 fill_filespec(one, old_sha1, old_mode);
1032         if (new_mode)
1033                 fill_filespec(two, new_sha1, new_mode);
1034         dp = diff_queue(&diff_queued_diff, one, two);
1035         dp->score = score * MAX_SCORE / 100;
1036         dp->status = status;
1037 }
1038
1039 void diff_change(unsigned old_mode, unsigned new_mode,
1040                  const unsigned char *old_sha1,
1041                  const unsigned char *new_sha1,
1042                  const char *base, const char *path)
1043 {
1044         char concatpath[PATH_MAX];
1045         struct diff_filespec *one, *two;
1046
1047         if (reverse_diff) {
1048                 unsigned tmp;
1049                 const unsigned char *tmp_c;
1050                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
1051                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
1052         }
1053         if (!path) path = "";
1054         sprintf(concatpath, "%s%s", base, path);
1055         one = alloc_filespec(concatpath);
1056         two = alloc_filespec(concatpath);
1057         fill_filespec(one, old_sha1, old_mode);
1058         fill_filespec(two, new_sha1, new_mode);
1059
1060         diff_queue(&diff_queued_diff, one, two);
1061 }
1062
1063 void diff_unmerge(const char *path)
1064 {
1065         struct diff_filespec *one, *two;
1066         one = alloc_filespec(path);
1067         two = alloc_filespec(path);
1068         diff_queue(&diff_queued_diff, one, two);
1069 }