read-tree.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #define DBRT_DEBUG 1
   7
   8 #include "cache.h"
   9
  10 #include "object.h"
  11 #include "tree.h"
  12
  13 static int merge = 0;
  14 static int update = 0;
  15 static int index_only = 0;
  16 static int nontrivial_merge = 0;
  17 static int trivial_merges_only = 0;
  18
  19 static int head_idx = -1;
  20 static int merge_size = 0;
  21
  22 static struct object_list *trees = NULL;
  23
  24 static struct cache_entry df_conflict_entry = {
  25 };
  26
  27 static struct tree_entry_list df_conflict_list = {
  28         .name = NULL,
  29         .next = &df_conflict_list
  30 };
  31
  32 typedef int (*merge_fn_t)(struct cache_entry **src);
  33
  34 static int entcmp(char *name1, int dir1, char *name2, int dir2)
  35 {
  36         int len1 = strlen(name1);
  37         int len2 = strlen(name2);
  38         int len = len1 < len2 ? len1 : len2;
  39         int ret = memcmp(name1, name2, len);
  40         unsigned char c1, c2;
  41         if (ret)
  42                 return ret;
  43         c1 = name1[len];
  44         c2 = name2[len];
  45         if (!c1 && dir1)
  46                 c1 = '/';
  47         if (!c2 && dir2)
  48                 c2 = '/';
  49         ret = (c1 < c2) ? -1 : (c1 > c2) ? 1 : 0;
  50         if (c1 && c2 && !ret)
  51                 ret = len1 - len2;
  52         return ret;
  53 }
  54
  55 static int unpack_trees_rec(struct tree_entry_list **posns, int len,
  56                             const char *base, merge_fn_t fn, int *indpos)
  57 {
  58         int baselen = strlen(base);
  59         int src_size = len + 1;
  60         do {
  61                 int i;
  62                 char *first;
  63                 int firstdir = 0;
  64                 int pathlen;
  65                 unsigned ce_size;
  66                 struct tree_entry_list **subposns;
  67                 struct cache_entry **src;
  68                 int any_files = 0;
  69                 int any_dirs = 0;
  70                 char *cache_name;
  71                 int ce_stage;
  72
  73                 /* Find the first name in the input. */
  74
  75                 first = NULL;
  76                 cache_name = NULL;
  77
  78                 /* Check the cache */
  79                 if (merge && *indpos < active_nr) {
  80                         /* This is a bit tricky: */
  81                         /* If the index has a subdirectory (with
  82                          * contents) as the first name, it'll get a
  83                          * filename like "foo/bar". But that's after
  84                          * "foo", so the entry in trees will get
  85                          * handled first, at which point we'll go into
  86                          * "foo", and deal with "bar" from the index,
  87                          * because the base will be "foo/". The only
  88                          * way we can actually have "foo/bar" first of
  89                          * all the things is if the trees don't
  90                          * contain "foo" at all, in which case we'll
  91                          * handle "foo/bar" without going into the
  92                          * directory, but that's fine (and will return
  93                          * an error anyway, with the added unknown
  94                          * file case.
  95                          */
  96
  97                         cache_name = active_cache[*indpos]->name;
  98                         if (strlen(cache_name) > baselen &&
  99                             !memcmp(cache_name, base, baselen)) {
 100                                 cache_name += baselen;
 101                                 first = cache_name;
 102                         } else {
 103                                 cache_name = NULL;
 104                         }
 105                 }
 106
 107 #if DBRT_DEBUG > 1
 108                 if (first)
 109                         printf("index %s\n", first);
 110 #endif
 111                 for (i = 0; i < len; i++) {
 112                         if (!posns[i] || posns[i] == &df_conflict_list)
 113                                 continue;
 114 #if DBRT_DEBUG > 1
 115                         printf("%d %s\n", i + 1, posns[i]->name);
 116 #endif
 117                         if (!first || entcmp(first, firstdir,
 118                                              posns[i]->name,
 119                                              posns[i]->directory) > 0) {
 120                                 first = posns[i]->name;
 121                                 firstdir = posns[i]->directory;
 122                         }
 123                 }
 124                 /* No name means we're done */
 125                 if (!first)
 126                         return 0;
 127
 128                 pathlen = strlen(first);
 129                 ce_size = cache_entry_size(baselen + pathlen);
 130
 131                 src = xmalloc(sizeof(struct cache_entry *) * src_size);
 132                 memset(src, 0, sizeof(struct cache_entry *) * src_size);
 133
 134                 subposns = xmalloc(sizeof(struct tree_list_entry *) * len);
 135                 memset(subposns, 0, sizeof(struct tree_list_entry *) * len);
 136
 137                 if (cache_name && !strcmp(cache_name, first)) {
 138                         any_files = 1;
 139                         src[0] = active_cache[*indpos];
 140                         remove_cache_entry_at(*indpos);
 141                 }
 142
 143                 for (i = 0; i < len; i++) {
 144                         struct cache_entry *ce;
 145
 146                         if (!posns[i] ||
 147                             (posns[i] != &df_conflict_list &&
 148                              strcmp(first, posns[i]->name))) {
 149                                 continue;
 150                         }
 151
 152                         if (posns[i] == &df_conflict_list) {
 153                                 src[i + merge] = &df_conflict_entry;
 154                                 continue;
 155                         }
 156
 157                         if (posns[i]->directory) {
 158                                 any_dirs = 1;
 159                                 parse_tree(posns[i]->item.tree);
 160                                 subposns[i] = posns[i]->item.tree->entries;
 161                                 posns[i] = posns[i]->next;
 162                                 src[i + merge] = &df_conflict_entry;
 163                                 continue;
 164                         }
 165
 166                         if (!merge)
 167                                 ce_stage = 0;
 168                         else if (i + 1 < head_idx)
 169                                 ce_stage = 1;
 170                         else if (i + 1 > head_idx)
 171                                 ce_stage = 3;
 172                         else
 173                                 ce_stage = 2;
 174
 175                         ce = xmalloc(ce_size);
 176                         memset(ce, 0, ce_size);
 177                         ce->ce_mode = create_ce_mode(posns[i]->mode);
 178                         ce->ce_flags = create_ce_flags(baselen + pathlen,
 179                                                        ce_stage);
 180                         memcpy(ce->name, base, baselen);
 181                         memcpy(ce->name + baselen, first, pathlen + 1);
 182
 183                         any_files = 1;
 184
 185                         memcpy(ce->sha1, posns[i]->item.any->sha1, 20);
 186                         src[i + merge] = ce;
 187                         subposns[i] = &df_conflict_list;
 188                         posns[i] = posns[i]->next;
 189                 }
 190                 if (any_files) {
 191                         if (merge) {
 192                                 int ret;
 193
 194 #if DBRT_DEBUG > 1
 195                                 printf("%s:\n", first);
 196                                 for (i = 0; i < src_size; i++) {
 197                                         printf(" %d ", i);
 198                                         if (src[i])
 199                                                 printf("%s\n", sha1_to_hex(src[i]->sha1));
 200                                         else
 201                                                 printf("\n");
 202                                 }
 203 #endif
 204                                 ret = fn(src);
 205
 206 #if DBRT_DEBUG > 1
 207                                 printf("Added %d entries\n", ret);
 208 #endif
 209                                 *indpos += ret;
 210                         } else {
 211                                 for (i = 0; i < src_size; i++) {
 212                                         if (src[i]) {
 213                                                 add_cache_entry(src[i], ADD_CACHE_OK_TO_ADD|ADD_CACHE_SKIP_DFCHECK);
 214                                         }
 215                                 }
 216                         }
 217                 }
 218                 if (any_dirs) {
 219                         char *newbase = xmalloc(baselen + 2 + pathlen);
 220                         memcpy(newbase, base, baselen);
 221                         memcpy(newbase + baselen, first, pathlen);
 222                         newbase[baselen + pathlen] = '/';
 223                         newbase[baselen + pathlen + 1] = '\0';
 224                         if (unpack_trees_rec(subposns, len, newbase, fn,
 225                                              indpos))
 226                                 return -1;
 227                         free(newbase);
 228                 }
 229                 free(subposns);
 230                 free(src);
 231         } while (1);
 232 }
 233
 234 static void reject_merge(struct cache_entry *ce)
 235 {
 236         die("Entry '%s' would be overwritten by merge. Cannot merge.",
 237             ce->name);
 238 }
 239
 240 /* Unlink the last component and attempt to remove leading
 241  * directories, in case this unlink is the removal of the
 242  * last entry in the directory -- empty directories are removed.
 243  */
 244 static void unlink_entry(char *name)
 245 {
 246         char *cp, *prev;
 247
 248         if (unlink(name))
 249                 return;
 250         prev = NULL;
 251         while (1) {
 252                 int status;
 253                 cp = strrchr(name, '/');
 254                 if (prev)
 255                         *prev = '/';
 256                 if (!cp)
 257                         break;
 258
 259                 *cp = 0;
 260                 status = rmdir(name);
 261                 if (status) {
 262                         *cp = '/';
 263                         break;
 264                 }
 265                 prev = cp;
 266         }
 267 }
 268
 269 static void check_updates(struct cache_entry **src, int nr)
 270 {
 271         static struct checkout state = {
 272                 .base_dir = "",
 273                 .force = 1,
 274                 .quiet = 1,
 275                 .refresh_cache = 1,
 276         };
 277         unsigned short mask = htons(CE_UPDATE);
 278         while (nr--) {
 279                 struct cache_entry *ce = *src++;
 280                 if (!ce->ce_mode) {
 281                         if (update)
 282                                 unlink_entry(ce->name);
 283                         continue;
 284                 }
 285                 if (ce->ce_flags & mask) {
 286                         ce->ce_flags &= ~mask;
 287                         if (update)
 288                                 checkout_entry(ce, &state);
 289                 }
 290         }
 291 }
 292
 293 static int unpack_trees(merge_fn_t fn)
 294 {
 295         int indpos = 0;
 296         unsigned len = object_list_length(trees);
 297         struct tree_entry_list **posns;
 298         int i;
 299         struct object_list *posn = trees;
 300         merge_size = len;
 301
 302         if (len) {
 303                 posns = xmalloc(len * sizeof(struct tree_entry_list *));
 304                 for (i = 0; i < len; i++) {
 305                         posns[i] = ((struct tree *) posn->item)->entries;
 306                         posn = posn->next;
 307                 }
 308                 if (unpack_trees_rec(posns, len, "", fn, &indpos))
 309                         return -1;
 310         }
 311
 312         if (trivial_merges_only && nontrivial_merge)
 313                 die("Merge requires file-level merging");
 314
 315         check_updates(active_cache, active_nr);
 316         return 0;
 317 }
 318
 319 static int list_tree(unsigned char *sha1)
 320 {
 321         struct tree *tree = parse_tree_indirect(sha1);
 322         if (!tree)
 323                 return -1;
 324         object_list_append(&tree->object, &trees);
 325         return 0;
 326 }
 327
 328 static int same(struct cache_entry *a, struct cache_entry *b)
 329 {
 330         if (!!a != !!b)
 331                 return 0;
 332         if (!a && !b)
 333                 return 1;
 334         return a->ce_mode == b->ce_mode &&
 335                 !memcmp(a->sha1, b->sha1, 20);
 336 }
 337
 338
 339 /*
 340  * When a CE gets turned into an unmerged entry, we
 341  * want it to be up-to-date
 342  */
 343 static void verify_uptodate(struct cache_entry *ce)
 344 {
 345         struct stat st;
 346
 347         if (index_only)
 348                 return;
 349
 350         if (!lstat(ce->name, &st)) {
 351                 unsigned changed = ce_match_stat(ce, &st);
 352                 if (!changed)
 353                         return;
 354                 errno = 0;
 355         }
 356         if (errno == ENOENT)
 357                 return;
 358         die("Entry '%s' not uptodate. Cannot merge.", ce->name);
 359 }
 360
 361 static int merged_entry(struct cache_entry *merge, struct cache_entry *old)
 362 {
 363         merge->ce_flags |= htons(CE_UPDATE);
 364         if (old) {
 365                 /*
 366                  * See if we can re-use the old CE directly?
 367                  * That way we get the uptodate stat info.
 368                  *
 369                  * This also removes the UPDATE flag on
 370                  * a match.
 371                  */
 372                 if (same(old, merge)) {
 373                         *merge = *old;
 374                 } else {
 375                         verify_uptodate(old);
 376                 }
 377         }
 378         merge->ce_flags &= ~htons(CE_STAGEMASK);
 379         add_cache_entry(merge, ADD_CACHE_OK_TO_ADD);
 380         return 1;
 381 }
 382
 383 static int deleted_entry(struct cache_entry *ce, struct cache_entry *old)
 384 {
 385         if (old)
 386                 verify_uptodate(old);
 387         ce->ce_mode = 0;
 388         add_cache_entry(ce, ADD_CACHE_OK_TO_ADD);
 389         return 1;
 390 }
 391
 392 static int keep_entry(struct cache_entry *ce)
 393 {
 394         add_cache_entry(ce, ADD_CACHE_OK_TO_ADD);
 395         return 1;
 396 }
 397
 398 #if DBRT_DEBUG
 399 static void show_stage_entry(FILE *o,
 400                              const char *label, const struct cache_entry *ce)
 401 {
 402         if (!ce)
 403                 fprintf(o, "%s (missing)\n", label);
 404         else
 405                 fprintf(o, "%s%06o %s %d\t%s\n",
 406                         label,
 407                         ntohl(ce->ce_mode),
 408                         sha1_to_hex(ce->sha1),
 409                         ce_stage(ce),
 410                         ce->name);
 411 }
 412 #endif
 413
 414 static int threeway_merge(struct cache_entry **stages)
 415 {
 416         struct cache_entry *index;
 417         struct cache_entry *head;
 418         struct cache_entry *remote = stages[head_idx + 1];
 419         int count;
 420         int head_match = 0;
 421         int remote_match = 0;
 422
 423         int df_conflict_head = 0;
 424         int df_conflict_remote = 0;
 425
 426         int any_anc_missing = 0;
 427         int i;
 428
 429         for (i = 1; i < head_idx; i++) {
 430                 if (!stages[i])
 431                         any_anc_missing = 1;
 432         }
 433
 434         index = stages[0];
 435         head = stages[head_idx];
 436
 437         if (head == &df_conflict_entry) {
 438                 df_conflict_head = 1;
 439                 head = NULL;
 440         }
 441
 442         if (remote == &df_conflict_entry) {
 443                 df_conflict_remote = 1;
 444                 remote = NULL;
 445         }
 446
 447         /* First, if there's a #16 situation, note that to prevent #13
 448          * and #14.
 449          */
 450         if (!same(remote, head)) {
 451                 for (i = 1; i < head_idx; i++) {
 452                         if (same(stages[i], head)) {
 453                                 head_match = i;
 454                         }
 455                         if (same(stages[i], remote)) {
 456                                 remote_match = i;
 457                         }
 458                 }
 459         }
 460
 461         /* We start with cases where the index is allowed to match
 462          * something other than the head: #14(ALT) and #2ALT, where it
 463          * is permitted to match the result instead.
 464          */
 465         /* #14, #14ALT, #2ALT */
 466         if (remote && !df_conflict_head && head_match && !remote_match) {
 467                 if (index && !same(index, remote) && !same(index, head))
 468                         reject_merge(index);
 469                 return merged_entry(remote, index);
 470         }
 471         /*
 472          * If we have an entry in the index cache, then we want to
 473          * make sure that it matches head.
 474          */
 475         if (index && !same(index, head)) {
 476                 reject_merge(index);
 477         }
 478
 479         if (head) {
 480                 /* #5ALT, #15 */
 481                 if (same(head, remote))
 482                         return merged_entry(head, index);
 483                 /* #13, #3ALT */
 484                 if (!df_conflict_remote && remote_match && !head_match)
 485                         return merged_entry(head, index);
 486         }
 487
 488         /* #1 */
 489         if (!head && !remote && any_anc_missing)
 490                 return 0;
 491
 492         /* Below are "no merge" cases, which require that the index be
 493          * up-to-date to avoid the files getting overwritten with
 494          * conflict resolution files.
 495          */
 496         if (index) {
 497                 verify_uptodate(index);
 498         }
 499
 500         nontrivial_merge = 1;
 501
 502         /* #2, #3, #4, #6, #7, #9, #11. */
 503         count = 0;
 504         if (!head_match || !remote_match) {
 505                 for (i = 1; i < head_idx; i++) {
 506                         if (stages[i]) {
 507                                 keep_entry(stages[i]);
 508                                 count++;
 509                                 break;
 510                         }
 511                 }
 512         }
 513 #if DBRT_DEBUG
 514         else {
 515                 fprintf(stderr, "read-tree: warning #16 detected\n");
 516                 show_stage_entry(stderr, "head   ", stages[head_match]);
 517                 show_stage_entry(stderr, "remote ", stages[remote_match]);
 518         }
 519 #endif
 520         if (head) { count += keep_entry(head); }
 521         if (remote) { count += keep_entry(remote); }
 522         return count;
 523 }
 524
 525 /*
 526  * Two-way merge.
 527  *
 528  * The rule is to "carry forward" what is in the index without losing
 529  * information across a "fast forward", favoring a successful merge
 530  * over a merge failure when it makes sense.  For details of the
 531  * "carry forward" rule, please see <Documentation/git-read-tree.txt>.
 532  *
 533  */
 534 static int twoway_merge(struct cache_entry **src)
 535 {
 536         struct cache_entry *current = src[0];
 537         struct cache_entry *oldtree = src[1], *newtree = src[2];
 538
 539         if (merge_size != 2)
 540                 return error("Cannot do a twoway merge of %d trees\n",
 541                              merge_size);
 542
 543         if (current) {
 544                 if ((!oldtree && !newtree) || /* 4 and 5 */
 545                     (!oldtree && newtree &&
 546                      same(current, newtree)) || /* 6 and 7 */
 547                     (oldtree && newtree &&
 548                      same(oldtree, newtree)) || /* 14 and 15 */
 549                     (oldtree && newtree &&
 550                      !same(oldtree, newtree) && /* 18 and 19*/
 551                      same(current, newtree))) {
 552                         return keep_entry(current);
 553                 }
 554                 else if (oldtree && !newtree && same(current, oldtree)) {
 555                         /* 10 or 11 */
 556                         return deleted_entry(oldtree, current);
 557                 }
 558                 else if (oldtree && newtree &&
 559                          same(current, oldtree) && !same(current, newtree)) {
 560                         /* 20 or 21 */
 561                         return merged_entry(newtree, current);
 562                 }
 563                 else {
 564                         /* all other failures */
 565                         if (oldtree)
 566                                 reject_merge(oldtree);
 567                         if (current)
 568                                 reject_merge(current);
 569                         if (newtree)
 570                                 reject_merge(newtree);
 571                         return -1;
 572                 }
 573         }
 574         else if (newtree)
 575                 return merged_entry(newtree, current);
 576         else
 577                 return deleted_entry(oldtree, current);
 578 }
 579
 580 /*
 581  * One-way merge.
 582  *
 583  * The rule is:
 584  * - take the stat information from stage0, take the data from stage1
 585  */
 586 static int oneway_merge(struct cache_entry **src)
 587 {
 588         struct cache_entry *old = src[0];
 589         struct cache_entry *a = src[1];
 590
 591         if (merge_size != 1)
 592                 return error("Cannot do a oneway merge of %d trees\n",
 593                              merge_size);
 594
 595         if (!a)
 596                 return 0;
 597         if (old && same(old, a)) {
 598                 return keep_entry(old);
 599         }
 600         return merged_entry(a, NULL);
 601 }
 602
 603 static int read_cache_unmerged(void)
 604 {
 605         int i, deleted;
 606         struct cache_entry **dst;
 607
 608         read_cache();
 609         dst = active_cache;
 610         deleted = 0;
 611         for (i = 0; i < active_nr; i++) {
 612                 struct cache_entry *ce = active_cache[i];
 613                 if (ce_stage(ce)) {
 614                         deleted++;
 615                         continue;
 616                 }
 617                 if (deleted)
 618                         *dst = ce;
 619                 dst++;
 620         }
 621         active_nr -= deleted;
 622         return deleted;
 623 }
 624
 625 static const char read_tree_usage[] = "git-read-tree (<sha> | -m [-u | -i] <sha1> [<sha2> [<sha3>]])";
 626
 627 static struct cache_file cache_file;
 628
 629 int main(int argc, char **argv)
 630 {
 631         int i, newfd, reset, stage = 0;
 632         unsigned char sha1[20];
 633         merge_fn_t fn = NULL;
 634
 635         setup_git_directory();
 636
 637         newfd = hold_index_file_for_update(&cache_file, get_index_file());
 638         if (newfd < 0)
 639                 die("unable to create new cachefile");
 640
 641         git_config(git_default_config);
 642
 643         merge = 0;
 644         reset = 0;
 645         for (i = 1; i < argc; i++) {
 646                 const char *arg = argv[i];
 647
 648                 /* "-u" means "update", meaning that a merge will update
 649                  * the working tree.
 650                  */
 651                 if (!strcmp(arg, "-u")) {
 652                         update = 1;
 653                         continue;
 654                 }
 655
 656                 /* "-i" means "index only", meaning that a merge will
 657                  * not even look at the working tree.
 658                  */
 659                 if (!strcmp(arg, "-i")) {
 660                         index_only = 1;
 661                         continue;
 662                 }
 663
 664                 /* This differs from "-m" in that we'll silently ignore unmerged entries */
 665                 if (!strcmp(arg, "--reset")) {
 666                         if (stage || merge)
 667                                 usage(read_tree_usage);
 668                         reset = 1;
 669                         merge = 1;
 670                         stage = 1;
 671                         read_cache_unmerged();
 672                         continue;
 673                 }
 674
 675                 if (!strcmp(arg, "--trivial")) {
 676                         trivial_merges_only = 1;
 677                         continue;
 678                 }
 679
 680                 /* "-m" stands for "merge", meaning we start in stage 1 */
 681                 if (!strcmp(arg, "-m")) {
 682                         if (stage || merge)
 683                                 usage(read_tree_usage);
 684                         if (read_cache_unmerged())
 685                                 die("you need to resolve your current index first");
 686                         stage = 1;
 687                         merge = 1;
 688                         continue;
 689                 }
 690
 691                 /* using -u and -i at the same time makes no sense */
 692                 if (1 < index_only + update)
 693                         usage(read_tree_usage);
 694
 695                 if (get_sha1(arg, sha1) < 0)
 696                         usage(read_tree_usage);
 697                 if (list_tree(sha1) < 0)
 698                         die("failed to unpack tree object %s", arg);
 699                 stage++;
 700         }
 701         if ((update||index_only) && !merge)
 702                 usage(read_tree_usage);
 703
 704         if (merge) {
 705                 if (stage < 2)
 706                         die("just how do you expect me to merge %d trees?", stage-1);
 707                 switch (stage - 1) {
 708                 case 1:
 709                         fn = oneway_merge;
 710                         break;
 711                 case 2:
 712                         fn = twoway_merge;
 713                         break;
 714                 case 3:
 715                         fn = threeway_merge;
 716                         break;
 717                 default:
 718                         fn = threeway_merge;
 719                         break;
 720                 }
 721
 722                 if (stage - 1 >= 3)
 723                         head_idx = stage - 2;
 724                 else
 725                         head_idx = 1;
 726         }
 727
 728         unpack_trees(fn);
 729         if (write_cache(newfd, active_cache, active_nr) ||
 730             commit_index_file(&cache_file))
 731                 die("unable to write new index file");
 732         return 0;
 733 }