tree-walk.c

   1 #include "cache.h"
   2 #include "tree-walk.h"
   3 #include "unpack-trees.h"
   4 #include "dir.h"
   5 #include "tree.h"
   6 #include "pathspec.h"
   7
   8 static const char *get_mode(const char *str, unsigned int *modep)
   9 {
  10         unsigned char c;
  11         unsigned int mode = 0;
  12
  13         if (*str == ' ')
  14                 return NULL;
  15
  16         while ((c = *str++) != ' ') {
  17                 if (c < '0' || c > '7')
  18                         return NULL;
  19                 mode = (mode << 3) + (c - '0');
  20         }
  21         *modep = mode;
  22         return str;
  23 }
  24
  25 static int decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size, struct strbuf *err)
  26 {
  27         const char *path;
  28         unsigned int mode, len;
  29
  30         if (size < 23 || buf[size - 21]) {
  31                 strbuf_addstr(err, _("too-short tree object"));
  32                 return -1;
  33         }
  34
  35         path = get_mode(buf, &mode);
  36         if (!path) {
  37                 strbuf_addstr(err, _("malformed mode in tree entry"));
  38                 return -1;
  39         }
  40         if (!*path) {
  41                 strbuf_addstr(err, _("empty filename in tree entry"));
  42                 return -1;
  43         }
  44 #ifdef GIT_WINDOWS_NATIVE
  45         if (protect_ntfs && strchr(path, '\\')) {
  46                 strbuf_addf(err, _("filename in tree entry contains backslash: '%s'"), path);
  47                 return -1;
  48         }
  49 #endif
  50         len = strlen(path) + 1;
  51
  52         /* Initialize the descriptor entry */
  53         desc->entry.path = path;
  54         desc->entry.mode = canon_mode(mode);
  55         desc->entry.oid  = (const struct object_id *)(path + len);
  56
  57         return 0;
  58 }
  59
  60 static int init_tree_desc_internal(struct tree_desc *desc, const void *buffer, unsigned long size, struct strbuf *err)
  61 {
  62         desc->buffer = buffer;
  63         desc->size = size;
  64         if (size)
  65                 return decode_tree_entry(desc, buffer, size, err);
  66         return 0;
  67 }
  68
  69 void init_tree_desc(struct tree_desc *desc, const void *buffer, unsigned long size)
  70 {
  71         struct strbuf err = STRBUF_INIT;
  72         if (init_tree_desc_internal(desc, buffer, size, &err))
  73                 die("%s", err.buf);
  74         strbuf_release(&err);
  75 }
  76
  77 int init_tree_desc_gently(struct tree_desc *desc, const void *buffer, unsigned long size)
  78 {
  79         struct strbuf err = STRBUF_INIT;
  80         int result = init_tree_desc_internal(desc, buffer, size, &err);
  81         if (result)
  82                 error("%s", err.buf);
  83         strbuf_release(&err);
  84         return result;
  85 }
  86
  87 void *fill_tree_descriptor(struct tree_desc *desc, const struct object_id *oid)
  88 {
  89         unsigned long size = 0;
  90         void *buf = NULL;
  91
  92         if (oid) {
  93                 buf = read_object_with_reference(oid->hash, tree_type, &size,
  94                                                  NULL);
  95                 if (!buf)
  96                         die("unable to read tree %s", oid_to_hex(oid));
  97         }
  98         init_tree_desc(desc, buf, size);
  99         return buf;
 100 }
 101
 102 static void entry_clear(struct name_entry *a)
 103 {
 104         memset(a, 0, sizeof(*a));
 105 }
 106
 107 static void entry_extract(struct tree_desc *t, struct name_entry *a)
 108 {
 109         *a = t->entry;
 110 }
 111
 112 static int update_tree_entry_internal(struct tree_desc *desc, struct strbuf *err)
 113 {
 114         const void *buf = desc->buffer;
 115         const unsigned char *end = desc->entry.oid->hash + 20;
 116         unsigned long size = desc->size;
 117         unsigned long len = end - (const unsigned char *)buf;
 118
 119         if (size < len)
 120                 die(_("too-short tree file"));
 121         buf = end;
 122         size -= len;
 123         desc->buffer = buf;
 124         desc->size = size;
 125         if (size)
 126                 return decode_tree_entry(desc, buf, size, err);
 127         return 0;
 128 }
 129
 130 void update_tree_entry(struct tree_desc *desc)
 131 {
 132         struct strbuf err = STRBUF_INIT;
 133         if (update_tree_entry_internal(desc, &err))
 134                 die("%s", err.buf);
 135         strbuf_release(&err);
 136 }
 137
 138 int update_tree_entry_gently(struct tree_desc *desc)
 139 {
 140         struct strbuf err = STRBUF_INIT;
 141         if (update_tree_entry_internal(desc, &err)) {
 142                 error("%s", err.buf);
 143                 strbuf_release(&err);
 144                 /* Stop processing this tree after error */
 145                 desc->size = 0;
 146                 return -1;
 147         }
 148         strbuf_release(&err);
 149         return 0;
 150 }
 151
 152 int tree_entry(struct tree_desc *desc, struct name_entry *entry)
 153 {
 154         if (!desc->size)
 155                 return 0;
 156
 157         *entry = desc->entry;
 158         update_tree_entry(desc);
 159         return 1;
 160 }
 161
 162 int tree_entry_gently(struct tree_desc *desc, struct name_entry *entry)
 163 {
 164         if (!desc->size)
 165                 return 0;
 166
 167         *entry = desc->entry;
 168         if (update_tree_entry_gently(desc))
 169                 return 0;
 170         return 1;
 171 }
 172
 173 void setup_traverse_info(struct traverse_info *info, const char *base)
 174 {
 175         int pathlen = strlen(base);
 176         static struct traverse_info dummy;
 177
 178         memset(info, 0, sizeof(*info));
 179         if (pathlen && base[pathlen-1] == '/')
 180                 pathlen--;
 181         info->pathlen = pathlen ? pathlen + 1 : 0;
 182         info->name.path = base;
 183         info->name.oid = (void *)(base + pathlen + 1);
 184         if (pathlen)
 185                 info->prev = &dummy;
 186 }
 187
 188 char *make_traverse_path(char *path, const struct traverse_info *info, const struct name_entry *n)
 189 {
 190         int len = tree_entry_len(n);
 191         int pathlen = info->pathlen;
 192
 193         path[pathlen + len] = 0;
 194         for (;;) {
 195                 memcpy(path + pathlen, n->path, len);
 196                 if (!pathlen)
 197                         break;
 198                 path[--pathlen] = '/';
 199                 n = &info->name;
 200                 len = tree_entry_len(n);
 201                 info = info->prev;
 202                 pathlen -= len;
 203         }
 204         return path;
 205 }
 206
 207 struct tree_desc_skip {
 208         struct tree_desc_skip *prev;
 209         const void *ptr;
 210 };
 211
 212 struct tree_desc_x {
 213         struct tree_desc d;
 214         struct tree_desc_skip *skip;
 215 };
 216
 217 static int check_entry_match(const char *a, int a_len, const char *b, int b_len)
 218 {
 219         /*
 220          * The caller wants to pick *a* from a tree or nothing.
 221          * We are looking at *b* in a tree.
 222          *
 223          * (0) If a and b are the same name, we are trivially happy.
 224          *
 225          * There are three possibilities where *a* could be hiding
 226          * behind *b*.
 227          *
 228          * (1) *a* == "t",   *b* == "ab"  i.e. *b* sorts earlier than *a* no
 229          *                                matter what.
 230          * (2) *a* == "t",   *b* == "t-2" and "t" is a subtree in the tree;
 231          * (3) *a* == "t-2", *b* == "t"   and "t-2" is a blob in the tree.
 232          *
 233          * Otherwise we know *a* won't appear in the tree without
 234          * scanning further.
 235          */
 236
 237         int cmp = name_compare(a, a_len, b, b_len);
 238
 239         /* Most common case first -- reading sync'd trees */
 240         if (!cmp)
 241                 return cmp;
 242
 243         if (0 < cmp) {
 244                 /* a comes after b; it does not matter if it is case (3)
 245                 if (b_len < a_len && !memcmp(a, b, b_len) && a[b_len] < '/')
 246                         return 1;
 247                 */
 248                 return 1; /* keep looking */
 249         }
 250
 251         /* b comes after a; are we looking at case (2)? */
 252         if (a_len < b_len && !memcmp(a, b, a_len) && b[a_len] < '/')
 253                 return 1; /* keep looking */
 254
 255         return -1; /* a cannot appear in the tree */
 256 }
 257
 258 /*
 259  * From the extended tree_desc, extract the first name entry, while
 260  * paying attention to the candidate "first" name.  Most importantly,
 261  * when looking for an entry, if there are entries that sorts earlier
 262  * in the tree object representation than that name, skip them and
 263  * process the named entry first.  We will remember that we haven't
 264  * processed the first entry yet, and in the later call skip the
 265  * entry we processed early when update_extended_entry() is called.
 266  *
 267  * E.g. if the underlying tree object has these entries:
 268  *
 269  *    blob    "t-1"
 270  *    blob    "t-2"
 271  *    tree    "t"
 272  *    blob    "t=1"
 273  *
 274  * and the "first" asks for "t", remember that we still need to
 275  * process "t-1" and "t-2" but extract "t".  After processing the
 276  * entry "t" from this call, the caller will let us know by calling
 277  * update_extended_entry() that we can remember "t" has been processed
 278  * already.
 279  */
 280
 281 static void extended_entry_extract(struct tree_desc_x *t,
 282                                    struct name_entry *a,
 283                                    const char *first,
 284                                    int first_len)
 285 {
 286         const char *path;
 287         int len;
 288         struct tree_desc probe;
 289         struct tree_desc_skip *skip;
 290
 291         /*
 292          * Extract the first entry from the tree_desc, but skip the
 293          * ones that we already returned in earlier rounds.
 294          */
 295         while (1) {
 296                 if (!t->d.size) {
 297                         entry_clear(a);
 298                         break; /* not found */
 299                 }
 300                 entry_extract(&t->d, a);
 301                 for (skip = t->skip; skip; skip = skip->prev)
 302                         if (a->path == skip->ptr)
 303                                 break; /* found */
 304                 if (!skip)
 305                         break;
 306                 /* We have processed this entry already. */
 307                 update_tree_entry(&t->d);
 308         }
 309
 310         if (!first || !a->path)
 311                 return;
 312
 313         /*
 314          * The caller wants "first" from this tree, or nothing.
 315          */
 316         path = a->path;
 317         len = tree_entry_len(a);
 318         switch (check_entry_match(first, first_len, path, len)) {
 319         case -1:
 320                 entry_clear(a);
 321         case 0:
 322                 return;
 323         default:
 324                 break;
 325         }
 326
 327         /*
 328          * We need to look-ahead -- we suspect that a subtree whose
 329          * name is "first" may be hiding behind the current entry "path".
 330          */
 331         probe = t->d;
 332         while (probe.size) {
 333                 entry_extract(&probe, a);
 334                 path = a->path;
 335                 len = tree_entry_len(a);
 336                 switch (check_entry_match(first, first_len, path, len)) {
 337                 case -1:
 338                         entry_clear(a);
 339                 case 0:
 340                         return;
 341                 default:
 342                         update_tree_entry(&probe);
 343                         break;
 344                 }
 345                 /* keep looking */
 346         }
 347         entry_clear(a);
 348 }
 349
 350 static void update_extended_entry(struct tree_desc_x *t, struct name_entry *a)
 351 {
 352         if (t->d.entry.path == a->path) {
 353                 update_tree_entry(&t->d);
 354         } else {
 355                 /* we have returned this entry early */
 356                 struct tree_desc_skip *skip = xmalloc(sizeof(*skip));
 357                 skip->ptr = a->path;
 358                 skip->prev = t->skip;
 359                 t->skip = skip;
 360         }
 361 }
 362
 363 static void free_extended_entry(struct tree_desc_x *t)
 364 {
 365         struct tree_desc_skip *p, *s;
 366
 367         for (s = t->skip; s; s = p) {
 368                 p = s->prev;
 369                 free(s);
 370         }
 371 }
 372
 373 static inline int prune_traversal(struct name_entry *e,
 374                                   struct traverse_info *info,
 375                                   struct strbuf *base,
 376                                   int still_interesting)
 377 {
 378         if (!info->pathspec || still_interesting == 2)
 379                 return 2;
 380         if (still_interesting < 0)
 381                 return still_interesting;
 382         return tree_entry_interesting(e, base, 0, info->pathspec);
 383 }
 384
 385 int traverse_trees(int n, struct tree_desc *t, struct traverse_info *info)
 386 {
 387         int error = 0;
 388         struct name_entry *entry = xmalloc(n*sizeof(*entry));
 389         int i;
 390         struct tree_desc_x *tx = xcalloc(n, sizeof(*tx));
 391         struct strbuf base = STRBUF_INIT;
 392         int interesting = 1;
 393         char *traverse_path;
 394
 395         for (i = 0; i < n; i++)
 396                 tx[i].d = t[i];
 397
 398         if (info->prev) {
 399                 strbuf_grow(&base, info->pathlen);
 400                 make_traverse_path(base.buf, info->prev, &info->name);
 401                 base.buf[info->pathlen-1] = '/';
 402                 strbuf_setlen(&base, info->pathlen);
 403                 traverse_path = xstrndup(base.buf, info->pathlen);
 404         } else {
 405                 traverse_path = xstrndup(info->name.path, info->pathlen);
 406         }
 407         info->traverse_path = traverse_path;
 408         for (;;) {
 409                 int trees_used;
 410                 unsigned long mask, dirmask;
 411                 const char *first = NULL;
 412                 int first_len = 0;
 413                 struct name_entry *e = NULL;
 414                 int len;
 415
 416                 for (i = 0; i < n; i++) {
 417                         e = entry + i;
 418                         extended_entry_extract(tx + i, e, NULL, 0);
 419                 }
 420
 421                 /*
 422                  * A tree may have "t-2" at the current location even
 423                  * though it may have "t" that is a subtree behind it,
 424                  * and another tree may return "t".  We want to grab
 425                  * all "t" from all trees to match in such a case.
 426                  */
 427                 for (i = 0; i < n; i++) {
 428                         e = entry + i;
 429                         if (!e->path)
 430                                 continue;
 431                         len = tree_entry_len(e);
 432                         if (!first) {
 433                                 first = e->path;
 434                                 first_len = len;
 435                                 continue;
 436                         }
 437                         if (name_compare(e->path, len, first, first_len) < 0) {
 438                                 first = e->path;
 439                                 first_len = len;
 440                         }
 441                 }
 442
 443                 if (first) {
 444                         for (i = 0; i < n; i++) {
 445                                 e = entry + i;
 446                                 extended_entry_extract(tx + i, e, first, first_len);
 447                                 /* Cull the ones that are not the earliest */
 448                                 if (!e->path)
 449                                         continue;
 450                                 len = tree_entry_len(e);
 451                                 if (name_compare(e->path, len, first, first_len))
 452                                         entry_clear(e);
 453                         }
 454                 }
 455
 456                 /* Now we have in entry[i] the earliest name from the trees */
 457                 mask = 0;
 458                 dirmask = 0;
 459                 for (i = 0; i < n; i++) {
 460                         if (!entry[i].path)
 461                                 continue;
 462                         mask |= 1ul << i;
 463                         if (S_ISDIR(entry[i].mode))
 464                                 dirmask |= 1ul << i;
 465                         e = &entry[i];
 466                 }
 467                 if (!mask)
 468                         break;
 469                 interesting = prune_traversal(e, info, &base, interesting);
 470                 if (interesting < 0)
 471                         break;
 472                 if (interesting) {
 473                         trees_used = info->fn(n, mask, dirmask, entry, info);
 474                         if (trees_used < 0) {
 475                                 error = trees_used;
 476                                 if (!info->show_all_errors)
 477                                         break;
 478                         }
 479                         mask &= trees_used;
 480                 }
 481                 for (i = 0; i < n; i++)
 482                         if (mask & (1ul << i))
 483                                 update_extended_entry(tx + i, entry + i);
 484         }
 485         free(entry);
 486         for (i = 0; i < n; i++)
 487                 free_extended_entry(tx + i);
 488         free(tx);
 489         free(traverse_path);
 490         info->traverse_path = NULL;
 491         strbuf_release(&base);
 492         return error;
 493 }
 494
 495 struct dir_state {
 496         void *tree;
 497         unsigned long size;
 498         unsigned char sha1[20];
 499 };
 500
 501 static int find_tree_entry(struct tree_desc *t, const char *name, unsigned char *result, unsigned *mode)
 502 {
 503         int namelen = strlen(name);
 504         while (t->size) {
 505                 const char *entry;
 506                 const struct object_id *oid;
 507                 int entrylen, cmp;
 508
 509                 oid = tree_entry_extract(t, &entry, mode);
 510                 entrylen = tree_entry_len(&t->entry);
 511                 update_tree_entry(t);
 512                 if (entrylen > namelen)
 513                         continue;
 514                 cmp = memcmp(name, entry, entrylen);
 515                 if (cmp > 0)
 516                         continue;
 517                 if (cmp < 0)
 518                         break;
 519                 if (entrylen == namelen) {
 520                         hashcpy(result, oid->hash);
 521                         return 0;
 522                 }
 523                 if (name[entrylen] != '/')
 524                         continue;
 525                 if (!S_ISDIR(*mode))
 526                         break;
 527                 if (++entrylen == namelen) {
 528                         hashcpy(result, oid->hash);
 529                         return 0;
 530                 }
 531                 return get_tree_entry(oid->hash, name + entrylen, result, mode);
 532         }
 533         return -1;
 534 }
 535
 536 int get_tree_entry(const unsigned char *tree_sha1, const char *name, unsigned char *sha1, unsigned *mode)
 537 {
 538         int retval;
 539         void *tree;
 540         unsigned long size;
 541         unsigned char root[20];
 542
 543         tree = read_object_with_reference(tree_sha1, tree_type, &size, root);
 544         if (!tree)
 545                 return -1;
 546
 547         if (name[0] == '\0') {
 548                 hashcpy(sha1, root);
 549                 free(tree);
 550                 return 0;
 551         }
 552
 553         if (!size) {
 554                 retval = -1;
 555         } else {
 556                 struct tree_desc t;
 557                 init_tree_desc(&t, tree, size);
 558                 retval = find_tree_entry(&t, name, sha1, mode);
 559         }
 560         free(tree);
 561         return retval;
 562 }
 563
 564 /*
 565  * This is Linux's built-in max for the number of symlinks to follow.
 566  * That limit, of course, does not affect git, but it's a reasonable
 567  * choice.
 568  */
 569 #define GET_TREE_ENTRY_FOLLOW_SYMLINKS_MAX_LINKS 40
 570
 571 /**
 572  * Find a tree entry by following symlinks in tree_sha (which is
 573  * assumed to be the root of the repository).  In the event that a
 574  * symlink points outside the repository (e.g. a link to /foo or a
 575  * root-level link to ../foo), the portion of the link which is
 576  * outside the repository will be returned in result_path, and *mode
 577  * will be set to 0.  It is assumed that result_path is uninitialized.
 578  * If there are no symlinks, or the end result of the symlink chain
 579  * points to an object inside the repository, result will be filled in
 580  * with the sha1 of the found object, and *mode will hold the mode of
 581  * the object.
 582  *
 583  * See the code for enum follow_symlink_result for a description of
 584  * the return values.
 585  */
 586 enum follow_symlinks_result get_tree_entry_follow_symlinks(unsigned char *tree_sha1, const char *name, unsigned char *result, struct strbuf *result_path, unsigned *mode)
 587 {
 588         int retval = MISSING_OBJECT;
 589         struct dir_state *parents = NULL;
 590         size_t parents_alloc = 0;
 591         size_t i, parents_nr = 0;
 592         unsigned char current_tree_sha1[20];
 593         struct strbuf namebuf = STRBUF_INIT;
 594         struct tree_desc t;
 595         int follows_remaining = GET_TREE_ENTRY_FOLLOW_SYMLINKS_MAX_LINKS;
 596
 597         init_tree_desc(&t, NULL, 0UL);
 598         strbuf_addstr(&namebuf, name);
 599         hashcpy(current_tree_sha1, tree_sha1);
 600
 601         while (1) {
 602                 int find_result;
 603                 char *first_slash;
 604                 char *remainder = NULL;
 605
 606                 if (!t.buffer) {
 607                         void *tree;
 608                         unsigned char root[20];
 609                         unsigned long size;
 610                         tree = read_object_with_reference(current_tree_sha1,
 611                                                           tree_type, &size,
 612                                                           root);
 613                         if (!tree)
 614                                 goto done;
 615
 616                         ALLOC_GROW(parents, parents_nr + 1, parents_alloc);
 617                         parents[parents_nr].tree = tree;
 618                         parents[parents_nr].size = size;
 619                         hashcpy(parents[parents_nr].sha1, root);
 620                         parents_nr++;
 621
 622                         if (namebuf.buf[0] == '\0') {
 623                                 hashcpy(result, root);
 624                                 retval = FOUND;
 625                                 goto done;
 626                         }
 627
 628                         if (!size)
 629                                 goto done;
 630
 631                         /* descend */
 632                         init_tree_desc(&t, tree, size);
 633                 }
 634
 635                 /* Handle symlinks to e.g. a//b by removing leading slashes */
 636                 while (namebuf.buf[0] == '/') {
 637                         strbuf_remove(&namebuf, 0, 1);
 638                 }
 639
 640                 /* Split namebuf into a first component and a remainder */
 641                 if ((first_slash = strchr(namebuf.buf, '/'))) {
 642                         *first_slash = 0;
 643                         remainder = first_slash + 1;
 644                 }
 645
 646                 if (!strcmp(namebuf.buf, "..")) {
 647                         struct dir_state *parent;
 648                         /*
 649                          * We could end up with .. in the namebuf if it
 650                          * appears in a symlink.
 651                          */
 652
 653                         if (parents_nr == 1) {
 654                                 if (remainder)
 655                                         *first_slash = '/';
 656                                 strbuf_add(result_path, namebuf.buf,
 657                                            namebuf.len);
 658                                 *mode = 0;
 659                                 retval = FOUND;
 660                                 goto done;
 661                         }
 662                         parent = &parents[parents_nr - 1];
 663                         free(parent->tree);
 664                         parents_nr--;
 665                         parent = &parents[parents_nr - 1];
 666                         init_tree_desc(&t, parent->tree, parent->size);
 667                         strbuf_remove(&namebuf, 0, remainder ? 3 : 2);
 668                         continue;
 669                 }
 670
 671                 /* We could end up here via a symlink to dir/.. */
 672                 if (namebuf.buf[0] == '\0') {
 673                         hashcpy(result, parents[parents_nr - 1].sha1);
 674                         retval = FOUND;
 675                         goto done;
 676                 }
 677
 678                 /* Look up the first (or only) path component in the tree. */
 679                 find_result = find_tree_entry(&t, namebuf.buf,
 680                                               current_tree_sha1, mode);
 681                 if (find_result) {
 682                         goto done;
 683                 }
 684
 685                 if (S_ISDIR(*mode)) {
 686                         if (!remainder) {
 687                                 hashcpy(result, current_tree_sha1);
 688                                 retval = FOUND;
 689                                 goto done;
 690                         }
 691                         /* Descend the tree */
 692                         t.buffer = NULL;
 693                         strbuf_remove(&namebuf, 0,
 694                                       1 + first_slash - namebuf.buf);
 695                 } else if (S_ISREG(*mode)) {
 696                         if (!remainder) {
 697                                 hashcpy(result, current_tree_sha1);
 698                                 retval = FOUND;
 699                         } else {
 700                                 retval = NOT_DIR;
 701                         }
 702                         goto done;
 703                 } else if (S_ISLNK(*mode)) {
 704                         /* Follow a symlink */
 705                         unsigned long link_len;
 706                         size_t len;
 707                         char *contents, *contents_start;
 708                         struct dir_state *parent;
 709                         enum object_type type;
 710
 711                         if (follows_remaining-- == 0) {
 712                                 /* Too many symlinks followed */
 713                                 retval = SYMLINK_LOOP;
 714                                 goto done;
 715                         }
 716
 717                         /*
 718                          * At this point, we have followed at a least
 719                          * one symlink, so on error we need to report this.
 720                          */
 721                         retval = DANGLING_SYMLINK;
 722
 723                         contents = read_sha1_file(current_tree_sha1, &type,
 724                                                   &link_len);
 725
 726                         if (!contents)
 727                                 goto done;
 728
 729                         if (contents[0] == '/') {
 730                                 strbuf_addstr(result_path, contents);
 731                                 free(contents);
 732                                 *mode = 0;
 733                                 retval = FOUND;
 734                                 goto done;
 735                         }
 736
 737                         if (remainder)
 738                                 len = first_slash - namebuf.buf;
 739                         else
 740                                 len = namebuf.len;
 741
 742                         contents_start = contents;
 743
 744                         parent = &parents[parents_nr - 1];
 745                         init_tree_desc(&t, parent->tree, parent->size);
 746                         strbuf_splice(&namebuf, 0, len,
 747                                       contents_start, link_len);
 748                         if (remainder)
 749                                 namebuf.buf[link_len] = '/';
 750                         free(contents);
 751                 }
 752         }
 753 done:
 754         for (i = 0; i < parents_nr; i++)
 755                 free(parents[i].tree);
 756         free(parents);
 757
 758         strbuf_release(&namebuf);
 759         return retval;
 760 }
 761
 762 static int match_entry(const struct pathspec_item *item,
 763                        const struct name_entry *entry, int pathlen,
 764                        const char *match, int matchlen,
 765                        enum interesting *never_interesting)
 766 {
 767         int m = -1; /* signals that we haven't called strncmp() */
 768
 769         if (item->magic & PATHSPEC_ICASE)
 770                 /*
 771                  * "Never interesting" trick requires exact
 772                  * matching. We could do something clever with inexact
 773                  * matching, but it's trickier (and not to forget that
 774                  * strcasecmp is locale-dependent, at least in
 775                  * glibc). Just disable it for now. It can't be worse
 776                  * than the wildcard's codepath of '[Tt][Hi][Is][Ss]'
 777                  * pattern.
 778                  */
 779                 *never_interesting = entry_not_interesting;
 780         else if (*never_interesting != entry_not_interesting) {
 781                 /*
 782                  * We have not seen any match that sorts later
 783                  * than the current path.
 784                  */
 785
 786                 /*
 787                  * Does match sort strictly earlier than path
 788                  * with their common parts?
 789                  */
 790                 m = strncmp(match, entry->path,
 791                             (matchlen < pathlen) ? matchlen : pathlen);
 792                 if (m < 0)
 793                         return 0;
 794
 795                 /*
 796                  * If we come here even once, that means there is at
 797                  * least one pathspec that would sort equal to or
 798                  * later than the path we are currently looking at.
 799                  * In other words, if we have never reached this point
 800                  * after iterating all pathspecs, it means all
 801                  * pathspecs are either outside of base, or inside the
 802                  * base but sorts strictly earlier than the current
 803                  * one.  In either case, they will never match the
 804                  * subsequent entries.  In such a case, we initialized
 805                  * the variable to -1 and that is what will be
 806                  * returned, allowing the caller to terminate early.
 807                  */
 808                 *never_interesting = entry_not_interesting;
 809         }
 810
 811         if (pathlen > matchlen)
 812                 return 0;
 813
 814         if (matchlen > pathlen) {
 815                 if (match[pathlen] != '/')
 816                         return 0;
 817                 if (!S_ISDIR(entry->mode) && !S_ISGITLINK(entry->mode))
 818                         return 0;
 819         }
 820
 821         if (m == -1)
 822                 /*
 823                  * we cheated and did not do strncmp(), so we do
 824                  * that here.
 825                  */
 826                 m = ps_strncmp(item, match, entry->path, pathlen);
 827
 828         /*
 829          * If common part matched earlier then it is a hit,
 830          * because we rejected the case where path is not a
 831          * leading directory and is shorter than match.
 832          */
 833         if (!m)
 834                 /*
 835                  * match_entry does not check if the prefix part is
 836                  * matched case-sensitively. If the entry is a
 837                  * directory and part of prefix, it'll be rematched
 838                  * eventually by basecmp with special treatment for
 839                  * the prefix.
 840                  */
 841                 return 1;
 842
 843         return 0;
 844 }
 845
 846 /* :(icase)-aware string compare */
 847 static int basecmp(const struct pathspec_item *item,
 848                    const char *base, const char *match, int len)
 849 {
 850         if (item->magic & PATHSPEC_ICASE) {
 851                 int ret, n = len > item->prefix ? item->prefix : len;
 852                 ret = strncmp(base, match, n);
 853                 if (ret)
 854                         return ret;
 855                 base += n;
 856                 match += n;
 857                 len -= n;
 858         }
 859         return ps_strncmp(item, base, match, len);
 860 }
 861
 862 static int match_dir_prefix(const struct pathspec_item *item,
 863                             const char *base,
 864                             const char *match, int matchlen)
 865 {
 866         if (basecmp(item, base, match, matchlen))
 867                 return 0;
 868
 869         /*
 870          * If the base is a subdirectory of a path which
 871          * was specified, all of them are interesting.
 872          */
 873         if (!matchlen ||
 874             base[matchlen] == '/' ||
 875             match[matchlen - 1] == '/')
 876                 return 1;
 877
 878         /* Just a random prefix match */
 879         return 0;
 880 }
 881
 882 /*
 883  * Perform matching on the leading non-wildcard part of
 884  * pathspec. item->nowildcard_len must be greater than zero. Return
 885  * non-zero if base is matched.
 886  */
 887 static int match_wildcard_base(const struct pathspec_item *item,
 888                                const char *base, int baselen,
 889                                int *matched)
 890 {
 891         const char *match = item->match;
 892         /* the wildcard part is not considered in this function */
 893         int matchlen = item->nowildcard_len;
 894
 895         if (baselen) {
 896                 int dirlen;
 897                 /*
 898                  * Return early if base is longer than the
 899                  * non-wildcard part but it does not match.
 900                  */
 901                 if (baselen >= matchlen) {
 902                         *matched = matchlen;
 903                         return !basecmp(item, base, match, matchlen);
 904                 }
 905
 906                 dirlen = matchlen;
 907                 while (dirlen && match[dirlen - 1] != '/')
 908                         dirlen--;
 909
 910                 /*
 911                  * Return early if base is shorter than the
 912                  * non-wildcard part but it does not match. Note that
 913                  * base ends with '/' so we are sure it really matches
 914                  * directory
 915                  */
 916                 if (basecmp(item, base, match, baselen))
 917                         return 0;
 918                 *matched = baselen;
 919         } else
 920                 *matched = 0;
 921         /*
 922          * we could have checked entry against the non-wildcard part
 923          * that is not in base and does similar never_interesting
 924          * optimization as in match_entry. For now just be happy with
 925          * base comparison.
 926          */
 927         return entry_interesting;
 928 }
 929
 930 /*
 931  * Is a tree entry interesting given the pathspec we have?
 932  *
 933  * Pre-condition: either baselen == base_offset (i.e. empty path)
 934  * or base[baselen-1] == '/' (i.e. with trailing slash).
 935  */
 936 static enum interesting do_match(const struct name_entry *entry,
 937                                  struct strbuf *base, int base_offset,
 938                                  const struct pathspec *ps,
 939                                  int exclude)
 940 {
 941         int i;
 942         int pathlen, baselen = base->len - base_offset;
 943         enum interesting never_interesting = ps->has_wildcard ?
 944                 entry_not_interesting : all_entries_not_interesting;
 945
 946         GUARD_PATHSPEC(ps,
 947                        PATHSPEC_FROMTOP |
 948                        PATHSPEC_MAXDEPTH |
 949                        PATHSPEC_LITERAL |
 950                        PATHSPEC_GLOB |
 951                        PATHSPEC_ICASE |
 952                        PATHSPEC_EXCLUDE);
 953
 954         if (!ps->nr) {
 955                 if (!ps->recursive ||
 956                     !(ps->magic & PATHSPEC_MAXDEPTH) ||
 957                     ps->max_depth == -1)
 958                         return all_entries_interesting;
 959                 return within_depth(base->buf + base_offset, baselen,
 960                                     !!S_ISDIR(entry->mode),
 961                                     ps->max_depth) ?
 962                         entry_interesting : entry_not_interesting;
 963         }
 964
 965         pathlen = tree_entry_len(entry);
 966
 967         for (i = ps->nr - 1; i >= 0; i--) {
 968                 const struct pathspec_item *item = ps->items+i;
 969                 const char *match = item->match;
 970                 const char *base_str = base->buf + base_offset;
 971                 int matchlen = item->len, matched = 0;
 972
 973                 if ((!exclude &&   item->magic & PATHSPEC_EXCLUDE) ||
 974                     ( exclude && !(item->magic & PATHSPEC_EXCLUDE)))
 975                         continue;
 976
 977                 if (baselen >= matchlen) {
 978                         /* If it doesn't match, move along... */
 979                         if (!match_dir_prefix(item, base_str, match, matchlen))
 980                                 goto match_wildcards;
 981
 982                         if (!ps->recursive ||
 983                             !(ps->magic & PATHSPEC_MAXDEPTH) ||
 984                             ps->max_depth == -1)
 985                                 return all_entries_interesting;
 986
 987                         return within_depth(base_str + matchlen + 1,
 988                                             baselen - matchlen - 1,
 989                                             !!S_ISDIR(entry->mode),
 990                                             ps->max_depth) ?
 991                                 entry_interesting : entry_not_interesting;
 992                 }
 993
 994                 /* Either there must be no base, or the base must match. */
 995                 if (baselen == 0 || !basecmp(item, base_str, match, baselen)) {
 996                         if (match_entry(item, entry, pathlen,
 997                                         match + baselen, matchlen - baselen,
 998                                         &never_interesting))
 999                                 return entry_interesting;
1000
1001                         if (item->nowildcard_len < item->len) {
1002                                 if (!git_fnmatch(item, match + baselen, entry->path,
1003                                                  item->nowildcard_len - baselen))
1004                                         return entry_interesting;
1005
1006                                 /*
1007                                  * Match all directories. We'll try to
1008                                  * match files later on.
1009                                  */
1010                                 if (ps->recursive && S_ISDIR(entry->mode))
1011                                         return entry_interesting;
1012
1013                                 /*
1014                                  * When matching against submodules with
1015                                  * wildcard characters, ensure that the entry
1016                                  * at least matches up to the first wild
1017                                  * character.  More accurate matching can then
1018                                  * be performed in the submodule itself.
1019                                  */
1020                                 if (ps->recursive && S_ISGITLINK(entry->mode) &&
1021                                     !ps_strncmp(item, match + baselen,
1022                                                 entry->path,
1023                                                 item->nowildcard_len - baselen))
1024                                         return entry_interesting;
1025                         }
1026
1027                         continue;
1028                 }
1029
1030 match_wildcards:
1031                 if (item->nowildcard_len == item->len)
1032                         continue;
1033
1034                 if (item->nowildcard_len &&
1035                     !match_wildcard_base(item, base_str, baselen, &matched))
1036                         continue;
1037
1038                 /*
1039                  * Concatenate base and entry->path into one and do
1040                  * fnmatch() on it.
1041                  *
1042                  * While we could avoid concatenation in certain cases
1043                  * [1], which saves a memcpy and potentially a
1044                  * realloc, it turns out not worth it. Measurement on
1045                  * linux-2.6 does not show any clear improvements,
1046                  * partly because of the nowildcard_len optimization
1047                  * in git_fnmatch(). Avoid micro-optimizations here.
1048                  *
1049                  * [1] if match_wildcard_base() says the base
1050                  * directory is already matched, we only need to match
1051                  * the rest, which is shorter so _in theory_ faster.
1052                  */
1053
1054                 strbuf_add(base, entry->path, pathlen);
1055
1056                 if (!git_fnmatch(item, match, base->buf + base_offset,
1057                                  item->nowildcard_len)) {
1058                         strbuf_setlen(base, base_offset + baselen);
1059                         return entry_interesting;
1060                 }
1061
1062                 /*
1063                  * When matching against submodules with
1064                  * wildcard characters, ensure that the entry
1065                  * at least matches up to the first wild
1066                  * character.  More accurate matching can then
1067                  * be performed in the submodule itself.
1068                  */
1069                 if (ps->recursive && S_ISGITLINK(entry->mode) &&
1070                     !ps_strncmp(item, match, base->buf + base_offset,
1071                                 item->nowildcard_len)) {
1072                         strbuf_setlen(base, base_offset + baselen);
1073                         return entry_interesting;
1074                 }
1075
1076                 strbuf_setlen(base, base_offset + baselen);
1077
1078                 /*
1079                  * Match all directories. We'll try to match files
1080                  * later on.
1081                  * max_depth is ignored but we may consider support it
1082                  * in future, see
1083                  * https://public-inbox.org/git/7vmxo5l2g4.fsf@alter.siamese.dyndns.org/
1084                  */
1085                 if (ps->recursive && S_ISDIR(entry->mode))
1086                         return entry_interesting;
1087         }
1088         return never_interesting; /* No matches */
1089 }
1090
1091 /*
1092  * Is a tree entry interesting given the pathspec we have?
1093  *
1094  * Pre-condition: either baselen == base_offset (i.e. empty path)
1095  * or base[baselen-1] == '/' (i.e. with trailing slash).
1096  */
1097 enum interesting tree_entry_interesting(const struct name_entry *entry,
1098                                         struct strbuf *base, int base_offset,
1099                                         const struct pathspec *ps)
1100 {
1101         enum interesting positive, negative;
1102         positive = do_match(entry, base, base_offset, ps, 0);
1103
1104         /*
1105          * case | entry | positive | negative | result
1106          * -----+-------+----------+----------+-------
1107          *   1  |  file |   -1     |  -1..2   |  -1
1108          *   2  |  file |    0     |  -1..2   |   0
1109          *   3  |  file |    1     |   -1     |   1
1110          *   4  |  file |    1     |    0     |   1
1111          *   5  |  file |    1     |    1     |   0
1112          *   6  |  file |    1     |    2     |   0
1113          *   7  |  file |    2     |   -1     |   2
1114          *   8  |  file |    2     |    0     |   2
1115          *   9  |  file |    2     |    1     |   0
1116          *  10  |  file |    2     |    2     |  -1
1117          * -----+-------+----------+----------+-------
1118          *  11  |  dir  |   -1     |  -1..2   |  -1
1119          *  12  |  dir  |    0     |  -1..2   |   0
1120          *  13  |  dir  |    1     |   -1     |   1
1121          *  14  |  dir  |    1     |    0     |   1
1122          *  15  |  dir  |    1     |    1     |   1 (*)
1123          *  16  |  dir  |    1     |    2     |   0
1124          *  17  |  dir  |    2     |   -1     |   2
1125          *  18  |  dir  |    2     |    0     |   2
1126          *  19  |  dir  |    2     |    1     |   1 (*)
1127          *  20  |  dir  |    2     |    2     |  -1
1128          *
1129          * (*) An exclude pattern interested in a directory does not
1130          * necessarily mean it will exclude all of the directory. In
1131          * wildcard case, it can't decide until looking at individual
1132          * files inside. So don't write such directories off yet.
1133          */
1134
1135         if (!(ps->magic & PATHSPEC_EXCLUDE) ||
1136             positive <= entry_not_interesting) /* #1, #2, #11, #12 */
1137                 return positive;
1138
1139         negative = do_match(entry, base, base_offset, ps, 1);
1140
1141         /* #3, #4, #7, #8, #13, #14, #17, #18 */
1142         if (negative <= entry_not_interesting)
1143                 return positive;
1144
1145         /* #15, #19 */
1146         if (S_ISDIR(entry->mode) &&
1147             positive >= entry_interesting &&
1148             negative == entry_interesting)
1149                 return entry_interesting;
1150
1151         if ((positive == entry_interesting &&
1152              negative >= entry_interesting) || /* #5, #6, #16 */
1153             (positive == all_entries_interesting &&
1154              negative == entry_interesting)) /* #9 */
1155                 return entry_not_interesting;
1156
1157         return all_entries_not_interesting; /* #10, #20 */
1158 }