tree-walk.c

   1 #include "cache.h"
   2 #include "tree-walk.h"
   3 #include "unpack-trees.h"
   4 #include "dir.h"
   5 #include "object-store.h"
   6 #include "tree.h"
   7 #include "pathspec.h"
   8
   9 static const char *get_mode(const char *str, unsigned int *modep)
  10 {
  11         unsigned char c;
  12         unsigned int mode = 0;
  13
  14         if (*str == ' ')
  15                 return NULL;
  16
  17         while ((c = *str++) != ' ') {
  18                 if (c < '0' || c > '7')
  19                         return NULL;
  20                 mode = (mode << 3) + (c - '0');
  21         }
  22         *modep = mode;
  23         return str;
  24 }
  25
  26 static int decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size, struct strbuf *err)
  27 {
  28         const char *path;
  29         unsigned int mode, len;
  30         const unsigned hashsz = the_hash_algo->rawsz;
  31
  32         if (size < hashsz + 3 || buf[size - (hashsz + 1)]) {
  33                 strbuf_addstr(err, _("too-short tree object"));
  34                 return -1;
  35         }
  36
  37         path = get_mode(buf, &mode);
  38         if (!path) {
  39                 strbuf_addstr(err, _("malformed mode in tree entry"));
  40                 return -1;
  41         }
  42         if (!*path) {
  43                 strbuf_addstr(err, _("empty filename in tree entry"));
  44                 return -1;
  45         }
  46 #ifdef GIT_WINDOWS_NATIVE
  47         if (protect_ntfs && strchr(path, '\\')) {
  48                 strbuf_addf(err, _("filename in tree entry contains backslash: '%s'"), path);
  49                 return -1;
  50         }
  51 #endif
  52         len = strlen(path) + 1;
  53
  54         /* Initialize the descriptor entry */
  55         desc->entry.path = path;
  56         desc->entry.mode = canon_mode(mode);
  57         desc->entry.oid  = (const struct object_id *)(path + len);
  58
  59         return 0;
  60 }
  61
  62 static int init_tree_desc_internal(struct tree_desc *desc, const void *buffer, unsigned long size, struct strbuf *err)
  63 {
  64         desc->buffer = buffer;
  65         desc->size = size;
  66         if (size)
  67                 return decode_tree_entry(desc, buffer, size, err);
  68         return 0;
  69 }
  70
  71 void init_tree_desc(struct tree_desc *desc, const void *buffer, unsigned long size)
  72 {
  73         struct strbuf err = STRBUF_INIT;
  74         if (init_tree_desc_internal(desc, buffer, size, &err))
  75                 die("%s", err.buf);
  76         strbuf_release(&err);
  77 }
  78
  79 int init_tree_desc_gently(struct tree_desc *desc, const void *buffer, unsigned long size)
  80 {
  81         struct strbuf err = STRBUF_INIT;
  82         int result = init_tree_desc_internal(desc, buffer, size, &err);
  83         if (result)
  84                 error("%s", err.buf);
  85         strbuf_release(&err);
  86         return result;
  87 }
  88
  89 void *fill_tree_descriptor(struct tree_desc *desc, const struct object_id *oid)
  90 {
  91         unsigned long size = 0;
  92         void *buf = NULL;
  93
  94         if (oid) {
  95                 buf = read_object_with_reference(oid, tree_type, &size, NULL);
  96                 if (!buf)
  97                         die("unable to read tree %s", oid_to_hex(oid));
  98         }
  99         init_tree_desc(desc, buf, size);
 100         return buf;
 101 }
 102
 103 static void entry_clear(struct name_entry *a)
 104 {
 105         memset(a, 0, sizeof(*a));
 106 }
 107
 108 static void entry_extract(struct tree_desc *t, struct name_entry *a)
 109 {
 110         *a = t->entry;
 111 }
 112
 113 static int update_tree_entry_internal(struct tree_desc *desc, struct strbuf *err)
 114 {
 115         const void *buf = desc->buffer;
 116         const unsigned char *end = desc->entry.oid->hash + the_hash_algo->rawsz;
 117         unsigned long size = desc->size;
 118         unsigned long len = end - (const unsigned char *)buf;
 119
 120         if (size < len)
 121                 die(_("too-short tree file"));
 122         buf = end;
 123         size -= len;
 124         desc->buffer = buf;
 125         desc->size = size;
 126         if (size)
 127                 return decode_tree_entry(desc, buf, size, err);
 128         return 0;
 129 }
 130
 131 void update_tree_entry(struct tree_desc *desc)
 132 {
 133         struct strbuf err = STRBUF_INIT;
 134         if (update_tree_entry_internal(desc, &err))
 135                 die("%s", err.buf);
 136         strbuf_release(&err);
 137 }
 138
 139 int update_tree_entry_gently(struct tree_desc *desc)
 140 {
 141         struct strbuf err = STRBUF_INIT;
 142         if (update_tree_entry_internal(desc, &err)) {
 143                 error("%s", err.buf);
 144                 strbuf_release(&err);
 145                 /* Stop processing this tree after error */
 146                 desc->size = 0;
 147                 return -1;
 148         }
 149         strbuf_release(&err);
 150         return 0;
 151 }
 152
 153 int tree_entry(struct tree_desc *desc, struct name_entry *entry)
 154 {
 155         if (!desc->size)
 156                 return 0;
 157
 158         *entry = desc->entry;
 159         update_tree_entry(desc);
 160         return 1;
 161 }
 162
 163 int tree_entry_gently(struct tree_desc *desc, struct name_entry *entry)
 164 {
 165         if (!desc->size)
 166                 return 0;
 167
 168         *entry = desc->entry;
 169         if (update_tree_entry_gently(desc))
 170                 return 0;
 171         return 1;
 172 }
 173
 174 void setup_traverse_info(struct traverse_info *info, const char *base)
 175 {
 176         int pathlen = strlen(base);
 177         static struct traverse_info dummy;
 178
 179         memset(info, 0, sizeof(*info));
 180         if (pathlen && base[pathlen-1] == '/')
 181                 pathlen--;
 182         info->pathlen = pathlen ? pathlen + 1 : 0;
 183         info->name.path = base;
 184         info->name.oid = (void *)(base + pathlen + 1);
 185         if (pathlen)
 186                 info->prev = &dummy;
 187 }
 188
 189 char *make_traverse_path(char *path, const struct traverse_info *info, const struct name_entry *n)
 190 {
 191         int len = tree_entry_len(n);
 192         int pathlen = info->pathlen;
 193
 194         path[pathlen + len] = 0;
 195         for (;;) {
 196                 memcpy(path + pathlen, n->path, len);
 197                 if (!pathlen)
 198                         break;
 199                 path[--pathlen] = '/';
 200                 n = &info->name;
 201                 len = tree_entry_len(n);
 202                 info = info->prev;
 203                 pathlen -= len;
 204         }
 205         return path;
 206 }
 207
 208 struct tree_desc_skip {
 209         struct tree_desc_skip *prev;
 210         const void *ptr;
 211 };
 212
 213 struct tree_desc_x {
 214         struct tree_desc d;
 215         struct tree_desc_skip *skip;
 216 };
 217
 218 static int check_entry_match(const char *a, int a_len, const char *b, int b_len)
 219 {
 220         /*
 221          * The caller wants to pick *a* from a tree or nothing.
 222          * We are looking at *b* in a tree.
 223          *
 224          * (0) If a and b are the same name, we are trivially happy.
 225          *
 226          * There are three possibilities where *a* could be hiding
 227          * behind *b*.
 228          *
 229          * (1) *a* == "t",   *b* == "ab"  i.e. *b* sorts earlier than *a* no
 230          *                                matter what.
 231          * (2) *a* == "t",   *b* == "t-2" and "t" is a subtree in the tree;
 232          * (3) *a* == "t-2", *b* == "t"   and "t-2" is a blob in the tree.
 233          *
 234          * Otherwise we know *a* won't appear in the tree without
 235          * scanning further.
 236          */
 237
 238         int cmp = name_compare(a, a_len, b, b_len);
 239
 240         /* Most common case first -- reading sync'd trees */
 241         if (!cmp)
 242                 return cmp;
 243
 244         if (0 < cmp) {
 245                 /* a comes after b; it does not matter if it is case (3)
 246                 if (b_len < a_len && !memcmp(a, b, b_len) && a[b_len] < '/')
 247                         return 1;
 248                 */
 249                 return 1; /* keep looking */
 250         }
 251
 252         /* b comes after a; are we looking at case (2)? */
 253         if (a_len < b_len && !memcmp(a, b, a_len) && b[a_len] < '/')
 254                 return 1; /* keep looking */
 255
 256         return -1; /* a cannot appear in the tree */
 257 }
 258
 259 /*
 260  * From the extended tree_desc, extract the first name entry, while
 261  * paying attention to the candidate "first" name.  Most importantly,
 262  * when looking for an entry, if there are entries that sorts earlier
 263  * in the tree object representation than that name, skip them and
 264  * process the named entry first.  We will remember that we haven't
 265  * processed the first entry yet, and in the later call skip the
 266  * entry we processed early when update_extended_entry() is called.
 267  *
 268  * E.g. if the underlying tree object has these entries:
 269  *
 270  *    blob    "t-1"
 271  *    blob    "t-2"
 272  *    tree    "t"
 273  *    blob    "t=1"
 274  *
 275  * and the "first" asks for "t", remember that we still need to
 276  * process "t-1" and "t-2" but extract "t".  After processing the
 277  * entry "t" from this call, the caller will let us know by calling
 278  * update_extended_entry() that we can remember "t" has been processed
 279  * already.
 280  */
 281
 282 static void extended_entry_extract(struct tree_desc_x *t,
 283                                    struct name_entry *a,
 284                                    const char *first,
 285                                    int first_len)
 286 {
 287         const char *path;
 288         int len;
 289         struct tree_desc probe;
 290         struct tree_desc_skip *skip;
 291
 292         /*
 293          * Extract the first entry from the tree_desc, but skip the
 294          * ones that we already returned in earlier rounds.
 295          */
 296         while (1) {
 297                 if (!t->d.size) {
 298                         entry_clear(a);
 299                         break; /* not found */
 300                 }
 301                 entry_extract(&t->d, a);
 302                 for (skip = t->skip; skip; skip = skip->prev)
 303                         if (a->path == skip->ptr)
 304                                 break; /* found */
 305                 if (!skip)
 306                         break;
 307                 /* We have processed this entry already. */
 308                 update_tree_entry(&t->d);
 309         }
 310
 311         if (!first || !a->path)
 312                 return;
 313
 314         /*
 315          * The caller wants "first" from this tree, or nothing.
 316          */
 317         path = a->path;
 318         len = tree_entry_len(a);
 319         switch (check_entry_match(first, first_len, path, len)) {
 320         case -1:
 321                 entry_clear(a);
 322         case 0:
 323                 return;
 324         default:
 325                 break;
 326         }
 327
 328         /*
 329          * We need to look-ahead -- we suspect that a subtree whose
 330          * name is "first" may be hiding behind the current entry "path".
 331          */
 332         probe = t->d;
 333         while (probe.size) {
 334                 entry_extract(&probe, a);
 335                 path = a->path;
 336                 len = tree_entry_len(a);
 337                 switch (check_entry_match(first, first_len, path, len)) {
 338                 case -1:
 339                         entry_clear(a);
 340                 case 0:
 341                         return;
 342                 default:
 343                         update_tree_entry(&probe);
 344                         break;
 345                 }
 346                 /* keep looking */
 347         }
 348         entry_clear(a);
 349 }
 350
 351 static void update_extended_entry(struct tree_desc_x *t, struct name_entry *a)
 352 {
 353         if (t->d.entry.path == a->path) {
 354                 update_tree_entry(&t->d);
 355         } else {
 356                 /* we have returned this entry early */
 357                 struct tree_desc_skip *skip = xmalloc(sizeof(*skip));
 358                 skip->ptr = a->path;
 359                 skip->prev = t->skip;
 360                 t->skip = skip;
 361         }
 362 }
 363
 364 static void free_extended_entry(struct tree_desc_x *t)
 365 {
 366         struct tree_desc_skip *p, *s;
 367
 368         for (s = t->skip; s; s = p) {
 369                 p = s->prev;
 370                 free(s);
 371         }
 372 }
 373
 374 static inline int prune_traversal(struct name_entry *e,
 375                                   struct traverse_info *info,
 376                                   struct strbuf *base,
 377                                   int still_interesting)
 378 {
 379         if (!info->pathspec || still_interesting == 2)
 380                 return 2;
 381         if (still_interesting < 0)
 382                 return still_interesting;
 383         return tree_entry_interesting(e, base, 0, info->pathspec);
 384 }
 385
 386 int traverse_trees(int n, struct tree_desc *t, struct traverse_info *info)
 387 {
 388         int error = 0;
 389         struct name_entry *entry = xmalloc(n*sizeof(*entry));
 390         int i;
 391         struct tree_desc_x *tx = xcalloc(n, sizeof(*tx));
 392         struct strbuf base = STRBUF_INIT;
 393         int interesting = 1;
 394         char *traverse_path;
 395
 396         for (i = 0; i < n; i++)
 397                 tx[i].d = t[i];
 398
 399         if (info->prev) {
 400                 strbuf_grow(&base, info->pathlen);
 401                 make_traverse_path(base.buf, info->prev, &info->name);
 402                 base.buf[info->pathlen-1] = '/';
 403                 strbuf_setlen(&base, info->pathlen);
 404                 traverse_path = xstrndup(base.buf, info->pathlen);
 405         } else {
 406                 traverse_path = xstrndup(info->name.path, info->pathlen);
 407         }
 408         info->traverse_path = traverse_path;
 409         for (;;) {
 410                 int trees_used;
 411                 unsigned long mask, dirmask;
 412                 const char *first = NULL;
 413                 int first_len = 0;
 414                 struct name_entry *e = NULL;
 415                 int len;
 416
 417                 for (i = 0; i < n; i++) {
 418                         e = entry + i;
 419                         extended_entry_extract(tx + i, e, NULL, 0);
 420                 }
 421
 422                 /*
 423                  * A tree may have "t-2" at the current location even
 424                  * though it may have "t" that is a subtree behind it,
 425                  * and another tree may return "t".  We want to grab
 426                  * all "t" from all trees to match in such a case.
 427                  */
 428                 for (i = 0; i < n; i++) {
 429                         e = entry + i;
 430                         if (!e->path)
 431                                 continue;
 432                         len = tree_entry_len(e);
 433                         if (!first) {
 434                                 first = e->path;
 435                                 first_len = len;
 436                                 continue;
 437                         }
 438                         if (name_compare(e->path, len, first, first_len) < 0) {
 439                                 first = e->path;
 440                                 first_len = len;
 441                         }
 442                 }
 443
 444                 if (first) {
 445                         for (i = 0; i < n; i++) {
 446                                 e = entry + i;
 447                                 extended_entry_extract(tx + i, e, first, first_len);
 448                                 /* Cull the ones that are not the earliest */
 449                                 if (!e->path)
 450                                         continue;
 451                                 len = tree_entry_len(e);
 452                                 if (name_compare(e->path, len, first, first_len))
 453                                         entry_clear(e);
 454                         }
 455                 }
 456
 457                 /* Now we have in entry[i] the earliest name from the trees */
 458                 mask = 0;
 459                 dirmask = 0;
 460                 for (i = 0; i < n; i++) {
 461                         if (!entry[i].path)
 462                                 continue;
 463                         mask |= 1ul << i;
 464                         if (S_ISDIR(entry[i].mode))
 465                                 dirmask |= 1ul << i;
 466                         e = &entry[i];
 467                 }
 468                 if (!mask)
 469                         break;
 470                 interesting = prune_traversal(e, info, &base, interesting);
 471                 if (interesting < 0)
 472                         break;
 473                 if (interesting) {
 474                         trees_used = info->fn(n, mask, dirmask, entry, info);
 475                         if (trees_used < 0) {
 476                                 error = trees_used;
 477                                 if (!info->show_all_errors)
 478                                         break;
 479                         }
 480                         mask &= trees_used;
 481                 }
 482                 for (i = 0; i < n; i++)
 483                         if (mask & (1ul << i))
 484                                 update_extended_entry(tx + i, entry + i);
 485         }
 486         free(entry);
 487         for (i = 0; i < n; i++)
 488                 free_extended_entry(tx + i);
 489         free(tx);
 490         free(traverse_path);
 491         info->traverse_path = NULL;
 492         strbuf_release(&base);
 493         return error;
 494 }
 495
 496 struct dir_state {
 497         void *tree;
 498         unsigned long size;
 499         struct object_id oid;
 500 };
 501
 502 static int find_tree_entry(struct tree_desc *t, const char *name, struct object_id *result, unsigned *mode)
 503 {
 504         int namelen = strlen(name);
 505         while (t->size) {
 506                 const char *entry;
 507                 const struct object_id *oid;
 508                 int entrylen, cmp;
 509
 510                 oid = tree_entry_extract(t, &entry, mode);
 511                 entrylen = tree_entry_len(&t->entry);
 512                 update_tree_entry(t);
 513                 if (entrylen > namelen)
 514                         continue;
 515                 cmp = memcmp(name, entry, entrylen);
 516                 if (cmp > 0)
 517                         continue;
 518                 if (cmp < 0)
 519                         break;
 520                 if (entrylen == namelen) {
 521                         oidcpy(result, oid);
 522                         return 0;
 523                 }
 524                 if (name[entrylen] != '/')
 525                         continue;
 526                 if (!S_ISDIR(*mode))
 527                         break;
 528                 if (++entrylen == namelen) {
 529                         oidcpy(result, oid);
 530                         return 0;
 531                 }
 532                 return get_tree_entry(oid, name + entrylen, result, mode);
 533         }
 534         return -1;
 535 }
 536
 537 int get_tree_entry(const struct object_id *tree_oid, const char *name, struct object_id *oid, unsigned *mode)
 538 {
 539         int retval;
 540         void *tree;
 541         unsigned long size;
 542         struct object_id root;
 543
 544         tree = read_object_with_reference(tree_oid, tree_type, &size, &root);
 545         if (!tree)
 546                 return -1;
 547
 548         if (name[0] == '\0') {
 549                 oidcpy(oid, &root);
 550                 free(tree);
 551                 return 0;
 552         }
 553
 554         if (!size) {
 555                 retval = -1;
 556         } else {
 557                 struct tree_desc t;
 558                 init_tree_desc(&t, tree, size);
 559                 retval = find_tree_entry(&t, name, oid, mode);
 560         }
 561         free(tree);
 562         return retval;
 563 }
 564
 565 /*
 566  * This is Linux's built-in max for the number of symlinks to follow.
 567  * That limit, of course, does not affect git, but it's a reasonable
 568  * choice.
 569  */
 570 #define GET_TREE_ENTRY_FOLLOW_SYMLINKS_MAX_LINKS 40
 571
 572 /**
 573  * Find a tree entry by following symlinks in tree_sha (which is
 574  * assumed to be the root of the repository).  In the event that a
 575  * symlink points outside the repository (e.g. a link to /foo or a
 576  * root-level link to ../foo), the portion of the link which is
 577  * outside the repository will be returned in result_path, and *mode
 578  * will be set to 0.  It is assumed that result_path is uninitialized.
 579  * If there are no symlinks, or the end result of the symlink chain
 580  * points to an object inside the repository, result will be filled in
 581  * with the sha1 of the found object, and *mode will hold the mode of
 582  * the object.
 583  *
 584  * See the code for enum follow_symlink_result for a description of
 585  * the return values.
 586  */
 587 enum follow_symlinks_result get_tree_entry_follow_symlinks(struct object_id *tree_oid, const char *name, struct object_id *result, struct strbuf *result_path, unsigned *mode)
 588 {
 589         int retval = MISSING_OBJECT;
 590         struct dir_state *parents = NULL;
 591         size_t parents_alloc = 0;
 592         size_t i, parents_nr = 0;
 593         struct object_id current_tree_oid;
 594         struct strbuf namebuf = STRBUF_INIT;
 595         struct tree_desc t;
 596         int follows_remaining = GET_TREE_ENTRY_FOLLOW_SYMLINKS_MAX_LINKS;
 597
 598         init_tree_desc(&t, NULL, 0UL);
 599         strbuf_addstr(&namebuf, name);
 600         oidcpy(&current_tree_oid, tree_oid);
 601
 602         while (1) {
 603                 int find_result;
 604                 char *first_slash;
 605                 char *remainder = NULL;
 606
 607                 if (!t.buffer) {
 608                         void *tree;
 609                         struct object_id root;
 610                         unsigned long size;
 611                         tree = read_object_with_reference(&current_tree_oid,
 612                                                           tree_type, &size,
 613                                                           &root);
 614                         if (!tree)
 615                                 goto done;
 616
 617                         ALLOC_GROW(parents, parents_nr + 1, parents_alloc);
 618                         parents[parents_nr].tree = tree;
 619                         parents[parents_nr].size = size;
 620                         oidcpy(&parents[parents_nr].oid, &root);
 621                         parents_nr++;
 622
 623                         if (namebuf.buf[0] == '\0') {
 624                                 oidcpy(result, &root);
 625                                 retval = FOUND;
 626                                 goto done;
 627                         }
 628
 629                         if (!size)
 630                                 goto done;
 631
 632                         /* descend */
 633                         init_tree_desc(&t, tree, size);
 634                 }
 635
 636                 /* Handle symlinks to e.g. a//b by removing leading slashes */
 637                 while (namebuf.buf[0] == '/') {
 638                         strbuf_remove(&namebuf, 0, 1);
 639                 }
 640
 641                 /* Split namebuf into a first component and a remainder */
 642                 if ((first_slash = strchr(namebuf.buf, '/'))) {
 643                         *first_slash = 0;
 644                         remainder = first_slash + 1;
 645                 }
 646
 647                 if (!strcmp(namebuf.buf, "..")) {
 648                         struct dir_state *parent;
 649                         /*
 650                          * We could end up with .. in the namebuf if it
 651                          * appears in a symlink.
 652                          */
 653
 654                         if (parents_nr == 1) {
 655                                 if (remainder)
 656                                         *first_slash = '/';
 657                                 strbuf_add(result_path, namebuf.buf,
 658                                            namebuf.len);
 659                                 *mode = 0;
 660                                 retval = FOUND;
 661                                 goto done;
 662                         }
 663                         parent = &parents[parents_nr - 1];
 664                         free(parent->tree);
 665                         parents_nr--;
 666                         parent = &parents[parents_nr - 1];
 667                         init_tree_desc(&t, parent->tree, parent->size);
 668                         strbuf_remove(&namebuf, 0, remainder ? 3 : 2);
 669                         continue;
 670                 }
 671
 672                 /* We could end up here via a symlink to dir/.. */
 673                 if (namebuf.buf[0] == '\0') {
 674                         oidcpy(result, &parents[parents_nr - 1].oid);
 675                         retval = FOUND;
 676                         goto done;
 677                 }
 678
 679                 /* Look up the first (or only) path component in the tree. */
 680                 find_result = find_tree_entry(&t, namebuf.buf,
 681                                               &current_tree_oid, mode);
 682                 if (find_result) {
 683                         goto done;
 684                 }
 685
 686                 if (S_ISDIR(*mode)) {
 687                         if (!remainder) {
 688                                 oidcpy(result, &current_tree_oid);
 689                                 retval = FOUND;
 690                                 goto done;
 691                         }
 692                         /* Descend the tree */
 693                         t.buffer = NULL;
 694                         strbuf_remove(&namebuf, 0,
 695                                       1 + first_slash - namebuf.buf);
 696                 } else if (S_ISREG(*mode)) {
 697                         if (!remainder) {
 698                                 oidcpy(result, &current_tree_oid);
 699                                 retval = FOUND;
 700                         } else {
 701                                 retval = NOT_DIR;
 702                         }
 703                         goto done;
 704                 } else if (S_ISLNK(*mode)) {
 705                         /* Follow a symlink */
 706                         unsigned long link_len;
 707                         size_t len;
 708                         char *contents, *contents_start;
 709                         struct dir_state *parent;
 710                         enum object_type type;
 711
 712                         if (follows_remaining-- == 0) {
 713                                 /* Too many symlinks followed */
 714                                 retval = SYMLINK_LOOP;
 715                                 goto done;
 716                         }
 717
 718                         /*
 719                          * At this point, we have followed at a least
 720                          * one symlink, so on error we need to report this.
 721                          */
 722                         retval = DANGLING_SYMLINK;
 723
 724                         contents = read_object_file(&current_tree_oid, &type,
 725                                                     &link_len);
 726
 727                         if (!contents)
 728                                 goto done;
 729
 730                         if (contents[0] == '/') {
 731                                 strbuf_addstr(result_path, contents);
 732                                 free(contents);
 733                                 *mode = 0;
 734                                 retval = FOUND;
 735                                 goto done;
 736                         }
 737
 738                         if (remainder)
 739                                 len = first_slash - namebuf.buf;
 740                         else
 741                                 len = namebuf.len;
 742
 743                         contents_start = contents;
 744
 745                         parent = &parents[parents_nr - 1];
 746                         init_tree_desc(&t, parent->tree, parent->size);
 747                         strbuf_splice(&namebuf, 0, len,
 748                                       contents_start, link_len);
 749                         if (remainder)
 750                                 namebuf.buf[link_len] = '/';
 751                         free(contents);
 752                 }
 753         }
 754 done:
 755         for (i = 0; i < parents_nr; i++)
 756                 free(parents[i].tree);
 757         free(parents);
 758
 759         strbuf_release(&namebuf);
 760         return retval;
 761 }
 762
 763 static int match_entry(const struct pathspec_item *item,
 764                        const struct name_entry *entry, int pathlen,
 765                        const char *match, int matchlen,
 766                        enum interesting *never_interesting)
 767 {
 768         int m = -1; /* signals that we haven't called strncmp() */
 769
 770         if (item->magic & PATHSPEC_ICASE)
 771                 /*
 772                  * "Never interesting" trick requires exact
 773                  * matching. We could do something clever with inexact
 774                  * matching, but it's trickier (and not to forget that
 775                  * strcasecmp is locale-dependent, at least in
 776                  * glibc). Just disable it for now. It can't be worse
 777                  * than the wildcard's codepath of '[Tt][Hi][Is][Ss]'
 778                  * pattern.
 779                  */
 780                 *never_interesting = entry_not_interesting;
 781         else if (*never_interesting != entry_not_interesting) {
 782                 /*
 783                  * We have not seen any match that sorts later
 784                  * than the current path.
 785                  */
 786
 787                 /*
 788                  * Does match sort strictly earlier than path
 789                  * with their common parts?
 790                  */
 791                 m = strncmp(match, entry->path,
 792                             (matchlen < pathlen) ? matchlen : pathlen);
 793                 if (m < 0)
 794                         return 0;
 795
 796                 /*
 797                  * If we come here even once, that means there is at
 798                  * least one pathspec that would sort equal to or
 799                  * later than the path we are currently looking at.
 800                  * In other words, if we have never reached this point
 801                  * after iterating all pathspecs, it means all
 802                  * pathspecs are either outside of base, or inside the
 803                  * base but sorts strictly earlier than the current
 804                  * one.  In either case, they will never match the
 805                  * subsequent entries.  In such a case, we initialized
 806                  * the variable to -1 and that is what will be
 807                  * returned, allowing the caller to terminate early.
 808                  */
 809                 *never_interesting = entry_not_interesting;
 810         }
 811
 812         if (pathlen > matchlen)
 813                 return 0;
 814
 815         if (matchlen > pathlen) {
 816                 if (match[pathlen] != '/')
 817                         return 0;
 818                 if (!S_ISDIR(entry->mode) && !S_ISGITLINK(entry->mode))
 819                         return 0;
 820         }
 821
 822         if (m == -1)
 823                 /*
 824                  * we cheated and did not do strncmp(), so we do
 825                  * that here.
 826                  */
 827                 m = ps_strncmp(item, match, entry->path, pathlen);
 828
 829         /*
 830          * If common part matched earlier then it is a hit,
 831          * because we rejected the case where path is not a
 832          * leading directory and is shorter than match.
 833          */
 834         if (!m)
 835                 /*
 836                  * match_entry does not check if the prefix part is
 837                  * matched case-sensitively. If the entry is a
 838                  * directory and part of prefix, it'll be rematched
 839                  * eventually by basecmp with special treatment for
 840                  * the prefix.
 841                  */
 842                 return 1;
 843
 844         return 0;
 845 }
 846
 847 /* :(icase)-aware string compare */
 848 static int basecmp(const struct pathspec_item *item,
 849                    const char *base, const char *match, int len)
 850 {
 851         if (item->magic & PATHSPEC_ICASE) {
 852                 int ret, n = len > item->prefix ? item->prefix : len;
 853                 ret = strncmp(base, match, n);
 854                 if (ret)
 855                         return ret;
 856                 base += n;
 857                 match += n;
 858                 len -= n;
 859         }
 860         return ps_strncmp(item, base, match, len);
 861 }
 862
 863 static int match_dir_prefix(const struct pathspec_item *item,
 864                             const char *base,
 865                             const char *match, int matchlen)
 866 {
 867         if (basecmp(item, base, match, matchlen))
 868                 return 0;
 869
 870         /*
 871          * If the base is a subdirectory of a path which
 872          * was specified, all of them are interesting.
 873          */
 874         if (!matchlen ||
 875             base[matchlen] == '/' ||
 876             match[matchlen - 1] == '/')
 877                 return 1;
 878
 879         /* Just a random prefix match */
 880         return 0;
 881 }
 882
 883 /*
 884  * Perform matching on the leading non-wildcard part of
 885  * pathspec. item->nowildcard_len must be greater than zero. Return
 886  * non-zero if base is matched.
 887  */
 888 static int match_wildcard_base(const struct pathspec_item *item,
 889                                const char *base, int baselen,
 890                                int *matched)
 891 {
 892         const char *match = item->match;
 893         /* the wildcard part is not considered in this function */
 894         int matchlen = item->nowildcard_len;
 895
 896         if (baselen) {
 897                 int dirlen;
 898                 /*
 899                  * Return early if base is longer than the
 900                  * non-wildcard part but it does not match.
 901                  */
 902                 if (baselen >= matchlen) {
 903                         *matched = matchlen;
 904                         return !basecmp(item, base, match, matchlen);
 905                 }
 906
 907                 dirlen = matchlen;
 908                 while (dirlen && match[dirlen - 1] != '/')
 909                         dirlen--;
 910
 911                 /*
 912                  * Return early if base is shorter than the
 913                  * non-wildcard part but it does not match. Note that
 914                  * base ends with '/' so we are sure it really matches
 915                  * directory
 916                  */
 917                 if (basecmp(item, base, match, baselen))
 918                         return 0;
 919                 *matched = baselen;
 920         } else
 921                 *matched = 0;
 922         /*
 923          * we could have checked entry against the non-wildcard part
 924          * that is not in base and does similar never_interesting
 925          * optimization as in match_entry. For now just be happy with
 926          * base comparison.
 927          */
 928         return entry_interesting;
 929 }
 930
 931 /*
 932  * Is a tree entry interesting given the pathspec we have?
 933  *
 934  * Pre-condition: either baselen == base_offset (i.e. empty path)
 935  * or base[baselen-1] == '/' (i.e. with trailing slash).
 936  */
 937 static enum interesting do_match(const struct name_entry *entry,
 938                                  struct strbuf *base, int base_offset,
 939                                  const struct pathspec *ps,
 940                                  int exclude)
 941 {
 942         int i;
 943         int pathlen, baselen = base->len - base_offset;
 944         enum interesting never_interesting = ps->has_wildcard ?
 945                 entry_not_interesting : all_entries_not_interesting;
 946
 947         GUARD_PATHSPEC(ps,
 948                        PATHSPEC_FROMTOP |
 949                        PATHSPEC_MAXDEPTH |
 950                        PATHSPEC_LITERAL |
 951                        PATHSPEC_GLOB |
 952                        PATHSPEC_ICASE |
 953                        PATHSPEC_EXCLUDE);
 954
 955         if (!ps->nr) {
 956                 if (!ps->recursive ||
 957                     !(ps->magic & PATHSPEC_MAXDEPTH) ||
 958                     ps->max_depth == -1)
 959                         return all_entries_interesting;
 960                 return within_depth(base->buf + base_offset, baselen,
 961                                     !!S_ISDIR(entry->mode),
 962                                     ps->max_depth) ?
 963                         entry_interesting : entry_not_interesting;
 964         }
 965
 966         pathlen = tree_entry_len(entry);
 967
 968         for (i = ps->nr - 1; i >= 0; i--) {
 969                 const struct pathspec_item *item = ps->items+i;
 970                 const char *match = item->match;
 971                 const char *base_str = base->buf + base_offset;
 972                 int matchlen = item->len, matched = 0;
 973
 974                 if ((!exclude &&   item->magic & PATHSPEC_EXCLUDE) ||
 975                     ( exclude && !(item->magic & PATHSPEC_EXCLUDE)))
 976                         continue;
 977
 978                 if (baselen >= matchlen) {
 979                         /* If it doesn't match, move along... */
 980                         if (!match_dir_prefix(item, base_str, match, matchlen))
 981                                 goto match_wildcards;
 982
 983                         if (!ps->recursive ||
 984                             !(ps->magic & PATHSPEC_MAXDEPTH) ||
 985                             ps->max_depth == -1)
 986                                 return all_entries_interesting;
 987
 988                         return within_depth(base_str + matchlen + 1,
 989                                             baselen - matchlen - 1,
 990                                             !!S_ISDIR(entry->mode),
 991                                             ps->max_depth) ?
 992                                 entry_interesting : entry_not_interesting;
 993                 }
 994
 995                 /* Either there must be no base, or the base must match. */
 996                 if (baselen == 0 || !basecmp(item, base_str, match, baselen)) {
 997                         if (match_entry(item, entry, pathlen,
 998                                         match + baselen, matchlen - baselen,
 999                                         &never_interesting))
1000                                 return entry_interesting;
1001
1002                         if (item->nowildcard_len < item->len) {
1003                                 if (!git_fnmatch(item, match + baselen, entry->path,
1004                                                  item->nowildcard_len - baselen))
1005                                         return entry_interesting;
1006
1007                                 /*
1008                                  * Match all directories. We'll try to
1009                                  * match files later on.
1010                                  */
1011                                 if (ps->recursive && S_ISDIR(entry->mode))
1012                                         return entry_interesting;
1013
1014                                 /*
1015                                  * When matching against submodules with
1016                                  * wildcard characters, ensure that the entry
1017                                  * at least matches up to the first wild
1018                                  * character.  More accurate matching can then
1019                                  * be performed in the submodule itself.
1020                                  */
1021                                 if (ps->recurse_submodules &&
1022                                     S_ISGITLINK(entry->mode) &&
1023                                     !ps_strncmp(item, match + baselen,
1024                                                 entry->path,
1025                                                 item->nowildcard_len - baselen))
1026                                         return entry_interesting;
1027                         }
1028
1029                         continue;
1030                 }
1031
1032 match_wildcards:
1033                 if (item->nowildcard_len == item->len)
1034                         continue;
1035
1036                 if (item->nowildcard_len &&
1037                     !match_wildcard_base(item, base_str, baselen, &matched))
1038                         continue;
1039
1040                 /*
1041                  * Concatenate base and entry->path into one and do
1042                  * fnmatch() on it.
1043                  *
1044                  * While we could avoid concatenation in certain cases
1045                  * [1], which saves a memcpy and potentially a
1046                  * realloc, it turns out not worth it. Measurement on
1047                  * linux-2.6 does not show any clear improvements,
1048                  * partly because of the nowildcard_len optimization
1049                  * in git_fnmatch(). Avoid micro-optimizations here.
1050                  *
1051                  * [1] if match_wildcard_base() says the base
1052                  * directory is already matched, we only need to match
1053                  * the rest, which is shorter so _in theory_ faster.
1054                  */
1055
1056                 strbuf_add(base, entry->path, pathlen);
1057
1058                 if (!git_fnmatch(item, match, base->buf + base_offset,
1059                                  item->nowildcard_len)) {
1060                         strbuf_setlen(base, base_offset + baselen);
1061                         return entry_interesting;
1062                 }
1063
1064                 /*
1065                  * When matching against submodules with
1066                  * wildcard characters, ensure that the entry
1067                  * at least matches up to the first wild
1068                  * character.  More accurate matching can then
1069                  * be performed in the submodule itself.
1070                  */
1071                 if (ps->recurse_submodules && S_ISGITLINK(entry->mode) &&
1072                     !ps_strncmp(item, match, base->buf + base_offset,
1073                                 item->nowildcard_len)) {
1074                         strbuf_setlen(base, base_offset + baselen);
1075                         return entry_interesting;
1076                 }
1077
1078                 strbuf_setlen(base, base_offset + baselen);
1079
1080                 /*
1081                  * Match all directories. We'll try to match files
1082                  * later on.
1083                  * max_depth is ignored but we may consider support it
1084                  * in future, see
1085                  * https://public-inbox.org/git/7vmxo5l2g4.fsf@alter.siamese.dyndns.org/
1086                  */
1087                 if (ps->recursive && S_ISDIR(entry->mode))
1088                         return entry_interesting;
1089         }
1090         return never_interesting; /* No matches */
1091 }
1092
1093 /*
1094  * Is a tree entry interesting given the pathspec we have?
1095  *
1096  * Pre-condition: either baselen == base_offset (i.e. empty path)
1097  * or base[baselen-1] == '/' (i.e. with trailing slash).
1098  */
1099 enum interesting tree_entry_interesting(const struct name_entry *entry,
1100                                         struct strbuf *base, int base_offset,
1101                                         const struct pathspec *ps)
1102 {
1103         enum interesting positive, negative;
1104         positive = do_match(entry, base, base_offset, ps, 0);
1105
1106         /*
1107          * case | entry | positive | negative | result
1108          * -----+-------+----------+----------+-------
1109          *   1  |  file |   -1     |  -1..2   |  -1
1110          *   2  |  file |    0     |  -1..2   |   0
1111          *   3  |  file |    1     |   -1     |   1
1112          *   4  |  file |    1     |    0     |   1
1113          *   5  |  file |    1     |    1     |   0
1114          *   6  |  file |    1     |    2     |   0
1115          *   7  |  file |    2     |   -1     |   2
1116          *   8  |  file |    2     |    0     |   2
1117          *   9  |  file |    2     |    1     |   0
1118          *  10  |  file |    2     |    2     |  -1
1119          * -----+-------+----------+----------+-------
1120          *  11  |  dir  |   -1     |  -1..2   |  -1
1121          *  12  |  dir  |    0     |  -1..2   |   0
1122          *  13  |  dir  |    1     |   -1     |   1
1123          *  14  |  dir  |    1     |    0     |   1
1124          *  15  |  dir  |    1     |    1     |   1 (*)
1125          *  16  |  dir  |    1     |    2     |   0
1126          *  17  |  dir  |    2     |   -1     |   2
1127          *  18  |  dir  |    2     |    0     |   2
1128          *  19  |  dir  |    2     |    1     |   1 (*)
1129          *  20  |  dir  |    2     |    2     |  -1
1130          *
1131          * (*) An exclude pattern interested in a directory does not
1132          * necessarily mean it will exclude all of the directory. In
1133          * wildcard case, it can't decide until looking at individual
1134          * files inside. So don't write such directories off yet.
1135          */
1136
1137         if (!(ps->magic & PATHSPEC_EXCLUDE) ||
1138             positive <= entry_not_interesting) /* #1, #2, #11, #12 */
1139                 return positive;
1140
1141         negative = do_match(entry, base, base_offset, ps, 1);
1142
1143         /* #3, #4, #7, #8, #13, #14, #17, #18 */
1144         if (negative <= entry_not_interesting)
1145                 return positive;
1146
1147         /* #15, #19 */
1148         if (S_ISDIR(entry->mode) &&
1149             positive >= entry_interesting &&
1150             negative == entry_interesting)
1151                 return entry_interesting;
1152
1153         if ((positive == entry_interesting &&
1154              negative >= entry_interesting) || /* #5, #6, #16 */
1155             (positive == all_entries_interesting &&
1156              negative == entry_interesting)) /* #9 */
1157                 return entry_not_interesting;
1158
1159         return all_entries_not_interesting; /* #10, #20 */
1160 }