cache-tree.c

   1 #include "cache.h"
   2 #include "tree.h"
   3 #include "cache-tree.h"
   4
   5 #ifndef DEBUG
   6 #define DEBUG 0
   7 #endif
   8
   9 struct cache_tree *cache_tree(void)
  10 {
  11         struct cache_tree *it = xcalloc(1, sizeof(struct cache_tree));
  12         it->entry_count = -1;
  13         return it;
  14 }
  15
  16 void cache_tree_free(struct cache_tree **it_p)
  17 {
  18         int i;
  19         struct cache_tree *it = *it_p;
  20
  21         if (!it)
  22                 return;
  23         for (i = 0; i < it->subtree_nr; i++)
  24                 if (it->down[i])
  25                         cache_tree_free(&it->down[i]->cache_tree);
  26         free(it->down);
  27         free(it);
  28         *it_p = NULL;
  29 }
  30
  31 static int subtree_name_cmp(const char *one, int onelen,
  32                             const char *two, int twolen)
  33 {
  34         if (onelen < twolen)
  35                 return -1;
  36         if (twolen < onelen)
  37                 return 1;
  38         return memcmp(one, two, onelen);
  39 }
  40
  41 static int subtree_pos(struct cache_tree *it, const char *path, int pathlen)
  42 {
  43         struct cache_tree_sub **down = it->down;
  44         int lo, hi;
  45         lo = 0;
  46         hi = it->subtree_nr;
  47         while (lo < hi) {
  48                 int mi = (lo + hi) / 2;
  49                 struct cache_tree_sub *mdl = down[mi];
  50                 int cmp = subtree_name_cmp(path, pathlen,
  51                                            mdl->name, mdl->namelen);
  52                 if (!cmp)
  53                         return mi;
  54                 if (cmp < 0)
  55                         hi = mi;
  56                 else
  57                         lo = mi + 1;
  58         }
  59         return -lo-1;
  60 }
  61
  62 static struct cache_tree_sub *find_subtree(struct cache_tree *it,
  63                                            const char *path,
  64                                            int pathlen,
  65                                            int create)
  66 {
  67         struct cache_tree_sub *down;
  68         int pos = subtree_pos(it, path, pathlen);
  69         if (0 <= pos)
  70                 return it->down[pos];
  71         if (!create)
  72                 return NULL;
  73
  74         pos = -pos-1;
  75         if (it->subtree_alloc <= it->subtree_nr) {
  76                 it->subtree_alloc = alloc_nr(it->subtree_alloc);
  77                 it->down = xrealloc(it->down, it->subtree_alloc *
  78                                     sizeof(*it->down));
  79         }
  80         it->subtree_nr++;
  81
  82         down = xmalloc(sizeof(*down) + pathlen + 1);
  83         down->cache_tree = NULL;
  84         down->namelen = pathlen;
  85         memcpy(down->name, path, pathlen);
  86         down->name[pathlen] = 0;
  87
  88         if (pos < it->subtree_nr)
  89                 memmove(it->down + pos + 1,
  90                         it->down + pos,
  91                         sizeof(down) * (it->subtree_nr - pos - 1));
  92         it->down[pos] = down;
  93         return down;
  94 }
  95
  96 struct cache_tree_sub *cache_tree_sub(struct cache_tree *it, const char *path)
  97 {
  98         int pathlen = strlen(path);
  99         return find_subtree(it, path, pathlen, 1);
 100 }
 101
 102 void cache_tree_invalidate_path(struct cache_tree *it, const char *path)
 103 {
 104         /* a/b/c
 105          * ==> invalidate self
 106          * ==> find "a", have it invalidate "b/c"
 107          * a
 108          * ==> invalidate self
 109          * ==> if "a" exists as a subtree, remove it.
 110          */
 111         const char *slash;
 112         int namelen;
 113         struct cache_tree_sub *down;
 114
 115 #if DEBUG
 116         fprintf(stderr, "cache-tree invalidate <%s>\n", path);
 117 #endif
 118
 119         if (!it)
 120                 return;
 121         slash = strchr(path, '/');
 122         it->entry_count = -1;
 123         if (!slash) {
 124                 int pos;
 125                 namelen = strlen(path);
 126                 pos = subtree_pos(it, path, namelen);
 127                 if (0 <= pos) {
 128                         cache_tree_free(&it->down[pos]->cache_tree);
 129                         free(it->down[pos]);
 130                         /* 0 1 2 3 4 5
 131                          *       ^     ^subtree_nr = 6
 132                          *       pos
 133                          * move 4 and 5 up one place (2 entries)
 134                          * 2 = 6 - 3 - 1 = subtree_nr - pos - 1
 135                          */
 136                         memmove(it->down+pos, it->down+pos+1,
 137                                 sizeof(struct cache_tree_sub *) *
 138                                 (it->subtree_nr - pos - 1));
 139                         it->subtree_nr--;
 140                 }
 141                 return;
 142         }
 143         namelen = slash - path;
 144         down = find_subtree(it, path, namelen, 0);
 145         if (down)
 146                 cache_tree_invalidate_path(down->cache_tree, slash + 1);
 147 }
 148
 149 static int verify_cache(struct cache_entry **cache,
 150                         int entries)
 151 {
 152         int i, funny;
 153
 154         /* Verify that the tree is merged */
 155         funny = 0;
 156         for (i = 0; i < entries; i++) {
 157                 struct cache_entry *ce = cache[i];
 158                 if (ce_stage(ce)) {
 159                         if (10 < ++funny) {
 160                                 fprintf(stderr, "...\n");
 161                                 break;
 162                         }
 163                         fprintf(stderr, "%s: unmerged (%s)\n",
 164                                 ce->name, sha1_to_hex(ce->sha1));
 165                 }
 166         }
 167         if (funny)
 168                 return -1;
 169
 170         /* Also verify that the cache does not have path and path/file
 171          * at the same time.  At this point we know the cache has only
 172          * stage 0 entries.
 173          */
 174         funny = 0;
 175         for (i = 0; i < entries - 1; i++) {
 176                 /* path/file always comes after path because of the way
 177                  * the cache is sorted.  Also path can appear only once,
 178                  * which means conflicting one would immediately follow.
 179                  */
 180                 const char *this_name = cache[i]->name;
 181                 const char *next_name = cache[i+1]->name;
 182                 int this_len = strlen(this_name);
 183                 if (this_len < strlen(next_name) &&
 184                     strncmp(this_name, next_name, this_len) == 0 &&
 185                     next_name[this_len] == '/') {
 186                         if (10 < ++funny) {
 187                                 fprintf(stderr, "...\n");
 188                                 break;
 189                         }
 190                         fprintf(stderr, "You have both %s and %s\n",
 191                                 this_name, next_name);
 192                 }
 193         }
 194         if (funny)
 195                 return -1;
 196         return 0;
 197 }
 198
 199 static void discard_unused_subtrees(struct cache_tree *it)
 200 {
 201         struct cache_tree_sub **down = it->down;
 202         int nr = it->subtree_nr;
 203         int dst, src;
 204         for (dst = src = 0; src < nr; src++) {
 205                 struct cache_tree_sub *s = down[src];
 206                 if (s->used)
 207                         down[dst++] = s;
 208                 else {
 209                         cache_tree_free(&s->cache_tree);
 210                         free(s);
 211                         it->subtree_nr--;
 212                 }
 213         }
 214 }
 215
 216 int cache_tree_fully_valid(struct cache_tree *it)
 217 {
 218         int i;
 219         if (!it)
 220                 return 0;
 221         if (it->entry_count < 0 || !has_sha1_file(it->sha1))
 222                 return 0;
 223         for (i = 0; i < it->subtree_nr; i++) {
 224                 if (!cache_tree_fully_valid(it->down[i]->cache_tree))
 225                         return 0;
 226         }
 227         return 1;
 228 }
 229
 230 static int update_one(struct cache_tree *it,
 231                       struct cache_entry **cache,
 232                       int entries,
 233                       const char *base,
 234                       int baselen,
 235                       int missing_ok,
 236                       int dryrun)
 237 {
 238         unsigned long size, offset;
 239         char *buffer;
 240         int i;
 241
 242         if (0 <= it->entry_count && has_sha1_file(it->sha1))
 243                 return it->entry_count;
 244
 245         /*
 246          * We first scan for subtrees and update them; we start by
 247          * marking existing subtrees -- the ones that are unmarked
 248          * should not be in the result.
 249          */
 250         for (i = 0; i < it->subtree_nr; i++)
 251                 it->down[i]->used = 0;
 252
 253         /*
 254          * Find the subtrees and update them.
 255          */
 256         for (i = 0; i < entries; i++) {
 257                 struct cache_entry *ce = cache[i];
 258                 struct cache_tree_sub *sub;
 259                 const char *path, *slash;
 260                 int pathlen, sublen, subcnt;
 261
 262                 path = ce->name;
 263                 pathlen = ce_namelen(ce);
 264                 if (pathlen <= baselen || memcmp(base, path, baselen))
 265                         break; /* at the end of this level */
 266
 267                 slash = strchr(path + baselen, '/');
 268                 if (!slash)
 269                         continue;
 270                 /*
 271                  * a/bbb/c (base = a/, slash = /c)
 272                  * ==>
 273                  * path+baselen = bbb/c, sublen = 3
 274                  */
 275                 sublen = slash - (path + baselen);
 276                 sub = find_subtree(it, path + baselen, sublen, 1);
 277                 if (!sub->cache_tree)
 278                         sub->cache_tree = cache_tree();
 279                 subcnt = update_one(sub->cache_tree,
 280                                     cache + i, entries - i,
 281                                     path,
 282                                     baselen + sublen + 1,
 283                                     missing_ok,
 284                                     dryrun);
 285                 i += subcnt - 1;
 286                 sub->used = 1;
 287         }
 288
 289         discard_unused_subtrees(it);
 290
 291         /*
 292          * Then write out the tree object for this level.
 293          */
 294         size = 8192;
 295         buffer = xmalloc(size);
 296         offset = 0;
 297
 298         for (i = 0; i < entries; i++) {
 299                 struct cache_entry *ce = cache[i];
 300                 struct cache_tree_sub *sub;
 301                 const char *path, *slash;
 302                 int pathlen, entlen;
 303                 const unsigned char *sha1;
 304                 unsigned mode;
 305
 306                 path = ce->name;
 307                 pathlen = ce_namelen(ce);
 308                 if (pathlen <= baselen || memcmp(base, path, baselen))
 309                         break; /* at the end of this level */
 310
 311                 slash = strchr(path + baselen, '/');
 312                 if (slash) {
 313                         entlen = slash - (path + baselen);
 314                         sub = find_subtree(it, path + baselen, entlen, 0);
 315                         if (!sub)
 316                                 die("cache-tree.c: '%.*s' in '%s' not found",
 317                                     entlen, path + baselen, path);
 318                         i += sub->cache_tree->entry_count - 1;
 319                         sha1 = sub->cache_tree->sha1;
 320                         mode = S_IFDIR;
 321                 }
 322                 else {
 323                         sha1 = ce->sha1;
 324                         mode = ntohl(ce->ce_mode);
 325                         entlen = pathlen - baselen;
 326                 }
 327                 if (!missing_ok && !has_sha1_file(sha1))
 328                         return error("invalid object %s", sha1_to_hex(sha1));
 329
 330                 if (!ce->ce_mode)
 331                         continue; /* entry being removed */
 332
 333                 if (size < offset + entlen + 100) {
 334                         size = alloc_nr(offset + entlen + 100);
 335                         buffer = xrealloc(buffer, size);
 336                 }
 337                 offset += sprintf(buffer + offset,
 338                                   "%o %.*s", mode, entlen, path + baselen);
 339                 buffer[offset++] = 0;
 340                 hashcpy((unsigned char*)buffer + offset, sha1);
 341                 offset += 20;
 342
 343 #if DEBUG
 344                 fprintf(stderr, "cache-tree update-one %o %.*s\n",
 345                         mode, entlen, path + baselen);
 346 #endif
 347         }
 348
 349         if (dryrun)
 350                 hash_sha1_file(buffer, offset, tree_type, it->sha1);
 351         else
 352                 write_sha1_file(buffer, offset, tree_type, it->sha1);
 353         free(buffer);
 354         it->entry_count = i;
 355 #if DEBUG
 356         fprintf(stderr, "cache-tree update-one (%d ent, %d subtree) %s\n",
 357                 it->entry_count, it->subtree_nr,
 358                 sha1_to_hex(it->sha1));
 359 #endif
 360         return i;
 361 }
 362
 363 int cache_tree_update(struct cache_tree *it,
 364                       struct cache_entry **cache,
 365                       int entries,
 366                       int missing_ok,
 367                       int dryrun)
 368 {
 369         int i;
 370         i = verify_cache(cache, entries);
 371         if (i)
 372                 return i;
 373         i = update_one(it, cache, entries, "", 0, missing_ok, dryrun);
 374         if (i < 0)
 375                 return i;
 376         return 0;
 377 }
 378
 379 static void *write_one(struct cache_tree *it,
 380                        char *path,
 381                        int pathlen,
 382                        char *buffer,
 383                        unsigned long *size,
 384                        unsigned long *offset)
 385 {
 386         int i;
 387
 388         /* One "cache-tree" entry consists of the following:
 389          * path (NUL terminated)
 390          * entry_count, subtree_nr ("%d %d\n")
 391          * tree-sha1 (missing if invalid)
 392          * subtree_nr "cache-tree" entries for subtrees.
 393          */
 394         if (*size < *offset + pathlen + 100) {
 395                 *size = alloc_nr(*offset + pathlen + 100);
 396                 buffer = xrealloc(buffer, *size);
 397         }
 398         *offset += sprintf(buffer + *offset, "%.*s%c%d %d\n",
 399                            pathlen, path, 0,
 400                            it->entry_count, it->subtree_nr);
 401
 402 #if DEBUG
 403         if (0 <= it->entry_count)
 404                 fprintf(stderr, "cache-tree <%.*s> (%d ent, %d subtree) %s\n",
 405                         pathlen, path, it->entry_count, it->subtree_nr,
 406                         sha1_to_hex(it->sha1));
 407         else
 408                 fprintf(stderr, "cache-tree <%.*s> (%d subtree) invalid\n",
 409                         pathlen, path, it->subtree_nr);
 410 #endif
 411
 412         if (0 <= it->entry_count) {
 413                 hashcpy((unsigned char*)buffer + *offset, it->sha1);
 414                 *offset += 20;
 415         }
 416         for (i = 0; i < it->subtree_nr; i++) {
 417                 struct cache_tree_sub *down = it->down[i];
 418                 if (i) {
 419                         struct cache_tree_sub *prev = it->down[i-1];
 420                         if (subtree_name_cmp(down->name, down->namelen,
 421                                              prev->name, prev->namelen) <= 0)
 422                                 die("fatal - unsorted cache subtree");
 423                 }
 424                 buffer = write_one(down->cache_tree, down->name, down->namelen,
 425                                    buffer, size, offset);
 426         }
 427         return buffer;
 428 }
 429
 430 void *cache_tree_write(struct cache_tree *root, unsigned long *size_p)
 431 {
 432         char path[PATH_MAX];
 433         unsigned long size = 8192;
 434         char *buffer = xmalloc(size);
 435
 436         *size_p = 0;
 437         path[0] = 0;
 438         return write_one(root, path, 0, buffer, &size, size_p);
 439 }
 440
 441 static struct cache_tree *read_one(const char **buffer, unsigned long *size_p)
 442 {
 443         const char *buf = *buffer;
 444         unsigned long size = *size_p;
 445         const char *cp;
 446         char *ep;
 447         struct cache_tree *it;
 448         int i, subtree_nr;
 449
 450         it = NULL;
 451         /* skip name, but make sure name exists */
 452         while (size && *buf) {
 453                 size--;
 454                 buf++;
 455         }
 456         if (!size)
 457                 goto free_return;
 458         buf++; size--;
 459         it = cache_tree();
 460
 461         cp = buf;
 462         it->entry_count = strtol(cp, &ep, 10);
 463         if (cp == ep)
 464                 goto free_return;
 465         cp = ep;
 466         subtree_nr = strtol(cp, &ep, 10);
 467         if (cp == ep)
 468                 goto free_return;
 469         while (size && *buf && *buf != '\n') {
 470                 size--;
 471                 buf++;
 472         }
 473         if (!size)
 474                 goto free_return;
 475         buf++; size--;
 476         if (0 <= it->entry_count) {
 477                 if (size < 20)
 478                         goto free_return;
 479                 hashcpy(it->sha1, (unsigned char*)buf);
 480                 buf += 20;
 481                 size -= 20;
 482         }
 483
 484 #if DEBUG
 485         if (0 <= it->entry_count)
 486                 fprintf(stderr, "cache-tree <%s> (%d ent, %d subtree) %s\n",
 487                         *buffer, it->entry_count, subtree_nr,
 488                         sha1_to_hex(it->sha1));
 489         else
 490                 fprintf(stderr, "cache-tree <%s> (%d subtrees) invalid\n",
 491                         *buffer, subtree_nr);
 492 #endif
 493
 494         /*
 495          * Just a heuristic -- we do not add directories that often but
 496          * we do not want to have to extend it immediately when we do,
 497          * hence +2.
 498          */
 499         it->subtree_alloc = subtree_nr + 2;
 500         it->down = xcalloc(it->subtree_alloc, sizeof(struct cache_tree_sub *));
 501         for (i = 0; i < subtree_nr; i++) {
 502                 /* read each subtree */
 503                 struct cache_tree *sub;
 504                 struct cache_tree_sub *subtree;
 505                 const char *name = buf;
 506
 507                 sub = read_one(&buf, &size);
 508                 if (!sub)
 509                         goto free_return;
 510                 subtree = cache_tree_sub(it, name);
 511                 subtree->cache_tree = sub;
 512         }
 513         if (subtree_nr != it->subtree_nr)
 514                 die("cache-tree: internal error");
 515         *buffer = buf;
 516         *size_p = size;
 517         return it;
 518
 519  free_return:
 520         cache_tree_free(&it);
 521         return NULL;
 522 }
 523
 524 struct cache_tree *cache_tree_read(const char *buffer, unsigned long size)
 525 {
 526         if (buffer[0])
 527                 return NULL; /* not the whole tree */
 528         return read_one(&buffer, &size);
 529 }
 530
 531 struct cache_tree *cache_tree_find(struct cache_tree *it, const char *path)
 532 {
 533         while (*path) {
 534                 const char *slash;
 535                 struct cache_tree_sub *sub;
 536
 537                 slash = strchr(path, '/');
 538                 if (!slash)
 539                         slash = path + strlen(path);
 540                 /* between path and slash is the name of the
 541                  * subtree to look for.
 542                  */
 543                 sub = find_subtree(it, path, slash - path, 0);
 544                 if (!sub)
 545                         return NULL;
 546                 it = sub->cache_tree;
 547                 if (slash)
 548                         while (*slash && *slash == '/')
 549                                 slash++;
 550                 if (!slash || !*slash)
 551                         return it; /* prefix ended with slashes */
 552                 path = slash;
 553         }
 554         return it;
 555 }