cache-tree.c

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