sha1_file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  *
   6  * This handles basic git sha1 object files - packing, unpacking,
   7  * creation etc.
   8  */
   9 #include <sys/types.h>
  10 #include <dirent.h>
  11 #include "cache.h"
  12 #include "delta.h"
  13 #include "pack.h"
  14
  15 #ifndef O_NOATIME
  16 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  17 #define O_NOATIME 01000000
  18 #else
  19 #define O_NOATIME 0
  20 #endif
  21 #endif
  22
  23 static unsigned int sha1_file_open_flag = O_NOATIME;
  24
  25 static unsigned hexval(char c)
  26 {
  27         if (c >= '0' && c <= '9')
  28                 return c - '0';
  29         if (c >= 'a' && c <= 'f')
  30                 return c - 'a' + 10;
  31         if (c >= 'A' && c <= 'F')
  32                 return c - 'A' + 10;
  33         return ~0;
  34 }
  35
  36 int get_sha1_hex(const char *hex, unsigned char *sha1)
  37 {
  38         int i;
  39         for (i = 0; i < 20; i++) {
  40                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  41                 if (val & ~0xff)
  42                         return -1;
  43                 *sha1++ = val;
  44                 hex += 2;
  45         }
  46         return 0;
  47 }
  48
  49 static int get_sha1_file(const char *path, unsigned char *result)
  50 {
  51         char buffer[60];
  52         int fd = open(path, O_RDONLY);
  53         int len;
  54
  55         if (fd < 0)
  56                 return -1;
  57         len = read(fd, buffer, sizeof(buffer));
  58         close(fd);
  59         if (len < 40)
  60                 return -1;
  61         return get_sha1_hex(buffer, result);
  62 }
  63
  64 static char *git_dir, *git_object_dir, *git_index_file, *git_refs_dir;
  65 static void setup_git_env(void)
  66 {
  67         git_dir = gitenv(GIT_DIR_ENVIRONMENT);
  68         if (!git_dir)
  69                 git_dir = DEFAULT_GIT_DIR_ENVIRONMENT;
  70         git_object_dir = gitenv(DB_ENVIRONMENT);
  71         if (!git_object_dir) {
  72                 git_object_dir = xmalloc(strlen(git_dir) + 9);
  73                 sprintf(git_object_dir, "%s/objects", git_dir);
  74         }
  75         git_refs_dir = xmalloc(strlen(git_dir) + 6);
  76         sprintf(git_refs_dir, "%s/refs", git_dir);
  77         git_index_file = gitenv(INDEX_ENVIRONMENT);
  78         if (!git_index_file) {
  79                 git_index_file = xmalloc(strlen(git_dir) + 7);
  80                 sprintf(git_index_file, "%s/index", git_dir);
  81         }
  82 }
  83
  84 char *get_object_directory(void)
  85 {
  86         if (!git_object_dir)
  87                 setup_git_env();
  88         return git_object_dir;
  89 }
  90
  91 char *get_refs_directory(void)
  92 {
  93         if (!git_refs_dir)
  94                 setup_git_env();
  95         return git_refs_dir;
  96 }
  97
  98 char *get_index_file(void)
  99 {
 100         if (!git_index_file)
 101                 setup_git_env();
 102         return git_index_file;
 103 }
 104
 105 int get_sha1(const char *str, unsigned char *sha1)
 106 {
 107         static char pathname[PATH_MAX];
 108         static const char *prefix[] = {
 109                 "",
 110                 "refs",
 111                 "refs/tags",
 112                 "refs/heads",
 113                 "refs/snap",
 114                 NULL
 115         };
 116         const char **p;
 117
 118         if (!get_sha1_hex(str, sha1))
 119                 return 0;
 120
 121         if (!git_dir)
 122                 setup_git_env();
 123         for (p = prefix; *p; p++) {
 124                 snprintf(pathname, sizeof(pathname), "%s/%s/%s",
 125                          git_dir, *p, str);
 126                 if (!get_sha1_file(pathname, sha1))
 127                         return 0;
 128         }
 129
 130         return -1;
 131 }
 132
 133 char * sha1_to_hex(const unsigned char *sha1)
 134 {
 135         static char buffer[50];
 136         static const char hex[] = "0123456789abcdef";
 137         char *buf = buffer;
 138         int i;
 139
 140         for (i = 0; i < 20; i++) {
 141                 unsigned int val = *sha1++;
 142                 *buf++ = hex[val >> 4];
 143                 *buf++ = hex[val & 0xf];
 144         }
 145         return buffer;
 146 }
 147
 148 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 149 {
 150         int i;
 151         for (i = 0; i < 20; i++) {
 152                 static char hex[] = "0123456789abcdef";
 153                 unsigned int val = sha1[i];
 154                 char *pos = pathbuf + i*2 + (i > 0);
 155                 *pos++ = hex[val >> 4];
 156                 *pos = hex[val & 0xf];
 157         }
 158 }
 159
 160 /*
 161  * NOTE! This returns a statically allocated buffer, so you have to be
 162  * careful about using it. Do a "strdup()" if you need to save the
 163  * filename.
 164  *
 165  * Also note that this returns the location for creating.  Reading
 166  * SHA1 file can happen from any alternate directory listed in the
 167  * DB_ENVIRONMENT environment variable if it is not found in
 168  * the primary object database.
 169  */
 170 char *sha1_file_name(const unsigned char *sha1)
 171 {
 172         static char *name, *base;
 173
 174         if (!base) {
 175                 const char *sha1_file_directory = get_object_directory();
 176                 int len = strlen(sha1_file_directory);
 177                 base = xmalloc(len + 60);
 178                 memcpy(base, sha1_file_directory, len);
 179                 memset(base+len, 0, 60);
 180                 base[len] = '/';
 181                 base[len+3] = '/';
 182                 name = base + len + 1;
 183         }
 184         fill_sha1_path(name, sha1);
 185         return base;
 186 }
 187
 188 struct alternate_object_database *alt_odb;
 189
 190 /*
 191  * Prepare alternate object database registry.
 192  * alt_odb points at an array of struct alternate_object_database.
 193  * This array is terminated with an element that has both its base
 194  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 195  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 196  * variable, and its base points at a statically allocated buffer
 197  * that contains "/the/directory/corresponding/to/.git/objects/...",
 198  * while its name points just after the slash at the end of
 199  * ".git/objects/" in the example above, and has enough space to hold
 200  * 40-byte hex SHA1, an extra slash for the first level indirection,
 201  * and the terminating NUL.
 202  * This function allocates the alt_odb array and all the strings
 203  * pointed by base fields of the array elements with one xmalloc();
 204  * the string pool immediately follows the array.
 205  */
 206 void prepare_alt_odb(void)
 207 {
 208         int pass, totlen, i;
 209         const char *cp, *last;
 210         char *op = NULL;
 211         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 212
 213         if (alt_odb)
 214                 return;
 215         /* The first pass counts how large an area to allocate to
 216          * hold the entire alt_odb structure, including array of
 217          * structs and path buffers for them.  The second pass fills
 218          * the structure and prepares the path buffers for use by
 219          * fill_sha1_path().
 220          */
 221         for (totlen = pass = 0; pass < 2; pass++) {
 222                 last = alt;
 223                 i = 0;
 224                 do {
 225                         cp = strchr(last, ':') ? : last + strlen(last);
 226                         if (last != cp) {
 227                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 228                                 int pfxlen = cp - last;
 229                                 int entlen = pfxlen + 43;
 230                                 if (pass == 0)
 231                                         totlen += entlen;
 232                                 else {
 233                                         alt_odb[i].base = op;
 234                                         alt_odb[i].name = op + pfxlen + 1;
 235                                         memcpy(op, last, pfxlen);
 236                                         op[pfxlen] = op[pfxlen + 3] = '/';
 237                                         op[entlen-1] = 0;
 238                                         op += entlen;
 239                                 }
 240                                 i++;
 241                         }
 242                         while (*cp && *cp == ':')
 243                                 cp++;
 244                         last = cp;
 245                 } while (*cp);
 246                 if (pass)
 247                         break;
 248                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 249                 alt_odb[i].base = alt_odb[i].name = NULL;
 250                 op = (char*)(&alt_odb[i+1]);
 251         }
 252 }
 253
 254 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 255 {
 256         int i;
 257         char *name = sha1_file_name(sha1);
 258
 259         if (!stat(name, st))
 260                 return name;
 261         prepare_alt_odb();
 262         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 263                 fill_sha1_path(name, sha1);
 264                 if (!stat(alt_odb[i].base, st))
 265                         return alt_odb[i].base;
 266         }
 267         return NULL;
 268 }
 269
 270 #define PACK_MAX_SZ (1<<26)
 271 static int pack_used_ctr;
 272 static unsigned long pack_mapped;
 273 struct packed_git *packed_git;
 274
 275 struct pack_entry {
 276         unsigned int offset;
 277         unsigned char sha1[20];
 278         struct packed_git *p;
 279 };
 280
 281 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 282                                 void **idx_map_)
 283 {
 284         void *idx_map;
 285         unsigned int *index;
 286         unsigned long idx_size;
 287         int nr, i;
 288         int fd = open(path, O_RDONLY);
 289         struct stat st;
 290         if (fd < 0)
 291                 return -1;
 292         if (fstat(fd, &st)) {
 293                 close(fd);
 294                 return -1;
 295         }
 296         idx_size = st.st_size;
 297         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 298         close(fd);
 299         if (idx_map == MAP_FAILED)
 300                 return -1;
 301
 302         index = idx_map;
 303
 304         /* check index map */
 305         if (idx_size < 4*256 + 20)
 306                 return error("index file too small");
 307         nr = 0;
 308         for (i = 0; i < 256; i++) {
 309                 unsigned int n = ntohl(index[i]);
 310                 if (n < nr)
 311                         return error("non-monotonic index");
 312                 nr = n;
 313         }
 314
 315         /*
 316          * Total size:
 317          *  - 256 index entries 4 bytes each
 318          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 319          *  - 20-byte SHA1 of the packfile
 320          *  - 20-byte SHA1 file checksum
 321          */
 322         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 323                 return error("wrong index file size");
 324
 325         *idx_map_ = idx_map;
 326         *idx_size_ = idx_size;
 327         return 0;
 328 }
 329
 330 static void unuse_one_packed_git(void)
 331 {
 332         /* NOTYET */
 333 }
 334
 335 static int use_packed_git(struct packed_git *p)
 336 {
 337         if (!p->pack_base) {
 338                 int fd;
 339                 struct stat st;
 340                 void *map;
 341
 342                 pack_mapped += p->pack_size;
 343                 while (PACK_MAX_SZ < pack_mapped)
 344                         unuse_one_packed_git();
 345                 fd = open(p->pack_name, O_RDONLY);
 346                 if (fd < 0)
 347                         return -1;
 348                 if (fstat(fd, &st)) {
 349                         close(fd);
 350                         return -1;
 351                 }
 352                 if (st.st_size != p->pack_size)
 353                         return -1;
 354                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 355                 close(fd);
 356                 if (map == MAP_FAILED)
 357                         return -1;
 358                 p->pack_base = map;
 359         }
 360         p->pack_last_used = pack_used_ctr++;
 361         return 0;
 362 }
 363
 364 static struct packed_git *add_packed_git(char *path, int path_len)
 365 {
 366         struct stat st;
 367         struct packed_git *p;
 368         unsigned long idx_size;
 369         void *idx_map;
 370
 371         if (check_packed_git_idx(path, &idx_size, &idx_map))
 372                 return NULL;
 373
 374         /* do we have a corresponding .pack file? */
 375         strcpy(path + path_len - 4, ".pack");
 376         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 377                 munmap(idx_map, idx_size);
 378                 return NULL;
 379         }
 380         /* ok, it looks sane as far as we can check without
 381          * actually mapping the pack file.
 382          */
 383         p = xmalloc(sizeof(*p) + path_len + 2);
 384         strcpy(p->pack_name, path);
 385         p->index_size = idx_size;
 386         p->pack_size = st.st_size;
 387         p->index_base = idx_map;
 388         p->next = NULL;
 389         p->pack_base = NULL;
 390         p->pack_last_used = 0;
 391         return p;
 392 }
 393
 394 static void prepare_packed_git_one(char *objdir)
 395 {
 396         char path[PATH_MAX];
 397         int len;
 398         DIR *dir;
 399         struct dirent *de;
 400
 401         sprintf(path, "%s/pack", objdir);
 402         len = strlen(path);
 403         dir = opendir(path);
 404         if (!dir)
 405                 return;
 406         path[len++] = '/';
 407         while ((de = readdir(dir)) != NULL) {
 408                 int namelen = strlen(de->d_name);
 409                 struct packed_git *p;
 410
 411                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 412                         continue;
 413
 414                 /* we have .idx.  Is it a file we can map? */
 415                 strcpy(path + len, de->d_name);
 416                 p = add_packed_git(path, len + namelen);
 417                 if (!p)
 418                         continue;
 419                 p->next = packed_git;
 420                 packed_git = p;
 421         }
 422 }
 423
 424 void prepare_packed_git(void)
 425 {
 426         int i;
 427         static int run_once = 0;
 428
 429         if (run_once++)
 430                 return;
 431
 432         prepare_packed_git_one(get_object_directory());
 433         prepare_alt_odb();
 434         for (i = 0; alt_odb[i].base != NULL; i++) {
 435                 alt_odb[i].name[0] = 0;
 436                 prepare_packed_git_one(alt_odb[i].base);
 437         }
 438 }
 439
 440 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 441 {
 442         char header[100];
 443         unsigned char real_sha1[20];
 444         SHA_CTX c;
 445
 446         SHA1_Init(&c);
 447         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 448         SHA1_Update(&c, map, size);
 449         SHA1_Final(real_sha1, &c);
 450         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 451 }
 452
 453 static void *map_sha1_file_internal(const unsigned char *sha1,
 454                                     unsigned long *size,
 455                                     int say_error)
 456 {
 457         struct stat st;
 458         void *map;
 459         int fd;
 460         char *filename = find_sha1_file(sha1, &st);
 461
 462         if (!filename) {
 463                 if (say_error)
 464                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 465                 return NULL;
 466         }
 467
 468         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 469         if (fd < 0) {
 470                 /* See if it works without O_NOATIME */
 471                 switch (sha1_file_open_flag) {
 472                 default:
 473                         fd = open(filename, O_RDONLY);
 474                         if (fd >= 0)
 475                                 break;
 476                 /* Fallthrough */
 477                 case 0:
 478                         if (say_error)
 479                                 perror(filename);
 480                         return NULL;
 481                 }
 482
 483                 /* If it failed once, it will probably fail again.
 484                  * Stop using O_NOATIME
 485                  */
 486                 sha1_file_open_flag = 0;
 487         }
 488         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 489         close(fd);
 490         if (-1 == (int)(long)map)
 491                 return NULL;
 492         *size = st.st_size;
 493         return map;
 494 }
 495
 496 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 497 {
 498         return map_sha1_file_internal(sha1, size, 1);
 499 }
 500
 501 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 502 {
 503         /* Get the data stream */
 504         memset(stream, 0, sizeof(*stream));
 505         stream->next_in = map;
 506         stream->avail_in = mapsize;
 507         stream->next_out = buffer;
 508         stream->avail_out = size;
 509
 510         inflateInit(stream);
 511         return inflate(stream, 0);
 512 }
 513
 514 void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 515 {
 516         int bytes = strlen(buffer) + 1;
 517         unsigned char *buf = xmalloc(1+size);
 518
 519         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 520         bytes = stream->total_out - bytes;
 521         if (bytes < size) {
 522                 stream->next_out = buf + bytes;
 523                 stream->avail_out = size - bytes;
 524                 while (inflate(stream, Z_FINISH) == Z_OK)
 525                         /* nothing */;
 526         }
 527         buf[size] = 0;
 528         inflateEnd(stream);
 529         return buf;
 530 }
 531
 532 /*
 533  * We used to just use "sscanf()", but that's actually way
 534  * too permissive for what we want to check. So do an anal
 535  * object header parse by hand.
 536  */
 537 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 538 {
 539         int i;
 540         unsigned long size;
 541
 542         /*
 543          * The type can be at most ten bytes (including the
 544          * terminating '\0' that we add), and is followed by
 545          * a space.
 546          */
 547         i = 10;
 548         for (;;) {
 549                 char c = *hdr++;
 550                 if (c == ' ')
 551                         break;
 552                 if (!--i)
 553                         return -1;
 554                 *type++ = c;
 555         }
 556         *type = 0;
 557
 558         /*
 559          * The length must follow immediately, and be in canonical
 560          * decimal format (ie "010" is not valid).
 561          */
 562         size = *hdr++ - '0';
 563         if (size > 9)
 564                 return -1;
 565         if (size) {
 566                 for (;;) {
 567                         unsigned long c = *hdr - '0';
 568                         if (c > 9)
 569                                 break;
 570                         hdr++;
 571                         size = size * 10 + c;
 572                 }
 573         }
 574         *sizep = size;
 575
 576         /*
 577          * The length must be followed by a zero byte
 578          */
 579         return *hdr ? -1 : 0;
 580 }
 581
 582 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 583 {
 584         int ret;
 585         z_stream stream;
 586         char hdr[8192];
 587
 588         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 589         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 590                 return NULL;
 591
 592         return unpack_sha1_rest(&stream, hdr, *size);
 593 }
 594
 595 static int packed_delta_info(unsigned char *base_sha1,
 596                              unsigned long delta_size,
 597                              unsigned long left,
 598                              char *type,
 599                              unsigned long *sizep)
 600 {
 601         const unsigned char *data;
 602         unsigned char delta_head[64];
 603         unsigned long result_size, base_size, verify_base_size;
 604         z_stream stream;
 605         int st;
 606
 607         if (left < 20)
 608                 die("truncated pack file");
 609         if (sha1_object_info(base_sha1, type, &base_size))
 610                 die("cannot get info for delta-pack base");
 611
 612         memset(&stream, 0, sizeof(stream));
 613
 614         data = stream.next_in = base_sha1 + 20;
 615         stream.avail_in = left - 20;
 616         stream.next_out = delta_head;
 617         stream.avail_out = sizeof(delta_head);
 618
 619         inflateInit(&stream);
 620         st = inflate(&stream, Z_FINISH);
 621         inflateEnd(&stream);
 622         if ((st != Z_STREAM_END) && stream.total_out != sizeof(delta_head))
 623                 die("delta data unpack-initial failed");
 624
 625         /* Examine the initial part of the delta to figure out
 626          * the result size.  Verify the base size while we are at it.
 627          */
 628         data = delta_head;
 629         verify_base_size = get_delta_hdr_size(&data);
 630         if (verify_base_size != base_size)
 631                 die("delta base size mismatch");
 632
 633         /* Read the result size */
 634         result_size = get_delta_hdr_size(&data);
 635         *sizep = result_size;
 636         return 0;
 637 }
 638
 639 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 640         enum object_type *type, unsigned long *sizep)
 641 {
 642         unsigned shift;
 643         unsigned char *pack, c;
 644         unsigned long size;
 645
 646         if (offset >= p->pack_size)
 647                 die("object offset outside of pack file");
 648
 649         pack =  p->pack_base + offset;
 650         c = *pack++;
 651         offset++;
 652         *type = (c >> 4) & 7;
 653         size = c & 15;
 654         shift = 4;
 655         while (c & 0x80) {
 656                 if (offset >= p->pack_size)
 657                         die("object offset outside of pack file");
 658                 c = *pack++;
 659                 offset++;
 660                 size += (c & 0x7f) << shift;
 661                 shift += 7;
 662         }
 663         *sizep = size;
 664         return offset;
 665 }
 666
 667 static int packed_object_info(struct pack_entry *entry,
 668                               char *type, unsigned long *sizep)
 669 {
 670         struct packed_git *p = entry->p;
 671         unsigned long offset, size, left;
 672         unsigned char *pack;
 673         enum object_type kind;
 674
 675         if (use_packed_git(p))
 676                 die("cannot map packed file");
 677
 678         offset = unpack_object_header(p, entry->offset, &kind, &size);
 679         pack = p->pack_base + offset;
 680         left = p->pack_size - offset;
 681
 682         switch (kind) {
 683         case OBJ_DELTA:
 684                 return packed_delta_info(pack, size, left, type, sizep);
 685                 break;
 686         case OBJ_COMMIT:
 687                 strcpy(type, "commit");
 688                 break;
 689         case OBJ_TREE:
 690                 strcpy(type, "tree");
 691                 break;
 692         case OBJ_BLOB:
 693                 strcpy(type, "blob");
 694                 break;
 695         case OBJ_TAG:
 696                 strcpy(type, "tag");
 697                 break;
 698         default:
 699                 die("corrupted pack file");
 700         }
 701         *sizep = size;
 702         return 0;
 703 }
 704
 705 /* forward declaration for a mutually recursive function */
 706 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 707
 708 static void *unpack_delta_entry(unsigned char *base_sha1,
 709                                 unsigned long delta_size,
 710                                 unsigned long left,
 711                                 char *type,
 712                                 unsigned long *sizep)
 713 {
 714         void *data, *delta_data, *result, *base;
 715         unsigned long data_size, result_size, base_size;
 716         z_stream stream;
 717         int st;
 718
 719         if (left < 20)
 720                 die("truncated pack file");
 721         data = base_sha1 + 20;
 722         data_size = left - 20;
 723         delta_data = xmalloc(delta_size);
 724
 725         memset(&stream, 0, sizeof(stream));
 726
 727         stream.next_in = data;
 728         stream.avail_in = data_size;
 729         stream.next_out = delta_data;
 730         stream.avail_out = delta_size;
 731
 732         inflateInit(&stream);
 733         st = inflate(&stream, Z_FINISH);
 734         inflateEnd(&stream);
 735         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 736                 die("delta data unpack failed");
 737
 738         /* This may recursively unpack the base, which is what we want */
 739         base = read_sha1_file(base_sha1, type, &base_size);
 740         if (!base)
 741                 die("failed to read delta-pack base object %s",
 742                     sha1_to_hex(base_sha1));
 743         result = patch_delta(base, base_size,
 744                              delta_data, delta_size,
 745                              &result_size);
 746         if (!result)
 747                 die("failed to apply delta");
 748         free(delta_data);
 749         free(base);
 750         *sizep = result_size;
 751         return result;
 752 }
 753
 754 static void *unpack_non_delta_entry(unsigned char *data,
 755                                     unsigned long size,
 756                                     unsigned long left)
 757 {
 758         int st;
 759         z_stream stream;
 760         char *buffer;
 761
 762         buffer = xmalloc(size + 1);
 763         buffer[size] = 0;
 764         memset(&stream, 0, sizeof(stream));
 765         stream.next_in = data;
 766         stream.avail_in = left;
 767         stream.next_out = buffer;
 768         stream.avail_out = size;
 769
 770         inflateInit(&stream);
 771         st = inflate(&stream, Z_FINISH);
 772         inflateEnd(&stream);
 773         if ((st != Z_STREAM_END) || stream.total_out != size) {
 774                 free(buffer);
 775                 return NULL;
 776         }
 777
 778         return buffer;
 779 }
 780
 781 static void *unpack_entry(struct pack_entry *entry,
 782                           char *type, unsigned long *sizep)
 783 {
 784         struct packed_git *p = entry->p;
 785         unsigned long offset, size, left;
 786         unsigned char *pack;
 787         enum object_type kind;
 788
 789         if (use_packed_git(p))
 790                 die("cannot map packed file");
 791
 792         offset = unpack_object_header(p, entry->offset, &kind, &size);
 793         pack = p->pack_base + offset;
 794         left = p->pack_size - offset;
 795         switch (kind) {
 796         case OBJ_DELTA:
 797                 return unpack_delta_entry(pack, size, left, type, sizep);
 798         case OBJ_COMMIT:
 799                 strcpy(type, "commit");
 800                 break;
 801         case OBJ_TREE:
 802                 strcpy(type, "tree");
 803                 break;
 804         case OBJ_BLOB:
 805                 strcpy(type, "blob");
 806                 break;
 807         case OBJ_TAG:
 808                 strcpy(type, "tag");
 809                 break;
 810         default:
 811                 die("corrupted pack file");
 812         }
 813         *sizep = size;
 814         return unpack_non_delta_entry(pack, size, left);
 815 }
 816
 817 int num_packed_objects(const struct packed_git *p)
 818 {
 819         /* See check_packed_git_idx and pack-objects.c */
 820         return (p->index_size - 20 - 20 - 4*256) / 24;
 821 }
 822
 823 int nth_packed_object_sha1(const struct packed_git *p, int n,
 824                            unsigned char* sha1)
 825 {
 826         void *index = p->index_base + 256;
 827         if (n < 0 || num_packed_objects(p) <= n)
 828                 return -1;
 829         memcpy(sha1, (index + 24 * n + 4), 20);
 830         return 0;
 831 }
 832
 833 static int find_pack_entry_1(const unsigned char *sha1,
 834                              struct pack_entry *e, struct packed_git *p)
 835 {
 836         int *level1_ofs = p->index_base;
 837         int hi = ntohl(level1_ofs[*sha1]);
 838         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 839         void *index = p->index_base + 256;
 840
 841         do {
 842                 int mi = (lo + hi) / 2;
 843                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 844                 if (!cmp) {
 845                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 846                         memcpy(e->sha1, sha1, 20);
 847                         e->p = p;
 848                         return 1;
 849                 }
 850                 if (cmp > 0)
 851                         hi = mi;
 852                 else
 853                         lo = mi+1;
 854         } while (lo < hi);
 855         return 0;
 856 }
 857
 858 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 859 {
 860         struct packed_git *p;
 861         prepare_packed_git();
 862
 863         for (p = packed_git; p; p = p->next) {
 864                 if (find_pack_entry_1(sha1, e, p))
 865                         return 1;
 866         }
 867         return 0;
 868 }
 869
 870 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
 871 {
 872         int status;
 873         unsigned long mapsize, size;
 874         void *map;
 875         z_stream stream;
 876         char hdr[128];
 877
 878         map = map_sha1_file_internal(sha1, &mapsize, 0);
 879         if (!map) {
 880                 struct pack_entry e;
 881
 882                 if (!find_pack_entry(sha1, &e))
 883                         return error("unable to find %s", sha1_to_hex(sha1));
 884                 if (!packed_object_info(&e, type, sizep))
 885                         return 0;
 886                 /* sheesh */
 887                 map = unpack_entry(&e, type, sizep);
 888                 free(map);
 889                 return (map == NULL) ? 0 : -1;
 890         }
 891         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
 892                 status = error("unable to unpack %s header",
 893                                sha1_to_hex(sha1));
 894         if (parse_sha1_header(hdr, type, &size) < 0)
 895                 status = error("unable to parse %s header", sha1_to_hex(sha1));
 896         else {
 897                 status = 0;
 898                 *sizep = size;
 899         }
 900         inflateEnd(&stream);
 901         munmap(map, mapsize);
 902         return status;
 903 }
 904
 905 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
 906 {
 907         struct pack_entry e;
 908
 909         if (!find_pack_entry(sha1, &e)) {
 910                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
 911                 return NULL;
 912         }
 913         return unpack_entry(&e, type, size);
 914 }
 915
 916 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
 917 {
 918         unsigned long mapsize;
 919         void *map, *buf;
 920
 921         map = map_sha1_file_internal(sha1, &mapsize, 0);
 922         if (map) {
 923                 buf = unpack_sha1_file(map, mapsize, type, size);
 924                 munmap(map, mapsize);
 925                 return buf;
 926         }
 927         return read_packed_sha1(sha1, type, size);
 928 }
 929
 930 void *read_object_with_reference(const unsigned char *sha1,
 931                                  const char *required_type,
 932                                  unsigned long *size,
 933                                  unsigned char *actual_sha1_return)
 934 {
 935         char type[20];
 936         void *buffer;
 937         unsigned long isize;
 938         unsigned char actual_sha1[20];
 939
 940         memcpy(actual_sha1, sha1, 20);
 941         while (1) {
 942                 int ref_length = -1;
 943                 const char *ref_type = NULL;
 944
 945                 buffer = read_sha1_file(actual_sha1, type, &isize);
 946                 if (!buffer)
 947                         return NULL;
 948                 if (!strcmp(type, required_type)) {
 949                         *size = isize;
 950                         if (actual_sha1_return)
 951                                 memcpy(actual_sha1_return, actual_sha1, 20);
 952                         return buffer;
 953                 }
 954                 /* Handle references */
 955                 else if (!strcmp(type, "commit"))
 956                         ref_type = "tree ";
 957                 else if (!strcmp(type, "tag"))
 958                         ref_type = "object ";
 959                 else {
 960                         free(buffer);
 961                         return NULL;
 962                 }
 963                 ref_length = strlen(ref_type);
 964
 965                 if (memcmp(buffer, ref_type, ref_length) ||
 966                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
 967                         free(buffer);
 968                         return NULL;
 969                 }
 970                 /* Now we have the ID of the referred-to object in
 971                  * actual_sha1.  Check again. */
 972         }
 973 }
 974
 975 static char *write_sha1_file_prepare(void *buf,
 976                                      unsigned long len,
 977                                      const char *type,
 978                                      unsigned char *sha1,
 979                                      unsigned char *hdr,
 980                                      int *hdrlen)
 981 {
 982         SHA_CTX c;
 983
 984         /* Generate the header */
 985         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
 986
 987         /* Sha1.. */
 988         SHA1_Init(&c);
 989         SHA1_Update(&c, hdr, *hdrlen);
 990         SHA1_Update(&c, buf, len);
 991         SHA1_Final(sha1, &c);
 992
 993         return sha1_file_name(sha1);
 994 }
 995
 996 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
 997 {
 998         int size;
 999         unsigned char *compressed;
1000         z_stream stream;
1001         unsigned char sha1[20];
1002         char *filename;
1003         static char tmpfile[PATH_MAX];
1004         unsigned char hdr[50];
1005         int fd, hdrlen, ret;
1006
1007         /* Normally if we have it in the pack then we do not bother writing
1008          * it out into .git/objects/??/?{38} file.
1009          */
1010         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1011         if (returnsha1)
1012                 memcpy(returnsha1, sha1, 20);
1013         if (has_sha1_file(sha1))
1014                 return 0;
1015         fd = open(filename, O_RDONLY);
1016         if (fd >= 0) {
1017                 /*
1018                  * FIXME!!! We might do collision checking here, but we'd
1019                  * need to uncompress the old file and check it. Later.
1020                  */
1021                 close(fd);
1022                 return 0;
1023         }
1024
1025         if (errno != ENOENT) {
1026                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1027                 return -1;
1028         }
1029
1030         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1031
1032         fd = mkstemp(tmpfile);
1033         if (fd < 0) {
1034                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1035                 return -1;
1036         }
1037
1038         /* Set it up */
1039         memset(&stream, 0, sizeof(stream));
1040         deflateInit(&stream, Z_BEST_COMPRESSION);
1041         size = deflateBound(&stream, len+hdrlen);
1042         compressed = xmalloc(size);
1043
1044         /* Compress it */
1045         stream.next_out = compressed;
1046         stream.avail_out = size;
1047
1048         /* First header.. */
1049         stream.next_in = hdr;
1050         stream.avail_in = hdrlen;
1051         while (deflate(&stream, 0) == Z_OK)
1052                 /* nothing */;
1053
1054         /* Then the data itself.. */
1055         stream.next_in = buf;
1056         stream.avail_in = len;
1057         while (deflate(&stream, Z_FINISH) == Z_OK)
1058                 /* nothing */;
1059         deflateEnd(&stream);
1060         size = stream.total_out;
1061
1062         if (write(fd, compressed, size) != size)
1063                 die("unable to write file");
1064         fchmod(fd, 0444);
1065         close(fd);
1066         free(compressed);
1067
1068         ret = link(tmpfile, filename);
1069         if (ret < 0) {
1070                 ret = errno;
1071
1072                 /*
1073                  * Coda hack - coda doesn't like cross-directory links,
1074                  * so we fall back to a rename, which will mean that it
1075                  * won't be able to check collisions, but that's not a
1076                  * big deal.
1077                  *
1078                  * When this succeeds, we just return 0. We have nothing
1079                  * left to unlink.
1080                  */
1081                 if (ret == EXDEV && !rename(tmpfile, filename))
1082                         return 0;
1083         }
1084         unlink(tmpfile);
1085         if (ret) {
1086                 if (ret != EEXIST) {
1087                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1088                         return -1;
1089                 }
1090                 /* FIXME!!! Collision check here ? */
1091         }
1092
1093         return 0;
1094 }
1095
1096 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1097 {
1098         char *filename = sha1_file_name(sha1);
1099
1100         int local;
1101         z_stream stream;
1102         unsigned char real_sha1[20];
1103         unsigned char buf[4096];
1104         unsigned char discard[4096];
1105         int ret;
1106         SHA_CTX c;
1107
1108         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1109
1110         if (local < 0)
1111                 return error("Couldn't open %s\n", filename);
1112
1113         memset(&stream, 0, sizeof(stream));
1114
1115         inflateInit(&stream);
1116
1117         SHA1_Init(&c);
1118
1119         do {
1120                 ssize_t size;
1121                 size = read(fd, buf, 4096);
1122                 if (size <= 0) {
1123                         close(local);
1124                         unlink(filename);
1125                         if (!size)
1126                                 return error("Connection closed?");
1127                         perror("Reading from connection");
1128                         return -1;
1129                 }
1130                 write(local, buf, size);
1131                 stream.avail_in = size;
1132                 stream.next_in = buf;
1133                 do {
1134                         stream.next_out = discard;
1135                         stream.avail_out = sizeof(discard);
1136                         ret = inflate(&stream, Z_SYNC_FLUSH);
1137                         SHA1_Update(&c, discard, sizeof(discard) -
1138                                     stream.avail_out);
1139                 } while (stream.avail_in && ret == Z_OK);
1140
1141         } while (ret == Z_OK);
1142         inflateEnd(&stream);
1143
1144         close(local);
1145         SHA1_Final(real_sha1, &c);
1146         if (ret != Z_STREAM_END) {
1147                 unlink(filename);
1148                 return error("File %s corrupted", sha1_to_hex(sha1));
1149         }
1150         if (memcmp(sha1, real_sha1, 20)) {
1151                 unlink(filename);
1152                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1153         }
1154
1155         return 0;
1156 }
1157
1158 int has_sha1_file(const unsigned char *sha1)
1159 {
1160         struct stat st;
1161         struct pack_entry e;
1162
1163         if (find_sha1_file(sha1, &st))
1164                 return 1;
1165         return find_pack_entry(sha1, &e);
1166 }
1167
1168 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1169 {
1170         unsigned long size = st->st_size;
1171         void *buf;
1172         int ret;
1173
1174         buf = "";
1175         if (size)
1176                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1177         close(fd);
1178         if ((int)(long)buf == -1)
1179                 return -1;
1180
1181         ret = write_sha1_file(buf, size, "blob", sha1);
1182         if (size)
1183                 munmap(buf, size);
1184         return ret;
1185 }