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