index-pack.c

   1 #include "cache.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include <sys/time.h>
  10 #include <signal.h>
  11
  12 static const char index_pack_usage[] =
  13 "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  14
  15 struct object_entry
  16 {
  17         unsigned long offset;
  18         unsigned long size;
  19         unsigned int hdr_size;
  20         enum object_type type;
  21         enum object_type real_type;
  22         unsigned char sha1[20];
  23 };
  24
  25 union delta_base {
  26         unsigned char sha1[20];
  27         unsigned long offset;
  28 };
  29
  30 /*
  31  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  32  * to memcmp() only the first 20 bytes.
  33  */
  34 #define UNION_BASE_SZ   20
  35
  36 struct delta_entry
  37 {
  38         union delta_base base;
  39         int obj_no;
  40 };
  41
  42 static struct object_entry *objects;
  43 static struct delta_entry *deltas;
  44 static int nr_objects;
  45 static int nr_deltas;
  46 static int nr_resolved_deltas;
  47
  48 static int from_stdin;
  49 static int verbose;
  50
  51 static volatile sig_atomic_t progress_update;
  52
  53 static void progress_interval(int signum)
  54 {
  55         progress_update = 1;
  56 }
  57
  58 static void setup_progress_signal(void)
  59 {
  60         struct sigaction sa;
  61         struct itimerval v;
  62
  63         memset(&sa, 0, sizeof(sa));
  64         sa.sa_handler = progress_interval;
  65         sigemptyset(&sa.sa_mask);
  66         sa.sa_flags = SA_RESTART;
  67         sigaction(SIGALRM, &sa, NULL);
  68
  69         v.it_interval.tv_sec = 1;
  70         v.it_interval.tv_usec = 0;
  71         v.it_value = v.it_interval;
  72         setitimer(ITIMER_REAL, &v, NULL);
  73
  74 }
  75
  76 static unsigned display_progress(unsigned n, unsigned total, unsigned last_pc)
  77 {
  78         unsigned percent = n * 100 / total;
  79         if (percent != last_pc || progress_update) {
  80                 fprintf(stderr, "%4u%% (%u/%u) done\r", percent, n, total);
  81                 progress_update = 0;
  82         }
  83         return percent;
  84 }
  85
  86 /* We always read in 4kB chunks. */
  87 static unsigned char input_buffer[4096];
  88 static unsigned long input_offset, input_len, consumed_bytes;
  89 static SHA_CTX input_ctx;
  90 static int input_fd, output_fd, mmap_fd;
  91
  92 /* Discard current buffer used content. */
  93 static void flush(void)
  94 {
  95         if (input_offset) {
  96                 if (output_fd >= 0)
  97                         write_or_die(output_fd, input_buffer, input_offset);
  98                 SHA1_Update(&input_ctx, input_buffer, input_offset);
  99                 memmove(input_buffer, input_buffer + input_offset, input_len);
 100                 input_offset = 0;
 101         }
 102 }
 103
 104 /*
 105  * Make sure at least "min" bytes are available in the buffer, and
 106  * return the pointer to the buffer.
 107  */
 108 static void *fill(int min)
 109 {
 110         if (min <= input_len)
 111                 return input_buffer + input_offset;
 112         if (min > sizeof(input_buffer))
 113                 die("cannot fill %d bytes", min);
 114         flush();
 115         do {
 116                 int ret = xread(input_fd, input_buffer + input_len,
 117                                 sizeof(input_buffer) - input_len);
 118                 if (ret <= 0) {
 119                         if (!ret)
 120                                 die("early EOF");
 121                         die("read error on input: %s", strerror(errno));
 122                 }
 123                 input_len += ret;
 124         } while (input_len < min);
 125         return input_buffer;
 126 }
 127
 128 static void use(int bytes)
 129 {
 130         if (bytes > input_len)
 131                 die("used more bytes than were available");
 132         input_len -= bytes;
 133         input_offset += bytes;
 134         consumed_bytes += bytes;
 135 }
 136
 137 static const char *open_pack_file(const char *pack_name)
 138 {
 139         if (from_stdin) {
 140                 input_fd = 0;
 141                 if (!pack_name) {
 142                         static char tmpfile[PATH_MAX];
 143                         snprintf(tmpfile, sizeof(tmpfile),
 144                                  "%s/pack_XXXXXX", get_object_directory());
 145                         output_fd = mkstemp(tmpfile);
 146                         pack_name = xstrdup(tmpfile);
 147                 } else
 148                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 149                 if (output_fd < 0)
 150                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 151                 mmap_fd = output_fd;
 152         } else {
 153                 input_fd = open(pack_name, O_RDONLY);
 154                 if (input_fd < 0)
 155                         die("cannot open packfile '%s': %s",
 156                             pack_name, strerror(errno));
 157                 output_fd = -1;
 158                 mmap_fd = input_fd;
 159         }
 160         SHA1_Init(&input_ctx);
 161         return pack_name;
 162 }
 163
 164 static void parse_pack_header(void)
 165 {
 166         struct pack_header *hdr = fill(sizeof(struct pack_header));
 167
 168         /* Header consistency check */
 169         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 170                 die("pack signature mismatch");
 171         if (!pack_version_ok(hdr->hdr_version))
 172                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 173
 174         nr_objects = ntohl(hdr->hdr_entries);
 175         use(sizeof(struct pack_header));
 176 }
 177
 178 static void bad_object(unsigned long offset, const char *format,
 179                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 180
 181 static void bad_object(unsigned long offset, const char *format, ...)
 182 {
 183         va_list params;
 184         char buf[1024];
 185
 186         va_start(params, format);
 187         vsnprintf(buf, sizeof(buf), format, params);
 188         va_end(params);
 189         die("pack has bad object at offset %lu: %s", offset, buf);
 190 }
 191
 192 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 193 {
 194         z_stream stream;
 195         void *buf = xmalloc(size);
 196
 197         memset(&stream, 0, sizeof(stream));
 198         stream.next_out = buf;
 199         stream.avail_out = size;
 200         stream.next_in = fill(1);
 201         stream.avail_in = input_len;
 202         inflateInit(&stream);
 203
 204         for (;;) {
 205                 int ret = inflate(&stream, 0);
 206                 use(input_len - stream.avail_in);
 207                 if (stream.total_out == size && ret == Z_STREAM_END)
 208                         break;
 209                 if (ret != Z_OK)
 210                         bad_object(offset, "inflate returned %d", ret);
 211                 stream.next_in = fill(1);
 212                 stream.avail_in = input_len;
 213         }
 214         inflateEnd(&stream);
 215         return buf;
 216 }
 217
 218 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 219 {
 220         unsigned char *p, c;
 221         unsigned long size, base_offset;
 222         unsigned shift;
 223
 224         obj->offset = consumed_bytes;
 225
 226         p = fill(1);
 227         c = *p;
 228         use(1);
 229         obj->type = (c >> 4) & 7;
 230         size = (c & 15);
 231         shift = 4;
 232         while (c & 0x80) {
 233                 p = fill(1);
 234                 c = *p;
 235                 use(1);
 236                 size += (c & 0x7fUL) << shift;
 237                 shift += 7;
 238         }
 239         obj->size = size;
 240
 241         switch (obj->type) {
 242         case OBJ_REF_DELTA:
 243                 hashcpy(delta_base->sha1, fill(20));
 244                 use(20);
 245                 break;
 246         case OBJ_OFS_DELTA:
 247                 memset(delta_base, 0, sizeof(*delta_base));
 248                 p = fill(1);
 249                 c = *p;
 250                 use(1);
 251                 base_offset = c & 127;
 252                 while (c & 128) {
 253                         base_offset += 1;
 254                         if (!base_offset || base_offset & ~(~0UL >> 7))
 255                                 bad_object(obj->offset, "offset value overflow for delta base object");
 256                         p = fill(1);
 257                         c = *p;
 258                         use(1);
 259                         base_offset = (base_offset << 7) + (c & 127);
 260                 }
 261                 delta_base->offset = obj->offset - base_offset;
 262                 if (delta_base->offset >= obj->offset)
 263                         bad_object(obj->offset, "delta base offset is out of bound");
 264                 break;
 265         case OBJ_COMMIT:
 266         case OBJ_TREE:
 267         case OBJ_BLOB:
 268         case OBJ_TAG:
 269                 break;
 270         default:
 271                 bad_object(obj->offset, "bad object type %d", obj->type);
 272         }
 273         obj->hdr_size = consumed_bytes - obj->offset;
 274
 275         return unpack_entry_data(obj->offset, obj->size);
 276 }
 277
 278 static void *get_data_from_pack(struct object_entry *obj)
 279 {
 280         unsigned long from = obj[0].offset + obj[0].hdr_size;
 281         unsigned long len = obj[1].offset - from;
 282         unsigned pg_offset = from % getpagesize();
 283         unsigned char *map, *data;
 284         z_stream stream;
 285         int st;
 286
 287         map = mmap(NULL, len + pg_offset, PROT_READ, MAP_PRIVATE,
 288                    mmap_fd, from - pg_offset);
 289         if (map == MAP_FAILED)
 290                 die("cannot mmap pack file: %s", strerror(errno));
 291         data = xmalloc(obj->size);
 292         memset(&stream, 0, sizeof(stream));
 293         stream.next_out = data;
 294         stream.avail_out = obj->size;
 295         stream.next_in = map + pg_offset;
 296         stream.avail_in = len;
 297         inflateInit(&stream);
 298         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 299         inflateEnd(&stream);
 300         if (st != Z_STREAM_END || stream.total_out != obj->size)
 301                 die("serious inflate inconsistency");
 302         munmap(map, len + pg_offset);
 303         return data;
 304 }
 305
 306 static int find_delta(const union delta_base *base)
 307 {
 308         int first = 0, last = nr_deltas;
 309
 310         while (first < last) {
 311                 int next = (first + last) / 2;
 312                 struct delta_entry *delta = &deltas[next];
 313                 int cmp;
 314
 315                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 316                 if (!cmp)
 317                         return next;
 318                 if (cmp < 0) {
 319                         last = next;
 320                         continue;
 321                 }
 322                 first = next+1;
 323         }
 324         return -first-1;
 325 }
 326
 327 static int find_delta_children(const union delta_base *base,
 328                                int *first_index, int *last_index)
 329 {
 330         int first = find_delta(base);
 331         int last = first;
 332         int end = nr_deltas - 1;
 333
 334         if (first < 0)
 335                 return -1;
 336         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 337                 --first;
 338         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 339                 ++last;
 340         *first_index = first;
 341         *last_index = last;
 342         return 0;
 343 }
 344
 345 static void sha1_object(const void *data, unsigned long size,
 346                         enum object_type type, unsigned char *sha1)
 347 {
 348         SHA_CTX ctx;
 349         char header[50];
 350         int header_size;
 351         const char *type_str;
 352
 353         switch (type) {
 354         case OBJ_COMMIT: type_str = commit_type; break;
 355         case OBJ_TREE:   type_str = tree_type; break;
 356         case OBJ_BLOB:   type_str = blob_type; break;
 357         case OBJ_TAG:    type_str = tag_type; break;
 358         default:
 359                 die("bad type %d", type);
 360         }
 361
 362         header_size = sprintf(header, "%s %lu", type_str, size) + 1;
 363
 364         SHA1_Init(&ctx);
 365         SHA1_Update(&ctx, header, header_size);
 366         SHA1_Update(&ctx, data, size);
 367         SHA1_Final(sha1, &ctx);
 368 }
 369
 370 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 371                           unsigned long base_size, enum object_type type)
 372 {
 373         void *delta_data;
 374         unsigned long delta_size;
 375         void *result;
 376         unsigned long result_size;
 377         union delta_base delta_base;
 378         int j, first, last;
 379
 380         delta_obj->real_type = type;
 381         delta_data = get_data_from_pack(delta_obj);
 382         delta_size = delta_obj->size;
 383         result = patch_delta(base_data, base_size, delta_data, delta_size,
 384                              &result_size);
 385         free(delta_data);
 386         if (!result)
 387                 bad_object(delta_obj->offset, "failed to apply delta");
 388         sha1_object(result, result_size, type, delta_obj->sha1);
 389         nr_resolved_deltas++;
 390
 391         hashcpy(delta_base.sha1, delta_obj->sha1);
 392         if (!find_delta_children(&delta_base, &first, &last)) {
 393                 for (j = first; j <= last; j++) {
 394                         struct object_entry *child = objects + deltas[j].obj_no;
 395                         if (child->real_type == OBJ_REF_DELTA)
 396                                 resolve_delta(child, result, result_size, type);
 397                 }
 398         }
 399
 400         memset(&delta_base, 0, sizeof(delta_base));
 401         delta_base.offset = delta_obj->offset;
 402         if (!find_delta_children(&delta_base, &first, &last)) {
 403                 for (j = first; j <= last; j++) {
 404                         struct object_entry *child = objects + deltas[j].obj_no;
 405                         if (child->real_type == OBJ_OFS_DELTA)
 406                                 resolve_delta(child, result, result_size, type);
 407                 }
 408         }
 409
 410         free(result);
 411 }
 412
 413 static int compare_delta_entry(const void *a, const void *b)
 414 {
 415         const struct delta_entry *delta_a = a;
 416         const struct delta_entry *delta_b = b;
 417         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 418 }
 419
 420 /* Parse all objects and return the pack content SHA1 hash */
 421 static void parse_pack_objects(unsigned char *sha1)
 422 {
 423         int i, percent = -1;
 424         struct delta_entry *delta = deltas;
 425         void *data;
 426         struct stat st;
 427
 428         /*
 429          * First pass:
 430          * - find locations of all objects;
 431          * - calculate SHA1 of all non-delta objects;
 432          * - remember base (SHA1 or offset) for all deltas.
 433          */
 434         if (verbose)
 435                 fprintf(stderr, "Indexing %d objects.\n", nr_objects);
 436         for (i = 0; i < nr_objects; i++) {
 437                 struct object_entry *obj = &objects[i];
 438                 data = unpack_raw_entry(obj, &delta->base);
 439                 obj->real_type = obj->type;
 440                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 441                         nr_deltas++;
 442                         delta->obj_no = i;
 443                         delta++;
 444                 } else
 445                         sha1_object(data, obj->size, obj->type, obj->sha1);
 446                 free(data);
 447                 if (verbose)
 448                         percent = display_progress(i+1, nr_objects, percent);
 449         }
 450         objects[i].offset = consumed_bytes;
 451         if (verbose)
 452                 fputc('\n', stderr);
 453
 454         /* Check pack integrity */
 455         flush();
 456         SHA1_Final(sha1, &input_ctx);
 457         if (hashcmp(fill(20), sha1))
 458                 die("pack is corrupted (SHA1 mismatch)");
 459         use(20);
 460
 461         /* If input_fd is a file, we should have reached its end now. */
 462         if (fstat(input_fd, &st))
 463                 die("cannot fstat packfile: %s", strerror(errno));
 464         if (S_ISREG(st.st_mode) && st.st_size != consumed_bytes)
 465                 die("pack has junk at the end");
 466
 467         if (!nr_deltas)
 468                 return;
 469
 470         /* Sort deltas by base SHA1/offset for fast searching */
 471         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 472               compare_delta_entry);
 473
 474         /*
 475          * Second pass:
 476          * - for all non-delta objects, look if it is used as a base for
 477          *   deltas;
 478          * - if used as a base, uncompress the object and apply all deltas,
 479          *   recursively checking if the resulting object is used as a base
 480          *   for some more deltas.
 481          */
 482         if (verbose)
 483                 fprintf(stderr, "Resolving %d deltas.\n", nr_deltas);
 484         for (i = 0; i < nr_objects; i++) {
 485                 struct object_entry *obj = &objects[i];
 486                 union delta_base base;
 487                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 488
 489                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 490                         continue;
 491                 hashcpy(base.sha1, obj->sha1);
 492                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 493                 memset(&base, 0, sizeof(base));
 494                 base.offset = obj->offset;
 495                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 496                 if (!ref && !ofs)
 497                         continue;
 498                 data = get_data_from_pack(obj);
 499                 if (ref)
 500                         for (j = ref_first; j <= ref_last; j++) {
 501                                 struct object_entry *child = objects + deltas[j].obj_no;
 502                                 if (child->real_type == OBJ_REF_DELTA)
 503                                         resolve_delta(child, data,
 504                                                       obj->size, obj->type);
 505                         }
 506                 if (ofs)
 507                         for (j = ofs_first; j <= ofs_last; j++) {
 508                                 struct object_entry *child = objects + deltas[j].obj_no;
 509                                 if (child->real_type == OBJ_OFS_DELTA)
 510                                         resolve_delta(child, data,
 511                                                       obj->size, obj->type);
 512                         }
 513                 free(data);
 514                 if (verbose)
 515                         percent = display_progress(nr_resolved_deltas,
 516                                                    nr_deltas, percent);
 517         }
 518         if (verbose && nr_resolved_deltas == nr_deltas)
 519                 fputc('\n', stderr);
 520 }
 521
 522 static int write_compressed(int fd, void *in, unsigned int size)
 523 {
 524         z_stream stream;
 525         unsigned long maxsize;
 526         void *out;
 527
 528         memset(&stream, 0, sizeof(stream));
 529         deflateInit(&stream, zlib_compression_level);
 530         maxsize = deflateBound(&stream, size);
 531         out = xmalloc(maxsize);
 532
 533         /* Compress it */
 534         stream.next_in = in;
 535         stream.avail_in = size;
 536         stream.next_out = out;
 537         stream.avail_out = maxsize;
 538         while (deflate(&stream, Z_FINISH) == Z_OK);
 539         deflateEnd(&stream);
 540
 541         size = stream.total_out;
 542         write_or_die(fd, out, size);
 543         free(out);
 544         return size;
 545 }
 546
 547 static void append_obj_to_pack(void *buf,
 548                                unsigned long size, enum object_type type)
 549 {
 550         struct object_entry *obj = &objects[nr_objects++];
 551         unsigned char header[10];
 552         unsigned long s = size;
 553         int n = 0;
 554         unsigned char c = (type << 4) | (s & 15);
 555         s >>= 4;
 556         while (s) {
 557                 header[n++] = c | 0x80;
 558                 c = s & 0x7f;
 559                 s >>= 7;
 560         }
 561         header[n++] = c;
 562         write_or_die(output_fd, header, n);
 563         obj[1].offset = obj[0].offset + n;
 564         obj[1].offset += write_compressed(output_fd, buf, size);
 565         sha1_object(buf, size, type, obj->sha1);
 566 }
 567
 568 static int delta_pos_compare(const void *_a, const void *_b)
 569 {
 570         struct delta_entry *a = *(struct delta_entry **)_a;
 571         struct delta_entry *b = *(struct delta_entry **)_b;
 572         return a->obj_no - b->obj_no;
 573 }
 574
 575 static void fix_unresolved_deltas(int nr_unresolved)
 576 {
 577         struct delta_entry **sorted_by_pos;
 578         int i, n = 0, percent = -1;
 579
 580         /*
 581          * Since many unresolved deltas may well be themselves base objects
 582          * for more unresolved deltas, we really want to include the
 583          * smallest number of base objects that would cover as much delta
 584          * as possible by picking the
 585          * trunc deltas first, allowing for other deltas to resolve without
 586          * additional base objects.  Since most base objects are to be found
 587          * before deltas depending on them, a good heuristic is to start
 588          * resolving deltas in the same order as their position in the pack.
 589          */
 590         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 591         for (i = 0; i < nr_deltas; i++) {
 592                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 593                         continue;
 594                 sorted_by_pos[n++] = &deltas[i];
 595         }
 596         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 597
 598         for (i = 0; i < n; i++) {
 599                 struct delta_entry *d = sorted_by_pos[i];
 600                 void *data;
 601                 unsigned long size;
 602                 char type[10];
 603                 enum object_type obj_type;
 604                 int j, first, last;
 605
 606                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 607                         continue;
 608                 data = read_sha1_file(d->base.sha1, type, &size);
 609                 if (!data)
 610                         continue;
 611                 if      (!strcmp(type, blob_type))   obj_type = OBJ_BLOB;
 612                 else if (!strcmp(type, tree_type))   obj_type = OBJ_TREE;
 613                 else if (!strcmp(type, commit_type)) obj_type = OBJ_COMMIT;
 614                 else if (!strcmp(type, tag_type))    obj_type = OBJ_TAG;
 615                 else die("base object %s is of type '%s'",
 616                          sha1_to_hex(d->base.sha1), type);
 617
 618                 find_delta_children(&d->base, &first, &last);
 619                 for (j = first; j <= last; j++) {
 620                         struct object_entry *child = objects + deltas[j].obj_no;
 621                         if (child->real_type == OBJ_REF_DELTA)
 622                                 resolve_delta(child, data, size, obj_type);
 623                 }
 624
 625                 append_obj_to_pack(data, size, obj_type);
 626                 free(data);
 627                 if (verbose)
 628                         percent = display_progress(nr_resolved_deltas,
 629                                                    nr_deltas, percent);
 630         }
 631         free(sorted_by_pos);
 632         if (verbose)
 633                 fputc('\n', stderr);
 634 }
 635
 636 static void readjust_pack_header_and_sha1(unsigned char *sha1)
 637 {
 638         struct pack_header hdr;
 639         SHA_CTX ctx;
 640         int size;
 641
 642         /* Rewrite pack header with updated object number */
 643         if (lseek(output_fd, 0, SEEK_SET) != 0)
 644                 die("cannot seek back: %s", strerror(errno));
 645         if (xread(output_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 646                 die("cannot read pack header back: %s", strerror(errno));
 647         hdr.hdr_entries = htonl(nr_objects);
 648         if (lseek(output_fd, 0, SEEK_SET) != 0)
 649                 die("cannot seek back: %s", strerror(errno));
 650         write_or_die(output_fd, &hdr, sizeof(hdr));
 651         if (lseek(output_fd, 0, SEEK_SET) != 0)
 652                 die("cannot seek back: %s", strerror(errno));
 653
 654         /* Recompute and store the new pack's SHA1 */
 655         SHA1_Init(&ctx);
 656         do {
 657                 unsigned char *buf[4096];
 658                 size = xread(output_fd, buf, sizeof(buf));
 659                 if (size < 0)
 660                         die("cannot read pack data back: %s", strerror(errno));
 661                 SHA1_Update(&ctx, buf, size);
 662         } while (size > 0);
 663         SHA1_Final(sha1, &ctx);
 664         write_or_die(output_fd, sha1, 20);
 665 }
 666
 667 static int sha1_compare(const void *_a, const void *_b)
 668 {
 669         struct object_entry *a = *(struct object_entry **)_a;
 670         struct object_entry *b = *(struct object_entry **)_b;
 671         return hashcmp(a->sha1, b->sha1);
 672 }
 673
 674 /*
 675  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 676  * the SHA1 hash of sorted object names.
 677  */
 678 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 679 {
 680         struct sha1file *f;
 681         struct object_entry **sorted_by_sha, **list, **last;
 682         unsigned int array[256];
 683         int i, fd;
 684         SHA_CTX ctx;
 685
 686         if (nr_objects) {
 687                 sorted_by_sha =
 688                         xcalloc(nr_objects, sizeof(struct object_entry *));
 689                 list = sorted_by_sha;
 690                 last = sorted_by_sha + nr_objects;
 691                 for (i = 0; i < nr_objects; ++i)
 692                         sorted_by_sha[i] = &objects[i];
 693                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 694                       sha1_compare);
 695
 696         }
 697         else
 698                 sorted_by_sha = list = last = NULL;
 699
 700         if (!index_name) {
 701                 static char tmpfile[PATH_MAX];
 702                 snprintf(tmpfile, sizeof(tmpfile),
 703                          "%s/index_XXXXXX", get_object_directory());
 704                 fd = mkstemp(tmpfile);
 705                 index_name = xstrdup(tmpfile);
 706         } else {
 707                 unlink(index_name);
 708                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 709         }
 710         if (fd < 0)
 711                 die("unable to create %s: %s", index_name, strerror(errno));
 712         f = sha1fd(fd, index_name);
 713
 714         /*
 715          * Write the first-level table (the list is sorted,
 716          * but we use a 256-entry lookup to be able to avoid
 717          * having to do eight extra binary search iterations).
 718          */
 719         for (i = 0; i < 256; i++) {
 720                 struct object_entry **next = list;
 721                 while (next < last) {
 722                         struct object_entry *obj = *next;
 723                         if (obj->sha1[0] != i)
 724                                 break;
 725                         next++;
 726                 }
 727                 array[i] = htonl(next - sorted_by_sha);
 728                 list = next;
 729         }
 730         sha1write(f, array, 256 * sizeof(int));
 731
 732         /* recompute the SHA1 hash of sorted object names.
 733          * currently pack-objects does not do this, but that
 734          * can be fixed.
 735          */
 736         SHA1_Init(&ctx);
 737         /*
 738          * Write the actual SHA1 entries..
 739          */
 740         list = sorted_by_sha;
 741         for (i = 0; i < nr_objects; i++) {
 742                 struct object_entry *obj = *list++;
 743                 unsigned int offset = htonl(obj->offset);
 744                 sha1write(f, &offset, 4);
 745                 sha1write(f, obj->sha1, 20);
 746                 SHA1_Update(&ctx, obj->sha1, 20);
 747         }
 748         sha1write(f, sha1, 20);
 749         sha1close(f, NULL, 1);
 750         free(sorted_by_sha);
 751         SHA1_Final(sha1, &ctx);
 752         return index_name;
 753 }
 754
 755 static void final(const char *final_pack_name, const char *curr_pack_name,
 756                   const char *final_index_name, const char *curr_index_name,
 757                   const char *keep_name, const char *keep_msg,
 758                   unsigned char *sha1)
 759 {
 760         char *report = "pack";
 761         char name[PATH_MAX];
 762         int err;
 763
 764         if (!from_stdin) {
 765                 close(input_fd);
 766         } else {
 767                 err = close(output_fd);
 768                 if (err)
 769                         die("error while closing pack file: %s", strerror(errno));
 770                 chmod(curr_pack_name, 0444);
 771         }
 772
 773         if (keep_msg) {
 774                 int keep_fd, keep_msg_len = strlen(keep_msg);
 775                 if (!keep_name) {
 776                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 777                                  get_object_directory(), sha1_to_hex(sha1));
 778                         keep_name = name;
 779                 }
 780                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 781                 if (keep_fd < 0) {
 782                         if (errno != EEXIST)
 783                                 die("cannot write keep file");
 784                 } else {
 785                         if (keep_msg_len > 0) {
 786                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 787                                 write_or_die(keep_fd, "\n", 1);
 788                         }
 789                         close(keep_fd);
 790                         report = "keep";
 791                 }
 792         }
 793
 794         if (final_pack_name != curr_pack_name) {
 795                 if (!final_pack_name) {
 796                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 797                                  get_object_directory(), sha1_to_hex(sha1));
 798                         final_pack_name = name;
 799                 }
 800                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 801                         die("cannot store pack file");
 802         }
 803
 804         chmod(curr_index_name, 0444);
 805         if (final_index_name != curr_index_name) {
 806                 if (!final_index_name) {
 807                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 808                                  get_object_directory(), sha1_to_hex(sha1));
 809                         final_index_name = name;
 810                 }
 811                 if (move_temp_to_file(curr_index_name, final_index_name))
 812                         die("cannot store index file");
 813         }
 814
 815         if (!from_stdin) {
 816                 printf("%s\n", sha1_to_hex(sha1));
 817         } else {
 818                 char buf[48];
 819                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 820                                    report, sha1_to_hex(sha1));
 821                 xwrite(1, buf, len);
 822
 823                 /*
 824                  * Let's just mimic git-unpack-objects here and write
 825                  * the last part of the input buffer to stdout.
 826                  */
 827                 while (input_len) {
 828                         err = xwrite(1, input_buffer + input_offset, input_len);
 829                         if (err <= 0)
 830                                 break;
 831                         input_len -= err;
 832                         input_offset += err;
 833                 }
 834         }
 835 }
 836
 837 int main(int argc, char **argv)
 838 {
 839         int i, fix_thin_pack = 0;
 840         const char *curr_pack, *pack_name = NULL;
 841         const char *curr_index, *index_name = NULL;
 842         const char *keep_name = NULL, *keep_msg = NULL;
 843         char *index_name_buf = NULL, *keep_name_buf = NULL;
 844         unsigned char sha1[20];
 845
 846         for (i = 1; i < argc; i++) {
 847                 const char *arg = argv[i];
 848
 849                 if (*arg == '-') {
 850                         if (!strcmp(arg, "--stdin")) {
 851                                 from_stdin = 1;
 852                         } else if (!strcmp(arg, "--fix-thin")) {
 853                                 fix_thin_pack = 1;
 854                         } else if (!strcmp(arg, "--keep")) {
 855                                 keep_msg = "";
 856                         } else if (!strncmp(arg, "--keep=", 7)) {
 857                                 keep_msg = arg + 7;
 858                         } else if (!strncmp(arg, "--pack_header=", 14)) {
 859                                 struct pack_header *hdr;
 860                                 char *c;
 861
 862                                 hdr = (struct pack_header *)input_buffer;
 863                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 864                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 865                                 if (*c != ',')
 866                                         die("bad %s", arg);
 867                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 868                                 if (*c)
 869                                         die("bad %s", arg);
 870                                 input_len = sizeof(*hdr);
 871                         } else if (!strcmp(arg, "-v")) {
 872                                 verbose = 1;
 873                         } else if (!strcmp(arg, "-o")) {
 874                                 if (index_name || (i+1) >= argc)
 875                                         usage(index_pack_usage);
 876                                 index_name = argv[++i];
 877                         } else
 878                                 usage(index_pack_usage);
 879                         continue;
 880                 }
 881
 882                 if (pack_name)
 883                         usage(index_pack_usage);
 884                 pack_name = arg;
 885         }
 886
 887         if (!pack_name && !from_stdin)
 888                 usage(index_pack_usage);
 889         if (fix_thin_pack && !from_stdin)
 890                 die("--fix-thin cannot be used without --stdin");
 891         if (!index_name && pack_name) {
 892                 int len = strlen(pack_name);
 893                 if (!has_extension(pack_name, ".pack"))
 894                         die("packfile name '%s' does not end with '.pack'",
 895                             pack_name);
 896                 index_name_buf = xmalloc(len);
 897                 memcpy(index_name_buf, pack_name, len - 5);
 898                 strcpy(index_name_buf + len - 5, ".idx");
 899                 index_name = index_name_buf;
 900         }
 901         if (keep_msg && !keep_name && pack_name) {
 902                 int len = strlen(pack_name);
 903                 if (!has_extension(pack_name, ".pack"))
 904                         die("packfile name '%s' does not end with '.pack'",
 905                             pack_name);
 906                 keep_name_buf = xmalloc(len);
 907                 memcpy(keep_name_buf, pack_name, len - 5);
 908                 strcpy(keep_name_buf + len - 5, ".keep");
 909                 keep_name = keep_name_buf;
 910         }
 911
 912         curr_pack = open_pack_file(pack_name);
 913         parse_pack_header();
 914         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 915         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 916         if (verbose)
 917                 setup_progress_signal();
 918         parse_pack_objects(sha1);
 919         if (nr_deltas != nr_resolved_deltas) {
 920                 if (fix_thin_pack) {
 921                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 922                         int nr_objects_initial = nr_objects;
 923                         if (nr_unresolved <= 0)
 924                                 die("confusion beyond insanity");
 925                         objects = xrealloc(objects,
 926                                            (nr_objects + nr_unresolved + 1)
 927                                            * sizeof(*objects));
 928                         fix_unresolved_deltas(nr_unresolved);
 929                         if (verbose)
 930                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 931                                         nr_objects - nr_objects_initial);
 932                         readjust_pack_header_and_sha1(sha1);
 933                 }
 934                 if (nr_deltas != nr_resolved_deltas)
 935                         die("pack has %d unresolved deltas",
 936                             nr_deltas - nr_resolved_deltas);
 937         } else {
 938                 /* Flush remaining pack final 20-byte SHA1. */
 939                 flush();
 940         }
 941         free(deltas);
 942         curr_index = write_index_file(index_name, sha1);
 943         final(pack_name, curr_pack,
 944                 index_name, curr_index,
 945                 keep_name, keep_msg,
 946                 sha1);
 947         free(objects);
 948         free(index_name_buf);
 949         free(keep_name_buf);
 950
 951         return 0;
 952 }