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
  10 static const char index_pack_usage[] =
  11 "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  12
  13 struct object_entry
  14 {
  15         off_t offset;
  16         unsigned long size;
  17         unsigned int hdr_size;
  18         enum object_type type;
  19         enum object_type real_type;
  20         unsigned char sha1[20];
  21 };
  22
  23 union delta_base {
  24         unsigned char sha1[20];
  25         off_t offset;
  26 };
  27
  28 /*
  29  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  30  * to memcmp() only the first 20 bytes.
  31  */
  32 #define UNION_BASE_SZ   20
  33
  34 struct delta_entry
  35 {
  36         union delta_base base;
  37         int obj_no;
  38 };
  39
  40 static struct object_entry *objects;
  41 static struct delta_entry *deltas;
  42 static int nr_objects;
  43 static int nr_deltas;
  44 static int nr_resolved_deltas;
  45
  46 static int from_stdin;
  47 static int verbose;
  48
  49 static volatile sig_atomic_t progress_update;
  50
  51 static void progress_interval(int signum)
  52 {
  53         progress_update = 1;
  54 }
  55
  56 static void setup_progress_signal(void)
  57 {
  58         struct sigaction sa;
  59         struct itimerval v;
  60
  61         memset(&sa, 0, sizeof(sa));
  62         sa.sa_handler = progress_interval;
  63         sigemptyset(&sa.sa_mask);
  64         sa.sa_flags = SA_RESTART;
  65         sigaction(SIGALRM, &sa, NULL);
  66
  67         v.it_interval.tv_sec = 1;
  68         v.it_interval.tv_usec = 0;
  69         v.it_value = v.it_interval;
  70         setitimer(ITIMER_REAL, &v, NULL);
  71
  72 }
  73
  74 static unsigned display_progress(unsigned n, unsigned total, unsigned last_pc)
  75 {
  76         unsigned percent = n * 100 / total;
  77         if (percent != last_pc || progress_update) {
  78                 fprintf(stderr, "%4u%% (%u/%u) done\r", percent, n, total);
  79                 progress_update = 0;
  80         }
  81         return percent;
  82 }
  83
  84 /* We always read in 4kB chunks. */
  85 static unsigned char input_buffer[4096];
  86 static unsigned int input_offset, input_len;
  87 static off_t consumed_bytes;
  88 static SHA_CTX input_ctx;
  89 static int input_fd, output_fd, pack_fd;
  90
  91 /* Discard current buffer used content. */
  92 static void flush(void)
  93 {
  94         if (input_offset) {
  95                 if (output_fd >= 0)
  96                         write_or_die(output_fd, input_buffer, input_offset);
  97                 SHA1_Update(&input_ctx, input_buffer, input_offset);
  98                 memmove(input_buffer, input_buffer + input_offset, input_len);
  99                 input_offset = 0;
 100         }
 101 }
 102
 103 /*
 104  * Make sure at least "min" bytes are available in the buffer, and
 105  * return the pointer to the buffer.
 106  */
 107 static void *fill(int min)
 108 {
 109         if (min <= input_len)
 110                 return input_buffer + input_offset;
 111         if (min > sizeof(input_buffer))
 112                 die("cannot fill %d bytes", min);
 113         flush();
 114         do {
 115                 int ret = xread(input_fd, input_buffer + input_len,
 116                                 sizeof(input_buffer) - input_len);
 117                 if (ret <= 0) {
 118                         if (!ret)
 119                                 die("early EOF");
 120                         die("read error on input: %s", strerror(errno));
 121                 }
 122                 input_len += ret;
 123         } while (input_len < min);
 124         return input_buffer;
 125 }
 126
 127 static void use(int bytes)
 128 {
 129         if (bytes > input_len)
 130                 die("used more bytes than were available");
 131         input_len -= bytes;
 132         input_offset += bytes;
 133
 134         /* make sure off_t is sufficiently large not to wrap */
 135         if (consumed_bytes > consumed_bytes + bytes)
 136                 die("pack too large for current definition of off_t");
 137         consumed_bytes += bytes;
 138 }
 139
 140 static const char *open_pack_file(const char *pack_name)
 141 {
 142         if (from_stdin) {
 143                 input_fd = 0;
 144                 if (!pack_name) {
 145                         static char tmpfile[PATH_MAX];
 146                         snprintf(tmpfile, sizeof(tmpfile),
 147                                  "%s/tmp_pack_XXXXXX", get_object_directory());
 148                         output_fd = mkstemp(tmpfile);
 149                         pack_name = xstrdup(tmpfile);
 150                 } else
 151                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 152                 if (output_fd < 0)
 153                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 154                 pack_fd = output_fd;
 155         } else {
 156                 input_fd = open(pack_name, O_RDONLY);
 157                 if (input_fd < 0)
 158                         die("cannot open packfile '%s': %s",
 159                             pack_name, strerror(errno));
 160                 output_fd = -1;
 161                 pack_fd = input_fd;
 162         }
 163         SHA1_Init(&input_ctx);
 164         return pack_name;
 165 }
 166
 167 static void parse_pack_header(void)
 168 {
 169         struct pack_header *hdr = fill(sizeof(struct pack_header));
 170
 171         /* Header consistency check */
 172         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 173                 die("pack signature mismatch");
 174         if (!pack_version_ok(hdr->hdr_version))
 175                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 176
 177         nr_objects = ntohl(hdr->hdr_entries);
 178         use(sizeof(struct pack_header));
 179 }
 180
 181 static void bad_object(unsigned long offset, const char *format,
 182                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 183
 184 static void bad_object(unsigned long offset, const char *format, ...)
 185 {
 186         va_list params;
 187         char buf[1024];
 188
 189         va_start(params, format);
 190         vsnprintf(buf, sizeof(buf), format, params);
 191         va_end(params);
 192         die("pack has bad object at offset %lu: %s", offset, buf);
 193 }
 194
 195 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 196 {
 197         z_stream stream;
 198         void *buf = xmalloc(size);
 199
 200         memset(&stream, 0, sizeof(stream));
 201         stream.next_out = buf;
 202         stream.avail_out = size;
 203         stream.next_in = fill(1);
 204         stream.avail_in = input_len;
 205         inflateInit(&stream);
 206
 207         for (;;) {
 208                 int ret = inflate(&stream, 0);
 209                 use(input_len - stream.avail_in);
 210                 if (stream.total_out == size && ret == Z_STREAM_END)
 211                         break;
 212                 if (ret != Z_OK)
 213                         bad_object(offset, "inflate returned %d", ret);
 214                 stream.next_in = fill(1);
 215                 stream.avail_in = input_len;
 216         }
 217         inflateEnd(&stream);
 218         return buf;
 219 }
 220
 221 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 222 {
 223         unsigned char *p, c;
 224         unsigned long size;
 225         off_t base_offset;
 226         unsigned shift;
 227
 228         obj->offset = consumed_bytes;
 229
 230         p = fill(1);
 231         c = *p;
 232         use(1);
 233         obj->type = (c >> 4) & 7;
 234         size = (c & 15);
 235         shift = 4;
 236         while (c & 0x80) {
 237                 p = fill(1);
 238                 c = *p;
 239                 use(1);
 240                 size += (c & 0x7fUL) << shift;
 241                 shift += 7;
 242         }
 243         obj->size = size;
 244
 245         switch (obj->type) {
 246         case OBJ_REF_DELTA:
 247                 hashcpy(delta_base->sha1, fill(20));
 248                 use(20);
 249                 break;
 250         case OBJ_OFS_DELTA:
 251                 memset(delta_base, 0, sizeof(*delta_base));
 252                 p = fill(1);
 253                 c = *p;
 254                 use(1);
 255                 base_offset = c & 127;
 256                 while (c & 128) {
 257                         base_offset += 1;
 258                         if (!base_offset || MSB(base_offset, 7))
 259                                 bad_object(obj->offset, "offset value overflow for delta base object");
 260                         p = fill(1);
 261                         c = *p;
 262                         use(1);
 263                         base_offset = (base_offset << 7) + (c & 127);
 264                 }
 265                 delta_base->offset = obj->offset - base_offset;
 266                 if (delta_base->offset >= obj->offset)
 267                         bad_object(obj->offset, "delta base offset is out of bound");
 268                 break;
 269         case OBJ_COMMIT:
 270         case OBJ_TREE:
 271         case OBJ_BLOB:
 272         case OBJ_TAG:
 273                 break;
 274         default:
 275                 bad_object(obj->offset, "unknown object type %d", obj->type);
 276         }
 277         obj->hdr_size = consumed_bytes - obj->offset;
 278
 279         return unpack_entry_data(obj->offset, obj->size);
 280 }
 281
 282 static void *get_data_from_pack(struct object_entry *obj)
 283 {
 284         unsigned long from = obj[0].offset + obj[0].hdr_size;
 285         unsigned long len = obj[1].offset - from;
 286         unsigned long rdy = 0;
 287         unsigned char *src, *data;
 288         z_stream stream;
 289         int st;
 290
 291         src = xmalloc(len);
 292         data = src;
 293         do {
 294                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 295                 if (n <= 0)
 296                         die("cannot pread pack file: %s", strerror(errno));
 297                 rdy += n;
 298         } while (rdy < len);
 299         data = xmalloc(obj->size);
 300         memset(&stream, 0, sizeof(stream));
 301         stream.next_out = data;
 302         stream.avail_out = obj->size;
 303         stream.next_in = src;
 304         stream.avail_in = len;
 305         inflateInit(&stream);
 306         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 307         inflateEnd(&stream);
 308         if (st != Z_STREAM_END || stream.total_out != obj->size)
 309                 die("serious inflate inconsistency");
 310         free(src);
 311         return data;
 312 }
 313
 314 static int find_delta(const union delta_base *base)
 315 {
 316         int first = 0, last = nr_deltas;
 317
 318         while (first < last) {
 319                 int next = (first + last) / 2;
 320                 struct delta_entry *delta = &deltas[next];
 321                 int cmp;
 322
 323                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 324                 if (!cmp)
 325                         return next;
 326                 if (cmp < 0) {
 327                         last = next;
 328                         continue;
 329                 }
 330                 first = next+1;
 331         }
 332         return -first-1;
 333 }
 334
 335 static int find_delta_children(const union delta_base *base,
 336                                int *first_index, int *last_index)
 337 {
 338         int first = find_delta(base);
 339         int last = first;
 340         int end = nr_deltas - 1;
 341
 342         if (first < 0)
 343                 return -1;
 344         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 345                 --first;
 346         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 347                 ++last;
 348         *first_index = first;
 349         *last_index = last;
 350         return 0;
 351 }
 352
 353 static void sha1_object(const void *data, unsigned long size,
 354                         enum object_type type, unsigned char *sha1)
 355 {
 356         hash_sha1_file(data, size, typename(type), sha1);
 357         if (has_sha1_file(sha1)) {
 358                 void *has_data;
 359                 enum object_type has_type;
 360                 unsigned long has_size;
 361                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 362                 if (!has_data)
 363                         die("cannot read existing object %s", sha1_to_hex(sha1));
 364                 if (size != has_size || type != has_type ||
 365                     memcmp(data, has_data, size) != 0)
 366                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 367                 free(has_data);
 368         }
 369 }
 370
 371 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 372                           unsigned long base_size, enum object_type type)
 373 {
 374         void *delta_data;
 375         unsigned long delta_size;
 376         void *result;
 377         unsigned long result_size;
 378         union delta_base delta_base;
 379         int j, first, last;
 380
 381         delta_obj->real_type = type;
 382         delta_data = get_data_from_pack(delta_obj);
 383         delta_size = delta_obj->size;
 384         result = patch_delta(base_data, base_size, delta_data, delta_size,
 385                              &result_size);
 386         free(delta_data);
 387         if (!result)
 388                 bad_object(delta_obj->offset, "failed to apply delta");
 389         sha1_object(result, result_size, type, delta_obj->sha1);
 390         nr_resolved_deltas++;
 391
 392         hashcpy(delta_base.sha1, delta_obj->sha1);
 393         if (!find_delta_children(&delta_base, &first, &last)) {
 394                 for (j = first; j <= last; j++) {
 395                         struct object_entry *child = objects + deltas[j].obj_no;
 396                         if (child->real_type == OBJ_REF_DELTA)
 397                                 resolve_delta(child, result, result_size, type);
 398                 }
 399         }
 400
 401         memset(&delta_base, 0, sizeof(delta_base));
 402         delta_base.offset = delta_obj->offset;
 403         if (!find_delta_children(&delta_base, &first, &last)) {
 404                 for (j = first; j <= last; j++) {
 405                         struct object_entry *child = objects + deltas[j].obj_no;
 406                         if (child->real_type == OBJ_OFS_DELTA)
 407                                 resolve_delta(child, result, result_size, type);
 408                 }
 409         }
 410
 411         free(result);
 412 }
 413
 414 static int compare_delta_entry(const void *a, const void *b)
 415 {
 416         const struct delta_entry *delta_a = a;
 417         const struct delta_entry *delta_b = b;
 418         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 419 }
 420
 421 /* Parse all objects and return the pack content SHA1 hash */
 422 static void parse_pack_objects(unsigned char *sha1)
 423 {
 424         int i, percent = -1;
 425         struct delta_entry *delta = deltas;
 426         void *data;
 427         struct stat st;
 428
 429         /*
 430          * First pass:
 431          * - find locations of all objects;
 432          * - calculate SHA1 of all non-delta objects;
 433          * - remember base (SHA1 or offset) for all deltas.
 434          */
 435         if (verbose)
 436                 fprintf(stderr, "Indexing %d objects.\n", nr_objects);
 437         for (i = 0; i < nr_objects; i++) {
 438                 struct object_entry *obj = &objects[i];
 439                 data = unpack_raw_entry(obj, &delta->base);
 440                 obj->real_type = obj->type;
 441                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 442                         nr_deltas++;
 443                         delta->obj_no = i;
 444                         delta++;
 445                 } else
 446                         sha1_object(data, obj->size, obj->type, obj->sha1);
 447                 free(data);
 448                 if (verbose)
 449                         percent = display_progress(i+1, nr_objects, percent);
 450         }
 451         objects[i].offset = consumed_bytes;
 452         if (verbose)
 453                 fputc('\n', stderr);
 454
 455         /* Check pack integrity */
 456         flush();
 457         SHA1_Final(sha1, &input_ctx);
 458         if (hashcmp(fill(20), sha1))
 459                 die("pack is corrupted (SHA1 mismatch)");
 460         use(20);
 461
 462         /* If input_fd is a file, we should have reached its end now. */
 463         if (fstat(input_fd, &st))
 464                 die("cannot fstat packfile: %s", strerror(errno));
 465         if (S_ISREG(st.st_mode) &&
 466                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 467                 die("pack has junk at the end");
 468
 469         if (!nr_deltas)
 470                 return;
 471
 472         /* Sort deltas by base SHA1/offset for fast searching */
 473         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 474               compare_delta_entry);
 475
 476         /*
 477          * Second pass:
 478          * - for all non-delta objects, look if it is used as a base for
 479          *   deltas;
 480          * - if used as a base, uncompress the object and apply all deltas,
 481          *   recursively checking if the resulting object is used as a base
 482          *   for some more deltas.
 483          */
 484         if (verbose)
 485                 fprintf(stderr, "Resolving %d deltas.\n", nr_deltas);
 486         for (i = 0; i < nr_objects; i++) {
 487                 struct object_entry *obj = &objects[i];
 488                 union delta_base base;
 489                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 490
 491                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 492                         continue;
 493                 hashcpy(base.sha1, obj->sha1);
 494                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 495                 memset(&base, 0, sizeof(base));
 496                 base.offset = obj->offset;
 497                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 498                 if (!ref && !ofs)
 499                         continue;
 500                 data = get_data_from_pack(obj);
 501                 if (ref)
 502                         for (j = ref_first; j <= ref_last; j++) {
 503                                 struct object_entry *child = objects + deltas[j].obj_no;
 504                                 if (child->real_type == OBJ_REF_DELTA)
 505                                         resolve_delta(child, data,
 506                                                       obj->size, obj->type);
 507                         }
 508                 if (ofs)
 509                         for (j = ofs_first; j <= ofs_last; j++) {
 510                                 struct object_entry *child = objects + deltas[j].obj_no;
 511                                 if (child->real_type == OBJ_OFS_DELTA)
 512                                         resolve_delta(child, data,
 513                                                       obj->size, obj->type);
 514                         }
 515                 free(data);
 516                 if (verbose)
 517                         percent = display_progress(nr_resolved_deltas,
 518                                                    nr_deltas, percent);
 519         }
 520         if (verbose && nr_resolved_deltas == nr_deltas)
 521                 fputc('\n', stderr);
 522 }
 523
 524 static int write_compressed(int fd, void *in, unsigned int size)
 525 {
 526         z_stream stream;
 527         unsigned long maxsize;
 528         void *out;
 529
 530         memset(&stream, 0, sizeof(stream));
 531         deflateInit(&stream, zlib_compression_level);
 532         maxsize = deflateBound(&stream, size);
 533         out = xmalloc(maxsize);
 534
 535         /* Compress it */
 536         stream.next_in = in;
 537         stream.avail_in = size;
 538         stream.next_out = out;
 539         stream.avail_out = maxsize;
 540         while (deflate(&stream, Z_FINISH) == Z_OK);
 541         deflateEnd(&stream);
 542
 543         size = stream.total_out;
 544         write_or_die(fd, out, size);
 545         free(out);
 546         return size;
 547 }
 548
 549 static void append_obj_to_pack(const unsigned char *sha1, void *buf,
 550                                unsigned long size, enum object_type type)
 551 {
 552         struct object_entry *obj = &objects[nr_objects++];
 553         unsigned char header[10];
 554         unsigned long s = size;
 555         int n = 0;
 556         unsigned char c = (type << 4) | (s & 15);
 557         s >>= 4;
 558         while (s) {
 559                 header[n++] = c | 0x80;
 560                 c = s & 0x7f;
 561                 s >>= 7;
 562         }
 563         header[n++] = c;
 564         write_or_die(output_fd, header, n);
 565         obj[1].offset = obj[0].offset + n;
 566         obj[1].offset += write_compressed(output_fd, buf, size);
 567         hashcpy(obj->sha1, sha1);
 568 }
 569
 570 static int delta_pos_compare(const void *_a, const void *_b)
 571 {
 572         struct delta_entry *a = *(struct delta_entry **)_a;
 573         struct delta_entry *b = *(struct delta_entry **)_b;
 574         return a->obj_no - b->obj_no;
 575 }
 576
 577 static void fix_unresolved_deltas(int nr_unresolved)
 578 {
 579         struct delta_entry **sorted_by_pos;
 580         int i, n = 0, percent = -1;
 581
 582         /*
 583          * Since many unresolved deltas may well be themselves base objects
 584          * for more unresolved deltas, we really want to include the
 585          * smallest number of base objects that would cover as much delta
 586          * as possible by picking the
 587          * trunc deltas first, allowing for other deltas to resolve without
 588          * additional base objects.  Since most base objects are to be found
 589          * before deltas depending on them, a good heuristic is to start
 590          * resolving deltas in the same order as their position in the pack.
 591          */
 592         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 593         for (i = 0; i < nr_deltas; i++) {
 594                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 595                         continue;
 596                 sorted_by_pos[n++] = &deltas[i];
 597         }
 598         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 599
 600         for (i = 0; i < n; i++) {
 601                 struct delta_entry *d = sorted_by_pos[i];
 602                 void *data;
 603                 unsigned long size;
 604                 enum object_type type;
 605                 int j, first, last;
 606
 607                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 608                         continue;
 609                 data = read_sha1_file(d->base.sha1, &type, &size);
 610                 if (!data)
 611                         continue;
 612
 613                 find_delta_children(&d->base, &first, &last);
 614                 for (j = first; j <= last; j++) {
 615                         struct object_entry *child = objects + deltas[j].obj_no;
 616                         if (child->real_type == OBJ_REF_DELTA)
 617                                 resolve_delta(child, data, size, type);
 618                 }
 619
 620                 if (check_sha1_signature(d->base.sha1, data, size, typename(type)))
 621                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 622                 append_obj_to_pack(d->base.sha1, data, size, type);
 623                 free(data);
 624                 if (verbose)
 625                         percent = display_progress(nr_resolved_deltas,
 626                                                    nr_deltas, percent);
 627         }
 628         free(sorted_by_pos);
 629         if (verbose)
 630                 fputc('\n', stderr);
 631 }
 632
 633 static void readjust_pack_header_and_sha1(unsigned char *sha1)
 634 {
 635         struct pack_header hdr;
 636         SHA_CTX ctx;
 637         int size;
 638
 639         /* Rewrite pack header with updated object number */
 640         if (lseek(output_fd, 0, SEEK_SET) != 0)
 641                 die("cannot seek back: %s", strerror(errno));
 642         if (read_in_full(output_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 643                 die("cannot read pack header back: %s", strerror(errno));
 644         hdr.hdr_entries = htonl(nr_objects);
 645         if (lseek(output_fd, 0, SEEK_SET) != 0)
 646                 die("cannot seek back: %s", strerror(errno));
 647         write_or_die(output_fd, &hdr, sizeof(hdr));
 648         if (lseek(output_fd, 0, SEEK_SET) != 0)
 649                 die("cannot seek back: %s", strerror(errno));
 650
 651         /* Recompute and store the new pack's SHA1 */
 652         SHA1_Init(&ctx);
 653         do {
 654                 unsigned char *buf[4096];
 655                 size = xread(output_fd, buf, sizeof(buf));
 656                 if (size < 0)
 657                         die("cannot read pack data back: %s", strerror(errno));
 658                 SHA1_Update(&ctx, buf, size);
 659         } while (size > 0);
 660         SHA1_Final(sha1, &ctx);
 661         write_or_die(output_fd, sha1, 20);
 662 }
 663
 664 static int sha1_compare(const void *_a, const void *_b)
 665 {
 666         struct object_entry *a = *(struct object_entry **)_a;
 667         struct object_entry *b = *(struct object_entry **)_b;
 668         return hashcmp(a->sha1, b->sha1);
 669 }
 670
 671 /*
 672  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 673  * the SHA1 hash of sorted object names.
 674  */
 675 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 676 {
 677         struct sha1file *f;
 678         struct object_entry **sorted_by_sha, **list, **last;
 679         unsigned int array[256];
 680         int i, fd;
 681         SHA_CTX ctx;
 682
 683         if (nr_objects) {
 684                 sorted_by_sha =
 685                         xcalloc(nr_objects, sizeof(struct object_entry *));
 686                 list = sorted_by_sha;
 687                 last = sorted_by_sha + nr_objects;
 688                 for (i = 0; i < nr_objects; ++i)
 689                         sorted_by_sha[i] = &objects[i];
 690                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 691                       sha1_compare);
 692
 693         }
 694         else
 695                 sorted_by_sha = list = last = NULL;
 696
 697         if (!index_name) {
 698                 static char tmpfile[PATH_MAX];
 699                 snprintf(tmpfile, sizeof(tmpfile),
 700                          "%s/tmp_idx_XXXXXX", get_object_directory());
 701                 fd = mkstemp(tmpfile);
 702                 index_name = xstrdup(tmpfile);
 703         } else {
 704                 unlink(index_name);
 705                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 706         }
 707         if (fd < 0)
 708                 die("unable to create %s: %s", index_name, strerror(errno));
 709         f = sha1fd(fd, index_name);
 710
 711         /*
 712          * Write the first-level table (the list is sorted,
 713          * but we use a 256-entry lookup to be able to avoid
 714          * having to do eight extra binary search iterations).
 715          */
 716         for (i = 0; i < 256; i++) {
 717                 struct object_entry **next = list;
 718                 while (next < last) {
 719                         struct object_entry *obj = *next;
 720                         if (obj->sha1[0] != i)
 721                                 break;
 722                         next++;
 723                 }
 724                 array[i] = htonl(next - sorted_by_sha);
 725                 list = next;
 726         }
 727         sha1write(f, array, 256 * sizeof(int));
 728
 729         /* recompute the SHA1 hash of sorted object names.
 730          * currently pack-objects does not do this, but that
 731          * can be fixed.
 732          */
 733         SHA1_Init(&ctx);
 734         /*
 735          * Write the actual SHA1 entries..
 736          */
 737         list = sorted_by_sha;
 738         for (i = 0; i < nr_objects; i++) {
 739                 struct object_entry *obj = *list++;
 740                 unsigned int offset = htonl(obj->offset);
 741                 sha1write(f, &offset, 4);
 742                 sha1write(f, obj->sha1, 20);
 743                 SHA1_Update(&ctx, obj->sha1, 20);
 744         }
 745         sha1write(f, sha1, 20);
 746         sha1close(f, NULL, 1);
 747         free(sorted_by_sha);
 748         SHA1_Final(sha1, &ctx);
 749         return index_name;
 750 }
 751
 752 static void final(const char *final_pack_name, const char *curr_pack_name,
 753                   const char *final_index_name, const char *curr_index_name,
 754                   const char *keep_name, const char *keep_msg,
 755                   unsigned char *sha1)
 756 {
 757         const char *report = "pack";
 758         char name[PATH_MAX];
 759         int err;
 760
 761         if (!from_stdin) {
 762                 close(input_fd);
 763         } else {
 764                 err = close(output_fd);
 765                 if (err)
 766                         die("error while closing pack file: %s", strerror(errno));
 767                 chmod(curr_pack_name, 0444);
 768         }
 769
 770         if (keep_msg) {
 771                 int keep_fd, keep_msg_len = strlen(keep_msg);
 772                 if (!keep_name) {
 773                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 774                                  get_object_directory(), sha1_to_hex(sha1));
 775                         keep_name = name;
 776                 }
 777                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 778                 if (keep_fd < 0) {
 779                         if (errno != EEXIST)
 780                                 die("cannot write keep file");
 781                 } else {
 782                         if (keep_msg_len > 0) {
 783                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 784                                 write_or_die(keep_fd, "\n", 1);
 785                         }
 786                         close(keep_fd);
 787                         report = "keep";
 788                 }
 789         }
 790
 791         if (final_pack_name != curr_pack_name) {
 792                 if (!final_pack_name) {
 793                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 794                                  get_object_directory(), sha1_to_hex(sha1));
 795                         final_pack_name = name;
 796                 }
 797                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 798                         die("cannot store pack file");
 799         }
 800
 801         chmod(curr_index_name, 0444);
 802         if (final_index_name != curr_index_name) {
 803                 if (!final_index_name) {
 804                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 805                                  get_object_directory(), sha1_to_hex(sha1));
 806                         final_index_name = name;
 807                 }
 808                 if (move_temp_to_file(curr_index_name, final_index_name))
 809                         die("cannot store index file");
 810         }
 811
 812         if (!from_stdin) {
 813                 printf("%s\n", sha1_to_hex(sha1));
 814         } else {
 815                 char buf[48];
 816                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 817                                    report, sha1_to_hex(sha1));
 818                 write_or_die(1, buf, len);
 819
 820                 /*
 821                  * Let's just mimic git-unpack-objects here and write
 822                  * the last part of the input buffer to stdout.
 823                  */
 824                 while (input_len) {
 825                         err = xwrite(1, input_buffer + input_offset, input_len);
 826                         if (err <= 0)
 827                                 break;
 828                         input_len -= err;
 829                         input_offset += err;
 830                 }
 831         }
 832 }
 833
 834 int main(int argc, char **argv)
 835 {
 836         int i, fix_thin_pack = 0;
 837         const char *curr_pack, *pack_name = NULL;
 838         const char *curr_index, *index_name = NULL;
 839         const char *keep_name = NULL, *keep_msg = NULL;
 840         char *index_name_buf = NULL, *keep_name_buf = NULL;
 841         unsigned char sha1[20];
 842
 843         for (i = 1; i < argc; i++) {
 844                 const char *arg = argv[i];
 845
 846                 if (*arg == '-') {
 847                         if (!strcmp(arg, "--stdin")) {
 848                                 from_stdin = 1;
 849                         } else if (!strcmp(arg, "--fix-thin")) {
 850                                 fix_thin_pack = 1;
 851                         } else if (!strcmp(arg, "--keep")) {
 852                                 keep_msg = "";
 853                         } else if (!prefixcmp(arg, "--keep=")) {
 854                                 keep_msg = arg + 7;
 855                         } else if (!prefixcmp(arg, "--pack_header=")) {
 856                                 struct pack_header *hdr;
 857                                 char *c;
 858
 859                                 hdr = (struct pack_header *)input_buffer;
 860                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 861                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 862                                 if (*c != ',')
 863                                         die("bad %s", arg);
 864                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 865                                 if (*c)
 866                                         die("bad %s", arg);
 867                                 input_len = sizeof(*hdr);
 868                         } else if (!strcmp(arg, "-v")) {
 869                                 verbose = 1;
 870                         } else if (!strcmp(arg, "-o")) {
 871                                 if (index_name || (i+1) >= argc)
 872                                         usage(index_pack_usage);
 873                                 index_name = argv[++i];
 874                         } else
 875                                 usage(index_pack_usage);
 876                         continue;
 877                 }
 878
 879                 if (pack_name)
 880                         usage(index_pack_usage);
 881                 pack_name = arg;
 882         }
 883
 884         if (!pack_name && !from_stdin)
 885                 usage(index_pack_usage);
 886         if (fix_thin_pack && !from_stdin)
 887                 die("--fix-thin cannot be used without --stdin");
 888         if (!index_name && pack_name) {
 889                 int len = strlen(pack_name);
 890                 if (!has_extension(pack_name, ".pack"))
 891                         die("packfile name '%s' does not end with '.pack'",
 892                             pack_name);
 893                 index_name_buf = xmalloc(len);
 894                 memcpy(index_name_buf, pack_name, len - 5);
 895                 strcpy(index_name_buf + len - 5, ".idx");
 896                 index_name = index_name_buf;
 897         }
 898         if (keep_msg && !keep_name && pack_name) {
 899                 int len = strlen(pack_name);
 900                 if (!has_extension(pack_name, ".pack"))
 901                         die("packfile name '%s' does not end with '.pack'",
 902                             pack_name);
 903                 keep_name_buf = xmalloc(len);
 904                 memcpy(keep_name_buf, pack_name, len - 5);
 905                 strcpy(keep_name_buf + len - 5, ".keep");
 906                 keep_name = keep_name_buf;
 907         }
 908
 909         curr_pack = open_pack_file(pack_name);
 910         parse_pack_header();
 911         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 912         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 913         if (verbose)
 914                 setup_progress_signal();
 915         parse_pack_objects(sha1);
 916         if (nr_deltas != nr_resolved_deltas) {
 917                 if (fix_thin_pack) {
 918                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 919                         int nr_objects_initial = nr_objects;
 920                         if (nr_unresolved <= 0)
 921                                 die("confusion beyond insanity");
 922                         objects = xrealloc(objects,
 923                                            (nr_objects + nr_unresolved + 1)
 924                                            * sizeof(*objects));
 925                         fix_unresolved_deltas(nr_unresolved);
 926                         if (verbose)
 927                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 928                                         nr_objects - nr_objects_initial);
 929                         readjust_pack_header_and_sha1(sha1);
 930                 }
 931                 if (nr_deltas != nr_resolved_deltas)
 932                         die("pack has %d unresolved deltas",
 933                             nr_deltas - nr_resolved_deltas);
 934         } else {
 935                 /* Flush remaining pack final 20-byte SHA1. */
 936                 flush();
 937         }
 938         free(deltas);
 939         curr_index = write_index_file(index_name, sha1);
 940         final(pack_name, curr_pack,
 941                 index_name, curr_index,
 942                 keep_name, keep_msg,
 943                 sha1);
 944         free(objects);
 945         free(index_name_buf);
 946         free(keep_name_buf);
 947
 948         return 0;
 949 }