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