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/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 char *src, *data;
 281         z_stream stream;
 282         int st;
 283
 284         src = xmalloc(len);
 285         if (pread(pack_fd, src, len, from) != len)
 286                 die("cannot pread pack file: %s", strerror(errno));
 287         data = xmalloc(obj->size);
 288         memset(&stream, 0, sizeof(stream));
 289         stream.next_out = data;
 290         stream.avail_out = obj->size;
 291         stream.next_in = src;
 292         stream.avail_in = len;
 293         inflateInit(&stream);
 294         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 295         inflateEnd(&stream);
 296         if (st != Z_STREAM_END || stream.total_out != obj->size)
 297                 die("serious inflate inconsistency");
 298         free(src);
 299         return data;
 300 }
 301
 302 static int find_delta(const union delta_base *base)
 303 {
 304         int first = 0, last = nr_deltas;
 305
 306         while (first < last) {
 307                 int next = (first + last) / 2;
 308                 struct delta_entry *delta = &deltas[next];
 309                 int cmp;
 310
 311                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 312                 if (!cmp)
 313                         return next;
 314                 if (cmp < 0) {
 315                         last = next;
 316                         continue;
 317                 }
 318                 first = next+1;
 319         }
 320         return -first-1;
 321 }
 322
 323 static int find_delta_children(const union delta_base *base,
 324                                int *first_index, int *last_index)
 325 {
 326         int first = find_delta(base);
 327         int last = first;
 328         int end = nr_deltas - 1;
 329
 330         if (first < 0)
 331                 return -1;
 332         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 333                 --first;
 334         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 335                 ++last;
 336         *first_index = first;
 337         *last_index = last;
 338         return 0;
 339 }
 340
 341 static void sha1_object(const void *data, unsigned long size,
 342                         enum object_type type, unsigned char *sha1)
 343 {
 344         SHA_CTX ctx;
 345         char header[50];
 346         int header_size;
 347         const char *type_str;
 348
 349         switch (type) {
 350         case OBJ_COMMIT: type_str = commit_type; break;
 351         case OBJ_TREE:   type_str = tree_type; break;
 352         case OBJ_BLOB:   type_str = blob_type; break;
 353         case OBJ_TAG:    type_str = tag_type; break;
 354         default:
 355                 die("bad type %d", type);
 356         }
 357
 358         header_size = sprintf(header, "%s %lu", type_str, size) + 1;
 359
 360         SHA1_Init(&ctx);
 361         SHA1_Update(&ctx, header, header_size);
 362         SHA1_Update(&ctx, data, size);
 363         SHA1_Final(sha1, &ctx);
 364 }
 365
 366 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 367                           unsigned long base_size, enum object_type type)
 368 {
 369         void *delta_data;
 370         unsigned long delta_size;
 371         void *result;
 372         unsigned long result_size;
 373         union delta_base delta_base;
 374         int j, first, last;
 375
 376         delta_obj->real_type = type;
 377         delta_data = get_data_from_pack(delta_obj);
 378         delta_size = delta_obj->size;
 379         result = patch_delta(base_data, base_size, delta_data, delta_size,
 380                              &result_size);
 381         free(delta_data);
 382         if (!result)
 383                 bad_object(delta_obj->offset, "failed to apply delta");
 384         sha1_object(result, result_size, type, delta_obj->sha1);
 385         nr_resolved_deltas++;
 386
 387         hashcpy(delta_base.sha1, delta_obj->sha1);
 388         if (!find_delta_children(&delta_base, &first, &last)) {
 389                 for (j = first; j <= last; j++) {
 390                         struct object_entry *child = objects + deltas[j].obj_no;
 391                         if (child->real_type == OBJ_REF_DELTA)
 392                                 resolve_delta(child, result, result_size, type);
 393                 }
 394         }
 395
 396         memset(&delta_base, 0, sizeof(delta_base));
 397         delta_base.offset = delta_obj->offset;
 398         if (!find_delta_children(&delta_base, &first, &last)) {
 399                 for (j = first; j <= last; j++) {
 400                         struct object_entry *child = objects + deltas[j].obj_no;
 401                         if (child->real_type == OBJ_OFS_DELTA)
 402                                 resolve_delta(child, result, result_size, type);
 403                 }
 404         }
 405
 406         free(result);
 407 }
 408
 409 static int compare_delta_entry(const void *a, const void *b)
 410 {
 411         const struct delta_entry *delta_a = a;
 412         const struct delta_entry *delta_b = b;
 413         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 414 }
 415
 416 /* Parse all objects and return the pack content SHA1 hash */
 417 static void parse_pack_objects(unsigned char *sha1)
 418 {
 419         int i, percent = -1;
 420         struct delta_entry *delta = deltas;
 421         void *data;
 422         struct stat st;
 423
 424         /*
 425          * First pass:
 426          * - find locations of all objects;
 427          * - calculate SHA1 of all non-delta objects;
 428          * - remember base (SHA1 or offset) for all deltas.
 429          */
 430         if (verbose)
 431                 fprintf(stderr, "Indexing %d objects.\n", nr_objects);
 432         for (i = 0; i < nr_objects; i++) {
 433                 struct object_entry *obj = &objects[i];
 434                 data = unpack_raw_entry(obj, &delta->base);
 435                 obj->real_type = obj->type;
 436                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 437                         nr_deltas++;
 438                         delta->obj_no = i;
 439                         delta++;
 440                 } else
 441                         sha1_object(data, obj->size, obj->type, obj->sha1);
 442                 free(data);
 443                 if (verbose)
 444                         percent = display_progress(i+1, nr_objects, percent);
 445         }
 446         objects[i].offset = consumed_bytes;
 447         if (verbose)
 448                 fputc('\n', stderr);
 449
 450         /* Check pack integrity */
 451         flush();
 452         SHA1_Final(sha1, &input_ctx);
 453         if (hashcmp(fill(20), sha1))
 454                 die("pack is corrupted (SHA1 mismatch)");
 455         use(20);
 456
 457         /* If input_fd is a file, we should have reached its end now. */
 458         if (fstat(input_fd, &st))
 459                 die("cannot fstat packfile: %s", strerror(errno));
 460         if (S_ISREG(st.st_mode) && st.st_size != consumed_bytes)
 461                 die("pack has junk at the end");
 462
 463         if (!nr_deltas)
 464                 return;
 465
 466         /* Sort deltas by base SHA1/offset for fast searching */
 467         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 468               compare_delta_entry);
 469
 470         /*
 471          * Second pass:
 472          * - for all non-delta objects, look if it is used as a base for
 473          *   deltas;
 474          * - if used as a base, uncompress the object and apply all deltas,
 475          *   recursively checking if the resulting object is used as a base
 476          *   for some more deltas.
 477          */
 478         if (verbose)
 479                 fprintf(stderr, "Resolving %d deltas.\n", nr_deltas);
 480         for (i = 0; i < nr_objects; i++) {
 481                 struct object_entry *obj = &objects[i];
 482                 union delta_base base;
 483                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 484
 485                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 486                         continue;
 487                 hashcpy(base.sha1, obj->sha1);
 488                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 489                 memset(&base, 0, sizeof(base));
 490                 base.offset = obj->offset;
 491                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 492                 if (!ref && !ofs)
 493                         continue;
 494                 data = get_data_from_pack(obj);
 495                 if (ref)
 496                         for (j = ref_first; j <= ref_last; j++) {
 497                                 struct object_entry *child = objects + deltas[j].obj_no;
 498                                 if (child->real_type == OBJ_REF_DELTA)
 499                                         resolve_delta(child, data,
 500                                                       obj->size, obj->type);
 501                         }
 502                 if (ofs)
 503                         for (j = ofs_first; j <= ofs_last; j++) {
 504                                 struct object_entry *child = objects + deltas[j].obj_no;
 505                                 if (child->real_type == OBJ_OFS_DELTA)
 506                                         resolve_delta(child, data,
 507                                                       obj->size, obj->type);
 508                         }
 509                 free(data);
 510                 if (verbose)
 511                         percent = display_progress(nr_resolved_deltas,
 512                                                    nr_deltas, percent);
 513         }
 514         if (verbose && nr_resolved_deltas == nr_deltas)
 515                 fputc('\n', stderr);
 516 }
 517
 518 static int write_compressed(int fd, void *in, unsigned int size)
 519 {
 520         z_stream stream;
 521         unsigned long maxsize;
 522         void *out;
 523
 524         memset(&stream, 0, sizeof(stream));
 525         deflateInit(&stream, zlib_compression_level);
 526         maxsize = deflateBound(&stream, size);
 527         out = xmalloc(maxsize);
 528
 529         /* Compress it */
 530         stream.next_in = in;
 531         stream.avail_in = size;
 532         stream.next_out = out;
 533         stream.avail_out = maxsize;
 534         while (deflate(&stream, Z_FINISH) == Z_OK);
 535         deflateEnd(&stream);
 536
 537         size = stream.total_out;
 538         write_or_die(fd, out, size);
 539         free(out);
 540         return size;
 541 }
 542
 543 static void append_obj_to_pack(void *buf,
 544                                unsigned long size, enum object_type type)
 545 {
 546         struct object_entry *obj = &objects[nr_objects++];
 547         unsigned char header[10];
 548         unsigned long s = size;
 549         int n = 0;
 550         unsigned char c = (type << 4) | (s & 15);
 551         s >>= 4;
 552         while (s) {
 553                 header[n++] = c | 0x80;
 554                 c = s & 0x7f;
 555                 s >>= 7;
 556         }
 557         header[n++] = c;
 558         write_or_die(output_fd, header, n);
 559         obj[1].offset = obj[0].offset + n;
 560         obj[1].offset += write_compressed(output_fd, buf, size);
 561         sha1_object(buf, size, type, obj->sha1);
 562 }
 563
 564 static int delta_pos_compare(const void *_a, const void *_b)
 565 {
 566         struct delta_entry *a = *(struct delta_entry **)_a;
 567         struct delta_entry *b = *(struct delta_entry **)_b;
 568         return a->obj_no - b->obj_no;
 569 }
 570
 571 static void fix_unresolved_deltas(int nr_unresolved)
 572 {
 573         struct delta_entry **sorted_by_pos;
 574         int i, n = 0, percent = -1;
 575
 576         /*
 577          * Since many unresolved deltas may well be themselves base objects
 578          * for more unresolved deltas, we really want to include the
 579          * smallest number of base objects that would cover as much delta
 580          * as possible by picking the
 581          * trunc deltas first, allowing for other deltas to resolve without
 582          * additional base objects.  Since most base objects are to be found
 583          * before deltas depending on them, a good heuristic is to start
 584          * resolving deltas in the same order as their position in the pack.
 585          */
 586         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 587         for (i = 0; i < nr_deltas; i++) {
 588                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 589                         continue;
 590                 sorted_by_pos[n++] = &deltas[i];
 591         }
 592         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 593
 594         for (i = 0; i < n; i++) {
 595                 struct delta_entry *d = sorted_by_pos[i];
 596                 void *data;
 597                 unsigned long size;
 598                 char type[10];
 599                 enum object_type obj_type;
 600                 int j, first, last;
 601
 602                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 603                         continue;
 604                 data = read_sha1_file(d->base.sha1, type, &size);
 605                 if (!data)
 606                         continue;
 607                 if      (!strcmp(type, blob_type))   obj_type = OBJ_BLOB;
 608                 else if (!strcmp(type, tree_type))   obj_type = OBJ_TREE;
 609                 else if (!strcmp(type, commit_type)) obj_type = OBJ_COMMIT;
 610                 else if (!strcmp(type, tag_type))    obj_type = OBJ_TAG;
 611                 else die("base object %s is of type '%s'",
 612                          sha1_to_hex(d->base.sha1), type);
 613
 614                 find_delta_children(&d->base, &first, &last);
 615                 for (j = first; j <= last; j++) {
 616                         struct object_entry *child = objects + deltas[j].obj_no;
 617                         if (child->real_type == OBJ_REF_DELTA)
 618                                 resolve_delta(child, data, size, obj_type);
 619                 }
 620
 621                 append_obj_to_pack(data, size, obj_type);
 622                 free(data);
 623                 if (verbose)
 624                         percent = display_progress(nr_resolved_deltas,
 625                                                    nr_deltas, percent);
 626         }
 627         free(sorted_by_pos);
 628         if (verbose)
 629                 fputc('\n', stderr);
 630 }
 631
 632 static void readjust_pack_header_and_sha1(unsigned char *sha1)
 633 {
 634         struct pack_header hdr;
 635         SHA_CTX ctx;
 636         int size;
 637
 638         /* Rewrite pack header with updated object number */
 639         if (lseek(output_fd, 0, SEEK_SET) != 0)
 640                 die("cannot seek back: %s", strerror(errno));
 641         if (xread(output_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 642                 die("cannot read pack header back: %s", strerror(errno));
 643         hdr.hdr_entries = htonl(nr_objects);
 644         if (lseek(output_fd, 0, SEEK_SET) != 0)
 645                 die("cannot seek back: %s", strerror(errno));
 646         write_or_die(output_fd, &hdr, sizeof(hdr));
 647         if (lseek(output_fd, 0, SEEK_SET) != 0)
 648                 die("cannot seek back: %s", strerror(errno));
 649
 650         /* Recompute and store the new pack's SHA1 */
 651         SHA1_Init(&ctx);
 652         do {
 653                 unsigned char *buf[4096];
 654                 size = xread(output_fd, buf, sizeof(buf));
 655                 if (size < 0)
 656                         die("cannot read pack data back: %s", strerror(errno));
 657                 SHA1_Update(&ctx, buf, size);
 658         } while (size > 0);
 659         SHA1_Final(sha1, &ctx);
 660         write_or_die(output_fd, sha1, 20);
 661 }
 662
 663 static int sha1_compare(const void *_a, const void *_b)
 664 {
 665         struct object_entry *a = *(struct object_entry **)_a;
 666         struct object_entry *b = *(struct object_entry **)_b;
 667         return hashcmp(a->sha1, b->sha1);
 668 }
 669
 670 /*
 671  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 672  * the SHA1 hash of sorted object names.
 673  */
 674 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 675 {
 676         struct sha1file *f;
 677         struct object_entry **sorted_by_sha, **list, **last;
 678         unsigned int array[256];
 679         int i, fd;
 680         SHA_CTX ctx;
 681
 682         if (nr_objects) {
 683                 sorted_by_sha =
 684                         xcalloc(nr_objects, sizeof(struct object_entry *));
 685                 list = sorted_by_sha;
 686                 last = sorted_by_sha + nr_objects;
 687                 for (i = 0; i < nr_objects; ++i)
 688                         sorted_by_sha[i] = &objects[i];
 689                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 690                       sha1_compare);
 691
 692         }
 693         else
 694                 sorted_by_sha = list = last = NULL;
 695
 696         if (!index_name) {
 697                 static char tmpfile[PATH_MAX];
 698                 snprintf(tmpfile, sizeof(tmpfile),
 699                          "%s/index_XXXXXX", get_object_directory());
 700                 fd = mkstemp(tmpfile);
 701                 index_name = xstrdup(tmpfile);
 702         } else {
 703                 unlink(index_name);
 704                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 705         }
 706         if (fd < 0)
 707                 die("unable to create %s: %s", index_name, strerror(errno));
 708         f = sha1fd(fd, index_name);
 709
 710         /*
 711          * Write the first-level table (the list is sorted,
 712          * but we use a 256-entry lookup to be able to avoid
 713          * having to do eight extra binary search iterations).
 714          */
 715         for (i = 0; i < 256; i++) {
 716                 struct object_entry **next = list;
 717                 while (next < last) {
 718                         struct object_entry *obj = *next;
 719                         if (obj->sha1[0] != i)
 720                                 break;
 721                         next++;
 722                 }
 723                 array[i] = htonl(next - sorted_by_sha);
 724                 list = next;
 725         }
 726         sha1write(f, array, 256 * sizeof(int));
 727
 728         /* recompute the SHA1 hash of sorted object names.
 729          * currently pack-objects does not do this, but that
 730          * can be fixed.
 731          */
 732         SHA1_Init(&ctx);
 733         /*
 734          * Write the actual SHA1 entries..
 735          */
 736         list = sorted_by_sha;
 737         for (i = 0; i < nr_objects; i++) {
 738                 struct object_entry *obj = *list++;
 739                 unsigned int offset = htonl(obj->offset);
 740                 sha1write(f, &offset, 4);
 741                 sha1write(f, obj->sha1, 20);
 742                 SHA1_Update(&ctx, obj->sha1, 20);
 743         }
 744         sha1write(f, sha1, 20);
 745         sha1close(f, NULL, 1);
 746         free(sorted_by_sha);
 747         SHA1_Final(sha1, &ctx);
 748         return index_name;
 749 }
 750
 751 static void final(const char *final_pack_name, const char *curr_pack_name,
 752                   const char *final_index_name, const char *curr_index_name,
 753                   const char *keep_name, const char *keep_msg,
 754                   unsigned char *sha1)
 755 {
 756         char *report = "pack";
 757         char name[PATH_MAX];
 758         int err;
 759
 760         if (!from_stdin) {
 761                 close(input_fd);
 762         } else {
 763                 err = close(output_fd);
 764                 if (err)
 765                         die("error while closing pack file: %s", strerror(errno));
 766                 chmod(curr_pack_name, 0444);
 767         }
 768
 769         if (keep_msg) {
 770                 int keep_fd, keep_msg_len = strlen(keep_msg);
 771                 if (!keep_name) {
 772                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 773                                  get_object_directory(), sha1_to_hex(sha1));
 774                         keep_name = name;
 775                 }
 776                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 777                 if (keep_fd < 0) {
 778                         if (errno != EEXIST)
 779                                 die("cannot write keep file");
 780                 } else {
 781                         if (keep_msg_len > 0) {
 782                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 783                                 write_or_die(keep_fd, "\n", 1);
 784                         }
 785                         close(keep_fd);
 786                         report = "keep";
 787                 }
 788         }
 789
 790         if (final_pack_name != curr_pack_name) {
 791                 if (!final_pack_name) {
 792                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 793                                  get_object_directory(), sha1_to_hex(sha1));
 794                         final_pack_name = name;
 795                 }
 796                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 797                         die("cannot store pack file");
 798         }
 799
 800         chmod(curr_index_name, 0444);
 801         if (final_index_name != curr_index_name) {
 802                 if (!final_index_name) {
 803                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 804                                  get_object_directory(), sha1_to_hex(sha1));
 805                         final_index_name = name;
 806                 }
 807                 if (move_temp_to_file(curr_index_name, final_index_name))
 808                         die("cannot store index file");
 809         }
 810
 811         if (!from_stdin) {
 812                 printf("%s\n", sha1_to_hex(sha1));
 813         } else {
 814                 char buf[48];
 815                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 816                                    report, sha1_to_hex(sha1));
 817                 xwrite(1, buf, len);
 818
 819                 /*
 820                  * Let's just mimic git-unpack-objects here and write
 821                  * the last part of the input buffer to stdout.
 822                  */
 823                 while (input_len) {
 824                         err = xwrite(1, input_buffer + input_offset, input_len);
 825                         if (err <= 0)
 826                                 break;
 827                         input_len -= err;
 828                         input_offset += err;
 829                 }
 830         }
 831 }
 832
 833 int main(int argc, char **argv)
 834 {
 835         int i, fix_thin_pack = 0;
 836         const char *curr_pack, *pack_name = NULL;
 837         const char *curr_index, *index_name = NULL;
 838         const char *keep_name = NULL, *keep_msg = NULL;
 839         char *index_name_buf = NULL, *keep_name_buf = NULL;
 840         unsigned char sha1[20];
 841
 842         for (i = 1; i < argc; i++) {
 843                 const char *arg = argv[i];
 844
 845                 if (*arg == '-') {
 846                         if (!strcmp(arg, "--stdin")) {
 847                                 from_stdin = 1;
 848                         } else if (!strcmp(arg, "--fix-thin")) {
 849                                 fix_thin_pack = 1;
 850                         } else if (!strcmp(arg, "--keep")) {
 851                                 keep_msg = "";
 852                         } else if (!strncmp(arg, "--keep=", 7)) {
 853                                 keep_msg = arg + 7;
 854                         } else if (!strncmp(arg, "--pack_header=", 14)) {
 855                                 struct pack_header *hdr;
 856                                 char *c;
 857
 858                                 hdr = (struct pack_header *)input_buffer;
 859                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 860                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 861                                 if (*c != ',')
 862                                         die("bad %s", arg);
 863                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 864                                 if (*c)
 865                                         die("bad %s", arg);
 866                                 input_len = sizeof(*hdr);
 867                         } else if (!strcmp(arg, "-v")) {
 868                                 verbose = 1;
 869                         } else if (!strcmp(arg, "-o")) {
 870                                 if (index_name || (i+1) >= argc)
 871                                         usage(index_pack_usage);
 872                                 index_name = argv[++i];
 873                         } else
 874                                 usage(index_pack_usage);
 875                         continue;
 876                 }
 877
 878                 if (pack_name)
 879                         usage(index_pack_usage);
 880                 pack_name = arg;
 881         }
 882
 883         if (!pack_name && !from_stdin)
 884                 usage(index_pack_usage);
 885         if (fix_thin_pack && !from_stdin)
 886                 die("--fix-thin cannot be used without --stdin");
 887         if (!index_name && pack_name) {
 888                 int len = strlen(pack_name);
 889                 if (!has_extension(pack_name, ".pack"))
 890                         die("packfile name '%s' does not end with '.pack'",
 891                             pack_name);
 892                 index_name_buf = xmalloc(len);
 893                 memcpy(index_name_buf, pack_name, len - 5);
 894                 strcpy(index_name_buf + len - 5, ".idx");
 895                 index_name = index_name_buf;
 896         }
 897         if (keep_msg && !keep_name && pack_name) {
 898                 int len = strlen(pack_name);
 899                 if (!has_extension(pack_name, ".pack"))
 900                         die("packfile name '%s' does not end with '.pack'",
 901                             pack_name);
 902                 keep_name_buf = xmalloc(len);
 903                 memcpy(keep_name_buf, pack_name, len - 5);
 904                 strcpy(keep_name_buf + len - 5, ".keep");
 905                 keep_name = keep_name_buf;
 906         }
 907
 908         curr_pack = open_pack_file(pack_name);
 909         parse_pack_header();
 910         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 911         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 912         if (verbose)
 913                 setup_progress_signal();
 914         parse_pack_objects(sha1);
 915         if (nr_deltas != nr_resolved_deltas) {
 916                 if (fix_thin_pack) {
 917                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 918                         int nr_objects_initial = nr_objects;
 919                         if (nr_unresolved <= 0)
 920                                 die("confusion beyond insanity");
 921                         objects = xrealloc(objects,
 922                                            (nr_objects + nr_unresolved + 1)
 923                                            * sizeof(*objects));
 924                         fix_unresolved_deltas(nr_unresolved);
 925                         if (verbose)
 926                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 927                                         nr_objects - nr_objects_initial);
 928                         readjust_pack_header_and_sha1(sha1);
 929                 }
 930                 if (nr_deltas != nr_resolved_deltas)
 931                         die("pack has %d unresolved deltas",
 932                             nr_deltas - nr_resolved_deltas);
 933         } else {
 934                 /* Flush remaining pack final 20-byte SHA1. */
 935                 flush();
 936         }
 937         free(deltas);
 938         curr_index = write_index_file(index_name, sha1);
 939         final(pack_name, curr_pack,
 940                 index_name, curr_index,
 941                 keep_name, keep_msg,
 942                 sha1);
 943         free(objects);
 944         free(index_name_buf);
 945         free(keep_name_buf);
 946
 947         return 0;
 948 }