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 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         SHA_CTX ctx;
 351         char header[50];
 352         int header_size;
 353         const char *type_str;
 354
 355         switch (type) {
 356         case OBJ_COMMIT: type_str = commit_type; break;
 357         case OBJ_TREE:   type_str = tree_type; break;
 358         case OBJ_BLOB:   type_str = blob_type; break;
 359         case OBJ_TAG:    type_str = tag_type; break;
 360         default:
 361                 die("bad type %d", type);
 362         }
 363
 364         header_size = sprintf(header, "%s %lu", type_str, size) + 1;
 365
 366         SHA1_Init(&ctx);
 367         SHA1_Update(&ctx, header, header_size);
 368         SHA1_Update(&ctx, data, size);
 369         SHA1_Final(sha1, &ctx);
 370 }
 371
 372 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 373                           unsigned long base_size, enum object_type type)
 374 {
 375         void *delta_data;
 376         unsigned long delta_size;
 377         void *result;
 378         unsigned long result_size;
 379         union delta_base delta_base;
 380         int j, first, last;
 381
 382         delta_obj->real_type = type;
 383         delta_data = get_data_from_pack(delta_obj);
 384         delta_size = delta_obj->size;
 385         result = patch_delta(base_data, base_size, delta_data, delta_size,
 386                              &result_size);
 387         free(delta_data);
 388         if (!result)
 389                 bad_object(delta_obj->offset, "failed to apply delta");
 390         sha1_object(result, result_size, type, delta_obj->sha1);
 391         nr_resolved_deltas++;
 392
 393         hashcpy(delta_base.sha1, delta_obj->sha1);
 394         if (!find_delta_children(&delta_base, &first, &last)) {
 395                 for (j = first; j <= last; j++) {
 396                         struct object_entry *child = objects + deltas[j].obj_no;
 397                         if (child->real_type == OBJ_REF_DELTA)
 398                                 resolve_delta(child, result, result_size, type);
 399                 }
 400         }
 401
 402         memset(&delta_base, 0, sizeof(delta_base));
 403         delta_base.offset = delta_obj->offset;
 404         if (!find_delta_children(&delta_base, &first, &last)) {
 405                 for (j = first; j <= last; j++) {
 406                         struct object_entry *child = objects + deltas[j].obj_no;
 407                         if (child->real_type == OBJ_OFS_DELTA)
 408                                 resolve_delta(child, result, result_size, type);
 409                 }
 410         }
 411
 412         free(result);
 413 }
 414
 415 static int compare_delta_entry(const void *a, const void *b)
 416 {
 417         const struct delta_entry *delta_a = a;
 418         const struct delta_entry *delta_b = b;
 419         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 420 }
 421
 422 /* Parse all objects and return the pack content SHA1 hash */
 423 static void parse_pack_objects(unsigned char *sha1)
 424 {
 425         int i, percent = -1;
 426         struct delta_entry *delta = deltas;
 427         void *data;
 428         struct stat st;
 429
 430         /*
 431          * First pass:
 432          * - find locations of all objects;
 433          * - calculate SHA1 of all non-delta objects;
 434          * - remember base (SHA1 or offset) for all deltas.
 435          */
 436         if (verbose)
 437                 fprintf(stderr, "Indexing %d objects.\n", nr_objects);
 438         for (i = 0; i < nr_objects; i++) {
 439                 struct object_entry *obj = &objects[i];
 440                 data = unpack_raw_entry(obj, &delta->base);
 441                 obj->real_type = obj->type;
 442                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 443                         nr_deltas++;
 444                         delta->obj_no = i;
 445                         delta++;
 446                 } else
 447                         sha1_object(data, obj->size, obj->type, obj->sha1);
 448                 free(data);
 449                 if (verbose)
 450                         percent = display_progress(i+1, nr_objects, percent);
 451         }
 452         objects[i].offset = consumed_bytes;
 453         if (verbose)
 454                 fputc('\n', stderr);
 455
 456         /* Check pack integrity */
 457         flush();
 458         SHA1_Final(sha1, &input_ctx);
 459         if (hashcmp(fill(20), sha1))
 460                 die("pack is corrupted (SHA1 mismatch)");
 461         use(20);
 462
 463         /* If input_fd is a file, we should have reached its end now. */
 464         if (fstat(input_fd, &st))
 465                 die("cannot fstat packfile: %s", strerror(errno));
 466         if (S_ISREG(st.st_mode) && st.st_size != consumed_bytes)
 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(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         sha1_object(buf, size, type, obj->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                 char type[10];
 605                 enum object_type obj_type;
 606                 int j, first, last;
 607
 608                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 609                         continue;
 610                 data = read_sha1_file(d->base.sha1, type, &size);
 611                 if (!data)
 612                         continue;
 613                 if      (!strcmp(type, blob_type))   obj_type = OBJ_BLOB;
 614                 else if (!strcmp(type, tree_type))   obj_type = OBJ_TREE;
 615                 else if (!strcmp(type, commit_type)) obj_type = OBJ_COMMIT;
 616                 else if (!strcmp(type, tag_type))    obj_type = OBJ_TAG;
 617                 else die("base object %s is of type '%s'",
 618                          sha1_to_hex(d->base.sha1), type);
 619
 620                 find_delta_children(&d->base, &first, &last);
 621                 for (j = first; j <= last; j++) {
 622                         struct object_entry *child = objects + deltas[j].obj_no;
 623                         if (child->real_type == OBJ_REF_DELTA)
 624                                 resolve_delta(child, data, size, obj_type);
 625                 }
 626
 627                 append_obj_to_pack(data, size, obj_type);
 628                 free(data);
 629                 if (verbose)
 630                         percent = display_progress(nr_resolved_deltas,
 631                                                    nr_deltas, percent);
 632         }
 633         free(sorted_by_pos);
 634         if (verbose)
 635                 fputc('\n', stderr);
 636 }
 637
 638 static void readjust_pack_header_and_sha1(unsigned char *sha1)
 639 {
 640         struct pack_header hdr;
 641         SHA_CTX ctx;
 642         int size;
 643
 644         /* Rewrite pack header with updated object number */
 645         if (lseek(output_fd, 0, SEEK_SET) != 0)
 646                 die("cannot seek back: %s", strerror(errno));
 647         if (read_in_full(output_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 648                 die("cannot read pack header back: %s", strerror(errno));
 649         hdr.hdr_entries = htonl(nr_objects);
 650         if (lseek(output_fd, 0, SEEK_SET) != 0)
 651                 die("cannot seek back: %s", strerror(errno));
 652         write_or_die(output_fd, &hdr, sizeof(hdr));
 653         if (lseek(output_fd, 0, SEEK_SET) != 0)
 654                 die("cannot seek back: %s", strerror(errno));
 655
 656         /* Recompute and store the new pack's SHA1 */
 657         SHA1_Init(&ctx);
 658         do {
 659                 unsigned char *buf[4096];
 660                 size = xread(output_fd, buf, sizeof(buf));
 661                 if (size < 0)
 662                         die("cannot read pack data back: %s", strerror(errno));
 663                 SHA1_Update(&ctx, buf, size);
 664         } while (size > 0);
 665         SHA1_Final(sha1, &ctx);
 666         write_or_die(output_fd, sha1, 20);
 667 }
 668
 669 static int sha1_compare(const void *_a, const void *_b)
 670 {
 671         struct object_entry *a = *(struct object_entry **)_a;
 672         struct object_entry *b = *(struct object_entry **)_b;
 673         return hashcmp(a->sha1, b->sha1);
 674 }
 675
 676 /*
 677  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 678  * the SHA1 hash of sorted object names.
 679  */
 680 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 681 {
 682         struct sha1file *f;
 683         struct object_entry **sorted_by_sha, **list, **last;
 684         unsigned int array[256];
 685         int i, fd;
 686         SHA_CTX ctx;
 687
 688         if (nr_objects) {
 689                 sorted_by_sha =
 690                         xcalloc(nr_objects, sizeof(struct object_entry *));
 691                 list = sorted_by_sha;
 692                 last = sorted_by_sha + nr_objects;
 693                 for (i = 0; i < nr_objects; ++i)
 694                         sorted_by_sha[i] = &objects[i];
 695                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 696                       sha1_compare);
 697
 698         }
 699         else
 700                 sorted_by_sha = list = last = NULL;
 701
 702         if (!index_name) {
 703                 static char tmpfile[PATH_MAX];
 704                 snprintf(tmpfile, sizeof(tmpfile),
 705                          "%s/index_XXXXXX", get_object_directory());
 706                 fd = mkstemp(tmpfile);
 707                 index_name = xstrdup(tmpfile);
 708         } else {
 709                 unlink(index_name);
 710                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 711         }
 712         if (fd < 0)
 713                 die("unable to create %s: %s", index_name, strerror(errno));
 714         f = sha1fd(fd, index_name);
 715
 716         /*
 717          * Write the first-level table (the list is sorted,
 718          * but we use a 256-entry lookup to be able to avoid
 719          * having to do eight extra binary search iterations).
 720          */
 721         for (i = 0; i < 256; i++) {
 722                 struct object_entry **next = list;
 723                 while (next < last) {
 724                         struct object_entry *obj = *next;
 725                         if (obj->sha1[0] != i)
 726                                 break;
 727                         next++;
 728                 }
 729                 array[i] = htonl(next - sorted_by_sha);
 730                 list = next;
 731         }
 732         sha1write(f, array, 256 * sizeof(int));
 733
 734         /* recompute the SHA1 hash of sorted object names.
 735          * currently pack-objects does not do this, but that
 736          * can be fixed.
 737          */
 738         SHA1_Init(&ctx);
 739         /*
 740          * Write the actual SHA1 entries..
 741          */
 742         list = sorted_by_sha;
 743         for (i = 0; i < nr_objects; i++) {
 744                 struct object_entry *obj = *list++;
 745                 unsigned int offset = htonl(obj->offset);
 746                 sha1write(f, &offset, 4);
 747                 sha1write(f, obj->sha1, 20);
 748                 SHA1_Update(&ctx, obj->sha1, 20);
 749         }
 750         sha1write(f, sha1, 20);
 751         sha1close(f, NULL, 1);
 752         free(sorted_by_sha);
 753         SHA1_Final(sha1, &ctx);
 754         return index_name;
 755 }
 756
 757 static void final(const char *final_pack_name, const char *curr_pack_name,
 758                   const char *final_index_name, const char *curr_index_name,
 759                   const char *keep_name, const char *keep_msg,
 760                   unsigned char *sha1)
 761 {
 762         char *report = "pack";
 763         char name[PATH_MAX];
 764         int err;
 765
 766         if (!from_stdin) {
 767                 close(input_fd);
 768         } else {
 769                 err = close(output_fd);
 770                 if (err)
 771                         die("error while closing pack file: %s", strerror(errno));
 772                 chmod(curr_pack_name, 0444);
 773         }
 774
 775         if (keep_msg) {
 776                 int keep_fd, keep_msg_len = strlen(keep_msg);
 777                 if (!keep_name) {
 778                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 779                                  get_object_directory(), sha1_to_hex(sha1));
 780                         keep_name = name;
 781                 }
 782                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 783                 if (keep_fd < 0) {
 784                         if (errno != EEXIST)
 785                                 die("cannot write keep file");
 786                 } else {
 787                         if (keep_msg_len > 0) {
 788                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 789                                 write_or_die(keep_fd, "\n", 1);
 790                         }
 791                         close(keep_fd);
 792                         report = "keep";
 793                 }
 794         }
 795
 796         if (final_pack_name != curr_pack_name) {
 797                 if (!final_pack_name) {
 798                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 799                                  get_object_directory(), sha1_to_hex(sha1));
 800                         final_pack_name = name;
 801                 }
 802                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 803                         die("cannot store pack file");
 804         }
 805
 806         chmod(curr_index_name, 0444);
 807         if (final_index_name != curr_index_name) {
 808                 if (!final_index_name) {
 809                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 810                                  get_object_directory(), sha1_to_hex(sha1));
 811                         final_index_name = name;
 812                 }
 813                 if (move_temp_to_file(curr_index_name, final_index_name))
 814                         die("cannot store index file");
 815         }
 816
 817         if (!from_stdin) {
 818                 printf("%s\n", sha1_to_hex(sha1));
 819         } else {
 820                 char buf[48];
 821                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 822                                    report, sha1_to_hex(sha1));
 823                 write_or_die(1, buf, len);
 824
 825                 /*
 826                  * Let's just mimic git-unpack-objects here and write
 827                  * the last part of the input buffer to stdout.
 828                  */
 829                 while (input_len) {
 830                         err = xwrite(1, input_buffer + input_offset, input_len);
 831                         if (err <= 0)
 832                                 break;
 833                         input_len -= err;
 834                         input_offset += err;
 835                 }
 836         }
 837 }
 838
 839 int main(int argc, char **argv)
 840 {
 841         int i, fix_thin_pack = 0;
 842         const char *curr_pack, *pack_name = NULL;
 843         const char *curr_index, *index_name = NULL;
 844         const char *keep_name = NULL, *keep_msg = NULL;
 845         char *index_name_buf = NULL, *keep_name_buf = NULL;
 846         unsigned char sha1[20];
 847
 848         for (i = 1; i < argc; i++) {
 849                 const char *arg = argv[i];
 850
 851                 if (*arg == '-') {
 852                         if (!strcmp(arg, "--stdin")) {
 853                                 from_stdin = 1;
 854                         } else if (!strcmp(arg, "--fix-thin")) {
 855                                 fix_thin_pack = 1;
 856                         } else if (!strcmp(arg, "--keep")) {
 857                                 keep_msg = "";
 858                         } else if (!strncmp(arg, "--keep=", 7)) {
 859                                 keep_msg = arg + 7;
 860                         } else if (!strncmp(arg, "--pack_header=", 14)) {
 861                                 struct pack_header *hdr;
 862                                 char *c;
 863
 864                                 hdr = (struct pack_header *)input_buffer;
 865                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 866                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 867                                 if (*c != ',')
 868                                         die("bad %s", arg);
 869                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 870                                 if (*c)
 871                                         die("bad %s", arg);
 872                                 input_len = sizeof(*hdr);
 873                         } else if (!strcmp(arg, "-v")) {
 874                                 verbose = 1;
 875                         } else if (!strcmp(arg, "-o")) {
 876                                 if (index_name || (i+1) >= argc)
 877                                         usage(index_pack_usage);
 878                                 index_name = argv[++i];
 879                         } else
 880                                 usage(index_pack_usage);
 881                         continue;
 882                 }
 883
 884                 if (pack_name)
 885                         usage(index_pack_usage);
 886                 pack_name = arg;
 887         }
 888
 889         if (!pack_name && !from_stdin)
 890                 usage(index_pack_usage);
 891         if (fix_thin_pack && !from_stdin)
 892                 die("--fix-thin cannot be used without --stdin");
 893         if (!index_name && pack_name) {
 894                 int len = strlen(pack_name);
 895                 if (!has_extension(pack_name, ".pack"))
 896                         die("packfile name '%s' does not end with '.pack'",
 897                             pack_name);
 898                 index_name_buf = xmalloc(len);
 899                 memcpy(index_name_buf, pack_name, len - 5);
 900                 strcpy(index_name_buf + len - 5, ".idx");
 901                 index_name = index_name_buf;
 902         }
 903         if (keep_msg && !keep_name && pack_name) {
 904                 int len = strlen(pack_name);
 905                 if (!has_extension(pack_name, ".pack"))
 906                         die("packfile name '%s' does not end with '.pack'",
 907                             pack_name);
 908                 keep_name_buf = xmalloc(len);
 909                 memcpy(keep_name_buf, pack_name, len - 5);
 910                 strcpy(keep_name_buf + len - 5, ".keep");
 911                 keep_name = keep_name_buf;
 912         }
 913
 914         curr_pack = open_pack_file(pack_name);
 915         parse_pack_header();
 916         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 917         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 918         if (verbose)
 919                 setup_progress_signal();
 920         parse_pack_objects(sha1);
 921         if (nr_deltas != nr_resolved_deltas) {
 922                 if (fix_thin_pack) {
 923                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 924                         int nr_objects_initial = nr_objects;
 925                         if (nr_unresolved <= 0)
 926                                 die("confusion beyond insanity");
 927                         objects = xrealloc(objects,
 928                                            (nr_objects + nr_unresolved + 1)
 929                                            * sizeof(*objects));
 930                         fix_unresolved_deltas(nr_unresolved);
 931                         if (verbose)
 932                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 933                                         nr_objects - nr_objects_initial);
 934                         readjust_pack_header_and_sha1(sha1);
 935                 }
 936                 if (nr_deltas != nr_resolved_deltas)
 937                         die("pack has %d unresolved deltas",
 938                             nr_deltas - nr_resolved_deltas);
 939         } else {
 940                 /* Flush remaining pack final 20-byte SHA1. */
 941                 flush();
 942         }
 943         free(deltas);
 944         curr_index = write_index_file(index_name, sha1);
 945         final(pack_name, curr_pack,
 946                 index_name, curr_index,
 947                 keep_name, keep_msg,
 948                 sha1);
 949         free(objects);
 950         free(index_name_buf);
 951         free(keep_name_buf);
 952
 953         return 0;
 954 }