index-pack.c

   1 #define _XOPEN_SOURCE 600
   2 #include <unistd.h>
   3 #include <sys/time.h>
   4 #include <signal.h>
   5
   6 #include "cache.h"
   7 #include "delta.h"
   8 #include "pack.h"
   9 #include "csum-file.h"
  10 #include "blob.h"
  11 #include "commit.h"
  12 #include "tag.h"
  13 #include "tree.h"
  14
  15 static const char index_pack_usage[] =
  16 "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  17
  18 struct object_entry
  19 {
  20         unsigned long offset;
  21         unsigned long size;
  22         unsigned int hdr_size;
  23         enum object_type type;
  24         enum object_type real_type;
  25         unsigned char sha1[20];
  26 };
  27
  28 union delta_base {
  29         unsigned char sha1[20];
  30         unsigned long offset;
  31 };
  32
  33 /*
  34  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  35  * to memcmp() only the first 20 bytes.
  36  */
  37 #define UNION_BASE_SZ   20
  38
  39 struct delta_entry
  40 {
  41         union delta_base base;
  42         int obj_no;
  43 };
  44
  45 static struct object_entry *objects;
  46 static struct delta_entry *deltas;
  47 static int nr_objects;
  48 static int nr_deltas;
  49 static int nr_resolved_deltas;
  50
  51 static int from_stdin;
  52 static int verbose;
  53
  54 static volatile sig_atomic_t progress_update;
  55
  56 static void progress_interval(int signum)
  57 {
  58         progress_update = 1;
  59 }
  60
  61 static void setup_progress_signal(void)
  62 {
  63         struct sigaction sa;
  64         struct itimerval v;
  65
  66         memset(&sa, 0, sizeof(sa));
  67         sa.sa_handler = progress_interval;
  68         sigemptyset(&sa.sa_mask);
  69         sa.sa_flags = SA_RESTART;
  70         sigaction(SIGALRM, &sa, NULL);
  71
  72         v.it_interval.tv_sec = 1;
  73         v.it_interval.tv_usec = 0;
  74         v.it_value = v.it_interval;
  75         setitimer(ITIMER_REAL, &v, NULL);
  76
  77 }
  78
  79 static unsigned display_progress(unsigned n, unsigned total, unsigned last_pc)
  80 {
  81         unsigned percent = n * 100 / total;
  82         if (percent != last_pc || progress_update) {
  83                 fprintf(stderr, "%4u%% (%u/%u) done\r", percent, n, total);
  84                 progress_update = 0;
  85         }
  86         return percent;
  87 }
  88
  89 /* We always read in 4kB chunks. */
  90 static unsigned char input_buffer[4096];
  91 static unsigned long input_offset, input_len, consumed_bytes;
  92 static SHA_CTX input_ctx;
  93 static int input_fd, output_fd, pack_fd;
  94
  95 /* Discard current buffer used content. */
  96 static void flush(void)
  97 {
  98         if (input_offset) {
  99                 if (output_fd >= 0)
 100                         write_or_die(output_fd, input_buffer, input_offset);
 101                 SHA1_Update(&input_ctx, input_buffer, input_offset);
 102                 memmove(input_buffer, input_buffer + input_offset, input_len);
 103                 input_offset = 0;
 104         }
 105 }
 106
 107 /*
 108  * Make sure at least "min" bytes are available in the buffer, and
 109  * return the pointer to the buffer.
 110  */
 111 static void *fill(int min)
 112 {
 113         if (min <= input_len)
 114                 return input_buffer + input_offset;
 115         if (min > sizeof(input_buffer))
 116                 die("cannot fill %d bytes", min);
 117         flush();
 118         do {
 119                 int ret = xread(input_fd, input_buffer + input_len,
 120                                 sizeof(input_buffer) - input_len);
 121                 if (ret <= 0) {
 122                         if (!ret)
 123                                 die("early EOF");
 124                         die("read error on input: %s", strerror(errno));
 125                 }
 126                 input_len += ret;
 127         } while (input_len < min);
 128         return input_buffer;
 129 }
 130
 131 static void use(int bytes)
 132 {
 133         if (bytes > input_len)
 134                 die("used more bytes than were available");
 135         input_len -= bytes;
 136         input_offset += bytes;
 137         consumed_bytes += bytes;
 138 }
 139
 140 static const char *open_pack_file(const char *pack_name)
 141 {
 142         if (from_stdin) {
 143                 input_fd = 0;
 144                 if (!pack_name) {
 145                         static char tmpfile[PATH_MAX];
 146                         snprintf(tmpfile, sizeof(tmpfile),
 147                                  "%s/pack_XXXXXX", get_object_directory());
 148                         output_fd = mkstemp(tmpfile);
 149                         pack_name = xstrdup(tmpfile);
 150                 } else
 151                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 152                 if (output_fd < 0)
 153                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 154                 pack_fd = output_fd;
 155         } else {
 156                 input_fd = open(pack_name, O_RDONLY);
 157                 if (input_fd < 0)
 158                         die("cannot open packfile '%s': %s",
 159                             pack_name, strerror(errno));
 160                 output_fd = -1;
 161                 pack_fd = input_fd;
 162         }
 163         SHA1_Init(&input_ctx);
 164         return pack_name;
 165 }
 166
 167 static void parse_pack_header(void)
 168 {
 169         struct pack_header *hdr = fill(sizeof(struct pack_header));
 170
 171         /* Header consistency check */
 172         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 173                 die("pack signature mismatch");
 174         if (!pack_version_ok(hdr->hdr_version))
 175                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 176
 177         nr_objects = ntohl(hdr->hdr_entries);
 178         use(sizeof(struct pack_header));
 179 }
 180
 181 static void bad_object(unsigned long offset, const char *format,
 182                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 183
 184 static void bad_object(unsigned long offset, const char *format, ...)
 185 {
 186         va_list params;
 187         char buf[1024];
 188
 189         va_start(params, format);
 190         vsnprintf(buf, sizeof(buf), format, params);
 191         va_end(params);
 192         die("pack has bad object at offset %lu: %s", offset, buf);
 193 }
 194
 195 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 196 {
 197         z_stream stream;
 198         void *buf = xmalloc(size);
 199
 200         memset(&stream, 0, sizeof(stream));
 201         stream.next_out = buf;
 202         stream.avail_out = size;
 203         stream.next_in = fill(1);
 204         stream.avail_in = input_len;
 205         inflateInit(&stream);
 206
 207         for (;;) {
 208                 int ret = inflate(&stream, 0);
 209                 use(input_len - stream.avail_in);
 210                 if (stream.total_out == size && ret == Z_STREAM_END)
 211                         break;
 212                 if (ret != Z_OK)
 213                         bad_object(offset, "inflate returned %d", ret);
 214                 stream.next_in = fill(1);
 215                 stream.avail_in = input_len;
 216         }
 217         inflateEnd(&stream);
 218         return buf;
 219 }
 220
 221 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 222 {
 223         unsigned char *p, c;
 224         unsigned long size, base_offset;
 225         unsigned shift;
 226
 227         obj->offset = consumed_bytes;
 228
 229         p = fill(1);
 230         c = *p;
 231         use(1);
 232         obj->type = (c >> 4) & 7;
 233         size = (c & 15);
 234         shift = 4;
 235         while (c & 0x80) {
 236                 p = fill(1);
 237                 c = *p;
 238                 use(1);
 239                 size += (c & 0x7fUL) << shift;
 240                 shift += 7;
 241         }
 242         obj->size = size;
 243
 244         switch (obj->type) {
 245         case OBJ_REF_DELTA:
 246                 hashcpy(delta_base->sha1, fill(20));
 247                 use(20);
 248                 break;
 249         case OBJ_OFS_DELTA:
 250                 memset(delta_base, 0, sizeof(*delta_base));
 251                 p = fill(1);
 252                 c = *p;
 253                 use(1);
 254                 base_offset = c & 127;
 255                 while (c & 128) {
 256                         base_offset += 1;
 257                         if (!base_offset || base_offset & ~(~0UL >> 7))
 258                                 bad_object(obj->offset, "offset value overflow for delta base object");
 259                         p = fill(1);
 260                         c = *p;
 261                         use(1);
 262                         base_offset = (base_offset << 7) + (c & 127);
 263                 }
 264                 delta_base->offset = obj->offset - base_offset;
 265                 if (delta_base->offset >= obj->offset)
 266                         bad_object(obj->offset, "delta base offset is out of bound");
 267                 break;
 268         case OBJ_COMMIT:
 269         case OBJ_TREE:
 270         case OBJ_BLOB:
 271         case OBJ_TAG:
 272                 break;
 273         default:
 274                 bad_object(obj->offset, "bad object type %d", obj->type);
 275         }
 276         obj->hdr_size = consumed_bytes - obj->offset;
 277
 278         return unpack_entry_data(obj->offset, obj->size);
 279 }
 280
 281 static void *get_data_from_pack(struct object_entry *obj)
 282 {
 283         unsigned long from = obj[0].offset + obj[0].hdr_size;
 284         unsigned long len = obj[1].offset - from;
 285         unsigned char *src, *data;
 286         z_stream stream;
 287         int st;
 288
 289         src = xmalloc(len);
 290         if (pread(pack_fd, src, len, from) != len)
 291                 die("cannot pread pack file: %s", strerror(errno));
 292         data = xmalloc(obj->size);
 293         memset(&stream, 0, sizeof(stream));
 294         stream.next_out = data;
 295         stream.avail_out = obj->size;
 296         stream.next_in = src;
 297         stream.avail_in = len;
 298         inflateInit(&stream);
 299         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 300         inflateEnd(&stream);
 301         if (st != Z_STREAM_END || stream.total_out != obj->size)
 302                 die("serious inflate inconsistency");
 303         free(src);
 304         return data;
 305 }
 306
 307 static int find_delta(const union delta_base *base)
 308 {
 309         int first = 0, last = nr_deltas;
 310
 311         while (first < last) {
 312                 int next = (first + last) / 2;
 313                 struct delta_entry *delta = &deltas[next];
 314                 int cmp;
 315
 316                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 317                 if (!cmp)
 318                         return next;
 319                 if (cmp < 0) {
 320                         last = next;
 321                         continue;
 322                 }
 323                 first = next+1;
 324         }
 325         return -first-1;
 326 }
 327
 328 static int find_delta_children(const union delta_base *base,
 329                                int *first_index, int *last_index)
 330 {
 331         int first = find_delta(base);
 332         int last = first;
 333         int end = nr_deltas - 1;
 334
 335         if (first < 0)
 336                 return -1;
 337         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 338                 --first;
 339         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 340                 ++last;
 341         *first_index = first;
 342         *last_index = last;
 343         return 0;
 344 }
 345
 346 static void sha1_object(const void *data, unsigned long size,
 347                         enum object_type type, unsigned char *sha1)
 348 {
 349         SHA_CTX ctx;
 350         char header[50];
 351         int header_size;
 352         const char *type_str;
 353
 354         switch (type) {
 355         case OBJ_COMMIT: type_str = commit_type; break;
 356         case OBJ_TREE:   type_str = tree_type; break;
 357         case OBJ_BLOB:   type_str = blob_type; break;
 358         case OBJ_TAG:    type_str = tag_type; break;
 359         default:
 360                 die("bad type %d", type);
 361         }
 362
 363         header_size = sprintf(header, "%s %lu", type_str, size) + 1;
 364
 365         SHA1_Init(&ctx);
 366         SHA1_Update(&ctx, header, header_size);
 367         SHA1_Update(&ctx, data, size);
 368         SHA1_Final(sha1, &ctx);
 369 }
 370
 371 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 372                           unsigned long base_size, enum object_type type)
 373 {
 374         void *delta_data;
 375         unsigned long delta_size;
 376         void *result;
 377         unsigned long result_size;
 378         union delta_base delta_base;
 379         int j, first, last;
 380
 381         delta_obj->real_type = type;
 382         delta_data = get_data_from_pack(delta_obj);
 383         delta_size = delta_obj->size;
 384         result = patch_delta(base_data, base_size, delta_data, delta_size,
 385                              &result_size);
 386         free(delta_data);
 387         if (!result)
 388                 bad_object(delta_obj->offset, "failed to apply delta");
 389         sha1_object(result, result_size, type, delta_obj->sha1);
 390         nr_resolved_deltas++;
 391
 392         hashcpy(delta_base.sha1, delta_obj->sha1);
 393         if (!find_delta_children(&delta_base, &first, &last)) {
 394                 for (j = first; j <= last; j++) {
 395                         struct object_entry *child = objects + deltas[j].obj_no;
 396                         if (child->real_type == OBJ_REF_DELTA)
 397                                 resolve_delta(child, result, result_size, type);
 398                 }
 399         }
 400
 401         memset(&delta_base, 0, sizeof(delta_base));
 402         delta_base.offset = delta_obj->offset;
 403         if (!find_delta_children(&delta_base, &first, &last)) {
 404                 for (j = first; j <= last; j++) {
 405                         struct object_entry *child = objects + deltas[j].obj_no;
 406                         if (child->real_type == OBJ_OFS_DELTA)
 407                                 resolve_delta(child, result, result_size, type);
 408                 }
 409         }
 410
 411         free(result);
 412 }
 413
 414 static int compare_delta_entry(const void *a, const void *b)
 415 {
 416         const struct delta_entry *delta_a = a;
 417         const struct delta_entry *delta_b = b;
 418         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 419 }
 420
 421 /* Parse all objects and return the pack content SHA1 hash */
 422 static void parse_pack_objects(unsigned char *sha1)
 423 {
 424         int i, percent = -1;
 425         struct delta_entry *delta = deltas;
 426         void *data;
 427         struct stat st;
 428
 429         /*
 430          * First pass:
 431          * - find locations of all objects;
 432          * - calculate SHA1 of all non-delta objects;
 433          * - remember base (SHA1 or offset) for all deltas.
 434          */
 435         if (verbose)
 436                 fprintf(stderr, "Indexing %d objects.\n", nr_objects);
 437         for (i = 0; i < nr_objects; i++) {
 438                 struct object_entry *obj = &objects[i];
 439                 data = unpack_raw_entry(obj, &delta->base);
 440                 obj->real_type = obj->type;
 441                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 442                         nr_deltas++;
 443                         delta->obj_no = i;
 444                         delta++;
 445                 } else
 446                         sha1_object(data, obj->size, obj->type, obj->sha1);
 447                 free(data);
 448                 if (verbose)
 449                         percent = display_progress(i+1, nr_objects, percent);
 450         }
 451         objects[i].offset = consumed_bytes;
 452         if (verbose)
 453                 fputc('\n', stderr);
 454
 455         /* Check pack integrity */
 456         flush();
 457         SHA1_Final(sha1, &input_ctx);
 458         if (hashcmp(fill(20), sha1))
 459                 die("pack is corrupted (SHA1 mismatch)");
 460         use(20);
 461
 462         /* If input_fd is a file, we should have reached its end now. */
 463         if (fstat(input_fd, &st))
 464                 die("cannot fstat packfile: %s", strerror(errno));
 465         if (S_ISREG(st.st_mode) && st.st_size != consumed_bytes)
 466                 die("pack has junk at the end");
 467
 468         if (!nr_deltas)
 469                 return;
 470
 471         /* Sort deltas by base SHA1/offset for fast searching */
 472         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 473               compare_delta_entry);
 474
 475         /*
 476          * Second pass:
 477          * - for all non-delta objects, look if it is used as a base for
 478          *   deltas;
 479          * - if used as a base, uncompress the object and apply all deltas,
 480          *   recursively checking if the resulting object is used as a base
 481          *   for some more deltas.
 482          */
 483         if (verbose)
 484                 fprintf(stderr, "Resolving %d deltas.\n", nr_deltas);
 485         for (i = 0; i < nr_objects; i++) {
 486                 struct object_entry *obj = &objects[i];
 487                 union delta_base base;
 488                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 489
 490                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 491                         continue;
 492                 hashcpy(base.sha1, obj->sha1);
 493                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 494                 memset(&base, 0, sizeof(base));
 495                 base.offset = obj->offset;
 496                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 497                 if (!ref && !ofs)
 498                         continue;
 499                 data = get_data_from_pack(obj);
 500                 if (ref)
 501                         for (j = ref_first; j <= ref_last; j++) {
 502                                 struct object_entry *child = objects + deltas[j].obj_no;
 503                                 if (child->real_type == OBJ_REF_DELTA)
 504                                         resolve_delta(child, data,
 505                                                       obj->size, obj->type);
 506                         }
 507                 if (ofs)
 508                         for (j = ofs_first; j <= ofs_last; j++) {
 509                                 struct object_entry *child = objects + deltas[j].obj_no;
 510                                 if (child->real_type == OBJ_OFS_DELTA)
 511                                         resolve_delta(child, data,
 512                                                       obj->size, obj->type);
 513                         }
 514                 free(data);
 515                 if (verbose)
 516                         percent = display_progress(nr_resolved_deltas,
 517                                                    nr_deltas, percent);
 518         }
 519         if (verbose && nr_resolved_deltas == nr_deltas)
 520                 fputc('\n', stderr);
 521 }
 522
 523 static int write_compressed(int fd, void *in, unsigned int size)
 524 {
 525         z_stream stream;
 526         unsigned long maxsize;
 527         void *out;
 528
 529         memset(&stream, 0, sizeof(stream));
 530         deflateInit(&stream, zlib_compression_level);
 531         maxsize = deflateBound(&stream, size);
 532         out = xmalloc(maxsize);
 533
 534         /* Compress it */
 535         stream.next_in = in;
 536         stream.avail_in = size;
 537         stream.next_out = out;
 538         stream.avail_out = maxsize;
 539         while (deflate(&stream, Z_FINISH) == Z_OK);
 540         deflateEnd(&stream);
 541
 542         size = stream.total_out;
 543         write_or_die(fd, out, size);
 544         free(out);
 545         return size;
 546 }
 547
 548 static void append_obj_to_pack(void *buf,
 549                                unsigned long size, enum object_type type)
 550 {
 551         struct object_entry *obj = &objects[nr_objects++];
 552         unsigned char header[10];
 553         unsigned long s = size;
 554         int n = 0;
 555         unsigned char c = (type << 4) | (s & 15);
 556         s >>= 4;
 557         while (s) {
 558                 header[n++] = c | 0x80;
 559                 c = s & 0x7f;
 560                 s >>= 7;
 561         }
 562         header[n++] = c;
 563         write_or_die(output_fd, header, n);
 564         obj[1].offset = obj[0].offset + n;
 565         obj[1].offset += write_compressed(output_fd, buf, size);
 566         sha1_object(buf, size, type, obj->sha1);
 567 }
 568
 569 static int delta_pos_compare(const void *_a, const void *_b)
 570 {
 571         struct delta_entry *a = *(struct delta_entry **)_a;
 572         struct delta_entry *b = *(struct delta_entry **)_b;
 573         return a->obj_no - b->obj_no;
 574 }
 575
 576 static void fix_unresolved_deltas(int nr_unresolved)
 577 {
 578         struct delta_entry **sorted_by_pos;
 579         int i, n = 0, percent = -1;
 580
 581         /*
 582          * Since many unresolved deltas may well be themselves base objects
 583          * for more unresolved deltas, we really want to include the
 584          * smallest number of base objects that would cover as much delta
 585          * as possible by picking the
 586          * trunc deltas first, allowing for other deltas to resolve without
 587          * additional base objects.  Since most base objects are to be found
 588          * before deltas depending on them, a good heuristic is to start
 589          * resolving deltas in the same order as their position in the pack.
 590          */
 591         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 592         for (i = 0; i < nr_deltas; i++) {
 593                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 594                         continue;
 595                 sorted_by_pos[n++] = &deltas[i];
 596         }
 597         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 598
 599         for (i = 0; i < n; i++) {
 600                 struct delta_entry *d = sorted_by_pos[i];
 601                 void *data;
 602                 unsigned long size;
 603                 char type[10];
 604                 enum object_type obj_type;
 605                 int j, first, last;
 606
 607                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 608                         continue;
 609                 data = read_sha1_file(d->base.sha1, type, &size);
 610                 if (!data)
 611                         continue;
 612                 if      (!strcmp(type, blob_type))   obj_type = OBJ_BLOB;
 613                 else if (!strcmp(type, tree_type))   obj_type = OBJ_TREE;
 614                 else if (!strcmp(type, commit_type)) obj_type = OBJ_COMMIT;
 615                 else if (!strcmp(type, tag_type))    obj_type = OBJ_TAG;
 616                 else die("base object %s is of type '%s'",
 617                          sha1_to_hex(d->base.sha1), type);
 618
 619                 find_delta_children(&d->base, &first, &last);
 620                 for (j = first; j <= last; j++) {
 621                         struct object_entry *child = objects + deltas[j].obj_no;
 622                         if (child->real_type == OBJ_REF_DELTA)
 623                                 resolve_delta(child, data, size, obj_type);
 624                 }
 625
 626                 append_obj_to_pack(data, size, obj_type);
 627                 free(data);
 628                 if (verbose)
 629                         percent = display_progress(nr_resolved_deltas,
 630                                                    nr_deltas, percent);
 631         }
 632         free(sorted_by_pos);
 633         if (verbose)
 634                 fputc('\n', stderr);
 635 }
 636
 637 static void readjust_pack_header_and_sha1(unsigned char *sha1)
 638 {
 639         struct pack_header hdr;
 640         SHA_CTX ctx;
 641         int size;
 642
 643         /* Rewrite pack header with updated object number */
 644         if (lseek(output_fd, 0, SEEK_SET) != 0)
 645                 die("cannot seek back: %s", strerror(errno));
 646         if (xread(output_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 647                 die("cannot read pack header back: %s", strerror(errno));
 648         hdr.hdr_entries = htonl(nr_objects);
 649         if (lseek(output_fd, 0, SEEK_SET) != 0)
 650                 die("cannot seek back: %s", strerror(errno));
 651         write_or_die(output_fd, &hdr, sizeof(hdr));
 652         if (lseek(output_fd, 0, SEEK_SET) != 0)
 653                 die("cannot seek back: %s", strerror(errno));
 654
 655         /* Recompute and store the new pack's SHA1 */
 656         SHA1_Init(&ctx);
 657         do {
 658                 unsigned char *buf[4096];
 659                 size = xread(output_fd, buf, sizeof(buf));
 660                 if (size < 0)
 661                         die("cannot read pack data back: %s", strerror(errno));
 662                 SHA1_Update(&ctx, buf, size);
 663         } while (size > 0);
 664         SHA1_Final(sha1, &ctx);
 665         write_or_die(output_fd, sha1, 20);
 666 }
 667
 668 static int sha1_compare(const void *_a, const void *_b)
 669 {
 670         struct object_entry *a = *(struct object_entry **)_a;
 671         struct object_entry *b = *(struct object_entry **)_b;
 672         return hashcmp(a->sha1, b->sha1);
 673 }
 674
 675 /*
 676  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 677  * the SHA1 hash of sorted object names.
 678  */
 679 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 680 {
 681         struct sha1file *f;
 682         struct object_entry **sorted_by_sha, **list, **last;
 683         unsigned int array[256];
 684         int i, fd;
 685         SHA_CTX ctx;
 686
 687         if (nr_objects) {
 688                 sorted_by_sha =
 689                         xcalloc(nr_objects, sizeof(struct object_entry *));
 690                 list = sorted_by_sha;
 691                 last = sorted_by_sha + nr_objects;
 692                 for (i = 0; i < nr_objects; ++i)
 693                         sorted_by_sha[i] = &objects[i];
 694                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 695                       sha1_compare);
 696
 697         }
 698         else
 699                 sorted_by_sha = list = last = NULL;
 700
 701         if (!index_name) {
 702                 static char tmpfile[PATH_MAX];
 703                 snprintf(tmpfile, sizeof(tmpfile),
 704                          "%s/index_XXXXXX", get_object_directory());
 705                 fd = mkstemp(tmpfile);
 706                 index_name = xstrdup(tmpfile);
 707         } else {
 708                 unlink(index_name);
 709                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 710         }
 711         if (fd < 0)
 712                 die("unable to create %s: %s", index_name, strerror(errno));
 713         f = sha1fd(fd, index_name);
 714
 715         /*
 716          * Write the first-level table (the list is sorted,
 717          * but we use a 256-entry lookup to be able to avoid
 718          * having to do eight extra binary search iterations).
 719          */
 720         for (i = 0; i < 256; i++) {
 721                 struct object_entry **next = list;
 722                 while (next < last) {
 723                         struct object_entry *obj = *next;
 724                         if (obj->sha1[0] != i)
 725                                 break;
 726                         next++;
 727                 }
 728                 array[i] = htonl(next - sorted_by_sha);
 729                 list = next;
 730         }
 731         sha1write(f, array, 256 * sizeof(int));
 732
 733         /* recompute the SHA1 hash of sorted object names.
 734          * currently pack-objects does not do this, but that
 735          * can be fixed.
 736          */
 737         SHA1_Init(&ctx);
 738         /*
 739          * Write the actual SHA1 entries..
 740          */
 741         list = sorted_by_sha;
 742         for (i = 0; i < nr_objects; i++) {
 743                 struct object_entry *obj = *list++;
 744                 unsigned int offset = htonl(obj->offset);
 745                 sha1write(f, &offset, 4);
 746                 sha1write(f, obj->sha1, 20);
 747                 SHA1_Update(&ctx, obj->sha1, 20);
 748         }
 749         sha1write(f, sha1, 20);
 750         sha1close(f, NULL, 1);
 751         free(sorted_by_sha);
 752         SHA1_Final(sha1, &ctx);
 753         return index_name;
 754 }
 755
 756 static void final(const char *final_pack_name, const char *curr_pack_name,
 757                   const char *final_index_name, const char *curr_index_name,
 758                   const char *keep_name, const char *keep_msg,
 759                   unsigned char *sha1)
 760 {
 761         char *report = "pack";
 762         char name[PATH_MAX];
 763         int err;
 764
 765         if (!from_stdin) {
 766                 close(input_fd);
 767         } else {
 768                 err = close(output_fd);
 769                 if (err)
 770                         die("error while closing pack file: %s", strerror(errno));
 771                 chmod(curr_pack_name, 0444);
 772         }
 773
 774         if (keep_msg) {
 775                 int keep_fd, keep_msg_len = strlen(keep_msg);
 776                 if (!keep_name) {
 777                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 778                                  get_object_directory(), sha1_to_hex(sha1));
 779                         keep_name = name;
 780                 }
 781                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 782                 if (keep_fd < 0) {
 783                         if (errno != EEXIST)
 784                                 die("cannot write keep file");
 785                 } else {
 786                         if (keep_msg_len > 0) {
 787                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 788                                 write_or_die(keep_fd, "\n", 1);
 789                         }
 790                         close(keep_fd);
 791                         report = "keep";
 792                 }
 793         }
 794
 795         if (final_pack_name != curr_pack_name) {
 796                 if (!final_pack_name) {
 797                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 798                                  get_object_directory(), sha1_to_hex(sha1));
 799                         final_pack_name = name;
 800                 }
 801                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 802                         die("cannot store pack file");
 803         }
 804
 805         chmod(curr_index_name, 0444);
 806         if (final_index_name != curr_index_name) {
 807                 if (!final_index_name) {
 808                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 809                                  get_object_directory(), sha1_to_hex(sha1));
 810                         final_index_name = name;
 811                 }
 812                 if (move_temp_to_file(curr_index_name, final_index_name))
 813                         die("cannot store index file");
 814         }
 815
 816         if (!from_stdin) {
 817                 printf("%s\n", sha1_to_hex(sha1));
 818         } else {
 819                 char buf[48];
 820                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 821                                    report, sha1_to_hex(sha1));
 822                 xwrite(1, buf, len);
 823
 824                 /*
 825                  * Let's just mimic git-unpack-objects here and write
 826                  * the last part of the input buffer to stdout.
 827                  */
 828                 while (input_len) {
 829                         err = xwrite(1, input_buffer + input_offset, input_len);
 830                         if (err <= 0)
 831                                 break;
 832                         input_len -= err;
 833                         input_offset += err;
 834                 }
 835         }
 836 }
 837
 838 int main(int argc, char **argv)
 839 {
 840         int i, fix_thin_pack = 0;
 841         const char *curr_pack, *pack_name = NULL;
 842         const char *curr_index, *index_name = NULL;
 843         const char *keep_name = NULL, *keep_msg = NULL;
 844         char *index_name_buf = NULL, *keep_name_buf = NULL;
 845         unsigned char sha1[20];
 846
 847         for (i = 1; i < argc; i++) {
 848                 const char *arg = argv[i];
 849
 850                 if (*arg == '-') {
 851                         if (!strcmp(arg, "--stdin")) {
 852                                 from_stdin = 1;
 853                         } else if (!strcmp(arg, "--fix-thin")) {
 854                                 fix_thin_pack = 1;
 855                         } else if (!strcmp(arg, "--keep")) {
 856                                 keep_msg = "";
 857                         } else if (!strncmp(arg, "--keep=", 7)) {
 858                                 keep_msg = arg + 7;
 859                         } else if (!strncmp(arg, "--pack_header=", 14)) {
 860                                 struct pack_header *hdr;
 861                                 char *c;
 862
 863                                 hdr = (struct pack_header *)input_buffer;
 864                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 865                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 866                                 if (*c != ',')
 867                                         die("bad %s", arg);
 868                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 869                                 if (*c)
 870                                         die("bad %s", arg);
 871                                 input_len = sizeof(*hdr);
 872                         } else if (!strcmp(arg, "-v")) {
 873                                 verbose = 1;
 874                         } else if (!strcmp(arg, "-o")) {
 875                                 if (index_name || (i+1) >= argc)
 876                                         usage(index_pack_usage);
 877                                 index_name = argv[++i];
 878                         } else
 879                                 usage(index_pack_usage);
 880                         continue;
 881                 }
 882
 883                 if (pack_name)
 884                         usage(index_pack_usage);
 885                 pack_name = arg;
 886         }
 887
 888         if (!pack_name && !from_stdin)
 889                 usage(index_pack_usage);
 890         if (fix_thin_pack && !from_stdin)
 891                 die("--fix-thin cannot be used without --stdin");
 892         if (!index_name && pack_name) {
 893                 int len = strlen(pack_name);
 894                 if (!has_extension(pack_name, ".pack"))
 895                         die("packfile name '%s' does not end with '.pack'",
 896                             pack_name);
 897                 index_name_buf = xmalloc(len);
 898                 memcpy(index_name_buf, pack_name, len - 5);
 899                 strcpy(index_name_buf + len - 5, ".idx");
 900                 index_name = index_name_buf;
 901         }
 902         if (keep_msg && !keep_name && pack_name) {
 903                 int len = strlen(pack_name);
 904                 if (!has_extension(pack_name, ".pack"))
 905                         die("packfile name '%s' does not end with '.pack'",
 906                             pack_name);
 907                 keep_name_buf = xmalloc(len);
 908                 memcpy(keep_name_buf, pack_name, len - 5);
 909                 strcpy(keep_name_buf + len - 5, ".keep");
 910                 keep_name = keep_name_buf;
 911         }
 912
 913         curr_pack = open_pack_file(pack_name);
 914         parse_pack_header();
 915         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 916         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 917         if (verbose)
 918                 setup_progress_signal();
 919         parse_pack_objects(sha1);
 920         if (nr_deltas != nr_resolved_deltas) {
 921                 if (fix_thin_pack) {
 922                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 923                         int nr_objects_initial = nr_objects;
 924                         if (nr_unresolved <= 0)
 925                                 die("confusion beyond insanity");
 926                         objects = xrealloc(objects,
 927                                            (nr_objects + nr_unresolved + 1)
 928                                            * sizeof(*objects));
 929                         fix_unresolved_deltas(nr_unresolved);
 930                         if (verbose)
 931                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 932                                         nr_objects - nr_objects_initial);
 933                         readjust_pack_header_and_sha1(sha1);
 934                 }
 935                 if (nr_deltas != nr_resolved_deltas)
 936                         die("pack has %d unresolved deltas",
 937                             nr_deltas - nr_resolved_deltas);
 938         } else {
 939                 /* Flush remaining pack final 20-byte SHA1. */
 940                 flush();
 941         }
 942         free(deltas);
 943         curr_index = write_index_file(index_name, sha1);
 944         final(pack_name, curr_pack,
 945                 index_name, curr_index,
 946                 keep_name, keep_msg,
 947                 sha1);
 948         free(objects);
 949         free(index_name_buf);
 950         free(keep_name_buf);
 951
 952         return 0;
 953 }