pack-objects.c

   1 #include "cache.h"
   2 #include "object.h"
   3 #include "blob.h"
   4 #include "commit.h"
   5 #include "tag.h"
   6 #include "tree.h"
   7 #include "delta.h"
   8 #include "pack.h"
   9 #include "csum-file.h"
  10 #include "tree-walk.h"
  11 #include <sys/time.h>
  12 #include <signal.h>
  13
  14 static const char pack_usage[] = "git-pack-objects [-q] [--no-reuse-delta] [--non-empty] [--local] [--incremental] [--window=N] [--depth=N] {--stdout | base-name} < object-list";
  15
  16 struct object_entry {
  17         unsigned char sha1[20];
  18         unsigned long size;     /* uncompressed size */
  19         unsigned long offset;   /* offset into the final pack file;
  20                                  * nonzero if already written.
  21                                  */
  22         unsigned int depth;     /* delta depth */
  23         unsigned int delta_limit;       /* base adjustment for in-pack delta */
  24         unsigned int hash;      /* name hint hash */
  25         enum object_type type;
  26         enum object_type in_pack_type;  /* could be delta */
  27         unsigned long delta_size;       /* delta data size (uncompressed) */
  28         struct object_entry *delta;     /* delta base object */
  29         struct packed_git *in_pack;     /* already in pack */
  30         unsigned int in_pack_offset;
  31         struct object_entry *delta_child; /* delitified objects who bases me */
  32         struct object_entry *delta_sibling; /* other deltified objects who
  33                                              * uses the same base as me
  34                                              */
  35         int preferred_base;     /* we do not pack this, but is encouraged to
  36                                  * be used as the base objectto delta huge
  37                                  * objects against.
  38                                  */
  39 };
  40
  41 /*
  42  * Objects we are going to pack are colected in objects array (dynamically
  43  * expanded).  nr_objects & nr_alloc controls this array.  They are stored
  44  * in the order we see -- typically rev-list --objects order that gives us
  45  * nice "minimum seek" order.
  46  *
  47  * sorted-by-sha ans sorted-by-type are arrays of pointers that point at
  48  * elements in the objects array.  The former is used to build the pack
  49  * index (lists object names in the ascending order to help offset lookup),
  50  * and the latter is used to group similar things together by try_delta()
  51  * heuristics.
  52  */
  53
  54 static unsigned char object_list_sha1[20];
  55 static int non_empty = 0;
  56 static int no_reuse_delta = 0;
  57 static int local = 0;
  58 static int incremental = 0;
  59 static struct object_entry **sorted_by_sha, **sorted_by_type;
  60 static struct object_entry *objects = NULL;
  61 static int nr_objects = 0, nr_alloc = 0, nr_result = 0;
  62 static const char *base_name;
  63 static unsigned char pack_file_sha1[20];
  64 static int progress = 1;
  65 static volatile int progress_update = 0;
  66
  67 /*
  68  * The object names in objects array are hashed with this hashtable,
  69  * to help looking up the entry by object name.  Binary search from
  70  * sorted_by_sha is also possible but this was easier to code and faster.
  71  * This hashtable is built after all the objects are seen.
  72  */
  73 static int *object_ix = NULL;
  74 static int object_ix_hashsz = 0;
  75
  76 /*
  77  * Pack index for existing packs give us easy access to the offsets into
  78  * corresponding pack file where each object's data starts, but the entries
  79  * do not store the size of the compressed representation (uncompressed
  80  * size is easily available by examining the pack entry header).  We build
  81  * a hashtable of existing packs (pack_revindex), and keep reverse index
  82  * here -- pack index file is sorted by object name mapping to offset; this
  83  * pack_revindex[].revindex array is an ordered list of offsets, so if you
  84  * know the offset of an object, next offset is where its packed
  85  * representation ends.
  86  */
  87 struct pack_revindex {
  88         struct packed_git *p;
  89         unsigned long *revindex;
  90 } *pack_revindex = NULL;
  91 static int pack_revindex_hashsz = 0;
  92
  93 /*
  94  * stats
  95  */
  96 static int written = 0;
  97 static int written_delta = 0;
  98 static int reused = 0;
  99 static int reused_delta = 0;
 100
 101 static int pack_revindex_ix(struct packed_git *p)
 102 {
 103         unsigned long ui = (unsigned long)p;
 104         int i;
 105
 106         ui = ui ^ (ui >> 16); /* defeat structure alignment */
 107         i = (int)(ui % pack_revindex_hashsz);
 108         while (pack_revindex[i].p) {
 109                 if (pack_revindex[i].p == p)
 110                         return i;
 111                 if (++i == pack_revindex_hashsz)
 112                         i = 0;
 113         }
 114         return -1 - i;
 115 }
 116
 117 static void prepare_pack_ix(void)
 118 {
 119         int num;
 120         struct packed_git *p;
 121         for (num = 0, p = packed_git; p; p = p->next)
 122                 num++;
 123         if (!num)
 124                 return;
 125         pack_revindex_hashsz = num * 11;
 126         pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
 127         for (p = packed_git; p; p = p->next) {
 128                 num = pack_revindex_ix(p);
 129                 num = - 1 - num;
 130                 pack_revindex[num].p = p;
 131         }
 132         /* revindex elements are lazily initialized */
 133 }
 134
 135 static int cmp_offset(const void *a_, const void *b_)
 136 {
 137         unsigned long a = *(unsigned long *) a_;
 138         unsigned long b = *(unsigned long *) b_;
 139         if (a < b)
 140                 return -1;
 141         else if (a == b)
 142                 return 0;
 143         else
 144                 return 1;
 145 }
 146
 147 /*
 148  * Ordered list of offsets of objects in the pack.
 149  */
 150 static void prepare_pack_revindex(struct pack_revindex *rix)
 151 {
 152         struct packed_git *p = rix->p;
 153         int num_ent = num_packed_objects(p);
 154         int i;
 155         void *index = p->index_base + 256;
 156
 157         rix->revindex = xmalloc(sizeof(unsigned long) * (num_ent + 1));
 158         for (i = 0; i < num_ent; i++) {
 159                 long hl = *((long *)(index + 24 * i));
 160                 rix->revindex[i] = ntohl(hl);
 161         }
 162         /* This knows the pack format -- the 20-byte trailer
 163          * follows immediately after the last object data.
 164          */
 165         rix->revindex[num_ent] = p->pack_size - 20;
 166         qsort(rix->revindex, num_ent, sizeof(unsigned long), cmp_offset);
 167 }
 168
 169 static unsigned long find_packed_object_size(struct packed_git *p,
 170                                              unsigned long ofs)
 171 {
 172         int num;
 173         int lo, hi;
 174         struct pack_revindex *rix;
 175         unsigned long *revindex;
 176         num = pack_revindex_ix(p);
 177         if (num < 0)
 178                 die("internal error: pack revindex uninitialized");
 179         rix = &pack_revindex[num];
 180         if (!rix->revindex)
 181                 prepare_pack_revindex(rix);
 182         revindex = rix->revindex;
 183         lo = 0;
 184         hi = num_packed_objects(p) + 1;
 185         do {
 186                 int mi = (lo + hi) / 2;
 187                 if (revindex[mi] == ofs) {
 188                         return revindex[mi+1] - ofs;
 189                 }
 190                 else if (ofs < revindex[mi])
 191                         hi = mi;
 192                 else
 193                         lo = mi + 1;
 194         } while (lo < hi);
 195         die("internal error: pack revindex corrupt");
 196 }
 197
 198 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 199 {
 200         unsigned long othersize, delta_size;
 201         char type[10];
 202         void *otherbuf = read_sha1_file(entry->delta->sha1, type, &othersize);
 203         void *delta_buf;
 204
 205         if (!otherbuf)
 206                 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
 207         delta_buf = diff_delta(otherbuf, othersize,
 208                                buf, size, &delta_size, 0);
 209         if (!delta_buf || delta_size != entry->delta_size)
 210                 die("delta size changed");
 211         free(buf);
 212         free(otherbuf);
 213         return delta_buf;
 214 }
 215
 216 /*
 217  * The per-object header is a pretty dense thing, which is
 218  *  - first byte: low four bits are "size", then three bits of "type",
 219  *    and the high bit is "size continues".
 220  *  - each byte afterwards: low seven bits are size continuation,
 221  *    with the high bit being "size continues"
 222  */
 223 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
 224 {
 225         int n = 1;
 226         unsigned char c;
 227
 228         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 229                 die("bad type %d", type);
 230
 231         c = (type << 4) | (size & 15);
 232         size >>= 4;
 233         while (size) {
 234                 *hdr++ = c | 0x80;
 235                 c = size & 0x7f;
 236                 size >>= 7;
 237                 n++;
 238         }
 239         *hdr = c;
 240         return n;
 241 }
 242
 243 static unsigned long write_object(struct sha1file *f,
 244                                   struct object_entry *entry)
 245 {
 246         unsigned long size;
 247         char type[10];
 248         void *buf;
 249         unsigned char header[10];
 250         unsigned hdrlen, datalen;
 251         enum object_type obj_type;
 252         int to_reuse = 0;
 253
 254         if (entry->preferred_base)
 255                 return 0;
 256
 257         obj_type = entry->type;
 258         if (! entry->in_pack)
 259                 to_reuse = 0;   /* can't reuse what we don't have */
 260         else if (obj_type == OBJ_DELTA)
 261                 to_reuse = 1;   /* check_object() decided it for us */
 262         else if (obj_type != entry->in_pack_type)
 263                 to_reuse = 0;   /* pack has delta which is unusable */
 264         else if (entry->delta)
 265                 to_reuse = 0;   /* we want to pack afresh */
 266         else
 267                 to_reuse = 1;   /* we have it in-pack undeltified,
 268                                  * and we do not need to deltify it.
 269                                  */
 270
 271         if (! to_reuse) {
 272                 buf = read_sha1_file(entry->sha1, type, &size);
 273                 if (!buf)
 274                         die("unable to read %s", sha1_to_hex(entry->sha1));
 275                 if (size != entry->size)
 276                         die("object %s size inconsistency (%lu vs %lu)",
 277                             sha1_to_hex(entry->sha1), size, entry->size);
 278                 if (entry->delta) {
 279                         buf = delta_against(buf, size, entry);
 280                         size = entry->delta_size;
 281                         obj_type = OBJ_DELTA;
 282                 }
 283                 /*
 284                  * The object header is a byte of 'type' followed by zero or
 285                  * more bytes of length.  For deltas, the 20 bytes of delta
 286                  * sha1 follows that.
 287                  */
 288                 hdrlen = encode_header(obj_type, size, header);
 289                 sha1write(f, header, hdrlen);
 290
 291                 if (entry->delta) {
 292                         sha1write(f, entry->delta, 20);
 293                         hdrlen += 20;
 294                 }
 295                 datalen = sha1write_compressed(f, buf, size);
 296                 free(buf);
 297         }
 298         else {
 299                 struct packed_git *p = entry->in_pack;
 300                 use_packed_git(p);
 301
 302                 datalen = find_packed_object_size(p, entry->in_pack_offset);
 303                 buf = p->pack_base + entry->in_pack_offset;
 304                 sha1write(f, buf, datalen);
 305                 unuse_packed_git(p);
 306                 hdrlen = 0; /* not really */
 307                 if (obj_type == OBJ_DELTA)
 308                         reused_delta++;
 309                 reused++;
 310         }
 311         if (obj_type == OBJ_DELTA)
 312                 written_delta++;
 313         written++;
 314         return hdrlen + datalen;
 315 }
 316
 317 static unsigned long write_one(struct sha1file *f,
 318                                struct object_entry *e,
 319                                unsigned long offset)
 320 {
 321         if (e->offset)
 322                 /* offset starts from header size and cannot be zero
 323                  * if it is written already.
 324                  */
 325                 return offset;
 326         e->offset = offset;
 327         offset += write_object(f, e);
 328         /* if we are deltified, write out its base object. */
 329         if (e->delta)
 330                 offset = write_one(f, e->delta, offset);
 331         return offset;
 332 }
 333
 334 static void write_pack_file(void)
 335 {
 336         int i;
 337         struct sha1file *f;
 338         unsigned long offset;
 339         struct pack_header hdr;
 340         unsigned last_percent = 999;
 341         int do_progress = 0;
 342
 343         if (!base_name)
 344                 f = sha1fd(1, "<stdout>");
 345         else {
 346                 f = sha1create("%s-%s.%s", base_name,
 347                                sha1_to_hex(object_list_sha1), "pack");
 348                 do_progress = progress;
 349         }
 350         if (do_progress)
 351                 fprintf(stderr, "Writing %d objects.\n", nr_result);
 352
 353         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 354         hdr.hdr_version = htonl(PACK_VERSION);
 355         hdr.hdr_entries = htonl(nr_result);
 356         sha1write(f, &hdr, sizeof(hdr));
 357         offset = sizeof(hdr);
 358         if (!nr_result)
 359                 goto done;
 360         for (i = 0; i < nr_objects; i++) {
 361                 offset = write_one(f, objects + i, offset);
 362                 if (do_progress) {
 363                         unsigned percent = written * 100 / nr_result;
 364                         if (progress_update || percent != last_percent) {
 365                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
 366                                         percent, written, nr_result);
 367                                 progress_update = 0;
 368                                 last_percent = percent;
 369                         }
 370                 }
 371         }
 372         if (do_progress)
 373                 fputc('\n', stderr);
 374  done:
 375         sha1close(f, pack_file_sha1, 1);
 376 }
 377
 378 static void write_index_file(void)
 379 {
 380         int i;
 381         struct sha1file *f = sha1create("%s-%s.%s", base_name,
 382                                         sha1_to_hex(object_list_sha1), "idx");
 383         struct object_entry **list = sorted_by_sha;
 384         struct object_entry **last = list + nr_result;
 385         unsigned int array[256];
 386
 387         /*
 388          * Write the first-level table (the list is sorted,
 389          * but we use a 256-entry lookup to be able to avoid
 390          * having to do eight extra binary search iterations).
 391          */
 392         for (i = 0; i < 256; i++) {
 393                 struct object_entry **next = list;
 394                 while (next < last) {
 395                         struct object_entry *entry = *next;
 396                         if (entry->sha1[0] != i)
 397                                 break;
 398                         next++;
 399                 }
 400                 array[i] = htonl(next - sorted_by_sha);
 401                 list = next;
 402         }
 403         sha1write(f, array, 256 * sizeof(int));
 404
 405         /*
 406          * Write the actual SHA1 entries..
 407          */
 408         list = sorted_by_sha;
 409         for (i = 0; i < nr_result; i++) {
 410                 struct object_entry *entry = *list++;
 411                 unsigned int offset = htonl(entry->offset);
 412                 sha1write(f, &offset, 4);
 413                 sha1write(f, entry->sha1, 20);
 414         }
 415         sha1write(f, pack_file_sha1, 20);
 416         sha1close(f, NULL, 1);
 417 }
 418
 419 static int locate_object_entry_hash(const unsigned char *sha1)
 420 {
 421         int i;
 422         unsigned int ui;
 423         memcpy(&ui, sha1, sizeof(unsigned int));
 424         i = ui % object_ix_hashsz;
 425         while (0 < object_ix[i]) {
 426                 if (!memcmp(sha1, objects[object_ix[i]-1].sha1, 20))
 427                         return i;
 428                 if (++i == object_ix_hashsz)
 429                         i = 0;
 430         }
 431         return -1 - i;
 432 }
 433
 434 static struct object_entry *locate_object_entry(const unsigned char *sha1)
 435 {
 436         int i;
 437
 438         if (!object_ix_hashsz)
 439                 return NULL;
 440
 441         i = locate_object_entry_hash(sha1);
 442         if (0 <= i)
 443                 return &objects[object_ix[i]-1];
 444         return NULL;
 445 }
 446
 447 static void rehash_objects(void)
 448 {
 449         int i;
 450         struct object_entry *oe;
 451
 452         object_ix_hashsz = nr_objects * 3;
 453         if (object_ix_hashsz < 1024)
 454                 object_ix_hashsz = 1024;
 455         object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
 456         object_ix = memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
 457         for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
 458                 int ix = locate_object_entry_hash(oe->sha1);
 459                 if (0 <= ix)
 460                         continue;
 461                 ix = -1 - ix;
 462                 object_ix[ix] = i + 1;
 463         }
 464 }
 465
 466 struct name_path {
 467         struct name_path *up;
 468         const char *elem;
 469         int len;
 470 };
 471
 472 #define DIRBITS 12
 473
 474 static unsigned name_hash(struct name_path *path, const char *name)
 475 {
 476         struct name_path *p = path;
 477         const char *n = name + strlen(name);
 478         unsigned hash = 0, name_hash = 0, name_done = 0;
 479
 480         if (n != name && n[-1] == '\n')
 481                 n--;
 482         while (name <= --n) {
 483                 unsigned char c = *n;
 484                 if (c == '/' && !name_done) {
 485                         name_hash = hash;
 486                         name_done = 1;
 487                         hash = 0;
 488                 }
 489                 hash = hash * 11 + c;
 490         }
 491         if (!name_done) {
 492                 name_hash = hash;
 493                 hash = 0;
 494         }
 495         for (p = path; p; p = p->up) {
 496                 hash = hash * 11 + '/';
 497                 n = p->elem + p->len;
 498                 while (p->elem <= --n) {
 499                         unsigned char c = *n;
 500                         hash = hash * 11 + c;
 501                 }
 502         }
 503         /*
 504          * Make sure "Makefile" and "t/Makefile" are hashed separately
 505          * but close enough.
 506          */
 507         hash = (name_hash<<DIRBITS) | (hash & ((1U<<DIRBITS )-1));
 508
 509         if (0) { /* debug */
 510                 n = name + strlen(name);
 511                 if (n != name && n[-1] == '\n')
 512                         n--;
 513                 while (name <= --n)
 514                         fputc(*n, stderr);
 515                 for (p = path; p; p = p->up) {
 516                         fputc('/', stderr);
 517                         n = p->elem + p->len;
 518                         while (p->elem <= --n)
 519                                 fputc(*n, stderr);
 520                 }
 521                 fprintf(stderr, "\t%08x\n", hash);
 522         }
 523         return hash;
 524 }
 525
 526 static int add_object_entry(const unsigned char *sha1, unsigned hash, int exclude)
 527 {
 528         unsigned int idx = nr_objects;
 529         struct object_entry *entry;
 530         struct packed_git *p;
 531         unsigned int found_offset = 0;
 532         struct packed_git *found_pack = NULL;
 533         int ix, status = 0;
 534
 535         if (!exclude) {
 536                 for (p = packed_git; p; p = p->next) {
 537                         struct pack_entry e;
 538                         if (find_pack_entry_one(sha1, &e, p)) {
 539                                 if (incremental)
 540                                         return 0;
 541                                 if (local && !p->pack_local)
 542                                         return 0;
 543                                 if (!found_pack) {
 544                                         found_offset = e.offset;
 545                                         found_pack = e.p;
 546                                 }
 547                         }
 548                 }
 549         }
 550         if ((entry = locate_object_entry(sha1)) != NULL)
 551                 goto already_added;
 552
 553         if (idx >= nr_alloc) {
 554                 unsigned int needed = (idx + 1024) * 3 / 2;
 555                 objects = xrealloc(objects, needed * sizeof(*entry));
 556                 nr_alloc = needed;
 557         }
 558         entry = objects + idx;
 559         nr_objects = idx + 1;
 560         memset(entry, 0, sizeof(*entry));
 561         memcpy(entry->sha1, sha1, 20);
 562         entry->hash = hash;
 563
 564         if (object_ix_hashsz * 3 <= nr_objects * 4)
 565                 rehash_objects();
 566         else {
 567                 ix = locate_object_entry_hash(entry->sha1);
 568                 if (0 <= ix)
 569                         die("internal error in object hashing.");
 570                 object_ix[-1 - ix] = idx + 1;
 571         }
 572         status = 1;
 573
 574  already_added:
 575         if (progress_update) {
 576                 fprintf(stderr, "Counting objects...%d\r", nr_objects);
 577                 progress_update = 0;
 578         }
 579         if (exclude)
 580                 entry->preferred_base = 1;
 581         else {
 582                 if (found_pack) {
 583                         entry->in_pack = found_pack;
 584                         entry->in_pack_offset = found_offset;
 585                 }
 586         }
 587         return status;
 588 }
 589
 590 static void add_pbase_tree(struct tree_desc *tree, struct name_path *up)
 591 {
 592         while (tree->size) {
 593                 const unsigned char *sha1;
 594                 const char *name;
 595                 unsigned mode, hash;
 596                 unsigned long size;
 597                 char type[20];
 598
 599                 sha1 = tree_entry_extract(tree, &name, &mode);
 600                 update_tree_entry(tree);
 601                 if (!has_sha1_file(sha1))
 602                         continue;
 603                 if (sha1_object_info(sha1, type, &size))
 604                         continue;
 605
 606                 hash = name_hash(up, name);
 607                 if (!add_object_entry(sha1, hash, 1))
 608                         continue;
 609
 610                 if (!strcmp(type, tree_type)) {
 611                         struct tree_desc sub;
 612                         void *elem;
 613                         struct name_path me;
 614
 615                         elem = read_sha1_file(sha1, type, &sub.size);
 616                         sub.buf = elem;
 617                         if (sub.buf) {
 618                                 me.up = up;
 619                                 me.elem = name;
 620                                 me.len = strlen(name);
 621                                 add_pbase_tree(&sub, &me);
 622                                 free(elem);
 623                         }
 624                 }
 625         }
 626 }
 627
 628 static void add_preferred_base(unsigned char *sha1)
 629 {
 630         struct tree_desc tree;
 631         void *elem;
 632
 633         elem = read_object_with_reference(sha1, tree_type, &tree.size, NULL);
 634         tree.buf = elem;
 635         if (!tree.buf)
 636                 return;
 637         if (add_object_entry(sha1, name_hash(NULL, ""), 1))
 638                 add_pbase_tree(&tree, NULL);
 639         free(elem);
 640 }
 641
 642 static void check_object(struct object_entry *entry)
 643 {
 644         char type[20];
 645
 646         if (entry->in_pack && !entry->preferred_base) {
 647                 unsigned char base[20];
 648                 unsigned long size;
 649                 struct object_entry *base_entry;
 650
 651                 /* We want in_pack_type even if we do not reuse delta.
 652                  * There is no point not reusing non-delta representations.
 653                  */
 654                 check_reuse_pack_delta(entry->in_pack,
 655                                        entry->in_pack_offset,
 656                                        base, &size,
 657                                        &entry->in_pack_type);
 658
 659                 /* Check if it is delta, and the base is also an object
 660                  * we are going to pack.  If so we will reuse the existing
 661                  * delta.
 662                  */
 663                 if (!no_reuse_delta &&
 664                     entry->in_pack_type == OBJ_DELTA &&
 665                     (base_entry = locate_object_entry(base)) &&
 666                     (!base_entry->preferred_base)) {
 667
 668                         /* Depth value does not matter - find_deltas()
 669                          * will never consider reused delta as the
 670                          * base object to deltify other objects
 671                          * against, in order to avoid circular deltas.
 672                          */
 673
 674                         /* uncompressed size of the delta data */
 675                         entry->size = entry->delta_size = size;
 676                         entry->delta = base_entry;
 677                         entry->type = OBJ_DELTA;
 678
 679                         entry->delta_sibling = base_entry->delta_child;
 680                         base_entry->delta_child = entry;
 681
 682                         return;
 683                 }
 684                 /* Otherwise we would do the usual */
 685         }
 686
 687         if (sha1_object_info(entry->sha1, type, &entry->size))
 688                 die("unable to get type of object %s",
 689                     sha1_to_hex(entry->sha1));
 690
 691         if (!strcmp(type, commit_type)) {
 692                 entry->type = OBJ_COMMIT;
 693         } else if (!strcmp(type, tree_type)) {
 694                 entry->type = OBJ_TREE;
 695         } else if (!strcmp(type, blob_type)) {
 696                 entry->type = OBJ_BLOB;
 697         } else if (!strcmp(type, tag_type)) {
 698                 entry->type = OBJ_TAG;
 699         } else
 700                 die("unable to pack object %s of type %s",
 701                     sha1_to_hex(entry->sha1), type);
 702 }
 703
 704 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
 705 {
 706         struct object_entry *child = me->delta_child;
 707         unsigned int m = n;
 708         while (child) {
 709                 unsigned int c = check_delta_limit(child, n + 1);
 710                 if (m < c)
 711                         m = c;
 712                 child = child->delta_sibling;
 713         }
 714         return m;
 715 }
 716
 717 static void get_object_details(void)
 718 {
 719         int i;
 720         struct object_entry *entry;
 721
 722         prepare_pack_ix();
 723         for (i = 0, entry = objects; i < nr_objects; i++, entry++)
 724                 check_object(entry);
 725
 726         if (nr_objects == nr_result) {
 727                 /*
 728                  * Depth of objects that depend on the entry -- this
 729                  * is subtracted from depth-max to break too deep
 730                  * delta chain because of delta data reusing.
 731                  * However, we loosen this restriction when we know we
 732                  * are creating a thin pack -- it will have to be
 733                  * expanded on the other end anyway, so do not
 734                  * artificially cut the delta chain and let it go as
 735                  * deep as it wants.
 736                  */
 737                 for (i = 0, entry = objects; i < nr_objects; i++, entry++)
 738                         if (!entry->delta && entry->delta_child)
 739                                 entry->delta_limit =
 740                                         check_delta_limit(entry, 1);
 741         }
 742 }
 743
 744 typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *);
 745
 746 static entry_sort_t current_sort;
 747
 748 static int sort_comparator(const void *_a, const void *_b)
 749 {
 750         struct object_entry *a = *(struct object_entry **)_a;
 751         struct object_entry *b = *(struct object_entry **)_b;
 752         return current_sort(a,b);
 753 }
 754
 755 static struct object_entry **create_sorted_list(entry_sort_t sort)
 756 {
 757         struct object_entry **list = xmalloc(nr_objects * sizeof(struct object_entry *));
 758         int i;
 759
 760         for (i = 0; i < nr_objects; i++)
 761                 list[i] = objects + i;
 762         current_sort = sort;
 763         qsort(list, nr_objects, sizeof(struct object_entry *), sort_comparator);
 764         return list;
 765 }
 766
 767 static int sha1_sort(const struct object_entry *a, const struct object_entry *b)
 768 {
 769         return memcmp(a->sha1, b->sha1, 20);
 770 }
 771
 772 static struct object_entry **create_final_object_list(void)
 773 {
 774         struct object_entry **list;
 775         int i, j;
 776
 777         for (i = nr_result = 0; i < nr_objects; i++)
 778                 if (!objects[i].preferred_base)
 779                         nr_result++;
 780         list = xmalloc(nr_result * sizeof(struct object_entry *));
 781         for (i = j = 0; i < nr_objects; i++) {
 782                 if (!objects[i].preferred_base)
 783                         list[j++] = objects + i;
 784         }
 785         current_sort = sha1_sort;
 786         qsort(list, nr_result, sizeof(struct object_entry *), sort_comparator);
 787         return list;
 788 }
 789
 790 static int type_size_sort(const struct object_entry *a, const struct object_entry *b)
 791 {
 792         if (a->type < b->type)
 793                 return -1;
 794         if (a->type > b->type)
 795                 return 1;
 796         if (a->hash < b->hash)
 797                 return -1;
 798         if (a->hash > b->hash)
 799                 return 1;
 800         if (a->preferred_base < b->preferred_base)
 801                 return -1;
 802         if (a->preferred_base > b->preferred_base)
 803                 return 1;
 804         if (a->size < b->size)
 805                 return -1;
 806         if (a->size > b->size)
 807                 return 1;
 808         return a < b ? -1 : (a > b);
 809 }
 810
 811 struct unpacked {
 812         struct object_entry *entry;
 813         void *data;
 814 };
 815
 816 /*
 817  * We search for deltas _backwards_ in a list sorted by type and
 818  * by size, so that we see progressively smaller and smaller files.
 819  * That's because we prefer deltas to be from the bigger file
 820  * to the smaller - deletes are potentially cheaper, but perhaps
 821  * more importantly, the bigger file is likely the more recent
 822  * one.
 823  */
 824 static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_depth)
 825 {
 826         struct object_entry *cur_entry = cur->entry;
 827         struct object_entry *old_entry = old->entry;
 828         unsigned long size, oldsize, delta_size, sizediff;
 829         long max_size;
 830         void *delta_buf;
 831
 832         /* Don't bother doing diffs between different types */
 833         if (cur_entry->type != old_entry->type)
 834                 return -1;
 835
 836         /* We do not compute delta to *create* objects we are not
 837          * going to pack.
 838          */
 839         if (cur_entry->preferred_base)
 840                 return -1;
 841
 842         /* If the current object is at pack edge, take the depth the
 843          * objects that depend on the current object into account --
 844          * otherwise they would become too deep.
 845          */
 846         if (cur_entry->delta_child) {
 847                 if (max_depth <= cur_entry->delta_limit)
 848                         return 0;
 849                 max_depth -= cur_entry->delta_limit;
 850         }
 851
 852         size = cur_entry->size;
 853         oldsize = old_entry->size;
 854         sizediff = oldsize > size ? oldsize - size : size - oldsize;
 855
 856         if (size < 50)
 857                 return -1;
 858         if (old_entry->depth >= max_depth)
 859                 return 0;
 860
 861         /*
 862          * NOTE!
 863          *
 864          * We always delta from the bigger to the smaller, since that's
 865          * more space-efficient (deletes don't have to say _what_ they
 866          * delete).
 867          */
 868         max_size = size / 2 - 20;
 869         if (cur_entry->delta)
 870                 max_size = cur_entry->delta_size-1;
 871         if (sizediff >= max_size)
 872                 return -1;
 873         delta_buf = diff_delta(old->data, oldsize,
 874                                cur->data, size, &delta_size, max_size);
 875         if (!delta_buf)
 876                 return 0;
 877         cur_entry->delta = old_entry;
 878         cur_entry->delta_size = delta_size;
 879         cur_entry->depth = old_entry->depth + 1;
 880         free(delta_buf);
 881         return 0;
 882 }
 883
 884 static void progress_interval(int signum)
 885 {
 886         signal(SIGALRM, progress_interval);
 887         progress_update = 1;
 888 }
 889
 890 static void find_deltas(struct object_entry **list, int window, int depth)
 891 {
 892         int i, idx;
 893         unsigned int array_size = window * sizeof(struct unpacked);
 894         struct unpacked *array = xmalloc(array_size);
 895         unsigned processed = 0;
 896         unsigned last_percent = 999;
 897
 898         memset(array, 0, array_size);
 899         i = nr_objects;
 900         idx = 0;
 901         if (progress)
 902                 fprintf(stderr, "Deltifying %d objects.\n", nr_result);
 903
 904         while (--i >= 0) {
 905                 struct object_entry *entry = list[i];
 906                 struct unpacked *n = array + idx;
 907                 unsigned long size;
 908                 char type[10];
 909                 int j;
 910
 911                 if (!entry->preferred_base)
 912                         processed++;
 913
 914                 if (progress) {
 915                         unsigned percent = processed * 100 / nr_result;
 916                         if (percent != last_percent || progress_update) {
 917                                 fprintf(stderr, "%4u%% (%u/%u) done\r",
 918                                         percent, processed, nr_result);
 919                                 progress_update = 0;
 920                                 last_percent = percent;
 921                         }
 922                 }
 923
 924                 if (entry->delta)
 925                         /* This happens if we decided to reuse existing
 926                          * delta from a pack.  "!no_reuse_delta &&" is implied.
 927                          */
 928                         continue;
 929
 930                 free(n->data);
 931                 n->entry = entry;
 932                 n->data = read_sha1_file(entry->sha1, type, &size);
 933                 if (size != entry->size)
 934                         die("object %s inconsistent object length (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
 935
 936                 j = window;
 937                 while (--j > 0) {
 938                         unsigned int other_idx = idx + j;
 939                         struct unpacked *m;
 940                         if (other_idx >= window)
 941                                 other_idx -= window;
 942                         m = array + other_idx;
 943                         if (!m->entry)
 944                                 break;
 945                         if (try_delta(n, m, depth) < 0)
 946                                 break;
 947                 }
 948 #if 0
 949                 /* if we made n a delta, and if n is already at max
 950                  * depth, leaving it in the window is pointless.  we
 951                  * should evict it first.
 952                  * ... in theory only; somehow this makes things worse.
 953                  */
 954                 if (entry->delta && depth <= entry->depth)
 955                         continue;
 956 #endif
 957                 idx++;
 958                 if (idx >= window)
 959                         idx = 0;
 960         }
 961
 962         if (progress)
 963                 fputc('\n', stderr);
 964
 965         for (i = 0; i < window; ++i)
 966                 free(array[i].data);
 967         free(array);
 968 }
 969
 970 static void prepare_pack(int window, int depth)
 971 {
 972         get_object_details();
 973         sorted_by_type = create_sorted_list(type_size_sort);
 974         if (window && depth)
 975                 find_deltas(sorted_by_type, window+1, depth);
 976 }
 977
 978 static int reuse_cached_pack(unsigned char *sha1, int pack_to_stdout)
 979 {
 980         static const char cache[] = "pack-cache/pack-%s.%s";
 981         char *cached_pack, *cached_idx;
 982         int ifd, ofd, ifd_ix = -1;
 983
 984         cached_pack = git_path(cache, sha1_to_hex(sha1), "pack");
 985         ifd = open(cached_pack, O_RDONLY);
 986         if (ifd < 0)
 987                 return 0;
 988
 989         if (!pack_to_stdout) {
 990                 cached_idx = git_path(cache, sha1_to_hex(sha1), "idx");
 991                 ifd_ix = open(cached_idx, O_RDONLY);
 992                 if (ifd_ix < 0) {
 993                         close(ifd);
 994                         return 0;
 995                 }
 996         }
 997
 998         if (progress)
 999                 fprintf(stderr, "Reusing %d objects pack %s\n", nr_objects,
1000                         sha1_to_hex(sha1));
1001
1002         if (pack_to_stdout) {
1003                 if (copy_fd(ifd, 1))
1004                         exit(1);
1005                 close(ifd);
1006         }
1007         else {
1008                 char name[PATH_MAX];
1009                 snprintf(name, sizeof(name),
1010                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "pack");
1011                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1012                 if (ofd < 0)
1013                         die("unable to open %s (%s)", name, strerror(errno));
1014                 if (copy_fd(ifd, ofd))
1015                         exit(1);
1016                 close(ifd);
1017
1018                 snprintf(name, sizeof(name),
1019                          "%s-%s.%s", base_name, sha1_to_hex(sha1), "idx");
1020                 ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
1021                 if (ofd < 0)
1022                         die("unable to open %s (%s)", name, strerror(errno));
1023                 if (copy_fd(ifd_ix, ofd))
1024                         exit(1);
1025                 close(ifd_ix);
1026                 puts(sha1_to_hex(sha1));
1027         }
1028
1029         return 1;
1030 }
1031
1032 int main(int argc, char **argv)
1033 {
1034         SHA_CTX ctx;
1035         char line[PATH_MAX + 20];
1036         int window = 10, depth = 10, pack_to_stdout = 0;
1037         struct object_entry **list;
1038         int i;
1039
1040         setup_git_directory();
1041
1042         for (i = 1; i < argc; i++) {
1043                 const char *arg = argv[i];
1044
1045                 if (*arg == '-') {
1046                         if (!strcmp("--non-empty", arg)) {
1047                                 non_empty = 1;
1048                                 continue;
1049                         }
1050                         if (!strcmp("--local", arg)) {
1051                                 local = 1;
1052                                 continue;
1053                         }
1054                         if (!strcmp("--incremental", arg)) {
1055                                 incremental = 1;
1056                                 continue;
1057                         }
1058                         if (!strncmp("--window=", arg, 9)) {
1059                                 char *end;
1060                                 window = strtoul(arg+9, &end, 0);
1061                                 if (!arg[9] || *end)
1062                                         usage(pack_usage);
1063                                 continue;
1064                         }
1065                         if (!strncmp("--depth=", arg, 8)) {
1066                                 char *end;
1067                                 depth = strtoul(arg+8, &end, 0);
1068                                 if (!arg[8] || *end)
1069                                         usage(pack_usage);
1070                                 continue;
1071                         }
1072                         if (!strcmp("-q", arg)) {
1073                                 progress = 0;
1074                                 continue;
1075                         }
1076                         if (!strcmp("--no-reuse-delta", arg)) {
1077                                 no_reuse_delta = 1;
1078                                 continue;
1079                         }
1080                         if (!strcmp("--stdout", arg)) {
1081                                 pack_to_stdout = 1;
1082                                 continue;
1083                         }
1084                         usage(pack_usage);
1085                 }
1086                 if (base_name)
1087                         usage(pack_usage);
1088                 base_name = arg;
1089         }
1090
1091         if (pack_to_stdout != !base_name)
1092                 usage(pack_usage);
1093
1094         prepare_packed_git();
1095
1096         if (progress) {
1097                 struct itimerval v;
1098                 v.it_interval.tv_sec = 1;
1099                 v.it_interval.tv_usec = 0;
1100                 v.it_value = v.it_interval;
1101                 signal(SIGALRM, progress_interval);
1102                 setitimer(ITIMER_REAL, &v, NULL);
1103                 fprintf(stderr, "Generating pack...\n");
1104         }
1105
1106         while (fgets(line, sizeof(line), stdin) != NULL) {
1107                 unsigned char sha1[20];
1108
1109                 if (line[0] == '-') {
1110                         if (get_sha1_hex(line+1, sha1))
1111                                 die("expected edge sha1, got garbage:\n %s",
1112                                     line+1);
1113                         add_preferred_base(sha1);
1114                         continue;
1115                 }
1116                 if (get_sha1_hex(line, sha1))
1117                         die("expected sha1, got garbage:\n %s", line);
1118                 add_object_entry(sha1, name_hash(NULL, line+41), 0);
1119         }
1120         if (progress)
1121                 fprintf(stderr, "Done counting %d objects.\n", nr_objects);
1122         sorted_by_sha = create_final_object_list();
1123         if (non_empty && !nr_result)
1124                 return 0;
1125
1126         SHA1_Init(&ctx);
1127         list = sorted_by_sha;
1128         for (i = 0; i < nr_result; i++) {
1129                 struct object_entry *entry = *list++;
1130                 SHA1_Update(&ctx, entry->sha1, 20);
1131         }
1132         SHA1_Final(object_list_sha1, &ctx);
1133         if (progress && (nr_objects != nr_result))
1134                 fprintf(stderr, "Result has %d objects.\n", nr_result);
1135
1136         if (reuse_cached_pack(object_list_sha1, pack_to_stdout))
1137                 ;
1138         else {
1139                 if (nr_result)
1140                         prepare_pack(window, depth);
1141                 if (progress && pack_to_stdout) {
1142                         /* the other end usually displays progress itself */
1143                         struct itimerval v = {{0,},};
1144                         setitimer(ITIMER_REAL, &v, NULL);
1145                         signal(SIGALRM, SIG_IGN );
1146                         progress_update = 0;
1147                 }
1148                 write_pack_file();
1149                 if (!pack_to_stdout) {
1150                         write_index_file();
1151                         puts(sha1_to_hex(object_list_sha1));
1152                 }
1153         }
1154         if (progress)
1155                 fprintf(stderr, "Total %d, written %d (delta %d), reused %d (delta %d)\n",
1156                         nr_result, written, written_delta, reused, reused_delta);
1157         return 0;
1158 }