pack-objects.c

   1 #include <ctype.h>
   2 #include "cache.h"
   3 #include "object.h"
   4 #include "delta.h"
   5 #include "pack.h"
   6 #include "csum-file.h"
   7
   8 static const char pack_usage[] = "git-pack-objects [--local] [--incremental] [--window=N] [--depth=N] {--stdout | base-name} < object-list";
   9
  10 struct object_entry {
  11         unsigned char sha1[20];
  12         unsigned long size;
  13         unsigned long offset;
  14         unsigned int depth;
  15         unsigned int hash;
  16         enum object_type type;
  17         unsigned long delta_size;
  18         struct object_entry *delta;
  19 };
  20
  21 static unsigned char object_list_sha1[20];
  22 static int non_empty = 0;
  23 static int local = 0;
  24 static int incremental = 0;
  25 static struct object_entry **sorted_by_sha, **sorted_by_type;
  26 static struct object_entry *objects = NULL;
  27 static int nr_objects = 0, nr_alloc = 0;
  28 static const char *base_name;
  29 static unsigned char pack_file_sha1[20];
  30
  31 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
  32 {
  33         unsigned long othersize, delta_size;
  34         char type[10];
  35         void *otherbuf = read_sha1_file(entry->delta->sha1, type, &othersize);
  36         void *delta_buf;
  37
  38         if (!otherbuf)
  39                 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
  40         delta_buf = diff_delta(otherbuf, othersize,
  41                                buf, size, &delta_size, 0);
  42         if (!delta_buf || delta_size != entry->delta_size)
  43                 die("delta size changed");
  44         free(buf);
  45         free(otherbuf);
  46         return delta_buf;
  47 }
  48
  49 /*
  50  * The per-object header is a pretty dense thing, which is
  51  *  - first byte: low four bits are "size", then three bits of "type",
  52  *    and the high bit is "size continues".
  53  *  - each byte afterwards: low seven bits are size continuation,
  54  *    with the high bit being "size continues"
  55  */
  56 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
  57 {
  58         int n = 1;
  59         unsigned char c;
  60
  61         if (type < OBJ_COMMIT || type > OBJ_DELTA)
  62                 die("bad type %d", type);
  63
  64         c = (type << 4) | (size & 15);
  65         size >>= 4;
  66         while (size) {
  67                 *hdr++ = c | 0x80;
  68                 c = size & 0x7f;
  69                 size >>= 7;
  70                 n++;
  71         }
  72         *hdr = c;
  73         return n;
  74 }
  75
  76 static unsigned long write_object(struct sha1file *f, struct object_entry *entry)
  77 {
  78         unsigned long size;
  79         char type[10];
  80         void *buf = read_sha1_file(entry->sha1, type, &size);
  81         unsigned char header[10];
  82         unsigned hdrlen, datalen;
  83         enum object_type obj_type;
  84
  85         if (!buf)
  86                 die("unable to read %s", sha1_to_hex(entry->sha1));
  87         if (size != entry->size)
  88                 die("object %s size inconsistency (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
  89
  90         /*
  91          * The object header is a byte of 'type' followed by zero or
  92          * more bytes of length.  For deltas, the 20 bytes of delta sha1
  93          * follows that.
  94          */
  95         obj_type = entry->type;
  96         if (entry->delta) {
  97                 buf = delta_against(buf, size, entry);
  98                 size = entry->delta_size;
  99                 obj_type = OBJ_DELTA;
 100         }
 101         hdrlen = encode_header(obj_type, size, header);
 102         sha1write(f, header, hdrlen);
 103         if (entry->delta) {
 104                 sha1write(f, entry->delta, 20);
 105                 hdrlen += 20;
 106         }
 107         datalen = sha1write_compressed(f, buf, size);
 108         free(buf);
 109         return hdrlen + datalen;
 110 }
 111
 112 static unsigned long write_one(struct sha1file *f,
 113                                struct object_entry *e,
 114                                unsigned long offset)
 115 {
 116         if (e->offset)
 117                 /* offset starts from header size and cannot be zero
 118                  * if it is written already.
 119                  */
 120                 return offset;
 121         e->offset = offset;
 122         offset += write_object(f, e);
 123         /* if we are delitified, write out its base object. */
 124         if (e->delta)
 125                 offset = write_one(f, e->delta, offset);
 126         return offset;
 127 }
 128
 129 static void write_pack_file(void)
 130 {
 131         int i;
 132         struct sha1file *f;
 133         unsigned long offset;
 134         unsigned long mb;
 135         struct pack_header hdr;
 136
 137         if (!base_name)
 138                 f = sha1fd(1, "<stdout>");
 139         else
 140                 f = sha1create("%s-%s.%s", base_name, sha1_to_hex(object_list_sha1), "pack");
 141         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 142         hdr.hdr_version = htonl(PACK_VERSION);
 143         hdr.hdr_entries = htonl(nr_objects);
 144         sha1write(f, &hdr, sizeof(hdr));
 145         offset = sizeof(hdr);
 146         for (i = 0; i < nr_objects; i++)
 147                 offset = write_one(f, objects + i, offset);
 148
 149         sha1close(f, pack_file_sha1, 1);
 150         mb = offset >> 20;
 151         offset &= 0xfffff;
 152 }
 153
 154 static void write_index_file(void)
 155 {
 156         int i;
 157         struct sha1file *f = sha1create("%s-%s.%s", base_name, sha1_to_hex(object_list_sha1), "idx");
 158         struct object_entry **list = sorted_by_sha;
 159         struct object_entry **last = list + nr_objects;
 160         unsigned int array[256];
 161
 162         /*
 163          * Write the first-level table (the list is sorted,
 164          * but we use a 256-entry lookup to be able to avoid
 165          * having to do eight extra binary search iterations).
 166          */
 167         for (i = 0; i < 256; i++) {
 168                 struct object_entry **next = list;
 169                 while (next < last) {
 170                         struct object_entry *entry = *next;
 171                         if (entry->sha1[0] != i)
 172                                 break;
 173                         next++;
 174                 }
 175                 array[i] = htonl(next - sorted_by_sha);
 176                 list = next;
 177         }
 178         sha1write(f, array, 256 * sizeof(int));
 179
 180         /*
 181          * Write the actual SHA1 entries..
 182          */
 183         list = sorted_by_sha;
 184         for (i = 0; i < nr_objects; i++) {
 185                 struct object_entry *entry = *list++;
 186                 unsigned int offset = htonl(entry->offset);
 187                 sha1write(f, &offset, 4);
 188                 sha1write(f, entry->sha1, 20);
 189         }
 190         sha1write(f, pack_file_sha1, 20);
 191         sha1close(f, NULL, 1);
 192 }
 193
 194 static int add_object_entry(unsigned char *sha1, unsigned int hash)
 195 {
 196         unsigned int idx = nr_objects;
 197         struct object_entry *entry;
 198
 199         if (incremental || local) {
 200                 struct packed_git *p;
 201
 202                 for (p = packed_git; p; p = p->next) {
 203                         struct pack_entry e;
 204
 205                         if (find_pack_entry_one(sha1, &e, p)) {
 206                                 if (incremental)
 207                                         return 0;
 208                                 if (local && !p->pack_local)
 209                                         return 0;
 210                         }
 211                 }
 212         }
 213
 214         if (idx >= nr_alloc) {
 215                 unsigned int needed = (idx + 1024) * 3 / 2;
 216                 objects = xrealloc(objects, needed * sizeof(*entry));
 217                 nr_alloc = needed;
 218         }
 219         entry = objects + idx;
 220         memset(entry, 0, sizeof(*entry));
 221         memcpy(entry->sha1, sha1, 20);
 222         entry->hash = hash;
 223         nr_objects = idx+1;
 224         return 1;
 225 }
 226
 227 static void check_object(struct object_entry *entry)
 228 {
 229         char type[20];
 230
 231         if (!sha1_object_info(entry->sha1, type, &entry->size)) {
 232                 if (!strcmp(type, "commit")) {
 233                         entry->type = OBJ_COMMIT;
 234                 } else if (!strcmp(type, "tree")) {
 235                         entry->type = OBJ_TREE;
 236                 } else if (!strcmp(type, "blob")) {
 237                         entry->type = OBJ_BLOB;
 238                 } else if (!strcmp(type, "tag")) {
 239                         entry->type = OBJ_TAG;
 240                 } else
 241                         die("unable to pack object %s of type %s",
 242                             sha1_to_hex(entry->sha1), type);
 243         }
 244         else
 245                 die("unable to get type of object %s",
 246                     sha1_to_hex(entry->sha1));
 247 }
 248
 249 static void get_object_details(void)
 250 {
 251         int i;
 252         struct object_entry *entry = objects;
 253
 254         for (i = 0; i < nr_objects; i++)
 255                 check_object(entry++);
 256 }
 257
 258 typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *);
 259
 260 static entry_sort_t current_sort;
 261
 262 static int sort_comparator(const void *_a, const void *_b)
 263 {
 264         struct object_entry *a = *(struct object_entry **)_a;
 265         struct object_entry *b = *(struct object_entry **)_b;
 266         return current_sort(a,b);
 267 }
 268
 269 static struct object_entry **create_sorted_list(entry_sort_t sort)
 270 {
 271         struct object_entry **list = xmalloc(nr_objects * sizeof(struct object_entry *));
 272         int i;
 273
 274         for (i = 0; i < nr_objects; i++)
 275                 list[i] = objects + i;
 276         current_sort = sort;
 277         qsort(list, nr_objects, sizeof(struct object_entry *), sort_comparator);
 278         return list;
 279 }
 280
 281 static int sha1_sort(const struct object_entry *a, const struct object_entry *b)
 282 {
 283         return memcmp(a->sha1, b->sha1, 20);
 284 }
 285
 286 static int type_size_sort(const struct object_entry *a, const struct object_entry *b)
 287 {
 288         if (a->type < b->type)
 289                 return -1;
 290         if (a->type > b->type)
 291                 return 1;
 292         if (a->hash < b->hash)
 293                 return -1;
 294         if (a->hash > b->hash)
 295                 return 1;
 296         if (a->size < b->size)
 297                 return -1;
 298         if (a->size > b->size)
 299                 return 1;
 300         return a < b ? -1 : (a > b);
 301 }
 302
 303 struct unpacked {
 304         struct object_entry *entry;
 305         void *data;
 306 };
 307
 308 /*
 309  * We search for deltas _backwards_ in a list sorted by type and
 310  * by size, so that we see progressively smaller and smaller files.
 311  * That's because we prefer deltas to be from the bigger file
 312  * to the smaller - deletes are potentially cheaper, but perhaps
 313  * more importantly, the bigger file is likely the more recent
 314  * one.
 315  */
 316 static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_depth)
 317 {
 318         struct object_entry *cur_entry = cur->entry;
 319         struct object_entry *old_entry = old->entry;
 320         unsigned long size, oldsize, delta_size, sizediff;
 321         long max_size;
 322         void *delta_buf;
 323
 324         /* Don't bother doing diffs between different types */
 325         if (cur_entry->type != old_entry->type)
 326                 return -1;
 327
 328         size = cur_entry->size;
 329         if (size < 50)
 330                 return -1;
 331         oldsize = old_entry->size;
 332         sizediff = oldsize > size ? oldsize - size : size - oldsize;
 333         if (sizediff > size / 8)
 334                 return -1;
 335         if (old_entry->depth >= max_depth)
 336                 return 0;
 337
 338         /*
 339          * NOTE!
 340          *
 341          * We always delta from the bigger to the smaller, since that's
 342          * more space-efficient (deletes don't have to say _what_ they
 343          * delete).
 344          */
 345         max_size = size / 2 - 20;
 346         if (cur_entry->delta)
 347                 max_size = cur_entry->delta_size-1;
 348         if (sizediff >= max_size)
 349                 return -1;
 350         delta_buf = diff_delta(old->data, oldsize,
 351                                cur->data, size, &delta_size, max_size);
 352         if (!delta_buf)
 353                 return 0;
 354         cur_entry->delta = old_entry;
 355         cur_entry->delta_size = delta_size;
 356         cur_entry->depth = old_entry->depth + 1;
 357         free(delta_buf);
 358         return 0;
 359 }
 360
 361 static void find_deltas(struct object_entry **list, int window, int depth)
 362 {
 363         int i, idx;
 364         unsigned int array_size = window * sizeof(struct unpacked);
 365         struct unpacked *array = xmalloc(array_size);
 366
 367         memset(array, 0, array_size);
 368         i = nr_objects;
 369         idx = 0;
 370         while (--i >= 0) {
 371                 struct object_entry *entry = list[i];
 372                 struct unpacked *n = array + idx;
 373                 unsigned long size;
 374                 char type[10];
 375                 int j;
 376
 377                 free(n->data);
 378                 n->entry = entry;
 379                 n->data = read_sha1_file(entry->sha1, type, &size);
 380                 if (size != entry->size)
 381                         die("object %s inconsistent object length (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
 382                 j = window;
 383                 while (--j > 0) {
 384                         unsigned int other_idx = idx + j;
 385                         struct unpacked *m;
 386                         if (other_idx >= window)
 387                                 other_idx -= window;
 388                         m = array + other_idx;
 389                         if (!m->entry)
 390                                 break;
 391                         if (try_delta(n, m, depth) < 0)
 392                                 break;
 393                 }
 394                 idx++;
 395                 if (idx >= window)
 396                         idx = 0;
 397         }
 398
 399         for (i = 0; i < window; ++i)
 400                 free(array[i].data);
 401         free(array);
 402 }
 403
 404 int main(int argc, char **argv)
 405 {
 406         SHA_CTX ctx;
 407         char line[PATH_MAX + 20];
 408         int window = 10, depth = 10, pack_to_stdout = 0;
 409         struct object_entry **list;
 410         int i;
 411
 412         for (i = 1; i < argc; i++) {
 413                 const char *arg = argv[i];
 414
 415                 if (*arg == '-') {
 416                         if (!strcmp("--non-empty", arg)) {
 417                                 non_empty = 1;
 418                                 continue;
 419                         }
 420                         if (!strcmp("--local", arg)) {
 421                                 local = 1;
 422                                 continue;
 423                         }
 424                         if (!strcmp("--incremental", arg)) {
 425                                 incremental = 1;
 426                                 continue;
 427                         }
 428                         if (!strncmp("--window=", arg, 9)) {
 429                                 char *end;
 430                                 window = strtoul(arg+9, &end, 0);
 431                                 if (!arg[9] || *end)
 432                                         usage(pack_usage);
 433                                 continue;
 434                         }
 435                         if (!strncmp("--depth=", arg, 8)) {
 436                                 char *end;
 437                                 depth = strtoul(arg+8, &end, 0);
 438                                 if (!arg[8] || *end)
 439                                         usage(pack_usage);
 440                                 continue;
 441                         }
 442                         if (!strcmp("--stdout", arg)) {
 443                                 pack_to_stdout = 1;
 444                                 continue;
 445                         }
 446                         usage(pack_usage);
 447                 }
 448                 if (base_name)
 449                         usage(pack_usage);
 450                 base_name = arg;
 451         }
 452
 453         if (pack_to_stdout != !base_name)
 454                 usage(pack_usage);
 455
 456         prepare_packed_git();
 457         while (fgets(line, sizeof(line), stdin) != NULL) {
 458                 unsigned int hash;
 459                 char *p;
 460                 unsigned char sha1[20];
 461
 462                 if (get_sha1_hex(line, sha1))
 463                         die("expected sha1, got garbage");
 464                 hash = 0;
 465                 p = line+40;
 466                 while (*p) {
 467                         unsigned char c = *p++;
 468                         if (isspace(c))
 469                                 continue;
 470                         hash = hash * 11 + c;
 471                 }
 472                 add_object_entry(sha1, hash);
 473         }
 474         if (non_empty && !nr_objects)
 475                 return 0;
 476         get_object_details();
 477
 478         fprintf(stderr, "Packing %d objects\n", nr_objects);
 479
 480         sorted_by_sha = create_sorted_list(sha1_sort);
 481         SHA1_Init(&ctx);
 482         list = sorted_by_sha;
 483         for (i = 0; i < nr_objects; i++) {
 484                 struct object_entry *entry = *list++;
 485                 SHA1_Update(&ctx, entry->sha1, 20);
 486         }
 487         SHA1_Final(object_list_sha1, &ctx);
 488
 489         sorted_by_type = create_sorted_list(type_size_sort);
 490         if (window && depth)
 491                 find_deltas(sorted_by_type, window+1, depth);
 492
 493         write_pack_file();
 494         if (!pack_to_stdout) {
 495                 write_index_file();
 496                 puts(sha1_to_hex(object_list_sha1));
 497         }
 498         return 0;
 499 }