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 [--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 int non_empty = 0;
  22 static int incremental = 0;
  23 static struct object_entry **sorted_by_sha, **sorted_by_type;
  24 static struct object_entry *objects = NULL;
  25 static int nr_objects = 0, nr_alloc = 0;
  26 static const char *base_name;
  27 static unsigned char pack_file_sha1[20];
  28
  29 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
  30 {
  31         unsigned long othersize, delta_size;
  32         char type[10];
  33         void *otherbuf = read_sha1_file(entry->delta->sha1, type, &othersize);
  34         void *delta_buf;
  35
  36         if (!otherbuf)
  37                 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
  38         delta_buf = diff_delta(otherbuf, othersize,
  39                                buf, size, &delta_size, 0);
  40         if (!delta_buf || delta_size != entry->delta_size)
  41                 die("delta size changed");
  42         free(buf);
  43         free(otherbuf);
  44         return delta_buf;
  45 }
  46
  47 /*
  48  * The per-object header is a pretty dense thing, which is
  49  *  - first byte: low four bits are "size", then three bits of "type",
  50  *    and the high bit is "size continues".
  51  *  - each byte afterwards: low seven bits are size continuation,
  52  *    with the high bit being "size continues"
  53  */
  54 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
  55 {
  56         int n = 1;
  57         unsigned char c;
  58
  59         if (type < OBJ_COMMIT || type > OBJ_DELTA)
  60                 die("bad type %d", type);
  61
  62         c = (type << 4) | (size & 15);
  63         size >>= 4;
  64         while (size) {
  65                 *hdr++ = c | 0x80;
  66                 c = size & 0x7f;
  67                 size >>= 7;
  68                 n++;
  69         }
  70         *hdr = c;
  71         return n;
  72 }
  73
  74 static unsigned long write_object(struct sha1file *f, struct object_entry *entry)
  75 {
  76         unsigned long size;
  77         char type[10];
  78         void *buf = read_sha1_file(entry->sha1, type, &size);
  79         unsigned char header[10];
  80         unsigned hdrlen, datalen;
  81         enum object_type obj_type;
  82
  83         if (!buf)
  84                 die("unable to read %s", sha1_to_hex(entry->sha1));
  85         if (size != entry->size)
  86                 die("object %s size inconsistency (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
  87
  88         /*
  89          * The object header is a byte of 'type' followed by zero or
  90          * more bytes of length.  For deltas, the 20 bytes of delta sha1
  91          * follows that.
  92          */
  93         obj_type = entry->type;
  94         if (entry->delta) {
  95                 buf = delta_against(buf, size, entry);
  96                 size = entry->delta_size;
  97                 obj_type = OBJ_DELTA;
  98         }
  99         hdrlen = encode_header(obj_type, size, header);
 100         sha1write(f, header, hdrlen);
 101         if (entry->delta) {
 102                 sha1write(f, entry->delta, 20);
 103                 hdrlen += 20;
 104         }
 105         datalen = sha1write_compressed(f, buf, size);
 106         free(buf);
 107         return hdrlen + datalen;
 108 }
 109
 110 static unsigned long write_one(struct sha1file *f,
 111                                struct object_entry *e,
 112                                unsigned long offset)
 113 {
 114         if (e->offset)
 115                 /* offset starts from header size and cannot be zero
 116                  * if it is written already.
 117                  */
 118                 return offset;
 119         e->offset = offset;
 120         offset += write_object(f, e);
 121         /* if we are delitified, write out its base object. */
 122         if (e->delta)
 123                 offset = write_one(f, e->delta, offset);
 124         return offset;
 125 }
 126
 127 static void write_pack_file(void)
 128 {
 129         int i;
 130         struct sha1file *f;
 131         unsigned long offset;
 132         unsigned long mb;
 133         struct pack_header hdr;
 134
 135         if (!base_name)
 136                 f = sha1fd(1, "<stdout>");
 137         else
 138                 f = sha1create("%s.%s", base_name, "pack");
 139         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 140         hdr.hdr_version = htonl(PACK_VERSION);
 141         hdr.hdr_entries = htonl(nr_objects);
 142         sha1write(f, &hdr, sizeof(hdr));
 143         offset = sizeof(hdr);
 144         for (i = 0; i < nr_objects; i++)
 145                 offset = write_one(f, objects + i, offset);
 146
 147         sha1close(f, pack_file_sha1, 1);
 148         mb = offset >> 20;
 149         offset &= 0xfffff;
 150 }
 151
 152 static void write_index_file(void)
 153 {
 154         int i;
 155         struct sha1file *f = sha1create("%s.%s", base_name, "idx");
 156         struct object_entry **list = sorted_by_sha;
 157         struct object_entry **last = list + nr_objects;
 158         unsigned int array[256];
 159
 160         /*
 161          * Write the first-level table (the list is sorted,
 162          * but we use a 256-entry lookup to be able to avoid
 163          * having to do eight extra binary search iterations).
 164          */
 165         for (i = 0; i < 256; i++) {
 166                 struct object_entry **next = list;
 167                 while (next < last) {
 168                         struct object_entry *entry = *next;
 169                         if (entry->sha1[0] != i)
 170                                 break;
 171                         next++;
 172                 }
 173                 array[i] = htonl(next - sorted_by_sha);
 174                 list = next;
 175         }
 176         sha1write(f, array, 256 * sizeof(int));
 177
 178         /*
 179          * Write the actual SHA1 entries..
 180          */
 181         list = sorted_by_sha;
 182         for (i = 0; i < nr_objects; i++) {
 183                 struct object_entry *entry = *list++;
 184                 unsigned int offset = htonl(entry->offset);
 185                 sha1write(f, &offset, 4);
 186                 sha1write(f, entry->sha1, 20);
 187         }
 188         sha1write(f, pack_file_sha1, 20);
 189         sha1close(f, NULL, 1);
 190 }
 191
 192 static void add_object_entry(unsigned char *sha1, unsigned int hash)
 193 {
 194         unsigned int idx = nr_objects;
 195         struct object_entry *entry;
 196
 197         if (incremental && has_sha1_pack(sha1))
 198                 return;
 199
 200         if (idx >= nr_alloc) {
 201                 unsigned int needed = (idx + 1024) * 3 / 2;
 202                 objects = xrealloc(objects, needed * sizeof(*entry));
 203                 nr_alloc = needed;
 204         }
 205         entry = objects + idx;
 206         memset(entry, 0, sizeof(*entry));
 207         memcpy(entry->sha1, sha1, 20);
 208         entry->hash = hash;
 209         nr_objects = idx+1;
 210 }
 211
 212 static void check_object(struct object_entry *entry)
 213 {
 214         char type[20];
 215
 216         if (!sha1_object_info(entry->sha1, type, &entry->size)) {
 217                 if (!strcmp(type, "commit")) {
 218                         entry->type = OBJ_COMMIT;
 219                 } else if (!strcmp(type, "tree")) {
 220                         entry->type = OBJ_TREE;
 221                 } else if (!strcmp(type, "blob")) {
 222                         entry->type = OBJ_BLOB;
 223                 } else if (!strcmp(type, "tag")) {
 224                         entry->type = OBJ_TAG;
 225                 } else
 226                         die("unable to pack object %s of type %s",
 227                             sha1_to_hex(entry->sha1), type);
 228         }
 229         else
 230                 die("unable to get type of object %s",
 231                     sha1_to_hex(entry->sha1));
 232 }
 233
 234 static void get_object_details(void)
 235 {
 236         int i;
 237         struct object_entry *entry = objects;
 238
 239         for (i = 0; i < nr_objects; i++)
 240                 check_object(entry++);
 241 }
 242
 243 typedef int (*entry_sort_t)(const struct object_entry *, const struct object_entry *);
 244
 245 static entry_sort_t current_sort;
 246
 247 static int sort_comparator(const void *_a, const void *_b)
 248 {
 249         struct object_entry *a = *(struct object_entry **)_a;
 250         struct object_entry *b = *(struct object_entry **)_b;
 251         return current_sort(a,b);
 252 }
 253
 254 static struct object_entry **create_sorted_list(entry_sort_t sort)
 255 {
 256         struct object_entry **list = xmalloc(nr_objects * sizeof(struct object_entry *));
 257         int i;
 258
 259         for (i = 0; i < nr_objects; i++)
 260                 list[i] = objects + i;
 261         current_sort = sort;
 262         qsort(list, nr_objects, sizeof(struct object_entry *), sort_comparator);
 263         return list;
 264 }
 265
 266 static int sha1_sort(const struct object_entry *a, const struct object_entry *b)
 267 {
 268         return memcmp(a->sha1, b->sha1, 20);
 269 }
 270
 271 static int type_size_sort(const struct object_entry *a, const struct object_entry *b)
 272 {
 273         if (a->type < b->type)
 274                 return -1;
 275         if (a->type > b->type)
 276                 return 1;
 277         if (a->hash < b->hash)
 278                 return -1;
 279         if (a->hash > b->hash)
 280                 return 1;
 281         if (a->size < b->size)
 282                 return -1;
 283         if (a->size > b->size)
 284                 return 1;
 285         return a < b ? -1 : (a > b);
 286 }
 287
 288 struct unpacked {
 289         struct object_entry *entry;
 290         void *data;
 291 };
 292
 293 /*
 294  * We search for deltas _backwards_ in a list sorted by type and
 295  * by size, so that we see progressively smaller and smaller files.
 296  * That's because we prefer deltas to be from the bigger file
 297  * to the smaller - deletes are potentially cheaper, but perhaps
 298  * more importantly, the bigger file is likely the more recent
 299  * one.
 300  */
 301 static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_depth)
 302 {
 303         struct object_entry *cur_entry = cur->entry;
 304         struct object_entry *old_entry = old->entry;
 305         unsigned long size, oldsize, delta_size, sizediff;
 306         long max_size;
 307         void *delta_buf;
 308
 309         /* Don't bother doing diffs between different types */
 310         if (cur_entry->type != old_entry->type)
 311                 return -1;
 312
 313         size = cur_entry->size;
 314         if (size < 50)
 315                 return -1;
 316         oldsize = old_entry->size;
 317         sizediff = oldsize > size ? oldsize - size : size - oldsize;
 318         if (sizediff > size / 8)
 319                 return -1;
 320         if (old_entry->depth >= max_depth)
 321                 return 0;
 322
 323         /*
 324          * NOTE!
 325          *
 326          * We always delta from the bigger to the smaller, since that's
 327          * more space-efficient (deletes don't have to say _what_ they
 328          * delete).
 329          */
 330         max_size = size / 2 - 20;
 331         if (cur_entry->delta)
 332                 max_size = cur_entry->delta_size-1;
 333         if (sizediff >= max_size)
 334                 return -1;
 335         delta_buf = diff_delta(old->data, oldsize,
 336                                cur->data, size, &delta_size, max_size);
 337         if (!delta_buf)
 338                 return 0;
 339         cur_entry->delta = old_entry;
 340         cur_entry->delta_size = delta_size;
 341         cur_entry->depth = old_entry->depth + 1;
 342         free(delta_buf);
 343         return 0;
 344 }
 345
 346 static void find_deltas(struct object_entry **list, int window, int depth)
 347 {
 348         int i, idx;
 349         unsigned int array_size = window * sizeof(struct unpacked);
 350         struct unpacked *array = xmalloc(array_size);
 351
 352         memset(array, 0, array_size);
 353         i = nr_objects;
 354         idx = 0;
 355         while (--i >= 0) {
 356                 struct object_entry *entry = list[i];
 357                 struct unpacked *n = array + idx;
 358                 unsigned long size;
 359                 char type[10];
 360                 int j;
 361
 362                 free(n->data);
 363                 n->entry = entry;
 364                 n->data = read_sha1_file(entry->sha1, type, &size);
 365                 if (size != entry->size)
 366                         die("object %s inconsistent object length (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
 367                 j = window;
 368                 while (--j > 0) {
 369                         unsigned int other_idx = idx + j;
 370                         struct unpacked *m;
 371                         if (other_idx >= window)
 372                                 other_idx -= window;
 373                         m = array + other_idx;
 374                         if (!m->entry)
 375                                 break;
 376                         if (try_delta(n, m, depth) < 0)
 377                                 break;
 378                 }
 379                 idx++;
 380                 if (idx >= window)
 381                         idx = 0;
 382         }
 383 }
 384
 385 int main(int argc, char **argv)
 386 {
 387         char line[PATH_MAX + 20];
 388         int window = 10, depth = 10, pack_to_stdout = 0;
 389         int i;
 390
 391         for (i = 1; i < argc; i++) {
 392                 const char *arg = argv[i];
 393
 394                 if (*arg == '-') {
 395                         if (!strcmp("--non-empty", arg)) {
 396                                 non_empty = 1;
 397                                 continue;
 398                         }
 399                         if (!strcmp("--incremental", arg)) {
 400                                 incremental = 1;
 401                                 continue;
 402                         }
 403                         if (!strncmp("--window=", arg, 9)) {
 404                                 char *end;
 405                                 window = strtoul(arg+9, &end, 0);
 406                                 if (!arg[9] || *end)
 407                                         usage(pack_usage);
 408                                 continue;
 409                         }
 410                         if (!strncmp("--depth=", arg, 8)) {
 411                                 char *end;
 412                                 depth = strtoul(arg+8, &end, 0);
 413                                 if (!arg[8] || *end)
 414                                         usage(pack_usage);
 415                                 continue;
 416                         }
 417                         if (!strcmp("--stdout", arg)) {
 418                                 pack_to_stdout = 1;
 419                                 continue;
 420                         }
 421                         usage(pack_usage);
 422                 }
 423                 if (base_name)
 424                         usage(pack_usage);
 425                 base_name = arg;
 426         }
 427
 428         if (pack_to_stdout != !base_name)
 429                 usage(pack_usage);
 430
 431         while (fgets(line, sizeof(line), stdin) != NULL) {
 432                 unsigned int hash;
 433                 char *p;
 434                 unsigned char sha1[20];
 435
 436                 if (get_sha1_hex(line, sha1))
 437                         die("expected sha1, got garbage");
 438                 hash = 0;
 439                 p = line+40;
 440                 while (*p) {
 441                         unsigned char c = *p++;
 442                         if (isspace(c))
 443                                 continue;
 444                         hash = hash * 11 + c;
 445                 }
 446                 add_object_entry(sha1, hash);
 447         }
 448         if (non_empty && !nr_objects)
 449                 return 0;
 450         get_object_details();
 451
 452         fprintf(stderr, "Packing %d objects\n", nr_objects);
 453
 454         sorted_by_sha = create_sorted_list(sha1_sort);
 455         sorted_by_type = create_sorted_list(type_size_sort);
 456         if (window && depth)
 457                 find_deltas(sorted_by_type, window+1, depth);
 458
 459         write_pack_file();
 460         if (!pack_to_stdout)
 461                 write_index_file();
 462         return 0;
 463 }