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