builtin-unpack-objects.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "object.h"
   4 #include "delta.h"
   5 #include "pack.h"
   6 #include "blob.h"
   7 #include "commit.h"
   8 #include "tag.h"
   9 #include "tree.h"
  10
  11 static int dry_run, quiet, recover, has_errors;
  12 static const char unpack_usage[] = "git-unpack-objects [-n] [-q] [-r] < pack-file";
  13
  14 /* We always read in 4kB chunks. */
  15 static unsigned char buffer[4096];
  16 static unsigned long offset, len, consumed_bytes;
  17 static SHA_CTX ctx;
  18
  19 /*
  20  * Make sure at least "min" bytes are available in the buffer, and
  21  * return the pointer to the buffer.
  22  */
  23 static void *fill(int min)
  24 {
  25         if (min <= len)
  26                 return buffer + offset;
  27         if (min > sizeof(buffer))
  28                 die("cannot fill %d bytes", min);
  29         if (offset) {
  30                 SHA1_Update(&ctx, buffer, offset);
  31                 memmove(buffer, buffer + offset, len);
  32                 offset = 0;
  33         }
  34         do {
  35                 int ret = xread(0, buffer + len, sizeof(buffer) - len);
  36                 if (ret <= 0) {
  37                         if (!ret)
  38                                 die("early EOF");
  39                         die("read error on input: %s", strerror(errno));
  40                 }
  41                 len += ret;
  42         } while (len < min);
  43         return buffer;
  44 }
  45
  46 static void use(int bytes)
  47 {
  48         if (bytes > len)
  49                 die("used more bytes than were available");
  50         len -= bytes;
  51         offset += bytes;
  52         consumed_bytes += bytes;
  53 }
  54
  55 static void *get_data(unsigned long size)
  56 {
  57         z_stream stream;
  58         void *buf = xmalloc(size);
  59
  60         memset(&stream, 0, sizeof(stream));
  61
  62         stream.next_out = buf;
  63         stream.avail_out = size;
  64         stream.next_in = fill(1);
  65         stream.avail_in = len;
  66         inflateInit(&stream);
  67
  68         for (;;) {
  69                 int ret = inflate(&stream, 0);
  70                 use(len - stream.avail_in);
  71                 if (stream.total_out == size && ret == Z_STREAM_END)
  72                         break;
  73                 if (ret != Z_OK) {
  74                         error("inflate returned %d\n", ret);
  75                         free(buf);
  76                         buf = NULL;
  77                         if (!recover)
  78                                 exit(1);
  79                         has_errors = 1;
  80                         break;
  81                 }
  82                 stream.next_in = fill(1);
  83                 stream.avail_in = len;
  84         }
  85         inflateEnd(&stream);
  86         return buf;
  87 }
  88
  89 struct delta_info {
  90         unsigned char base_sha1[20];
  91         unsigned long base_offset;
  92         unsigned long size;
  93         void *delta;
  94         unsigned nr;
  95         struct delta_info *next;
  96 };
  97
  98 static struct delta_info *delta_list;
  99
 100 static void add_delta_to_list(unsigned nr, unsigned const char *base_sha1,
 101                               unsigned long base_offset,
 102                               void *delta, unsigned long size)
 103 {
 104         struct delta_info *info = xmalloc(sizeof(*info));
 105
 106         hashcpy(info->base_sha1, base_sha1);
 107         info->base_offset = base_offset;
 108         info->size = size;
 109         info->delta = delta;
 110         info->nr = nr;
 111         info->next = delta_list;
 112         delta_list = info;
 113 }
 114
 115 struct obj_info {
 116         unsigned long offset;
 117         unsigned char sha1[20];
 118 };
 119
 120 static struct obj_info *obj_list;
 121
 122 static void added_object(unsigned nr, const char *type, void *data,
 123                          unsigned long size);
 124
 125 static void write_object(unsigned nr, void *buf, unsigned long size,
 126                          const char *type)
 127 {
 128         if (write_sha1_file(buf, size, type, obj_list[nr].sha1) < 0)
 129                 die("failed to write object");
 130         added_object(nr, type, buf, size);
 131 }
 132
 133 static void resolve_delta(unsigned nr, const char *type,
 134                           void *base, unsigned long base_size,
 135                           void *delta, unsigned long delta_size)
 136 {
 137         void *result;
 138         unsigned long result_size;
 139
 140         result = patch_delta(base, base_size,
 141                              delta, delta_size,
 142                              &result_size);
 143         if (!result)
 144                 die("failed to apply delta");
 145         free(delta);
 146         write_object(nr, result, result_size, type);
 147         free(result);
 148 }
 149
 150 static void added_object(unsigned nr, const char *type, void *data,
 151                          unsigned long size)
 152 {
 153         struct delta_info **p = &delta_list;
 154         struct delta_info *info;
 155
 156         while ((info = *p) != NULL) {
 157                 if (!hashcmp(info->base_sha1, obj_list[nr].sha1) ||
 158                     info->base_offset == obj_list[nr].offset) {
 159                         *p = info->next;
 160                         p = &delta_list;
 161                         resolve_delta(info->nr, type, data, size,
 162                                       info->delta, info->size);
 163                         free(info);
 164                         continue;
 165                 }
 166                 p = &info->next;
 167         }
 168 }
 169
 170 static void unpack_non_delta_entry(enum object_type kind, unsigned long size,
 171                                    unsigned nr)
 172 {
 173         void *buf = get_data(size);
 174         const char *type;
 175
 176         switch (kind) {
 177         case OBJ_COMMIT: type = commit_type; break;
 178         case OBJ_TREE:   type = tree_type; break;
 179         case OBJ_BLOB:   type = blob_type; break;
 180         case OBJ_TAG:    type = tag_type; break;
 181         default: die("bad type %d", kind);
 182         }
 183         if (!dry_run && buf)
 184                 write_object(nr, buf, size, type);
 185         free(buf);
 186 }
 187
 188 static void unpack_delta_entry(enum object_type kind, unsigned long delta_size,
 189                                unsigned nr)
 190 {
 191         void *delta_data, *base;
 192         unsigned long base_size;
 193         char type[20];
 194         unsigned char base_sha1[20];
 195
 196         if (kind == OBJ_REF_DELTA) {
 197                 hashcpy(base_sha1, fill(20));
 198                 use(20);
 199                 delta_data = get_data(delta_size);
 200                 if (dry_run || !delta_data) {
 201                         free(delta_data);
 202                         return;
 203                 }
 204                 if (!has_sha1_file(base_sha1)) {
 205                         hashcpy(obj_list[nr].sha1, null_sha1);
 206                         add_delta_to_list(nr, base_sha1, 0, delta_data, delta_size);
 207                         return;
 208                 }
 209         } else {
 210                 unsigned base_found = 0;
 211                 unsigned char *pack, c;
 212                 unsigned long base_offset;
 213                 unsigned lo, mid, hi;
 214
 215                 pack = fill(1);
 216                 c = *pack;
 217                 use(1);
 218                 base_offset = c & 127;
 219                 while (c & 128) {
 220                         base_offset += 1;
 221                         if (!base_offset || base_offset & ~(~0UL >> 7))
 222                                 die("offset value overflow for delta base object");
 223                         pack = fill(1);
 224                         c = *pack;
 225                         use(1);
 226                         base_offset = (base_offset << 7) + (c & 127);
 227                 }
 228                 base_offset = obj_list[nr].offset - base_offset;
 229
 230                 delta_data = get_data(delta_size);
 231                 if (dry_run || !delta_data) {
 232                         free(delta_data);
 233                         return;
 234                 }
 235                 lo = 0;
 236                 hi = nr;
 237                 while (lo < hi) {
 238                         mid = (lo + hi)/2;
 239                         if (base_offset < obj_list[mid].offset) {
 240                                 hi = mid;
 241                         } else if (base_offset > obj_list[mid].offset) {
 242                                 lo = mid + 1;
 243                         } else {
 244                                 hashcpy(base_sha1, obj_list[mid].sha1);
 245                                 base_found = !is_null_sha1(base_sha1);
 246                                 break;
 247                         }
 248                 }
 249                 if (!base_found) {
 250                         /* The delta base object is itself a delta that
 251                            has not been resolved yet. */
 252                         hashcpy(obj_list[nr].sha1, null_sha1);
 253                         add_delta_to_list(nr, null_sha1, base_offset, delta_data, delta_size);
 254                         return;
 255                 }
 256         }
 257
 258         base = read_sha1_file(base_sha1, type, &base_size);
 259         if (!base) {
 260                 error("failed to read delta-pack base object %s",
 261                       sha1_to_hex(base_sha1));
 262                 if (!recover)
 263                         exit(1);
 264                 has_errors = 1;
 265                 return;
 266         }
 267         resolve_delta(nr, type, base, base_size, delta_data, delta_size);
 268         free(base);
 269 }
 270
 271 static void unpack_one(unsigned nr, unsigned total)
 272 {
 273         unsigned shift;
 274         unsigned char *pack, c;
 275         unsigned long size;
 276         enum object_type type;
 277
 278         obj_list[nr].offset = consumed_bytes;
 279
 280         pack = fill(1);
 281         c = *pack;
 282         use(1);
 283         type = (c >> 4) & 7;
 284         size = (c & 15);
 285         shift = 4;
 286         while (c & 0x80) {
 287                 pack = fill(1);
 288                 c = *pack;
 289                 use(1);
 290                 size += (c & 0x7f) << shift;
 291                 shift += 7;
 292         }
 293         if (!quiet) {
 294                 static unsigned long last_sec;
 295                 static unsigned last_percent;
 296                 struct timeval now;
 297                 unsigned percentage = ((nr+1) * 100) / total;
 298
 299                 gettimeofday(&now, NULL);
 300                 if (percentage != last_percent || now.tv_sec != last_sec) {
 301                         last_sec = now.tv_sec;
 302                         last_percent = percentage;
 303                         fprintf(stderr, "%4u%% (%u/%u) done\r",
 304                                         percentage, (nr+1), total);
 305                 }
 306         }
 307         switch (type) {
 308         case OBJ_COMMIT:
 309         case OBJ_TREE:
 310         case OBJ_BLOB:
 311         case OBJ_TAG:
 312                 unpack_non_delta_entry(type, size, nr);
 313                 return;
 314         case OBJ_REF_DELTA:
 315         case OBJ_OFS_DELTA:
 316                 unpack_delta_entry(type, size, nr);
 317                 return;
 318         default:
 319                 error("bad object type %d", type);
 320                 has_errors = 1;
 321                 if (recover)
 322                         return;
 323                 exit(1);
 324         }
 325 }
 326
 327 static void unpack_all(void)
 328 {
 329         int i;
 330         struct pack_header *hdr = fill(sizeof(struct pack_header));
 331         unsigned nr_objects = ntohl(hdr->hdr_entries);
 332
 333         if (ntohl(hdr->hdr_signature) != PACK_SIGNATURE)
 334                 die("bad pack file");
 335         if (!pack_version_ok(hdr->hdr_version))
 336                 die("unknown pack file version %d", ntohl(hdr->hdr_version));
 337         fprintf(stderr, "Unpacking %d objects\n", nr_objects);
 338
 339         obj_list = xmalloc(nr_objects * sizeof(*obj_list));
 340         use(sizeof(struct pack_header));
 341         for (i = 0; i < nr_objects; i++)
 342                 unpack_one(i, nr_objects);
 343         if (delta_list)
 344                 die("unresolved deltas left after unpacking");
 345 }
 346
 347 int cmd_unpack_objects(int argc, const char **argv, const char *prefix)
 348 {
 349         int i;
 350         unsigned char sha1[20];
 351
 352         git_config(git_default_config);
 353
 354         quiet = !isatty(2);
 355
 356         for (i = 1 ; i < argc; i++) {
 357                 const char *arg = argv[i];
 358
 359                 if (*arg == '-') {
 360                         if (!strcmp(arg, "-n")) {
 361                                 dry_run = 1;
 362                                 continue;
 363                         }
 364                         if (!strcmp(arg, "-q")) {
 365                                 quiet = 1;
 366                                 continue;
 367                         }
 368                         if (!strcmp(arg, "-r")) {
 369                                 recover = 1;
 370                                 continue;
 371                         }
 372                         if (!strncmp(arg, "--pack_header=", 14)) {
 373                                 struct pack_header *hdr;
 374                                 char *c;
 375
 376                                 hdr = (struct pack_header *)buffer;
 377                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 378                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 379                                 if (*c != ',')
 380                                         die("bad %s", arg);
 381                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 382                                 if (*c)
 383                                         die("bad %s", arg);
 384                                 len = sizeof(*hdr);
 385                                 continue;
 386                         }
 387                         usage(unpack_usage);
 388                 }
 389
 390                 /* We don't take any non-flag arguments now.. Maybe some day */
 391                 usage(unpack_usage);
 392         }
 393         SHA1_Init(&ctx);
 394         unpack_all();
 395         SHA1_Update(&ctx, buffer, offset);
 396         SHA1_Final(sha1, &ctx);
 397         if (hashcmp(fill(20), sha1))
 398                 die("final sha1 did not match");
 399         use(20);
 400
 401         /* Write the last part of the buffer to stdout */
 402         while (len) {
 403                 int ret = xwrite(1, buffer + offset, len);
 404                 if (ret <= 0)
 405                         break;
 406                 len -= ret;
 407                 offset += ret;
 408         }
 409
 410         /* All done */
 411         if (!quiet)
 412                 fprintf(stderr, "\n");
 413         return has_errors;
 414 }