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 #include <sys/time.h>
  12
  13 static int dry_run, quiet;
  14 static const char unpack_usage[] = "git-unpack-objects [-n] [-q] < pack-file";
  15
  16 /* We always read in 4kB chunks. */
  17 static unsigned char buffer[4096];
  18 static unsigned long offset, len, eof;
  19 static SHA_CTX ctx;
  20
  21 /*
  22  * Make sure at least "min" bytes are available in the buffer, and
  23  * return the pointer to the buffer.
  24  */
  25 static void * fill(int min)
  26 {
  27         if (min <= len)
  28                 return buffer + offset;
  29         if (eof)
  30                 die("unable to fill input");
  31         if (min > sizeof(buffer))
  32                 die("cannot fill %d bytes", min);
  33         if (offset) {
  34                 SHA1_Update(&ctx, buffer, offset);
  35                 memcpy(buffer, buffer + offset, len);
  36                 offset = 0;
  37         }
  38         do {
  39                 int ret = xread(0, buffer + len, sizeof(buffer) - len);
  40                 if (ret <= 0) {
  41                         if (!ret)
  42                                 die("early EOF");
  43                         die("read error on input: %s", strerror(errno));
  44                 }
  45                 len += ret;
  46         } while (len < min);
  47         return buffer;
  48 }
  49
  50 static void use(int bytes)
  51 {
  52         if (bytes > len)
  53                 die("used more bytes than were available");
  54         len -= bytes;
  55         offset += bytes;
  56 }
  57
  58 static void *get_data(unsigned long size)
  59 {
  60         z_stream stream;
  61         void *buf = xmalloc(size);
  62
  63         memset(&stream, 0, sizeof(stream));
  64
  65         stream.next_out = buf;
  66         stream.avail_out = size;
  67         stream.next_in = fill(1);
  68         stream.avail_in = len;
  69         inflateInit(&stream);
  70
  71         for (;;) {
  72                 int ret = inflate(&stream, 0);
  73                 use(len - stream.avail_in);
  74                 if (stream.total_out == size && ret == Z_STREAM_END)
  75                         break;
  76                 if (ret != Z_OK)
  77                         die("inflate returned %d\n", ret);
  78                 stream.next_in = fill(1);
  79                 stream.avail_in = len;
  80         }
  81         inflateEnd(&stream);
  82         return buf;
  83 }
  84
  85 struct delta_info {
  86         unsigned char base_sha1[20];
  87         unsigned long size;
  88         void *delta;
  89         struct delta_info *next;
  90 };
  91
  92 static struct delta_info *delta_list;
  93
  94 static void add_delta_to_list(unsigned char *base_sha1, void *delta, unsigned long size)
  95 {
  96         struct delta_info *info = xmalloc(sizeof(*info));
  97
  98         memcpy(info->base_sha1, base_sha1, 20);
  99         info->size = size;
 100         info->delta = delta;
 101         info->next = delta_list;
 102         delta_list = info;
 103 }
 104
 105 static void added_object(unsigned char *sha1, const char *type, void *data, unsigned long size);
 106
 107 static void write_object(void *buf, unsigned long size, const char *type)
 108 {
 109         unsigned char sha1[20];
 110         if (write_sha1_file(buf, size, type, sha1) < 0)
 111                 die("failed to write object");
 112         added_object(sha1, type, buf, size);
 113 }
 114
 115 static int resolve_delta(const char *type,
 116         void *base, unsigned long base_size,
 117         void *delta, unsigned long delta_size)
 118 {
 119         void *result;
 120         unsigned long result_size;
 121
 122         result = patch_delta(base, base_size,
 123                              delta, delta_size,
 124                              &result_size);
 125         if (!result)
 126                 die("failed to apply delta");
 127         free(delta);
 128         write_object(result, result_size, type);
 129         free(result);
 130         return 0;
 131 }
 132
 133 static void added_object(unsigned char *sha1, const char *type, void *data, unsigned long size)
 134 {
 135         struct delta_info **p = &delta_list;
 136         struct delta_info *info;
 137
 138         while ((info = *p) != NULL) {
 139                 if (!memcmp(info->base_sha1, sha1, 20)) {
 140                         *p = info->next;
 141                         p = &delta_list;
 142                         resolve_delta(type, data, size, info->delta, info->size);
 143                         free(info);
 144                         continue;
 145                 }
 146                 p = &info->next;
 147         }
 148 }
 149
 150 static int unpack_non_delta_entry(enum object_type kind, unsigned long size)
 151 {
 152         void *buf = get_data(size);
 153         const char *type;
 154
 155         switch (kind) {
 156         case OBJ_COMMIT: type = commit_type; break;
 157         case OBJ_TREE:   type = tree_type; break;
 158         case OBJ_BLOB:   type = blob_type; break;
 159         case OBJ_TAG:    type = tag_type; break;
 160         default: die("bad type %d", kind);
 161         }
 162         if (!dry_run)
 163                 write_object(buf, size, type);
 164         free(buf);
 165         return 0;
 166 }
 167
 168 static int unpack_delta_entry(unsigned long delta_size)
 169 {
 170         void *delta_data, *base;
 171         unsigned long base_size;
 172         char type[20];
 173         unsigned char base_sha1[20];
 174         int result;
 175
 176         memcpy(base_sha1, fill(20), 20);
 177         use(20);
 178
 179         delta_data = get_data(delta_size);
 180         if (dry_run) {
 181                 free(delta_data);
 182                 return 0;
 183         }
 184
 185         if (!has_sha1_file(base_sha1)) {
 186                 add_delta_to_list(base_sha1, delta_data, delta_size);
 187                 return 0;
 188         }
 189         base = read_sha1_file(base_sha1, type, &base_size);
 190         if (!base)
 191                 die("failed to read delta-pack base object %s", sha1_to_hex(base_sha1));
 192         result = resolve_delta(type, base, base_size, delta_data, delta_size);
 193         free(base);
 194         return result;
 195 }
 196
 197 static void unpack_one(unsigned nr, unsigned total)
 198 {
 199         unsigned shift;
 200         unsigned char *pack, c;
 201         unsigned long size;
 202         enum object_type type;
 203
 204         pack = fill(1);
 205         c = *pack;
 206         use(1);
 207         type = (c >> 4) & 7;
 208         size = (c & 15);
 209         shift = 4;
 210         while (c & 0x80) {
 211                 pack = fill(1);
 212                 c = *pack++;
 213                 use(1);
 214                 size += (c & 0x7f) << shift;
 215                 shift += 7;
 216         }
 217         if (!quiet) {
 218                 static unsigned long last_sec;
 219                 static unsigned last_percent;
 220                 struct timeval now;
 221                 unsigned percentage = (nr * 100) / total;
 222
 223                 gettimeofday(&now, NULL);
 224                 if (percentage != last_percent || now.tv_sec != last_sec) {
 225                         last_sec = now.tv_sec;
 226                         last_percent = percentage;
 227                         fprintf(stderr, "%4u%% (%u/%u) done\r", percentage, nr, total);
 228                 }
 229         }
 230         switch (type) {
 231         case OBJ_COMMIT:
 232         case OBJ_TREE:
 233         case OBJ_BLOB:
 234         case OBJ_TAG:
 235                 unpack_non_delta_entry(type, size);
 236                 return;
 237         case OBJ_DELTA:
 238                 unpack_delta_entry(size);
 239                 return;
 240         default:
 241                 die("bad object type %d", type);
 242         }
 243 }
 244
 245 static void unpack_all(void)
 246 {
 247         int i;
 248         struct pack_header *hdr = fill(sizeof(struct pack_header));
 249         unsigned nr_objects = ntohl(hdr->hdr_entries);
 250
 251         if (ntohl(hdr->hdr_signature) != PACK_SIGNATURE)
 252                 die("bad pack file");
 253         if (!pack_version_ok(hdr->hdr_version))
 254                 die("unknown pack file version %d", ntohl(hdr->hdr_version));
 255         fprintf(stderr, "Unpacking %d objects\n", nr_objects);
 256
 257         use(sizeof(struct pack_header));
 258         for (i = 0; i < nr_objects; i++)
 259                 unpack_one(i+1, nr_objects);
 260         if (delta_list)
 261                 die("unresolved deltas left after unpacking");
 262 }
 263
 264 int cmd_unpack_objects(int argc, const char **argv, const char *prefix)
 265 {
 266         int i;
 267         unsigned char sha1[20];
 268
 269         quiet = !isatty(2);
 270
 271         for (i = 1 ; i < argc; i++) {
 272                 const char *arg = argv[i];
 273
 274                 if (*arg == '-') {
 275                         if (!strcmp(arg, "-n")) {
 276                                 dry_run = 1;
 277                                 continue;
 278                         }
 279                         if (!strcmp(arg, "-q")) {
 280                                 quiet = 1;
 281                                 continue;
 282                         }
 283                         usage(unpack_usage);
 284                 }
 285
 286                 /* We don't take any non-flag arguments now.. Maybe some day */
 287                 usage(unpack_usage);
 288         }
 289         SHA1_Init(&ctx);
 290         unpack_all();
 291         SHA1_Update(&ctx, buffer, offset);
 292         SHA1_Final(sha1, &ctx);
 293         if (memcmp(fill(20), sha1, 20))
 294                 die("final sha1 did not match");
 295         use(20);
 296
 297         /* Write the last part of the buffer to stdout */
 298         while (len) {
 299                 int ret = xwrite(1, buffer + offset, len);
 300                 if (ret <= 0)
 301                         break;
 302                 len -= ret;
 303                 offset += ret;
 304         }
 305
 306         /* All done */
 307         if (!quiet)
 308                 fprintf(stderr, "\n");
 309         return 0;
 310 }