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;
  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 (min > sizeof(buffer))
  30                 die("cannot fill %d bytes", min);
  31         if (offset) {
  32                 SHA1_Update(&ctx, buffer, offset);
  33                 memcpy(buffer, buffer + offset, len);
  34                 offset = 0;
  35         }
  36         do {
  37                 int ret = xread(0, buffer + len, sizeof(buffer) - len);
  38                 if (ret <= 0) {
  39                         if (!ret)
  40                                 die("early EOF");
  41                         die("read error on input: %s", strerror(errno));
  42                 }
  43                 len += ret;
  44         } while (len < min);
  45         return buffer;
  46 }
  47
  48 static void use(int bytes)
  49 {
  50         if (bytes > len)
  51                 die("used more bytes than were available");
  52         len -= bytes;
  53         offset += bytes;
  54 }
  55
  56 static void *get_data(unsigned long size)
  57 {
  58         z_stream stream;
  59         void *buf = xmalloc(size);
  60
  61         memset(&stream, 0, sizeof(stream));
  62
  63         stream.next_out = buf;
  64         stream.avail_out = size;
  65         stream.next_in = fill(1);
  66         stream.avail_in = len;
  67         inflateInit(&stream);
  68
  69         for (;;) {
  70                 int ret = inflate(&stream, 0);
  71                 use(len - stream.avail_in);
  72                 if (stream.total_out == size && ret == Z_STREAM_END)
  73                         break;
  74                 if (ret != Z_OK)
  75                         die("inflate returned %d\n", ret);
  76                 stream.next_in = fill(1);
  77                 stream.avail_in = len;
  78         }
  79         inflateEnd(&stream);
  80         return buf;
  81 }
  82
  83 struct delta_info {
  84         unsigned char base_sha1[20];
  85         unsigned long size;
  86         void *delta;
  87         struct delta_info *next;
  88 };
  89
  90 static struct delta_info *delta_list;
  91
  92 static void add_delta_to_list(unsigned char *base_sha1, void *delta, unsigned long size)
  93 {
  94         struct delta_info *info = xmalloc(sizeof(*info));
  95
  96         hashcpy(info->base_sha1, base_sha1);
  97         info->size = size;
  98         info->delta = delta;
  99         info->next = delta_list;
 100         delta_list = info;
 101 }
 102
 103 static void added_object(unsigned char *sha1, const char *type, void *data, unsigned long size);
 104
 105 static void write_object(void *buf, unsigned long size, const char *type)
 106 {
 107         unsigned char sha1[20];
 108         if (write_sha1_file(buf, size, type, sha1) < 0)
 109                 die("failed to write object");
 110         added_object(sha1, type, buf, size);
 111 }
 112
 113 static int resolve_delta(const char *type,
 114         void *base, unsigned long base_size,
 115         void *delta, unsigned long delta_size)
 116 {
 117         void *result;
 118         unsigned long result_size;
 119
 120         result = patch_delta(base, base_size,
 121                              delta, delta_size,
 122                              &result_size);
 123         if (!result)
 124                 die("failed to apply delta");
 125         free(delta);
 126         write_object(result, result_size, type);
 127         free(result);
 128         return 0;
 129 }
 130
 131 static void added_object(unsigned char *sha1, const char *type, void *data, unsigned long size)
 132 {
 133         struct delta_info **p = &delta_list;
 134         struct delta_info *info;
 135
 136         while ((info = *p) != NULL) {
 137                 if (!hashcmp(info->base_sha1, sha1)) {
 138                         *p = info->next;
 139                         p = &delta_list;
 140                         resolve_delta(type, data, size, info->delta, info->size);
 141                         free(info);
 142                         continue;
 143                 }
 144                 p = &info->next;
 145         }
 146 }
 147
 148 static int unpack_non_delta_entry(enum object_type kind, unsigned long size)
 149 {
 150         void *buf = get_data(size);
 151         const char *type;
 152
 153         switch (kind) {
 154         case OBJ_COMMIT: type = commit_type; break;
 155         case OBJ_TREE:   type = tree_type; break;
 156         case OBJ_BLOB:   type = blob_type; break;
 157         case OBJ_TAG:    type = tag_type; break;
 158         default: die("bad type %d", kind);
 159         }
 160         if (!dry_run)
 161                 write_object(buf, size, type);
 162         free(buf);
 163         return 0;
 164 }
 165
 166 static int unpack_delta_entry(unsigned long delta_size)
 167 {
 168         void *delta_data, *base;
 169         unsigned long base_size;
 170         char type[20];
 171         unsigned char base_sha1[20];
 172         int result;
 173
 174         hashcpy(base_sha1, fill(20));
 175         use(20);
 176
 177         delta_data = get_data(delta_size);
 178         if (dry_run) {
 179                 free(delta_data);
 180                 return 0;
 181         }
 182
 183         if (!has_sha1_file(base_sha1)) {
 184                 add_delta_to_list(base_sha1, delta_data, delta_size);
 185                 return 0;
 186         }
 187         base = read_sha1_file(base_sha1, type, &base_size);
 188         if (!base)
 189                 die("failed to read delta-pack base object %s", sha1_to_hex(base_sha1));
 190         result = resolve_delta(type, base, base_size, delta_data, delta_size);
 191         free(base);
 192         return result;
 193 }
 194
 195 static void unpack_one(unsigned nr, unsigned total)
 196 {
 197         unsigned shift;
 198         unsigned char *pack, c;
 199         unsigned long size;
 200         enum object_type type;
 201
 202         pack = fill(1);
 203         c = *pack;
 204         use(1);
 205         type = (c >> 4) & 7;
 206         size = (c & 15);
 207         shift = 4;
 208         while (c & 0x80) {
 209                 pack = fill(1);
 210                 c = *pack++;
 211                 use(1);
 212                 size += (c & 0x7f) << shift;
 213                 shift += 7;
 214         }
 215         if (!quiet) {
 216                 static unsigned long last_sec;
 217                 static unsigned last_percent;
 218                 struct timeval now;
 219                 unsigned percentage = (nr * 100) / total;
 220
 221                 gettimeofday(&now, NULL);
 222                 if (percentage != last_percent || now.tv_sec != last_sec) {
 223                         last_sec = now.tv_sec;
 224                         last_percent = percentage;
 225                         fprintf(stderr, "%4u%% (%u/%u) done\r", percentage, nr, total);
 226                 }
 227         }
 228         switch (type) {
 229         case OBJ_COMMIT:
 230         case OBJ_TREE:
 231         case OBJ_BLOB:
 232         case OBJ_TAG:
 233                 unpack_non_delta_entry(type, size);
 234                 return;
 235         case OBJ_DELTA:
 236                 unpack_delta_entry(size);
 237                 return;
 238         default:
 239                 die("bad object type %d", type);
 240         }
 241 }
 242
 243 static void unpack_all(void)
 244 {
 245         int i;
 246         struct pack_header *hdr = fill(sizeof(struct pack_header));
 247         unsigned nr_objects = ntohl(hdr->hdr_entries);
 248
 249         if (ntohl(hdr->hdr_signature) != PACK_SIGNATURE)
 250                 die("bad pack file");
 251         if (!pack_version_ok(hdr->hdr_version))
 252                 die("unknown pack file version %d", ntohl(hdr->hdr_version));
 253         fprintf(stderr, "Unpacking %d objects\n", nr_objects);
 254
 255         use(sizeof(struct pack_header));
 256         for (i = 0; i < nr_objects; i++)
 257                 unpack_one(i+1, nr_objects);
 258         if (delta_list)
 259                 die("unresolved deltas left after unpacking");
 260 }
 261
 262 int cmd_unpack_objects(int argc, const char **argv, const char *prefix)
 263 {
 264         int i;
 265         unsigned char sha1[20];
 266
 267         git_config(git_default_config);
 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 (hashcmp(fill(20), sha1))
 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 }