archive-zip.c

   1 /*
   2  * Copyright (c) 2006 Rene Scharfe
   3  */
   4 #include "cache.h"
   5 #include "archive.h"
   6 #include "streaming.h"
   7 #include "commit.h"
   8 #include "utf8.h"
   9
  10 static int zip_date;
  11 static int zip_time;
  12
  13 static unsigned char *zip_dir;
  14 static unsigned int zip_dir_size;
  15
  16 static unsigned int zip_offset;
  17 static unsigned int zip_dir_offset;
  18 static unsigned int zip_dir_entries;
  19
  20 #define ZIP_DIRECTORY_MIN_SIZE  (1024 * 1024)
  21 #define ZIP_STREAM      (1 <<  3)
  22 #define ZIP_UTF8        (1 << 11)
  23
  24 struct zip_local_header {
  25         unsigned char magic[4];
  26         unsigned char version[2];
  27         unsigned char flags[2];
  28         unsigned char compression_method[2];
  29         unsigned char mtime[2];
  30         unsigned char mdate[2];
  31         unsigned char crc32[4];
  32         unsigned char compressed_size[4];
  33         unsigned char size[4];
  34         unsigned char filename_length[2];
  35         unsigned char extra_length[2];
  36         unsigned char _end[1];
  37 };
  38
  39 struct zip_data_desc {
  40         unsigned char magic[4];
  41         unsigned char crc32[4];
  42         unsigned char compressed_size[4];
  43         unsigned char size[4];
  44         unsigned char _end[1];
  45 };
  46
  47 struct zip_dir_header {
  48         unsigned char magic[4];
  49         unsigned char creator_version[2];
  50         unsigned char version[2];
  51         unsigned char flags[2];
  52         unsigned char compression_method[2];
  53         unsigned char mtime[2];
  54         unsigned char mdate[2];
  55         unsigned char crc32[4];
  56         unsigned char compressed_size[4];
  57         unsigned char size[4];
  58         unsigned char filename_length[2];
  59         unsigned char extra_length[2];
  60         unsigned char comment_length[2];
  61         unsigned char disk[2];
  62         unsigned char attr1[2];
  63         unsigned char attr2[4];
  64         unsigned char offset[4];
  65         unsigned char _end[1];
  66 };
  67
  68 struct zip_dir_trailer {
  69         unsigned char magic[4];
  70         unsigned char disk[2];
  71         unsigned char directory_start_disk[2];
  72         unsigned char entries_on_this_disk[2];
  73         unsigned char entries[2];
  74         unsigned char size[4];
  75         unsigned char offset[4];
  76         unsigned char comment_length[2];
  77         unsigned char _end[1];
  78 };
  79
  80 /*
  81  * On ARM, padding is added at the end of the struct, so a simple
  82  * sizeof(struct ...) reports two bytes more than the payload size
  83  * we're interested in.
  84  */
  85 #define ZIP_LOCAL_HEADER_SIZE   offsetof(struct zip_local_header, _end)
  86 #define ZIP_DATA_DESC_SIZE      offsetof(struct zip_data_desc, _end)
  87 #define ZIP_DIR_HEADER_SIZE     offsetof(struct zip_dir_header, _end)
  88 #define ZIP_DIR_TRAILER_SIZE    offsetof(struct zip_dir_trailer, _end)
  89
  90 static void copy_le16(unsigned char *dest, unsigned int n)
  91 {
  92         dest[0] = 0xff & n;
  93         dest[1] = 0xff & (n >> 010);
  94 }
  95
  96 static void copy_le32(unsigned char *dest, unsigned int n)
  97 {
  98         dest[0] = 0xff & n;
  99         dest[1] = 0xff & (n >> 010);
 100         dest[2] = 0xff & (n >> 020);
 101         dest[3] = 0xff & (n >> 030);
 102 }
 103
 104 static void *zlib_deflate(void *data, unsigned long size,
 105                 int compression_level, unsigned long *compressed_size)
 106 {
 107         git_zstream stream;
 108         unsigned long maxsize;
 109         void *buffer;
 110         int result;
 111
 112         memset(&stream, 0, sizeof(stream));
 113         git_deflate_init(&stream, compression_level);
 114         maxsize = git_deflate_bound(&stream, size);
 115         buffer = xmalloc(maxsize);
 116
 117         stream.next_in = data;
 118         stream.avail_in = size;
 119         stream.next_out = buffer;
 120         stream.avail_out = maxsize;
 121
 122         do {
 123                 result = git_deflate(&stream, Z_FINISH);
 124         } while (result == Z_OK);
 125
 126         if (result != Z_STREAM_END) {
 127                 free(buffer);
 128                 return NULL;
 129         }
 130
 131         git_deflate_end(&stream);
 132         *compressed_size = stream.total_out;
 133
 134         return buffer;
 135 }
 136
 137 static void write_zip_data_desc(unsigned long size,
 138                                 unsigned long compressed_size,
 139                                 unsigned long crc)
 140 {
 141         struct zip_data_desc trailer;
 142
 143         copy_le32(trailer.magic, 0x08074b50);
 144         copy_le32(trailer.crc32, crc);
 145         copy_le32(trailer.compressed_size, compressed_size);
 146         copy_le32(trailer.size, size);
 147         write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
 148 }
 149
 150 static void set_zip_dir_data_desc(struct zip_dir_header *header,
 151                                   unsigned long size,
 152                                   unsigned long compressed_size,
 153                                   unsigned long crc)
 154 {
 155         copy_le32(header->crc32, crc);
 156         copy_le32(header->compressed_size, compressed_size);
 157         copy_le32(header->size, size);
 158 }
 159
 160 static void set_zip_header_data_desc(struct zip_local_header *header,
 161                                      unsigned long size,
 162                                      unsigned long compressed_size,
 163                                      unsigned long crc)
 164 {
 165         copy_le32(header->crc32, crc);
 166         copy_le32(header->compressed_size, compressed_size);
 167         copy_le32(header->size, size);
 168 }
 169
 170 #define STREAM_BUFFER_SIZE (1024 * 16)
 171
 172 static int write_zip_entry(struct archiver_args *args,
 173                            const unsigned char *sha1,
 174                            const char *path, size_t pathlen,
 175                            unsigned int mode)
 176 {
 177         struct zip_local_header header;
 178         struct zip_dir_header dirent;
 179         unsigned int creator_version = 0;
 180         unsigned long attr2 = 0;
 181         unsigned long compressed_size;
 182         unsigned long crc;
 183         unsigned long direntsize;
 184         int method;
 185         unsigned char *out;
 186         void *deflated = NULL;
 187         void *buffer;
 188         struct git_istream *stream = NULL;
 189         unsigned long flags = 0;
 190         unsigned long size;
 191
 192         crc = crc32(0, NULL, 0);
 193
 194         if (has_non_ascii(path)) {
 195                 if (is_utf8(path))
 196                         flags |= ZIP_UTF8;
 197                 else
 198                         warning("Path is not valid UTF-8: %s", path);
 199         }
 200
 201         if (pathlen > 0xffff) {
 202                 return error("path too long (%d chars, SHA1: %s): %s",
 203                                 (int)pathlen, sha1_to_hex(sha1), path);
 204         }
 205
 206         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 207                 method = 0;
 208                 attr2 = 16;
 209                 out = NULL;
 210                 size = 0;
 211                 compressed_size = 0;
 212                 buffer = NULL;
 213                 size = 0;
 214         } else if (S_ISREG(mode) || S_ISLNK(mode)) {
 215                 enum object_type type = sha1_object_info(sha1, &size);
 216
 217                 method = 0;
 218                 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
 219                         method = 8;
 220                 if (S_ISLNK(mode) || (mode & 0111) || (flags & ZIP_UTF8)) {
 221                         creator_version = 0x033f;
 222                         attr2 = mode;
 223                         if (S_ISLNK(mode))
 224                                 attr2 |= 0777;
 225                         attr2 <<= 16;
 226                 }
 227                 compressed_size = size;
 228
 229                 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
 230                     size > big_file_threshold) {
 231                         stream = open_istream(sha1, &type, &size, NULL);
 232                         if (!stream)
 233                                 return error("cannot stream blob %s",
 234                                              sha1_to_hex(sha1));
 235                         flags |= ZIP_STREAM;
 236                         out = buffer = NULL;
 237                 } else {
 238                         buffer = sha1_file_to_archive(args, path, sha1, mode,
 239                                                       &type, &size);
 240                         if (!buffer)
 241                                 return error("cannot read %s",
 242                                              sha1_to_hex(sha1));
 243                         crc = crc32(crc, buffer, size);
 244                         out = buffer;
 245                 }
 246         } else {
 247                 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
 248                                 sha1_to_hex(sha1));
 249         }
 250
 251         if (buffer && method == 8) {
 252                 deflated = zlib_deflate(buffer, size, args->compression_level,
 253                                 &compressed_size);
 254                 if (deflated && compressed_size - 6 < size) {
 255                         /* ZLIB --> raw compressed data (see RFC 1950) */
 256                         /* CMF and FLG ... */
 257                         out = (unsigned char *)deflated + 2;
 258                         compressed_size -= 6;   /* ... and ADLER32 */
 259                 } else {
 260                         method = 0;
 261                         compressed_size = size;
 262                 }
 263         }
 264
 265         /* make sure we have enough free space in the dictionary */
 266         direntsize = ZIP_DIR_HEADER_SIZE + pathlen;
 267         while (zip_dir_size < zip_dir_offset + direntsize) {
 268                 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
 269                 zip_dir = xrealloc(zip_dir, zip_dir_size);
 270         }
 271
 272         copy_le32(dirent.magic, 0x02014b50);
 273         copy_le16(dirent.creator_version, creator_version);
 274         copy_le16(dirent.version, 10);
 275         copy_le16(dirent.flags, flags);
 276         copy_le16(dirent.compression_method, method);
 277         copy_le16(dirent.mtime, zip_time);
 278         copy_le16(dirent.mdate, zip_date);
 279         set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 280         copy_le16(dirent.filename_length, pathlen);
 281         copy_le16(dirent.extra_length, 0);
 282         copy_le16(dirent.comment_length, 0);
 283         copy_le16(dirent.disk, 0);
 284         copy_le16(dirent.attr1, 0);
 285         copy_le32(dirent.attr2, attr2);
 286         copy_le32(dirent.offset, zip_offset);
 287
 288         copy_le32(header.magic, 0x04034b50);
 289         copy_le16(header.version, 10);
 290         copy_le16(header.flags, flags);
 291         copy_le16(header.compression_method, method);
 292         copy_le16(header.mtime, zip_time);
 293         copy_le16(header.mdate, zip_date);
 294         if (flags & ZIP_STREAM)
 295                 set_zip_header_data_desc(&header, 0, 0, 0);
 296         else
 297                 set_zip_header_data_desc(&header, size, compressed_size, crc);
 298         copy_le16(header.filename_length, pathlen);
 299         copy_le16(header.extra_length, 0);
 300         write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
 301         zip_offset += ZIP_LOCAL_HEADER_SIZE;
 302         write_or_die(1, path, pathlen);
 303         zip_offset += pathlen;
 304         if (stream && method == 0) {
 305                 unsigned char buf[STREAM_BUFFER_SIZE];
 306                 ssize_t readlen;
 307
 308                 for (;;) {
 309                         readlen = read_istream(stream, buf, sizeof(buf));
 310                         if (readlen <= 0)
 311                                 break;
 312                         crc = crc32(crc, buf, readlen);
 313                         write_or_die(1, buf, readlen);
 314                 }
 315                 close_istream(stream);
 316                 if (readlen)
 317                         return readlen;
 318
 319                 compressed_size = size;
 320                 zip_offset += compressed_size;
 321
 322                 write_zip_data_desc(size, compressed_size, crc);
 323                 zip_offset += ZIP_DATA_DESC_SIZE;
 324
 325                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 326         } else if (stream && method == 8) {
 327                 unsigned char buf[STREAM_BUFFER_SIZE];
 328                 ssize_t readlen;
 329                 git_zstream zstream;
 330                 int result;
 331                 size_t out_len;
 332                 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
 333
 334                 memset(&zstream, 0, sizeof(zstream));
 335                 git_deflate_init(&zstream, args->compression_level);
 336
 337                 compressed_size = 0;
 338                 zstream.next_out = compressed;
 339                 zstream.avail_out = sizeof(compressed);
 340
 341                 for (;;) {
 342                         readlen = read_istream(stream, buf, sizeof(buf));
 343                         if (readlen <= 0)
 344                                 break;
 345                         crc = crc32(crc, buf, readlen);
 346
 347                         zstream.next_in = buf;
 348                         zstream.avail_in = readlen;
 349                         result = git_deflate(&zstream, 0);
 350                         if (result != Z_OK)
 351                                 die("deflate error (%d)", result);
 352                         out = compressed;
 353                         if (!compressed_size)
 354                                 out += 2;
 355                         out_len = zstream.next_out - out;
 356
 357                         if (out_len > 0) {
 358                                 write_or_die(1, out, out_len);
 359                                 compressed_size += out_len;
 360                                 zstream.next_out = compressed;
 361                                 zstream.avail_out = sizeof(compressed);
 362                         }
 363
 364                 }
 365                 close_istream(stream);
 366                 if (readlen)
 367                         return readlen;
 368
 369                 zstream.next_in = buf;
 370                 zstream.avail_in = 0;
 371                 result = git_deflate(&zstream, Z_FINISH);
 372                 if (result != Z_STREAM_END)
 373                         die("deflate error (%d)", result);
 374
 375                 git_deflate_end(&zstream);
 376                 out = compressed;
 377                 if (!compressed_size)
 378                         out += 2;
 379                 out_len = zstream.next_out - out - 4;
 380                 write_or_die(1, out, out_len);
 381                 compressed_size += out_len;
 382                 zip_offset += compressed_size;
 383
 384                 write_zip_data_desc(size, compressed_size, crc);
 385                 zip_offset += ZIP_DATA_DESC_SIZE;
 386
 387                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 388         } else if (compressed_size > 0) {
 389                 write_or_die(1, out, compressed_size);
 390                 zip_offset += compressed_size;
 391         }
 392
 393         free(deflated);
 394         free(buffer);
 395
 396         memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
 397         zip_dir_offset += ZIP_DIR_HEADER_SIZE;
 398         memcpy(zip_dir + zip_dir_offset, path, pathlen);
 399         zip_dir_offset += pathlen;
 400         zip_dir_entries++;
 401
 402         return 0;
 403 }
 404
 405 static void write_zip_trailer(const unsigned char *sha1)
 406 {
 407         struct zip_dir_trailer trailer;
 408
 409         copy_le32(trailer.magic, 0x06054b50);
 410         copy_le16(trailer.disk, 0);
 411         copy_le16(trailer.directory_start_disk, 0);
 412         copy_le16(trailer.entries_on_this_disk, zip_dir_entries);
 413         copy_le16(trailer.entries, zip_dir_entries);
 414         copy_le32(trailer.size, zip_dir_offset);
 415         copy_le32(trailer.offset, zip_offset);
 416         copy_le16(trailer.comment_length, sha1 ? 40 : 0);
 417
 418         write_or_die(1, zip_dir, zip_dir_offset);
 419         write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
 420         if (sha1)
 421                 write_or_die(1, sha1_to_hex(sha1), 40);
 422 }
 423
 424 static void dos_time(time_t *time, int *dos_date, int *dos_time)
 425 {
 426         struct tm *t = localtime(time);
 427
 428         *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
 429                     (t->tm_year + 1900 - 1980) * 512;
 430         *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
 431 }
 432
 433 static int write_zip_archive(const struct archiver *ar,
 434                              struct archiver_args *args)
 435 {
 436         int err;
 437
 438         dos_time(&args->time, &zip_date, &zip_time);
 439
 440         zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
 441         zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
 442
 443         err = write_archive_entries(args, write_zip_entry);
 444         if (!err)
 445                 write_zip_trailer(args->commit_sha1);
 446
 447         free(zip_dir);
 448
 449         return err;
 450 }
 451
 452 static struct archiver zip_archiver = {
 453         "zip",
 454         write_zip_archive,
 455         ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
 456 };
 457
 458 void init_zip_archiver(void)
 459 {
 460         register_archiver(&zip_archiver);
 461 }