archive-zip.c

   1 /*
   2  * Copyright (c) 2006 Rene Scharfe
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "archive.h"
   7 #include "streaming.h"
   8 #include "utf8.h"
   9 #include "userdiff.h"
  10 #include "xdiff-interface.h"
  11
  12 static int zip_date;
  13 static int zip_time;
  14
  15 static unsigned char *zip_dir;
  16 static unsigned int zip_dir_size;
  17
  18 static unsigned int zip_offset;
  19 static unsigned int zip_dir_offset;
  20 static uint64_t zip_dir_entries;
  21
  22 static unsigned int max_creator_version;
  23
  24 #define ZIP_DIRECTORY_MIN_SIZE  (1024 * 1024)
  25 #define ZIP_STREAM      (1 <<  3)
  26 #define ZIP_UTF8        (1 << 11)
  27
  28 struct zip_local_header {
  29         unsigned char magic[4];
  30         unsigned char version[2];
  31         unsigned char flags[2];
  32         unsigned char compression_method[2];
  33         unsigned char mtime[2];
  34         unsigned char mdate[2];
  35         unsigned char crc32[4];
  36         unsigned char compressed_size[4];
  37         unsigned char size[4];
  38         unsigned char filename_length[2];
  39         unsigned char extra_length[2];
  40         unsigned char _end[1];
  41 };
  42
  43 struct zip_data_desc {
  44         unsigned char magic[4];
  45         unsigned char crc32[4];
  46         unsigned char compressed_size[4];
  47         unsigned char size[4];
  48         unsigned char _end[1];
  49 };
  50
  51 struct zip_dir_header {
  52         unsigned char magic[4];
  53         unsigned char creator_version[2];
  54         unsigned char version[2];
  55         unsigned char flags[2];
  56         unsigned char compression_method[2];
  57         unsigned char mtime[2];
  58         unsigned char mdate[2];
  59         unsigned char crc32[4];
  60         unsigned char compressed_size[4];
  61         unsigned char size[4];
  62         unsigned char filename_length[2];
  63         unsigned char extra_length[2];
  64         unsigned char comment_length[2];
  65         unsigned char disk[2];
  66         unsigned char attr1[2];
  67         unsigned char attr2[4];
  68         unsigned char offset[4];
  69         unsigned char _end[1];
  70 };
  71
  72 struct zip_dir_trailer {
  73         unsigned char magic[4];
  74         unsigned char disk[2];
  75         unsigned char directory_start_disk[2];
  76         unsigned char entries_on_this_disk[2];
  77         unsigned char entries[2];
  78         unsigned char size[4];
  79         unsigned char offset[4];
  80         unsigned char comment_length[2];
  81         unsigned char _end[1];
  82 };
  83
  84 struct zip_extra_mtime {
  85         unsigned char magic[2];
  86         unsigned char extra_size[2];
  87         unsigned char flags[1];
  88         unsigned char mtime[4];
  89         unsigned char _end[1];
  90 };
  91
  92 struct zip64_dir_trailer {
  93         unsigned char magic[4];
  94         unsigned char record_size[8];
  95         unsigned char creator_version[2];
  96         unsigned char version[2];
  97         unsigned char disk[4];
  98         unsigned char directory_start_disk[4];
  99         unsigned char entries_on_this_disk[8];
 100         unsigned char entries[8];
 101         unsigned char size[8];
 102         unsigned char offset[8];
 103         unsigned char _end[1];
 104 };
 105
 106 struct zip64_dir_trailer_locator {
 107         unsigned char magic[4];
 108         unsigned char disk[4];
 109         unsigned char offset[8];
 110         unsigned char number_of_disks[4];
 111         unsigned char _end[1];
 112 };
 113
 114 /*
 115  * On ARM, padding is added at the end of the struct, so a simple
 116  * sizeof(struct ...) reports two bytes more than the payload size
 117  * we're interested in.
 118  */
 119 #define ZIP_LOCAL_HEADER_SIZE   offsetof(struct zip_local_header, _end)
 120 #define ZIP_DATA_DESC_SIZE      offsetof(struct zip_data_desc, _end)
 121 #define ZIP_DIR_HEADER_SIZE     offsetof(struct zip_dir_header, _end)
 122 #define ZIP_DIR_TRAILER_SIZE    offsetof(struct zip_dir_trailer, _end)
 123 #define ZIP_EXTRA_MTIME_SIZE    offsetof(struct zip_extra_mtime, _end)
 124 #define ZIP_EXTRA_MTIME_PAYLOAD_SIZE \
 125         (ZIP_EXTRA_MTIME_SIZE - offsetof(struct zip_extra_mtime, flags))
 126 #define ZIP64_DIR_TRAILER_SIZE  offsetof(struct zip64_dir_trailer, _end)
 127 #define ZIP64_DIR_TRAILER_RECORD_SIZE \
 128         (ZIP64_DIR_TRAILER_SIZE - \
 129          offsetof(struct zip64_dir_trailer, creator_version))
 130 #define ZIP64_DIR_TRAILER_LOCATOR_SIZE \
 131         offsetof(struct zip64_dir_trailer_locator, _end)
 132
 133 static void copy_le16(unsigned char *dest, unsigned int n)
 134 {
 135         dest[0] = 0xff & n;
 136         dest[1] = 0xff & (n >> 010);
 137 }
 138
 139 static void copy_le32(unsigned char *dest, unsigned int n)
 140 {
 141         dest[0] = 0xff & n;
 142         dest[1] = 0xff & (n >> 010);
 143         dest[2] = 0xff & (n >> 020);
 144         dest[3] = 0xff & (n >> 030);
 145 }
 146
 147 static void copy_le64(unsigned char *dest, uint64_t n)
 148 {
 149         dest[0] = 0xff & n;
 150         dest[1] = 0xff & (n >> 010);
 151         dest[2] = 0xff & (n >> 020);
 152         dest[3] = 0xff & (n >> 030);
 153         dest[4] = 0xff & (n >> 040);
 154         dest[5] = 0xff & (n >> 050);
 155         dest[6] = 0xff & (n >> 060);
 156         dest[7] = 0xff & (n >> 070);
 157 }
 158
 159 static uint64_t clamp_max(uint64_t n, uint64_t max, int *clamped)
 160 {
 161         if (n <= max)
 162                 return n;
 163         *clamped = 1;
 164         return max;
 165 }
 166
 167 static void copy_le16_clamp(unsigned char *dest, uint64_t n, int *clamped)
 168 {
 169         copy_le16(dest, clamp_max(n, 0xffff, clamped));
 170 }
 171
 172 static void *zlib_deflate_raw(void *data, unsigned long size,
 173                               int compression_level,
 174                               unsigned long *compressed_size)
 175 {
 176         git_zstream stream;
 177         unsigned long maxsize;
 178         void *buffer;
 179         int result;
 180
 181         git_deflate_init_raw(&stream, compression_level);
 182         maxsize = git_deflate_bound(&stream, size);
 183         buffer = xmalloc(maxsize);
 184
 185         stream.next_in = data;
 186         stream.avail_in = size;
 187         stream.next_out = buffer;
 188         stream.avail_out = maxsize;
 189
 190         do {
 191                 result = git_deflate(&stream, Z_FINISH);
 192         } while (result == Z_OK);
 193
 194         if (result != Z_STREAM_END) {
 195                 free(buffer);
 196                 return NULL;
 197         }
 198
 199         git_deflate_end(&stream);
 200         *compressed_size = stream.total_out;
 201
 202         return buffer;
 203 }
 204
 205 static void write_zip_data_desc(unsigned long size,
 206                                 unsigned long compressed_size,
 207                                 unsigned long crc)
 208 {
 209         struct zip_data_desc trailer;
 210
 211         copy_le32(trailer.magic, 0x08074b50);
 212         copy_le32(trailer.crc32, crc);
 213         copy_le32(trailer.compressed_size, compressed_size);
 214         copy_le32(trailer.size, size);
 215         write_or_die(1, &trailer, ZIP_DATA_DESC_SIZE);
 216 }
 217
 218 static void set_zip_dir_data_desc(struct zip_dir_header *header,
 219                                   unsigned long size,
 220                                   unsigned long compressed_size,
 221                                   unsigned long crc)
 222 {
 223         copy_le32(header->crc32, crc);
 224         copy_le32(header->compressed_size, compressed_size);
 225         copy_le32(header->size, size);
 226 }
 227
 228 static void set_zip_header_data_desc(struct zip_local_header *header,
 229                                      unsigned long size,
 230                                      unsigned long compressed_size,
 231                                      unsigned long crc)
 232 {
 233         copy_le32(header->crc32, crc);
 234         copy_le32(header->compressed_size, compressed_size);
 235         copy_le32(header->size, size);
 236 }
 237
 238 static int has_only_ascii(const char *s)
 239 {
 240         for (;;) {
 241                 int c = *s++;
 242                 if (c == '\0')
 243                         return 1;
 244                 if (!isascii(c))
 245                         return 0;
 246         }
 247 }
 248
 249 static int entry_is_binary(const char *path, const void *buffer, size_t size)
 250 {
 251         struct userdiff_driver *driver = userdiff_find_by_path(path);
 252         if (!driver)
 253                 driver = userdiff_find_by_name("default");
 254         if (driver->binary != -1)
 255                 return driver->binary;
 256         return buffer_is_binary(buffer, size);
 257 }
 258
 259 #define STREAM_BUFFER_SIZE (1024 * 16)
 260
 261 static int write_zip_entry(struct archiver_args *args,
 262                            const unsigned char *sha1,
 263                            const char *path, size_t pathlen,
 264                            unsigned int mode)
 265 {
 266         struct zip_local_header header;
 267         struct zip_dir_header dirent;
 268         struct zip_extra_mtime extra;
 269         unsigned long attr2;
 270         unsigned long compressed_size;
 271         unsigned long crc;
 272         unsigned long direntsize;
 273         int method;
 274         unsigned char *out;
 275         void *deflated = NULL;
 276         void *buffer;
 277         struct git_istream *stream = NULL;
 278         unsigned long flags = 0;
 279         unsigned long size;
 280         int is_binary = -1;
 281         const char *path_without_prefix = path + args->baselen;
 282         unsigned int creator_version = 0;
 283
 284         crc = crc32(0, NULL, 0);
 285
 286         if (!has_only_ascii(path)) {
 287                 if (is_utf8(path))
 288                         flags |= ZIP_UTF8;
 289                 else
 290                         warning("Path is not valid UTF-8: %s", path);
 291         }
 292
 293         if (pathlen > 0xffff) {
 294                 return error("path too long (%d chars, SHA1: %s): %s",
 295                                 (int)pathlen, sha1_to_hex(sha1), path);
 296         }
 297
 298         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 299                 method = 0;
 300                 attr2 = 16;
 301                 out = NULL;
 302                 size = 0;
 303                 compressed_size = 0;
 304                 buffer = NULL;
 305         } else if (S_ISREG(mode) || S_ISLNK(mode)) {
 306                 enum object_type type = sha1_object_info(sha1, &size);
 307
 308                 method = 0;
 309                 attr2 = S_ISLNK(mode) ? ((mode | 0777) << 16) :
 310                         (mode & 0111) ? ((mode) << 16) : 0;
 311                 if (S_ISLNK(mode) || (mode & 0111))
 312                         creator_version = 0x0317;
 313                 if (S_ISREG(mode) && args->compression_level != 0 && size > 0)
 314                         method = 8;
 315
 316                 if (S_ISREG(mode) && type == OBJ_BLOB && !args->convert &&
 317                     size > big_file_threshold) {
 318                         stream = open_istream(sha1, &type, &size, NULL);
 319                         if (!stream)
 320                                 return error("cannot stream blob %s",
 321                                              sha1_to_hex(sha1));
 322                         flags |= ZIP_STREAM;
 323                         out = buffer = NULL;
 324                 } else {
 325                         buffer = sha1_file_to_archive(args, path, sha1, mode,
 326                                                       &type, &size);
 327                         if (!buffer)
 328                                 return error("cannot read %s",
 329                                              sha1_to_hex(sha1));
 330                         crc = crc32(crc, buffer, size);
 331                         is_binary = entry_is_binary(path_without_prefix,
 332                                                     buffer, size);
 333                         out = buffer;
 334                 }
 335                 compressed_size = (method == 0) ? size : 0;
 336         } else {
 337                 return error("unsupported file mode: 0%o (SHA1: %s)", mode,
 338                                 sha1_to_hex(sha1));
 339         }
 340
 341         if (creator_version > max_creator_version)
 342                 max_creator_version = creator_version;
 343
 344         if (buffer && method == 8) {
 345                 out = deflated = zlib_deflate_raw(buffer, size,
 346                                                   args->compression_level,
 347                                                   &compressed_size);
 348                 if (!out || compressed_size >= size) {
 349                         out = buffer;
 350                         method = 0;
 351                         compressed_size = size;
 352                 }
 353         }
 354
 355         copy_le16(extra.magic, 0x5455);
 356         copy_le16(extra.extra_size, ZIP_EXTRA_MTIME_PAYLOAD_SIZE);
 357         extra.flags[0] = 1;     /* just mtime */
 358         copy_le32(extra.mtime, args->time);
 359
 360         /* make sure we have enough free space in the dictionary */
 361         direntsize = ZIP_DIR_HEADER_SIZE + pathlen + ZIP_EXTRA_MTIME_SIZE;
 362         while (zip_dir_size < zip_dir_offset + direntsize) {
 363                 zip_dir_size += ZIP_DIRECTORY_MIN_SIZE;
 364                 zip_dir = xrealloc(zip_dir, zip_dir_size);
 365         }
 366
 367         copy_le32(dirent.magic, 0x02014b50);
 368         copy_le16(dirent.creator_version, creator_version);
 369         copy_le16(dirent.version, 10);
 370         copy_le16(dirent.flags, flags);
 371         copy_le16(dirent.compression_method, method);
 372         copy_le16(dirent.mtime, zip_time);
 373         copy_le16(dirent.mdate, zip_date);
 374         set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 375         copy_le16(dirent.filename_length, pathlen);
 376         copy_le16(dirent.extra_length, ZIP_EXTRA_MTIME_SIZE);
 377         copy_le16(dirent.comment_length, 0);
 378         copy_le16(dirent.disk, 0);
 379         copy_le32(dirent.attr2, attr2);
 380         copy_le32(dirent.offset, zip_offset);
 381
 382         copy_le32(header.magic, 0x04034b50);
 383         copy_le16(header.version, 10);
 384         copy_le16(header.flags, flags);
 385         copy_le16(header.compression_method, method);
 386         copy_le16(header.mtime, zip_time);
 387         copy_le16(header.mdate, zip_date);
 388         set_zip_header_data_desc(&header, size, compressed_size, crc);
 389         copy_le16(header.filename_length, pathlen);
 390         copy_le16(header.extra_length, ZIP_EXTRA_MTIME_SIZE);
 391         write_or_die(1, &header, ZIP_LOCAL_HEADER_SIZE);
 392         zip_offset += ZIP_LOCAL_HEADER_SIZE;
 393         write_or_die(1, path, pathlen);
 394         zip_offset += pathlen;
 395         write_or_die(1, &extra, ZIP_EXTRA_MTIME_SIZE);
 396         zip_offset += ZIP_EXTRA_MTIME_SIZE;
 397         if (stream && method == 0) {
 398                 unsigned char buf[STREAM_BUFFER_SIZE];
 399                 ssize_t readlen;
 400
 401                 for (;;) {
 402                         readlen = read_istream(stream, buf, sizeof(buf));
 403                         if (readlen <= 0)
 404                                 break;
 405                         crc = crc32(crc, buf, readlen);
 406                         if (is_binary == -1)
 407                                 is_binary = entry_is_binary(path_without_prefix,
 408                                                             buf, readlen);
 409                         write_or_die(1, buf, readlen);
 410                 }
 411                 close_istream(stream);
 412                 if (readlen)
 413                         return readlen;
 414
 415                 compressed_size = size;
 416                 zip_offset += compressed_size;
 417
 418                 write_zip_data_desc(size, compressed_size, crc);
 419                 zip_offset += ZIP_DATA_DESC_SIZE;
 420
 421                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 422         } else if (stream && method == 8) {
 423                 unsigned char buf[STREAM_BUFFER_SIZE];
 424                 ssize_t readlen;
 425                 git_zstream zstream;
 426                 int result;
 427                 size_t out_len;
 428                 unsigned char compressed[STREAM_BUFFER_SIZE * 2];
 429
 430                 git_deflate_init_raw(&zstream, args->compression_level);
 431
 432                 compressed_size = 0;
 433                 zstream.next_out = compressed;
 434                 zstream.avail_out = sizeof(compressed);
 435
 436                 for (;;) {
 437                         readlen = read_istream(stream, buf, sizeof(buf));
 438                         if (readlen <= 0)
 439                                 break;
 440                         crc = crc32(crc, buf, readlen);
 441                         if (is_binary == -1)
 442                                 is_binary = entry_is_binary(path_without_prefix,
 443                                                             buf, readlen);
 444
 445                         zstream.next_in = buf;
 446                         zstream.avail_in = readlen;
 447                         result = git_deflate(&zstream, 0);
 448                         if (result != Z_OK)
 449                                 die("deflate error (%d)", result);
 450                         out_len = zstream.next_out - compressed;
 451
 452                         if (out_len > 0) {
 453                                 write_or_die(1, compressed, out_len);
 454                                 compressed_size += out_len;
 455                                 zstream.next_out = compressed;
 456                                 zstream.avail_out = sizeof(compressed);
 457                         }
 458
 459                 }
 460                 close_istream(stream);
 461                 if (readlen)
 462                         return readlen;
 463
 464                 zstream.next_in = buf;
 465                 zstream.avail_in = 0;
 466                 result = git_deflate(&zstream, Z_FINISH);
 467                 if (result != Z_STREAM_END)
 468                         die("deflate error (%d)", result);
 469
 470                 git_deflate_end(&zstream);
 471                 out_len = zstream.next_out - compressed;
 472                 write_or_die(1, compressed, out_len);
 473                 compressed_size += out_len;
 474                 zip_offset += compressed_size;
 475
 476                 write_zip_data_desc(size, compressed_size, crc);
 477                 zip_offset += ZIP_DATA_DESC_SIZE;
 478
 479                 set_zip_dir_data_desc(&dirent, size, compressed_size, crc);
 480         } else if (compressed_size > 0) {
 481                 write_or_die(1, out, compressed_size);
 482                 zip_offset += compressed_size;
 483         }
 484
 485         free(deflated);
 486         free(buffer);
 487
 488         copy_le16(dirent.attr1, !is_binary);
 489
 490         memcpy(zip_dir + zip_dir_offset, &dirent, ZIP_DIR_HEADER_SIZE);
 491         zip_dir_offset += ZIP_DIR_HEADER_SIZE;
 492         memcpy(zip_dir + zip_dir_offset, path, pathlen);
 493         zip_dir_offset += pathlen;
 494         memcpy(zip_dir + zip_dir_offset, &extra, ZIP_EXTRA_MTIME_SIZE);
 495         zip_dir_offset += ZIP_EXTRA_MTIME_SIZE;
 496         zip_dir_entries++;
 497
 498         return 0;
 499 }
 500
 501 static void write_zip64_trailer(void)
 502 {
 503         struct zip64_dir_trailer trailer64;
 504         struct zip64_dir_trailer_locator locator64;
 505
 506         copy_le32(trailer64.magic, 0x06064b50);
 507         copy_le64(trailer64.record_size, ZIP64_DIR_TRAILER_RECORD_SIZE);
 508         copy_le16(trailer64.creator_version, max_creator_version);
 509         copy_le16(trailer64.version, 45);
 510         copy_le32(trailer64.disk, 0);
 511         copy_le32(trailer64.directory_start_disk, 0);
 512         copy_le64(trailer64.entries_on_this_disk, zip_dir_entries);
 513         copy_le64(trailer64.entries, zip_dir_entries);
 514         copy_le64(trailer64.size, zip_dir_offset);
 515         copy_le64(trailer64.offset, zip_offset);
 516
 517         copy_le32(locator64.magic, 0x07064b50);
 518         copy_le32(locator64.disk, 0);
 519         copy_le64(locator64.offset, zip_offset + zip_dir_offset);
 520         copy_le32(locator64.number_of_disks, 1);
 521
 522         write_or_die(1, &trailer64, ZIP64_DIR_TRAILER_SIZE);
 523         write_or_die(1, &locator64, ZIP64_DIR_TRAILER_LOCATOR_SIZE);
 524 }
 525
 526 static void write_zip_trailer(const unsigned char *sha1)
 527 {
 528         struct zip_dir_trailer trailer;
 529         int clamped = 0;
 530
 531         copy_le32(trailer.magic, 0x06054b50);
 532         copy_le16(trailer.disk, 0);
 533         copy_le16(trailer.directory_start_disk, 0);
 534         copy_le16_clamp(trailer.entries_on_this_disk, zip_dir_entries,
 535                         &clamped);
 536         copy_le16_clamp(trailer.entries, zip_dir_entries, &clamped);
 537         copy_le32(trailer.size, zip_dir_offset);
 538         copy_le32(trailer.offset, zip_offset);
 539         copy_le16(trailer.comment_length, sha1 ? GIT_SHA1_HEXSZ : 0);
 540
 541         write_or_die(1, zip_dir, zip_dir_offset);
 542         if (clamped)
 543                 write_zip64_trailer();
 544         write_or_die(1, &trailer, ZIP_DIR_TRAILER_SIZE);
 545         if (sha1)
 546                 write_or_die(1, sha1_to_hex(sha1), GIT_SHA1_HEXSZ);
 547 }
 548
 549 static void dos_time(time_t *time, int *dos_date, int *dos_time)
 550 {
 551         struct tm *t = localtime(time);
 552
 553         *dos_date = t->tm_mday + (t->tm_mon + 1) * 32 +
 554                     (t->tm_year + 1900 - 1980) * 512;
 555         *dos_time = t->tm_sec / 2 + t->tm_min * 32 + t->tm_hour * 2048;
 556 }
 557
 558 static int archive_zip_config(const char *var, const char *value, void *data)
 559 {
 560         return userdiff_config(var, value);
 561 }
 562
 563 static int write_zip_archive(const struct archiver *ar,
 564                              struct archiver_args *args)
 565 {
 566         int err;
 567
 568         git_config(archive_zip_config, NULL);
 569
 570         dos_time(&args->time, &zip_date, &zip_time);
 571
 572         zip_dir = xmalloc(ZIP_DIRECTORY_MIN_SIZE);
 573         zip_dir_size = ZIP_DIRECTORY_MIN_SIZE;
 574
 575         err = write_archive_entries(args, write_zip_entry);
 576         if (!err)
 577                 write_zip_trailer(args->commit_sha1);
 578
 579         free(zip_dir);
 580
 581         return err;
 582 }
 583
 584 static struct archiver zip_archiver = {
 585         "zip",
 586         write_zip_archive,
 587         ARCHIVER_WANT_COMPRESSION_LEVELS|ARCHIVER_REMOTE
 588 };
 589
 590 void init_zip_archiver(void)
 591 {
 592         register_archiver(&zip_archiver);
 593 }