1 /*- 2 * Copyright (c) 2004-2013 Tim Kientzle 3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA 4 * Copyright (c) 2013 Konrad Kleine 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28 #include "archive_platform.h" 29 30 /* 31 * The definitive documentation of the Zip file format is: 32 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT 33 * 34 * The Info-Zip project has pioneered various extensions to better 35 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 36 * "Ux", and 0x7875 "ux" extensions for time and ownership 37 * information. 38 * 39 * History of this code: The streaming Zip reader was first added to 40 * libarchive in January 2005. Support for seekable input sources was 41 * added in Nov 2011. Zip64 support (including a significant code 42 * refactoring) was added in 2014. 43 */ 44 45 #ifdef HAVE_ERRNO_H 46 #include <errno.h> 47 #endif 48 #ifdef HAVE_STDLIB_H 49 #include <stdlib.h> 50 #endif 51 #ifdef HAVE_ZLIB_H 52 #include <zlib.h> 53 #endif 54 #ifdef HAVE_BZLIB_H 55 #include <bzlib.h> 56 #endif 57 #ifdef HAVE_LZMA_H 58 #include <lzma.h> 59 #endif 60 #ifdef HAVE_ZSTD_H 61 #include <zstd.h> 62 #endif 63 64 #include "archive.h" 65 #include "archive_digest_private.h" 66 #include "archive_cryptor_private.h" 67 #include "archive_endian.h" 68 #include "archive_entry.h" 69 #include "archive_entry_locale.h" 70 #include "archive_hmac_private.h" 71 #include "archive_private.h" 72 #include "archive_rb.h" 73 #include "archive_read_private.h" 74 #include "archive_ppmd8_private.h" 75 76 #ifndef HAVE_ZLIB_H 77 #include "archive_crc32.h" 78 #endif 79 80 struct zip_entry { 81 struct archive_rb_node node; 82 struct zip_entry *next; 83 int64_t local_header_offset; 84 int64_t compressed_size; 85 int64_t uncompressed_size; 86 int64_t gid; 87 int64_t uid; 88 struct archive_string rsrcname; 89 time_t mtime; 90 time_t atime; 91 time_t ctime; 92 uint32_t crc32; 93 uint16_t mode; 94 uint16_t zip_flags; /* From GP Flags Field */ 95 unsigned char compression; 96 unsigned char system; /* From "version written by" */ 97 unsigned char flags; /* Our extra markers. */ 98 unsigned char decdat;/* Used for Decryption check */ 99 100 /* WinZip AES encryption extra field should be available 101 * when compression is 99. */ 102 struct { 103 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 104 unsigned vendor; 105 #define AES_VENDOR_AE_1 0x0001 106 #define AES_VENDOR_AE_2 0x0002 107 /* AES encryption strength: 108 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 109 unsigned strength; 110 /* Actual compression method. */ 111 unsigned char compression; 112 } aes_extra; 113 }; 114 115 struct trad_enc_ctx { 116 uint32_t keys[3]; 117 }; 118 119 /* Bits used in zip_flags. */ 120 #define ZIP_ENCRYPTED (1 << 0) 121 #define ZIP_LENGTH_AT_END (1 << 3) /* Also called "Streaming bit" */ 122 #define ZIP_STRONG_ENCRYPTED (1 << 6) 123 #define ZIP_UTF8_NAME (1 << 11) 124 /* See "7.2 Single Password Symmetric Encryption Method" 125 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 126 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 127 128 /* Bits used in flags. */ 129 #define LA_USED_ZIP64 (1 << 0) 130 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 131 132 /* 133 * See "WinZip - AES Encryption Information" 134 * http://www.winzip.com/aes_info.htm 135 */ 136 /* Value used in compression method. */ 137 #define WINZIP_AES_ENCRYPTION 99 138 /* Authentication code size. */ 139 #define AUTH_CODE_SIZE 10 140 /**/ 141 #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 142 143 struct zip { 144 /* Structural information about the archive. */ 145 struct archive_string format_name; 146 int64_t central_directory_offset; 147 int64_t central_directory_offset_adjusted; 148 size_t central_directory_entries_total; 149 size_t central_directory_entries_on_this_disk; 150 int has_encrypted_entries; 151 152 /* List of entries (seekable Zip only) */ 153 struct zip_entry *zip_entries; 154 struct archive_rb_tree tree; 155 struct archive_rb_tree tree_rsrc; 156 157 /* Bytes read but not yet consumed via __archive_read_consume() */ 158 size_t unconsumed; 159 160 /* Information about entry we're currently reading. */ 161 struct zip_entry *entry; 162 int64_t entry_bytes_remaining; 163 164 /* These count the number of bytes actually read for the entry. */ 165 int64_t entry_compressed_bytes_read; 166 int64_t entry_uncompressed_bytes_read; 167 168 /* Running CRC32 of the decompressed and decrypted data */ 169 unsigned long computed_crc32; 170 unsigned long (*crc32func)(unsigned long, const void *, 171 size_t); 172 char ignore_crc32; 173 174 /* Flags to mark progress of decompression. */ 175 char decompress_init; 176 char end_of_entry; 177 178 unsigned char *uncompressed_buffer; 179 size_t uncompressed_buffer_size; 180 181 #ifdef HAVE_ZLIB_H 182 z_stream stream; 183 char stream_valid; 184 #endif 185 186 #if HAVE_LZMA_H && HAVE_LIBLZMA 187 lzma_stream zipx_lzma_stream; 188 char zipx_lzma_valid; 189 #endif 190 191 #ifdef HAVE_BZLIB_H 192 bz_stream bzstream; 193 char bzstream_valid; 194 #endif 195 196 #if HAVE_ZSTD_H && HAVE_LIBZSTD 197 ZSTD_DStream *zstdstream; 198 char zstdstream_valid; 199 #endif 200 201 IByteIn zipx_ppmd_stream; 202 ssize_t zipx_ppmd_read_compressed; 203 CPpmd8 ppmd8; 204 char ppmd8_valid; 205 char ppmd8_stream_failed; 206 207 struct archive_string_conv *sconv; 208 struct archive_string_conv *sconv_default; 209 struct archive_string_conv *sconv_utf8; 210 int init_default_conversion; 211 int process_mac_extensions; 212 213 char init_decryption; 214 215 /* Decryption buffer. */ 216 /* 217 * The decrypted data starts at decrypted_ptr and 218 * extends for decrypted_bytes_remaining. Decryption 219 * adds new data to the end of this block, data is returned 220 * to clients from the beginning. When the block hits the 221 * end of decrypted_buffer, it has to be shuffled back to 222 * the beginning of the buffer. 223 */ 224 unsigned char *decrypted_buffer; 225 unsigned char *decrypted_ptr; 226 size_t decrypted_buffer_size; 227 size_t decrypted_bytes_remaining; 228 size_t decrypted_unconsumed_bytes; 229 230 /* Traditional PKWARE decryption. */ 231 struct trad_enc_ctx tctx; 232 char tctx_valid; 233 234 /* WinZip AES decryption. */ 235 /* Contexts used for AES decryption. */ 236 archive_crypto_ctx cctx; 237 char cctx_valid; 238 archive_hmac_sha1_ctx hctx; 239 char hctx_valid; 240 241 /* Strong encryption's decryption header information. */ 242 unsigned iv_size; 243 unsigned alg_id; 244 unsigned bit_len; 245 unsigned flags; 246 unsigned erd_size; 247 unsigned v_size; 248 unsigned v_crc32; 249 uint8_t *iv; 250 uint8_t *erd; 251 uint8_t *v_data; 252 }; 253 254 /* Many systems define min or MIN, but not all. */ 255 #define zipmin(a,b) ((a) < (b) ? (a) : (b)) 256 257 #ifdef HAVE_ZLIB_H 258 static int 259 zip_read_data_deflate(struct archive_read *a, const void **buff, 260 size_t *size, int64_t *offset); 261 #endif 262 #if HAVE_LZMA_H && HAVE_LIBLZMA 263 static int 264 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, 265 size_t *size, int64_t *offset); 266 #endif 267 268 /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8 269 * streams inside ZIP files. It has 2 purposes: one is to fetch the next 270 * compressed byte from the stream, second one is to increase the counter how 271 * many compressed bytes were read. */ 272 static Byte 273 ppmd_read(void* p) { 274 /* Get the handle to current decompression context. */ 275 struct archive_read *a = ((IByteIn*)p)->a; 276 struct zip *zip = (struct zip*) a->format->data; 277 ssize_t bytes_avail = 0; 278 279 /* Fetch next byte. */ 280 const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail); 281 if(bytes_avail < 1) { 282 zip->ppmd8_stream_failed = 1; 283 return 0; 284 } 285 286 __archive_read_consume(a, 1); 287 288 /* Increment the counter. */ 289 ++zip->zipx_ppmd_read_compressed; 290 291 /* Return the next compressed byte. */ 292 return data[0]; 293 } 294 295 /* ------------------------------------------------------------------------ */ 296 297 /* 298 Traditional PKWARE Decryption functions. 299 */ 300 301 static void 302 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 303 { 304 uint8_t t; 305 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 306 307 ctx->keys[0] = CRC32(ctx->keys[0], c); 308 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 309 t = (ctx->keys[1] >> 24) & 0xff; 310 ctx->keys[2] = CRC32(ctx->keys[2], t); 311 #undef CRC32 312 } 313 314 static uint8_t 315 trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) 316 { 317 unsigned temp = ctx->keys[2] | 2; 318 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 319 } 320 321 static void 322 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 323 size_t in_len, uint8_t *out, size_t out_len) 324 { 325 unsigned i, max; 326 327 max = (unsigned)((in_len < out_len)? in_len: out_len); 328 329 for (i = 0; i < max; i++) { 330 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); 331 out[i] = t; 332 trad_enc_update_keys(ctx, t); 333 } 334 } 335 336 static int 337 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 338 const uint8_t *key, size_t key_len, uint8_t *crcchk) 339 { 340 uint8_t header[12]; 341 342 if (key_len < 12) { 343 *crcchk = 0xff; 344 return -1; 345 } 346 347 ctx->keys[0] = 305419896L; 348 ctx->keys[1] = 591751049L; 349 ctx->keys[2] = 878082192L; 350 351 for (;pw_len; --pw_len) 352 trad_enc_update_keys(ctx, *pw++); 353 354 trad_enc_decrypt_update(ctx, key, 12, header, 12); 355 /* Return the last byte for CRC check. */ 356 *crcchk = header[11]; 357 return 0; 358 } 359 360 #if 0 361 static void 362 crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 363 int key_size) 364 { 365 #define MD_SIZE 20 366 archive_sha1_ctx ctx; 367 unsigned char md1[MD_SIZE]; 368 unsigned char md2[MD_SIZE * 2]; 369 unsigned char mkb[64]; 370 int i; 371 372 archive_sha1_init(&ctx); 373 archive_sha1_update(&ctx, p, size); 374 archive_sha1_final(&ctx, md1); 375 376 memset(mkb, 0x36, sizeof(mkb)); 377 for (i = 0; i < MD_SIZE; i++) 378 mkb[i] ^= md1[i]; 379 archive_sha1_init(&ctx); 380 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 381 archive_sha1_final(&ctx, md2); 382 383 memset(mkb, 0x5C, sizeof(mkb)); 384 for (i = 0; i < MD_SIZE; i++) 385 mkb[i] ^= md1[i]; 386 archive_sha1_init(&ctx); 387 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 388 archive_sha1_final(&ctx, md2 + MD_SIZE); 389 390 if (key_size > 32) 391 key_size = 32; 392 memcpy(key, md2, key_size); 393 #undef MD_SIZE 394 } 395 #endif 396 397 /* 398 * Common code for streaming or seeking modes. 399 * 400 * Includes code to read local file headers, decompress data 401 * from entry bodies, and common API. 402 */ 403 404 static unsigned long 405 real_crc32(unsigned long crc, const void *buff, size_t len) 406 { 407 return crc32(crc, buff, (unsigned int)len); 408 } 409 410 /* Used by "ignorecrc32" option to speed up tests. */ 411 static unsigned long 412 fake_crc32(unsigned long crc, const void *buff, size_t len) 413 { 414 (void)crc; /* UNUSED */ 415 (void)buff; /* UNUSED */ 416 (void)len; /* UNUSED */ 417 return 0; 418 } 419 420 static const struct { 421 int id; 422 const char * name; 423 } compression_methods[] = { 424 {0, "uncompressed"}, /* The file is stored (no compression) */ 425 {1, "shrinking"}, /* The file is Shrunk */ 426 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 427 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 428 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 429 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 430 {6, "imploded"}, /* The file is Imploded */ 431 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 432 {8, "deflation"}, /* The file is Deflated */ 433 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 434 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 435 * (old IBM TERSE) */ 436 {11, "reserved"}, /* Reserved by PKWARE */ 437 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 438 {13, "reserved"}, /* Reserved by PKWARE */ 439 {14, "lzma"}, /* LZMA (EFS) */ 440 {15, "reserved"}, /* Reserved by PKWARE */ 441 {16, "reserved"}, /* Reserved by PKWARE */ 442 {17, "reserved"}, /* Reserved by PKWARE */ 443 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 444 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 445 {93, "zstd"}, /* Zstandard (zstd) Compression */ 446 {95, "xz"}, /* XZ compressed data */ 447 {96, "jpeg"}, /* JPEG compressed data */ 448 {97, "wav-pack"}, /* WavPack compressed data */ 449 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 450 {99, "aes"} /* WinZip AES encryption */ 451 }; 452 453 static const char * 454 compression_name(const int compression) 455 { 456 static const int num_compression_methods = 457 sizeof(compression_methods)/sizeof(compression_methods[0]); 458 int i=0; 459 460 while(compression >= 0 && i < num_compression_methods) { 461 if (compression_methods[i].id == compression) 462 return compression_methods[i].name; 463 i++; 464 } 465 return "??"; 466 } 467 468 /* Convert an MSDOS-style date/time into Unix-style time. */ 469 static time_t 470 zip_time(const char *p) 471 { 472 int msTime, msDate; 473 struct tm ts; 474 475 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 476 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 477 478 memset(&ts, 0, sizeof(ts)); 479 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 480 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 481 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 482 ts.tm_hour = (msTime >> 11) & 0x1f; 483 ts.tm_min = (msTime >> 5) & 0x3f; 484 ts.tm_sec = (msTime << 1) & 0x3e; 485 ts.tm_isdst = -1; 486 return mktime(&ts); 487 } 488 489 /* 490 * The extra data is stored as a list of 491 * id1+size1+data1 + id2+size2+data2 ... 492 * triplets. id and size are 2 bytes each. 493 */ 494 static int 495 process_extra(struct archive_read *a, struct archive_entry *entry, 496 const char *p, size_t extra_length, struct zip_entry* zip_entry) 497 { 498 unsigned offset = 0; 499 struct zip *zip = (struct zip *)(a->format->data); 500 501 if (extra_length == 0) { 502 return ARCHIVE_OK; 503 } 504 505 if (extra_length < 4) { 506 size_t i = 0; 507 /* Some ZIP files may have trailing 0 bytes. Let's check they 508 * are all 0 and ignore them instead of returning an error. 509 * 510 * This is not technically correct, but some ZIP files look 511 * like this and other tools support those files - so let's 512 * also support them. 513 */ 514 for (; i < extra_length; i++) { 515 if (p[i] != 0) { 516 archive_set_error(&a->archive, 517 ARCHIVE_ERRNO_FILE_FORMAT, 518 "Too-small extra data: " 519 "Need at least 4 bytes, " 520 "but only found %d bytes", 521 (int)extra_length); 522 return ARCHIVE_FAILED; 523 } 524 } 525 526 return ARCHIVE_OK; 527 } 528 529 while (offset <= extra_length - 4) { 530 unsigned short headerid = archive_le16dec(p + offset); 531 unsigned short datasize = archive_le16dec(p + offset + 2); 532 533 offset += 4; 534 if (offset + datasize > extra_length) { 535 archive_set_error(&a->archive, 536 ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: " 537 "Need %d bytes but only found %d bytes", 538 (int)datasize, (int)(extra_length - offset)); 539 return ARCHIVE_FAILED; 540 } 541 #ifdef DEBUG 542 fprintf(stderr, "Header id 0x%04x, length %d\n", 543 headerid, datasize); 544 #endif 545 switch (headerid) { 546 case 0x0001: 547 /* Zip64 extended information extra field. */ 548 zip_entry->flags |= LA_USED_ZIP64; 549 if (zip_entry->uncompressed_size == 0xffffffff) { 550 uint64_t t = 0; 551 if (datasize < 8 552 || (t = archive_le64dec(p + offset)) > 553 INT64_MAX) { 554 archive_set_error(&a->archive, 555 ARCHIVE_ERRNO_FILE_FORMAT, 556 "Malformed 64-bit " 557 "uncompressed size"); 558 return ARCHIVE_FAILED; 559 } 560 zip_entry->uncompressed_size = t; 561 offset += 8; 562 datasize -= 8; 563 } 564 if (zip_entry->compressed_size == 0xffffffff) { 565 uint64_t t = 0; 566 if (datasize < 8 567 || (t = archive_le64dec(p + offset)) > 568 INT64_MAX) { 569 archive_set_error(&a->archive, 570 ARCHIVE_ERRNO_FILE_FORMAT, 571 "Malformed 64-bit " 572 "compressed size"); 573 return ARCHIVE_FAILED; 574 } 575 zip_entry->compressed_size = t; 576 offset += 8; 577 datasize -= 8; 578 } 579 if (zip_entry->local_header_offset == 0xffffffff) { 580 uint64_t t = 0; 581 if (datasize < 8 582 || (t = archive_le64dec(p + offset)) > 583 INT64_MAX) { 584 archive_set_error(&a->archive, 585 ARCHIVE_ERRNO_FILE_FORMAT, 586 "Malformed 64-bit " 587 "local header offset"); 588 return ARCHIVE_FAILED; 589 } 590 zip_entry->local_header_offset = t; 591 offset += 8; 592 datasize -= 8; 593 } 594 /* archive_le32dec(p + offset) gives disk 595 * on which file starts, but we don't handle 596 * multi-volume Zip files. */ 597 break; 598 #ifdef DEBUG 599 case 0x0017: 600 { 601 /* Strong encryption field. */ 602 if (archive_le16dec(p + offset) == 2) { 603 unsigned algId = 604 archive_le16dec(p + offset + 2); 605 unsigned bitLen = 606 archive_le16dec(p + offset + 4); 607 int flags = 608 archive_le16dec(p + offset + 6); 609 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 610 "flgas=%d\n", algId, bitLen,flags); 611 } 612 break; 613 } 614 #endif 615 case 0x5455: 616 { 617 /* Extended time field "UT". */ 618 int flags; 619 if (datasize == 0) { 620 archive_set_error(&a->archive, 621 ARCHIVE_ERRNO_FILE_FORMAT, 622 "Incomplete extended time field"); 623 return ARCHIVE_FAILED; 624 } 625 flags = p[offset]; 626 offset++; 627 datasize--; 628 /* Flag bits indicate which dates are present. */ 629 if (flags & 0x01) 630 { 631 #ifdef DEBUG 632 fprintf(stderr, "mtime: %lld -> %d\n", 633 (long long)zip_entry->mtime, 634 archive_le32dec(p + offset)); 635 #endif 636 if (datasize < 4) 637 break; 638 zip_entry->mtime = archive_le32dec(p + offset); 639 offset += 4; 640 datasize -= 4; 641 } 642 if (flags & 0x02) 643 { 644 if (datasize < 4) 645 break; 646 zip_entry->atime = archive_le32dec(p + offset); 647 offset += 4; 648 datasize -= 4; 649 } 650 if (flags & 0x04) 651 { 652 if (datasize < 4) 653 break; 654 zip_entry->ctime = archive_le32dec(p + offset); 655 offset += 4; 656 datasize -= 4; 657 } 658 break; 659 } 660 case 0x5855: 661 { 662 /* Info-ZIP Unix Extra Field (old version) "UX". */ 663 if (datasize >= 8) { 664 zip_entry->atime = archive_le32dec(p + offset); 665 zip_entry->mtime = 666 archive_le32dec(p + offset + 4); 667 } 668 if (datasize >= 12) { 669 zip_entry->uid = 670 archive_le16dec(p + offset + 8); 671 zip_entry->gid = 672 archive_le16dec(p + offset + 10); 673 } 674 break; 675 } 676 case 0x6c78: 677 { 678 /* Experimental 'xl' field */ 679 /* 680 * Introduced Dec 2013 to provide a way to 681 * include external file attributes (and other 682 * fields that ordinarily appear only in 683 * central directory) in local file header. 684 * This provides file type and permission 685 * information necessary to support full 686 * streaming extraction. Currently being 687 * discussed with other Zip developers 688 * ... subject to change. 689 * 690 * Format: 691 * The field starts with a bitmap that specifies 692 * which additional fields are included. The 693 * bitmap is variable length and can be extended in 694 * the future. 695 * 696 * n bytes - feature bitmap: first byte has low-order 697 * 7 bits. If high-order bit is set, a subsequent 698 * byte holds the next 7 bits, etc. 699 * 700 * if bitmap & 1, 2 byte "version made by" 701 * if bitmap & 2, 2 byte "internal file attributes" 702 * if bitmap & 4, 4 byte "external file attributes" 703 * if bitmap & 8, 2 byte comment length + n byte 704 * comment 705 */ 706 int bitmap, bitmap_last; 707 708 if (datasize < 1) 709 break; 710 bitmap_last = bitmap = 0xff & p[offset]; 711 offset += 1; 712 datasize -= 1; 713 714 /* We only support first 7 bits of bitmap; skip rest. */ 715 while ((bitmap_last & 0x80) != 0 716 && datasize >= 1) { 717 bitmap_last = p[offset]; 718 offset += 1; 719 datasize -= 1; 720 } 721 722 if (bitmap & 1) { 723 /* 2 byte "version made by" */ 724 if (datasize < 2) 725 break; 726 zip_entry->system 727 = archive_le16dec(p + offset) >> 8; 728 offset += 2; 729 datasize -= 2; 730 } 731 if (bitmap & 2) { 732 /* 2 byte "internal file attributes" */ 733 uint32_t internal_attributes; 734 if (datasize < 2) 735 break; 736 internal_attributes 737 = archive_le16dec(p + offset); 738 /* Not used by libarchive at present. */ 739 (void)internal_attributes; /* UNUSED */ 740 offset += 2; 741 datasize -= 2; 742 } 743 if (bitmap & 4) { 744 /* 4 byte "external file attributes" */ 745 uint32_t external_attributes; 746 if (datasize < 4) 747 break; 748 external_attributes 749 = archive_le32dec(p + offset); 750 if (zip_entry->system == 3) { 751 zip_entry->mode 752 = external_attributes >> 16; 753 } else if (zip_entry->system == 0) { 754 // Interpret MSDOS directory bit 755 if (0x10 == (external_attributes & 756 0x10)) { 757 zip_entry->mode = 758 AE_IFDIR | 0775; 759 } else { 760 zip_entry->mode = 761 AE_IFREG | 0664; 762 } 763 if (0x01 == (external_attributes & 764 0x01)) { 765 /* Read-only bit; 766 * strip write permissions */ 767 zip_entry->mode &= 0555; 768 } 769 } else { 770 zip_entry->mode = 0; 771 } 772 offset += 4; 773 datasize -= 4; 774 } 775 if (bitmap & 8) { 776 /* 2 byte comment length + comment */ 777 uint32_t comment_length; 778 if (datasize < 2) 779 break; 780 comment_length 781 = archive_le16dec(p + offset); 782 offset += 2; 783 datasize -= 2; 784 785 if (datasize < comment_length) 786 break; 787 /* Comment is not supported by libarchive */ 788 offset += comment_length; 789 datasize -= comment_length; 790 } 791 break; 792 } 793 case 0x7075: 794 { 795 /* Info-ZIP Unicode Path Extra Field. */ 796 if (datasize < 5 || entry == NULL) 797 break; 798 offset += 5; 799 datasize -= 5; 800 801 /* The path name in this field is always encoded 802 * in UTF-8. */ 803 if (zip->sconv_utf8 == NULL) { 804 zip->sconv_utf8 = 805 archive_string_conversion_from_charset( 806 &a->archive, "UTF-8", 1); 807 /* If the converter from UTF-8 is not 808 * available, then the path name from the main 809 * field will more likely be correct. */ 810 if (zip->sconv_utf8 == NULL) 811 break; 812 } 813 814 /* Make sure the CRC32 of the filename matches. */ 815 if (!zip->ignore_crc32) { 816 const char *cp = archive_entry_pathname(entry); 817 if (cp) { 818 unsigned long file_crc = 819 zip->crc32func(0, cp, strlen(cp)); 820 unsigned long utf_crc = 821 archive_le32dec(p + offset - 4); 822 if (file_crc != utf_crc) { 823 #ifdef DEBUG 824 fprintf(stderr, 825 "CRC filename mismatch; " 826 "CDE is %lx, but UTF8 " 827 "is outdated with %lx\n", 828 file_crc, utf_crc); 829 #endif 830 break; 831 } 832 } 833 } 834 835 if (archive_entry_copy_pathname_l(entry, 836 p + offset, datasize, zip->sconv_utf8) != 0) { 837 /* Ignore the error, and fallback to the path 838 * name from the main field. */ 839 #ifdef DEBUG 840 fprintf(stderr, "Failed to read the ZIP " 841 "0x7075 extra field path.\n"); 842 #endif 843 } 844 break; 845 } 846 case 0x7855: 847 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 848 #ifdef DEBUG 849 fprintf(stderr, "uid %d gid %d\n", 850 archive_le16dec(p + offset), 851 archive_le16dec(p + offset + 2)); 852 #endif 853 if (datasize >= 2) 854 zip_entry->uid = archive_le16dec(p + offset); 855 if (datasize >= 4) 856 zip_entry->gid = 857 archive_le16dec(p + offset + 2); 858 break; 859 case 0x7875: 860 { 861 /* Info-Zip Unix Extra Field (type 3) "ux". */ 862 int uidsize = 0, gidsize = 0; 863 864 /* TODO: support arbitrary uidsize/gidsize. */ 865 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 866 if (datasize >= 4) { 867 /* get a uid size. */ 868 uidsize = 0xff & (int)p[offset+1]; 869 if (uidsize == 2) 870 zip_entry->uid = 871 archive_le16dec( 872 p + offset + 2); 873 else if (uidsize == 4 && datasize >= 6) 874 zip_entry->uid = 875 archive_le32dec( 876 p + offset + 2); 877 } 878 if (datasize >= (2 + uidsize + 3)) { 879 /* get a gid size. */ 880 gidsize = 0xff & 881 (int)p[offset+2+uidsize]; 882 if (gidsize == 2) 883 zip_entry->gid = 884 archive_le16dec( 885 p+offset+2+uidsize+1); 886 else if (gidsize == 4 && 887 datasize >= (2 + uidsize + 5)) 888 zip_entry->gid = 889 archive_le32dec( 890 p+offset+2+uidsize+1); 891 } 892 } 893 break; 894 } 895 case 0x9901: 896 /* WinZip AES extra data field. */ 897 if (datasize < 6) { 898 archive_set_error(&a->archive, 899 ARCHIVE_ERRNO_FILE_FORMAT, 900 "Incomplete AES field"); 901 return ARCHIVE_FAILED; 902 } 903 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 904 /* Vendor version. */ 905 zip_entry->aes_extra.vendor = 906 archive_le16dec(p + offset); 907 /* AES encryption strength. */ 908 zip_entry->aes_extra.strength = p[offset + 4]; 909 /* Actual compression method. */ 910 zip_entry->aes_extra.compression = 911 p[offset + 5]; 912 } 913 break; 914 default: 915 break; 916 } 917 offset += datasize; 918 } 919 return ARCHIVE_OK; 920 } 921 922 /* 923 * Assumes file pointer is at beginning of local file header. 924 */ 925 static int 926 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 927 struct zip *zip) 928 { 929 const char *p; 930 const void *h; 931 const wchar_t *wp; 932 const char *cp; 933 size_t len, filename_length, extra_length; 934 struct archive_string_conv *sconv; 935 struct zip_entry *zip_entry = zip->entry; 936 struct zip_entry zip_entry_central_dir; 937 int ret = ARCHIVE_OK; 938 char version; 939 940 /* Save a copy of the original for consistency checks. */ 941 zip_entry_central_dir = *zip_entry; 942 943 zip->decompress_init = 0; 944 zip->end_of_entry = 0; 945 zip->entry_uncompressed_bytes_read = 0; 946 zip->entry_compressed_bytes_read = 0; 947 zip->computed_crc32 = zip->crc32func(0, NULL, 0); 948 949 /* Setup default conversion. */ 950 if (zip->sconv == NULL && !zip->init_default_conversion) { 951 zip->sconv_default = 952 archive_string_default_conversion_for_read(&(a->archive)); 953 zip->init_default_conversion = 1; 954 } 955 956 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 957 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 958 "Truncated ZIP file header"); 959 return (ARCHIVE_FATAL); 960 } 961 962 if (memcmp(p, "PK\003\004", 4) != 0) { 963 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 964 return ARCHIVE_FATAL; 965 } 966 version = p[4]; 967 zip_entry->system = p[5]; 968 zip_entry->zip_flags = archive_le16dec(p + 6); 969 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 970 zip->has_encrypted_entries = 1; 971 archive_entry_set_is_data_encrypted(entry, 1); 972 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 973 zip_entry->zip_flags & ZIP_ENCRYPTED && 974 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 975 archive_entry_set_is_metadata_encrypted(entry, 1); 976 return ARCHIVE_FATAL; 977 } 978 } 979 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 980 zip_entry->compression = (char)archive_le16dec(p + 8); 981 zip_entry->mtime = zip_time(p + 10); 982 zip_entry->crc32 = archive_le32dec(p + 14); 983 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 984 zip_entry->decdat = p[11]; 985 else 986 zip_entry->decdat = p[17]; 987 zip_entry->compressed_size = archive_le32dec(p + 18); 988 zip_entry->uncompressed_size = archive_le32dec(p + 22); 989 filename_length = archive_le16dec(p + 26); 990 extra_length = archive_le16dec(p + 28); 991 992 __archive_read_consume(a, 30); 993 994 /* Read the filename. */ 995 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 996 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 997 "Truncated ZIP file header"); 998 return (ARCHIVE_FATAL); 999 } 1000 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 1001 /* The filename is stored to be UTF-8. */ 1002 if (zip->sconv_utf8 == NULL) { 1003 zip->sconv_utf8 = 1004 archive_string_conversion_from_charset( 1005 &a->archive, "UTF-8", 1); 1006 if (zip->sconv_utf8 == NULL) 1007 return (ARCHIVE_FATAL); 1008 } 1009 sconv = zip->sconv_utf8; 1010 } else if (zip->sconv != NULL) 1011 sconv = zip->sconv; 1012 else 1013 sconv = zip->sconv_default; 1014 1015 if (archive_entry_copy_pathname_l(entry, 1016 h, filename_length, sconv) != 0) { 1017 if (errno == ENOMEM) { 1018 archive_set_error(&a->archive, ENOMEM, 1019 "Can't allocate memory for Pathname"); 1020 return (ARCHIVE_FATAL); 1021 } 1022 archive_set_error(&a->archive, 1023 ARCHIVE_ERRNO_FILE_FORMAT, 1024 "Pathname cannot be converted " 1025 "from %s to current locale.", 1026 archive_string_conversion_charset_name(sconv)); 1027 ret = ARCHIVE_WARN; 1028 } 1029 __archive_read_consume(a, filename_length); 1030 1031 /* Read the extra data. */ 1032 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 1033 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1034 "Truncated ZIP file header"); 1035 return (ARCHIVE_FATAL); 1036 } 1037 1038 if (ARCHIVE_OK != process_extra(a, entry, h, extra_length, 1039 zip_entry)) { 1040 return ARCHIVE_FATAL; 1041 } 1042 __archive_read_consume(a, extra_length); 1043 1044 /* Work around a bug in Info-Zip: When reading from a pipe, it 1045 * stats the pipe instead of synthesizing a file entry. */ 1046 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 1047 zip_entry->mode &= ~ AE_IFMT; 1048 zip_entry->mode |= AE_IFREG; 1049 } 1050 1051 /* If the mode is totally empty, set some sane default. */ 1052 if (zip_entry->mode == 0) { 1053 zip_entry->mode |= 0664; 1054 } 1055 1056 /* Windows archivers sometimes use backslash as the directory 1057 * separator. Normalize to slash. */ 1058 if (zip_entry->system == 0 && 1059 (wp = archive_entry_pathname_w(entry)) != NULL) { 1060 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { 1061 size_t i; 1062 struct archive_wstring s; 1063 archive_string_init(&s); 1064 archive_wstrcpy(&s, wp); 1065 for (i = 0; i < archive_strlen(&s); i++) { 1066 if (s.s[i] == '\\') 1067 s.s[i] = '/'; 1068 } 1069 archive_entry_copy_pathname_w(entry, s.s); 1070 archive_wstring_free(&s); 1071 } 1072 } 1073 1074 /* Make sure that entries with a trailing '/' are marked as directories 1075 * even if the External File Attributes contains bogus values. If this 1076 * is not a directory and there is no type, assume a regular file. */ 1077 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 1078 int has_slash; 1079 1080 wp = archive_entry_pathname_w(entry); 1081 if (wp != NULL) { 1082 len = wcslen(wp); 1083 has_slash = len > 0 && wp[len - 1] == L'/'; 1084 } else { 1085 cp = archive_entry_pathname(entry); 1086 len = (cp != NULL)?strlen(cp):0; 1087 has_slash = len > 0 && cp[len - 1] == '/'; 1088 } 1089 /* Correct file type as needed. */ 1090 if (has_slash) { 1091 zip_entry->mode &= ~AE_IFMT; 1092 zip_entry->mode |= AE_IFDIR; 1093 zip_entry->mode |= 0111; 1094 } else if ((zip_entry->mode & AE_IFMT) == 0) { 1095 zip_entry->mode |= AE_IFREG; 1096 } 1097 } 1098 1099 /* Make sure directories end in '/' */ 1100 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 1101 wp = archive_entry_pathname_w(entry); 1102 if (wp != NULL) { 1103 len = wcslen(wp); 1104 if (len > 0 && wp[len - 1] != L'/') { 1105 struct archive_wstring s; 1106 archive_string_init(&s); 1107 archive_wstrcat(&s, wp); 1108 archive_wstrappend_wchar(&s, L'/'); 1109 archive_entry_copy_pathname_w(entry, s.s); 1110 archive_wstring_free(&s); 1111 } 1112 } else { 1113 cp = archive_entry_pathname(entry); 1114 len = (cp != NULL)?strlen(cp):0; 1115 if (len > 0 && cp[len - 1] != '/') { 1116 struct archive_string s; 1117 archive_string_init(&s); 1118 archive_strcat(&s, cp); 1119 archive_strappend_char(&s, '/'); 1120 archive_entry_set_pathname(entry, s.s); 1121 archive_string_free(&s); 1122 } 1123 } 1124 } 1125 1126 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 1127 /* If this came from the central dir, its size info 1128 * is definitive, so ignore the length-at-end flag. */ 1129 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 1130 /* If local header is missing a value, use the one from 1131 the central directory. If both have it, warn about 1132 mismatches. */ 1133 if (zip_entry->crc32 == 0) { 1134 zip_entry->crc32 = zip_entry_central_dir.crc32; 1135 } else if (!zip->ignore_crc32 1136 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 1137 archive_set_error(&a->archive, 1138 ARCHIVE_ERRNO_FILE_FORMAT, 1139 "Inconsistent CRC32 values"); 1140 ret = ARCHIVE_WARN; 1141 } 1142 if (zip_entry->compressed_size == 0 1143 || zip_entry->compressed_size == 0xffffffff) { 1144 zip_entry->compressed_size 1145 = zip_entry_central_dir.compressed_size; 1146 } else if (zip_entry->compressed_size 1147 != zip_entry_central_dir.compressed_size) { 1148 archive_set_error(&a->archive, 1149 ARCHIVE_ERRNO_FILE_FORMAT, 1150 "Inconsistent compressed size: " 1151 "%jd in central directory, %jd in local header", 1152 (intmax_t)zip_entry_central_dir.compressed_size, 1153 (intmax_t)zip_entry->compressed_size); 1154 ret = ARCHIVE_WARN; 1155 } 1156 if (zip_entry->uncompressed_size == 0 || 1157 zip_entry->uncompressed_size == 0xffffffff) { 1158 zip_entry->uncompressed_size 1159 = zip_entry_central_dir.uncompressed_size; 1160 } else if (zip_entry->uncompressed_size 1161 != zip_entry_central_dir.uncompressed_size) { 1162 archive_set_error(&a->archive, 1163 ARCHIVE_ERRNO_FILE_FORMAT, 1164 "Inconsistent uncompressed size: " 1165 "%jd in central directory, %jd in local header", 1166 (intmax_t)zip_entry_central_dir.uncompressed_size, 1167 (intmax_t)zip_entry->uncompressed_size); 1168 ret = ARCHIVE_WARN; 1169 } 1170 } 1171 1172 /* Populate some additional entry fields: */ 1173 archive_entry_set_mode(entry, zip_entry->mode); 1174 archive_entry_set_uid(entry, zip_entry->uid); 1175 archive_entry_set_gid(entry, zip_entry->gid); 1176 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 1177 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 1178 archive_entry_set_atime(entry, zip_entry->atime, 0); 1179 1180 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 1181 size_t linkname_length; 1182 1183 if (zip_entry->compressed_size > 64 * 1024) { 1184 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1185 "Zip file with oversized link entry"); 1186 return ARCHIVE_FATAL; 1187 } 1188 1189 linkname_length = (size_t)zip_entry->compressed_size; 1190 1191 archive_entry_set_size(entry, 0); 1192 1193 // take into account link compression if any 1194 size_t linkname_full_length = linkname_length; 1195 if (zip->entry->compression != 0) 1196 { 1197 // symlink target string appeared to be compressed 1198 int status = ARCHIVE_FATAL; 1199 const void *uncompressed_buffer = NULL; 1200 1201 switch (zip->entry->compression) 1202 { 1203 #if HAVE_ZLIB_H 1204 case 8: /* Deflate compression. */ 1205 zip->entry_bytes_remaining = zip_entry->compressed_size; 1206 status = zip_read_data_deflate(a, &uncompressed_buffer, 1207 &linkname_full_length, NULL); 1208 break; 1209 #endif 1210 #if HAVE_LZMA_H && HAVE_LIBLZMA 1211 case 14: /* ZIPx LZMA compression. */ 1212 /*(see zip file format specification, section 4.4.5)*/ 1213 zip->entry_bytes_remaining = zip_entry->compressed_size; 1214 status = zip_read_data_zipx_lzma_alone(a, &uncompressed_buffer, 1215 &linkname_full_length, NULL); 1216 break; 1217 #endif 1218 default: /* Unsupported compression. */ 1219 break; 1220 } 1221 if (status == ARCHIVE_OK) 1222 { 1223 p = uncompressed_buffer; 1224 } 1225 else 1226 { 1227 archive_set_error(&a->archive, 1228 ARCHIVE_ERRNO_FILE_FORMAT, 1229 "Unsupported ZIP compression method " 1230 "during decompression of link entry (%d: %s)", 1231 zip->entry->compression, 1232 compression_name(zip->entry->compression)); 1233 return ARCHIVE_FAILED; 1234 } 1235 } 1236 else 1237 { 1238 p = __archive_read_ahead(a, linkname_length, NULL); 1239 } 1240 1241 if (p == NULL) { 1242 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1243 "Truncated Zip file"); 1244 return ARCHIVE_FATAL; 1245 } 1246 1247 sconv = zip->sconv; 1248 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1249 sconv = zip->sconv_utf8; 1250 if (sconv == NULL) 1251 sconv = zip->sconv_default; 1252 if (archive_entry_copy_symlink_l(entry, p, linkname_full_length, 1253 sconv) != 0) { 1254 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1255 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1256 archive_entry_copy_symlink_l(entry, p, 1257 linkname_full_length, NULL); 1258 if (errno == ENOMEM) { 1259 archive_set_error(&a->archive, ENOMEM, 1260 "Can't allocate memory for Symlink"); 1261 return (ARCHIVE_FATAL); 1262 } 1263 /* 1264 * Since there is no character-set regulation for 1265 * symlink name, do not report the conversion error 1266 * in an automatic conversion. 1267 */ 1268 if (sconv != zip->sconv_utf8 || 1269 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1270 archive_set_error(&a->archive, 1271 ARCHIVE_ERRNO_FILE_FORMAT, 1272 "Symlink cannot be converted " 1273 "from %s to current locale.", 1274 archive_string_conversion_charset_name( 1275 sconv)); 1276 ret = ARCHIVE_WARN; 1277 } 1278 } 1279 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1280 1281 if (__archive_read_consume(a, linkname_length) < 0) { 1282 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1283 "Read error skipping symlink target name"); 1284 return ARCHIVE_FATAL; 1285 } 1286 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1287 || (zip_entry->uncompressed_size > 0 1288 && zip_entry->uncompressed_size != 0xffffffff)) { 1289 /* Set the size only if it's meaningful. */ 1290 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1291 } 1292 zip->entry_bytes_remaining = zip_entry->compressed_size; 1293 1294 /* If there's no body, force read_data() to return EOF immediately. */ 1295 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1296 && zip->entry_bytes_remaining < 1) 1297 zip->end_of_entry = 1; 1298 1299 /* Set up a more descriptive format name. */ 1300 archive_string_empty(&zip->format_name); 1301 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1302 version / 10, version % 10, 1303 compression_name(zip->entry->compression)); 1304 a->archive.archive_format_name = zip->format_name.s; 1305 1306 return (ret); 1307 } 1308 1309 static int 1310 check_authentication_code(struct archive_read *a, const void *_p) 1311 { 1312 struct zip *zip = (struct zip *)(a->format->data); 1313 1314 /* Check authentication code. */ 1315 if (zip->hctx_valid) { 1316 const void *p; 1317 uint8_t hmac[20]; 1318 size_t hmac_len = 20; 1319 int cmp; 1320 1321 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1322 if (_p == NULL) { 1323 /* Read authentication code. */ 1324 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1325 if (p == NULL) { 1326 archive_set_error(&a->archive, 1327 ARCHIVE_ERRNO_FILE_FORMAT, 1328 "Truncated ZIP file data"); 1329 return (ARCHIVE_FATAL); 1330 } 1331 } else { 1332 p = _p; 1333 } 1334 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1335 __archive_read_consume(a, AUTH_CODE_SIZE); 1336 if (cmp != 0) { 1337 archive_set_error(&a->archive, 1338 ARCHIVE_ERRNO_MISC, 1339 "ZIP bad Authentication code"); 1340 return (ARCHIVE_WARN); 1341 } 1342 } 1343 return (ARCHIVE_OK); 1344 } 1345 1346 /* 1347 * The Zip end-of-file marker is inherently ambiguous. The specification 1348 * in APPNOTE.TXT allows any of four possible formats, and there is no 1349 * guaranteed-correct way for a reader to know a priori which one the writer 1350 * will have used. The four formats are: 1351 * 1. 32-bit format with an initial PK78 marker 1352 * 2. 32-bit format without that marker 1353 * 3. 64-bit format with the marker 1354 * 4. 64-bit format without the marker 1355 * 1356 * Mark Adler's `sunzip` streaming unzip program solved this ambiguity 1357 * by just looking at every possible combination and accepting the 1358 * longest one that matches the expected values. His approach always 1359 * consumes the longest possible matching EOF marker, based on an 1360 * analysis of all the possible failures and how the values could 1361 * overlap. 1362 * 1363 * For example, suppose both of the first two formats listed 1364 * above match. In that case, we know the next four 1365 * 32-bit words match this pattern: 1366 * ``` 1367 * [PK\07\08] [CRC32] [compressed size] [uncompressed size] 1368 * ``` 1369 * but we know they must also match this pattern: 1370 * ``` 1371 * [CRC32] [compressed size] [uncompressed size] [other PK marker] 1372 * ``` 1373 * 1374 * Since the first word here matches both the PK78 signature in the 1375 * first form and the CRC32 in the second, we know those two values 1376 * are equal, the CRC32 must be exactly 0x08074b50. Similarly, the 1377 * compressed and uncompressed size must also be exactly this value. 1378 * So we know these four words are all 0x08074b50. If we were to 1379 * accept the shorter pattern, it would be immediately followed by 1380 * another PK78 marker, which is not possible in a well-formed ZIP 1381 * archive unless there is garbage between entries. This implies we 1382 * should not accept the shorter form in such a case; we should accept 1383 * the longer form. 1384 * 1385 * If the second and third possibilities above both match, we 1386 * have a slightly different situation. The following words 1387 * must match both the 32-bit format 1388 * ``` 1389 * [CRC32] [compressed size] [uncompressed size] [other PK marker] 1390 * ``` 1391 * and the 64-bit format 1392 * ``` 1393 * [CRC32] [compressed low] [compressed high] [uncompressed low] [uncompressed high] [other PK marker] 1394 * ``` 1395 * Since the 32-bit and 64-bit compressed sizes both match, the 1396 * actual size must fit in 32 bits, which implies the high-order 1397 * word of the compressed size is zero. So we know the uncompressed 1398 * low word is zero, which again implies that if we accept the shorter 1399 * format, there will not be a valid PK marker following it. 1400 * 1401 * Similar considerations rule out the shorter form in every other 1402 * possibly-ambiguous pair. So if two of the four possible formats 1403 * match, we should accept the longer option. 1404 * 1405 * If none of the four formats matches, we know the archive must be 1406 * corrupted in some fashion. In particular, it's possible that the 1407 * length-at-end bit was incorrect and we should not really be looking 1408 * for an EOF marker at all. To allow for this possibility, we 1409 * evaluate the following words to collect data for a later error 1410 * report but do not consume any bytes. We instead rely on the later 1411 * search for a new PK marker to re-sync to the next well-formed 1412 * entry. 1413 */ 1414 static void 1415 consume_end_of_file_marker(struct archive_read *a, struct zip *zip) 1416 { 1417 const char *marker; 1418 const char *p; 1419 uint64_t compressed32, uncompressed32; 1420 uint64_t compressed64, uncompressed64; 1421 uint64_t compressed_actual, uncompressed_actual; 1422 uint32_t crc32_actual; 1423 const uint32_t PK78 = 0x08074B50ULL; 1424 uint8_t crc32_ignored, crc32_may_be_zero; 1425 1426 /* If there shouldn't be a marker, don't consume it. */ 1427 if ((zip->entry->zip_flags & ZIP_LENGTH_AT_END) == 0) { 1428 return; 1429 } 1430 1431 /* The longest Zip end-of-file record is 24 bytes. Since an 1432 * end-of-file record can never appear at the end of the 1433 * archive, we know 24 bytes will be available unless 1434 * the archive is severely truncated. */ 1435 if (NULL == (marker = __archive_read_ahead(a, 24, NULL))) { 1436 return; 1437 } 1438 p = marker; 1439 1440 /* The end-of-file record comprises: 1441 * = Optional PK\007\010 marker 1442 * = 4-byte CRC32 1443 * = Compressed size 1444 * = Uncompressed size 1445 * 1446 * The last two fields are either both 32 bits or both 64 1447 * bits. We check all possible layouts and accept any one 1448 * that gives us a complete match, else we make a best-effort 1449 * attempt to parse out the pieces. 1450 */ 1451 1452 /* CRC32 checking can be tricky: 1453 * * Test suites sometimes ignore the CRC32 1454 * * AES AE-2 always writes zero for the CRC32 1455 * * AES AE-1 sometimes writes zero for the CRC32 1456 */ 1457 crc32_ignored = zip->ignore_crc32; 1458 crc32_may_be_zero = 0; 1459 crc32_actual = zip->computed_crc32; 1460 if (zip->hctx_valid) { 1461 switch (zip->entry->aes_extra.vendor) { 1462 case AES_VENDOR_AE_2: 1463 crc32_actual = 0; 1464 break; 1465 case AES_VENDOR_AE_1: 1466 default: 1467 crc32_may_be_zero = 1; 1468 break; 1469 } 1470 } 1471 1472 /* Values computed from the actual data in the archive. */ 1473 compressed_actual = (uint64_t)zip->entry_compressed_bytes_read; 1474 uncompressed_actual = (uint64_t)zip->entry_uncompressed_bytes_read; 1475 1476 1477 /* Longest: PK78 marker, all 64-bit fields (24 bytes total) */ 1478 if (archive_le32dec(p) == PK78 1479 && ((archive_le32dec(p + 4) == crc32_actual) 1480 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) 1481 || crc32_ignored) 1482 && (archive_le64dec(p + 8) == compressed_actual) 1483 && (archive_le64dec(p + 16) == uncompressed_actual)) { 1484 if (!crc32_ignored) { 1485 zip->entry->crc32 = crc32_actual; 1486 } 1487 zip->entry->compressed_size = compressed_actual; 1488 zip->entry->uncompressed_size = uncompressed_actual; 1489 zip->unconsumed += 24; 1490 return; 1491 } 1492 1493 /* No PK78 marker, 64-bit fields (20 bytes total) */ 1494 if (((archive_le32dec(p) == crc32_actual) 1495 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) 1496 || crc32_ignored) 1497 && (archive_le64dec(p + 4) == compressed_actual) 1498 && (archive_le64dec(p + 12) == uncompressed_actual)) { 1499 if (!crc32_ignored) { 1500 zip->entry->crc32 = crc32_actual; 1501 } 1502 zip->entry->compressed_size = compressed_actual; 1503 zip->entry->uncompressed_size = uncompressed_actual; 1504 zip->unconsumed += 20; 1505 return; 1506 } 1507 1508 /* PK78 marker and 32-bit fields (16 bytes total) */ 1509 if (archive_le32dec(p) == PK78 1510 && ((archive_le32dec(p + 4) == crc32_actual) 1511 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) 1512 || crc32_ignored) 1513 && (archive_le32dec(p + 8) == compressed_actual) 1514 && (archive_le32dec(p + 12) == uncompressed_actual)) { 1515 if (!crc32_ignored) { 1516 zip->entry->crc32 = crc32_actual; 1517 } 1518 zip->entry->compressed_size = compressed_actual; 1519 zip->entry->uncompressed_size = uncompressed_actual; 1520 zip->unconsumed += 16; 1521 return; 1522 } 1523 1524 /* Shortest: No PK78 marker, all 32-bit fields (12 bytes total) */ 1525 if (((archive_le32dec(p) == crc32_actual) 1526 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0)) 1527 || crc32_ignored) 1528 && (archive_le32dec(p + 4) == compressed_actual) 1529 && (archive_le32dec(p + 8) == uncompressed_actual)) { 1530 if (!crc32_ignored) { 1531 zip->entry->crc32 = crc32_actual; 1532 } 1533 zip->entry->compressed_size = compressed_actual; 1534 zip->entry->uncompressed_size = uncompressed_actual; 1535 zip->unconsumed += 12; 1536 return; 1537 } 1538 1539 /* If none of the above patterns gives us a full exact match, 1540 * then there's something definitely amiss. The fallback code 1541 * below will parse out some plausible values for error 1542 * reporting purposes. Note that this won't actually 1543 * consume anything: 1544 * 1545 * = If there really is a marker here, the logic to resync to 1546 * the next entry will suffice to skip it. 1547 * 1548 * = There might not really be a marker: Corruption or bugs 1549 * may have set the length-at-end bit without a marker ever 1550 * having actually been written. In this case, we 1551 * explicitly should not consume any bytes, since that would 1552 * prevent us from correctly reading the next entry. 1553 */ 1554 if (archive_le32dec(p) == PK78) { 1555 p += 4; /* Ignore PK78 if it appears to be present */ 1556 } 1557 zip->entry->crc32 = archive_le32dec(p); /* Parse CRC32 */ 1558 p += 4; 1559 1560 /* Consider both 32- and 64-bit interpretations */ 1561 compressed32 = archive_le32dec(p); 1562 uncompressed32 = archive_le32dec(p + 4); 1563 compressed64 = archive_le64dec(p); 1564 uncompressed64 = archive_le64dec(p + 8); 1565 1566 /* The earlier patterns may have failed because of CRC32 1567 * mismatch, so it's still possible that both sizes match. 1568 * Try to match as many as we can... 1569 */ 1570 if (compressed32 == compressed_actual 1571 && uncompressed32 == uncompressed_actual) { 1572 /* Both 32-bit fields match */ 1573 zip->entry->compressed_size = compressed32; 1574 zip->entry->uncompressed_size = uncompressed32; 1575 } else if (compressed64 == compressed_actual 1576 || uncompressed64 == uncompressed_actual) { 1577 /* One or both 64-bit fields match */ 1578 zip->entry->compressed_size = compressed64; 1579 zip->entry->uncompressed_size = uncompressed64; 1580 } else { 1581 /* Zero or one 32-bit fields match */ 1582 zip->entry->compressed_size = compressed32; 1583 zip->entry->uncompressed_size = uncompressed32; 1584 } 1585 } 1586 1587 /* 1588 * Read "uncompressed" data. 1589 * 1590 * This is straightforward if we know the size of the data. This is 1591 * always true for the seeking reader (we've examined the Central 1592 * Directory already), and will often be true for the streaming reader 1593 * (the writer was writing uncompressed so probably knows the size). 1594 * 1595 * If we don't know the size, then life is more interesting. Note 1596 * that a careful reading of the Zip specification says that a writer 1597 * must use ZIP_LENGTH_AT_END if it cannot write the CRC into the 1598 * local header. And if it uses ZIP_LENGTH_AT_END, then it is 1599 * prohibited from storing the sizes in the local header. This 1600 * prevents fully-compliant streaming writers from providing any size 1601 * clues to a streaming reader. In this case, we have to scan the 1602 * data as we read to try to locate the end-of-file marker. 1603 * 1604 * We assume here that the end-of-file marker always has the 1605 * PK\007\010 signature. Although it's technically optional, newer 1606 * writers seem to provide it pretty consistently, and it's not clear 1607 * how to efficiently recognize an end-of-file marker that lacks it. 1608 * 1609 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1610 * zip->end_of_entry if it consumes all of the data. 1611 */ 1612 static int 1613 zip_read_data_none(struct archive_read *a, const void **_buff, 1614 size_t *size, int64_t *offset) 1615 { 1616 struct zip *zip; 1617 const char *buff; 1618 ssize_t bytes_avail; 1619 ssize_t trailing_extra; 1620 int r; 1621 1622 (void)offset; /* UNUSED */ 1623 1624 zip = (struct zip *)(a->format->data); 1625 trailing_extra = zip->hctx_valid ? AUTH_CODE_SIZE : 0; 1626 1627 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1628 const char *p; 1629 ssize_t grabbing_bytes = 24 + trailing_extra; 1630 1631 /* Grab at least 24 bytes. */ 1632 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1633 if (bytes_avail < grabbing_bytes) { 1634 /* Zip archives have end-of-archive markers 1635 that are longer than this, so a failure to get at 1636 least 24 bytes really does indicate a truncated 1637 file. */ 1638 archive_set_error(&a->archive, 1639 ARCHIVE_ERRNO_FILE_FORMAT, 1640 "Truncated ZIP file data"); 1641 return (ARCHIVE_FATAL); 1642 } 1643 /* Check for a complete PK\007\010 signature, followed 1644 * by the correct 4-byte CRC. */ 1645 p = buff + trailing_extra; 1646 if (p[0] == 'P' && p[1] == 'K' 1647 && p[2] == '\007' && p[3] == '\010' 1648 && (archive_le32dec(p + 4) == zip->computed_crc32 1649 || zip->ignore_crc32 1650 || (zip->hctx_valid 1651 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1652 zip->end_of_entry = 1; 1653 if (zip->hctx_valid) { 1654 r = check_authentication_code(a, buff); 1655 if (r != ARCHIVE_OK) 1656 return (r); 1657 } 1658 return (ARCHIVE_OK); 1659 } 1660 /* If not at EOF, ensure we consume at least one byte. */ 1661 ++p; 1662 1663 /* Scan forward until we see where a PK\007\010 signature 1664 * might be. */ 1665 /* Return bytes up until that point. On the next call, 1666 * the code above will verify the data descriptor. */ 1667 while (p < buff + bytes_avail - 4) { 1668 if (p[3] == 'P') { p += 3; } 1669 else if (p[3] == 'K') { p += 2; } 1670 else if (p[3] == '\007') { p += 1; } 1671 else if (p[3] == '\010' && p[2] == '\007' 1672 && p[1] == 'K' && p[0] == 'P') { 1673 break; 1674 } else { p += 4; } 1675 } 1676 p -= trailing_extra; 1677 bytes_avail = p - buff; 1678 } else { 1679 if (zip->entry_bytes_remaining == 0) { 1680 zip->end_of_entry = 1; 1681 if (zip->hctx_valid) { 1682 r = check_authentication_code(a, NULL); 1683 if (r != ARCHIVE_OK) 1684 return (r); 1685 } 1686 return (ARCHIVE_OK); 1687 } 1688 /* Grab a bunch of bytes. */ 1689 buff = __archive_read_ahead(a, 1, &bytes_avail); 1690 if (bytes_avail <= 0) { 1691 archive_set_error(&a->archive, 1692 ARCHIVE_ERRNO_FILE_FORMAT, 1693 "Truncated ZIP file data"); 1694 return (ARCHIVE_FATAL); 1695 } 1696 if (bytes_avail > zip->entry_bytes_remaining) 1697 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1698 } 1699 if (zip->tctx_valid || zip->cctx_valid) { 1700 size_t dec_size = bytes_avail; 1701 1702 if (dec_size > zip->decrypted_buffer_size) 1703 dec_size = zip->decrypted_buffer_size; 1704 if (zip->tctx_valid) { 1705 trad_enc_decrypt_update(&zip->tctx, 1706 (const uint8_t *)buff, dec_size, 1707 zip->decrypted_buffer, dec_size); 1708 } else { 1709 size_t dsize = dec_size; 1710 archive_hmac_sha1_update(&zip->hctx, 1711 (const uint8_t *)buff, dec_size); 1712 archive_decrypto_aes_ctr_update(&zip->cctx, 1713 (const uint8_t *)buff, dec_size, 1714 zip->decrypted_buffer, &dsize); 1715 } 1716 bytes_avail = dec_size; 1717 buff = (const char *)zip->decrypted_buffer; 1718 } 1719 zip->entry_bytes_remaining -= bytes_avail; 1720 zip->entry_uncompressed_bytes_read += bytes_avail; 1721 zip->entry_compressed_bytes_read += bytes_avail; 1722 zip->unconsumed += bytes_avail; 1723 *size = bytes_avail; 1724 *_buff = buff; 1725 return (ARCHIVE_OK); 1726 } 1727 1728 #if HAVE_LZMA_H && HAVE_LIBLZMA 1729 static int 1730 zipx_xz_init(struct archive_read *a, struct zip *zip) 1731 { 1732 lzma_ret r; 1733 1734 if(zip->zipx_lzma_valid) { 1735 lzma_end(&zip->zipx_lzma_stream); 1736 zip->zipx_lzma_valid = 0; 1737 } 1738 1739 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); 1740 r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0); 1741 if (r != LZMA_OK) { 1742 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1743 "xz initialization failed(%d)", 1744 r); 1745 1746 return (ARCHIVE_FAILED); 1747 } 1748 1749 zip->zipx_lzma_valid = 1; 1750 1751 free(zip->uncompressed_buffer); 1752 1753 zip->uncompressed_buffer_size = 256 * 1024; 1754 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); 1755 if (zip->uncompressed_buffer == NULL) { 1756 archive_set_error(&a->archive, ENOMEM, 1757 "No memory for xz decompression"); 1758 return (ARCHIVE_FATAL); 1759 } 1760 1761 zip->decompress_init = 1; 1762 return (ARCHIVE_OK); 1763 } 1764 1765 static int 1766 zipx_lzma_alone_init(struct archive_read *a, struct zip *zip) 1767 { 1768 lzma_ret r; 1769 const uint8_t* p; 1770 1771 #pragma pack(push) 1772 #pragma pack(1) 1773 struct _alone_header { 1774 uint8_t bytes[5]; 1775 uint64_t uncompressed_size; 1776 } alone_header; 1777 #pragma pack(pop) 1778 1779 if(zip->zipx_lzma_valid) { 1780 lzma_end(&zip->zipx_lzma_stream); 1781 zip->zipx_lzma_valid = 0; 1782 } 1783 1784 /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma 1785 * that is a part of XZ Utils. The stream format stored inside ZIPX 1786 * file is a modified "lzma alone" file format, that was used by the 1787 * `lzma` utility which was later deprecated in favour of `xz` utility. 1788 * Since those formats are nearly the same, we can use a standard 1789 * "lzma alone" decoder from XZ Utils. */ 1790 1791 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); 1792 r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX); 1793 if (r != LZMA_OK) { 1794 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1795 "lzma initialization failed(%d)", r); 1796 1797 return (ARCHIVE_FAILED); 1798 } 1799 1800 /* Flag the cleanup function that we want our lzma-related structures 1801 * to be freed later. */ 1802 zip->zipx_lzma_valid = 1; 1803 1804 /* The "lzma alone" file format and the stream format inside ZIPx are 1805 * almost the same. Here's an example of a structure of "lzma alone" 1806 * format: 1807 * 1808 * $ cat /bin/ls | lzma | xxd | head -n 1 1809 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814 1810 * 1811 * 5 bytes 8 bytes n bytes 1812 * <lzma_params><uncompressed_size><data...> 1813 * 1814 * lzma_params is a 5-byte blob that has to be decoded to extract 1815 * parameters of this LZMA stream. The uncompressed_size field is an 1816 * uint64_t value that contains information about the size of the 1817 * uncompressed file, or UINT64_MAX if this value is unknown. 1818 * The <data...> part is the actual lzma-compressed data stream. 1819 * 1820 * Now here's the structure of the stream inside the ZIPX file: 1821 * 1822 * $ cat stream_inside_zipx | xxd | head -n 1 1823 * 00000000: 0914 0500 5d00 8000 0000 2814 .... .... 1824 * 1825 * 2byte 2byte 5 bytes n bytes 1826 * <magic1><magic2><lzma_params><data...> 1827 * 1828 * This means that the ZIPX file contains an additional magic1 and 1829 * magic2 headers, the lzma_params field contains the same parameter 1830 * set as in the "lzma alone" format, and the <data...> field is the 1831 * same as in the "lzma alone" format as well. Note that also the zipx 1832 * format is missing the uncompressed_size field. 1833 * 1834 * So, in order to use the "lzma alone" decoder for the zipx lzma 1835 * stream, we simply need to shuffle around some fields, prepare a new 1836 * lzma alone header, feed it into lzma alone decoder so it will 1837 * initialize itself properly, and then we can start feeding normal 1838 * zipx lzma stream into the decoder. 1839 */ 1840 1841 /* Read magic1,magic2,lzma_params from the ZIPX stream. */ 1842 if(zip->entry_bytes_remaining < 9 || (p = __archive_read_ahead(a, 9, NULL)) == NULL) { 1843 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1844 "Truncated lzma data"); 1845 return (ARCHIVE_FATAL); 1846 } 1847 1848 if(p[2] != 0x05 || p[3] != 0x00) { 1849 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1850 "Invalid lzma data"); 1851 return (ARCHIVE_FATAL); 1852 } 1853 1854 /* Prepare an lzma alone header: copy the lzma_params blob into 1855 * a proper place into the lzma alone header. */ 1856 memcpy(&alone_header.bytes[0], p + 4, 5); 1857 1858 /* Initialize the 'uncompressed size' field to unknown; we'll manually 1859 * monitor how many bytes there are still to be uncompressed. */ 1860 alone_header.uncompressed_size = UINT64_MAX; 1861 1862 if(!zip->uncompressed_buffer) { 1863 zip->uncompressed_buffer_size = 256 * 1024; 1864 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); 1865 1866 if (zip->uncompressed_buffer == NULL) { 1867 archive_set_error(&a->archive, ENOMEM, 1868 "No memory for lzma decompression"); 1869 return (ARCHIVE_FATAL); 1870 } 1871 } 1872 1873 zip->zipx_lzma_stream.next_in = (void*) &alone_header; 1874 zip->zipx_lzma_stream.avail_in = sizeof(alone_header); 1875 zip->zipx_lzma_stream.total_in = 0; 1876 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1877 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; 1878 zip->zipx_lzma_stream.total_out = 0; 1879 1880 /* Feed only the header into the lzma alone decoder. This will 1881 * effectively initialize the decoder, and will not produce any 1882 * output bytes yet. */ 1883 r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1884 if (r != LZMA_OK) { 1885 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1886 "lzma stream initialization error"); 1887 return ARCHIVE_FATAL; 1888 } 1889 1890 /* We've already consumed some bytes, so take this into account. */ 1891 __archive_read_consume(a, 9); 1892 zip->entry_bytes_remaining -= 9; 1893 zip->entry_compressed_bytes_read += 9; 1894 1895 zip->decompress_init = 1; 1896 return (ARCHIVE_OK); 1897 } 1898 1899 static int 1900 zip_read_data_zipx_xz(struct archive_read *a, const void **buff, 1901 size_t *size, int64_t *offset) 1902 { 1903 struct zip* zip = (struct zip *)(a->format->data); 1904 int ret; 1905 lzma_ret lz_ret; 1906 const void* compressed_buf; 1907 ssize_t bytes_avail, in_bytes, to_consume = 0; 1908 1909 (void) offset; /* UNUSED */ 1910 1911 /* Initialize decompressor if not yet initialized. */ 1912 if (!zip->decompress_init) { 1913 ret = zipx_xz_init(a, zip); 1914 if (ret != ARCHIVE_OK) 1915 return (ret); 1916 } 1917 1918 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); 1919 if (bytes_avail < 0) { 1920 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1921 "Truncated xz file body"); 1922 return (ARCHIVE_FATAL); 1923 } 1924 1925 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); 1926 zip->zipx_lzma_stream.next_in = compressed_buf; 1927 zip->zipx_lzma_stream.avail_in = in_bytes; 1928 zip->zipx_lzma_stream.total_in = 0; 1929 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1930 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; 1931 zip->zipx_lzma_stream.total_out = 0; 1932 1933 /* Perform the decompression. */ 1934 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1935 switch(lz_ret) { 1936 case LZMA_DATA_ERROR: 1937 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1938 "xz data error (error %d)", (int) lz_ret); 1939 return (ARCHIVE_FATAL); 1940 1941 case LZMA_NO_CHECK: 1942 case LZMA_OK: 1943 break; 1944 1945 default: 1946 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1947 "xz unknown error %d", (int) lz_ret); 1948 return (ARCHIVE_FATAL); 1949 1950 case LZMA_STREAM_END: 1951 lzma_end(&zip->zipx_lzma_stream); 1952 zip->zipx_lzma_valid = 0; 1953 1954 if((int64_t) zip->zipx_lzma_stream.total_in != 1955 zip->entry_bytes_remaining) 1956 { 1957 archive_set_error(&a->archive, 1958 ARCHIVE_ERRNO_MISC, 1959 "xz premature end of stream"); 1960 return (ARCHIVE_FATAL); 1961 } 1962 1963 zip->end_of_entry = 1; 1964 break; 1965 } 1966 1967 to_consume = (ssize_t)zip->zipx_lzma_stream.total_in; 1968 1969 __archive_read_consume(a, to_consume); 1970 zip->entry_bytes_remaining -= to_consume; 1971 zip->entry_compressed_bytes_read += to_consume; 1972 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; 1973 1974 *size = (size_t)zip->zipx_lzma_stream.total_out; 1975 *buff = zip->uncompressed_buffer; 1976 1977 return (ARCHIVE_OK); 1978 } 1979 1980 static int 1981 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, 1982 size_t *size, int64_t *offset) 1983 { 1984 struct zip* zip = (struct zip *)(a->format->data); 1985 int ret; 1986 lzma_ret lz_ret; 1987 const void* compressed_buf; 1988 ssize_t bytes_avail, in_bytes, to_consume; 1989 1990 (void) offset; /* UNUSED */ 1991 1992 /* Initialize decompressor if not yet initialized. */ 1993 if (!zip->decompress_init) { 1994 ret = zipx_lzma_alone_init(a, zip); 1995 if (ret != ARCHIVE_OK) 1996 return (ret); 1997 } 1998 1999 /* Fetch more compressed data. The same note as in deflate handler 2000 * applies here as well: 2001 * 2002 * Note: '1' here is a performance optimization. Recall that the 2003 * decompression layer returns a count of available bytes; asking for 2004 * more than that forces the decompressor to combine reads by copying 2005 * data. 2006 */ 2007 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); 2008 if (bytes_avail < 0) { 2009 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2010 "Truncated lzma file body"); 2011 return (ARCHIVE_FATAL); 2012 } 2013 2014 /* Set decompressor parameters. */ 2015 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); 2016 2017 zip->zipx_lzma_stream.next_in = compressed_buf; 2018 zip->zipx_lzma_stream.avail_in = in_bytes; 2019 zip->zipx_lzma_stream.total_in = 0; 2020 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 2021 zip->zipx_lzma_stream.avail_out = 2022 /* These lzma_alone streams lack end of stream marker, so let's 2023 * make sure the unpacker won't try to unpack more than it's 2024 * supposed to. */ 2025 (size_t)zipmin((int64_t) zip->uncompressed_buffer_size, 2026 zip->entry->uncompressed_size - 2027 zip->entry_uncompressed_bytes_read); 2028 zip->zipx_lzma_stream.total_out = 0; 2029 2030 /* Perform the decompression. */ 2031 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 2032 switch(lz_ret) { 2033 case LZMA_DATA_ERROR: 2034 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2035 "lzma data error (error %d)", (int) lz_ret); 2036 return (ARCHIVE_FATAL); 2037 2038 /* This case is optional in lzma alone format. It can happen, 2039 * but most of the files don't have it. (GitHub #1257) */ 2040 case LZMA_STREAM_END: 2041 if((int64_t) zip->zipx_lzma_stream.total_in != 2042 zip->entry_bytes_remaining) 2043 { 2044 archive_set_error(&a->archive, 2045 ARCHIVE_ERRNO_MISC, 2046 "lzma alone premature end of stream"); 2047 return (ARCHIVE_FATAL); 2048 } 2049 2050 zip->end_of_entry = 1; 2051 break; 2052 2053 case LZMA_OK: 2054 break; 2055 2056 default: 2057 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2058 "lzma unknown error %d", (int) lz_ret); 2059 return (ARCHIVE_FATAL); 2060 } 2061 2062 to_consume = (ssize_t)zip->zipx_lzma_stream.total_in; 2063 2064 /* Update pointers. */ 2065 __archive_read_consume(a, to_consume); 2066 zip->entry_bytes_remaining -= to_consume; 2067 zip->entry_compressed_bytes_read += to_consume; 2068 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; 2069 2070 if(zip->entry_bytes_remaining == 0) { 2071 zip->end_of_entry = 1; 2072 } 2073 2074 /* Free lzma decoder handle because we'll no longer need it. */ 2075 /* This cannot be folded into LZMA_STREAM_END handling above 2076 * because the stream end marker is not required in this format. */ 2077 if(zip->end_of_entry) { 2078 lzma_end(&zip->zipx_lzma_stream); 2079 zip->zipx_lzma_valid = 0; 2080 } 2081 2082 /* Return values. */ 2083 *size = (size_t)zip->zipx_lzma_stream.total_out; 2084 *buff = zip->uncompressed_buffer; 2085 2086 /* If we're here, then we're good! */ 2087 return (ARCHIVE_OK); 2088 } 2089 #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */ 2090 2091 static int 2092 zipx_ppmd8_init(struct archive_read *a, struct zip *zip) 2093 { 2094 const void* p; 2095 uint32_t val; 2096 uint32_t order; 2097 uint32_t mem; 2098 uint32_t restore_method; 2099 2100 /* Remove previous decompression context if it exists. */ 2101 if(zip->ppmd8_valid) { 2102 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 2103 zip->ppmd8_valid = 0; 2104 } 2105 2106 /* Create a new decompression context. */ 2107 __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8); 2108 zip->ppmd8_stream_failed = 0; 2109 2110 /* Setup function pointers required by Ppmd8 decompressor. The 2111 * 'ppmd_read' function will feed new bytes to the decompressor, 2112 * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */ 2113 zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream; 2114 zip->zipx_ppmd_stream.a = a; 2115 zip->zipx_ppmd_stream.Read = &ppmd_read; 2116 2117 /* Reset number of read bytes to 0. */ 2118 zip->zipx_ppmd_read_compressed = 0; 2119 2120 /* Read Ppmd8 header (2 bytes). */ 2121 p = __archive_read_ahead(a, 2, NULL); 2122 if(!p) { 2123 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2124 "Truncated file data in PPMd8 stream"); 2125 return (ARCHIVE_FATAL); 2126 } 2127 __archive_read_consume(a, 2); 2128 2129 /* Decode the stream's compression parameters. */ 2130 val = archive_le16dec(p); 2131 order = (val & 15) + 1; 2132 mem = ((val >> 4) & 0xff) + 1; 2133 restore_method = (val >> 12); 2134 2135 if(order < 2 || restore_method > 2) { 2136 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2137 "Invalid parameter set in PPMd8 stream (order=%" PRId32 ", " 2138 "restore=%" PRId32 ")", order, restore_method); 2139 return (ARCHIVE_FAILED); 2140 } 2141 2142 /* Allocate the memory needed to properly decompress the file. */ 2143 if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) { 2144 archive_set_error(&a->archive, ENOMEM, 2145 "Unable to allocate memory for PPMd8 stream: %" PRId32 " bytes", 2146 mem << 20); 2147 return (ARCHIVE_FATAL); 2148 } 2149 2150 /* Signal the cleanup function to release Ppmd8 context in the 2151 * cleanup phase. */ 2152 zip->ppmd8_valid = 1; 2153 2154 /* Perform further Ppmd8 initialization. */ 2155 if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) { 2156 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 2157 "PPMd8 stream range decoder initialization error"); 2158 return (ARCHIVE_FATAL); 2159 } 2160 2161 __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order, 2162 restore_method); 2163 2164 /* Allocate the buffer that will hold uncompressed data. */ 2165 free(zip->uncompressed_buffer); 2166 2167 zip->uncompressed_buffer_size = 256 * 1024; 2168 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); 2169 2170 if(zip->uncompressed_buffer == NULL) { 2171 archive_set_error(&a->archive, ENOMEM, 2172 "No memory for PPMd8 decompression"); 2173 return ARCHIVE_FATAL; 2174 } 2175 2176 /* Ppmd8 initialization is done. */ 2177 zip->decompress_init = 1; 2178 2179 /* We've already read 2 bytes in the output stream. Additionally, 2180 * Ppmd8 initialization code could read some data as well. So we 2181 * are advancing the stream by 2 bytes plus whatever number of 2182 * bytes Ppmd8 init function used. */ 2183 zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed; 2184 2185 return ARCHIVE_OK; 2186 } 2187 2188 static int 2189 zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff, 2190 size_t *size, int64_t *offset) 2191 { 2192 struct zip* zip = (struct zip *)(a->format->data); 2193 int ret; 2194 size_t consumed_bytes = 0; 2195 ssize_t bytes_avail = 0; 2196 2197 (void) offset; /* UNUSED */ 2198 2199 /* If we're here for the first time, initialize Ppmd8 decompression 2200 * context first. */ 2201 if(!zip->decompress_init) { 2202 ret = zipx_ppmd8_init(a, zip); 2203 if(ret != ARCHIVE_OK) 2204 return ret; 2205 } 2206 2207 /* Fetch for more data. We're reading 1 byte here, but libarchive 2208 * should prefetch more bytes. */ 2209 (void) __archive_read_ahead(a, 1, &bytes_avail); 2210 if(bytes_avail < 0) { 2211 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2212 "Truncated PPMd8 file body"); 2213 return (ARCHIVE_FATAL); 2214 } 2215 2216 /* This counter will be updated inside ppmd_read(), which at one 2217 * point will be called by Ppmd8_DecodeSymbol. */ 2218 zip->zipx_ppmd_read_compressed = 0; 2219 2220 /* Decompression loop. */ 2221 do { 2222 int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol( 2223 &zip->ppmd8); 2224 if(sym < 0) { 2225 zip->end_of_entry = 1; 2226 break; 2227 } 2228 2229 /* This field is set by ppmd_read() when there was no more data 2230 * to be read. */ 2231 if(zip->ppmd8_stream_failed) { 2232 archive_set_error(&a->archive, 2233 ARCHIVE_ERRNO_FILE_FORMAT, 2234 "Truncated PPMd8 file body"); 2235 return (ARCHIVE_FATAL); 2236 } 2237 2238 zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym; 2239 ++consumed_bytes; 2240 } while(consumed_bytes < zip->uncompressed_buffer_size); 2241 2242 /* Update pointers so we can continue decompression in another call. */ 2243 zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed; 2244 zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed; 2245 zip->entry_uncompressed_bytes_read += consumed_bytes; 2246 2247 /* If we're at the end of stream, deinitialize Ppmd8 context. */ 2248 if(zip->end_of_entry) { 2249 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 2250 zip->ppmd8_valid = 0; 2251 } 2252 2253 /* Update pointers for libarchive. */ 2254 *buff = zip->uncompressed_buffer; 2255 *size = consumed_bytes; 2256 2257 return ARCHIVE_OK; 2258 } 2259 2260 #ifdef HAVE_BZLIB_H 2261 static int 2262 zipx_bzip2_init(struct archive_read *a, struct zip *zip) 2263 { 2264 int r; 2265 2266 /* Deallocate already existing BZ2 decompression context if it 2267 * exists. */ 2268 if(zip->bzstream_valid) { 2269 BZ2_bzDecompressEnd(&zip->bzstream); 2270 zip->bzstream_valid = 0; 2271 } 2272 2273 /* Allocate a new BZ2 decompression context. */ 2274 memset(&zip->bzstream, 0, sizeof(bz_stream)); 2275 r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1); 2276 if(r != BZ_OK) { 2277 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 2278 "bzip2 initialization failed(%d)", 2279 r); 2280 2281 return ARCHIVE_FAILED; 2282 } 2283 2284 /* Mark the bzstream field to be released in cleanup phase. */ 2285 zip->bzstream_valid = 1; 2286 2287 /* (Re)allocate the buffer that will contain decompressed bytes. */ 2288 free(zip->uncompressed_buffer); 2289 2290 zip->uncompressed_buffer_size = 256 * 1024; 2291 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); 2292 if (zip->uncompressed_buffer == NULL) { 2293 archive_set_error(&a->archive, ENOMEM, 2294 "No memory for bzip2 decompression"); 2295 return ARCHIVE_FATAL; 2296 } 2297 2298 /* Initialization done. */ 2299 zip->decompress_init = 1; 2300 return ARCHIVE_OK; 2301 } 2302 2303 static int 2304 zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff, 2305 size_t *size, int64_t *offset) 2306 { 2307 struct zip *zip = (struct zip *)(a->format->data); 2308 ssize_t bytes_avail = 0, in_bytes, to_consume; 2309 const void *compressed_buff; 2310 int r; 2311 uint64_t total_out; 2312 2313 (void) offset; /* UNUSED */ 2314 2315 /* Initialize decompression context if we're here for the first time. */ 2316 if(!zip->decompress_init) { 2317 r = zipx_bzip2_init(a, zip); 2318 if(r != ARCHIVE_OK) 2319 return r; 2320 } 2321 2322 /* Fetch more compressed bytes. */ 2323 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); 2324 if(bytes_avail < 0) { 2325 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2326 "Truncated bzip2 file body"); 2327 return (ARCHIVE_FATAL); 2328 } 2329 2330 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); 2331 if(in_bytes < 1) { 2332 /* libbz2 doesn't complain when caller feeds avail_in == 0. 2333 * It will actually return success in this case, which is 2334 * undesirable. This is why we need to make this check 2335 * manually. */ 2336 2337 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2338 "Truncated bzip2 file body"); 2339 return (ARCHIVE_FATAL); 2340 } 2341 2342 /* Setup buffer boundaries. */ 2343 zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff; 2344 zip->bzstream.avail_in = (uint32_t)in_bytes; 2345 zip->bzstream.total_in_hi32 = 0; 2346 zip->bzstream.total_in_lo32 = 0; 2347 zip->bzstream.next_out = (char*) zip->uncompressed_buffer; 2348 zip->bzstream.avail_out = (uint32_t)zip->uncompressed_buffer_size; 2349 zip->bzstream.total_out_hi32 = 0; 2350 zip->bzstream.total_out_lo32 = 0; 2351 2352 /* Perform the decompression. */ 2353 r = BZ2_bzDecompress(&zip->bzstream); 2354 switch(r) { 2355 case BZ_STREAM_END: 2356 /* If we're at the end of the stream, deinitialize the 2357 * decompression context now. */ 2358 switch(BZ2_bzDecompressEnd(&zip->bzstream)) { 2359 case BZ_OK: 2360 break; 2361 default: 2362 archive_set_error(&a->archive, 2363 ARCHIVE_ERRNO_MISC, 2364 "Failed to clean up bzip2 " 2365 "decompressor"); 2366 return ARCHIVE_FATAL; 2367 } 2368 2369 zip->end_of_entry = 1; 2370 break; 2371 case BZ_OK: 2372 /* The decompressor has successfully decoded this 2373 * chunk of data, but more data is still in queue. */ 2374 break; 2375 default: 2376 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2377 "bzip2 decompression failed"); 2378 return ARCHIVE_FATAL; 2379 } 2380 2381 /* Update the pointers so decompressor can continue decoding. */ 2382 to_consume = zip->bzstream.total_in_lo32; 2383 __archive_read_consume(a, to_consume); 2384 2385 total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) | 2386 zip->bzstream.total_out_lo32; 2387 2388 zip->entry_bytes_remaining -= to_consume; 2389 zip->entry_compressed_bytes_read += to_consume; 2390 zip->entry_uncompressed_bytes_read += total_out; 2391 2392 /* Give libarchive its due. */ 2393 *size = (size_t)total_out; 2394 *buff = zip->uncompressed_buffer; 2395 2396 return ARCHIVE_OK; 2397 } 2398 2399 #endif 2400 2401 #if HAVE_ZSTD_H && HAVE_LIBZSTD 2402 static int 2403 zipx_zstd_init(struct archive_read *a, struct zip *zip) 2404 { 2405 size_t r; 2406 2407 /* Deallocate already existing Zstd decompression context if it 2408 * exists. */ 2409 if(zip->zstdstream_valid) { 2410 ZSTD_freeDStream(zip->zstdstream); 2411 zip->zstdstream_valid = 0; 2412 } 2413 2414 /* Allocate a new Zstd decompression context. */ 2415 zip->zstdstream = ZSTD_createDStream(); 2416 2417 r = ZSTD_initDStream(zip->zstdstream); 2418 if (ZSTD_isError(r)) { 2419 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2420 "Error initializing zstd decompressor: %s", 2421 ZSTD_getErrorName(r)); 2422 2423 return ARCHIVE_FAILED; 2424 } 2425 2426 /* Mark the zstdstream field to be released in cleanup phase. */ 2427 zip->zstdstream_valid = 1; 2428 2429 /* (Re)allocate the buffer that will contain decompressed bytes. */ 2430 free(zip->uncompressed_buffer); 2431 2432 zip->uncompressed_buffer_size = ZSTD_DStreamOutSize(); 2433 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); 2434 if (zip->uncompressed_buffer == NULL) { 2435 archive_set_error(&a->archive, ENOMEM, 2436 "No memory for Zstd decompression"); 2437 2438 return ARCHIVE_FATAL; 2439 } 2440 2441 /* Initialization done. */ 2442 zip->decompress_init = 1; 2443 return ARCHIVE_OK; 2444 } 2445 2446 static int 2447 zip_read_data_zipx_zstd(struct archive_read *a, const void **buff, 2448 size_t *size, int64_t *offset) 2449 { 2450 struct zip *zip = (struct zip *)(a->format->data); 2451 ssize_t bytes_avail = 0, in_bytes, to_consume; 2452 const void *compressed_buff; 2453 int r; 2454 size_t ret; 2455 uint64_t total_out; 2456 ZSTD_outBuffer out; 2457 ZSTD_inBuffer in; 2458 2459 (void) offset; /* UNUSED */ 2460 2461 /* Initialize decompression context if we're here for the first time. */ 2462 if(!zip->decompress_init) { 2463 r = zipx_zstd_init(a, zip); 2464 if(r != ARCHIVE_OK) 2465 return r; 2466 } 2467 2468 /* Fetch more compressed bytes */ 2469 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); 2470 if(bytes_avail < 0) { 2471 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2472 "Truncated zstd file body"); 2473 return (ARCHIVE_FATAL); 2474 } 2475 2476 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail); 2477 if(in_bytes < 1) { 2478 /* zstd doesn't complain when caller feeds avail_in == 0. 2479 * It will actually return success in this case, which is 2480 * undesirable. This is why we need to make this check 2481 * manually. */ 2482 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2483 "Truncated zstd file body"); 2484 return (ARCHIVE_FATAL); 2485 } 2486 2487 /* Setup buffer boundaries */ 2488 in.src = compressed_buff; 2489 in.size = in_bytes; 2490 in.pos = 0; 2491 out = (ZSTD_outBuffer) { zip->uncompressed_buffer, zip->uncompressed_buffer_size, 0 }; 2492 2493 /* Perform the decompression. */ 2494 ret = ZSTD_decompressStream(zip->zstdstream, &out, &in); 2495 if (ZSTD_isError(ret)) { 2496 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2497 "Error during zstd decompression: %s", 2498 ZSTD_getErrorName(ret)); 2499 return (ARCHIVE_FATAL); 2500 } 2501 2502 /* Check end of the stream. */ 2503 if (ret == 0) { 2504 if ((in.pos == in.size) && (out.pos < out.size)) { 2505 zip->end_of_entry = 1; 2506 ZSTD_freeDStream(zip->zstdstream); 2507 zip->zstdstream_valid = 0; 2508 } 2509 } 2510 2511 /* Update the pointers so decompressor can continue decoding. */ 2512 to_consume = in.pos; 2513 __archive_read_consume(a, to_consume); 2514 2515 total_out = out.pos; 2516 2517 zip->entry_bytes_remaining -= to_consume; 2518 zip->entry_compressed_bytes_read += to_consume; 2519 zip->entry_uncompressed_bytes_read += total_out; 2520 2521 /* Give libarchive its due. */ 2522 *size = (size_t)total_out; 2523 *buff = zip->uncompressed_buffer; 2524 2525 return ARCHIVE_OK; 2526 } 2527 #endif 2528 2529 #ifdef HAVE_ZLIB_H 2530 static int 2531 zip_deflate_init(struct archive_read *a, struct zip *zip) 2532 { 2533 int r; 2534 2535 /* If we haven't yet read any data, initialize the decompressor. */ 2536 if (!zip->decompress_init) { 2537 if (zip->stream_valid) 2538 r = inflateReset(&zip->stream); 2539 else 2540 r = inflateInit2(&zip->stream, 2541 -15 /* Don't check for zlib header */); 2542 if (r != Z_OK) { 2543 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2544 "Can't initialize ZIP decompression."); 2545 return (ARCHIVE_FATAL); 2546 } 2547 /* Stream structure has been set up. */ 2548 zip->stream_valid = 1; 2549 /* We've initialized decompression for this stream. */ 2550 zip->decompress_init = 1; 2551 } 2552 return (ARCHIVE_OK); 2553 } 2554 2555 static int 2556 zip_read_data_deflate(struct archive_read *a, const void **buff, 2557 size_t *size, int64_t *offset) 2558 { 2559 struct zip *zip; 2560 ssize_t bytes_avail, to_consume = 0; 2561 const void *compressed_buff, *sp; 2562 int r; 2563 2564 (void)offset; /* UNUSED */ 2565 2566 zip = (struct zip *)(a->format->data); 2567 2568 /* If the buffer hasn't been allocated, allocate it now. */ 2569 if (zip->uncompressed_buffer == NULL) { 2570 zip->uncompressed_buffer_size = 256 * 1024; 2571 zip->uncompressed_buffer 2572 = malloc(zip->uncompressed_buffer_size); 2573 if (zip->uncompressed_buffer == NULL) { 2574 archive_set_error(&a->archive, ENOMEM, 2575 "No memory for ZIP decompression"); 2576 return (ARCHIVE_FATAL); 2577 } 2578 } 2579 2580 r = zip_deflate_init(a, zip); 2581 if (r != ARCHIVE_OK) 2582 return (r); 2583 2584 /* 2585 * Note: '1' here is a performance optimization. 2586 * Recall that the decompression layer returns a count of 2587 * available bytes; asking for more than that forces the 2588 * decompressor to combine reads by copying data. 2589 */ 2590 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 2591 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2592 && bytes_avail > zip->entry_bytes_remaining) { 2593 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 2594 } 2595 if (bytes_avail < 0) { 2596 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2597 "Truncated ZIP file body"); 2598 return (ARCHIVE_FATAL); 2599 } 2600 2601 if (zip->tctx_valid || zip->cctx_valid) { 2602 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 2603 size_t buff_remaining = 2604 (zip->decrypted_buffer + 2605 zip->decrypted_buffer_size) 2606 - (zip->decrypted_ptr + 2607 zip->decrypted_bytes_remaining); 2608 2609 if (buff_remaining > (size_t)bytes_avail) 2610 buff_remaining = (size_t)bytes_avail; 2611 2612 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 2613 zip->entry_bytes_remaining > 0) { 2614 if ((int64_t)(zip->decrypted_bytes_remaining 2615 + buff_remaining) 2616 > zip->entry_bytes_remaining) { 2617 if (zip->entry_bytes_remaining < 2618 (int64_t)zip->decrypted_bytes_remaining) 2619 buff_remaining = 0; 2620 else 2621 buff_remaining = 2622 (size_t)zip->entry_bytes_remaining 2623 - zip->decrypted_bytes_remaining; 2624 } 2625 } 2626 if (buff_remaining > 0) { 2627 if (zip->tctx_valid) { 2628 trad_enc_decrypt_update(&zip->tctx, 2629 compressed_buff, buff_remaining, 2630 zip->decrypted_ptr 2631 + zip->decrypted_bytes_remaining, 2632 buff_remaining); 2633 } else { 2634 size_t dsize = buff_remaining; 2635 archive_decrypto_aes_ctr_update( 2636 &zip->cctx, 2637 compressed_buff, buff_remaining, 2638 zip->decrypted_ptr 2639 + zip->decrypted_bytes_remaining, 2640 &dsize); 2641 } 2642 zip->decrypted_bytes_remaining += 2643 buff_remaining; 2644 } 2645 } 2646 bytes_avail = zip->decrypted_bytes_remaining; 2647 compressed_buff = (const char *)zip->decrypted_ptr; 2648 } 2649 2650 /* 2651 * A bug in zlib.h: stream.next_in should be marked 'const' 2652 * but isn't (the library never alters data through the 2653 * next_in pointer, only reads it). The result: this ugly 2654 * cast to remove 'const'. 2655 */ 2656 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 2657 zip->stream.avail_in = (uInt)bytes_avail; 2658 zip->stream.total_in = 0; 2659 zip->stream.next_out = zip->uncompressed_buffer; 2660 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 2661 zip->stream.total_out = 0; 2662 2663 r = inflate(&zip->stream, 0); 2664 switch (r) { 2665 case Z_OK: 2666 break; 2667 case Z_STREAM_END: 2668 zip->end_of_entry = 1; 2669 break; 2670 case Z_MEM_ERROR: 2671 archive_set_error(&a->archive, ENOMEM, 2672 "Out of memory for ZIP decompression"); 2673 return (ARCHIVE_FATAL); 2674 default: 2675 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2676 "ZIP decompression failed (%d)", r); 2677 return (ARCHIVE_FATAL); 2678 } 2679 2680 /* Consume as much as the compressor actually used. */ 2681 to_consume = zip->stream.total_in; 2682 __archive_read_consume(a, to_consume); 2683 zip->entry_bytes_remaining -= to_consume; 2684 zip->entry_compressed_bytes_read += to_consume; 2685 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 2686 2687 if (zip->tctx_valid || zip->cctx_valid) { 2688 zip->decrypted_bytes_remaining -= to_consume; 2689 if (zip->decrypted_bytes_remaining == 0) 2690 zip->decrypted_ptr = zip->decrypted_buffer; 2691 else 2692 zip->decrypted_ptr += to_consume; 2693 } 2694 if (zip->hctx_valid) 2695 archive_hmac_sha1_update(&zip->hctx, sp, to_consume); 2696 2697 if (zip->end_of_entry) { 2698 if (zip->hctx_valid) { 2699 r = check_authentication_code(a, NULL); 2700 if (r != ARCHIVE_OK) { 2701 return (r); 2702 } 2703 } 2704 } 2705 2706 *size = zip->stream.total_out; 2707 *buff = zip->uncompressed_buffer; 2708 2709 return (ARCHIVE_OK); 2710 } 2711 #endif 2712 2713 static int 2714 read_decryption_header(struct archive_read *a) 2715 { 2716 struct zip *zip = (struct zip *)(a->format->data); 2717 const char *p; 2718 unsigned int remaining_size; 2719 unsigned int ts; 2720 2721 /* 2722 * Read an initialization vector data field. 2723 */ 2724 p = __archive_read_ahead(a, 2, NULL); 2725 if (p == NULL) 2726 goto truncated; 2727 ts = zip->iv_size; 2728 zip->iv_size = archive_le16dec(p); 2729 __archive_read_consume(a, 2); 2730 if (ts < zip->iv_size) { 2731 free(zip->iv); 2732 zip->iv = NULL; 2733 } 2734 p = __archive_read_ahead(a, zip->iv_size, NULL); 2735 if (p == NULL) 2736 goto truncated; 2737 if (zip->iv == NULL) { 2738 zip->iv = malloc(zip->iv_size); 2739 if (zip->iv == NULL) 2740 goto nomem; 2741 } 2742 memcpy(zip->iv, p, zip->iv_size); 2743 __archive_read_consume(a, zip->iv_size); 2744 2745 /* 2746 * Read a size of remaining decryption header field. 2747 */ 2748 p = __archive_read_ahead(a, 14, NULL); 2749 if (p == NULL) 2750 goto truncated; 2751 remaining_size = archive_le32dec(p); 2752 if (remaining_size < 16 || remaining_size > (1 << 18)) 2753 goto corrupted; 2754 2755 /* Check if format version is supported. */ 2756 if (archive_le16dec(p+4) != 3) { 2757 archive_set_error(&a->archive, 2758 ARCHIVE_ERRNO_FILE_FORMAT, 2759 "Unsupported encryption format version: %u", 2760 archive_le16dec(p+4)); 2761 return (ARCHIVE_FAILED); 2762 } 2763 2764 /* 2765 * Read an encryption algorithm field. 2766 */ 2767 zip->alg_id = archive_le16dec(p+6); 2768 switch (zip->alg_id) { 2769 case 0x6601:/* DES */ 2770 case 0x6602:/* RC2 */ 2771 case 0x6603:/* 3DES 168 */ 2772 case 0x6609:/* 3DES 112 */ 2773 case 0x660E:/* AES 128 */ 2774 case 0x660F:/* AES 192 */ 2775 case 0x6610:/* AES 256 */ 2776 case 0x6702:/* RC2 (version >= 5.2) */ 2777 case 0x6720:/* Blowfish */ 2778 case 0x6721:/* Twofish */ 2779 case 0x6801:/* RC4 */ 2780 /* Supported encryption algorithm. */ 2781 break; 2782 default: 2783 archive_set_error(&a->archive, 2784 ARCHIVE_ERRNO_FILE_FORMAT, 2785 "Unknown encryption algorithm: %u", zip->alg_id); 2786 return (ARCHIVE_FAILED); 2787 } 2788 2789 /* 2790 * Read a bit length field. 2791 */ 2792 zip->bit_len = archive_le16dec(p+8); 2793 2794 /* 2795 * Read a flags field. 2796 */ 2797 zip->flags = archive_le16dec(p+10); 2798 switch (zip->flags & 0xf000) { 2799 case 0x0001: /* Password is required to decrypt. */ 2800 case 0x0002: /* Certificates only. */ 2801 case 0x0003: /* Password or certificate required to decrypt. */ 2802 break; 2803 default: 2804 archive_set_error(&a->archive, 2805 ARCHIVE_ERRNO_FILE_FORMAT, 2806 "Unknown encryption flag: %u", zip->flags); 2807 return (ARCHIVE_FAILED); 2808 } 2809 if ((zip->flags & 0xf000) == 0 || 2810 (zip->flags & 0xf000) == 0x4000) { 2811 archive_set_error(&a->archive, 2812 ARCHIVE_ERRNO_FILE_FORMAT, 2813 "Unknown encryption flag: %u", zip->flags); 2814 return (ARCHIVE_FAILED); 2815 } 2816 2817 /* 2818 * Read an encrypted random data field. 2819 */ 2820 ts = zip->erd_size; 2821 zip->erd_size = archive_le16dec(p+12); 2822 __archive_read_consume(a, 14); 2823 if ((zip->erd_size & 0xf) != 0 || 2824 (zip->erd_size + 16) > remaining_size || 2825 (zip->erd_size + 16) < zip->erd_size) 2826 goto corrupted; 2827 2828 if (ts < zip->erd_size) { 2829 free(zip->erd); 2830 zip->erd = NULL; 2831 } 2832 p = __archive_read_ahead(a, zip->erd_size, NULL); 2833 if (p == NULL) 2834 goto truncated; 2835 if (zip->erd == NULL) { 2836 zip->erd = malloc(zip->erd_size); 2837 if (zip->erd == NULL) 2838 goto nomem; 2839 } 2840 memcpy(zip->erd, p, zip->erd_size); 2841 __archive_read_consume(a, zip->erd_size); 2842 2843 /* 2844 * Read a reserved data field. 2845 */ 2846 p = __archive_read_ahead(a, 4, NULL); 2847 if (p == NULL) 2848 goto truncated; 2849 /* Reserved data size should be zero. */ 2850 if (archive_le32dec(p) != 0) 2851 goto corrupted; 2852 __archive_read_consume(a, 4); 2853 2854 /* 2855 * Read a password validation data field. 2856 */ 2857 p = __archive_read_ahead(a, 2, NULL); 2858 if (p == NULL) 2859 goto truncated; 2860 ts = zip->v_size; 2861 zip->v_size = archive_le16dec(p); 2862 __archive_read_consume(a, 2); 2863 if ((zip->v_size & 0x0f) != 0 || 2864 (zip->erd_size + zip->v_size + 16) > remaining_size || 2865 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 2866 goto corrupted; 2867 if (ts < zip->v_size) { 2868 free(zip->v_data); 2869 zip->v_data = NULL; 2870 } 2871 p = __archive_read_ahead(a, zip->v_size, NULL); 2872 if (p == NULL) 2873 goto truncated; 2874 if (zip->v_data == NULL) { 2875 zip->v_data = malloc(zip->v_size); 2876 if (zip->v_data == NULL) 2877 goto nomem; 2878 } 2879 memcpy(zip->v_data, p, zip->v_size); 2880 __archive_read_consume(a, zip->v_size); 2881 2882 p = __archive_read_ahead(a, 4, NULL); 2883 if (p == NULL) 2884 goto truncated; 2885 zip->v_crc32 = archive_le32dec(p); 2886 __archive_read_consume(a, 4); 2887 2888 /*return (ARCHIVE_OK); 2889 * This is not fully implemented yet.*/ 2890 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2891 "Encrypted file is unsupported"); 2892 return (ARCHIVE_FAILED); 2893 truncated: 2894 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2895 "Truncated ZIP file data"); 2896 return (ARCHIVE_FATAL); 2897 corrupted: 2898 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2899 "Corrupted ZIP file data"); 2900 return (ARCHIVE_FATAL); 2901 nomem: 2902 archive_set_error(&a->archive, ENOMEM, 2903 "No memory for ZIP decryption"); 2904 return (ARCHIVE_FATAL); 2905 } 2906 2907 static int 2908 zip_alloc_decryption_buffer(struct archive_read *a) 2909 { 2910 struct zip *zip = (struct zip *)(a->format->data); 2911 size_t bs = 256 * 1024; 2912 2913 if (zip->decrypted_buffer == NULL) { 2914 zip->decrypted_buffer_size = bs; 2915 zip->decrypted_buffer = malloc(bs); 2916 if (zip->decrypted_buffer == NULL) { 2917 archive_set_error(&a->archive, ENOMEM, 2918 "No memory for ZIP decryption"); 2919 return (ARCHIVE_FATAL); 2920 } 2921 } 2922 zip->decrypted_ptr = zip->decrypted_buffer; 2923 return (ARCHIVE_OK); 2924 } 2925 2926 static int 2927 init_traditional_PKWARE_decryption(struct archive_read *a) 2928 { 2929 struct zip *zip = (struct zip *)(a->format->data); 2930 const void *p; 2931 int retry; 2932 int r; 2933 2934 if (zip->tctx_valid) 2935 return (ARCHIVE_OK); 2936 2937 /* 2938 Read the 12 bytes encryption header stored at 2939 the start of the data area. 2940 */ 2941 #define ENC_HEADER_SIZE 12 2942 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2943 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 2944 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2945 "Truncated Zip encrypted body: only %jd bytes available", 2946 (intmax_t)zip->entry_bytes_remaining); 2947 return (ARCHIVE_FATAL); 2948 } 2949 2950 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 2951 if (p == NULL) { 2952 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2953 "Truncated ZIP file data"); 2954 return (ARCHIVE_FATAL); 2955 } 2956 2957 for (retry = 0;; retry++) { 2958 const char *passphrase; 2959 uint8_t crcchk; 2960 2961 passphrase = __archive_read_next_passphrase(a); 2962 if (passphrase == NULL) { 2963 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2964 (retry > 0)? 2965 "Incorrect passphrase": 2966 "Passphrase required for this entry"); 2967 return (ARCHIVE_FAILED); 2968 } 2969 2970 /* 2971 * Initialize ctx for Traditional PKWARE Decryption. 2972 */ 2973 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 2974 p, ENC_HEADER_SIZE, &crcchk); 2975 if (r == 0 && crcchk == zip->entry->decdat) 2976 break;/* The passphrase is OK. */ 2977 if (retry > 10000) { 2978 /* Avoid infinity loop. */ 2979 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2980 "Too many incorrect passphrases"); 2981 return (ARCHIVE_FAILED); 2982 } 2983 } 2984 2985 __archive_read_consume(a, ENC_HEADER_SIZE); 2986 zip->tctx_valid = 1; 2987 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 2988 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 2989 } 2990 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 2991 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 2992 zip->decrypted_bytes_remaining = 0; 2993 2994 return (zip_alloc_decryption_buffer(a)); 2995 #undef ENC_HEADER_SIZE 2996 } 2997 2998 static int 2999 init_WinZip_AES_decryption(struct archive_read *a) 3000 { 3001 struct zip *zip = (struct zip *)(a->format->data); 3002 const void *p; 3003 const uint8_t *pv; 3004 size_t key_len, salt_len; 3005 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 3006 int retry; 3007 int r; 3008 3009 if (zip->cctx_valid || zip->hctx_valid) 3010 return (ARCHIVE_OK); 3011 3012 switch (zip->entry->aes_extra.strength) { 3013 case 1: salt_len = 8; key_len = 16; break; 3014 case 2: salt_len = 12; key_len = 24; break; 3015 case 3: salt_len = 16; key_len = 32; break; 3016 default: goto corrupted; 3017 } 3018 p = __archive_read_ahead(a, salt_len + 2, NULL); 3019 if (p == NULL) 3020 goto truncated; 3021 3022 for (retry = 0;; retry++) { 3023 const char *passphrase; 3024 3025 passphrase = __archive_read_next_passphrase(a); 3026 if (passphrase == NULL) { 3027 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3028 (retry > 0)? 3029 "Incorrect passphrase": 3030 "Passphrase required for this entry"); 3031 return (ARCHIVE_FAILED); 3032 } 3033 memset(derived_key, 0, sizeof(derived_key)); 3034 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 3035 p, salt_len, 1000, derived_key, key_len * 2 + 2); 3036 if (r != 0) { 3037 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3038 "Decryption is unsupported due to lack of " 3039 "crypto library"); 3040 return (ARCHIVE_FAILED); 3041 } 3042 3043 /* Check password verification value. */ 3044 pv = ((const uint8_t *)p) + salt_len; 3045 if (derived_key[key_len * 2] == pv[0] && 3046 derived_key[key_len * 2 + 1] == pv[1]) 3047 break;/* The passphrase is OK. */ 3048 if (retry > 10000) { 3049 /* Avoid infinity loop. */ 3050 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3051 "Too many incorrect passphrases"); 3052 return (ARCHIVE_FAILED); 3053 } 3054 } 3055 3056 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 3057 if (r != 0) { 3058 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3059 "Decryption is unsupported due to lack of crypto library"); 3060 return (ARCHIVE_FAILED); 3061 } 3062 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 3063 if (r != 0) { 3064 archive_decrypto_aes_ctr_release(&zip->cctx); 3065 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3066 "Failed to initialize HMAC-SHA1"); 3067 return (ARCHIVE_FAILED); 3068 } 3069 zip->cctx_valid = zip->hctx_valid = 1; 3070 __archive_read_consume(a, salt_len + 2); 3071 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 3072 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 3073 && zip->entry_bytes_remaining < 0) 3074 goto corrupted; 3075 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 3076 zip->decrypted_bytes_remaining = 0; 3077 3078 zip->entry->compression = zip->entry->aes_extra.compression; 3079 return (zip_alloc_decryption_buffer(a)); 3080 3081 truncated: 3082 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3083 "Truncated ZIP file data"); 3084 return (ARCHIVE_FATAL); 3085 corrupted: 3086 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3087 "Corrupted ZIP file data"); 3088 return (ARCHIVE_FATAL); 3089 } 3090 3091 static int 3092 archive_read_format_zip_read_data(struct archive_read *a, 3093 const void **buff, size_t *size, int64_t *offset) 3094 { 3095 int r; 3096 struct zip *zip = (struct zip *)(a->format->data); 3097 3098 if (zip->has_encrypted_entries == 3099 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 3100 zip->has_encrypted_entries = 0; 3101 } 3102 3103 *offset = zip->entry_uncompressed_bytes_read; 3104 *size = 0; 3105 *buff = NULL; 3106 3107 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 3108 if (zip->end_of_entry) 3109 return (ARCHIVE_EOF); 3110 3111 /* Return EOF immediately if this is a non-regular file. */ 3112 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 3113 return (ARCHIVE_EOF); 3114 3115 __archive_read_consume(a, zip->unconsumed); 3116 zip->unconsumed = 0; 3117 3118 if (zip->init_decryption) { 3119 zip->has_encrypted_entries = 1; 3120 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 3121 r = read_decryption_header(a); 3122 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 3123 r = init_WinZip_AES_decryption(a); 3124 else 3125 r = init_traditional_PKWARE_decryption(a); 3126 if (r != ARCHIVE_OK) 3127 return (r); 3128 zip->init_decryption = 0; 3129 } 3130 3131 switch(zip->entry->compression) { 3132 case 0: /* No compression. */ 3133 r = zip_read_data_none(a, buff, size, offset); 3134 break; 3135 #ifdef HAVE_BZLIB_H 3136 case 12: /* ZIPx bzip2 compression. */ 3137 r = zip_read_data_zipx_bzip2(a, buff, size, offset); 3138 break; 3139 #endif 3140 #if HAVE_LZMA_H && HAVE_LIBLZMA 3141 case 14: /* ZIPx LZMA compression. */ 3142 r = zip_read_data_zipx_lzma_alone(a, buff, size, offset); 3143 break; 3144 case 95: /* ZIPx XZ compression. */ 3145 r = zip_read_data_zipx_xz(a, buff, size, offset); 3146 break; 3147 #endif 3148 #if HAVE_ZSTD_H && HAVE_LIBZSTD 3149 case 93: /* ZIPx Zstd compression. */ 3150 r = zip_read_data_zipx_zstd(a, buff, size, offset); 3151 break; 3152 #endif 3153 /* PPMd support is built-in, so we don't need any #if guards. */ 3154 case 98: /* ZIPx PPMd compression. */ 3155 r = zip_read_data_zipx_ppmd(a, buff, size, offset); 3156 break; 3157 3158 #ifdef HAVE_ZLIB_H 3159 case 8: /* Deflate compression. */ 3160 r = zip_read_data_deflate(a, buff, size, offset); 3161 break; 3162 #endif 3163 default: /* Unsupported compression. */ 3164 /* Return a warning. */ 3165 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3166 "Unsupported ZIP compression method (%d: %s)", 3167 zip->entry->compression, compression_name(zip->entry->compression)); 3168 /* We can't decompress this entry, but we will 3169 * be able to skip() it and try the next entry. */ 3170 return (ARCHIVE_FAILED); 3171 break; 3172 } 3173 if (r != ARCHIVE_OK) 3174 return (r); 3175 if (*size > 0) { 3176 zip->computed_crc32 = zip->crc32func(zip->computed_crc32, *buff, 3177 (unsigned)*size); 3178 } 3179 /* If we hit the end, swallow any end-of-data marker and 3180 * verify the final check values. */ 3181 if (zip->end_of_entry) { 3182 consume_end_of_file_marker(a, zip); 3183 3184 /* Check computed CRC against header */ 3185 if ((!zip->hctx_valid || 3186 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 3187 zip->entry->crc32 != zip->computed_crc32 3188 && !zip->ignore_crc32) { 3189 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3190 "ZIP bad CRC: 0x%lx should be 0x%lx", 3191 (unsigned long)zip->computed_crc32, 3192 (unsigned long)zip->entry->crc32); 3193 return (ARCHIVE_FAILED); 3194 } 3195 /* Check file size against header. */ 3196 if (zip->entry->compressed_size != 3197 zip->entry_compressed_bytes_read) { 3198 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3199 "ZIP compressed data is wrong size " 3200 "(read %jd, expected %jd)", 3201 (intmax_t)zip->entry_compressed_bytes_read, 3202 (intmax_t)zip->entry->compressed_size); 3203 return (ARCHIVE_FAILED); 3204 } 3205 /* Size field only stores the lower 32 bits of the actual 3206 * size. */ 3207 if ((zip->entry->uncompressed_size & UINT32_MAX) 3208 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 3209 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3210 "ZIP uncompressed data is wrong size " 3211 "(read %jd, expected %jd)\n", 3212 (intmax_t)zip->entry_uncompressed_bytes_read, 3213 (intmax_t)zip->entry->uncompressed_size); 3214 return (ARCHIVE_FAILED); 3215 } 3216 } 3217 3218 return (ARCHIVE_OK); 3219 } 3220 3221 static int 3222 archive_read_format_zip_cleanup(struct archive_read *a) 3223 { 3224 struct zip *zip; 3225 struct zip_entry *zip_entry, *next_zip_entry; 3226 3227 zip = (struct zip *)(a->format->data); 3228 3229 #ifdef HAVE_ZLIB_H 3230 if (zip->stream_valid) 3231 inflateEnd(&zip->stream); 3232 #endif 3233 3234 #if HAVE_LZMA_H && HAVE_LIBLZMA 3235 if (zip->zipx_lzma_valid) { 3236 lzma_end(&zip->zipx_lzma_stream); 3237 } 3238 #endif 3239 3240 #ifdef HAVE_BZLIB_H 3241 if (zip->bzstream_valid) { 3242 BZ2_bzDecompressEnd(&zip->bzstream); 3243 } 3244 #endif 3245 3246 #if HAVE_ZSTD_H && HAVE_LIBZSTD 3247 if (zip->zstdstream_valid) { 3248 ZSTD_freeDStream(zip->zstdstream); 3249 } 3250 #endif 3251 3252 free(zip->uncompressed_buffer); 3253 3254 if (zip->ppmd8_valid) 3255 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 3256 3257 if (zip->zip_entries) { 3258 zip_entry = zip->zip_entries; 3259 while (zip_entry != NULL) { 3260 next_zip_entry = zip_entry->next; 3261 archive_string_free(&zip_entry->rsrcname); 3262 free(zip_entry); 3263 zip_entry = next_zip_entry; 3264 } 3265 } 3266 free(zip->decrypted_buffer); 3267 if (zip->cctx_valid) 3268 archive_decrypto_aes_ctr_release(&zip->cctx); 3269 if (zip->hctx_valid) 3270 archive_hmac_sha1_cleanup(&zip->hctx); 3271 free(zip->iv); 3272 free(zip->erd); 3273 free(zip->v_data); 3274 archive_string_free(&zip->format_name); 3275 free(zip); 3276 (a->format->data) = NULL; 3277 return (ARCHIVE_OK); 3278 } 3279 3280 static int 3281 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 3282 { 3283 if (_a && _a->format) { 3284 struct zip * zip = (struct zip *)_a->format->data; 3285 if (zip) { 3286 return zip->has_encrypted_entries; 3287 } 3288 } 3289 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3290 } 3291 3292 static int 3293 archive_read_format_zip_options(struct archive_read *a, 3294 const char *key, const char *val) 3295 { 3296 struct zip *zip; 3297 int ret = ARCHIVE_FAILED; 3298 3299 zip = (struct zip *)(a->format->data); 3300 if (strcmp(key, "compat-2x") == 0) { 3301 /* Handle filenames as libarchive 2.x */ 3302 zip->init_default_conversion = (val != NULL) ? 1 : 0; 3303 return (ARCHIVE_OK); 3304 } else if (strcmp(key, "hdrcharset") == 0) { 3305 if (val == NULL || val[0] == 0) 3306 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3307 "zip: hdrcharset option needs a character-set name" 3308 ); 3309 else { 3310 zip->sconv = archive_string_conversion_from_charset( 3311 &a->archive, val, 0); 3312 if (zip->sconv != NULL) { 3313 if (strcmp(val, "UTF-8") == 0) 3314 zip->sconv_utf8 = zip->sconv; 3315 ret = ARCHIVE_OK; 3316 } else 3317 ret = ARCHIVE_FATAL; 3318 } 3319 return (ret); 3320 } else if (strcmp(key, "ignorecrc32") == 0) { 3321 /* Mostly useful for testing. */ 3322 if (val == NULL || val[0] == 0) { 3323 zip->crc32func = real_crc32; 3324 zip->ignore_crc32 = 0; 3325 } else { 3326 zip->crc32func = fake_crc32; 3327 zip->ignore_crc32 = 1; 3328 } 3329 return (ARCHIVE_OK); 3330 } else if (strcmp(key, "mac-ext") == 0) { 3331 zip->process_mac_extensions = (val != NULL && val[0] != 0); 3332 return (ARCHIVE_OK); 3333 } 3334 3335 /* Note: The "warn" return is just to inform the options 3336 * supervisor that we didn't handle it. It will generate 3337 * a suitable error if no one used this option. */ 3338 return (ARCHIVE_WARN); 3339 } 3340 3341 int 3342 archive_read_support_format_zip(struct archive *a) 3343 { 3344 int r; 3345 r = archive_read_support_format_zip_streamable(a); 3346 if (r != ARCHIVE_OK) 3347 return r; 3348 return (archive_read_support_format_zip_seekable(a)); 3349 } 3350 3351 /* ------------------------------------------------------------------------ */ 3352 3353 /* 3354 * Streaming-mode support 3355 */ 3356 3357 3358 static int 3359 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 3360 { 3361 (void)a; /* UNUSED */ 3362 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 3363 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 3364 } 3365 3366 static int 3367 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 3368 { 3369 const char *p; 3370 3371 (void)best_bid; /* UNUSED */ 3372 3373 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 3374 return (-1); 3375 3376 /* 3377 * Bid of 29 here comes from: 3378 * + 16 bits for "PK", 3379 * + next 16-bit field has 6 options so contributes 3380 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 3381 * 3382 * So we've effectively verified ~29 total bits of check data. 3383 */ 3384 if (p[0] == 'P' && p[1] == 'K') { 3385 if ((p[2] == '\001' && p[3] == '\002') 3386 || (p[2] == '\003' && p[3] == '\004') 3387 || (p[2] == '\005' && p[3] == '\006') 3388 || (p[2] == '\006' && p[3] == '\006') 3389 || (p[2] == '\007' && p[3] == '\010') 3390 || (p[2] == '0' && p[3] == '0')) 3391 return (29); 3392 } 3393 3394 /* TODO: It's worth looking ahead a little bit for a valid 3395 * PK signature. In particular, that would make it possible 3396 * to read some UUEncoded SFX files or SFX files coming from 3397 * a network socket. */ 3398 3399 return (0); 3400 } 3401 3402 static int 3403 archive_read_format_zip_streamable_read_header(struct archive_read *a, 3404 struct archive_entry *entry) 3405 { 3406 struct zip *zip; 3407 3408 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 3409 if (a->archive.archive_format_name == NULL) 3410 a->archive.archive_format_name = "ZIP"; 3411 3412 zip = (struct zip *)(a->format->data); 3413 3414 /* 3415 * It should be sufficient to call archive_read_next_header() for 3416 * a reader to determine if an entry is encrypted or not. If the 3417 * encryption of an entry is only detectable when calling 3418 * archive_read_data(), so be it. We'll do the same check there 3419 * as well. 3420 */ 3421 if (zip->has_encrypted_entries == 3422 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 3423 zip->has_encrypted_entries = 0; 3424 3425 /* Make sure we have a zip_entry structure to use. */ 3426 if (zip->zip_entries == NULL) { 3427 zip->zip_entries = malloc(sizeof(struct zip_entry)); 3428 if (zip->zip_entries == NULL) { 3429 archive_set_error(&a->archive, ENOMEM, 3430 "Out of memory"); 3431 return ARCHIVE_FATAL; 3432 } 3433 } 3434 zip->entry = zip->zip_entries; 3435 memset(zip->entry, 0, sizeof(struct zip_entry)); 3436 3437 if (zip->cctx_valid) 3438 archive_decrypto_aes_ctr_release(&zip->cctx); 3439 if (zip->hctx_valid) 3440 archive_hmac_sha1_cleanup(&zip->hctx); 3441 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3442 __archive_read_reset_passphrase(a); 3443 3444 /* Search ahead for the next local file header. */ 3445 __archive_read_consume(a, zip->unconsumed); 3446 zip->unconsumed = 0; 3447 for (;;) { 3448 int64_t skipped = 0; 3449 const char *p, *end; 3450 ssize_t bytes; 3451 3452 p = __archive_read_ahead(a, 4, &bytes); 3453 if (p == NULL) 3454 return (ARCHIVE_FATAL); 3455 end = p + bytes; 3456 3457 while (p + 4 <= end) { 3458 if (p[0] == 'P' && p[1] == 'K') { 3459 if (p[2] == '\003' && p[3] == '\004') { 3460 /* Regular file entry. */ 3461 __archive_read_consume(a, skipped); 3462 return zip_read_local_file_header(a, 3463 entry, zip); 3464 } 3465 3466 /* 3467 * TODO: We cannot restore permissions 3468 * based only on the local file headers. 3469 * Consider scanning the central 3470 * directory and returning additional 3471 * entries for at least directories. 3472 * This would allow us to properly set 3473 * directory permissions. 3474 * 3475 * This won't help us fix symlinks 3476 * and may not help with regular file 3477 * permissions, either. <sigh> 3478 */ 3479 if (p[2] == '\001' && p[3] == '\002') { 3480 return (ARCHIVE_EOF); 3481 } 3482 3483 /* End of central directory? Must be an 3484 * empty archive. */ 3485 if ((p[2] == '\005' && p[3] == '\006') 3486 || (p[2] == '\006' && p[3] == '\006')) 3487 return (ARCHIVE_EOF); 3488 } 3489 ++p; 3490 ++skipped; 3491 } 3492 __archive_read_consume(a, skipped); 3493 } 3494 } 3495 3496 static int 3497 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 3498 { 3499 struct zip *zip; 3500 int64_t bytes_skipped; 3501 3502 zip = (struct zip *)(a->format->data); 3503 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 3504 zip->unconsumed = 0; 3505 if (bytes_skipped < 0) 3506 return (ARCHIVE_FATAL); 3507 3508 /* If we've already read to end of data, we're done. */ 3509 if (zip->end_of_entry) 3510 return (ARCHIVE_OK); 3511 3512 /* So we know we're streaming... */ 3513 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 3514 || zip->entry->compressed_size > 0) { 3515 /* We know the compressed length, so we can just skip. */ 3516 bytes_skipped = __archive_read_consume(a, 3517 zip->entry_bytes_remaining); 3518 if (bytes_skipped < 0) 3519 return (ARCHIVE_FATAL); 3520 return (ARCHIVE_OK); 3521 } 3522 3523 if (zip->init_decryption) { 3524 int r; 3525 3526 zip->has_encrypted_entries = 1; 3527 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 3528 r = read_decryption_header(a); 3529 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 3530 r = init_WinZip_AES_decryption(a); 3531 else 3532 r = init_traditional_PKWARE_decryption(a); 3533 if (r != ARCHIVE_OK) 3534 return (r); 3535 zip->init_decryption = 0; 3536 } 3537 3538 /* We're streaming and we don't know the length. */ 3539 /* If the body is compressed and we know the format, we can 3540 * find an exact end-of-entry by decompressing it. */ 3541 switch (zip->entry->compression) { 3542 #ifdef HAVE_ZLIB_H 3543 case 8: /* Deflate compression. */ 3544 while (!zip->end_of_entry) { 3545 int64_t offset = 0; 3546 const void *buff = NULL; 3547 size_t size = 0; 3548 int r; 3549 r = zip_read_data_deflate(a, &buff, &size, &offset); 3550 if (r != ARCHIVE_OK) 3551 return (r); 3552 } 3553 return ARCHIVE_OK; 3554 #endif 3555 default: /* Uncompressed or unknown. */ 3556 /* Scan for a PK\007\010 signature. */ 3557 for (;;) { 3558 const char *p, *buff; 3559 ssize_t bytes_avail; 3560 buff = __archive_read_ahead(a, 16, &bytes_avail); 3561 if (bytes_avail < 16) { 3562 archive_set_error(&a->archive, 3563 ARCHIVE_ERRNO_FILE_FORMAT, 3564 "Truncated ZIP file data"); 3565 return (ARCHIVE_FATAL); 3566 } 3567 p = buff; 3568 while (p <= buff + bytes_avail - 16) { 3569 if (p[3] == 'P') { p += 3; } 3570 else if (p[3] == 'K') { p += 2; } 3571 else if (p[3] == '\007') { p += 1; } 3572 else if (p[3] == '\010' && p[2] == '\007' 3573 && p[1] == 'K' && p[0] == 'P') { 3574 if (zip->entry->flags & LA_USED_ZIP64) 3575 __archive_read_consume(a, 3576 p - buff + 24); 3577 else 3578 __archive_read_consume(a, 3579 p - buff + 16); 3580 return ARCHIVE_OK; 3581 } else { p += 4; } 3582 } 3583 __archive_read_consume(a, p - buff); 3584 } 3585 } 3586 } 3587 3588 int 3589 archive_read_support_format_zip_streamable(struct archive *_a) 3590 { 3591 struct archive_read *a = (struct archive_read *)_a; 3592 struct zip *zip; 3593 int r; 3594 3595 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3596 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 3597 3598 zip = calloc(1, sizeof(*zip)); 3599 if (zip == NULL) { 3600 archive_set_error(&a->archive, ENOMEM, 3601 "Can't allocate zip data"); 3602 return (ARCHIVE_FATAL); 3603 } 3604 3605 /* Streamable reader doesn't support mac extensions. */ 3606 zip->process_mac_extensions = 0; 3607 3608 /* 3609 * Until enough data has been read, we cannot tell about 3610 * any encrypted entries yet. 3611 */ 3612 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3613 zip->crc32func = real_crc32; 3614 3615 r = __archive_read_register_format(a, 3616 zip, 3617 "zip", 3618 archive_read_format_zip_streamable_bid, 3619 archive_read_format_zip_options, 3620 archive_read_format_zip_streamable_read_header, 3621 archive_read_format_zip_read_data, 3622 archive_read_format_zip_read_data_skip_streamable, 3623 NULL, 3624 archive_read_format_zip_cleanup, 3625 archive_read_support_format_zip_capabilities_streamable, 3626 archive_read_format_zip_has_encrypted_entries); 3627 3628 if (r != ARCHIVE_OK) 3629 free(zip); 3630 return (ARCHIVE_OK); 3631 } 3632 3633 /* ------------------------------------------------------------------------ */ 3634 3635 /* 3636 * Seeking-mode support 3637 */ 3638 3639 static int 3640 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 3641 { 3642 (void)a; /* UNUSED */ 3643 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 3644 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 3645 } 3646 3647 /* 3648 * TODO: This is a performance sink because it forces the read core to 3649 * drop buffered data from the start of file, which will then have to 3650 * be re-read again if this bidder loses. 3651 * 3652 * We workaround this a little by passing in the best bid so far so 3653 * that later bidders can do nothing if they know they'll never 3654 * outbid. But we can certainly do better... 3655 */ 3656 static int 3657 read_eocd(struct zip *zip, const char *p, int64_t current_offset) 3658 { 3659 uint16_t disk_num; 3660 uint32_t cd_size, cd_offset; 3661 3662 disk_num = archive_le16dec(p + 4); 3663 cd_size = archive_le32dec(p + 12); 3664 cd_offset = archive_le32dec(p + 16); 3665 3666 /* Sanity-check the EOCD we've found. */ 3667 3668 /* This must be the first volume. */ 3669 if (disk_num != 0) 3670 return 0; 3671 /* Central directory must be on this volume. */ 3672 if (disk_num != archive_le16dec(p + 6)) 3673 return 0; 3674 /* All central directory entries must be on this volume. */ 3675 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 3676 return 0; 3677 /* Central directory can't extend beyond start of EOCD record. */ 3678 if ((int64_t)cd_offset + cd_size > current_offset) 3679 return 0; 3680 3681 /* Save the central directory location for later use. */ 3682 zip->central_directory_offset = cd_offset; 3683 zip->central_directory_offset_adjusted = current_offset - cd_size; 3684 3685 /* This is just a tiny bit higher than the maximum 3686 returned by the streaming Zip bidder. This ensures 3687 that the more accurate seeking Zip parser wins 3688 whenever seek is available. */ 3689 return 32; 3690 } 3691 3692 /* 3693 * Examine Zip64 EOCD locator: If it's valid, store the information 3694 * from it. 3695 */ 3696 static int 3697 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 3698 { 3699 int64_t eocd64_offset; 3700 int64_t eocd64_size; 3701 3702 /* Sanity-check the locator record. */ 3703 3704 /* Central dir must be on first volume. */ 3705 if (archive_le32dec(p + 4) != 0) 3706 return 0; 3707 /* Must be only a single volume. */ 3708 if (archive_le32dec(p + 16) != 1) 3709 return 0; 3710 3711 /* Find the Zip64 EOCD record. */ 3712 eocd64_offset = archive_le64dec(p + 8); 3713 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 3714 return 0; 3715 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 3716 return 0; 3717 /* Make sure we can read all of it. */ 3718 eocd64_size = archive_le64dec(p + 4) + 12; 3719 if (eocd64_size < 56 || eocd64_size > 16384) 3720 return 0; 3721 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 3722 return 0; 3723 3724 /* Sanity-check the EOCD64 */ 3725 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 3726 return 0; 3727 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 3728 return 0; 3729 /* CD can't be split. */ 3730 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 3731 return 0; 3732 3733 /* Save the central directory offset for later use. */ 3734 zip->central_directory_offset = archive_le64dec(p + 48); 3735 /* TODO: Needs scanning backwards to find the eocd64 instead of assuming */ 3736 zip->central_directory_offset_adjusted = zip->central_directory_offset; 3737 3738 return 32; 3739 } 3740 3741 static int 3742 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 3743 { 3744 struct zip *zip = (struct zip *)a->format->data; 3745 int64_t file_size, current_offset; 3746 const char *p; 3747 int i, tail; 3748 3749 /* If someone has already bid more than 32, then avoid 3750 trashing the look-ahead buffers with a seek. */ 3751 if (best_bid > 32) 3752 return (-1); 3753 3754 file_size = __archive_read_seek(a, 0, SEEK_END); 3755 if (file_size <= 0) 3756 return 0; 3757 3758 /* Search last 16k of file for end-of-central-directory 3759 * record (which starts with PK\005\006) */ 3760 tail = (int)zipmin(1024 * 16, file_size); 3761 current_offset = __archive_read_seek(a, -tail, SEEK_END); 3762 if (current_offset < 0) 3763 return 0; 3764 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 3765 return 0; 3766 /* Boyer-Moore search backwards from the end, since we want 3767 * to match the last EOCD in the file (there can be more than 3768 * one if there is an uncompressed Zip archive as a member 3769 * within this Zip archive). */ 3770 for (i = tail - 22; i > 0;) { 3771 switch (p[i]) { 3772 case 'P': 3773 if (memcmp(p + i, "PK\005\006", 4) == 0) { 3774 int ret = read_eocd(zip, p + i, 3775 current_offset + i); 3776 /* Zip64 EOCD locator precedes 3777 * regular EOCD if present. */ 3778 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 3779 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); 3780 if (ret_zip64 > ret) 3781 ret = ret_zip64; 3782 } 3783 return (ret); 3784 } 3785 i -= 4; 3786 break; 3787 case 'K': i -= 1; break; 3788 case 005: i -= 2; break; 3789 case 006: i -= 3; break; 3790 default: i -= 4; break; 3791 } 3792 } 3793 return 0; 3794 } 3795 3796 /* The red-black trees are only used in seeking mode to manage 3797 * the in-memory copy of the central directory. */ 3798 3799 static int 3800 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 3801 { 3802 const struct zip_entry *e1 = (const struct zip_entry *)n1; 3803 const struct zip_entry *e2 = (const struct zip_entry *)n2; 3804 3805 if (e1->local_header_offset > e2->local_header_offset) 3806 return -1; 3807 if (e1->local_header_offset < e2->local_header_offset) 3808 return 1; 3809 return 0; 3810 } 3811 3812 static int 3813 cmp_key(const struct archive_rb_node *n, const void *key) 3814 { 3815 /* This function won't be called */ 3816 (void)n; /* UNUSED */ 3817 (void)key; /* UNUSED */ 3818 return 1; 3819 } 3820 3821 static const struct archive_rb_tree_ops rb_ops = { 3822 &cmp_node, &cmp_key 3823 }; 3824 3825 static int 3826 rsrc_cmp_node(const struct archive_rb_node *n1, 3827 const struct archive_rb_node *n2) 3828 { 3829 const struct zip_entry *e1 = (const struct zip_entry *)n1; 3830 const struct zip_entry *e2 = (const struct zip_entry *)n2; 3831 3832 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 3833 } 3834 3835 static int 3836 rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 3837 { 3838 const struct zip_entry *e = (const struct zip_entry *)n; 3839 return (strcmp((const char *)key, e->rsrcname.s)); 3840 } 3841 3842 static const struct archive_rb_tree_ops rb_rsrc_ops = { 3843 &rsrc_cmp_node, &rsrc_cmp_key 3844 }; 3845 3846 static const char * 3847 rsrc_basename(const char *name, size_t name_length) 3848 { 3849 const char *s, *r; 3850 3851 r = s = name; 3852 for (;;) { 3853 s = memchr(s, '/', name_length - (s - name)); 3854 if (s == NULL) 3855 break; 3856 r = ++s; 3857 } 3858 return (r); 3859 } 3860 3861 static void 3862 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 3863 { 3864 struct archive_string str; 3865 struct zip_entry *dir; 3866 char *s; 3867 3868 archive_string_init(&str); 3869 archive_strncpy(&str, name, name_length); 3870 for (;;) { 3871 s = strrchr(str.s, '/'); 3872 if (s == NULL) 3873 break; 3874 *s = '\0'; 3875 /* Transfer the parent directory from zip->tree_rsrc RB 3876 * tree to zip->tree RB tree to expose. */ 3877 dir = (struct zip_entry *) 3878 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 3879 if (dir == NULL) 3880 break; 3881 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 3882 archive_string_free(&dir->rsrcname); 3883 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 3884 } 3885 archive_string_free(&str); 3886 } 3887 3888 static int 3889 slurp_central_directory(struct archive_read *a, struct archive_entry* entry, 3890 struct zip *zip) 3891 { 3892 ssize_t i; 3893 unsigned found; 3894 int64_t correction; 3895 ssize_t bytes_avail; 3896 const char *p; 3897 3898 /* 3899 * Find the start of the central directory. The end-of-CD 3900 * record has our starting point, but there are lots of 3901 * Zip archives which have had other data prepended to the 3902 * file, which makes the recorded offsets all too small. 3903 * So we search forward from the specified offset until we 3904 * find the real start of the central directory. Then we 3905 * know the correction we need to apply to account for leading 3906 * padding. 3907 */ 3908 if (__archive_read_seek(a, zip->central_directory_offset_adjusted, SEEK_SET) 3909 < 0) 3910 return ARCHIVE_FATAL; 3911 3912 found = 0; 3913 while (!found) { 3914 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 3915 return ARCHIVE_FATAL; 3916 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 3917 switch (p[i + 3]) { 3918 case 'P': i += 3; break; 3919 case 'K': i += 2; break; 3920 case 001: i += 1; break; 3921 case 002: 3922 if (memcmp(p + i, "PK\001\002", 4) == 0) { 3923 p += i; 3924 found = 1; 3925 } else 3926 i += 4; 3927 break; 3928 case 005: i += 1; break; 3929 case 006: 3930 if (memcmp(p + i, "PK\005\006", 4) == 0) { 3931 p += i; 3932 found = 1; 3933 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 3934 p += i; 3935 found = 1; 3936 } else 3937 i += 1; 3938 break; 3939 default: i += 4; break; 3940 } 3941 } 3942 __archive_read_consume(a, i); 3943 } 3944 correction = archive_filter_bytes(&a->archive, 0) 3945 - zip->central_directory_offset; 3946 3947 __archive_rb_tree_init(&zip->tree, &rb_ops); 3948 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 3949 3950 zip->central_directory_entries_total = 0; 3951 while (1) { 3952 struct zip_entry *zip_entry; 3953 size_t filename_length, extra_length, comment_length; 3954 uint32_t external_attributes; 3955 const char *name, *r; 3956 3957 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 3958 return ARCHIVE_FATAL; 3959 if (memcmp(p, "PK\006\006", 4) == 0 3960 || memcmp(p, "PK\005\006", 4) == 0) { 3961 break; 3962 } else if (memcmp(p, "PK\001\002", 4) != 0) { 3963 archive_set_error(&a->archive, 3964 -1, "Invalid central directory signature"); 3965 return ARCHIVE_FATAL; 3966 } 3967 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 3968 return ARCHIVE_FATAL; 3969 3970 zip_entry = calloc(1, sizeof(struct zip_entry)); 3971 if (zip_entry == NULL) { 3972 archive_set_error(&a->archive, ENOMEM, 3973 "Can't allocate zip entry"); 3974 return ARCHIVE_FATAL; 3975 } 3976 zip_entry->next = zip->zip_entries; 3977 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 3978 zip->zip_entries = zip_entry; 3979 zip->central_directory_entries_total++; 3980 3981 /* version = p[4]; */ 3982 zip_entry->system = p[5]; 3983 /* version_required = archive_le16dec(p + 6); */ 3984 zip_entry->zip_flags = archive_le16dec(p + 8); 3985 if (zip_entry->zip_flags 3986 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 3987 zip->has_encrypted_entries = 1; 3988 } 3989 zip_entry->compression = (char)archive_le16dec(p + 10); 3990 zip_entry->mtime = zip_time(p + 12); 3991 zip_entry->crc32 = archive_le32dec(p + 16); 3992 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 3993 zip_entry->decdat = p[13]; 3994 else 3995 zip_entry->decdat = p[19]; 3996 zip_entry->compressed_size = archive_le32dec(p + 20); 3997 zip_entry->uncompressed_size = archive_le32dec(p + 24); 3998 filename_length = archive_le16dec(p + 28); 3999 extra_length = archive_le16dec(p + 30); 4000 comment_length = archive_le16dec(p + 32); 4001 /* disk_start = archive_le16dec(p + 34); 4002 * Better be zero. 4003 * internal_attributes = archive_le16dec(p + 36); 4004 * text bit */ 4005 external_attributes = archive_le32dec(p + 38); 4006 zip_entry->local_header_offset = 4007 archive_le32dec(p + 42) + correction; 4008 4009 /* If we can't guess the mode, leave it zero here; 4010 when we read the local file header we might get 4011 more information. */ 4012 if (zip_entry->system == 3) { 4013 zip_entry->mode = external_attributes >> 16; 4014 } else if (zip_entry->system == 0) { 4015 // Interpret MSDOS directory bit 4016 if (0x10 == (external_attributes & 0x10)) { 4017 zip_entry->mode = AE_IFDIR | 0775; 4018 } else { 4019 zip_entry->mode = AE_IFREG | 0664; 4020 } 4021 if (0x01 == (external_attributes & 0x01)) { 4022 // Read-only bit; strip write permissions 4023 zip_entry->mode &= 0555; 4024 } 4025 } else { 4026 zip_entry->mode = 0; 4027 } 4028 4029 /* We're done with the regular data; get the filename and 4030 * extra data. */ 4031 __archive_read_consume(a, 46); 4032 p = __archive_read_ahead(a, filename_length + extra_length, 4033 NULL); 4034 if (p == NULL) { 4035 archive_set_error(&a->archive, 4036 ARCHIVE_ERRNO_FILE_FORMAT, 4037 "Truncated ZIP file header"); 4038 return ARCHIVE_FATAL; 4039 } 4040 if (ARCHIVE_OK != process_extra(a, entry, p + filename_length, 4041 extra_length, zip_entry)) { 4042 return ARCHIVE_FATAL; 4043 } 4044 4045 /* 4046 * Mac resource fork files are stored under the 4047 * "__MACOSX/" directory, so we should check if 4048 * it is. 4049 */ 4050 if (!zip->process_mac_extensions) { 4051 /* Treat every entry as a regular entry. */ 4052 __archive_rb_tree_insert_node(&zip->tree, 4053 &zip_entry->node); 4054 } else { 4055 name = p; 4056 r = rsrc_basename(name, filename_length); 4057 if (filename_length >= 9 && 4058 strncmp("__MACOSX/", name, 9) == 0) { 4059 /* If this file is not a resource fork nor 4060 * a directory. We should treat it as a non 4061 * resource fork file to expose it. */ 4062 if (name[filename_length-1] != '/' && 4063 (r - name < 3 || r[0] != '.' || 4064 r[1] != '_')) { 4065 __archive_rb_tree_insert_node( 4066 &zip->tree, &zip_entry->node); 4067 /* Expose its parent directories. */ 4068 expose_parent_dirs(zip, name, 4069 filename_length); 4070 } else { 4071 /* This file is a resource fork file or 4072 * a directory. */ 4073 archive_strncpy(&(zip_entry->rsrcname), 4074 name, filename_length); 4075 __archive_rb_tree_insert_node( 4076 &zip->tree_rsrc, &zip_entry->node); 4077 } 4078 } else { 4079 /* Generate resource fork name to find its 4080 * resource file at zip->tree_rsrc. */ 4081 4082 /* If this is an entry ending with slash, 4083 * make the resource for name slash-less 4084 * as the actual resource fork doesn't end with '/'. 4085 */ 4086 size_t tmp_length = filename_length; 4087 if (tmp_length > 0 && name[tmp_length - 1] == '/') { 4088 tmp_length--; 4089 r = rsrc_basename(name, tmp_length); 4090 } 4091 4092 archive_strcpy(&(zip_entry->rsrcname), 4093 "__MACOSX/"); 4094 archive_strncat(&(zip_entry->rsrcname), 4095 name, r - name); 4096 archive_strcat(&(zip_entry->rsrcname), "._"); 4097 archive_strncat(&(zip_entry->rsrcname), 4098 name + (r - name), 4099 tmp_length - (r - name)); 4100 /* Register an entry to RB tree to sort it by 4101 * file offset. */ 4102 __archive_rb_tree_insert_node(&zip->tree, 4103 &zip_entry->node); 4104 } 4105 } 4106 4107 /* Skip the comment too ... */ 4108 __archive_read_consume(a, 4109 filename_length + extra_length + comment_length); 4110 } 4111 4112 return ARCHIVE_OK; 4113 } 4114 4115 static ssize_t 4116 zip_get_local_file_header_size(struct archive_read *a, size_t extra) 4117 { 4118 const char *p; 4119 ssize_t filename_length, extra_length; 4120 4121 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 4122 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4123 "Truncated ZIP file header"); 4124 return (ARCHIVE_WARN); 4125 } 4126 p += extra; 4127 4128 if (memcmp(p, "PK\003\004", 4) != 0) { 4129 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 4130 return ARCHIVE_WARN; 4131 } 4132 filename_length = archive_le16dec(p + 26); 4133 extra_length = archive_le16dec(p + 28); 4134 4135 return (30 + filename_length + extra_length); 4136 } 4137 4138 static int 4139 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 4140 struct zip_entry *rsrc) 4141 { 4142 struct zip *zip = (struct zip *)a->format->data; 4143 unsigned char *metadata, *mp; 4144 int64_t offset = archive_filter_bytes(&a->archive, 0); 4145 size_t remaining_bytes, metadata_bytes; 4146 ssize_t hsize; 4147 int ret = ARCHIVE_OK, eof; 4148 4149 switch(rsrc->compression) { 4150 case 0: /* No compression. */ 4151 if (rsrc->uncompressed_size != rsrc->compressed_size) { 4152 archive_set_error(&a->archive, 4153 ARCHIVE_ERRNO_FILE_FORMAT, 4154 "Malformed OS X metadata entry: " 4155 "inconsistent size"); 4156 return (ARCHIVE_FATAL); 4157 } 4158 #ifdef HAVE_ZLIB_H 4159 case 8: /* Deflate compression. */ 4160 #endif 4161 break; 4162 default: /* Unsupported compression. */ 4163 /* Return a warning. */ 4164 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4165 "Unsupported ZIP compression method (%s)", 4166 compression_name(rsrc->compression)); 4167 /* We can't decompress this entry, but we will 4168 * be able to skip() it and try the next entry. */ 4169 return (ARCHIVE_WARN); 4170 } 4171 4172 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 4173 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4174 "Mac metadata is too large: %jd > 4M bytes", 4175 (intmax_t)rsrc->uncompressed_size); 4176 return (ARCHIVE_WARN); 4177 } 4178 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 4179 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4180 "Mac metadata is too large: %jd > 4M bytes", 4181 (intmax_t)rsrc->compressed_size); 4182 return (ARCHIVE_WARN); 4183 } 4184 4185 metadata = malloc((size_t)rsrc->uncompressed_size); 4186 if (metadata == NULL) { 4187 archive_set_error(&a->archive, ENOMEM, 4188 "Can't allocate memory for Mac metadata"); 4189 return (ARCHIVE_FATAL); 4190 } 4191 4192 if (offset < rsrc->local_header_offset) 4193 __archive_read_consume(a, rsrc->local_header_offset - offset); 4194 else if (offset != rsrc->local_header_offset) { 4195 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 4196 } 4197 4198 hsize = zip_get_local_file_header_size(a, 0); 4199 __archive_read_consume(a, hsize); 4200 4201 remaining_bytes = (size_t)rsrc->compressed_size; 4202 metadata_bytes = (size_t)rsrc->uncompressed_size; 4203 mp = metadata; 4204 eof = 0; 4205 while (!eof && remaining_bytes) { 4206 const unsigned char *p; 4207 ssize_t bytes_avail; 4208 size_t bytes_used; 4209 4210 p = __archive_read_ahead(a, 1, &bytes_avail); 4211 if (p == NULL) { 4212 archive_set_error(&a->archive, 4213 ARCHIVE_ERRNO_FILE_FORMAT, 4214 "Truncated ZIP file header"); 4215 ret = ARCHIVE_WARN; 4216 goto exit_mac_metadata; 4217 } 4218 if ((size_t)bytes_avail > remaining_bytes) 4219 bytes_avail = remaining_bytes; 4220 switch(rsrc->compression) { 4221 case 0: /* No compression. */ 4222 if ((size_t)bytes_avail > metadata_bytes) 4223 bytes_avail = metadata_bytes; 4224 memcpy(mp, p, bytes_avail); 4225 bytes_used = (size_t)bytes_avail; 4226 metadata_bytes -= bytes_used; 4227 mp += bytes_used; 4228 if (metadata_bytes == 0) 4229 eof = 1; 4230 break; 4231 #ifdef HAVE_ZLIB_H 4232 case 8: /* Deflate compression. */ 4233 { 4234 int r; 4235 4236 ret = zip_deflate_init(a, zip); 4237 if (ret != ARCHIVE_OK) 4238 goto exit_mac_metadata; 4239 zip->stream.next_in = 4240 (Bytef *)(uintptr_t)(const void *)p; 4241 zip->stream.avail_in = (uInt)bytes_avail; 4242 zip->stream.total_in = 0; 4243 zip->stream.next_out = mp; 4244 zip->stream.avail_out = (uInt)metadata_bytes; 4245 zip->stream.total_out = 0; 4246 4247 r = inflate(&zip->stream, 0); 4248 switch (r) { 4249 case Z_OK: 4250 break; 4251 case Z_STREAM_END: 4252 eof = 1; 4253 break; 4254 case Z_MEM_ERROR: 4255 archive_set_error(&a->archive, ENOMEM, 4256 "Out of memory for ZIP decompression"); 4257 ret = ARCHIVE_FATAL; 4258 goto exit_mac_metadata; 4259 default: 4260 archive_set_error(&a->archive, 4261 ARCHIVE_ERRNO_MISC, 4262 "ZIP decompression failed (%d)", r); 4263 ret = ARCHIVE_FATAL; 4264 goto exit_mac_metadata; 4265 } 4266 bytes_used = zip->stream.total_in; 4267 metadata_bytes -= zip->stream.total_out; 4268 mp += zip->stream.total_out; 4269 break; 4270 } 4271 #endif 4272 default: 4273 bytes_used = 0; 4274 break; 4275 } 4276 __archive_read_consume(a, bytes_used); 4277 remaining_bytes -= bytes_used; 4278 } 4279 archive_entry_copy_mac_metadata(entry, metadata, 4280 (size_t)rsrc->uncompressed_size - metadata_bytes); 4281 4282 exit_mac_metadata: 4283 __archive_read_seek(a, offset, SEEK_SET); 4284 zip->decompress_init = 0; 4285 free(metadata); 4286 return (ret); 4287 } 4288 4289 static int 4290 archive_read_format_zip_seekable_read_header(struct archive_read *a, 4291 struct archive_entry *entry) 4292 { 4293 struct zip *zip = (struct zip *)a->format->data; 4294 struct zip_entry *rsrc; 4295 int64_t offset; 4296 int r, ret = ARCHIVE_OK; 4297 4298 /* 4299 * It should be sufficient to call archive_read_next_header() for 4300 * a reader to determine if an entry is encrypted or not. If the 4301 * encryption of an entry is only detectable when calling 4302 * archive_read_data(), so be it. We'll do the same check there 4303 * as well. 4304 */ 4305 if (zip->has_encrypted_entries == 4306 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 4307 zip->has_encrypted_entries = 0; 4308 4309 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 4310 if (a->archive.archive_format_name == NULL) 4311 a->archive.archive_format_name = "ZIP"; 4312 4313 if (zip->zip_entries == NULL) { 4314 r = slurp_central_directory(a, entry, zip); 4315 if (r != ARCHIVE_OK) 4316 return r; 4317 /* Get first entry whose local header offset is lower than 4318 * other entries in the archive file. */ 4319 zip->entry = 4320 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 4321 } else if (zip->entry != NULL) { 4322 /* Get next entry in local header offset order. */ 4323 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 4324 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 4325 } 4326 4327 if (zip->entry == NULL) 4328 return ARCHIVE_EOF; 4329 4330 if (zip->entry->rsrcname.s) 4331 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 4332 &zip->tree_rsrc, zip->entry->rsrcname.s); 4333 else 4334 rsrc = NULL; 4335 4336 if (zip->cctx_valid) 4337 archive_decrypto_aes_ctr_release(&zip->cctx); 4338 if (zip->hctx_valid) 4339 archive_hmac_sha1_cleanup(&zip->hctx); 4340 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 4341 __archive_read_reset_passphrase(a); 4342 4343 /* File entries are sorted by the header offset, we should mostly 4344 * use __archive_read_consume to advance a read point to avoid 4345 * redundant data reading. */ 4346 offset = archive_filter_bytes(&a->archive, 0); 4347 if (offset < zip->entry->local_header_offset) 4348 __archive_read_consume(a, 4349 zip->entry->local_header_offset - offset); 4350 else if (offset != zip->entry->local_header_offset) { 4351 __archive_read_seek(a, zip->entry->local_header_offset, 4352 SEEK_SET); 4353 } 4354 zip->unconsumed = 0; 4355 r = zip_read_local_file_header(a, entry, zip); 4356 if (r != ARCHIVE_OK) 4357 return r; 4358 if (rsrc) { 4359 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 4360 if (ret2 < ret) 4361 ret = ret2; 4362 } 4363 return (ret); 4364 } 4365 4366 /* 4367 * We're going to seek for the next header anyway, so we don't 4368 * need to bother doing anything here. 4369 */ 4370 static int 4371 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 4372 { 4373 struct zip *zip; 4374 zip = (struct zip *)(a->format->data); 4375 4376 zip->unconsumed = 0; 4377 return (ARCHIVE_OK); 4378 } 4379 4380 int 4381 archive_read_support_format_zip_seekable(struct archive *_a) 4382 { 4383 struct archive_read *a = (struct archive_read *)_a; 4384 struct zip *zip; 4385 int r; 4386 4387 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 4388 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 4389 4390 zip = calloc(1, sizeof(*zip)); 4391 if (zip == NULL) { 4392 archive_set_error(&a->archive, ENOMEM, 4393 "Can't allocate zip data"); 4394 return (ARCHIVE_FATAL); 4395 } 4396 4397 #ifdef HAVE_COPYFILE_H 4398 /* Set this by default on Mac OS. */ 4399 zip->process_mac_extensions = 1; 4400 #endif 4401 4402 /* 4403 * Until enough data has been read, we cannot tell about 4404 * any encrypted entries yet. 4405 */ 4406 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 4407 zip->crc32func = real_crc32; 4408 4409 r = __archive_read_register_format(a, 4410 zip, 4411 "zip", 4412 archive_read_format_zip_seekable_bid, 4413 archive_read_format_zip_options, 4414 archive_read_format_zip_seekable_read_header, 4415 archive_read_format_zip_read_data, 4416 archive_read_format_zip_read_data_skip_seekable, 4417 NULL, 4418 archive_read_format_zip_cleanup, 4419 archive_read_support_format_zip_capabilities_seekable, 4420 archive_read_format_zip_has_encrypted_entries); 4421 4422 if (r != ARCHIVE_OK) 4423 free(zip); 4424 return (ARCHIVE_OK); 4425 } 4426 4427 /*# vim:set noet:*/ 4428