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 __FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_zip.c 201102 2009-12-28 03:11:36Z kientzle $"); 30 31 /* 32 * The definitive documentation of the Zip file format is: 33 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT 34 * 35 * The Info-Zip project has pioneered various extensions to better 36 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 37 * "Ux", and 0x7875 "ux" extensions for time and ownership 38 * information. 39 * 40 * History of this code: The streaming Zip reader was first added to 41 * libarchive in January 2005. Support for seekable input sources was 42 * added in Nov 2011. Zip64 support (including a significant code 43 * refactoring) was added in 2014. 44 */ 45 46 #ifdef HAVE_ERRNO_H 47 #include <errno.h> 48 #endif 49 #ifdef HAVE_STDLIB_H 50 #include <stdlib.h> 51 #endif 52 #ifdef HAVE_ZLIB_H 53 #include <zlib.h> 54 #endif 55 #ifdef HAVE_BZLIB_H 56 #include <bzlib.h> 57 #endif 58 #ifdef HAVE_LZMA_H 59 #include <lzma.h> 60 #endif 61 #ifdef HAVE_ZSTD_H 62 #include <zstd.h> 63 #endif 64 65 #include "archive.h" 66 #include "archive_digest_private.h" 67 #include "archive_cryptor_private.h" 68 #include "archive_endian.h" 69 #include "archive_entry.h" 70 #include "archive_entry_locale.h" 71 #include "archive_hmac_private.h" 72 #include "archive_private.h" 73 #include "archive_rb.h" 74 #include "archive_read_private.h" 75 #include "archive_ppmd8_private.h" 76 77 #ifndef HAVE_ZLIB_H 78 #include "archive_crc32.h" 79 #endif 80 81 struct zip_entry { 82 struct archive_rb_node node; 83 struct zip_entry *next; 84 int64_t local_header_offset; 85 int64_t compressed_size; 86 int64_t uncompressed_size; 87 int64_t gid; 88 int64_t uid; 89 struct archive_string rsrcname; 90 time_t mtime; 91 time_t atime; 92 time_t ctime; 93 uint32_t crc32; 94 uint16_t mode; 95 uint16_t zip_flags; /* From GP Flags Field */ 96 unsigned char compression; 97 unsigned char system; /* From "version written by" */ 98 unsigned char flags; /* Our extra markers. */ 99 unsigned char decdat;/* Used for Decryption check */ 100 101 /* WinZip AES encryption extra field should be available 102 * when compression is 99. */ 103 struct { 104 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 105 unsigned vendor; 106 #define AES_VENDOR_AE_1 0x0001 107 #define AES_VENDOR_AE_2 0x0002 108 /* AES encryption strength: 109 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 110 unsigned strength; 111 /* Actual compression method. */ 112 unsigned char compression; 113 } aes_extra; 114 }; 115 116 struct trad_enc_ctx { 117 uint32_t keys[3]; 118 }; 119 120 /* Bits used in zip_flags. */ 121 #define ZIP_ENCRYPTED (1 << 0) 122 #define ZIP_LENGTH_AT_END (1 << 3) 123 #define ZIP_STRONG_ENCRYPTED (1 << 6) 124 #define ZIP_UTF8_NAME (1 << 11) 125 /* See "7.2 Single Password Symmetric Encryption Method" 126 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 127 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 128 129 /* Bits used in flags. */ 130 #define LA_USED_ZIP64 (1 << 0) 131 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 132 133 /* 134 * See "WinZip - AES Encryption Information" 135 * http://www.winzip.com/aes_info.htm 136 */ 137 /* Value used in compression method. */ 138 #define WINZIP_AES_ENCRYPTION 99 139 /* Authentication code size. */ 140 #define AUTH_CODE_SIZE 10 141 /**/ 142 #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 143 144 struct zip { 145 /* Structural information about the archive. */ 146 struct archive_string format_name; 147 int64_t central_directory_offset; 148 int64_t central_directory_offset_adjusted; 149 size_t central_directory_entries_total; 150 size_t central_directory_entries_on_this_disk; 151 int has_encrypted_entries; 152 153 /* List of entries (seekable Zip only) */ 154 struct zip_entry *zip_entries; 155 struct archive_rb_tree tree; 156 struct archive_rb_tree tree_rsrc; 157 158 /* Bytes read but not yet consumed via __archive_read_consume() */ 159 size_t unconsumed; 160 161 /* Information about entry we're currently reading. */ 162 struct zip_entry *entry; 163 int64_t entry_bytes_remaining; 164 165 /* These count the number of bytes actually read for the entry. */ 166 int64_t entry_compressed_bytes_read; 167 int64_t entry_uncompressed_bytes_read; 168 169 /* Running CRC32 of the decompressed data */ 170 unsigned long entry_crc32; 171 unsigned long (*crc32func)(unsigned long, const void *, 172 size_t); 173 char ignore_crc32; 174 175 /* Flags to mark progress of decompression. */ 176 char decompress_init; 177 char end_of_entry; 178 179 unsigned char *uncompressed_buffer; 180 size_t uncompressed_buffer_size; 181 182 #ifdef HAVE_ZLIB_H 183 z_stream stream; 184 char stream_valid; 185 #endif 186 187 #if HAVE_LZMA_H && HAVE_LIBLZMA 188 lzma_stream zipx_lzma_stream; 189 char zipx_lzma_valid; 190 #endif 191 192 #ifdef HAVE_BZLIB_H 193 bz_stream bzstream; 194 char bzstream_valid; 195 #endif 196 197 #if HAVE_ZSTD_H && HAVE_LIBZSTD 198 ZSTD_DStream *zstdstream; 199 char zstdstream_valid; 200 #endif 201 202 IByteIn zipx_ppmd_stream; 203 ssize_t zipx_ppmd_read_compressed; 204 CPpmd8 ppmd8; 205 char ppmd8_valid; 206 char ppmd8_stream_failed; 207 208 struct archive_string_conv *sconv; 209 struct archive_string_conv *sconv_default; 210 struct archive_string_conv *sconv_utf8; 211 int init_default_conversion; 212 int process_mac_extensions; 213 214 char init_decryption; 215 216 /* Decryption buffer. */ 217 /* 218 * The decrypted data starts at decrypted_ptr and 219 * extends for decrypted_bytes_remaining. Decryption 220 * adds new data to the end of this block, data is returned 221 * to clients from the beginning. When the block hits the 222 * end of decrypted_buffer, it has to be shuffled back to 223 * the beginning of the buffer. 224 */ 225 unsigned char *decrypted_buffer; 226 unsigned char *decrypted_ptr; 227 size_t decrypted_buffer_size; 228 size_t decrypted_bytes_remaining; 229 size_t decrypted_unconsumed_bytes; 230 231 /* Traditional PKWARE decryption. */ 232 struct trad_enc_ctx tctx; 233 char tctx_valid; 234 235 /* WinZip AES decryption. */ 236 /* Contexts used for AES decryption. */ 237 archive_crypto_ctx cctx; 238 char cctx_valid; 239 archive_hmac_sha1_ctx hctx; 240 char hctx_valid; 241 242 /* Strong encryption's decryption header information. */ 243 unsigned iv_size; 244 unsigned alg_id; 245 unsigned bit_len; 246 unsigned flags; 247 unsigned erd_size; 248 unsigned v_size; 249 unsigned v_crc32; 250 uint8_t *iv; 251 uint8_t *erd; 252 uint8_t *v_data; 253 }; 254 255 /* Many systems define min or MIN, but not all. */ 256 #define zipmin(a,b) ((a) < (b) ? (a) : (b)) 257 258 #ifdef HAVE_ZLIB_H 259 static int 260 zip_read_data_deflate(struct archive_read *a, const void **buff, 261 size_t *size, int64_t *offset); 262 #endif 263 #if HAVE_LZMA_H && HAVE_LIBLZMA 264 static int 265 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, 266 size_t *size, int64_t *offset); 267 #endif 268 269 /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8 270 * streams inside ZIP files. It has 2 purposes: one is to fetch the next 271 * compressed byte from the stream, second one is to increase the counter how 272 * many compressed bytes were read. */ 273 static Byte 274 ppmd_read(void* p) { 275 /* Get the handle to current decompression context. */ 276 struct archive_read *a = ((IByteIn*)p)->a; 277 struct zip *zip = (struct zip*) a->format->data; 278 ssize_t bytes_avail = 0; 279 280 /* Fetch next byte. */ 281 const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail); 282 if(bytes_avail < 1) { 283 zip->ppmd8_stream_failed = 1; 284 return 0; 285 } 286 287 __archive_read_consume(a, 1); 288 289 /* Increment the counter. */ 290 ++zip->zipx_ppmd_read_compressed; 291 292 /* Return the next compressed byte. */ 293 return data[0]; 294 } 295 296 /* ------------------------------------------------------------------------ */ 297 298 /* 299 Traditional PKWARE Decryption functions. 300 */ 301 302 static void 303 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 304 { 305 uint8_t t; 306 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 307 308 ctx->keys[0] = CRC32(ctx->keys[0], c); 309 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 310 t = (ctx->keys[1] >> 24) & 0xff; 311 ctx->keys[2] = CRC32(ctx->keys[2], t); 312 #undef CRC32 313 } 314 315 static uint8_t 316 trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) 317 { 318 unsigned temp = ctx->keys[2] | 2; 319 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 320 } 321 322 static void 323 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 324 size_t in_len, uint8_t *out, size_t out_len) 325 { 326 unsigned i, max; 327 328 max = (unsigned)((in_len < out_len)? in_len: out_len); 329 330 for (i = 0; i < max; i++) { 331 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); 332 out[i] = t; 333 trad_enc_update_keys(ctx, t); 334 } 335 } 336 337 static int 338 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 339 const uint8_t *key, size_t key_len, uint8_t *crcchk) 340 { 341 uint8_t header[12]; 342 343 if (key_len < 12) { 344 *crcchk = 0xff; 345 return -1; 346 } 347 348 ctx->keys[0] = 305419896L; 349 ctx->keys[1] = 591751049L; 350 ctx->keys[2] = 878082192L; 351 352 for (;pw_len; --pw_len) 353 trad_enc_update_keys(ctx, *pw++); 354 355 trad_enc_decrypt_update(ctx, key, 12, header, 12); 356 /* Return the last byte for CRC check. */ 357 *crcchk = header[11]; 358 return 0; 359 } 360 361 #if 0 362 static void 363 crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 364 int key_size) 365 { 366 #define MD_SIZE 20 367 archive_sha1_ctx ctx; 368 unsigned char md1[MD_SIZE]; 369 unsigned char md2[MD_SIZE * 2]; 370 unsigned char mkb[64]; 371 int i; 372 373 archive_sha1_init(&ctx); 374 archive_sha1_update(&ctx, p, size); 375 archive_sha1_final(&ctx, md1); 376 377 memset(mkb, 0x36, sizeof(mkb)); 378 for (i = 0; i < MD_SIZE; i++) 379 mkb[i] ^= md1[i]; 380 archive_sha1_init(&ctx); 381 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 382 archive_sha1_final(&ctx, md2); 383 384 memset(mkb, 0x5C, sizeof(mkb)); 385 for (i = 0; i < MD_SIZE; i++) 386 mkb[i] ^= md1[i]; 387 archive_sha1_init(&ctx); 388 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 389 archive_sha1_final(&ctx, md2 + MD_SIZE); 390 391 if (key_size > 32) 392 key_size = 32; 393 memcpy(key, md2, key_size); 394 #undef MD_SIZE 395 } 396 #endif 397 398 /* 399 * Common code for streaming or seeking modes. 400 * 401 * Includes code to read local file headers, decompress data 402 * from entry bodies, and common API. 403 */ 404 405 static unsigned long 406 real_crc32(unsigned long crc, const void *buff, size_t len) 407 { 408 return crc32(crc, buff, (unsigned int)len); 409 } 410 411 /* Used by "ignorecrc32" option to speed up tests. */ 412 static unsigned long 413 fake_crc32(unsigned long crc, const void *buff, size_t len) 414 { 415 (void)crc; /* UNUSED */ 416 (void)buff; /* UNUSED */ 417 (void)len; /* UNUSED */ 418 return 0; 419 } 420 421 static const struct { 422 int id; 423 const char * name; 424 } compression_methods[] = { 425 {0, "uncompressed"}, /* The file is stored (no compression) */ 426 {1, "shrinking"}, /* The file is Shrunk */ 427 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 428 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 429 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 430 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 431 {6, "imploded"}, /* The file is Imploded */ 432 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 433 {8, "deflation"}, /* The file is Deflated */ 434 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 435 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 436 * (old IBM TERSE) */ 437 {11, "reserved"}, /* Reserved by PKWARE */ 438 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 439 {13, "reserved"}, /* Reserved by PKWARE */ 440 {14, "lzma"}, /* LZMA (EFS) */ 441 {15, "reserved"}, /* Reserved by PKWARE */ 442 {16, "reserved"}, /* Reserved by PKWARE */ 443 {17, "reserved"}, /* Reserved by PKWARE */ 444 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 445 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 446 {93, "zstd"}, /* Zstandard (zstd) Compression */ 447 {95, "xz"}, /* XZ compressed data */ 448 {96, "jpeg"}, /* JPEG compressed data */ 449 {97, "wav-pack"}, /* WavPack compressed data */ 450 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 451 {99, "aes"} /* WinZip AES encryption */ 452 }; 453 454 static const char * 455 compression_name(const int compression) 456 { 457 static const int num_compression_methods = 458 sizeof(compression_methods)/sizeof(compression_methods[0]); 459 int i=0; 460 461 while(compression >= 0 && i < num_compression_methods) { 462 if (compression_methods[i].id == compression) 463 return compression_methods[i].name; 464 i++; 465 } 466 return "??"; 467 } 468 469 /* Convert an MSDOS-style date/time into Unix-style time. */ 470 static time_t 471 zip_time(const char *p) 472 { 473 int msTime, msDate; 474 struct tm ts; 475 476 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 477 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 478 479 memset(&ts, 0, sizeof(ts)); 480 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 481 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 482 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 483 ts.tm_hour = (msTime >> 11) & 0x1f; 484 ts.tm_min = (msTime >> 5) & 0x3f; 485 ts.tm_sec = (msTime << 1) & 0x3e; 486 ts.tm_isdst = -1; 487 return mktime(&ts); 488 } 489 490 /* 491 * The extra data is stored as a list of 492 * id1+size1+data1 + id2+size2+data2 ... 493 * triplets. id and size are 2 bytes each. 494 */ 495 static int 496 process_extra(struct archive_read *a, struct archive_entry *entry, 497 const char *p, size_t extra_length, struct zip_entry* zip_entry) 498 { 499 unsigned offset = 0; 500 struct zip *zip = (struct zip *)(a->format->data); 501 502 if (extra_length == 0) { 503 return ARCHIVE_OK; 504 } 505 506 if (extra_length < 4) { 507 size_t i = 0; 508 /* Some ZIP files may have trailing 0 bytes. Let's check they 509 * are all 0 and ignore them instead of returning an error. 510 * 511 * This is not technically correct, but some ZIP files look 512 * like this and other tools support those files - so let's 513 * also support them. 514 */ 515 for (; i < extra_length; i++) { 516 if (p[i] != 0) { 517 archive_set_error(&a->archive, 518 ARCHIVE_ERRNO_FILE_FORMAT, 519 "Too-small extra data: " 520 "Need at least 4 bytes, " 521 "but only found %d bytes", 522 (int)extra_length); 523 return ARCHIVE_FAILED; 524 } 525 } 526 527 return ARCHIVE_OK; 528 } 529 530 while (offset <= extra_length - 4) { 531 unsigned short headerid = archive_le16dec(p + offset); 532 unsigned short datasize = archive_le16dec(p + offset + 2); 533 534 offset += 4; 535 if (offset + datasize > extra_length) { 536 archive_set_error(&a->archive, 537 ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: " 538 "Need %d bytes but only found %d bytes", 539 (int)datasize, (int)(extra_length - offset)); 540 return ARCHIVE_FAILED; 541 } 542 #ifdef DEBUG 543 fprintf(stderr, "Header id 0x%04x, length %d\n", 544 headerid, datasize); 545 #endif 546 switch (headerid) { 547 case 0x0001: 548 /* Zip64 extended information extra field. */ 549 zip_entry->flags |= LA_USED_ZIP64; 550 if (zip_entry->uncompressed_size == 0xffffffff) { 551 uint64_t t = 0; 552 if (datasize < 8 553 || (t = archive_le64dec(p + offset)) > 554 INT64_MAX) { 555 archive_set_error(&a->archive, 556 ARCHIVE_ERRNO_FILE_FORMAT, 557 "Malformed 64-bit " 558 "uncompressed size"); 559 return ARCHIVE_FAILED; 560 } 561 zip_entry->uncompressed_size = t; 562 offset += 8; 563 datasize -= 8; 564 } 565 if (zip_entry->compressed_size == 0xffffffff) { 566 uint64_t t = 0; 567 if (datasize < 8 568 || (t = archive_le64dec(p + offset)) > 569 INT64_MAX) { 570 archive_set_error(&a->archive, 571 ARCHIVE_ERRNO_FILE_FORMAT, 572 "Malformed 64-bit " 573 "compressed size"); 574 return ARCHIVE_FAILED; 575 } 576 zip_entry->compressed_size = t; 577 offset += 8; 578 datasize -= 8; 579 } 580 if (zip_entry->local_header_offset == 0xffffffff) { 581 uint64_t t = 0; 582 if (datasize < 8 583 || (t = archive_le64dec(p + offset)) > 584 INT64_MAX) { 585 archive_set_error(&a->archive, 586 ARCHIVE_ERRNO_FILE_FORMAT, 587 "Malformed 64-bit " 588 "local header offset"); 589 return ARCHIVE_FAILED; 590 } 591 zip_entry->local_header_offset = t; 592 offset += 8; 593 datasize -= 8; 594 } 595 /* archive_le32dec(p + offset) gives disk 596 * on which file starts, but we don't handle 597 * multi-volume Zip files. */ 598 break; 599 #ifdef DEBUG 600 case 0x0017: 601 { 602 /* Strong encryption field. */ 603 if (archive_le16dec(p + offset) == 2) { 604 unsigned algId = 605 archive_le16dec(p + offset + 2); 606 unsigned bitLen = 607 archive_le16dec(p + offset + 4); 608 int flags = 609 archive_le16dec(p + offset + 6); 610 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 611 "flgas=%d\n", algId, bitLen,flags); 612 } 613 break; 614 } 615 #endif 616 case 0x5455: 617 { 618 /* Extended time field "UT". */ 619 int flags; 620 if (datasize == 0) { 621 archive_set_error(&a->archive, 622 ARCHIVE_ERRNO_FILE_FORMAT, 623 "Incomplete extended time field"); 624 return ARCHIVE_FAILED; 625 } 626 flags = p[offset]; 627 offset++; 628 datasize--; 629 /* Flag bits indicate which dates are present. */ 630 if (flags & 0x01) 631 { 632 #ifdef DEBUG 633 fprintf(stderr, "mtime: %lld -> %d\n", 634 (long long)zip_entry->mtime, 635 archive_le32dec(p + offset)); 636 #endif 637 if (datasize < 4) 638 break; 639 zip_entry->mtime = archive_le32dec(p + offset); 640 offset += 4; 641 datasize -= 4; 642 } 643 if (flags & 0x02) 644 { 645 if (datasize < 4) 646 break; 647 zip_entry->atime = archive_le32dec(p + offset); 648 offset += 4; 649 datasize -= 4; 650 } 651 if (flags & 0x04) 652 { 653 if (datasize < 4) 654 break; 655 zip_entry->ctime = archive_le32dec(p + offset); 656 offset += 4; 657 datasize -= 4; 658 } 659 break; 660 } 661 case 0x5855: 662 { 663 /* Info-ZIP Unix Extra Field (old version) "UX". */ 664 if (datasize >= 8) { 665 zip_entry->atime = archive_le32dec(p + offset); 666 zip_entry->mtime = 667 archive_le32dec(p + offset + 4); 668 } 669 if (datasize >= 12) { 670 zip_entry->uid = 671 archive_le16dec(p + offset + 8); 672 zip_entry->gid = 673 archive_le16dec(p + offset + 10); 674 } 675 break; 676 } 677 case 0x6c78: 678 { 679 /* Experimental 'xl' field */ 680 /* 681 * Introduced Dec 2013 to provide a way to 682 * include external file attributes (and other 683 * fields that ordinarily appear only in 684 * central directory) in local file header. 685 * This provides file type and permission 686 * information necessary to support full 687 * streaming extraction. Currently being 688 * discussed with other Zip developers 689 * ... subject to change. 690 * 691 * Format: 692 * The field starts with a bitmap that specifies 693 * which additional fields are included. The 694 * bitmap is variable length and can be extended in 695 * the future. 696 * 697 * n bytes - feature bitmap: first byte has low-order 698 * 7 bits. If high-order bit is set, a subsequent 699 * byte holds the next 7 bits, etc. 700 * 701 * if bitmap & 1, 2 byte "version made by" 702 * if bitmap & 2, 2 byte "internal file attributes" 703 * if bitmap & 4, 4 byte "external file attributes" 704 * if bitmap & 8, 2 byte comment length + n byte 705 * comment 706 */ 707 int bitmap, bitmap_last; 708 709 if (datasize < 1) 710 break; 711 bitmap_last = bitmap = 0xff & p[offset]; 712 offset += 1; 713 datasize -= 1; 714 715 /* We only support first 7 bits of bitmap; skip rest. */ 716 while ((bitmap_last & 0x80) != 0 717 && datasize >= 1) { 718 bitmap_last = p[offset]; 719 offset += 1; 720 datasize -= 1; 721 } 722 723 if (bitmap & 1) { 724 /* 2 byte "version made by" */ 725 if (datasize < 2) 726 break; 727 zip_entry->system 728 = archive_le16dec(p + offset) >> 8; 729 offset += 2; 730 datasize -= 2; 731 } 732 if (bitmap & 2) { 733 /* 2 byte "internal file attributes" */ 734 uint32_t internal_attributes; 735 if (datasize < 2) 736 break; 737 internal_attributes 738 = archive_le16dec(p + offset); 739 /* Not used by libarchive at present. */ 740 (void)internal_attributes; /* UNUSED */ 741 offset += 2; 742 datasize -= 2; 743 } 744 if (bitmap & 4) { 745 /* 4 byte "external file attributes" */ 746 uint32_t external_attributes; 747 if (datasize < 4) 748 break; 749 external_attributes 750 = archive_le32dec(p + offset); 751 if (zip_entry->system == 3) { 752 zip_entry->mode 753 = external_attributes >> 16; 754 } else if (zip_entry->system == 0) { 755 // Interpret MSDOS directory bit 756 if (0x10 == (external_attributes & 757 0x10)) { 758 zip_entry->mode = 759 AE_IFDIR | 0775; 760 } else { 761 zip_entry->mode = 762 AE_IFREG | 0664; 763 } 764 if (0x01 == (external_attributes & 765 0x01)) { 766 /* Read-only bit; 767 * strip write permissions */ 768 zip_entry->mode &= 0555; 769 } 770 } else { 771 zip_entry->mode = 0; 772 } 773 offset += 4; 774 datasize -= 4; 775 } 776 if (bitmap & 8) { 777 /* 2 byte comment length + comment */ 778 uint32_t comment_length; 779 if (datasize < 2) 780 break; 781 comment_length 782 = archive_le16dec(p + offset); 783 offset += 2; 784 datasize -= 2; 785 786 if (datasize < comment_length) 787 break; 788 /* Comment is not supported by libarchive */ 789 offset += comment_length; 790 datasize -= comment_length; 791 } 792 break; 793 } 794 case 0x7075: 795 { 796 /* Info-ZIP Unicode Path Extra Field. */ 797 if (datasize < 5 || entry == NULL) 798 break; 799 offset += 5; 800 datasize -= 5; 801 802 /* The path name in this field is always encoded 803 * in UTF-8. */ 804 if (zip->sconv_utf8 == NULL) { 805 zip->sconv_utf8 = 806 archive_string_conversion_from_charset( 807 &a->archive, "UTF-8", 1); 808 /* If the converter from UTF-8 is not 809 * available, then the path name from the main 810 * field will more likely be correct. */ 811 if (zip->sconv_utf8 == NULL) 812 break; 813 } 814 815 /* Make sure the CRC32 of the filename matches. */ 816 if (!zip->ignore_crc32) { 817 const char *cp = archive_entry_pathname(entry); 818 if (cp) { 819 unsigned long file_crc = 820 zip->crc32func(0, cp, strlen(cp)); 821 unsigned long utf_crc = 822 archive_le32dec(p + offset - 4); 823 if (file_crc != utf_crc) { 824 #ifdef DEBUG 825 fprintf(stderr, 826 "CRC filename mismatch; " 827 "CDE is %lx, but UTF8 " 828 "is outdated with %lx\n", 829 file_crc, utf_crc); 830 #endif 831 break; 832 } 833 } 834 } 835 836 if (archive_entry_copy_pathname_l(entry, 837 p + offset, datasize, zip->sconv_utf8) != 0) { 838 /* Ignore the error, and fallback to the path 839 * name from the main field. */ 840 #ifdef DEBUG 841 fprintf(stderr, "Failed to read the ZIP " 842 "0x7075 extra field path.\n"); 843 #endif 844 } 845 break; 846 } 847 case 0x7855: 848 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 849 #ifdef DEBUG 850 fprintf(stderr, "uid %d gid %d\n", 851 archive_le16dec(p + offset), 852 archive_le16dec(p + offset + 2)); 853 #endif 854 if (datasize >= 2) 855 zip_entry->uid = archive_le16dec(p + offset); 856 if (datasize >= 4) 857 zip_entry->gid = 858 archive_le16dec(p + offset + 2); 859 break; 860 case 0x7875: 861 { 862 /* Info-Zip Unix Extra Field (type 3) "ux". */ 863 int uidsize = 0, gidsize = 0; 864 865 /* TODO: support arbitrary uidsize/gidsize. */ 866 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 867 if (datasize >= 4) { 868 /* get a uid size. */ 869 uidsize = 0xff & (int)p[offset+1]; 870 if (uidsize == 2) 871 zip_entry->uid = 872 archive_le16dec( 873 p + offset + 2); 874 else if (uidsize == 4 && datasize >= 6) 875 zip_entry->uid = 876 archive_le32dec( 877 p + offset + 2); 878 } 879 if (datasize >= (2 + uidsize + 3)) { 880 /* get a gid size. */ 881 gidsize = 0xff & 882 (int)p[offset+2+uidsize]; 883 if (gidsize == 2) 884 zip_entry->gid = 885 archive_le16dec( 886 p+offset+2+uidsize+1); 887 else if (gidsize == 4 && 888 datasize >= (2 + uidsize + 5)) 889 zip_entry->gid = 890 archive_le32dec( 891 p+offset+2+uidsize+1); 892 } 893 } 894 break; 895 } 896 case 0x9901: 897 /* WinZip AES extra data field. */ 898 if (datasize < 6) { 899 archive_set_error(&a->archive, 900 ARCHIVE_ERRNO_FILE_FORMAT, 901 "Incomplete AES field"); 902 return ARCHIVE_FAILED; 903 } 904 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 905 /* Vendor version. */ 906 zip_entry->aes_extra.vendor = 907 archive_le16dec(p + offset); 908 /* AES encryption strength. */ 909 zip_entry->aes_extra.strength = p[offset + 4]; 910 /* Actual compression method. */ 911 zip_entry->aes_extra.compression = 912 p[offset + 5]; 913 } 914 break; 915 default: 916 break; 917 } 918 offset += datasize; 919 } 920 return ARCHIVE_OK; 921 } 922 923 /* 924 * Assumes file pointer is at beginning of local file header. 925 */ 926 static int 927 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 928 struct zip *zip) 929 { 930 const char *p; 931 const void *h; 932 const wchar_t *wp; 933 const char *cp; 934 size_t len, filename_length, extra_length; 935 struct archive_string_conv *sconv; 936 struct zip_entry *zip_entry = zip->entry; 937 struct zip_entry zip_entry_central_dir; 938 int ret = ARCHIVE_OK; 939 char version; 940 941 /* Save a copy of the original for consistency checks. */ 942 zip_entry_central_dir = *zip_entry; 943 944 zip->decompress_init = 0; 945 zip->end_of_entry = 0; 946 zip->entry_uncompressed_bytes_read = 0; 947 zip->entry_compressed_bytes_read = 0; 948 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 949 950 /* Setup default conversion. */ 951 if (zip->sconv == NULL && !zip->init_default_conversion) { 952 zip->sconv_default = 953 archive_string_default_conversion_for_read(&(a->archive)); 954 zip->init_default_conversion = 1; 955 } 956 957 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 958 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 959 "Truncated ZIP file header"); 960 return (ARCHIVE_FATAL); 961 } 962 963 if (memcmp(p, "PK\003\004", 4) != 0) { 964 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 965 return ARCHIVE_FATAL; 966 } 967 version = p[4]; 968 zip_entry->system = p[5]; 969 zip_entry->zip_flags = archive_le16dec(p + 6); 970 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 971 zip->has_encrypted_entries = 1; 972 archive_entry_set_is_data_encrypted(entry, 1); 973 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 974 zip_entry->zip_flags & ZIP_ENCRYPTED && 975 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 976 archive_entry_set_is_metadata_encrypted(entry, 1); 977 return ARCHIVE_FATAL; 978 } 979 } 980 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 981 zip_entry->compression = (char)archive_le16dec(p + 8); 982 zip_entry->mtime = zip_time(p + 10); 983 zip_entry->crc32 = archive_le32dec(p + 14); 984 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 985 zip_entry->decdat = p[11]; 986 else 987 zip_entry->decdat = p[17]; 988 zip_entry->compressed_size = archive_le32dec(p + 18); 989 zip_entry->uncompressed_size = archive_le32dec(p + 22); 990 filename_length = archive_le16dec(p + 26); 991 extra_length = archive_le16dec(p + 28); 992 993 __archive_read_consume(a, 30); 994 995 /* Read the filename. */ 996 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 997 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 998 "Truncated ZIP file header"); 999 return (ARCHIVE_FATAL); 1000 } 1001 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 1002 /* The filename is stored to be UTF-8. */ 1003 if (zip->sconv_utf8 == NULL) { 1004 zip->sconv_utf8 = 1005 archive_string_conversion_from_charset( 1006 &a->archive, "UTF-8", 1); 1007 if (zip->sconv_utf8 == NULL) 1008 return (ARCHIVE_FATAL); 1009 } 1010 sconv = zip->sconv_utf8; 1011 } else if (zip->sconv != NULL) 1012 sconv = zip->sconv; 1013 else 1014 sconv = zip->sconv_default; 1015 1016 if (archive_entry_copy_pathname_l(entry, 1017 h, filename_length, sconv) != 0) { 1018 if (errno == ENOMEM) { 1019 archive_set_error(&a->archive, ENOMEM, 1020 "Can't allocate memory for Pathname"); 1021 return (ARCHIVE_FATAL); 1022 } 1023 archive_set_error(&a->archive, 1024 ARCHIVE_ERRNO_FILE_FORMAT, 1025 "Pathname cannot be converted " 1026 "from %s to current locale.", 1027 archive_string_conversion_charset_name(sconv)); 1028 ret = ARCHIVE_WARN; 1029 } 1030 __archive_read_consume(a, filename_length); 1031 1032 /* Read the extra data. */ 1033 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 1034 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1035 "Truncated ZIP file header"); 1036 return (ARCHIVE_FATAL); 1037 } 1038 1039 if (ARCHIVE_OK != process_extra(a, entry, h, extra_length, 1040 zip_entry)) { 1041 return ARCHIVE_FATAL; 1042 } 1043 __archive_read_consume(a, extra_length); 1044 1045 /* Work around a bug in Info-Zip: When reading from a pipe, it 1046 * stats the pipe instead of synthesizing a file entry. */ 1047 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 1048 zip_entry->mode &= ~ AE_IFMT; 1049 zip_entry->mode |= AE_IFREG; 1050 } 1051 1052 /* If the mode is totally empty, set some sane default. */ 1053 if (zip_entry->mode == 0) { 1054 zip_entry->mode |= 0664; 1055 } 1056 1057 /* Windows archivers sometimes use backslash as the directory 1058 * separator. Normalize to slash. */ 1059 if (zip_entry->system == 0 && 1060 (wp = archive_entry_pathname_w(entry)) != NULL) { 1061 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { 1062 size_t i; 1063 struct archive_wstring s; 1064 archive_string_init(&s); 1065 archive_wstrcpy(&s, wp); 1066 for (i = 0; i < archive_strlen(&s); i++) { 1067 if (s.s[i] == '\\') 1068 s.s[i] = '/'; 1069 } 1070 archive_entry_copy_pathname_w(entry, s.s); 1071 archive_wstring_free(&s); 1072 } 1073 } 1074 1075 /* Make sure that entries with a trailing '/' are marked as directories 1076 * even if the External File Attributes contains bogus values. If this 1077 * is not a directory and there is no type, assume a regular file. */ 1078 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 1079 int has_slash; 1080 1081 wp = archive_entry_pathname_w(entry); 1082 if (wp != NULL) { 1083 len = wcslen(wp); 1084 has_slash = len > 0 && wp[len - 1] == L'/'; 1085 } else { 1086 cp = archive_entry_pathname(entry); 1087 len = (cp != NULL)?strlen(cp):0; 1088 has_slash = len > 0 && cp[len - 1] == '/'; 1089 } 1090 /* Correct file type as needed. */ 1091 if (has_slash) { 1092 zip_entry->mode &= ~AE_IFMT; 1093 zip_entry->mode |= AE_IFDIR; 1094 zip_entry->mode |= 0111; 1095 } else if ((zip_entry->mode & AE_IFMT) == 0) { 1096 zip_entry->mode |= AE_IFREG; 1097 } 1098 } 1099 1100 /* Make sure directories end in '/' */ 1101 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 1102 wp = archive_entry_pathname_w(entry); 1103 if (wp != NULL) { 1104 len = wcslen(wp); 1105 if (len > 0 && wp[len - 1] != L'/') { 1106 struct archive_wstring s; 1107 archive_string_init(&s); 1108 archive_wstrcat(&s, wp); 1109 archive_wstrappend_wchar(&s, L'/'); 1110 archive_entry_copy_pathname_w(entry, s.s); 1111 archive_wstring_free(&s); 1112 } 1113 } else { 1114 cp = archive_entry_pathname(entry); 1115 len = (cp != NULL)?strlen(cp):0; 1116 if (len > 0 && cp[len - 1] != '/') { 1117 struct archive_string s; 1118 archive_string_init(&s); 1119 archive_strcat(&s, cp); 1120 archive_strappend_char(&s, '/'); 1121 archive_entry_set_pathname(entry, s.s); 1122 archive_string_free(&s); 1123 } 1124 } 1125 } 1126 1127 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 1128 /* If this came from the central dir, its size info 1129 * is definitive, so ignore the length-at-end flag. */ 1130 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 1131 /* If local header is missing a value, use the one from 1132 the central directory. If both have it, warn about 1133 mismatches. */ 1134 if (zip_entry->crc32 == 0) { 1135 zip_entry->crc32 = zip_entry_central_dir.crc32; 1136 } else if (!zip->ignore_crc32 1137 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 1138 archive_set_error(&a->archive, 1139 ARCHIVE_ERRNO_FILE_FORMAT, 1140 "Inconsistent CRC32 values"); 1141 ret = ARCHIVE_WARN; 1142 } 1143 if (zip_entry->compressed_size == 0) { 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 /* Set the size only if it's meaningful. */ 1289 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1290 } 1291 zip->entry_bytes_remaining = zip_entry->compressed_size; 1292 1293 /* If there's no body, force read_data() to return EOF immediately. */ 1294 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1295 && zip->entry_bytes_remaining < 1) 1296 zip->end_of_entry = 1; 1297 1298 /* Set up a more descriptive format name. */ 1299 archive_string_empty(&zip->format_name); 1300 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1301 version / 10, version % 10, 1302 compression_name(zip->entry->compression)); 1303 a->archive.archive_format_name = zip->format_name.s; 1304 1305 return (ret); 1306 } 1307 1308 static int 1309 check_authentication_code(struct archive_read *a, const void *_p) 1310 { 1311 struct zip *zip = (struct zip *)(a->format->data); 1312 1313 /* Check authentication code. */ 1314 if (zip->hctx_valid) { 1315 const void *p; 1316 uint8_t hmac[20]; 1317 size_t hmac_len = 20; 1318 int cmp; 1319 1320 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1321 if (_p == NULL) { 1322 /* Read authentication code. */ 1323 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1324 if (p == NULL) { 1325 archive_set_error(&a->archive, 1326 ARCHIVE_ERRNO_FILE_FORMAT, 1327 "Truncated ZIP file data"); 1328 return (ARCHIVE_FATAL); 1329 } 1330 } else { 1331 p = _p; 1332 } 1333 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1334 __archive_read_consume(a, AUTH_CODE_SIZE); 1335 if (cmp != 0) { 1336 archive_set_error(&a->archive, 1337 ARCHIVE_ERRNO_MISC, 1338 "ZIP bad Authentication code"); 1339 return (ARCHIVE_WARN); 1340 } 1341 } 1342 return (ARCHIVE_OK); 1343 } 1344 1345 /* 1346 * Read "uncompressed" data. There are three cases: 1347 * 1) We know the size of the data. This is always true for the 1348 * seeking reader (we've examined the Central Directory already). 1349 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1350 * Info-ZIP seems to do this; we know the size but have to grab 1351 * the CRC from the data descriptor afterwards. 1352 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1353 * we have no size information. In this case, we can do pretty 1354 * well by watching for the data descriptor record. The data 1355 * descriptor is 16 bytes and includes a computed CRC that should 1356 * provide a strong check. 1357 * 1358 * TODO: Technically, the PK\007\010 signature is optional. 1359 * In the original spec, the data descriptor contained CRC 1360 * and size fields but had no leading signature. In practice, 1361 * newer writers seem to provide the signature pretty consistently. 1362 * 1363 * For uncompressed data, the PK\007\010 marker seems essential 1364 * to be sure we've actually seen the end of the entry. 1365 * 1366 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1367 * zip->end_of_entry if it consumes all of the data. 1368 */ 1369 static int 1370 zip_read_data_none(struct archive_read *a, const void **_buff, 1371 size_t *size, int64_t *offset) 1372 { 1373 struct zip *zip; 1374 const char *buff; 1375 ssize_t bytes_avail; 1376 int r; 1377 1378 (void)offset; /* UNUSED */ 1379 1380 zip = (struct zip *)(a->format->data); 1381 1382 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1383 const char *p; 1384 ssize_t grabbing_bytes = 24; 1385 1386 if (zip->hctx_valid) 1387 grabbing_bytes += AUTH_CODE_SIZE; 1388 /* Grab at least 24 bytes. */ 1389 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1390 if (bytes_avail < grabbing_bytes) { 1391 /* Zip archives have end-of-archive markers 1392 that are longer than this, so a failure to get at 1393 least 24 bytes really does indicate a truncated 1394 file. */ 1395 archive_set_error(&a->archive, 1396 ARCHIVE_ERRNO_FILE_FORMAT, 1397 "Truncated ZIP file data"); 1398 return (ARCHIVE_FATAL); 1399 } 1400 /* Check for a complete PK\007\010 signature, followed 1401 * by the correct 4-byte CRC. */ 1402 p = buff; 1403 if (zip->hctx_valid) 1404 p += AUTH_CODE_SIZE; 1405 if (p[0] == 'P' && p[1] == 'K' 1406 && p[2] == '\007' && p[3] == '\010' 1407 && (archive_le32dec(p + 4) == zip->entry_crc32 1408 || zip->ignore_crc32 1409 || (zip->hctx_valid 1410 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1411 if (zip->entry->flags & LA_USED_ZIP64) { 1412 uint64_t compressed, uncompressed; 1413 zip->entry->crc32 = archive_le32dec(p + 4); 1414 compressed = archive_le64dec(p + 8); 1415 uncompressed = archive_le64dec(p + 16); 1416 if (compressed > INT64_MAX || uncompressed > 1417 INT64_MAX) { 1418 archive_set_error(&a->archive, 1419 ARCHIVE_ERRNO_FILE_FORMAT, 1420 "Overflow of 64-bit file sizes"); 1421 return ARCHIVE_FAILED; 1422 } 1423 zip->entry->compressed_size = compressed; 1424 zip->entry->uncompressed_size = uncompressed; 1425 zip->unconsumed = 24; 1426 } else { 1427 zip->entry->crc32 = archive_le32dec(p + 4); 1428 zip->entry->compressed_size = 1429 archive_le32dec(p + 8); 1430 zip->entry->uncompressed_size = 1431 archive_le32dec(p + 12); 1432 zip->unconsumed = 16; 1433 } 1434 if (zip->hctx_valid) { 1435 r = check_authentication_code(a, buff); 1436 if (r != ARCHIVE_OK) 1437 return (r); 1438 } 1439 zip->end_of_entry = 1; 1440 return (ARCHIVE_OK); 1441 } 1442 /* If not at EOF, ensure we consume at least one byte. */ 1443 ++p; 1444 1445 /* Scan forward until we see where a PK\007\010 signature 1446 * might be. */ 1447 /* Return bytes up until that point. On the next call, 1448 * the code above will verify the data descriptor. */ 1449 while (p < buff + bytes_avail - 4) { 1450 if (p[3] == 'P') { p += 3; } 1451 else if (p[3] == 'K') { p += 2; } 1452 else if (p[3] == '\007') { p += 1; } 1453 else if (p[3] == '\010' && p[2] == '\007' 1454 && p[1] == 'K' && p[0] == 'P') { 1455 if (zip->hctx_valid) 1456 p -= AUTH_CODE_SIZE; 1457 break; 1458 } else { p += 4; } 1459 } 1460 bytes_avail = p - buff; 1461 } else { 1462 if (zip->entry_bytes_remaining == 0) { 1463 zip->end_of_entry = 1; 1464 if (zip->hctx_valid) { 1465 r = check_authentication_code(a, NULL); 1466 if (r != ARCHIVE_OK) 1467 return (r); 1468 } 1469 return (ARCHIVE_OK); 1470 } 1471 /* Grab a bunch of bytes. */ 1472 buff = __archive_read_ahead(a, 1, &bytes_avail); 1473 if (bytes_avail <= 0) { 1474 archive_set_error(&a->archive, 1475 ARCHIVE_ERRNO_FILE_FORMAT, 1476 "Truncated ZIP file data"); 1477 return (ARCHIVE_FATAL); 1478 } 1479 if (bytes_avail > zip->entry_bytes_remaining) 1480 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1481 } 1482 if (zip->tctx_valid || zip->cctx_valid) { 1483 size_t dec_size = bytes_avail; 1484 1485 if (dec_size > zip->decrypted_buffer_size) 1486 dec_size = zip->decrypted_buffer_size; 1487 if (zip->tctx_valid) { 1488 trad_enc_decrypt_update(&zip->tctx, 1489 (const uint8_t *)buff, dec_size, 1490 zip->decrypted_buffer, dec_size); 1491 } else { 1492 size_t dsize = dec_size; 1493 archive_hmac_sha1_update(&zip->hctx, 1494 (const uint8_t *)buff, dec_size); 1495 archive_decrypto_aes_ctr_update(&zip->cctx, 1496 (const uint8_t *)buff, dec_size, 1497 zip->decrypted_buffer, &dsize); 1498 } 1499 bytes_avail = dec_size; 1500 buff = (const char *)zip->decrypted_buffer; 1501 } 1502 *size = bytes_avail; 1503 zip->entry_bytes_remaining -= bytes_avail; 1504 zip->entry_uncompressed_bytes_read += bytes_avail; 1505 zip->entry_compressed_bytes_read += bytes_avail; 1506 zip->unconsumed += bytes_avail; 1507 *_buff = buff; 1508 return (ARCHIVE_OK); 1509 } 1510 1511 static int 1512 consume_optional_marker(struct archive_read *a, struct zip *zip) 1513 { 1514 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1515 const char *p; 1516 1517 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1518 archive_set_error(&a->archive, 1519 ARCHIVE_ERRNO_FILE_FORMAT, 1520 "Truncated ZIP end-of-file record"); 1521 return (ARCHIVE_FATAL); 1522 } 1523 /* Consume the optional PK\007\010 marker. */ 1524 if (p[0] == 'P' && p[1] == 'K' && 1525 p[2] == '\007' && p[3] == '\010') { 1526 p += 4; 1527 zip->unconsumed = 4; 1528 } 1529 if (zip->entry->flags & LA_USED_ZIP64) { 1530 uint64_t compressed, uncompressed; 1531 zip->entry->crc32 = archive_le32dec(p); 1532 compressed = archive_le64dec(p + 4); 1533 uncompressed = archive_le64dec(p + 12); 1534 if (compressed > INT64_MAX || 1535 uncompressed > INT64_MAX) { 1536 archive_set_error(&a->archive, 1537 ARCHIVE_ERRNO_FILE_FORMAT, 1538 "Overflow of 64-bit file sizes"); 1539 return ARCHIVE_FAILED; 1540 } 1541 zip->entry->compressed_size = compressed; 1542 zip->entry->uncompressed_size = uncompressed; 1543 zip->unconsumed += 20; 1544 } else { 1545 zip->entry->crc32 = archive_le32dec(p); 1546 zip->entry->compressed_size = archive_le32dec(p + 4); 1547 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1548 zip->unconsumed += 12; 1549 } 1550 } 1551 1552 return (ARCHIVE_OK); 1553 } 1554 1555 #if HAVE_LZMA_H && HAVE_LIBLZMA 1556 static int 1557 zipx_xz_init(struct archive_read *a, struct zip *zip) 1558 { 1559 lzma_ret r; 1560 1561 if(zip->zipx_lzma_valid) { 1562 lzma_end(&zip->zipx_lzma_stream); 1563 zip->zipx_lzma_valid = 0; 1564 } 1565 1566 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); 1567 r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0); 1568 if (r != LZMA_OK) { 1569 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1570 "xz initialization failed(%d)", 1571 r); 1572 1573 return (ARCHIVE_FAILED); 1574 } 1575 1576 zip->zipx_lzma_valid = 1; 1577 1578 free(zip->uncompressed_buffer); 1579 1580 zip->uncompressed_buffer_size = 256 * 1024; 1581 zip->uncompressed_buffer = 1582 (uint8_t*) malloc(zip->uncompressed_buffer_size); 1583 if (zip->uncompressed_buffer == NULL) { 1584 archive_set_error(&a->archive, ENOMEM, 1585 "No memory for xz decompression"); 1586 return (ARCHIVE_FATAL); 1587 } 1588 1589 zip->decompress_init = 1; 1590 return (ARCHIVE_OK); 1591 } 1592 1593 static int 1594 zipx_lzma_alone_init(struct archive_read *a, struct zip *zip) 1595 { 1596 lzma_ret r; 1597 const uint8_t* p; 1598 1599 #pragma pack(push) 1600 #pragma pack(1) 1601 struct _alone_header { 1602 uint8_t bytes[5]; 1603 uint64_t uncompressed_size; 1604 } alone_header; 1605 #pragma pack(pop) 1606 1607 if(zip->zipx_lzma_valid) { 1608 lzma_end(&zip->zipx_lzma_stream); 1609 zip->zipx_lzma_valid = 0; 1610 } 1611 1612 /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma 1613 * that is a part of XZ Utils. The stream format stored inside ZIPX 1614 * file is a modified "lzma alone" file format, that was used by the 1615 * `lzma` utility which was later deprecated in favour of `xz` utility. 1616 * Since those formats are nearly the same, we can use a standard 1617 * "lzma alone" decoder from XZ Utils. */ 1618 1619 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); 1620 r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX); 1621 if (r != LZMA_OK) { 1622 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1623 "lzma initialization failed(%d)", r); 1624 1625 return (ARCHIVE_FAILED); 1626 } 1627 1628 /* Flag the cleanup function that we want our lzma-related structures 1629 * to be freed later. */ 1630 zip->zipx_lzma_valid = 1; 1631 1632 /* The "lzma alone" file format and the stream format inside ZIPx are 1633 * almost the same. Here's an example of a structure of "lzma alone" 1634 * format: 1635 * 1636 * $ cat /bin/ls | lzma | xxd | head -n 1 1637 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814 1638 * 1639 * 5 bytes 8 bytes n bytes 1640 * <lzma_params><uncompressed_size><data...> 1641 * 1642 * lzma_params is a 5-byte blob that has to be decoded to extract 1643 * parameters of this LZMA stream. The uncompressed_size field is an 1644 * uint64_t value that contains information about the size of the 1645 * uncompressed file, or UINT64_MAX if this value is unknown. 1646 * The <data...> part is the actual lzma-compressed data stream. 1647 * 1648 * Now here's the structure of the stream inside the ZIPX file: 1649 * 1650 * $ cat stream_inside_zipx | xxd | head -n 1 1651 * 00000000: 0914 0500 5d00 8000 0000 2814 .... .... 1652 * 1653 * 2byte 2byte 5 bytes n bytes 1654 * <magic1><magic2><lzma_params><data...> 1655 * 1656 * This means that the ZIPX file contains an additional magic1 and 1657 * magic2 headers, the lzma_params field contains the same parameter 1658 * set as in the "lzma alone" format, and the <data...> field is the 1659 * same as in the "lzma alone" format as well. Note that also the zipx 1660 * format is missing the uncompressed_size field. 1661 * 1662 * So, in order to use the "lzma alone" decoder for the zipx lzma 1663 * stream, we simply need to shuffle around some fields, prepare a new 1664 * lzma alone header, feed it into lzma alone decoder so it will 1665 * initialize itself properly, and then we can start feeding normal 1666 * zipx lzma stream into the decoder. 1667 */ 1668 1669 /* Read magic1,magic2,lzma_params from the ZIPX stream. */ 1670 if(zip->entry_bytes_remaining < 9 || (p = __archive_read_ahead(a, 9, NULL)) == NULL) { 1671 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1672 "Truncated lzma data"); 1673 return (ARCHIVE_FATAL); 1674 } 1675 1676 if(p[2] != 0x05 || p[3] != 0x00) { 1677 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1678 "Invalid lzma data"); 1679 return (ARCHIVE_FATAL); 1680 } 1681 1682 /* Prepare an lzma alone header: copy the lzma_params blob into 1683 * a proper place into the lzma alone header. */ 1684 memcpy(&alone_header.bytes[0], p + 4, 5); 1685 1686 /* Initialize the 'uncompressed size' field to unknown; we'll manually 1687 * monitor how many bytes there are still to be uncompressed. */ 1688 alone_header.uncompressed_size = UINT64_MAX; 1689 1690 if(!zip->uncompressed_buffer) { 1691 zip->uncompressed_buffer_size = 256 * 1024; 1692 zip->uncompressed_buffer = 1693 (uint8_t*) malloc(zip->uncompressed_buffer_size); 1694 1695 if (zip->uncompressed_buffer == NULL) { 1696 archive_set_error(&a->archive, ENOMEM, 1697 "No memory for lzma decompression"); 1698 return (ARCHIVE_FATAL); 1699 } 1700 } 1701 1702 zip->zipx_lzma_stream.next_in = (void*) &alone_header; 1703 zip->zipx_lzma_stream.avail_in = sizeof(alone_header); 1704 zip->zipx_lzma_stream.total_in = 0; 1705 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1706 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; 1707 zip->zipx_lzma_stream.total_out = 0; 1708 1709 /* Feed only the header into the lzma alone decoder. This will 1710 * effectively initialize the decoder, and will not produce any 1711 * output bytes yet. */ 1712 r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1713 if (r != LZMA_OK) { 1714 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1715 "lzma stream initialization error"); 1716 return ARCHIVE_FATAL; 1717 } 1718 1719 /* We've already consumed some bytes, so take this into account. */ 1720 __archive_read_consume(a, 9); 1721 zip->entry_bytes_remaining -= 9; 1722 zip->entry_compressed_bytes_read += 9; 1723 1724 zip->decompress_init = 1; 1725 return (ARCHIVE_OK); 1726 } 1727 1728 static int 1729 zip_read_data_zipx_xz(struct archive_read *a, const void **buff, 1730 size_t *size, int64_t *offset) 1731 { 1732 struct zip* zip = (struct zip *)(a->format->data); 1733 int ret; 1734 lzma_ret lz_ret; 1735 const void* compressed_buf; 1736 ssize_t bytes_avail, in_bytes, to_consume = 0; 1737 1738 (void) offset; /* UNUSED */ 1739 1740 /* Initialize decompressor if not yet initialized. */ 1741 if (!zip->decompress_init) { 1742 ret = zipx_xz_init(a, zip); 1743 if (ret != ARCHIVE_OK) 1744 return (ret); 1745 } 1746 1747 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); 1748 if (bytes_avail < 0) { 1749 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1750 "Truncated xz file body"); 1751 return (ARCHIVE_FATAL); 1752 } 1753 1754 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 1755 zip->zipx_lzma_stream.next_in = compressed_buf; 1756 zip->zipx_lzma_stream.avail_in = in_bytes; 1757 zip->zipx_lzma_stream.total_in = 0; 1758 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1759 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; 1760 zip->zipx_lzma_stream.total_out = 0; 1761 1762 /* Perform the decompression. */ 1763 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1764 switch(lz_ret) { 1765 case LZMA_DATA_ERROR: 1766 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1767 "xz data error (error %d)", (int) lz_ret); 1768 return (ARCHIVE_FATAL); 1769 1770 case LZMA_NO_CHECK: 1771 case LZMA_OK: 1772 break; 1773 1774 default: 1775 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1776 "xz unknown error %d", (int) lz_ret); 1777 return (ARCHIVE_FATAL); 1778 1779 case LZMA_STREAM_END: 1780 lzma_end(&zip->zipx_lzma_stream); 1781 zip->zipx_lzma_valid = 0; 1782 1783 if((int64_t) zip->zipx_lzma_stream.total_in != 1784 zip->entry_bytes_remaining) 1785 { 1786 archive_set_error(&a->archive, 1787 ARCHIVE_ERRNO_MISC, 1788 "xz premature end of stream"); 1789 return (ARCHIVE_FATAL); 1790 } 1791 1792 zip->end_of_entry = 1; 1793 break; 1794 } 1795 1796 to_consume = zip->zipx_lzma_stream.total_in; 1797 1798 __archive_read_consume(a, to_consume); 1799 zip->entry_bytes_remaining -= to_consume; 1800 zip->entry_compressed_bytes_read += to_consume; 1801 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; 1802 1803 *size = zip->zipx_lzma_stream.total_out; 1804 *buff = zip->uncompressed_buffer; 1805 1806 ret = consume_optional_marker(a, zip); 1807 if (ret != ARCHIVE_OK) 1808 return (ret); 1809 1810 return (ARCHIVE_OK); 1811 } 1812 1813 static int 1814 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, 1815 size_t *size, int64_t *offset) 1816 { 1817 struct zip* zip = (struct zip *)(a->format->data); 1818 int ret; 1819 lzma_ret lz_ret; 1820 const void* compressed_buf; 1821 ssize_t bytes_avail, in_bytes, to_consume; 1822 1823 (void) offset; /* UNUSED */ 1824 1825 /* Initialize decompressor if not yet initialized. */ 1826 if (!zip->decompress_init) { 1827 ret = zipx_lzma_alone_init(a, zip); 1828 if (ret != ARCHIVE_OK) 1829 return (ret); 1830 } 1831 1832 /* Fetch more compressed data. The same note as in deflate handler 1833 * applies here as well: 1834 * 1835 * Note: '1' here is a performance optimization. Recall that the 1836 * decompression layer returns a count of available bytes; asking for 1837 * more than that forces the decompressor to combine reads by copying 1838 * data. 1839 */ 1840 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); 1841 if (bytes_avail < 0) { 1842 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1843 "Truncated lzma file body"); 1844 return (ARCHIVE_FATAL); 1845 } 1846 1847 /* Set decompressor parameters. */ 1848 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 1849 1850 zip->zipx_lzma_stream.next_in = compressed_buf; 1851 zip->zipx_lzma_stream.avail_in = in_bytes; 1852 zip->zipx_lzma_stream.total_in = 0; 1853 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1854 zip->zipx_lzma_stream.avail_out = 1855 /* These lzma_alone streams lack end of stream marker, so let's 1856 * make sure the unpacker won't try to unpack more than it's 1857 * supposed to. */ 1858 zipmin((int64_t) zip->uncompressed_buffer_size, 1859 zip->entry->uncompressed_size - 1860 zip->entry_uncompressed_bytes_read); 1861 zip->zipx_lzma_stream.total_out = 0; 1862 1863 /* Perform the decompression. */ 1864 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1865 switch(lz_ret) { 1866 case LZMA_DATA_ERROR: 1867 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1868 "lzma data error (error %d)", (int) lz_ret); 1869 return (ARCHIVE_FATAL); 1870 1871 /* This case is optional in lzma alone format. It can happen, 1872 * but most of the files don't have it. (GitHub #1257) */ 1873 case LZMA_STREAM_END: 1874 lzma_end(&zip->zipx_lzma_stream); 1875 zip->zipx_lzma_valid = 0; 1876 if((int64_t) zip->zipx_lzma_stream.total_in != 1877 zip->entry_bytes_remaining) 1878 { 1879 archive_set_error(&a->archive, 1880 ARCHIVE_ERRNO_MISC, 1881 "lzma alone premature end of stream"); 1882 return (ARCHIVE_FATAL); 1883 } 1884 1885 zip->end_of_entry = 1; 1886 break; 1887 1888 case LZMA_OK: 1889 break; 1890 1891 default: 1892 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1893 "lzma unknown error %d", (int) lz_ret); 1894 return (ARCHIVE_FATAL); 1895 } 1896 1897 to_consume = zip->zipx_lzma_stream.total_in; 1898 1899 /* Update pointers. */ 1900 __archive_read_consume(a, to_consume); 1901 zip->entry_bytes_remaining -= to_consume; 1902 zip->entry_compressed_bytes_read += to_consume; 1903 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; 1904 1905 if(zip->entry_bytes_remaining == 0) { 1906 zip->end_of_entry = 1; 1907 } 1908 1909 /* Return values. */ 1910 *size = zip->zipx_lzma_stream.total_out; 1911 *buff = zip->uncompressed_buffer; 1912 1913 /* Behave the same way as during deflate decompression. */ 1914 ret = consume_optional_marker(a, zip); 1915 if (ret != ARCHIVE_OK) 1916 return (ret); 1917 1918 /* Free lzma decoder handle because we'll no longer need it. */ 1919 if(zip->end_of_entry) { 1920 lzma_end(&zip->zipx_lzma_stream); 1921 zip->zipx_lzma_valid = 0; 1922 } 1923 1924 /* If we're here, then we're good! */ 1925 return (ARCHIVE_OK); 1926 } 1927 #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */ 1928 1929 static int 1930 zipx_ppmd8_init(struct archive_read *a, struct zip *zip) 1931 { 1932 const void* p; 1933 uint32_t val; 1934 uint32_t order; 1935 uint32_t mem; 1936 uint32_t restore_method; 1937 1938 /* Remove previous decompression context if it exists. */ 1939 if(zip->ppmd8_valid) { 1940 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 1941 zip->ppmd8_valid = 0; 1942 } 1943 1944 /* Create a new decompression context. */ 1945 __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8); 1946 zip->ppmd8_stream_failed = 0; 1947 1948 /* Setup function pointers required by Ppmd8 decompressor. The 1949 * 'ppmd_read' function will feed new bytes to the decompressor, 1950 * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */ 1951 zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream; 1952 zip->zipx_ppmd_stream.a = a; 1953 zip->zipx_ppmd_stream.Read = &ppmd_read; 1954 1955 /* Reset number of read bytes to 0. */ 1956 zip->zipx_ppmd_read_compressed = 0; 1957 1958 /* Read Ppmd8 header (2 bytes). */ 1959 p = __archive_read_ahead(a, 2, NULL); 1960 if(!p) { 1961 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1962 "Truncated file data in PPMd8 stream"); 1963 return (ARCHIVE_FATAL); 1964 } 1965 __archive_read_consume(a, 2); 1966 1967 /* Decode the stream's compression parameters. */ 1968 val = archive_le16dec(p); 1969 order = (val & 15) + 1; 1970 mem = ((val >> 4) & 0xff) + 1; 1971 restore_method = (val >> 12); 1972 1973 if(order < 2 || restore_method > 2) { 1974 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1975 "Invalid parameter set in PPMd8 stream (order=%" PRId32 ", " 1976 "restore=%" PRId32 ")", order, restore_method); 1977 return (ARCHIVE_FAILED); 1978 } 1979 1980 /* Allocate the memory needed to properly decompress the file. */ 1981 if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) { 1982 archive_set_error(&a->archive, ENOMEM, 1983 "Unable to allocate memory for PPMd8 stream: %" PRId32 " bytes", 1984 mem << 20); 1985 return (ARCHIVE_FATAL); 1986 } 1987 1988 /* Signal the cleanup function to release Ppmd8 context in the 1989 * cleanup phase. */ 1990 zip->ppmd8_valid = 1; 1991 1992 /* Perform further Ppmd8 initialization. */ 1993 if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) { 1994 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1995 "PPMd8 stream range decoder initialization error"); 1996 return (ARCHIVE_FATAL); 1997 } 1998 1999 __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order, 2000 restore_method); 2001 2002 /* Allocate the buffer that will hold uncompressed data. */ 2003 free(zip->uncompressed_buffer); 2004 2005 zip->uncompressed_buffer_size = 256 * 1024; 2006 zip->uncompressed_buffer = 2007 (uint8_t*) malloc(zip->uncompressed_buffer_size); 2008 2009 if(zip->uncompressed_buffer == NULL) { 2010 archive_set_error(&a->archive, ENOMEM, 2011 "No memory for PPMd8 decompression"); 2012 return ARCHIVE_FATAL; 2013 } 2014 2015 /* Ppmd8 initialization is done. */ 2016 zip->decompress_init = 1; 2017 2018 /* We've already read 2 bytes in the output stream. Additionally, 2019 * Ppmd8 initialization code could read some data as well. So we 2020 * are advancing the stream by 2 bytes plus whatever number of 2021 * bytes Ppmd8 init function used. */ 2022 zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed; 2023 2024 return ARCHIVE_OK; 2025 } 2026 2027 static int 2028 zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff, 2029 size_t *size, int64_t *offset) 2030 { 2031 struct zip* zip = (struct zip *)(a->format->data); 2032 int ret; 2033 size_t consumed_bytes = 0; 2034 ssize_t bytes_avail = 0; 2035 2036 (void) offset; /* UNUSED */ 2037 2038 /* If we're here for the first time, initialize Ppmd8 decompression 2039 * context first. */ 2040 if(!zip->decompress_init) { 2041 ret = zipx_ppmd8_init(a, zip); 2042 if(ret != ARCHIVE_OK) 2043 return ret; 2044 } 2045 2046 /* Fetch for more data. We're reading 1 byte here, but libarchive 2047 * should prefetch more bytes. */ 2048 (void) __archive_read_ahead(a, 1, &bytes_avail); 2049 if(bytes_avail < 0) { 2050 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2051 "Truncated PPMd8 file body"); 2052 return (ARCHIVE_FATAL); 2053 } 2054 2055 /* This counter will be updated inside ppmd_read(), which at one 2056 * point will be called by Ppmd8_DecodeSymbol. */ 2057 zip->zipx_ppmd_read_compressed = 0; 2058 2059 /* Decompression loop. */ 2060 do { 2061 int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol( 2062 &zip->ppmd8); 2063 if(sym < 0) { 2064 zip->end_of_entry = 1; 2065 break; 2066 } 2067 2068 /* This field is set by ppmd_read() when there was no more data 2069 * to be read. */ 2070 if(zip->ppmd8_stream_failed) { 2071 archive_set_error(&a->archive, 2072 ARCHIVE_ERRNO_FILE_FORMAT, 2073 "Truncated PPMd8 file body"); 2074 return (ARCHIVE_FATAL); 2075 } 2076 2077 zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym; 2078 ++consumed_bytes; 2079 } while(consumed_bytes < zip->uncompressed_buffer_size); 2080 2081 /* Update pointers for libarchive. */ 2082 *buff = zip->uncompressed_buffer; 2083 *size = consumed_bytes; 2084 2085 /* Update pointers so we can continue decompression in another call. */ 2086 zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed; 2087 zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed; 2088 zip->entry_uncompressed_bytes_read += consumed_bytes; 2089 2090 /* If we're at the end of stream, deinitialize Ppmd8 context. */ 2091 if(zip->end_of_entry) { 2092 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 2093 zip->ppmd8_valid = 0; 2094 } 2095 2096 /* Seek for optional marker, same way as in each zip entry. */ 2097 ret = consume_optional_marker(a, zip); 2098 if (ret != ARCHIVE_OK) 2099 return ret; 2100 2101 return ARCHIVE_OK; 2102 } 2103 2104 #ifdef HAVE_BZLIB_H 2105 static int 2106 zipx_bzip2_init(struct archive_read *a, struct zip *zip) 2107 { 2108 int r; 2109 2110 /* Deallocate already existing BZ2 decompression context if it 2111 * exists. */ 2112 if(zip->bzstream_valid) { 2113 BZ2_bzDecompressEnd(&zip->bzstream); 2114 zip->bzstream_valid = 0; 2115 } 2116 2117 /* Allocate a new BZ2 decompression context. */ 2118 memset(&zip->bzstream, 0, sizeof(bz_stream)); 2119 r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1); 2120 if(r != BZ_OK) { 2121 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 2122 "bzip2 initialization failed(%d)", 2123 r); 2124 2125 return ARCHIVE_FAILED; 2126 } 2127 2128 /* Mark the bzstream field to be released in cleanup phase. */ 2129 zip->bzstream_valid = 1; 2130 2131 /* (Re)allocate the buffer that will contain decompressed bytes. */ 2132 free(zip->uncompressed_buffer); 2133 2134 zip->uncompressed_buffer_size = 256 * 1024; 2135 zip->uncompressed_buffer = 2136 (uint8_t*) malloc(zip->uncompressed_buffer_size); 2137 if (zip->uncompressed_buffer == NULL) { 2138 archive_set_error(&a->archive, ENOMEM, 2139 "No memory for bzip2 decompression"); 2140 return ARCHIVE_FATAL; 2141 } 2142 2143 /* Initialization done. */ 2144 zip->decompress_init = 1; 2145 return ARCHIVE_OK; 2146 } 2147 2148 static int 2149 zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff, 2150 size_t *size, int64_t *offset) 2151 { 2152 struct zip *zip = (struct zip *)(a->format->data); 2153 ssize_t bytes_avail = 0, in_bytes, to_consume; 2154 const void *compressed_buff; 2155 int r; 2156 uint64_t total_out; 2157 2158 (void) offset; /* UNUSED */ 2159 2160 /* Initialize decompression context if we're here for the first time. */ 2161 if(!zip->decompress_init) { 2162 r = zipx_bzip2_init(a, zip); 2163 if(r != ARCHIVE_OK) 2164 return r; 2165 } 2166 2167 /* Fetch more compressed bytes. */ 2168 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); 2169 if(bytes_avail < 0) { 2170 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2171 "Truncated bzip2 file body"); 2172 return (ARCHIVE_FATAL); 2173 } 2174 2175 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 2176 if(in_bytes < 1) { 2177 /* libbz2 doesn't complain when caller feeds avail_in == 0. 2178 * It will actually return success in this case, which is 2179 * undesirable. This is why we need to make this check 2180 * manually. */ 2181 2182 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2183 "Truncated bzip2 file body"); 2184 return (ARCHIVE_FATAL); 2185 } 2186 2187 /* Setup buffer boundaries. */ 2188 zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff; 2189 zip->bzstream.avail_in = in_bytes; 2190 zip->bzstream.total_in_hi32 = 0; 2191 zip->bzstream.total_in_lo32 = 0; 2192 zip->bzstream.next_out = (char*) zip->uncompressed_buffer; 2193 zip->bzstream.avail_out = zip->uncompressed_buffer_size; 2194 zip->bzstream.total_out_hi32 = 0; 2195 zip->bzstream.total_out_lo32 = 0; 2196 2197 /* Perform the decompression. */ 2198 r = BZ2_bzDecompress(&zip->bzstream); 2199 switch(r) { 2200 case BZ_STREAM_END: 2201 /* If we're at the end of the stream, deinitialize the 2202 * decompression context now. */ 2203 switch(BZ2_bzDecompressEnd(&zip->bzstream)) { 2204 case BZ_OK: 2205 break; 2206 default: 2207 archive_set_error(&a->archive, 2208 ARCHIVE_ERRNO_MISC, 2209 "Failed to clean up bzip2 " 2210 "decompressor"); 2211 return ARCHIVE_FATAL; 2212 } 2213 2214 zip->end_of_entry = 1; 2215 break; 2216 case BZ_OK: 2217 /* The decompressor has successfully decoded this 2218 * chunk of data, but more data is still in queue. */ 2219 break; 2220 default: 2221 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2222 "bzip2 decompression failed"); 2223 return ARCHIVE_FATAL; 2224 } 2225 2226 /* Update the pointers so decompressor can continue decoding. */ 2227 to_consume = zip->bzstream.total_in_lo32; 2228 __archive_read_consume(a, to_consume); 2229 2230 total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) + 2231 zip->bzstream.total_out_lo32; 2232 2233 zip->entry_bytes_remaining -= to_consume; 2234 zip->entry_compressed_bytes_read += to_consume; 2235 zip->entry_uncompressed_bytes_read += total_out; 2236 2237 /* Give libarchive its due. */ 2238 *size = total_out; 2239 *buff = zip->uncompressed_buffer; 2240 2241 /* Seek for optional marker, like in other entries. */ 2242 r = consume_optional_marker(a, zip); 2243 if(r != ARCHIVE_OK) 2244 return r; 2245 2246 return ARCHIVE_OK; 2247 } 2248 2249 #endif 2250 2251 #if HAVE_ZSTD_H && HAVE_LIBZSTD 2252 static int 2253 zipx_zstd_init(struct archive_read *a, struct zip *zip) 2254 { 2255 size_t r; 2256 2257 /* Deallocate already existing Zstd decompression context if it 2258 * exists. */ 2259 if(zip->zstdstream_valid) { 2260 ZSTD_freeDStream(zip->zstdstream); 2261 zip->zstdstream_valid = 0; 2262 } 2263 2264 /* Allocate a new Zstd decompression context. */ 2265 zip->zstdstream = ZSTD_createDStream(); 2266 2267 r = ZSTD_initDStream(zip->zstdstream); 2268 if (ZSTD_isError(r)) { 2269 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2270 "Error initializing zstd decompressor: %s", 2271 ZSTD_getErrorName(r)); 2272 2273 return ARCHIVE_FAILED; 2274 } 2275 2276 /* Mark the zstdstream field to be released in cleanup phase. */ 2277 zip->zstdstream_valid = 1; 2278 2279 /* (Re)allocate the buffer that will contain decompressed bytes. */ 2280 free(zip->uncompressed_buffer); 2281 2282 zip->uncompressed_buffer_size = ZSTD_DStreamOutSize(); 2283 zip->uncompressed_buffer = 2284 (uint8_t*) malloc(zip->uncompressed_buffer_size); 2285 if (zip->uncompressed_buffer == NULL) { 2286 archive_set_error(&a->archive, ENOMEM, 2287 "No memory for Zstd decompression"); 2288 2289 return ARCHIVE_FATAL; 2290 } 2291 2292 /* Initialization done. */ 2293 zip->decompress_init = 1; 2294 return ARCHIVE_OK; 2295 } 2296 2297 static int 2298 zip_read_data_zipx_zstd(struct archive_read *a, const void **buff, 2299 size_t *size, int64_t *offset) 2300 { 2301 struct zip *zip = (struct zip *)(a->format->data); 2302 ssize_t bytes_avail = 0, in_bytes, to_consume; 2303 const void *compressed_buff; 2304 int r; 2305 size_t ret; 2306 uint64_t total_out; 2307 ZSTD_outBuffer out; 2308 ZSTD_inBuffer in; 2309 2310 (void) offset; /* UNUSED */ 2311 2312 /* Initialize decompression context if we're here for the first time. */ 2313 if(!zip->decompress_init) { 2314 r = zipx_zstd_init(a, zip); 2315 if(r != ARCHIVE_OK) 2316 return r; 2317 } 2318 2319 /* Fetch more compressed bytes */ 2320 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); 2321 if(bytes_avail < 0) { 2322 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2323 "Truncated zstd file body"); 2324 return (ARCHIVE_FATAL); 2325 } 2326 2327 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 2328 if(in_bytes < 1) { 2329 /* zstd doesn't complain when caller feeds avail_in == 0. 2330 * It will actually return success in this case, which is 2331 * undesirable. This is why we need to make this check 2332 * manually. */ 2333 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2334 "Truncated zstd file body"); 2335 return (ARCHIVE_FATAL); 2336 } 2337 2338 /* Setup buffer boundaries */ 2339 in.src = compressed_buff; 2340 in.size = in_bytes; 2341 in.pos = 0; 2342 out = (ZSTD_outBuffer) { zip->uncompressed_buffer, zip->uncompressed_buffer_size, 0 }; 2343 2344 /* Perform the decompression. */ 2345 ret = ZSTD_decompressStream(zip->zstdstream, &out, &in); 2346 if (ZSTD_isError(ret)) { 2347 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2348 "Error during zstd decompression: %s", 2349 ZSTD_getErrorName(ret)); 2350 return (ARCHIVE_FATAL); 2351 } 2352 2353 /* Check end of the stream. */ 2354 if (ret == 0) { 2355 if ((in.pos == in.size) && (out.pos < out.size)) { 2356 zip->end_of_entry = 1; 2357 ZSTD_freeDStream(zip->zstdstream); 2358 zip->zstdstream_valid = 0; 2359 } 2360 } 2361 2362 /* Update the pointers so decompressor can continue decoding. */ 2363 to_consume = in.pos; 2364 __archive_read_consume(a, to_consume); 2365 2366 total_out = out.pos; 2367 2368 zip->entry_bytes_remaining -= to_consume; 2369 zip->entry_compressed_bytes_read += to_consume; 2370 zip->entry_uncompressed_bytes_read += total_out; 2371 2372 /* Give libarchive its due. */ 2373 *size = total_out; 2374 *buff = zip->uncompressed_buffer; 2375 2376 /* Seek for optional marker, like in other entries. */ 2377 r = consume_optional_marker(a, zip); 2378 if(r != ARCHIVE_OK) 2379 return r; 2380 2381 return ARCHIVE_OK; 2382 } 2383 #endif 2384 2385 #ifdef HAVE_ZLIB_H 2386 static int 2387 zip_deflate_init(struct archive_read *a, struct zip *zip) 2388 { 2389 int r; 2390 2391 /* If we haven't yet read any data, initialize the decompressor. */ 2392 if (!zip->decompress_init) { 2393 if (zip->stream_valid) 2394 r = inflateReset(&zip->stream); 2395 else 2396 r = inflateInit2(&zip->stream, 2397 -15 /* Don't check for zlib header */); 2398 if (r != Z_OK) { 2399 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2400 "Can't initialize ZIP decompression."); 2401 return (ARCHIVE_FATAL); 2402 } 2403 /* Stream structure has been set up. */ 2404 zip->stream_valid = 1; 2405 /* We've initialized decompression for this stream. */ 2406 zip->decompress_init = 1; 2407 } 2408 return (ARCHIVE_OK); 2409 } 2410 2411 static int 2412 zip_read_data_deflate(struct archive_read *a, const void **buff, 2413 size_t *size, int64_t *offset) 2414 { 2415 struct zip *zip; 2416 ssize_t bytes_avail; 2417 const void *compressed_buff, *sp; 2418 int r; 2419 2420 (void)offset; /* UNUSED */ 2421 2422 zip = (struct zip *)(a->format->data); 2423 2424 /* If the buffer hasn't been allocated, allocate it now. */ 2425 if (zip->uncompressed_buffer == NULL) { 2426 zip->uncompressed_buffer_size = 256 * 1024; 2427 zip->uncompressed_buffer 2428 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 2429 if (zip->uncompressed_buffer == NULL) { 2430 archive_set_error(&a->archive, ENOMEM, 2431 "No memory for ZIP decompression"); 2432 return (ARCHIVE_FATAL); 2433 } 2434 } 2435 2436 r = zip_deflate_init(a, zip); 2437 if (r != ARCHIVE_OK) 2438 return (r); 2439 2440 /* 2441 * Note: '1' here is a performance optimization. 2442 * Recall that the decompression layer returns a count of 2443 * available bytes; asking for more than that forces the 2444 * decompressor to combine reads by copying data. 2445 */ 2446 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 2447 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2448 && bytes_avail > zip->entry_bytes_remaining) { 2449 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 2450 } 2451 if (bytes_avail < 0) { 2452 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2453 "Truncated ZIP file body"); 2454 return (ARCHIVE_FATAL); 2455 } 2456 2457 if (zip->tctx_valid || zip->cctx_valid) { 2458 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 2459 size_t buff_remaining = 2460 (zip->decrypted_buffer + 2461 zip->decrypted_buffer_size) 2462 - (zip->decrypted_ptr + 2463 zip->decrypted_bytes_remaining); 2464 2465 if (buff_remaining > (size_t)bytes_avail) 2466 buff_remaining = (size_t)bytes_avail; 2467 2468 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 2469 zip->entry_bytes_remaining > 0) { 2470 if ((int64_t)(zip->decrypted_bytes_remaining 2471 + buff_remaining) 2472 > zip->entry_bytes_remaining) { 2473 if (zip->entry_bytes_remaining < 2474 (int64_t)zip->decrypted_bytes_remaining) 2475 buff_remaining = 0; 2476 else 2477 buff_remaining = 2478 (size_t)zip->entry_bytes_remaining 2479 - zip->decrypted_bytes_remaining; 2480 } 2481 } 2482 if (buff_remaining > 0) { 2483 if (zip->tctx_valid) { 2484 trad_enc_decrypt_update(&zip->tctx, 2485 compressed_buff, buff_remaining, 2486 zip->decrypted_ptr 2487 + zip->decrypted_bytes_remaining, 2488 buff_remaining); 2489 } else { 2490 size_t dsize = buff_remaining; 2491 archive_decrypto_aes_ctr_update( 2492 &zip->cctx, 2493 compressed_buff, buff_remaining, 2494 zip->decrypted_ptr 2495 + zip->decrypted_bytes_remaining, 2496 &dsize); 2497 } 2498 zip->decrypted_bytes_remaining += 2499 buff_remaining; 2500 } 2501 } 2502 bytes_avail = zip->decrypted_bytes_remaining; 2503 compressed_buff = (const char *)zip->decrypted_ptr; 2504 } 2505 2506 /* 2507 * A bug in zlib.h: stream.next_in should be marked 'const' 2508 * but isn't (the library never alters data through the 2509 * next_in pointer, only reads it). The result: this ugly 2510 * cast to remove 'const'. 2511 */ 2512 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 2513 zip->stream.avail_in = (uInt)bytes_avail; 2514 zip->stream.total_in = 0; 2515 zip->stream.next_out = zip->uncompressed_buffer; 2516 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 2517 zip->stream.total_out = 0; 2518 2519 r = inflate(&zip->stream, 0); 2520 switch (r) { 2521 case Z_OK: 2522 break; 2523 case Z_STREAM_END: 2524 zip->end_of_entry = 1; 2525 break; 2526 case Z_MEM_ERROR: 2527 archive_set_error(&a->archive, ENOMEM, 2528 "Out of memory for ZIP decompression"); 2529 return (ARCHIVE_FATAL); 2530 default: 2531 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2532 "ZIP decompression failed (%d)", r); 2533 return (ARCHIVE_FATAL); 2534 } 2535 2536 /* Consume as much as the compressor actually used. */ 2537 bytes_avail = zip->stream.total_in; 2538 if (zip->tctx_valid || zip->cctx_valid) { 2539 zip->decrypted_bytes_remaining -= bytes_avail; 2540 if (zip->decrypted_bytes_remaining == 0) 2541 zip->decrypted_ptr = zip->decrypted_buffer; 2542 else 2543 zip->decrypted_ptr += bytes_avail; 2544 } 2545 /* Calculate compressed data as much as we used.*/ 2546 if (zip->hctx_valid) 2547 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 2548 __archive_read_consume(a, bytes_avail); 2549 zip->entry_bytes_remaining -= bytes_avail; 2550 zip->entry_compressed_bytes_read += bytes_avail; 2551 2552 *size = zip->stream.total_out; 2553 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 2554 *buff = zip->uncompressed_buffer; 2555 2556 if (zip->end_of_entry && zip->hctx_valid) { 2557 r = check_authentication_code(a, NULL); 2558 if (r != ARCHIVE_OK) 2559 return (r); 2560 } 2561 2562 r = consume_optional_marker(a, zip); 2563 if (r != ARCHIVE_OK) 2564 return (r); 2565 2566 return (ARCHIVE_OK); 2567 } 2568 #endif 2569 2570 static int 2571 read_decryption_header(struct archive_read *a) 2572 { 2573 struct zip *zip = (struct zip *)(a->format->data); 2574 const char *p; 2575 unsigned int remaining_size; 2576 unsigned int ts; 2577 2578 /* 2579 * Read an initialization vector data field. 2580 */ 2581 p = __archive_read_ahead(a, 2, NULL); 2582 if (p == NULL) 2583 goto truncated; 2584 ts = zip->iv_size; 2585 zip->iv_size = archive_le16dec(p); 2586 __archive_read_consume(a, 2); 2587 if (ts < zip->iv_size) { 2588 free(zip->iv); 2589 zip->iv = NULL; 2590 } 2591 p = __archive_read_ahead(a, zip->iv_size, NULL); 2592 if (p == NULL) 2593 goto truncated; 2594 if (zip->iv == NULL) { 2595 zip->iv = malloc(zip->iv_size); 2596 if (zip->iv == NULL) 2597 goto nomem; 2598 } 2599 memcpy(zip->iv, p, zip->iv_size); 2600 __archive_read_consume(a, zip->iv_size); 2601 2602 /* 2603 * Read a size of remaining decryption header field. 2604 */ 2605 p = __archive_read_ahead(a, 14, NULL); 2606 if (p == NULL) 2607 goto truncated; 2608 remaining_size = archive_le32dec(p); 2609 if (remaining_size < 16 || remaining_size > (1 << 18)) 2610 goto corrupted; 2611 2612 /* Check if format version is supported. */ 2613 if (archive_le16dec(p+4) != 3) { 2614 archive_set_error(&a->archive, 2615 ARCHIVE_ERRNO_FILE_FORMAT, 2616 "Unsupported encryption format version: %u", 2617 archive_le16dec(p+4)); 2618 return (ARCHIVE_FAILED); 2619 } 2620 2621 /* 2622 * Read an encryption algorithm field. 2623 */ 2624 zip->alg_id = archive_le16dec(p+6); 2625 switch (zip->alg_id) { 2626 case 0x6601:/* DES */ 2627 case 0x6602:/* RC2 */ 2628 case 0x6603:/* 3DES 168 */ 2629 case 0x6609:/* 3DES 112 */ 2630 case 0x660E:/* AES 128 */ 2631 case 0x660F:/* AES 192 */ 2632 case 0x6610:/* AES 256 */ 2633 case 0x6702:/* RC2 (version >= 5.2) */ 2634 case 0x6720:/* Blowfish */ 2635 case 0x6721:/* Twofish */ 2636 case 0x6801:/* RC4 */ 2637 /* Supported encryption algorithm. */ 2638 break; 2639 default: 2640 archive_set_error(&a->archive, 2641 ARCHIVE_ERRNO_FILE_FORMAT, 2642 "Unknown encryption algorithm: %u", zip->alg_id); 2643 return (ARCHIVE_FAILED); 2644 } 2645 2646 /* 2647 * Read a bit length field. 2648 */ 2649 zip->bit_len = archive_le16dec(p+8); 2650 2651 /* 2652 * Read a flags field. 2653 */ 2654 zip->flags = archive_le16dec(p+10); 2655 switch (zip->flags & 0xf000) { 2656 case 0x0001: /* Password is required to decrypt. */ 2657 case 0x0002: /* Certificates only. */ 2658 case 0x0003: /* Password or certificate required to decrypt. */ 2659 break; 2660 default: 2661 archive_set_error(&a->archive, 2662 ARCHIVE_ERRNO_FILE_FORMAT, 2663 "Unknown encryption flag: %u", zip->flags); 2664 return (ARCHIVE_FAILED); 2665 } 2666 if ((zip->flags & 0xf000) == 0 || 2667 (zip->flags & 0xf000) == 0x4000) { 2668 archive_set_error(&a->archive, 2669 ARCHIVE_ERRNO_FILE_FORMAT, 2670 "Unknown encryption flag: %u", zip->flags); 2671 return (ARCHIVE_FAILED); 2672 } 2673 2674 /* 2675 * Read an encrypted random data field. 2676 */ 2677 ts = zip->erd_size; 2678 zip->erd_size = archive_le16dec(p+12); 2679 __archive_read_consume(a, 14); 2680 if ((zip->erd_size & 0xf) != 0 || 2681 (zip->erd_size + 16) > remaining_size || 2682 (zip->erd_size + 16) < zip->erd_size) 2683 goto corrupted; 2684 2685 if (ts < zip->erd_size) { 2686 free(zip->erd); 2687 zip->erd = NULL; 2688 } 2689 p = __archive_read_ahead(a, zip->erd_size, NULL); 2690 if (p == NULL) 2691 goto truncated; 2692 if (zip->erd == NULL) { 2693 zip->erd = malloc(zip->erd_size); 2694 if (zip->erd == NULL) 2695 goto nomem; 2696 } 2697 memcpy(zip->erd, p, zip->erd_size); 2698 __archive_read_consume(a, zip->erd_size); 2699 2700 /* 2701 * Read a reserved data field. 2702 */ 2703 p = __archive_read_ahead(a, 4, NULL); 2704 if (p == NULL) 2705 goto truncated; 2706 /* Reserved data size should be zero. */ 2707 if (archive_le32dec(p) != 0) 2708 goto corrupted; 2709 __archive_read_consume(a, 4); 2710 2711 /* 2712 * Read a password validation data field. 2713 */ 2714 p = __archive_read_ahead(a, 2, NULL); 2715 if (p == NULL) 2716 goto truncated; 2717 ts = zip->v_size; 2718 zip->v_size = archive_le16dec(p); 2719 __archive_read_consume(a, 2); 2720 if ((zip->v_size & 0x0f) != 0 || 2721 (zip->erd_size + zip->v_size + 16) > remaining_size || 2722 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 2723 goto corrupted; 2724 if (ts < zip->v_size) { 2725 free(zip->v_data); 2726 zip->v_data = NULL; 2727 } 2728 p = __archive_read_ahead(a, zip->v_size, NULL); 2729 if (p == NULL) 2730 goto truncated; 2731 if (zip->v_data == NULL) { 2732 zip->v_data = malloc(zip->v_size); 2733 if (zip->v_data == NULL) 2734 goto nomem; 2735 } 2736 memcpy(zip->v_data, p, zip->v_size); 2737 __archive_read_consume(a, zip->v_size); 2738 2739 p = __archive_read_ahead(a, 4, NULL); 2740 if (p == NULL) 2741 goto truncated; 2742 zip->v_crc32 = archive_le32dec(p); 2743 __archive_read_consume(a, 4); 2744 2745 /*return (ARCHIVE_OK); 2746 * This is not fully implemented yet.*/ 2747 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2748 "Encrypted file is unsupported"); 2749 return (ARCHIVE_FAILED); 2750 truncated: 2751 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2752 "Truncated ZIP file data"); 2753 return (ARCHIVE_FATAL); 2754 corrupted: 2755 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2756 "Corrupted ZIP file data"); 2757 return (ARCHIVE_FATAL); 2758 nomem: 2759 archive_set_error(&a->archive, ENOMEM, 2760 "No memory for ZIP decryption"); 2761 return (ARCHIVE_FATAL); 2762 } 2763 2764 static int 2765 zip_alloc_decryption_buffer(struct archive_read *a) 2766 { 2767 struct zip *zip = (struct zip *)(a->format->data); 2768 size_t bs = 256 * 1024; 2769 2770 if (zip->decrypted_buffer == NULL) { 2771 zip->decrypted_buffer_size = bs; 2772 zip->decrypted_buffer = malloc(bs); 2773 if (zip->decrypted_buffer == NULL) { 2774 archive_set_error(&a->archive, ENOMEM, 2775 "No memory for ZIP decryption"); 2776 return (ARCHIVE_FATAL); 2777 } 2778 } 2779 zip->decrypted_ptr = zip->decrypted_buffer; 2780 return (ARCHIVE_OK); 2781 } 2782 2783 static int 2784 init_traditional_PKWARE_decryption(struct archive_read *a) 2785 { 2786 struct zip *zip = (struct zip *)(a->format->data); 2787 const void *p; 2788 int retry; 2789 int r; 2790 2791 if (zip->tctx_valid) 2792 return (ARCHIVE_OK); 2793 2794 /* 2795 Read the 12 bytes encryption header stored at 2796 the start of the data area. 2797 */ 2798 #define ENC_HEADER_SIZE 12 2799 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2800 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 2801 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2802 "Truncated Zip encrypted body: only %jd bytes available", 2803 (intmax_t)zip->entry_bytes_remaining); 2804 return (ARCHIVE_FATAL); 2805 } 2806 2807 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 2808 if (p == NULL) { 2809 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2810 "Truncated ZIP file data"); 2811 return (ARCHIVE_FATAL); 2812 } 2813 2814 for (retry = 0;; retry++) { 2815 const char *passphrase; 2816 uint8_t crcchk; 2817 2818 passphrase = __archive_read_next_passphrase(a); 2819 if (passphrase == NULL) { 2820 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2821 (retry > 0)? 2822 "Incorrect passphrase": 2823 "Passphrase required for this entry"); 2824 return (ARCHIVE_FAILED); 2825 } 2826 2827 /* 2828 * Initialize ctx for Traditional PKWARE Decryption. 2829 */ 2830 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 2831 p, ENC_HEADER_SIZE, &crcchk); 2832 if (r == 0 && crcchk == zip->entry->decdat) 2833 break;/* The passphrase is OK. */ 2834 if (retry > 10000) { 2835 /* Avoid infinity loop. */ 2836 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2837 "Too many incorrect passphrases"); 2838 return (ARCHIVE_FAILED); 2839 } 2840 } 2841 2842 __archive_read_consume(a, ENC_HEADER_SIZE); 2843 zip->tctx_valid = 1; 2844 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 2845 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 2846 } 2847 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 2848 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 2849 zip->decrypted_bytes_remaining = 0; 2850 2851 return (zip_alloc_decryption_buffer(a)); 2852 #undef ENC_HEADER_SIZE 2853 } 2854 2855 static int 2856 init_WinZip_AES_decryption(struct archive_read *a) 2857 { 2858 struct zip *zip = (struct zip *)(a->format->data); 2859 const void *p; 2860 const uint8_t *pv; 2861 size_t key_len, salt_len; 2862 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 2863 int retry; 2864 int r; 2865 2866 if (zip->cctx_valid || zip->hctx_valid) 2867 return (ARCHIVE_OK); 2868 2869 switch (zip->entry->aes_extra.strength) { 2870 case 1: salt_len = 8; key_len = 16; break; 2871 case 2: salt_len = 12; key_len = 24; break; 2872 case 3: salt_len = 16; key_len = 32; break; 2873 default: goto corrupted; 2874 } 2875 p = __archive_read_ahead(a, salt_len + 2, NULL); 2876 if (p == NULL) 2877 goto truncated; 2878 2879 for (retry = 0;; retry++) { 2880 const char *passphrase; 2881 2882 passphrase = __archive_read_next_passphrase(a); 2883 if (passphrase == NULL) { 2884 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2885 (retry > 0)? 2886 "Incorrect passphrase": 2887 "Passphrase required for this entry"); 2888 return (ARCHIVE_FAILED); 2889 } 2890 memset(derived_key, 0, sizeof(derived_key)); 2891 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 2892 p, salt_len, 1000, derived_key, key_len * 2 + 2); 2893 if (r != 0) { 2894 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2895 "Decryption is unsupported due to lack of " 2896 "crypto library"); 2897 return (ARCHIVE_FAILED); 2898 } 2899 2900 /* Check password verification value. */ 2901 pv = ((const uint8_t *)p) + salt_len; 2902 if (derived_key[key_len * 2] == pv[0] && 2903 derived_key[key_len * 2 + 1] == pv[1]) 2904 break;/* The passphrase is OK. */ 2905 if (retry > 10000) { 2906 /* Avoid infinity loop. */ 2907 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2908 "Too many incorrect passphrases"); 2909 return (ARCHIVE_FAILED); 2910 } 2911 } 2912 2913 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 2914 if (r != 0) { 2915 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2916 "Decryption is unsupported due to lack of crypto library"); 2917 return (ARCHIVE_FAILED); 2918 } 2919 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 2920 if (r != 0) { 2921 archive_decrypto_aes_ctr_release(&zip->cctx); 2922 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2923 "Failed to initialize HMAC-SHA1"); 2924 return (ARCHIVE_FAILED); 2925 } 2926 zip->cctx_valid = zip->hctx_valid = 1; 2927 __archive_read_consume(a, salt_len + 2); 2928 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 2929 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2930 && zip->entry_bytes_remaining < 0) 2931 goto corrupted; 2932 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 2933 zip->decrypted_bytes_remaining = 0; 2934 2935 zip->entry->compression = zip->entry->aes_extra.compression; 2936 return (zip_alloc_decryption_buffer(a)); 2937 2938 truncated: 2939 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2940 "Truncated ZIP file data"); 2941 return (ARCHIVE_FATAL); 2942 corrupted: 2943 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2944 "Corrupted ZIP file data"); 2945 return (ARCHIVE_FATAL); 2946 } 2947 2948 static int 2949 archive_read_format_zip_read_data(struct archive_read *a, 2950 const void **buff, size_t *size, int64_t *offset) 2951 { 2952 int r; 2953 struct zip *zip = (struct zip *)(a->format->data); 2954 2955 if (zip->has_encrypted_entries == 2956 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 2957 zip->has_encrypted_entries = 0; 2958 } 2959 2960 *offset = zip->entry_uncompressed_bytes_read; 2961 *size = 0; 2962 *buff = NULL; 2963 2964 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 2965 if (zip->end_of_entry) 2966 return (ARCHIVE_EOF); 2967 2968 /* Return EOF immediately if this is a non-regular file. */ 2969 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 2970 return (ARCHIVE_EOF); 2971 2972 __archive_read_consume(a, zip->unconsumed); 2973 zip->unconsumed = 0; 2974 2975 if (zip->init_decryption) { 2976 zip->has_encrypted_entries = 1; 2977 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2978 r = read_decryption_header(a); 2979 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2980 r = init_WinZip_AES_decryption(a); 2981 else 2982 r = init_traditional_PKWARE_decryption(a); 2983 if (r != ARCHIVE_OK) 2984 return (r); 2985 zip->init_decryption = 0; 2986 } 2987 2988 switch(zip->entry->compression) { 2989 case 0: /* No compression. */ 2990 r = zip_read_data_none(a, buff, size, offset); 2991 break; 2992 #ifdef HAVE_BZLIB_H 2993 case 12: /* ZIPx bzip2 compression. */ 2994 r = zip_read_data_zipx_bzip2(a, buff, size, offset); 2995 break; 2996 #endif 2997 #if HAVE_LZMA_H && HAVE_LIBLZMA 2998 case 14: /* ZIPx LZMA compression. */ 2999 r = zip_read_data_zipx_lzma_alone(a, buff, size, offset); 3000 break; 3001 case 95: /* ZIPx XZ compression. */ 3002 r = zip_read_data_zipx_xz(a, buff, size, offset); 3003 break; 3004 #endif 3005 #if HAVE_ZSTD_H && HAVE_LIBZSTD 3006 case 93: /* ZIPx Zstd compression. */ 3007 r = zip_read_data_zipx_zstd(a, buff, size, offset); 3008 break; 3009 #endif 3010 /* PPMd support is built-in, so we don't need any #if guards. */ 3011 case 98: /* ZIPx PPMd compression. */ 3012 r = zip_read_data_zipx_ppmd(a, buff, size, offset); 3013 break; 3014 3015 #ifdef HAVE_ZLIB_H 3016 case 8: /* Deflate compression. */ 3017 r = zip_read_data_deflate(a, buff, size, offset); 3018 break; 3019 #endif 3020 default: /* Unsupported compression. */ 3021 /* Return a warning. */ 3022 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3023 "Unsupported ZIP compression method (%d: %s)", 3024 zip->entry->compression, compression_name(zip->entry->compression)); 3025 /* We can't decompress this entry, but we will 3026 * be able to skip() it and try the next entry. */ 3027 return (ARCHIVE_FAILED); 3028 break; 3029 } 3030 if (r != ARCHIVE_OK) 3031 return (r); 3032 /* Update checksum */ 3033 if (*size) 3034 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 3035 (unsigned)*size); 3036 /* If we hit the end, swallow any end-of-data marker. */ 3037 if (zip->end_of_entry) { 3038 /* Check file size, CRC against these values. */ 3039 if (zip->entry->compressed_size != 3040 zip->entry_compressed_bytes_read) { 3041 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3042 "ZIP compressed data is wrong size " 3043 "(read %jd, expected %jd)", 3044 (intmax_t)zip->entry_compressed_bytes_read, 3045 (intmax_t)zip->entry->compressed_size); 3046 return (ARCHIVE_WARN); 3047 } 3048 /* Size field only stores the lower 32 bits of the actual 3049 * size. */ 3050 if ((zip->entry->uncompressed_size & UINT32_MAX) 3051 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 3052 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3053 "ZIP uncompressed data is wrong size " 3054 "(read %jd, expected %jd)\n", 3055 (intmax_t)zip->entry_uncompressed_bytes_read, 3056 (intmax_t)zip->entry->uncompressed_size); 3057 return (ARCHIVE_WARN); 3058 } 3059 /* Check computed CRC against header */ 3060 if ((!zip->hctx_valid || 3061 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 3062 zip->entry->crc32 != zip->entry_crc32 3063 && !zip->ignore_crc32) { 3064 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3065 "ZIP bad CRC: 0x%lx should be 0x%lx", 3066 (unsigned long)zip->entry_crc32, 3067 (unsigned long)zip->entry->crc32); 3068 return (ARCHIVE_WARN); 3069 } 3070 } 3071 3072 return (ARCHIVE_OK); 3073 } 3074 3075 static int 3076 archive_read_format_zip_cleanup(struct archive_read *a) 3077 { 3078 struct zip *zip; 3079 struct zip_entry *zip_entry, *next_zip_entry; 3080 3081 zip = (struct zip *)(a->format->data); 3082 3083 #ifdef HAVE_ZLIB_H 3084 if (zip->stream_valid) 3085 inflateEnd(&zip->stream); 3086 #endif 3087 3088 #if HAVE_LZMA_H && HAVE_LIBLZMA 3089 if (zip->zipx_lzma_valid) { 3090 lzma_end(&zip->zipx_lzma_stream); 3091 } 3092 #endif 3093 3094 #ifdef HAVE_BZLIB_H 3095 if (zip->bzstream_valid) { 3096 BZ2_bzDecompressEnd(&zip->bzstream); 3097 } 3098 #endif 3099 3100 #if HAVE_ZSTD_H && HAVE_LIBZSTD 3101 if (zip->zstdstream_valid) { 3102 ZSTD_freeDStream(zip->zstdstream); 3103 } 3104 #endif 3105 3106 free(zip->uncompressed_buffer); 3107 3108 if (zip->ppmd8_valid) 3109 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 3110 3111 if (zip->zip_entries) { 3112 zip_entry = zip->zip_entries; 3113 while (zip_entry != NULL) { 3114 next_zip_entry = zip_entry->next; 3115 archive_string_free(&zip_entry->rsrcname); 3116 free(zip_entry); 3117 zip_entry = next_zip_entry; 3118 } 3119 } 3120 free(zip->decrypted_buffer); 3121 if (zip->cctx_valid) 3122 archive_decrypto_aes_ctr_release(&zip->cctx); 3123 if (zip->hctx_valid) 3124 archive_hmac_sha1_cleanup(&zip->hctx); 3125 free(zip->iv); 3126 free(zip->erd); 3127 free(zip->v_data); 3128 archive_string_free(&zip->format_name); 3129 free(zip); 3130 (a->format->data) = NULL; 3131 return (ARCHIVE_OK); 3132 } 3133 3134 static int 3135 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 3136 { 3137 if (_a && _a->format) { 3138 struct zip * zip = (struct zip *)_a->format->data; 3139 if (zip) { 3140 return zip->has_encrypted_entries; 3141 } 3142 } 3143 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3144 } 3145 3146 static int 3147 archive_read_format_zip_options(struct archive_read *a, 3148 const char *key, const char *val) 3149 { 3150 struct zip *zip; 3151 int ret = ARCHIVE_FAILED; 3152 3153 zip = (struct zip *)(a->format->data); 3154 if (strcmp(key, "compat-2x") == 0) { 3155 /* Handle filenames as libarchive 2.x */ 3156 zip->init_default_conversion = (val != NULL) ? 1 : 0; 3157 return (ARCHIVE_OK); 3158 } else if (strcmp(key, "hdrcharset") == 0) { 3159 if (val == NULL || val[0] == 0) 3160 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 3161 "zip: hdrcharset option needs a character-set name" 3162 ); 3163 else { 3164 zip->sconv = archive_string_conversion_from_charset( 3165 &a->archive, val, 0); 3166 if (zip->sconv != NULL) { 3167 if (strcmp(val, "UTF-8") == 0) 3168 zip->sconv_utf8 = zip->sconv; 3169 ret = ARCHIVE_OK; 3170 } else 3171 ret = ARCHIVE_FATAL; 3172 } 3173 return (ret); 3174 } else if (strcmp(key, "ignorecrc32") == 0) { 3175 /* Mostly useful for testing. */ 3176 if (val == NULL || val[0] == 0) { 3177 zip->crc32func = real_crc32; 3178 zip->ignore_crc32 = 0; 3179 } else { 3180 zip->crc32func = fake_crc32; 3181 zip->ignore_crc32 = 1; 3182 } 3183 return (ARCHIVE_OK); 3184 } else if (strcmp(key, "mac-ext") == 0) { 3185 zip->process_mac_extensions = (val != NULL && val[0] != 0); 3186 return (ARCHIVE_OK); 3187 } 3188 3189 /* Note: The "warn" return is just to inform the options 3190 * supervisor that we didn't handle it. It will generate 3191 * a suitable error if no one used this option. */ 3192 return (ARCHIVE_WARN); 3193 } 3194 3195 int 3196 archive_read_support_format_zip(struct archive *a) 3197 { 3198 int r; 3199 r = archive_read_support_format_zip_streamable(a); 3200 if (r != ARCHIVE_OK) 3201 return r; 3202 return (archive_read_support_format_zip_seekable(a)); 3203 } 3204 3205 /* ------------------------------------------------------------------------ */ 3206 3207 /* 3208 * Streaming-mode support 3209 */ 3210 3211 3212 static int 3213 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 3214 { 3215 (void)a; /* UNUSED */ 3216 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 3217 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 3218 } 3219 3220 static int 3221 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 3222 { 3223 const char *p; 3224 3225 (void)best_bid; /* UNUSED */ 3226 3227 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 3228 return (-1); 3229 3230 /* 3231 * Bid of 29 here comes from: 3232 * + 16 bits for "PK", 3233 * + next 16-bit field has 6 options so contributes 3234 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 3235 * 3236 * So we've effectively verified ~29 total bits of check data. 3237 */ 3238 if (p[0] == 'P' && p[1] == 'K') { 3239 if ((p[2] == '\001' && p[3] == '\002') 3240 || (p[2] == '\003' && p[3] == '\004') 3241 || (p[2] == '\005' && p[3] == '\006') 3242 || (p[2] == '\006' && p[3] == '\006') 3243 || (p[2] == '\007' && p[3] == '\010') 3244 || (p[2] == '0' && p[3] == '0')) 3245 return (29); 3246 } 3247 3248 /* TODO: It's worth looking ahead a little bit for a valid 3249 * PK signature. In particular, that would make it possible 3250 * to read some UUEncoded SFX files or SFX files coming from 3251 * a network socket. */ 3252 3253 return (0); 3254 } 3255 3256 static int 3257 archive_read_format_zip_streamable_read_header(struct archive_read *a, 3258 struct archive_entry *entry) 3259 { 3260 struct zip *zip; 3261 3262 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 3263 if (a->archive.archive_format_name == NULL) 3264 a->archive.archive_format_name = "ZIP"; 3265 3266 zip = (struct zip *)(a->format->data); 3267 3268 /* 3269 * It should be sufficient to call archive_read_next_header() for 3270 * a reader to determine if an entry is encrypted or not. If the 3271 * encryption of an entry is only detectable when calling 3272 * archive_read_data(), so be it. We'll do the same check there 3273 * as well. 3274 */ 3275 if (zip->has_encrypted_entries == 3276 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 3277 zip->has_encrypted_entries = 0; 3278 3279 /* Make sure we have a zip_entry structure to use. */ 3280 if (zip->zip_entries == NULL) { 3281 zip->zip_entries = malloc(sizeof(struct zip_entry)); 3282 if (zip->zip_entries == NULL) { 3283 archive_set_error(&a->archive, ENOMEM, 3284 "Out of memory"); 3285 return ARCHIVE_FATAL; 3286 } 3287 } 3288 zip->entry = zip->zip_entries; 3289 memset(zip->entry, 0, sizeof(struct zip_entry)); 3290 3291 if (zip->cctx_valid) 3292 archive_decrypto_aes_ctr_release(&zip->cctx); 3293 if (zip->hctx_valid) 3294 archive_hmac_sha1_cleanup(&zip->hctx); 3295 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3296 __archive_read_reset_passphrase(a); 3297 3298 /* Search ahead for the next local file header. */ 3299 __archive_read_consume(a, zip->unconsumed); 3300 zip->unconsumed = 0; 3301 for (;;) { 3302 int64_t skipped = 0; 3303 const char *p, *end; 3304 ssize_t bytes; 3305 3306 p = __archive_read_ahead(a, 4, &bytes); 3307 if (p == NULL) 3308 return (ARCHIVE_FATAL); 3309 end = p + bytes; 3310 3311 while (p + 4 <= end) { 3312 if (p[0] == 'P' && p[1] == 'K') { 3313 if (p[2] == '\003' && p[3] == '\004') { 3314 /* Regular file entry. */ 3315 __archive_read_consume(a, skipped); 3316 return zip_read_local_file_header(a, 3317 entry, zip); 3318 } 3319 3320 /* 3321 * TODO: We cannot restore permissions 3322 * based only on the local file headers. 3323 * Consider scanning the central 3324 * directory and returning additional 3325 * entries for at least directories. 3326 * This would allow us to properly set 3327 * directory permissions. 3328 * 3329 * This won't help us fix symlinks 3330 * and may not help with regular file 3331 * permissions, either. <sigh> 3332 */ 3333 if (p[2] == '\001' && p[3] == '\002') { 3334 return (ARCHIVE_EOF); 3335 } 3336 3337 /* End of central directory? Must be an 3338 * empty archive. */ 3339 if ((p[2] == '\005' && p[3] == '\006') 3340 || (p[2] == '\006' && p[3] == '\006')) 3341 return (ARCHIVE_EOF); 3342 } 3343 ++p; 3344 ++skipped; 3345 } 3346 __archive_read_consume(a, skipped); 3347 } 3348 } 3349 3350 static int 3351 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 3352 { 3353 struct zip *zip; 3354 int64_t bytes_skipped; 3355 3356 zip = (struct zip *)(a->format->data); 3357 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 3358 zip->unconsumed = 0; 3359 if (bytes_skipped < 0) 3360 return (ARCHIVE_FATAL); 3361 3362 /* If we've already read to end of data, we're done. */ 3363 if (zip->end_of_entry) 3364 return (ARCHIVE_OK); 3365 3366 /* So we know we're streaming... */ 3367 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 3368 || zip->entry->compressed_size > 0) { 3369 /* We know the compressed length, so we can just skip. */ 3370 bytes_skipped = __archive_read_consume(a, 3371 zip->entry_bytes_remaining); 3372 if (bytes_skipped < 0) 3373 return (ARCHIVE_FATAL); 3374 return (ARCHIVE_OK); 3375 } 3376 3377 if (zip->init_decryption) { 3378 int r; 3379 3380 zip->has_encrypted_entries = 1; 3381 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 3382 r = read_decryption_header(a); 3383 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 3384 r = init_WinZip_AES_decryption(a); 3385 else 3386 r = init_traditional_PKWARE_decryption(a); 3387 if (r != ARCHIVE_OK) 3388 return (r); 3389 zip->init_decryption = 0; 3390 } 3391 3392 /* We're streaming and we don't know the length. */ 3393 /* If the body is compressed and we know the format, we can 3394 * find an exact end-of-entry by decompressing it. */ 3395 switch (zip->entry->compression) { 3396 #ifdef HAVE_ZLIB_H 3397 case 8: /* Deflate compression. */ 3398 while (!zip->end_of_entry) { 3399 int64_t offset = 0; 3400 const void *buff = NULL; 3401 size_t size = 0; 3402 int r; 3403 r = zip_read_data_deflate(a, &buff, &size, &offset); 3404 if (r != ARCHIVE_OK) 3405 return (r); 3406 } 3407 return ARCHIVE_OK; 3408 #endif 3409 default: /* Uncompressed or unknown. */ 3410 /* Scan for a PK\007\010 signature. */ 3411 for (;;) { 3412 const char *p, *buff; 3413 ssize_t bytes_avail; 3414 buff = __archive_read_ahead(a, 16, &bytes_avail); 3415 if (bytes_avail < 16) { 3416 archive_set_error(&a->archive, 3417 ARCHIVE_ERRNO_FILE_FORMAT, 3418 "Truncated ZIP file data"); 3419 return (ARCHIVE_FATAL); 3420 } 3421 p = buff; 3422 while (p <= buff + bytes_avail - 16) { 3423 if (p[3] == 'P') { p += 3; } 3424 else if (p[3] == 'K') { p += 2; } 3425 else if (p[3] == '\007') { p += 1; } 3426 else if (p[3] == '\010' && p[2] == '\007' 3427 && p[1] == 'K' && p[0] == 'P') { 3428 if (zip->entry->flags & LA_USED_ZIP64) 3429 __archive_read_consume(a, 3430 p - buff + 24); 3431 else 3432 __archive_read_consume(a, 3433 p - buff + 16); 3434 return ARCHIVE_OK; 3435 } else { p += 4; } 3436 } 3437 __archive_read_consume(a, p - buff); 3438 } 3439 } 3440 } 3441 3442 int 3443 archive_read_support_format_zip_streamable(struct archive *_a) 3444 { 3445 struct archive_read *a = (struct archive_read *)_a; 3446 struct zip *zip; 3447 int r; 3448 3449 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3450 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 3451 3452 zip = (struct zip *)calloc(1, sizeof(*zip)); 3453 if (zip == NULL) { 3454 archive_set_error(&a->archive, ENOMEM, 3455 "Can't allocate zip data"); 3456 return (ARCHIVE_FATAL); 3457 } 3458 3459 /* Streamable reader doesn't support mac extensions. */ 3460 zip->process_mac_extensions = 0; 3461 3462 /* 3463 * Until enough data has been read, we cannot tell about 3464 * any encrypted entries yet. 3465 */ 3466 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3467 zip->crc32func = real_crc32; 3468 3469 r = __archive_read_register_format(a, 3470 zip, 3471 "zip", 3472 archive_read_format_zip_streamable_bid, 3473 archive_read_format_zip_options, 3474 archive_read_format_zip_streamable_read_header, 3475 archive_read_format_zip_read_data, 3476 archive_read_format_zip_read_data_skip_streamable, 3477 NULL, 3478 archive_read_format_zip_cleanup, 3479 archive_read_support_format_zip_capabilities_streamable, 3480 archive_read_format_zip_has_encrypted_entries); 3481 3482 if (r != ARCHIVE_OK) 3483 free(zip); 3484 return (ARCHIVE_OK); 3485 } 3486 3487 /* ------------------------------------------------------------------------ */ 3488 3489 /* 3490 * Seeking-mode support 3491 */ 3492 3493 static int 3494 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 3495 { 3496 (void)a; /* UNUSED */ 3497 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 3498 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 3499 } 3500 3501 /* 3502 * TODO: This is a performance sink because it forces the read core to 3503 * drop buffered data from the start of file, which will then have to 3504 * be re-read again if this bidder loses. 3505 * 3506 * We workaround this a little by passing in the best bid so far so 3507 * that later bidders can do nothing if they know they'll never 3508 * outbid. But we can certainly do better... 3509 */ 3510 static int 3511 read_eocd(struct zip *zip, const char *p, int64_t current_offset) 3512 { 3513 uint16_t disk_num; 3514 uint32_t cd_size, cd_offset; 3515 3516 disk_num = archive_le16dec(p + 4); 3517 cd_size = archive_le32dec(p + 12); 3518 cd_offset = archive_le32dec(p + 16); 3519 3520 /* Sanity-check the EOCD we've found. */ 3521 3522 /* This must be the first volume. */ 3523 if (disk_num != 0) 3524 return 0; 3525 /* Central directory must be on this volume. */ 3526 if (disk_num != archive_le16dec(p + 6)) 3527 return 0; 3528 /* All central directory entries must be on this volume. */ 3529 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 3530 return 0; 3531 /* Central directory can't extend beyond start of EOCD record. */ 3532 if (cd_offset + cd_size > current_offset) 3533 return 0; 3534 3535 /* Save the central directory location for later use. */ 3536 zip->central_directory_offset = cd_offset; 3537 zip->central_directory_offset_adjusted = current_offset - cd_size; 3538 3539 /* This is just a tiny bit higher than the maximum 3540 returned by the streaming Zip bidder. This ensures 3541 that the more accurate seeking Zip parser wins 3542 whenever seek is available. */ 3543 return 32; 3544 } 3545 3546 /* 3547 * Examine Zip64 EOCD locator: If it's valid, store the information 3548 * from it. 3549 */ 3550 static int 3551 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 3552 { 3553 int64_t eocd64_offset; 3554 int64_t eocd64_size; 3555 3556 /* Sanity-check the locator record. */ 3557 3558 /* Central dir must be on first volume. */ 3559 if (archive_le32dec(p + 4) != 0) 3560 return 0; 3561 /* Must be only a single volume. */ 3562 if (archive_le32dec(p + 16) != 1) 3563 return 0; 3564 3565 /* Find the Zip64 EOCD record. */ 3566 eocd64_offset = archive_le64dec(p + 8); 3567 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 3568 return 0; 3569 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 3570 return 0; 3571 /* Make sure we can read all of it. */ 3572 eocd64_size = archive_le64dec(p + 4) + 12; 3573 if (eocd64_size < 56 || eocd64_size > 16384) 3574 return 0; 3575 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 3576 return 0; 3577 3578 /* Sanity-check the EOCD64 */ 3579 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 3580 return 0; 3581 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 3582 return 0; 3583 /* CD can't be split. */ 3584 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 3585 return 0; 3586 3587 /* Save the central directory offset for later use. */ 3588 zip->central_directory_offset = archive_le64dec(p + 48); 3589 /* TODO: Needs scanning backwards to find the eocd64 instead of assuming */ 3590 zip->central_directory_offset_adjusted = zip->central_directory_offset; 3591 3592 return 32; 3593 } 3594 3595 static int 3596 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 3597 { 3598 struct zip *zip = (struct zip *)a->format->data; 3599 int64_t file_size, current_offset; 3600 const char *p; 3601 int i, tail; 3602 3603 /* If someone has already bid more than 32, then avoid 3604 trashing the look-ahead buffers with a seek. */ 3605 if (best_bid > 32) 3606 return (-1); 3607 3608 file_size = __archive_read_seek(a, 0, SEEK_END); 3609 if (file_size <= 0) 3610 return 0; 3611 3612 /* Search last 16k of file for end-of-central-directory 3613 * record (which starts with PK\005\006) */ 3614 tail = (int)zipmin(1024 * 16, file_size); 3615 current_offset = __archive_read_seek(a, -tail, SEEK_END); 3616 if (current_offset < 0) 3617 return 0; 3618 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 3619 return 0; 3620 /* Boyer-Moore search backwards from the end, since we want 3621 * to match the last EOCD in the file (there can be more than 3622 * one if there is an uncompressed Zip archive as a member 3623 * within this Zip archive). */ 3624 for (i = tail - 22; i > 0;) { 3625 switch (p[i]) { 3626 case 'P': 3627 if (memcmp(p + i, "PK\005\006", 4) == 0) { 3628 int ret = read_eocd(zip, p + i, 3629 current_offset + i); 3630 /* Zip64 EOCD locator precedes 3631 * regular EOCD if present. */ 3632 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 3633 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); 3634 if (ret_zip64 > ret) 3635 ret = ret_zip64; 3636 } 3637 return (ret); 3638 } 3639 i -= 4; 3640 break; 3641 case 'K': i -= 1; break; 3642 case 005: i -= 2; break; 3643 case 006: i -= 3; break; 3644 default: i -= 4; break; 3645 } 3646 } 3647 return 0; 3648 } 3649 3650 /* The red-black trees are only used in seeking mode to manage 3651 * the in-memory copy of the central directory. */ 3652 3653 static int 3654 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 3655 { 3656 const struct zip_entry *e1 = (const struct zip_entry *)n1; 3657 const struct zip_entry *e2 = (const struct zip_entry *)n2; 3658 3659 if (e1->local_header_offset > e2->local_header_offset) 3660 return -1; 3661 if (e1->local_header_offset < e2->local_header_offset) 3662 return 1; 3663 return 0; 3664 } 3665 3666 static int 3667 cmp_key(const struct archive_rb_node *n, const void *key) 3668 { 3669 /* This function won't be called */ 3670 (void)n; /* UNUSED */ 3671 (void)key; /* UNUSED */ 3672 return 1; 3673 } 3674 3675 static const struct archive_rb_tree_ops rb_ops = { 3676 &cmp_node, &cmp_key 3677 }; 3678 3679 static int 3680 rsrc_cmp_node(const struct archive_rb_node *n1, 3681 const struct archive_rb_node *n2) 3682 { 3683 const struct zip_entry *e1 = (const struct zip_entry *)n1; 3684 const struct zip_entry *e2 = (const struct zip_entry *)n2; 3685 3686 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 3687 } 3688 3689 static int 3690 rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 3691 { 3692 const struct zip_entry *e = (const struct zip_entry *)n; 3693 return (strcmp((const char *)key, e->rsrcname.s)); 3694 } 3695 3696 static const struct archive_rb_tree_ops rb_rsrc_ops = { 3697 &rsrc_cmp_node, &rsrc_cmp_key 3698 }; 3699 3700 static const char * 3701 rsrc_basename(const char *name, size_t name_length) 3702 { 3703 const char *s, *r; 3704 3705 r = s = name; 3706 for (;;) { 3707 s = memchr(s, '/', name_length - (s - name)); 3708 if (s == NULL) 3709 break; 3710 r = ++s; 3711 } 3712 return (r); 3713 } 3714 3715 static void 3716 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 3717 { 3718 struct archive_string str; 3719 struct zip_entry *dir; 3720 char *s; 3721 3722 archive_string_init(&str); 3723 archive_strncpy(&str, name, name_length); 3724 for (;;) { 3725 s = strrchr(str.s, '/'); 3726 if (s == NULL) 3727 break; 3728 *s = '\0'; 3729 /* Transfer the parent directory from zip->tree_rsrc RB 3730 * tree to zip->tree RB tree to expose. */ 3731 dir = (struct zip_entry *) 3732 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 3733 if (dir == NULL) 3734 break; 3735 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 3736 archive_string_free(&dir->rsrcname); 3737 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 3738 } 3739 archive_string_free(&str); 3740 } 3741 3742 static int 3743 slurp_central_directory(struct archive_read *a, struct archive_entry* entry, 3744 struct zip *zip) 3745 { 3746 ssize_t i; 3747 unsigned found; 3748 int64_t correction; 3749 ssize_t bytes_avail; 3750 const char *p; 3751 3752 /* 3753 * Find the start of the central directory. The end-of-CD 3754 * record has our starting point, but there are lots of 3755 * Zip archives which have had other data prepended to the 3756 * file, which makes the recorded offsets all too small. 3757 * So we search forward from the specified offset until we 3758 * find the real start of the central directory. Then we 3759 * know the correction we need to apply to account for leading 3760 * padding. 3761 */ 3762 if (__archive_read_seek(a, zip->central_directory_offset_adjusted, SEEK_SET) 3763 < 0) 3764 return ARCHIVE_FATAL; 3765 3766 found = 0; 3767 while (!found) { 3768 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 3769 return ARCHIVE_FATAL; 3770 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 3771 switch (p[i + 3]) { 3772 case 'P': i += 3; break; 3773 case 'K': i += 2; break; 3774 case 001: i += 1; break; 3775 case 002: 3776 if (memcmp(p + i, "PK\001\002", 4) == 0) { 3777 p += i; 3778 found = 1; 3779 } else 3780 i += 4; 3781 break; 3782 case 005: i += 1; break; 3783 case 006: 3784 if (memcmp(p + i, "PK\005\006", 4) == 0) { 3785 p += i; 3786 found = 1; 3787 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 3788 p += i; 3789 found = 1; 3790 } else 3791 i += 1; 3792 break; 3793 default: i += 4; break; 3794 } 3795 } 3796 __archive_read_consume(a, i); 3797 } 3798 correction = archive_filter_bytes(&a->archive, 0) 3799 - zip->central_directory_offset; 3800 3801 __archive_rb_tree_init(&zip->tree, &rb_ops); 3802 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 3803 3804 zip->central_directory_entries_total = 0; 3805 while (1) { 3806 struct zip_entry *zip_entry; 3807 size_t filename_length, extra_length, comment_length; 3808 uint32_t external_attributes; 3809 const char *name, *r; 3810 3811 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 3812 return ARCHIVE_FATAL; 3813 if (memcmp(p, "PK\006\006", 4) == 0 3814 || memcmp(p, "PK\005\006", 4) == 0) { 3815 break; 3816 } else if (memcmp(p, "PK\001\002", 4) != 0) { 3817 archive_set_error(&a->archive, 3818 -1, "Invalid central directory signature"); 3819 return ARCHIVE_FATAL; 3820 } 3821 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 3822 return ARCHIVE_FATAL; 3823 3824 zip_entry = calloc(1, sizeof(struct zip_entry)); 3825 if (zip_entry == NULL) { 3826 archive_set_error(&a->archive, ENOMEM, 3827 "Can't allocate zip entry"); 3828 return ARCHIVE_FATAL; 3829 } 3830 zip_entry->next = zip->zip_entries; 3831 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 3832 zip->zip_entries = zip_entry; 3833 zip->central_directory_entries_total++; 3834 3835 /* version = p[4]; */ 3836 zip_entry->system = p[5]; 3837 /* version_required = archive_le16dec(p + 6); */ 3838 zip_entry->zip_flags = archive_le16dec(p + 8); 3839 if (zip_entry->zip_flags 3840 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 3841 zip->has_encrypted_entries = 1; 3842 } 3843 zip_entry->compression = (char)archive_le16dec(p + 10); 3844 zip_entry->mtime = zip_time(p + 12); 3845 zip_entry->crc32 = archive_le32dec(p + 16); 3846 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 3847 zip_entry->decdat = p[13]; 3848 else 3849 zip_entry->decdat = p[19]; 3850 zip_entry->compressed_size = archive_le32dec(p + 20); 3851 zip_entry->uncompressed_size = archive_le32dec(p + 24); 3852 filename_length = archive_le16dec(p + 28); 3853 extra_length = archive_le16dec(p + 30); 3854 comment_length = archive_le16dec(p + 32); 3855 /* disk_start = archive_le16dec(p + 34); 3856 * Better be zero. 3857 * internal_attributes = archive_le16dec(p + 36); 3858 * text bit */ 3859 external_attributes = archive_le32dec(p + 38); 3860 zip_entry->local_header_offset = 3861 archive_le32dec(p + 42) + correction; 3862 3863 /* If we can't guess the mode, leave it zero here; 3864 when we read the local file header we might get 3865 more information. */ 3866 if (zip_entry->system == 3) { 3867 zip_entry->mode = external_attributes >> 16; 3868 } else if (zip_entry->system == 0) { 3869 // Interpret MSDOS directory bit 3870 if (0x10 == (external_attributes & 0x10)) { 3871 zip_entry->mode = AE_IFDIR | 0775; 3872 } else { 3873 zip_entry->mode = AE_IFREG | 0664; 3874 } 3875 if (0x01 == (external_attributes & 0x01)) { 3876 // Read-only bit; strip write permissions 3877 zip_entry->mode &= 0555; 3878 } 3879 } else { 3880 zip_entry->mode = 0; 3881 } 3882 3883 /* We're done with the regular data; get the filename and 3884 * extra data. */ 3885 __archive_read_consume(a, 46); 3886 p = __archive_read_ahead(a, filename_length + extra_length, 3887 NULL); 3888 if (p == NULL) { 3889 archive_set_error(&a->archive, 3890 ARCHIVE_ERRNO_FILE_FORMAT, 3891 "Truncated ZIP file header"); 3892 return ARCHIVE_FATAL; 3893 } 3894 if (ARCHIVE_OK != process_extra(a, entry, p + filename_length, 3895 extra_length, zip_entry)) { 3896 return ARCHIVE_FATAL; 3897 } 3898 3899 /* 3900 * Mac resource fork files are stored under the 3901 * "__MACOSX/" directory, so we should check if 3902 * it is. 3903 */ 3904 if (!zip->process_mac_extensions) { 3905 /* Treat every entry as a regular entry. */ 3906 __archive_rb_tree_insert_node(&zip->tree, 3907 &zip_entry->node); 3908 } else { 3909 name = p; 3910 r = rsrc_basename(name, filename_length); 3911 if (filename_length >= 9 && 3912 strncmp("__MACOSX/", name, 9) == 0) { 3913 /* If this file is not a resource fork nor 3914 * a directory. We should treat it as a non 3915 * resource fork file to expose it. */ 3916 if (name[filename_length-1] != '/' && 3917 (r - name < 3 || r[0] != '.' || 3918 r[1] != '_')) { 3919 __archive_rb_tree_insert_node( 3920 &zip->tree, &zip_entry->node); 3921 /* Expose its parent directories. */ 3922 expose_parent_dirs(zip, name, 3923 filename_length); 3924 } else { 3925 /* This file is a resource fork file or 3926 * a directory. */ 3927 archive_strncpy(&(zip_entry->rsrcname), 3928 name, filename_length); 3929 __archive_rb_tree_insert_node( 3930 &zip->tree_rsrc, &zip_entry->node); 3931 } 3932 } else { 3933 /* Generate resource fork name to find its 3934 * resource file at zip->tree_rsrc. */ 3935 archive_strcpy(&(zip_entry->rsrcname), 3936 "__MACOSX/"); 3937 archive_strncat(&(zip_entry->rsrcname), 3938 name, r - name); 3939 archive_strcat(&(zip_entry->rsrcname), "._"); 3940 archive_strncat(&(zip_entry->rsrcname), 3941 name + (r - name), 3942 filename_length - (r - name)); 3943 /* Register an entry to RB tree to sort it by 3944 * file offset. */ 3945 __archive_rb_tree_insert_node(&zip->tree, 3946 &zip_entry->node); 3947 } 3948 } 3949 3950 /* Skip the comment too ... */ 3951 __archive_read_consume(a, 3952 filename_length + extra_length + comment_length); 3953 } 3954 3955 return ARCHIVE_OK; 3956 } 3957 3958 static ssize_t 3959 zip_get_local_file_header_size(struct archive_read *a, size_t extra) 3960 { 3961 const char *p; 3962 ssize_t filename_length, extra_length; 3963 3964 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 3965 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3966 "Truncated ZIP file header"); 3967 return (ARCHIVE_WARN); 3968 } 3969 p += extra; 3970 3971 if (memcmp(p, "PK\003\004", 4) != 0) { 3972 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 3973 return ARCHIVE_WARN; 3974 } 3975 filename_length = archive_le16dec(p + 26); 3976 extra_length = archive_le16dec(p + 28); 3977 3978 return (30 + filename_length + extra_length); 3979 } 3980 3981 static int 3982 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 3983 struct zip_entry *rsrc) 3984 { 3985 struct zip *zip = (struct zip *)a->format->data; 3986 unsigned char *metadata, *mp; 3987 int64_t offset = archive_filter_bytes(&a->archive, 0); 3988 size_t remaining_bytes, metadata_bytes; 3989 ssize_t hsize; 3990 int ret = ARCHIVE_OK, eof; 3991 3992 switch(rsrc->compression) { 3993 case 0: /* No compression. */ 3994 if (rsrc->uncompressed_size != rsrc->compressed_size) { 3995 archive_set_error(&a->archive, 3996 ARCHIVE_ERRNO_FILE_FORMAT, 3997 "Malformed OS X metadata entry: " 3998 "inconsistent size"); 3999 return (ARCHIVE_FATAL); 4000 } 4001 #ifdef HAVE_ZLIB_H 4002 case 8: /* Deflate compression. */ 4003 #endif 4004 break; 4005 default: /* Unsupported compression. */ 4006 /* Return a warning. */ 4007 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4008 "Unsupported ZIP compression method (%s)", 4009 compression_name(rsrc->compression)); 4010 /* We can't decompress this entry, but we will 4011 * be able to skip() it and try the next entry. */ 4012 return (ARCHIVE_WARN); 4013 } 4014 4015 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 4016 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4017 "Mac metadata is too large: %jd > 4M bytes", 4018 (intmax_t)rsrc->uncompressed_size); 4019 return (ARCHIVE_WARN); 4020 } 4021 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 4022 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 4023 "Mac metadata is too large: %jd > 4M bytes", 4024 (intmax_t)rsrc->compressed_size); 4025 return (ARCHIVE_WARN); 4026 } 4027 4028 metadata = malloc((size_t)rsrc->uncompressed_size); 4029 if (metadata == NULL) { 4030 archive_set_error(&a->archive, ENOMEM, 4031 "Can't allocate memory for Mac metadata"); 4032 return (ARCHIVE_FATAL); 4033 } 4034 4035 if (offset < rsrc->local_header_offset) 4036 __archive_read_consume(a, rsrc->local_header_offset - offset); 4037 else if (offset != rsrc->local_header_offset) { 4038 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 4039 } 4040 4041 hsize = zip_get_local_file_header_size(a, 0); 4042 __archive_read_consume(a, hsize); 4043 4044 remaining_bytes = (size_t)rsrc->compressed_size; 4045 metadata_bytes = (size_t)rsrc->uncompressed_size; 4046 mp = metadata; 4047 eof = 0; 4048 while (!eof && remaining_bytes) { 4049 const unsigned char *p; 4050 ssize_t bytes_avail; 4051 size_t bytes_used; 4052 4053 p = __archive_read_ahead(a, 1, &bytes_avail); 4054 if (p == NULL) { 4055 archive_set_error(&a->archive, 4056 ARCHIVE_ERRNO_FILE_FORMAT, 4057 "Truncated ZIP file header"); 4058 ret = ARCHIVE_WARN; 4059 goto exit_mac_metadata; 4060 } 4061 if ((size_t)bytes_avail > remaining_bytes) 4062 bytes_avail = remaining_bytes; 4063 switch(rsrc->compression) { 4064 case 0: /* No compression. */ 4065 if ((size_t)bytes_avail > metadata_bytes) 4066 bytes_avail = metadata_bytes; 4067 memcpy(mp, p, bytes_avail); 4068 bytes_used = (size_t)bytes_avail; 4069 metadata_bytes -= bytes_used; 4070 mp += bytes_used; 4071 if (metadata_bytes == 0) 4072 eof = 1; 4073 break; 4074 #ifdef HAVE_ZLIB_H 4075 case 8: /* Deflate compression. */ 4076 { 4077 int r; 4078 4079 ret = zip_deflate_init(a, zip); 4080 if (ret != ARCHIVE_OK) 4081 goto exit_mac_metadata; 4082 zip->stream.next_in = 4083 (Bytef *)(uintptr_t)(const void *)p; 4084 zip->stream.avail_in = (uInt)bytes_avail; 4085 zip->stream.total_in = 0; 4086 zip->stream.next_out = mp; 4087 zip->stream.avail_out = (uInt)metadata_bytes; 4088 zip->stream.total_out = 0; 4089 4090 r = inflate(&zip->stream, 0); 4091 switch (r) { 4092 case Z_OK: 4093 break; 4094 case Z_STREAM_END: 4095 eof = 1; 4096 break; 4097 case Z_MEM_ERROR: 4098 archive_set_error(&a->archive, ENOMEM, 4099 "Out of memory for ZIP decompression"); 4100 ret = ARCHIVE_FATAL; 4101 goto exit_mac_metadata; 4102 default: 4103 archive_set_error(&a->archive, 4104 ARCHIVE_ERRNO_MISC, 4105 "ZIP decompression failed (%d)", r); 4106 ret = ARCHIVE_FATAL; 4107 goto exit_mac_metadata; 4108 } 4109 bytes_used = zip->stream.total_in; 4110 metadata_bytes -= zip->stream.total_out; 4111 mp += zip->stream.total_out; 4112 break; 4113 } 4114 #endif 4115 default: 4116 bytes_used = 0; 4117 break; 4118 } 4119 __archive_read_consume(a, bytes_used); 4120 remaining_bytes -= bytes_used; 4121 } 4122 archive_entry_copy_mac_metadata(entry, metadata, 4123 (size_t)rsrc->uncompressed_size - metadata_bytes); 4124 4125 exit_mac_metadata: 4126 __archive_read_seek(a, offset, SEEK_SET); 4127 zip->decompress_init = 0; 4128 free(metadata); 4129 return (ret); 4130 } 4131 4132 static int 4133 archive_read_format_zip_seekable_read_header(struct archive_read *a, 4134 struct archive_entry *entry) 4135 { 4136 struct zip *zip = (struct zip *)a->format->data; 4137 struct zip_entry *rsrc; 4138 int64_t offset; 4139 int r, ret = ARCHIVE_OK; 4140 4141 /* 4142 * It should be sufficient to call archive_read_next_header() for 4143 * a reader to determine if an entry is encrypted or not. If the 4144 * encryption of an entry is only detectable when calling 4145 * archive_read_data(), so be it. We'll do the same check there 4146 * as well. 4147 */ 4148 if (zip->has_encrypted_entries == 4149 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 4150 zip->has_encrypted_entries = 0; 4151 4152 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 4153 if (a->archive.archive_format_name == NULL) 4154 a->archive.archive_format_name = "ZIP"; 4155 4156 if (zip->zip_entries == NULL) { 4157 r = slurp_central_directory(a, entry, zip); 4158 if (r != ARCHIVE_OK) 4159 return r; 4160 /* Get first entry whose local header offset is lower than 4161 * other entries in the archive file. */ 4162 zip->entry = 4163 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 4164 } else if (zip->entry != NULL) { 4165 /* Get next entry in local header offset order. */ 4166 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 4167 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 4168 } 4169 4170 if (zip->entry == NULL) 4171 return ARCHIVE_EOF; 4172 4173 if (zip->entry->rsrcname.s) 4174 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 4175 &zip->tree_rsrc, zip->entry->rsrcname.s); 4176 else 4177 rsrc = NULL; 4178 4179 if (zip->cctx_valid) 4180 archive_decrypto_aes_ctr_release(&zip->cctx); 4181 if (zip->hctx_valid) 4182 archive_hmac_sha1_cleanup(&zip->hctx); 4183 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 4184 __archive_read_reset_passphrase(a); 4185 4186 /* File entries are sorted by the header offset, we should mostly 4187 * use __archive_read_consume to advance a read point to avoid 4188 * redundant data reading. */ 4189 offset = archive_filter_bytes(&a->archive, 0); 4190 if (offset < zip->entry->local_header_offset) 4191 __archive_read_consume(a, 4192 zip->entry->local_header_offset - offset); 4193 else if (offset != zip->entry->local_header_offset) { 4194 __archive_read_seek(a, zip->entry->local_header_offset, 4195 SEEK_SET); 4196 } 4197 zip->unconsumed = 0; 4198 r = zip_read_local_file_header(a, entry, zip); 4199 if (r != ARCHIVE_OK) 4200 return r; 4201 if (rsrc) { 4202 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 4203 if (ret2 < ret) 4204 ret = ret2; 4205 } 4206 return (ret); 4207 } 4208 4209 /* 4210 * We're going to seek for the next header anyway, so we don't 4211 * need to bother doing anything here. 4212 */ 4213 static int 4214 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 4215 { 4216 struct zip *zip; 4217 zip = (struct zip *)(a->format->data); 4218 4219 zip->unconsumed = 0; 4220 return (ARCHIVE_OK); 4221 } 4222 4223 int 4224 archive_read_support_format_zip_seekable(struct archive *_a) 4225 { 4226 struct archive_read *a = (struct archive_read *)_a; 4227 struct zip *zip; 4228 int r; 4229 4230 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 4231 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 4232 4233 zip = (struct zip *)calloc(1, sizeof(*zip)); 4234 if (zip == NULL) { 4235 archive_set_error(&a->archive, ENOMEM, 4236 "Can't allocate zip data"); 4237 return (ARCHIVE_FATAL); 4238 } 4239 4240 #ifdef HAVE_COPYFILE_H 4241 /* Set this by default on Mac OS. */ 4242 zip->process_mac_extensions = 1; 4243 #endif 4244 4245 /* 4246 * Until enough data has been read, we cannot tell about 4247 * any encrypted entries yet. 4248 */ 4249 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 4250 zip->crc32func = real_crc32; 4251 4252 r = __archive_read_register_format(a, 4253 zip, 4254 "zip", 4255 archive_read_format_zip_seekable_bid, 4256 archive_read_format_zip_options, 4257 archive_read_format_zip_seekable_read_header, 4258 archive_read_format_zip_read_data, 4259 archive_read_format_zip_read_data_skip_seekable, 4260 NULL, 4261 archive_read_format_zip_cleanup, 4262 archive_read_support_format_zip_capabilities_seekable, 4263 archive_read_format_zip_has_encrypted_entries); 4264 4265 if (r != ARCHIVE_OK) 4266 free(zip); 4267 return (ARCHIVE_OK); 4268 } 4269 4270 /*# vim:set noet:*/ 4271