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("$FreeBSD$"); 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 62 #include "archive.h" 63 #include "archive_digest_private.h" 64 #include "archive_cryptor_private.h" 65 #include "archive_endian.h" 66 #include "archive_entry.h" 67 #include "archive_entry_locale.h" 68 #include "archive_hmac_private.h" 69 #include "archive_private.h" 70 #include "archive_rb.h" 71 #include "archive_read_private.h" 72 #include "archive_ppmd8_private.h" 73 74 #ifndef HAVE_ZLIB_H 75 #include "archive_crc32.h" 76 #endif 77 78 struct zip_entry { 79 struct archive_rb_node node; 80 struct zip_entry *next; 81 int64_t local_header_offset; 82 int64_t compressed_size; 83 int64_t uncompressed_size; 84 int64_t gid; 85 int64_t uid; 86 struct archive_string rsrcname; 87 time_t mtime; 88 time_t atime; 89 time_t ctime; 90 uint32_t crc32; 91 uint16_t mode; 92 uint16_t zip_flags; /* From GP Flags Field */ 93 unsigned char compression; 94 unsigned char system; /* From "version written by" */ 95 unsigned char flags; /* Our extra markers. */ 96 unsigned char decdat;/* Used for Decryption check */ 97 98 /* WinZip AES encryption extra field should be available 99 * when compression is 99. */ 100 struct { 101 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 102 unsigned vendor; 103 #define AES_VENDOR_AE_1 0x0001 104 #define AES_VENDOR_AE_2 0x0002 105 /* AES encryption strength: 106 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 107 unsigned strength; 108 /* Actual compression method. */ 109 unsigned char compression; 110 } aes_extra; 111 }; 112 113 struct trad_enc_ctx { 114 uint32_t keys[3]; 115 }; 116 117 /* Bits used in zip_flags. */ 118 #define ZIP_ENCRYPTED (1 << 0) 119 #define ZIP_LENGTH_AT_END (1 << 3) 120 #define ZIP_STRONG_ENCRYPTED (1 << 6) 121 #define ZIP_UTF8_NAME (1 << 11) 122 /* See "7.2 Single Password Symmetric Encryption Method" 123 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 124 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 125 126 /* Bits used in flags. */ 127 #define LA_USED_ZIP64 (1 << 0) 128 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 129 130 /* 131 * See "WinZip - AES Encryption Information" 132 * http://www.winzip.com/aes_info.htm 133 */ 134 /* Value used in compression method. */ 135 #define WINZIP_AES_ENCRYPTION 99 136 /* Authentication code size. */ 137 #define AUTH_CODE_SIZE 10 138 /**/ 139 #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 140 141 struct zip { 142 /* Structural information about the archive. */ 143 struct archive_string format_name; 144 int64_t central_directory_offset; 145 size_t central_directory_entries_total; 146 size_t central_directory_entries_on_this_disk; 147 int has_encrypted_entries; 148 149 /* List of entries (seekable Zip only) */ 150 struct zip_entry *zip_entries; 151 struct archive_rb_tree tree; 152 struct archive_rb_tree tree_rsrc; 153 154 /* Bytes read but not yet consumed via __archive_read_consume() */ 155 size_t unconsumed; 156 157 /* Information about entry we're currently reading. */ 158 struct zip_entry *entry; 159 int64_t entry_bytes_remaining; 160 161 /* These count the number of bytes actually read for the entry. */ 162 int64_t entry_compressed_bytes_read; 163 int64_t entry_uncompressed_bytes_read; 164 165 /* Running CRC32 of the decompressed data */ 166 unsigned long entry_crc32; 167 unsigned long (*crc32func)(unsigned long, const void *, 168 size_t); 169 char ignore_crc32; 170 171 /* Flags to mark progress of decompression. */ 172 char decompress_init; 173 char end_of_entry; 174 175 unsigned char *uncompressed_buffer; 176 size_t uncompressed_buffer_size; 177 178 #ifdef HAVE_ZLIB_H 179 z_stream stream; 180 char stream_valid; 181 #endif 182 183 #if HAVE_LZMA_H && HAVE_LIBLZMA 184 lzma_stream zipx_lzma_stream; 185 char zipx_lzma_valid; 186 #endif 187 188 #ifdef HAVE_BZLIB_H 189 bz_stream bzstream; 190 char bzstream_valid; 191 #endif 192 193 IByteIn zipx_ppmd_stream; 194 ssize_t zipx_ppmd_read_compressed; 195 CPpmd8 ppmd8; 196 char ppmd8_valid; 197 198 struct archive_string_conv *sconv; 199 struct archive_string_conv *sconv_default; 200 struct archive_string_conv *sconv_utf8; 201 int init_default_conversion; 202 int process_mac_extensions; 203 204 char init_decryption; 205 206 /* Decryption buffer. */ 207 /* 208 * The decrypted data starts at decrypted_ptr and 209 * extends for decrypted_bytes_remaining. Decryption 210 * adds new data to the end of this block, data is returned 211 * to clients from the beginning. When the block hits the 212 * end of decrypted_buffer, it has to be shuffled back to 213 * the beginning of the buffer. 214 */ 215 unsigned char *decrypted_buffer; 216 unsigned char *decrypted_ptr; 217 size_t decrypted_buffer_size; 218 size_t decrypted_bytes_remaining; 219 size_t decrypted_unconsumed_bytes; 220 221 /* Traditional PKWARE decryption. */ 222 struct trad_enc_ctx tctx; 223 char tctx_valid; 224 225 /* WinZip AES decryption. */ 226 /* Contexts used for AES decryption. */ 227 archive_crypto_ctx cctx; 228 char cctx_valid; 229 archive_hmac_sha1_ctx hctx; 230 char hctx_valid; 231 232 /* Strong encryption's decryption header information. */ 233 unsigned iv_size; 234 unsigned alg_id; 235 unsigned bit_len; 236 unsigned flags; 237 unsigned erd_size; 238 unsigned v_size; 239 unsigned v_crc32; 240 uint8_t *iv; 241 uint8_t *erd; 242 uint8_t *v_data; 243 }; 244 245 /* Many systems define min or MIN, but not all. */ 246 #define zipmin(a,b) ((a) < (b) ? (a) : (b)) 247 248 /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8 249 * streams inside ZIP files. It has 2 purposes: one is to fetch the next 250 * compressed byte from the stream, second one is to increase the counter how 251 * many compressed bytes were read. */ 252 static Byte 253 ppmd_read(void* p) { 254 /* Get the handle to current decompression context. */ 255 struct archive_read *a = ((IByteIn*)p)->a; 256 struct zip *zip = (struct zip*) a->format->data; 257 258 /* Fetch next byte. */ 259 const uint8_t* data = __archive_read_ahead(a, 1, NULL); 260 __archive_read_consume(a, 1); 261 262 /* Increment the counter. */ 263 ++zip->zipx_ppmd_read_compressed; 264 265 /* Return the next compressed byte. */ 266 return data[0]; 267 } 268 269 /* ------------------------------------------------------------------------ */ 270 271 /* 272 Traditional PKWARE Decryption functions. 273 */ 274 275 static void 276 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 277 { 278 uint8_t t; 279 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 280 281 ctx->keys[0] = CRC32(ctx->keys[0], c); 282 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 283 t = (ctx->keys[1] >> 24) & 0xff; 284 ctx->keys[2] = CRC32(ctx->keys[2], t); 285 #undef CRC32 286 } 287 288 static uint8_t 289 trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) 290 { 291 unsigned temp = ctx->keys[2] | 2; 292 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 293 } 294 295 static void 296 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 297 size_t in_len, uint8_t *out, size_t out_len) 298 { 299 unsigned i, max; 300 301 max = (unsigned)((in_len < out_len)? in_len: out_len); 302 303 for (i = 0; i < max; i++) { 304 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); 305 out[i] = t; 306 trad_enc_update_keys(ctx, t); 307 } 308 } 309 310 static int 311 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 312 const uint8_t *key, size_t key_len, uint8_t *crcchk) 313 { 314 uint8_t header[12]; 315 316 if (key_len < 12) { 317 *crcchk = 0xff; 318 return -1; 319 } 320 321 ctx->keys[0] = 305419896L; 322 ctx->keys[1] = 591751049L; 323 ctx->keys[2] = 878082192L; 324 325 for (;pw_len; --pw_len) 326 trad_enc_update_keys(ctx, *pw++); 327 328 trad_enc_decrypt_update(ctx, key, 12, header, 12); 329 /* Return the last byte for CRC check. */ 330 *crcchk = header[11]; 331 return 0; 332 } 333 334 #if 0 335 static void 336 crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 337 int key_size) 338 { 339 #define MD_SIZE 20 340 archive_sha1_ctx ctx; 341 unsigned char md1[MD_SIZE]; 342 unsigned char md2[MD_SIZE * 2]; 343 unsigned char mkb[64]; 344 int i; 345 346 archive_sha1_init(&ctx); 347 archive_sha1_update(&ctx, p, size); 348 archive_sha1_final(&ctx, md1); 349 350 memset(mkb, 0x36, sizeof(mkb)); 351 for (i = 0; i < MD_SIZE; i++) 352 mkb[i] ^= md1[i]; 353 archive_sha1_init(&ctx); 354 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 355 archive_sha1_final(&ctx, md2); 356 357 memset(mkb, 0x5C, sizeof(mkb)); 358 for (i = 0; i < MD_SIZE; i++) 359 mkb[i] ^= md1[i]; 360 archive_sha1_init(&ctx); 361 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 362 archive_sha1_final(&ctx, md2 + MD_SIZE); 363 364 if (key_size > 32) 365 key_size = 32; 366 memcpy(key, md2, key_size); 367 #undef MD_SIZE 368 } 369 #endif 370 371 /* 372 * Common code for streaming or seeking modes. 373 * 374 * Includes code to read local file headers, decompress data 375 * from entry bodies, and common API. 376 */ 377 378 static unsigned long 379 real_crc32(unsigned long crc, const void *buff, size_t len) 380 { 381 return crc32(crc, buff, (unsigned int)len); 382 } 383 384 /* Used by "ignorecrc32" option to speed up tests. */ 385 static unsigned long 386 fake_crc32(unsigned long crc, const void *buff, size_t len) 387 { 388 (void)crc; /* UNUSED */ 389 (void)buff; /* UNUSED */ 390 (void)len; /* UNUSED */ 391 return 0; 392 } 393 394 static const struct { 395 int id; 396 const char * name; 397 } compression_methods[] = { 398 {0, "uncompressed"}, /* The file is stored (no compression) */ 399 {1, "shrinking"}, /* The file is Shrunk */ 400 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 401 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 402 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 403 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 404 {6, "imploded"}, /* The file is Imploded */ 405 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 406 {8, "deflation"}, /* The file is Deflated */ 407 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 408 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 409 * (old IBM TERSE) */ 410 {11, "reserved"}, /* Reserved by PKWARE */ 411 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 412 {13, "reserved"}, /* Reserved by PKWARE */ 413 {14, "lzma"}, /* LZMA (EFS) */ 414 {15, "reserved"}, /* Reserved by PKWARE */ 415 {16, "reserved"}, /* Reserved by PKWARE */ 416 {17, "reserved"}, /* Reserved by PKWARE */ 417 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 418 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 419 {95, "xz"}, /* XZ compressed data */ 420 {96, "jpeg"}, /* JPEG compressed data */ 421 {97, "wav-pack"}, /* WavPack compressed data */ 422 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 423 {99, "aes"} /* WinZip AES encryption */ 424 }; 425 426 static const char * 427 compression_name(const int compression) 428 { 429 static const int num_compression_methods = 430 sizeof(compression_methods)/sizeof(compression_methods[0]); 431 int i=0; 432 433 while(compression >= 0 && i < num_compression_methods) { 434 if (compression_methods[i].id == compression) 435 return compression_methods[i].name; 436 i++; 437 } 438 return "??"; 439 } 440 441 /* Convert an MSDOS-style date/time into Unix-style time. */ 442 static time_t 443 zip_time(const char *p) 444 { 445 int msTime, msDate; 446 struct tm ts; 447 448 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 449 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 450 451 memset(&ts, 0, sizeof(ts)); 452 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 453 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 454 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 455 ts.tm_hour = (msTime >> 11) & 0x1f; 456 ts.tm_min = (msTime >> 5) & 0x3f; 457 ts.tm_sec = (msTime << 1) & 0x3e; 458 ts.tm_isdst = -1; 459 return mktime(&ts); 460 } 461 462 /* 463 * The extra data is stored as a list of 464 * id1+size1+data1 + id2+size2+data2 ... 465 * triplets. id and size are 2 bytes each. 466 */ 467 static int 468 process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry) 469 { 470 unsigned offset = 0; 471 472 if (extra_length == 0) { 473 return ARCHIVE_OK; 474 } 475 476 if (extra_length < 4) { 477 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 478 "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length); 479 return ARCHIVE_FAILED; 480 } 481 while (offset <= extra_length - 4) { 482 unsigned short headerid = archive_le16dec(p + offset); 483 unsigned short datasize = archive_le16dec(p + offset + 2); 484 485 offset += 4; 486 if (offset + datasize > extra_length) { 487 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 488 "Extra data overflow: Need %d bytes but only found %d bytes", 489 (int)datasize, (int)(extra_length - offset)); 490 return ARCHIVE_FAILED; 491 } 492 #ifdef DEBUG 493 fprintf(stderr, "Header id 0x%04x, length %d\n", 494 headerid, datasize); 495 #endif 496 switch (headerid) { 497 case 0x0001: 498 /* Zip64 extended information extra field. */ 499 zip_entry->flags |= LA_USED_ZIP64; 500 if (zip_entry->uncompressed_size == 0xffffffff) { 501 uint64_t t = 0; 502 if (datasize < 8 503 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 504 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 505 "Malformed 64-bit uncompressed size"); 506 return ARCHIVE_FAILED; 507 } 508 zip_entry->uncompressed_size = t; 509 offset += 8; 510 datasize -= 8; 511 } 512 if (zip_entry->compressed_size == 0xffffffff) { 513 uint64_t t = 0; 514 if (datasize < 8 515 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 516 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 517 "Malformed 64-bit compressed size"); 518 return ARCHIVE_FAILED; 519 } 520 zip_entry->compressed_size = t; 521 offset += 8; 522 datasize -= 8; 523 } 524 if (zip_entry->local_header_offset == 0xffffffff) { 525 uint64_t t = 0; 526 if (datasize < 8 527 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 528 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 529 "Malformed 64-bit local header offset"); 530 return ARCHIVE_FAILED; 531 } 532 zip_entry->local_header_offset = t; 533 offset += 8; 534 datasize -= 8; 535 } 536 /* archive_le32dec(p + offset) gives disk 537 * on which file starts, but we don't handle 538 * multi-volume Zip files. */ 539 break; 540 #ifdef DEBUG 541 case 0x0017: 542 { 543 /* Strong encryption field. */ 544 if (archive_le16dec(p + offset) == 2) { 545 unsigned algId = 546 archive_le16dec(p + offset + 2); 547 unsigned bitLen = 548 archive_le16dec(p + offset + 4); 549 int flags = 550 archive_le16dec(p + offset + 6); 551 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 552 "flgas=%d\n", algId, bitLen,flags); 553 } 554 break; 555 } 556 #endif 557 case 0x5455: 558 { 559 /* Extended time field "UT". */ 560 int flags; 561 if (datasize == 0) { 562 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 563 "Incomplete extended time field"); 564 return ARCHIVE_FAILED; 565 } 566 flags = p[offset]; 567 offset++; 568 datasize--; 569 /* Flag bits indicate which dates are present. */ 570 if (flags & 0x01) 571 { 572 #ifdef DEBUG 573 fprintf(stderr, "mtime: %lld -> %d\n", 574 (long long)zip_entry->mtime, 575 archive_le32dec(p + offset)); 576 #endif 577 if (datasize < 4) 578 break; 579 zip_entry->mtime = archive_le32dec(p + offset); 580 offset += 4; 581 datasize -= 4; 582 } 583 if (flags & 0x02) 584 { 585 if (datasize < 4) 586 break; 587 zip_entry->atime = archive_le32dec(p + offset); 588 offset += 4; 589 datasize -= 4; 590 } 591 if (flags & 0x04) 592 { 593 if (datasize < 4) 594 break; 595 zip_entry->ctime = archive_le32dec(p + offset); 596 offset += 4; 597 datasize -= 4; 598 } 599 break; 600 } 601 case 0x5855: 602 { 603 /* Info-ZIP Unix Extra Field (old version) "UX". */ 604 if (datasize >= 8) { 605 zip_entry->atime = archive_le32dec(p + offset); 606 zip_entry->mtime = 607 archive_le32dec(p + offset + 4); 608 } 609 if (datasize >= 12) { 610 zip_entry->uid = 611 archive_le16dec(p + offset + 8); 612 zip_entry->gid = 613 archive_le16dec(p + offset + 10); 614 } 615 break; 616 } 617 case 0x6c78: 618 { 619 /* Experimental 'xl' field */ 620 /* 621 * Introduced Dec 2013 to provide a way to 622 * include external file attributes (and other 623 * fields that ordinarily appear only in 624 * central directory) in local file header. 625 * This provides file type and permission 626 * information necessary to support full 627 * streaming extraction. Currently being 628 * discussed with other Zip developers 629 * ... subject to change. 630 * 631 * Format: 632 * The field starts with a bitmap that specifies 633 * which additional fields are included. The 634 * bitmap is variable length and can be extended in 635 * the future. 636 * 637 * n bytes - feature bitmap: first byte has low-order 638 * 7 bits. If high-order bit is set, a subsequent 639 * byte holds the next 7 bits, etc. 640 * 641 * if bitmap & 1, 2 byte "version made by" 642 * if bitmap & 2, 2 byte "internal file attributes" 643 * if bitmap & 4, 4 byte "external file attributes" 644 * if bitmap & 8, 2 byte comment length + n byte comment 645 */ 646 int bitmap, bitmap_last; 647 648 if (datasize < 1) 649 break; 650 bitmap_last = bitmap = 0xff & p[offset]; 651 offset += 1; 652 datasize -= 1; 653 654 /* We only support first 7 bits of bitmap; skip rest. */ 655 while ((bitmap_last & 0x80) != 0 656 && datasize >= 1) { 657 bitmap_last = p[offset]; 658 offset += 1; 659 datasize -= 1; 660 } 661 662 if (bitmap & 1) { 663 /* 2 byte "version made by" */ 664 if (datasize < 2) 665 break; 666 zip_entry->system 667 = archive_le16dec(p + offset) >> 8; 668 offset += 2; 669 datasize -= 2; 670 } 671 if (bitmap & 2) { 672 /* 2 byte "internal file attributes" */ 673 uint32_t internal_attributes; 674 if (datasize < 2) 675 break; 676 internal_attributes 677 = archive_le16dec(p + offset); 678 /* Not used by libarchive at present. */ 679 (void)internal_attributes; /* UNUSED */ 680 offset += 2; 681 datasize -= 2; 682 } 683 if (bitmap & 4) { 684 /* 4 byte "external file attributes" */ 685 uint32_t external_attributes; 686 if (datasize < 4) 687 break; 688 external_attributes 689 = archive_le32dec(p + offset); 690 if (zip_entry->system == 3) { 691 zip_entry->mode 692 = external_attributes >> 16; 693 } else if (zip_entry->system == 0) { 694 // Interpret MSDOS directory bit 695 if (0x10 == (external_attributes & 0x10)) { 696 zip_entry->mode = AE_IFDIR | 0775; 697 } else { 698 zip_entry->mode = AE_IFREG | 0664; 699 } 700 if (0x01 == (external_attributes & 0x01)) { 701 // Read-only bit; strip write permissions 702 zip_entry->mode &= 0555; 703 } 704 } else { 705 zip_entry->mode = 0; 706 } 707 offset += 4; 708 datasize -= 4; 709 } 710 if (bitmap & 8) { 711 /* 2 byte comment length + comment */ 712 uint32_t comment_length; 713 if (datasize < 2) 714 break; 715 comment_length 716 = archive_le16dec(p + offset); 717 offset += 2; 718 datasize -= 2; 719 720 if (datasize < comment_length) 721 break; 722 /* Comment is not supported by libarchive */ 723 offset += comment_length; 724 datasize -= comment_length; 725 } 726 break; 727 } 728 case 0x7855: 729 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 730 #ifdef DEBUG 731 fprintf(stderr, "uid %d gid %d\n", 732 archive_le16dec(p + offset), 733 archive_le16dec(p + offset + 2)); 734 #endif 735 if (datasize >= 2) 736 zip_entry->uid = archive_le16dec(p + offset); 737 if (datasize >= 4) 738 zip_entry->gid = 739 archive_le16dec(p + offset + 2); 740 break; 741 case 0x7875: 742 { 743 /* Info-Zip Unix Extra Field (type 3) "ux". */ 744 int uidsize = 0, gidsize = 0; 745 746 /* TODO: support arbitrary uidsize/gidsize. */ 747 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 748 if (datasize >= 4) { 749 /* get a uid size. */ 750 uidsize = 0xff & (int)p[offset+1]; 751 if (uidsize == 2) 752 zip_entry->uid = 753 archive_le16dec( 754 p + offset + 2); 755 else if (uidsize == 4 && datasize >= 6) 756 zip_entry->uid = 757 archive_le32dec( 758 p + offset + 2); 759 } 760 if (datasize >= (2 + uidsize + 3)) { 761 /* get a gid size. */ 762 gidsize = 0xff & (int)p[offset+2+uidsize]; 763 if (gidsize == 2) 764 zip_entry->gid = 765 archive_le16dec( 766 p+offset+2+uidsize+1); 767 else if (gidsize == 4 && 768 datasize >= (2 + uidsize + 5)) 769 zip_entry->gid = 770 archive_le32dec( 771 p+offset+2+uidsize+1); 772 } 773 } 774 break; 775 } 776 case 0x9901: 777 /* WinZip AES extra data field. */ 778 if (datasize < 6) { 779 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 780 "Incomplete AES field"); 781 return ARCHIVE_FAILED; 782 } 783 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 784 /* Vendor version. */ 785 zip_entry->aes_extra.vendor = 786 archive_le16dec(p + offset); 787 /* AES encryption strength. */ 788 zip_entry->aes_extra.strength = p[offset + 4]; 789 /* Actual compression method. */ 790 zip_entry->aes_extra.compression = 791 p[offset + 5]; 792 } 793 break; 794 default: 795 break; 796 } 797 offset += datasize; 798 } 799 if (offset != extra_length) { 800 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 801 "Malformed extra data: Consumed %d bytes of %d bytes", 802 (int)offset, (int)extra_length); 803 return ARCHIVE_FAILED; 804 } 805 return ARCHIVE_OK; 806 } 807 808 /* 809 * Assumes file pointer is at beginning of local file header. 810 */ 811 static int 812 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 813 struct zip *zip) 814 { 815 const char *p; 816 const void *h; 817 const wchar_t *wp; 818 const char *cp; 819 size_t len, filename_length, extra_length; 820 struct archive_string_conv *sconv; 821 struct zip_entry *zip_entry = zip->entry; 822 struct zip_entry zip_entry_central_dir; 823 int ret = ARCHIVE_OK; 824 char version; 825 826 /* Save a copy of the original for consistency checks. */ 827 zip_entry_central_dir = *zip_entry; 828 829 zip->decompress_init = 0; 830 zip->end_of_entry = 0; 831 zip->entry_uncompressed_bytes_read = 0; 832 zip->entry_compressed_bytes_read = 0; 833 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 834 835 /* Setup default conversion. */ 836 if (zip->sconv == NULL && !zip->init_default_conversion) { 837 zip->sconv_default = 838 archive_string_default_conversion_for_read(&(a->archive)); 839 zip->init_default_conversion = 1; 840 } 841 842 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 843 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 844 "Truncated ZIP file header"); 845 return (ARCHIVE_FATAL); 846 } 847 848 if (memcmp(p, "PK\003\004", 4) != 0) { 849 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 850 return ARCHIVE_FATAL; 851 } 852 version = p[4]; 853 zip_entry->system = p[5]; 854 zip_entry->zip_flags = archive_le16dec(p + 6); 855 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 856 zip->has_encrypted_entries = 1; 857 archive_entry_set_is_data_encrypted(entry, 1); 858 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 859 zip_entry->zip_flags & ZIP_ENCRYPTED && 860 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 861 archive_entry_set_is_metadata_encrypted(entry, 1); 862 return ARCHIVE_FATAL; 863 } 864 } 865 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 866 zip_entry->compression = (char)archive_le16dec(p + 8); 867 zip_entry->mtime = zip_time(p + 10); 868 zip_entry->crc32 = archive_le32dec(p + 14); 869 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 870 zip_entry->decdat = p[11]; 871 else 872 zip_entry->decdat = p[17]; 873 zip_entry->compressed_size = archive_le32dec(p + 18); 874 zip_entry->uncompressed_size = archive_le32dec(p + 22); 875 filename_length = archive_le16dec(p + 26); 876 extra_length = archive_le16dec(p + 28); 877 878 __archive_read_consume(a, 30); 879 880 /* Read the filename. */ 881 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 882 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 883 "Truncated ZIP file header"); 884 return (ARCHIVE_FATAL); 885 } 886 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 887 /* The filename is stored to be UTF-8. */ 888 if (zip->sconv_utf8 == NULL) { 889 zip->sconv_utf8 = 890 archive_string_conversion_from_charset( 891 &a->archive, "UTF-8", 1); 892 if (zip->sconv_utf8 == NULL) 893 return (ARCHIVE_FATAL); 894 } 895 sconv = zip->sconv_utf8; 896 } else if (zip->sconv != NULL) 897 sconv = zip->sconv; 898 else 899 sconv = zip->sconv_default; 900 901 if (archive_entry_copy_pathname_l(entry, 902 h, filename_length, sconv) != 0) { 903 if (errno == ENOMEM) { 904 archive_set_error(&a->archive, ENOMEM, 905 "Can't allocate memory for Pathname"); 906 return (ARCHIVE_FATAL); 907 } 908 archive_set_error(&a->archive, 909 ARCHIVE_ERRNO_FILE_FORMAT, 910 "Pathname cannot be converted " 911 "from %s to current locale.", 912 archive_string_conversion_charset_name(sconv)); 913 ret = ARCHIVE_WARN; 914 } 915 __archive_read_consume(a, filename_length); 916 917 /* Read the extra data. */ 918 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 919 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 920 "Truncated ZIP file header"); 921 return (ARCHIVE_FATAL); 922 } 923 924 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 925 return ARCHIVE_FATAL; 926 } 927 __archive_read_consume(a, extra_length); 928 929 /* Work around a bug in Info-Zip: When reading from a pipe, it 930 * stats the pipe instead of synthesizing a file entry. */ 931 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 932 zip_entry->mode &= ~ AE_IFMT; 933 zip_entry->mode |= AE_IFREG; 934 } 935 936 /* If the mode is totally empty, set some sane default. */ 937 if (zip_entry->mode == 0) { 938 zip_entry->mode |= 0664; 939 } 940 941 /* Windows archivers sometimes use backslash as the directory separator. 942 Normalize to slash. */ 943 if (zip_entry->system == 0 && 944 (wp = archive_entry_pathname_w(entry)) != NULL) { 945 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { 946 size_t i; 947 struct archive_wstring s; 948 archive_string_init(&s); 949 archive_wstrcpy(&s, wp); 950 for (i = 0; i < archive_strlen(&s); i++) { 951 if (s.s[i] == '\\') 952 s.s[i] = '/'; 953 } 954 archive_entry_copy_pathname_w(entry, s.s); 955 archive_wstring_free(&s); 956 } 957 } 958 959 /* Make sure that entries with a trailing '/' are marked as directories 960 * even if the External File Attributes contains bogus values. If this 961 * is not a directory and there is no type, assume regularfile. */ 962 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 963 int has_slash; 964 965 wp = archive_entry_pathname_w(entry); 966 if (wp != NULL) { 967 len = wcslen(wp); 968 has_slash = len > 0 && wp[len - 1] == L'/'; 969 } else { 970 cp = archive_entry_pathname(entry); 971 len = (cp != NULL)?strlen(cp):0; 972 has_slash = len > 0 && cp[len - 1] == '/'; 973 } 974 /* Correct file type as needed. */ 975 if (has_slash) { 976 zip_entry->mode &= ~AE_IFMT; 977 zip_entry->mode |= AE_IFDIR; 978 zip_entry->mode |= 0111; 979 } else if ((zip_entry->mode & AE_IFMT) == 0) { 980 zip_entry->mode |= AE_IFREG; 981 } 982 } 983 984 /* Make sure directories end in '/' */ 985 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 986 wp = archive_entry_pathname_w(entry); 987 if (wp != NULL) { 988 len = wcslen(wp); 989 if (len > 0 && wp[len - 1] != L'/') { 990 struct archive_wstring s; 991 archive_string_init(&s); 992 archive_wstrcat(&s, wp); 993 archive_wstrappend_wchar(&s, L'/'); 994 archive_entry_copy_pathname_w(entry, s.s); 995 archive_wstring_free(&s); 996 } 997 } else { 998 cp = archive_entry_pathname(entry); 999 len = (cp != NULL)?strlen(cp):0; 1000 if (len > 0 && cp[len - 1] != '/') { 1001 struct archive_string s; 1002 archive_string_init(&s); 1003 archive_strcat(&s, cp); 1004 archive_strappend_char(&s, '/'); 1005 archive_entry_set_pathname(entry, s.s); 1006 archive_string_free(&s); 1007 } 1008 } 1009 } 1010 1011 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 1012 /* If this came from the central dir, it's size info 1013 * is definitive, so ignore the length-at-end flag. */ 1014 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 1015 /* If local header is missing a value, use the one from 1016 the central directory. If both have it, warn about 1017 mismatches. */ 1018 if (zip_entry->crc32 == 0) { 1019 zip_entry->crc32 = zip_entry_central_dir.crc32; 1020 } else if (!zip->ignore_crc32 1021 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 1022 archive_set_error(&a->archive, 1023 ARCHIVE_ERRNO_FILE_FORMAT, 1024 "Inconsistent CRC32 values"); 1025 ret = ARCHIVE_WARN; 1026 } 1027 if (zip_entry->compressed_size == 0) { 1028 zip_entry->compressed_size 1029 = zip_entry_central_dir.compressed_size; 1030 } else if (zip_entry->compressed_size 1031 != zip_entry_central_dir.compressed_size) { 1032 archive_set_error(&a->archive, 1033 ARCHIVE_ERRNO_FILE_FORMAT, 1034 "Inconsistent compressed size: " 1035 "%jd in central directory, %jd in local header", 1036 (intmax_t)zip_entry_central_dir.compressed_size, 1037 (intmax_t)zip_entry->compressed_size); 1038 ret = ARCHIVE_WARN; 1039 } 1040 if (zip_entry->uncompressed_size == 0) { 1041 zip_entry->uncompressed_size 1042 = zip_entry_central_dir.uncompressed_size; 1043 } else if (zip_entry->uncompressed_size 1044 != zip_entry_central_dir.uncompressed_size) { 1045 archive_set_error(&a->archive, 1046 ARCHIVE_ERRNO_FILE_FORMAT, 1047 "Inconsistent uncompressed size: " 1048 "%jd in central directory, %jd in local header", 1049 (intmax_t)zip_entry_central_dir.uncompressed_size, 1050 (intmax_t)zip_entry->uncompressed_size); 1051 ret = ARCHIVE_WARN; 1052 } 1053 } 1054 1055 /* Populate some additional entry fields: */ 1056 archive_entry_set_mode(entry, zip_entry->mode); 1057 archive_entry_set_uid(entry, zip_entry->uid); 1058 archive_entry_set_gid(entry, zip_entry->gid); 1059 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 1060 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 1061 archive_entry_set_atime(entry, zip_entry->atime, 0); 1062 1063 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 1064 size_t linkname_length; 1065 1066 if (zip_entry->compressed_size > 64 * 1024) { 1067 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1068 "Zip file with oversized link entry"); 1069 return ARCHIVE_FATAL; 1070 } 1071 1072 linkname_length = (size_t)zip_entry->compressed_size; 1073 1074 archive_entry_set_size(entry, 0); 1075 p = __archive_read_ahead(a, linkname_length, NULL); 1076 if (p == NULL) { 1077 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1078 "Truncated Zip file"); 1079 return ARCHIVE_FATAL; 1080 } 1081 1082 sconv = zip->sconv; 1083 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1084 sconv = zip->sconv_utf8; 1085 if (sconv == NULL) 1086 sconv = zip->sconv_default; 1087 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 1088 sconv) != 0) { 1089 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1090 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1091 archive_entry_copy_symlink_l(entry, p, 1092 linkname_length, NULL); 1093 if (errno == ENOMEM) { 1094 archive_set_error(&a->archive, ENOMEM, 1095 "Can't allocate memory for Symlink"); 1096 return (ARCHIVE_FATAL); 1097 } 1098 /* 1099 * Since there is no character-set regulation for 1100 * symlink name, do not report the conversion error 1101 * in an automatic conversion. 1102 */ 1103 if (sconv != zip->sconv_utf8 || 1104 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1105 archive_set_error(&a->archive, 1106 ARCHIVE_ERRNO_FILE_FORMAT, 1107 "Symlink cannot be converted " 1108 "from %s to current locale.", 1109 archive_string_conversion_charset_name( 1110 sconv)); 1111 ret = ARCHIVE_WARN; 1112 } 1113 } 1114 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1115 1116 if (__archive_read_consume(a, linkname_length) < 0) { 1117 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1118 "Read error skipping symlink target name"); 1119 return ARCHIVE_FATAL; 1120 } 1121 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1122 || zip_entry->uncompressed_size > 0) { 1123 /* Set the size only if it's meaningful. */ 1124 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1125 } 1126 zip->entry_bytes_remaining = zip_entry->compressed_size; 1127 1128 /* If there's no body, force read_data() to return EOF immediately. */ 1129 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1130 && zip->entry_bytes_remaining < 1) 1131 zip->end_of_entry = 1; 1132 1133 /* Set up a more descriptive format name. */ 1134 archive_string_empty(&zip->format_name); 1135 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1136 version / 10, version % 10, 1137 compression_name(zip->entry->compression)); 1138 a->archive.archive_format_name = zip->format_name.s; 1139 1140 return (ret); 1141 } 1142 1143 static int 1144 check_authentication_code(struct archive_read *a, const void *_p) 1145 { 1146 struct zip *zip = (struct zip *)(a->format->data); 1147 1148 /* Check authentication code. */ 1149 if (zip->hctx_valid) { 1150 const void *p; 1151 uint8_t hmac[20]; 1152 size_t hmac_len = 20; 1153 int cmp; 1154 1155 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1156 if (_p == NULL) { 1157 /* Read authentication code. */ 1158 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1159 if (p == NULL) { 1160 archive_set_error(&a->archive, 1161 ARCHIVE_ERRNO_FILE_FORMAT, 1162 "Truncated ZIP file data"); 1163 return (ARCHIVE_FATAL); 1164 } 1165 } else { 1166 p = _p; 1167 } 1168 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1169 __archive_read_consume(a, AUTH_CODE_SIZE); 1170 if (cmp != 0) { 1171 archive_set_error(&a->archive, 1172 ARCHIVE_ERRNO_MISC, 1173 "ZIP bad Authentication code"); 1174 return (ARCHIVE_WARN); 1175 } 1176 } 1177 return (ARCHIVE_OK); 1178 } 1179 1180 /* 1181 * Read "uncompressed" data. There are three cases: 1182 * 1) We know the size of the data. This is always true for the 1183 * seeking reader (we've examined the Central Directory already). 1184 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1185 * Info-ZIP seems to do this; we know the size but have to grab 1186 * the CRC from the data descriptor afterwards. 1187 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1188 * we have no size information. In this case, we can do pretty 1189 * well by watching for the data descriptor record. The data 1190 * descriptor is 16 bytes and includes a computed CRC that should 1191 * provide a strong check. 1192 * 1193 * TODO: Technically, the PK\007\010 signature is optional. 1194 * In the original spec, the data descriptor contained CRC 1195 * and size fields but had no leading signature. In practice, 1196 * newer writers seem to provide the signature pretty consistently. 1197 * 1198 * For uncompressed data, the PK\007\010 marker seems essential 1199 * to be sure we've actually seen the end of the entry. 1200 * 1201 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1202 * zip->end_of_entry if it consumes all of the data. 1203 */ 1204 static int 1205 zip_read_data_none(struct archive_read *a, const void **_buff, 1206 size_t *size, int64_t *offset) 1207 { 1208 struct zip *zip; 1209 const char *buff; 1210 ssize_t bytes_avail; 1211 int r; 1212 1213 (void)offset; /* UNUSED */ 1214 1215 zip = (struct zip *)(a->format->data); 1216 1217 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1218 const char *p; 1219 ssize_t grabbing_bytes = 24; 1220 1221 if (zip->hctx_valid) 1222 grabbing_bytes += AUTH_CODE_SIZE; 1223 /* Grab at least 24 bytes. */ 1224 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1225 if (bytes_avail < grabbing_bytes) { 1226 /* Zip archives have end-of-archive markers 1227 that are longer than this, so a failure to get at 1228 least 24 bytes really does indicate a truncated 1229 file. */ 1230 archive_set_error(&a->archive, 1231 ARCHIVE_ERRNO_FILE_FORMAT, 1232 "Truncated ZIP file data"); 1233 return (ARCHIVE_FATAL); 1234 } 1235 /* Check for a complete PK\007\010 signature, followed 1236 * by the correct 4-byte CRC. */ 1237 p = buff; 1238 if (zip->hctx_valid) 1239 p += AUTH_CODE_SIZE; 1240 if (p[0] == 'P' && p[1] == 'K' 1241 && p[2] == '\007' && p[3] == '\010' 1242 && (archive_le32dec(p + 4) == zip->entry_crc32 1243 || zip->ignore_crc32 1244 || (zip->hctx_valid 1245 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1246 if (zip->entry->flags & LA_USED_ZIP64) { 1247 uint64_t compressed, uncompressed; 1248 zip->entry->crc32 = archive_le32dec(p + 4); 1249 compressed = archive_le64dec(p + 8); 1250 uncompressed = archive_le64dec(p + 16); 1251 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1252 archive_set_error(&a->archive, 1253 ARCHIVE_ERRNO_FILE_FORMAT, 1254 "Overflow of 64-bit file sizes"); 1255 return ARCHIVE_FAILED; 1256 } 1257 zip->entry->compressed_size = compressed; 1258 zip->entry->uncompressed_size = uncompressed; 1259 zip->unconsumed = 24; 1260 } else { 1261 zip->entry->crc32 = archive_le32dec(p + 4); 1262 zip->entry->compressed_size = 1263 archive_le32dec(p + 8); 1264 zip->entry->uncompressed_size = 1265 archive_le32dec(p + 12); 1266 zip->unconsumed = 16; 1267 } 1268 if (zip->hctx_valid) { 1269 r = check_authentication_code(a, buff); 1270 if (r != ARCHIVE_OK) 1271 return (r); 1272 } 1273 zip->end_of_entry = 1; 1274 return (ARCHIVE_OK); 1275 } 1276 /* If not at EOF, ensure we consume at least one byte. */ 1277 ++p; 1278 1279 /* Scan forward until we see where a PK\007\010 signature 1280 * might be. */ 1281 /* Return bytes up until that point. On the next call, 1282 * the code above will verify the data descriptor. */ 1283 while (p < buff + bytes_avail - 4) { 1284 if (p[3] == 'P') { p += 3; } 1285 else if (p[3] == 'K') { p += 2; } 1286 else if (p[3] == '\007') { p += 1; } 1287 else if (p[3] == '\010' && p[2] == '\007' 1288 && p[1] == 'K' && p[0] == 'P') { 1289 if (zip->hctx_valid) 1290 p -= AUTH_CODE_SIZE; 1291 break; 1292 } else { p += 4; } 1293 } 1294 bytes_avail = p - buff; 1295 } else { 1296 if (zip->entry_bytes_remaining == 0) { 1297 zip->end_of_entry = 1; 1298 if (zip->hctx_valid) { 1299 r = check_authentication_code(a, NULL); 1300 if (r != ARCHIVE_OK) 1301 return (r); 1302 } 1303 return (ARCHIVE_OK); 1304 } 1305 /* Grab a bunch of bytes. */ 1306 buff = __archive_read_ahead(a, 1, &bytes_avail); 1307 if (bytes_avail <= 0) { 1308 archive_set_error(&a->archive, 1309 ARCHIVE_ERRNO_FILE_FORMAT, 1310 "Truncated ZIP file data"); 1311 return (ARCHIVE_FATAL); 1312 } 1313 if (bytes_avail > zip->entry_bytes_remaining) 1314 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1315 } 1316 if (zip->tctx_valid || zip->cctx_valid) { 1317 size_t dec_size = bytes_avail; 1318 1319 if (dec_size > zip->decrypted_buffer_size) 1320 dec_size = zip->decrypted_buffer_size; 1321 if (zip->tctx_valid) { 1322 trad_enc_decrypt_update(&zip->tctx, 1323 (const uint8_t *)buff, dec_size, 1324 zip->decrypted_buffer, dec_size); 1325 } else { 1326 size_t dsize = dec_size; 1327 archive_hmac_sha1_update(&zip->hctx, 1328 (const uint8_t *)buff, dec_size); 1329 archive_decrypto_aes_ctr_update(&zip->cctx, 1330 (const uint8_t *)buff, dec_size, 1331 zip->decrypted_buffer, &dsize); 1332 } 1333 bytes_avail = dec_size; 1334 buff = (const char *)zip->decrypted_buffer; 1335 } 1336 *size = bytes_avail; 1337 zip->entry_bytes_remaining -= bytes_avail; 1338 zip->entry_uncompressed_bytes_read += bytes_avail; 1339 zip->entry_compressed_bytes_read += bytes_avail; 1340 zip->unconsumed += bytes_avail; 1341 *_buff = buff; 1342 return (ARCHIVE_OK); 1343 } 1344 1345 static int 1346 consume_optional_marker(struct archive_read *a, struct zip *zip) 1347 { 1348 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1349 const char *p; 1350 1351 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1352 archive_set_error(&a->archive, 1353 ARCHIVE_ERRNO_FILE_FORMAT, 1354 "Truncated ZIP end-of-file record"); 1355 return (ARCHIVE_FATAL); 1356 } 1357 /* Consume the optional PK\007\010 marker. */ 1358 if (p[0] == 'P' && p[1] == 'K' && 1359 p[2] == '\007' && p[3] == '\010') { 1360 p += 4; 1361 zip->unconsumed = 4; 1362 } 1363 if (zip->entry->flags & LA_USED_ZIP64) { 1364 uint64_t compressed, uncompressed; 1365 zip->entry->crc32 = archive_le32dec(p); 1366 compressed = archive_le64dec(p + 4); 1367 uncompressed = archive_le64dec(p + 12); 1368 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1369 archive_set_error(&a->archive, 1370 ARCHIVE_ERRNO_FILE_FORMAT, 1371 "Overflow of 64-bit file sizes"); 1372 return ARCHIVE_FAILED; 1373 } 1374 zip->entry->compressed_size = compressed; 1375 zip->entry->uncompressed_size = uncompressed; 1376 zip->unconsumed += 20; 1377 } else { 1378 zip->entry->crc32 = archive_le32dec(p); 1379 zip->entry->compressed_size = archive_le32dec(p + 4); 1380 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1381 zip->unconsumed += 12; 1382 } 1383 } 1384 1385 return (ARCHIVE_OK); 1386 } 1387 1388 #if HAVE_LZMA_H && HAVE_LIBLZMA 1389 static int 1390 zipx_xz_init(struct archive_read *a, struct zip *zip) 1391 { 1392 lzma_ret r; 1393 1394 if(zip->zipx_lzma_valid) { 1395 lzma_end(&zip->zipx_lzma_stream); 1396 zip->zipx_lzma_valid = 0; 1397 } 1398 1399 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); 1400 r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0); 1401 if (r != LZMA_OK) { 1402 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1403 "xz initialization failed(%d)", 1404 r); 1405 1406 return (ARCHIVE_FAILED); 1407 } 1408 1409 zip->zipx_lzma_valid = 1; 1410 1411 free(zip->uncompressed_buffer); 1412 1413 zip->uncompressed_buffer_size = 256 * 1024; 1414 zip->uncompressed_buffer = 1415 (uint8_t*) malloc(zip->uncompressed_buffer_size); 1416 if (zip->uncompressed_buffer == NULL) { 1417 archive_set_error(&a->archive, ENOMEM, 1418 "No memory for xz decompression"); 1419 return (ARCHIVE_FATAL); 1420 } 1421 1422 zip->decompress_init = 1; 1423 return (ARCHIVE_OK); 1424 } 1425 1426 static int 1427 zipx_lzma_alone_init(struct archive_read *a, struct zip *zip) 1428 { 1429 lzma_ret r; 1430 const uint8_t* p; 1431 1432 #pragma pack(push) 1433 #pragma pack(1) 1434 struct _alone_header { 1435 uint8_t bytes[5]; 1436 uint64_t uncompressed_size; 1437 } alone_header; 1438 #pragma pack(pop) 1439 1440 /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma that 1441 * is a part of XZ Utils. The stream format stored inside ZIPX file is a 1442 * modified "lzma alone" file format, that was used by the `lzma` utility 1443 * which was later deprecated in favour of `xz` utility. Since those 1444 * formats are nearly the same, we can use a standard "lzma alone" decoder 1445 * from XZ Utils. */ 1446 1447 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); 1448 r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX); 1449 if (r != LZMA_OK) { 1450 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1451 "lzma initialization failed(%d)", r); 1452 1453 return (ARCHIVE_FAILED); 1454 } 1455 1456 /* Flag the cleanup function that we want our lzma-related structures 1457 * to be freed later. */ 1458 zip->zipx_lzma_valid = 1; 1459 1460 /* The "lzma alone" file format and the stream format inside ZIPx are 1461 * almost the same. Here's an example of a structure of "lzma alone" 1462 * format: 1463 * 1464 * $ cat /bin/ls | lzma | xxd | head -n 1 1465 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814 1466 * 1467 * 5 bytes 8 bytes n bytes 1468 * <lzma_params><uncompressed_size><data...> 1469 * 1470 * lzma_params is a 5-byte blob that has to be decoded to extract 1471 * parameters of this LZMA stream. The uncompressed_size field is an 1472 * uint64_t value that contains information about the size of the 1473 * uncompressed file, or UINT64_MAX if this value is unknown. The <data...> 1474 * part is the actual lzma-compressed data stream. 1475 * 1476 * Now here's the structure of the stream inside the ZIPX file: 1477 * 1478 * $ cat stream_inside_zipx | xxd | head -n 1 1479 * 00000000: 0914 0500 5d00 8000 0000 2814 .... .... 1480 * 1481 * 2byte 2byte 5 bytes n bytes 1482 * <magic1><magic2><lzma_params><data...> 1483 * 1484 * This means that the ZIPX file contains an additional magic1 and magic2 1485 * headers, the lzma_params field contains the same parameter set as in the 1486 * "lzma alone" format, and the <data...> field is the same as in the "lzma 1487 * alone" format as well. Note that also the zipx format is missing the 1488 * uncompressed_size field. 1489 * 1490 * So, in order to use the "lzma alone" decoder for the zipx lzma stream, 1491 * we simply need to shuffle around some fields, prepare a new lzma alone 1492 * header, feed it into lzma alone decoder so it will initialize itself 1493 * properly, and then we can start feeding normal zipx lzma stream into the 1494 * decoder. 1495 */ 1496 1497 /* Read magic1,magic2,lzma_params from the ZIPX stream. */ 1498 if((p = __archive_read_ahead(a, 9, NULL)) == NULL) { 1499 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1500 "Truncated lzma data"); 1501 return (ARCHIVE_FATAL); 1502 } 1503 1504 if(p[2] != 0x05 || p[3] != 0x00) { 1505 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1506 "Invalid lzma data"); 1507 return (ARCHIVE_FATAL); 1508 } 1509 1510 /* Prepare an lzma alone header: copy the lzma_params blob into a proper 1511 * place into the lzma alone header. */ 1512 memcpy(&alone_header.bytes[0], p + 4, 5); 1513 1514 /* Initialize the 'uncompressed size' field to unknown; we'll manually 1515 * monitor how many bytes there are still to be uncompressed. */ 1516 alone_header.uncompressed_size = UINT64_MAX; 1517 1518 if(!zip->uncompressed_buffer) { 1519 zip->uncompressed_buffer_size = 256 * 1024; 1520 zip->uncompressed_buffer = 1521 (uint8_t*) malloc(zip->uncompressed_buffer_size); 1522 1523 if (zip->uncompressed_buffer == NULL) { 1524 archive_set_error(&a->archive, ENOMEM, 1525 "No memory for lzma decompression"); 1526 return (ARCHIVE_FATAL); 1527 } 1528 } 1529 1530 zip->zipx_lzma_stream.next_in = (void*) &alone_header; 1531 zip->zipx_lzma_stream.avail_in = sizeof(alone_header); 1532 zip->zipx_lzma_stream.total_in = 0; 1533 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1534 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; 1535 zip->zipx_lzma_stream.total_out = 0; 1536 1537 /* Feed only the header into the lzma alone decoder. This will effectively 1538 * initialize the decoder, and will not produce any output bytes yet. */ 1539 r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1540 if (r != LZMA_OK) { 1541 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1542 "lzma stream initialization error"); 1543 return ARCHIVE_FATAL; 1544 } 1545 1546 /* We've already consumed some bytes, so take this into account. */ 1547 __archive_read_consume(a, 9); 1548 zip->entry_bytes_remaining -= 9; 1549 zip->entry_compressed_bytes_read += 9; 1550 1551 zip->decompress_init = 1; 1552 return (ARCHIVE_OK); 1553 } 1554 1555 static int 1556 zip_read_data_zipx_xz(struct archive_read *a, const void **buff, 1557 size_t *size, int64_t *offset) 1558 { 1559 struct zip* zip = (struct zip *)(a->format->data); 1560 int ret; 1561 lzma_ret lz_ret; 1562 const void* compressed_buf; 1563 ssize_t bytes_avail, in_bytes, to_consume = 0; 1564 1565 (void) offset; /* UNUSED */ 1566 1567 /* Initialize decompressor if not yet initialized. */ 1568 if (!zip->decompress_init) { 1569 ret = zipx_xz_init(a, zip); 1570 if (ret != ARCHIVE_OK) 1571 return (ret); 1572 } 1573 1574 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); 1575 if (bytes_avail < 0) { 1576 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1577 "Truncated xz file body"); 1578 return (ARCHIVE_FATAL); 1579 } 1580 1581 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 1582 zip->zipx_lzma_stream.next_in = compressed_buf; 1583 zip->zipx_lzma_stream.avail_in = in_bytes; 1584 zip->zipx_lzma_stream.total_in = 0; 1585 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1586 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; 1587 zip->zipx_lzma_stream.total_out = 0; 1588 1589 /* Perform the decompression. */ 1590 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1591 switch(lz_ret) { 1592 case LZMA_DATA_ERROR: 1593 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1594 "xz data error (error %d)", (int) lz_ret); 1595 return (ARCHIVE_FATAL); 1596 1597 case LZMA_NO_CHECK: 1598 case LZMA_OK: 1599 break; 1600 1601 default: 1602 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1603 "xz unknown error %d", (int) lz_ret); 1604 return (ARCHIVE_FATAL); 1605 1606 case LZMA_STREAM_END: 1607 lzma_end(&zip->zipx_lzma_stream); 1608 zip->zipx_lzma_valid = 0; 1609 1610 if((int64_t) zip->zipx_lzma_stream.total_in != 1611 zip->entry_bytes_remaining) 1612 { 1613 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1614 "xz premature end of stream"); 1615 return (ARCHIVE_FATAL); 1616 } 1617 1618 zip->end_of_entry = 1; 1619 break; 1620 } 1621 1622 to_consume = zip->zipx_lzma_stream.total_in; 1623 1624 __archive_read_consume(a, to_consume); 1625 zip->entry_bytes_remaining -= to_consume; 1626 zip->entry_compressed_bytes_read += to_consume; 1627 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; 1628 1629 *size = zip->zipx_lzma_stream.total_out; 1630 *buff = zip->uncompressed_buffer; 1631 1632 ret = consume_optional_marker(a, zip); 1633 if (ret != ARCHIVE_OK) 1634 return (ret); 1635 1636 return (ARCHIVE_OK); 1637 } 1638 1639 static int 1640 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, 1641 size_t *size, int64_t *offset) 1642 { 1643 struct zip* zip = (struct zip *)(a->format->data); 1644 int ret; 1645 lzma_ret lz_ret; 1646 const void* compressed_buf; 1647 ssize_t bytes_avail, in_bytes, to_consume; 1648 1649 (void) offset; /* UNUSED */ 1650 1651 /* Initialize decompressor if not yet initialized. */ 1652 if (!zip->decompress_init) { 1653 ret = zipx_lzma_alone_init(a, zip); 1654 if (ret != ARCHIVE_OK) 1655 return (ret); 1656 } 1657 1658 /* Fetch more compressed data. The same note as in deflate handler applies 1659 * here as well: 1660 * 1661 * Note: '1' here is a performance optimization. Recall that the 1662 * decompression layer returns a count of available bytes; asking for more 1663 * than that forces the decompressor to combine reads by copying data. 1664 */ 1665 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); 1666 if (bytes_avail < 0) { 1667 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1668 "Truncated lzma file body"); 1669 return (ARCHIVE_FATAL); 1670 } 1671 1672 /* Set decompressor parameters. */ 1673 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 1674 1675 zip->zipx_lzma_stream.next_in = compressed_buf; 1676 zip->zipx_lzma_stream.avail_in = in_bytes; 1677 zip->zipx_lzma_stream.total_in = 0; 1678 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; 1679 zip->zipx_lzma_stream.avail_out = 1680 /* These lzma_alone streams lack end of stream marker, so let's make 1681 * sure the unpacker won't try to unpack more than it's supposed to. */ 1682 zipmin((int64_t) zip->uncompressed_buffer_size, 1683 zip->entry->uncompressed_size - 1684 zip->entry_uncompressed_bytes_read); 1685 zip->zipx_lzma_stream.total_out = 0; 1686 1687 /* Perform the decompression. */ 1688 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); 1689 switch(lz_ret) { 1690 case LZMA_DATA_ERROR: 1691 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1692 "lzma data error (error %d)", (int) lz_ret); 1693 return (ARCHIVE_FATAL); 1694 1695 case LZMA_OK: 1696 break; 1697 1698 default: 1699 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1700 "lzma unknown error %d", (int) lz_ret); 1701 return (ARCHIVE_FATAL); 1702 } 1703 1704 to_consume = zip->zipx_lzma_stream.total_in; 1705 1706 /* Update pointers. */ 1707 __archive_read_consume(a, to_consume); 1708 zip->entry_bytes_remaining -= to_consume; 1709 zip->entry_compressed_bytes_read += to_consume; 1710 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; 1711 1712 if(zip->entry_bytes_remaining == 0) { 1713 zip->end_of_entry = 1; 1714 } 1715 1716 /* Return values. */ 1717 *size = zip->zipx_lzma_stream.total_out; 1718 *buff = zip->uncompressed_buffer; 1719 1720 /* Behave the same way as during deflate decompression. */ 1721 ret = consume_optional_marker(a, zip); 1722 if (ret != ARCHIVE_OK) 1723 return (ret); 1724 1725 /* Free lzma decoder handle because we'll no longer need it. */ 1726 if(zip->end_of_entry) { 1727 lzma_end(&zip->zipx_lzma_stream); 1728 zip->zipx_lzma_valid = 0; 1729 } 1730 1731 /* If we're here, then we're good! */ 1732 return (ARCHIVE_OK); 1733 } 1734 #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */ 1735 1736 static int 1737 zipx_ppmd8_init(struct archive_read *a, struct zip *zip) 1738 { 1739 const void* p; 1740 uint32_t val; 1741 uint32_t order; 1742 uint32_t mem; 1743 uint32_t restore_method; 1744 1745 /* Remove previous decompression context if it exists. */ 1746 if(zip->ppmd8_valid) { 1747 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 1748 zip->ppmd8_valid = 0; 1749 } 1750 1751 /* Create a new decompression context. */ 1752 __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8); 1753 1754 /* Setup function pointers required by Ppmd8 decompressor. The 1755 * 'ppmd_read' function will feed new bytes to the decompressor, 1756 * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */ 1757 zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream; 1758 zip->zipx_ppmd_stream.a = a; 1759 zip->zipx_ppmd_stream.Read = &ppmd_read; 1760 1761 /* Reset number of read bytes to 0. */ 1762 zip->zipx_ppmd_read_compressed = 0; 1763 1764 /* Read Ppmd8 header (2 bytes). */ 1765 p = __archive_read_ahead(a, 2, NULL); 1766 if(!p) { 1767 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1768 "Truncated file data in PPMd8 stream"); 1769 return (ARCHIVE_FATAL); 1770 } 1771 __archive_read_consume(a, 2); 1772 1773 /* Decode the stream's compression parameters. */ 1774 val = archive_le16dec(p); 1775 order = (val & 15) + 1; 1776 mem = ((val >> 4) & 0xff) + 1; 1777 restore_method = (val >> 12); 1778 1779 if(order < 2 || restore_method > 2) { 1780 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1781 "Invalid parameter set in PPMd8 stream (order=%d, " 1782 "restore=%d)", order, restore_method); 1783 return (ARCHIVE_FAILED); 1784 } 1785 1786 /* Allocate the memory needed to properly decompress the file. */ 1787 if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) { 1788 archive_set_error(&a->archive, ENOMEM, 1789 "Unable to allocate memory for PPMd8 stream: %d bytes", 1790 mem << 20); 1791 return (ARCHIVE_FATAL); 1792 } 1793 1794 /* Signal the cleanup function to release Ppmd8 context in the 1795 * cleanup phase. */ 1796 zip->ppmd8_valid = 1; 1797 1798 /* Perform further Ppmd8 initialization. */ 1799 if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) { 1800 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1801 "PPMd8 stream range decoder initialization error"); 1802 return (ARCHIVE_FATAL); 1803 } 1804 1805 __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order, restore_method); 1806 1807 /* Allocate the buffer that will hold uncompressed data. */ 1808 free(zip->uncompressed_buffer); 1809 1810 zip->uncompressed_buffer_size = 256 * 1024; 1811 zip->uncompressed_buffer = 1812 (uint8_t*) malloc(zip->uncompressed_buffer_size); 1813 1814 if(zip->uncompressed_buffer == NULL) { 1815 archive_set_error(&a->archive, ENOMEM, 1816 "No memory for PPMd8 decompression"); 1817 return ARCHIVE_FATAL; 1818 } 1819 1820 /* Ppmd8 initialization is done. */ 1821 zip->decompress_init = 1; 1822 1823 /* We've already read 2 bytes in the output stream. Additionally, 1824 * Ppmd8 initialization code could read some data as well. So we 1825 * are advancing the stream by 2 bytes plus whatever number of 1826 * bytes Ppmd8 init function used. */ 1827 zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed; 1828 1829 return ARCHIVE_OK; 1830 } 1831 1832 static int 1833 zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff, 1834 size_t *size, int64_t *offset) 1835 { 1836 struct zip* zip = (struct zip *)(a->format->data); 1837 int ret; 1838 size_t consumed_bytes = 0; 1839 ssize_t bytes_avail = 0; 1840 1841 (void) offset; /* UNUSED */ 1842 1843 /* If we're here for the first time, initialize Ppmd8 decompression 1844 * context first. */ 1845 if(!zip->decompress_init) { 1846 ret = zipx_ppmd8_init(a, zip); 1847 if(ret != ARCHIVE_OK) 1848 return ret; 1849 } 1850 1851 /* Fetch for more data. We're reading 1 byte here, but libarchive should 1852 * prefetch more bytes. */ 1853 (void) __archive_read_ahead(a, 1, &bytes_avail); 1854 if(bytes_avail < 0) { 1855 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1856 "Truncated PPMd8 file body"); 1857 return (ARCHIVE_FATAL); 1858 } 1859 1860 /* This counter will be updated inside ppmd_read(), which at one 1861 * point will be called by Ppmd8_DecodeSymbol. */ 1862 zip->zipx_ppmd_read_compressed = 0; 1863 1864 /* Decompression loop. */ 1865 do { 1866 int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol(&zip->ppmd8); 1867 if(sym < 0) { 1868 zip->end_of_entry = 1; 1869 break; 1870 } 1871 1872 zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym; 1873 ++consumed_bytes; 1874 } while(consumed_bytes < zip->uncompressed_buffer_size); 1875 1876 /* Update pointers for libarchive. */ 1877 *buff = zip->uncompressed_buffer; 1878 *size = consumed_bytes; 1879 1880 /* Update pointers so we can continue decompression in another call. */ 1881 zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed; 1882 zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed; 1883 zip->entry_uncompressed_bytes_read += consumed_bytes; 1884 1885 /* If we're at the end of stream, deinitialize Ppmd8 context. */ 1886 if(zip->end_of_entry) { 1887 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 1888 zip->ppmd8_valid = 0; 1889 } 1890 1891 /* Seek for optional marker, same way as in each zip entry. */ 1892 ret = consume_optional_marker(a, zip); 1893 if (ret != ARCHIVE_OK) 1894 return ret; 1895 1896 return ARCHIVE_OK; 1897 } 1898 1899 #ifdef HAVE_BZLIB_H 1900 static int 1901 zipx_bzip2_init(struct archive_read *a, struct zip *zip) 1902 { 1903 int r; 1904 1905 /* Deallocate already existing BZ2 decompression context if it 1906 * exists. */ 1907 if(zip->bzstream_valid) { 1908 BZ2_bzDecompressEnd(&zip->bzstream); 1909 zip->bzstream_valid = 0; 1910 } 1911 1912 /* Allocate a new BZ2 decompression context. */ 1913 memset(&zip->bzstream, 0, sizeof(bz_stream)); 1914 r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1); 1915 if(r != BZ_OK) { 1916 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, 1917 "bzip2 initialization failed(%d)", 1918 r); 1919 1920 return ARCHIVE_FAILED; 1921 } 1922 1923 /* Mark the bzstream field to be released in cleanup phase. */ 1924 zip->bzstream_valid = 1; 1925 1926 /* (Re)allocate the buffer that will contain decompressed bytes. */ 1927 free(zip->uncompressed_buffer); 1928 1929 zip->uncompressed_buffer_size = 256 * 1024; 1930 zip->uncompressed_buffer = 1931 (uint8_t*) malloc(zip->uncompressed_buffer_size); 1932 if (zip->uncompressed_buffer == NULL) { 1933 archive_set_error(&a->archive, ENOMEM, 1934 "No memory for bzip2 decompression"); 1935 return ARCHIVE_FATAL; 1936 } 1937 1938 /* Initialization done. */ 1939 zip->decompress_init = 1; 1940 return ARCHIVE_OK; 1941 } 1942 1943 static int 1944 zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff, 1945 size_t *size, int64_t *offset) 1946 { 1947 struct zip *zip = (struct zip *)(a->format->data); 1948 ssize_t bytes_avail = 0, in_bytes, to_consume; 1949 const void *compressed_buff; 1950 int r; 1951 uint64_t total_out; 1952 1953 (void) offset; /* UNUSED */ 1954 1955 /* Initialize decompression context if we're here for the first time. */ 1956 if(!zip->decompress_init) { 1957 r = zipx_bzip2_init(a, zip); 1958 if(r != ARCHIVE_OK) 1959 return r; 1960 } 1961 1962 /* Fetch more compressed bytes. */ 1963 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); 1964 if(bytes_avail < 0) { 1965 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1966 "Truncated bzip2 file body"); 1967 return (ARCHIVE_FATAL); 1968 } 1969 1970 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); 1971 1972 /* Setup buffer boundaries. */ 1973 zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff; 1974 zip->bzstream.avail_in = in_bytes; 1975 zip->bzstream.total_in_hi32 = 0; 1976 zip->bzstream.total_in_lo32 = 0; 1977 zip->bzstream.next_out = (char*) zip->uncompressed_buffer; 1978 zip->bzstream.avail_out = zip->uncompressed_buffer_size; 1979 zip->bzstream.total_out_hi32 = 0; 1980 zip->bzstream.total_out_lo32 = 0; 1981 1982 /* Perform the decompression. */ 1983 r = BZ2_bzDecompress(&zip->bzstream); 1984 switch(r) { 1985 case BZ_STREAM_END: 1986 /* If we're at the end of the stream, deinitialize the 1987 * decompression context now. */ 1988 switch(BZ2_bzDecompressEnd(&zip->bzstream)) { 1989 case BZ_OK: 1990 break; 1991 default: 1992 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1993 "Failed to clean up bzip2 decompressor"); 1994 return ARCHIVE_FATAL; 1995 } 1996 1997 zip->end_of_entry = 1; 1998 break; 1999 case BZ_OK: 2000 /* The decompressor has successfully decoded this chunk of 2001 * data, but more data is still in queue. */ 2002 break; 2003 default: 2004 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2005 "bzip2 decompression failed"); 2006 return ARCHIVE_FATAL; 2007 } 2008 2009 /* Update the pointers so decompressor can continue decoding. */ 2010 to_consume = zip->bzstream.total_in_lo32; 2011 __archive_read_consume(a, to_consume); 2012 2013 total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) + 2014 zip->bzstream.total_out_lo32; 2015 2016 zip->entry_bytes_remaining -= to_consume; 2017 zip->entry_compressed_bytes_read += to_consume; 2018 zip->entry_uncompressed_bytes_read += total_out; 2019 2020 /* Give libarchive its due. */ 2021 *size = total_out; 2022 *buff = zip->uncompressed_buffer; 2023 2024 /* Seek for optional marker, like in other entries. */ 2025 r = consume_optional_marker(a, zip); 2026 if(r != ARCHIVE_OK) 2027 return r; 2028 2029 return ARCHIVE_OK; 2030 } 2031 2032 #endif 2033 2034 #ifdef HAVE_ZLIB_H 2035 static int 2036 zip_deflate_init(struct archive_read *a, struct zip *zip) 2037 { 2038 int r; 2039 2040 /* If we haven't yet read any data, initialize the decompressor. */ 2041 if (!zip->decompress_init) { 2042 if (zip->stream_valid) 2043 r = inflateReset(&zip->stream); 2044 else 2045 r = inflateInit2(&zip->stream, 2046 -15 /* Don't check for zlib header */); 2047 if (r != Z_OK) { 2048 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2049 "Can't initialize ZIP decompression."); 2050 return (ARCHIVE_FATAL); 2051 } 2052 /* Stream structure has been set up. */ 2053 zip->stream_valid = 1; 2054 /* We've initialized decompression for this stream. */ 2055 zip->decompress_init = 1; 2056 } 2057 return (ARCHIVE_OK); 2058 } 2059 2060 static int 2061 zip_read_data_deflate(struct archive_read *a, const void **buff, 2062 size_t *size, int64_t *offset) 2063 { 2064 struct zip *zip; 2065 ssize_t bytes_avail; 2066 const void *compressed_buff, *sp; 2067 int r; 2068 2069 (void)offset; /* UNUSED */ 2070 2071 zip = (struct zip *)(a->format->data); 2072 2073 /* If the buffer hasn't been allocated, allocate it now. */ 2074 if (zip->uncompressed_buffer == NULL) { 2075 zip->uncompressed_buffer_size = 256 * 1024; 2076 zip->uncompressed_buffer 2077 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 2078 if (zip->uncompressed_buffer == NULL) { 2079 archive_set_error(&a->archive, ENOMEM, 2080 "No memory for ZIP decompression"); 2081 return (ARCHIVE_FATAL); 2082 } 2083 } 2084 2085 r = zip_deflate_init(a, zip); 2086 if (r != ARCHIVE_OK) 2087 return (r); 2088 2089 /* 2090 * Note: '1' here is a performance optimization. 2091 * Recall that the decompression layer returns a count of 2092 * available bytes; asking for more than that forces the 2093 * decompressor to combine reads by copying data. 2094 */ 2095 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 2096 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2097 && bytes_avail > zip->entry_bytes_remaining) { 2098 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 2099 } 2100 if (bytes_avail < 0) { 2101 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2102 "Truncated ZIP file body"); 2103 return (ARCHIVE_FATAL); 2104 } 2105 2106 if (zip->tctx_valid || zip->cctx_valid) { 2107 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 2108 size_t buff_remaining = 2109 (zip->decrypted_buffer + zip->decrypted_buffer_size) 2110 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 2111 2112 if (buff_remaining > (size_t)bytes_avail) 2113 buff_remaining = (size_t)bytes_avail; 2114 2115 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 2116 zip->entry_bytes_remaining > 0) { 2117 if ((int64_t)(zip->decrypted_bytes_remaining 2118 + buff_remaining) 2119 > zip->entry_bytes_remaining) { 2120 if (zip->entry_bytes_remaining < 2121 (int64_t)zip->decrypted_bytes_remaining) 2122 buff_remaining = 0; 2123 else 2124 buff_remaining = 2125 (size_t)zip->entry_bytes_remaining 2126 - zip->decrypted_bytes_remaining; 2127 } 2128 } 2129 if (buff_remaining > 0) { 2130 if (zip->tctx_valid) { 2131 trad_enc_decrypt_update(&zip->tctx, 2132 compressed_buff, buff_remaining, 2133 zip->decrypted_ptr 2134 + zip->decrypted_bytes_remaining, 2135 buff_remaining); 2136 } else { 2137 size_t dsize = buff_remaining; 2138 archive_decrypto_aes_ctr_update( 2139 &zip->cctx, 2140 compressed_buff, buff_remaining, 2141 zip->decrypted_ptr 2142 + zip->decrypted_bytes_remaining, 2143 &dsize); 2144 } 2145 zip->decrypted_bytes_remaining += buff_remaining; 2146 } 2147 } 2148 bytes_avail = zip->decrypted_bytes_remaining; 2149 compressed_buff = (const char *)zip->decrypted_ptr; 2150 } 2151 2152 /* 2153 * A bug in zlib.h: stream.next_in should be marked 'const' 2154 * but isn't (the library never alters data through the 2155 * next_in pointer, only reads it). The result: this ugly 2156 * cast to remove 'const'. 2157 */ 2158 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 2159 zip->stream.avail_in = (uInt)bytes_avail; 2160 zip->stream.total_in = 0; 2161 zip->stream.next_out = zip->uncompressed_buffer; 2162 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 2163 zip->stream.total_out = 0; 2164 2165 r = inflate(&zip->stream, 0); 2166 switch (r) { 2167 case Z_OK: 2168 break; 2169 case Z_STREAM_END: 2170 zip->end_of_entry = 1; 2171 break; 2172 case Z_MEM_ERROR: 2173 archive_set_error(&a->archive, ENOMEM, 2174 "Out of memory for ZIP decompression"); 2175 return (ARCHIVE_FATAL); 2176 default: 2177 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2178 "ZIP decompression failed (%d)", r); 2179 return (ARCHIVE_FATAL); 2180 } 2181 2182 /* Consume as much as the compressor actually used. */ 2183 bytes_avail = zip->stream.total_in; 2184 if (zip->tctx_valid || zip->cctx_valid) { 2185 zip->decrypted_bytes_remaining -= bytes_avail; 2186 if (zip->decrypted_bytes_remaining == 0) 2187 zip->decrypted_ptr = zip->decrypted_buffer; 2188 else 2189 zip->decrypted_ptr += bytes_avail; 2190 } 2191 /* Calculate compressed data as much as we used.*/ 2192 if (zip->hctx_valid) 2193 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 2194 __archive_read_consume(a, bytes_avail); 2195 zip->entry_bytes_remaining -= bytes_avail; 2196 zip->entry_compressed_bytes_read += bytes_avail; 2197 2198 *size = zip->stream.total_out; 2199 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 2200 *buff = zip->uncompressed_buffer; 2201 2202 if (zip->end_of_entry && zip->hctx_valid) { 2203 r = check_authentication_code(a, NULL); 2204 if (r != ARCHIVE_OK) 2205 return (r); 2206 } 2207 2208 r = consume_optional_marker(a, zip); 2209 if (r != ARCHIVE_OK) 2210 return (r); 2211 2212 return (ARCHIVE_OK); 2213 } 2214 #endif 2215 2216 static int 2217 read_decryption_header(struct archive_read *a) 2218 { 2219 struct zip *zip = (struct zip *)(a->format->data); 2220 const char *p; 2221 unsigned int remaining_size; 2222 unsigned int ts; 2223 2224 /* 2225 * Read an initialization vector data field. 2226 */ 2227 p = __archive_read_ahead(a, 2, NULL); 2228 if (p == NULL) 2229 goto truncated; 2230 ts = zip->iv_size; 2231 zip->iv_size = archive_le16dec(p); 2232 __archive_read_consume(a, 2); 2233 if (ts < zip->iv_size) { 2234 free(zip->iv); 2235 zip->iv = NULL; 2236 } 2237 p = __archive_read_ahead(a, zip->iv_size, NULL); 2238 if (p == NULL) 2239 goto truncated; 2240 if (zip->iv == NULL) { 2241 zip->iv = malloc(zip->iv_size); 2242 if (zip->iv == NULL) 2243 goto nomem; 2244 } 2245 memcpy(zip->iv, p, zip->iv_size); 2246 __archive_read_consume(a, zip->iv_size); 2247 2248 /* 2249 * Read a size of remaining decryption header field. 2250 */ 2251 p = __archive_read_ahead(a, 14, NULL); 2252 if (p == NULL) 2253 goto truncated; 2254 remaining_size = archive_le32dec(p); 2255 if (remaining_size < 16 || remaining_size > (1 << 18)) 2256 goto corrupted; 2257 2258 /* Check if format version is supported. */ 2259 if (archive_le16dec(p+4) != 3) { 2260 archive_set_error(&a->archive, 2261 ARCHIVE_ERRNO_FILE_FORMAT, 2262 "Unsupported encryption format version: %u", 2263 archive_le16dec(p+4)); 2264 return (ARCHIVE_FAILED); 2265 } 2266 2267 /* 2268 * Read an encryption algorithm field. 2269 */ 2270 zip->alg_id = archive_le16dec(p+6); 2271 switch (zip->alg_id) { 2272 case 0x6601:/* DES */ 2273 case 0x6602:/* RC2 */ 2274 case 0x6603:/* 3DES 168 */ 2275 case 0x6609:/* 3DES 112 */ 2276 case 0x660E:/* AES 128 */ 2277 case 0x660F:/* AES 192 */ 2278 case 0x6610:/* AES 256 */ 2279 case 0x6702:/* RC2 (version >= 5.2) */ 2280 case 0x6720:/* Blowfish */ 2281 case 0x6721:/* Twofish */ 2282 case 0x6801:/* RC4 */ 2283 /* Supported encryption algorithm. */ 2284 break; 2285 default: 2286 archive_set_error(&a->archive, 2287 ARCHIVE_ERRNO_FILE_FORMAT, 2288 "Unknown encryption algorithm: %u", zip->alg_id); 2289 return (ARCHIVE_FAILED); 2290 } 2291 2292 /* 2293 * Read a bit length field. 2294 */ 2295 zip->bit_len = archive_le16dec(p+8); 2296 2297 /* 2298 * Read a flags field. 2299 */ 2300 zip->flags = archive_le16dec(p+10); 2301 switch (zip->flags & 0xf000) { 2302 case 0x0001: /* Password is required to decrypt. */ 2303 case 0x0002: /* Certificates only. */ 2304 case 0x0003: /* Password or certificate required to decrypt. */ 2305 break; 2306 default: 2307 archive_set_error(&a->archive, 2308 ARCHIVE_ERRNO_FILE_FORMAT, 2309 "Unknown encryption flag: %u", zip->flags); 2310 return (ARCHIVE_FAILED); 2311 } 2312 if ((zip->flags & 0xf000) == 0 || 2313 (zip->flags & 0xf000) == 0x4000) { 2314 archive_set_error(&a->archive, 2315 ARCHIVE_ERRNO_FILE_FORMAT, 2316 "Unknown encryption flag: %u", zip->flags); 2317 return (ARCHIVE_FAILED); 2318 } 2319 2320 /* 2321 * Read an encrypted random data field. 2322 */ 2323 ts = zip->erd_size; 2324 zip->erd_size = archive_le16dec(p+12); 2325 __archive_read_consume(a, 14); 2326 if ((zip->erd_size & 0xf) != 0 || 2327 (zip->erd_size + 16) > remaining_size || 2328 (zip->erd_size + 16) < zip->erd_size) 2329 goto corrupted; 2330 2331 if (ts < zip->erd_size) { 2332 free(zip->erd); 2333 zip->erd = NULL; 2334 } 2335 p = __archive_read_ahead(a, zip->erd_size, NULL); 2336 if (p == NULL) 2337 goto truncated; 2338 if (zip->erd == NULL) { 2339 zip->erd = malloc(zip->erd_size); 2340 if (zip->erd == NULL) 2341 goto nomem; 2342 } 2343 memcpy(zip->erd, p, zip->erd_size); 2344 __archive_read_consume(a, zip->erd_size); 2345 2346 /* 2347 * Read a reserved data field. 2348 */ 2349 p = __archive_read_ahead(a, 4, NULL); 2350 if (p == NULL) 2351 goto truncated; 2352 /* Reserved data size should be zero. */ 2353 if (archive_le32dec(p) != 0) 2354 goto corrupted; 2355 __archive_read_consume(a, 4); 2356 2357 /* 2358 * Read a password validation data field. 2359 */ 2360 p = __archive_read_ahead(a, 2, NULL); 2361 if (p == NULL) 2362 goto truncated; 2363 ts = zip->v_size; 2364 zip->v_size = archive_le16dec(p); 2365 __archive_read_consume(a, 2); 2366 if ((zip->v_size & 0x0f) != 0 || 2367 (zip->erd_size + zip->v_size + 16) > remaining_size || 2368 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 2369 goto corrupted; 2370 if (ts < zip->v_size) { 2371 free(zip->v_data); 2372 zip->v_data = NULL; 2373 } 2374 p = __archive_read_ahead(a, zip->v_size, NULL); 2375 if (p == NULL) 2376 goto truncated; 2377 if (zip->v_data == NULL) { 2378 zip->v_data = malloc(zip->v_size); 2379 if (zip->v_data == NULL) 2380 goto nomem; 2381 } 2382 memcpy(zip->v_data, p, zip->v_size); 2383 __archive_read_consume(a, zip->v_size); 2384 2385 p = __archive_read_ahead(a, 4, NULL); 2386 if (p == NULL) 2387 goto truncated; 2388 zip->v_crc32 = archive_le32dec(p); 2389 __archive_read_consume(a, 4); 2390 2391 /*return (ARCHIVE_OK); 2392 * This is not fully implemented yet.*/ 2393 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2394 "Encrypted file is unsupported"); 2395 return (ARCHIVE_FAILED); 2396 truncated: 2397 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2398 "Truncated ZIP file data"); 2399 return (ARCHIVE_FATAL); 2400 corrupted: 2401 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2402 "Corrupted ZIP file data"); 2403 return (ARCHIVE_FATAL); 2404 nomem: 2405 archive_set_error(&a->archive, ENOMEM, 2406 "No memory for ZIP decryption"); 2407 return (ARCHIVE_FATAL); 2408 } 2409 2410 static int 2411 zip_alloc_decryption_buffer(struct archive_read *a) 2412 { 2413 struct zip *zip = (struct zip *)(a->format->data); 2414 size_t bs = 256 * 1024; 2415 2416 if (zip->decrypted_buffer == NULL) { 2417 zip->decrypted_buffer_size = bs; 2418 zip->decrypted_buffer = malloc(bs); 2419 if (zip->decrypted_buffer == NULL) { 2420 archive_set_error(&a->archive, ENOMEM, 2421 "No memory for ZIP decryption"); 2422 return (ARCHIVE_FATAL); 2423 } 2424 } 2425 zip->decrypted_ptr = zip->decrypted_buffer; 2426 return (ARCHIVE_OK); 2427 } 2428 2429 static int 2430 init_traditional_PKWARE_decryption(struct archive_read *a) 2431 { 2432 struct zip *zip = (struct zip *)(a->format->data); 2433 const void *p; 2434 int retry; 2435 int r; 2436 2437 if (zip->tctx_valid) 2438 return (ARCHIVE_OK); 2439 2440 /* 2441 Read the 12 bytes encryption header stored at 2442 the start of the data area. 2443 */ 2444 #define ENC_HEADER_SIZE 12 2445 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2446 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 2447 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2448 "Truncated Zip encrypted body: only %jd bytes available", 2449 (intmax_t)zip->entry_bytes_remaining); 2450 return (ARCHIVE_FATAL); 2451 } 2452 2453 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 2454 if (p == NULL) { 2455 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2456 "Truncated ZIP file data"); 2457 return (ARCHIVE_FATAL); 2458 } 2459 2460 for (retry = 0;; retry++) { 2461 const char *passphrase; 2462 uint8_t crcchk; 2463 2464 passphrase = __archive_read_next_passphrase(a); 2465 if (passphrase == NULL) { 2466 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2467 (retry > 0)? 2468 "Incorrect passphrase": 2469 "Passphrase required for this entry"); 2470 return (ARCHIVE_FAILED); 2471 } 2472 2473 /* 2474 * Initialize ctx for Traditional PKWARE Decryption. 2475 */ 2476 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 2477 p, ENC_HEADER_SIZE, &crcchk); 2478 if (r == 0 && crcchk == zip->entry->decdat) 2479 break;/* The passphrase is OK. */ 2480 if (retry > 10000) { 2481 /* Avoid infinity loop. */ 2482 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2483 "Too many incorrect passphrases"); 2484 return (ARCHIVE_FAILED); 2485 } 2486 } 2487 2488 __archive_read_consume(a, ENC_HEADER_SIZE); 2489 zip->tctx_valid = 1; 2490 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 2491 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 2492 } 2493 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 2494 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 2495 zip->decrypted_bytes_remaining = 0; 2496 2497 return (zip_alloc_decryption_buffer(a)); 2498 #undef ENC_HEADER_SIZE 2499 } 2500 2501 static int 2502 init_WinZip_AES_decryption(struct archive_read *a) 2503 { 2504 struct zip *zip = (struct zip *)(a->format->data); 2505 const void *p; 2506 const uint8_t *pv; 2507 size_t key_len, salt_len; 2508 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 2509 int retry; 2510 int r; 2511 2512 if (zip->cctx_valid || zip->hctx_valid) 2513 return (ARCHIVE_OK); 2514 2515 switch (zip->entry->aes_extra.strength) { 2516 case 1: salt_len = 8; key_len = 16; break; 2517 case 2: salt_len = 12; key_len = 24; break; 2518 case 3: salt_len = 16; key_len = 32; break; 2519 default: goto corrupted; 2520 } 2521 p = __archive_read_ahead(a, salt_len + 2, NULL); 2522 if (p == NULL) 2523 goto truncated; 2524 2525 for (retry = 0;; retry++) { 2526 const char *passphrase; 2527 2528 passphrase = __archive_read_next_passphrase(a); 2529 if (passphrase == NULL) { 2530 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2531 (retry > 0)? 2532 "Incorrect passphrase": 2533 "Passphrase required for this entry"); 2534 return (ARCHIVE_FAILED); 2535 } 2536 memset(derived_key, 0, sizeof(derived_key)); 2537 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 2538 p, salt_len, 1000, derived_key, key_len * 2 + 2); 2539 if (r != 0) { 2540 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2541 "Decryption is unsupported due to lack of " 2542 "crypto library"); 2543 return (ARCHIVE_FAILED); 2544 } 2545 2546 /* Check password verification value. */ 2547 pv = ((const uint8_t *)p) + salt_len; 2548 if (derived_key[key_len * 2] == pv[0] && 2549 derived_key[key_len * 2 + 1] == pv[1]) 2550 break;/* The passphrase is OK. */ 2551 if (retry > 10000) { 2552 /* Avoid infinity loop. */ 2553 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2554 "Too many incorrect passphrases"); 2555 return (ARCHIVE_FAILED); 2556 } 2557 } 2558 2559 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 2560 if (r != 0) { 2561 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2562 "Decryption is unsupported due to lack of crypto library"); 2563 return (ARCHIVE_FAILED); 2564 } 2565 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 2566 if (r != 0) { 2567 archive_decrypto_aes_ctr_release(&zip->cctx); 2568 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2569 "Failed to initialize HMAC-SHA1"); 2570 return (ARCHIVE_FAILED); 2571 } 2572 zip->cctx_valid = zip->hctx_valid = 1; 2573 __archive_read_consume(a, salt_len + 2); 2574 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 2575 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2576 && zip->entry_bytes_remaining < 0) 2577 goto corrupted; 2578 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 2579 zip->decrypted_bytes_remaining = 0; 2580 2581 zip->entry->compression = zip->entry->aes_extra.compression; 2582 return (zip_alloc_decryption_buffer(a)); 2583 2584 truncated: 2585 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2586 "Truncated ZIP file data"); 2587 return (ARCHIVE_FATAL); 2588 corrupted: 2589 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2590 "Corrupted ZIP file data"); 2591 return (ARCHIVE_FATAL); 2592 } 2593 2594 static int 2595 archive_read_format_zip_read_data(struct archive_read *a, 2596 const void **buff, size_t *size, int64_t *offset) 2597 { 2598 int r; 2599 struct zip *zip = (struct zip *)(a->format->data); 2600 2601 if (zip->has_encrypted_entries == 2602 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 2603 zip->has_encrypted_entries = 0; 2604 } 2605 2606 *offset = zip->entry_uncompressed_bytes_read; 2607 *size = 0; 2608 *buff = NULL; 2609 2610 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 2611 if (zip->end_of_entry) 2612 return (ARCHIVE_EOF); 2613 2614 /* Return EOF immediately if this is a non-regular file. */ 2615 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 2616 return (ARCHIVE_EOF); 2617 2618 __archive_read_consume(a, zip->unconsumed); 2619 zip->unconsumed = 0; 2620 2621 if (zip->init_decryption) { 2622 zip->has_encrypted_entries = 1; 2623 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2624 r = read_decryption_header(a); 2625 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2626 r = init_WinZip_AES_decryption(a); 2627 else 2628 r = init_traditional_PKWARE_decryption(a); 2629 if (r != ARCHIVE_OK) 2630 return (r); 2631 zip->init_decryption = 0; 2632 } 2633 2634 switch(zip->entry->compression) { 2635 case 0: /* No compression. */ 2636 r = zip_read_data_none(a, buff, size, offset); 2637 break; 2638 #ifdef HAVE_BZLIB_H 2639 case 12: /* ZIPx bzip2 compression. */ 2640 r = zip_read_data_zipx_bzip2(a, buff, size, offset); 2641 break; 2642 #endif 2643 #if HAVE_LZMA_H && HAVE_LIBLZMA 2644 case 14: /* ZIPx LZMA compression. */ 2645 r = zip_read_data_zipx_lzma_alone(a, buff, size, offset); 2646 break; 2647 case 95: /* ZIPx XZ compression. */ 2648 r = zip_read_data_zipx_xz(a, buff, size, offset); 2649 break; 2650 #endif 2651 /* PPMd support is built-in, so we don't need any #if guards. */ 2652 case 98: /* ZIPx PPMd compression. */ 2653 r = zip_read_data_zipx_ppmd(a, buff, size, offset); 2654 break; 2655 2656 #ifdef HAVE_ZLIB_H 2657 case 8: /* Deflate compression. */ 2658 r = zip_read_data_deflate(a, buff, size, offset); 2659 break; 2660 #endif 2661 default: /* Unsupported compression. */ 2662 /* Return a warning. */ 2663 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2664 "Unsupported ZIP compression method (%d: %s)", 2665 zip->entry->compression, compression_name(zip->entry->compression)); 2666 /* We can't decompress this entry, but we will 2667 * be able to skip() it and try the next entry. */ 2668 return (ARCHIVE_FAILED); 2669 break; 2670 } 2671 if (r != ARCHIVE_OK) 2672 return (r); 2673 /* Update checksum */ 2674 if (*size) 2675 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 2676 (unsigned)*size); 2677 /* If we hit the end, swallow any end-of-data marker. */ 2678 if (zip->end_of_entry) { 2679 /* Check file size, CRC against these values. */ 2680 if (zip->entry->compressed_size != 2681 zip->entry_compressed_bytes_read) { 2682 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2683 "ZIP compressed data is wrong size " 2684 "(read %jd, expected %jd)", 2685 (intmax_t)zip->entry_compressed_bytes_read, 2686 (intmax_t)zip->entry->compressed_size); 2687 return (ARCHIVE_WARN); 2688 } 2689 /* Size field only stores the lower 32 bits of the actual 2690 * size. */ 2691 if ((zip->entry->uncompressed_size & UINT32_MAX) 2692 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 2693 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2694 "ZIP uncompressed data is wrong size " 2695 "(read %jd, expected %jd)\n", 2696 (intmax_t)zip->entry_uncompressed_bytes_read, 2697 (intmax_t)zip->entry->uncompressed_size); 2698 return (ARCHIVE_WARN); 2699 } 2700 /* Check computed CRC against header */ 2701 if ((!zip->hctx_valid || 2702 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 2703 zip->entry->crc32 != zip->entry_crc32 2704 && !zip->ignore_crc32) { 2705 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2706 "ZIP bad CRC: 0x%lx should be 0x%lx", 2707 (unsigned long)zip->entry_crc32, 2708 (unsigned long)zip->entry->crc32); 2709 return (ARCHIVE_WARN); 2710 } 2711 } 2712 2713 return (ARCHIVE_OK); 2714 } 2715 2716 static int 2717 archive_read_format_zip_cleanup(struct archive_read *a) 2718 { 2719 struct zip *zip; 2720 struct zip_entry *zip_entry, *next_zip_entry; 2721 2722 zip = (struct zip *)(a->format->data); 2723 2724 #ifdef HAVE_ZLIB_H 2725 if (zip->stream_valid) 2726 inflateEnd(&zip->stream); 2727 #endif 2728 2729 #if HAVA_LZMA_H && HAVE_LIBLZMA 2730 if (zip->zipx_lzma_valid) { 2731 lzma_end(&zip->zipx_lzma_stream); 2732 } 2733 #endif 2734 2735 #ifdef HAVE_BZLIB_H 2736 if (zip->bzstream_valid) { 2737 BZ2_bzDecompressEnd(&zip->bzstream); 2738 } 2739 #endif 2740 2741 free(zip->uncompressed_buffer); 2742 2743 if (zip->ppmd8_valid) 2744 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); 2745 2746 if (zip->zip_entries) { 2747 zip_entry = zip->zip_entries; 2748 while (zip_entry != NULL) { 2749 next_zip_entry = zip_entry->next; 2750 archive_string_free(&zip_entry->rsrcname); 2751 free(zip_entry); 2752 zip_entry = next_zip_entry; 2753 } 2754 } 2755 free(zip->decrypted_buffer); 2756 if (zip->cctx_valid) 2757 archive_decrypto_aes_ctr_release(&zip->cctx); 2758 if (zip->hctx_valid) 2759 archive_hmac_sha1_cleanup(&zip->hctx); 2760 free(zip->iv); 2761 free(zip->erd); 2762 free(zip->v_data); 2763 archive_string_free(&zip->format_name); 2764 free(zip); 2765 (a->format->data) = NULL; 2766 return (ARCHIVE_OK); 2767 } 2768 2769 static int 2770 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 2771 { 2772 if (_a && _a->format) { 2773 struct zip * zip = (struct zip *)_a->format->data; 2774 if (zip) { 2775 return zip->has_encrypted_entries; 2776 } 2777 } 2778 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2779 } 2780 2781 static int 2782 archive_read_format_zip_options(struct archive_read *a, 2783 const char *key, const char *val) 2784 { 2785 struct zip *zip; 2786 int ret = ARCHIVE_FAILED; 2787 2788 zip = (struct zip *)(a->format->data); 2789 if (strcmp(key, "compat-2x") == 0) { 2790 /* Handle filenames as libarchive 2.x */ 2791 zip->init_default_conversion = (val != NULL) ? 1 : 0; 2792 return (ARCHIVE_OK); 2793 } else if (strcmp(key, "hdrcharset") == 0) { 2794 if (val == NULL || val[0] == 0) 2795 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2796 "zip: hdrcharset option needs a character-set name" 2797 ); 2798 else { 2799 zip->sconv = archive_string_conversion_from_charset( 2800 &a->archive, val, 0); 2801 if (zip->sconv != NULL) { 2802 if (strcmp(val, "UTF-8") == 0) 2803 zip->sconv_utf8 = zip->sconv; 2804 ret = ARCHIVE_OK; 2805 } else 2806 ret = ARCHIVE_FATAL; 2807 } 2808 return (ret); 2809 } else if (strcmp(key, "ignorecrc32") == 0) { 2810 /* Mostly useful for testing. */ 2811 if (val == NULL || val[0] == 0) { 2812 zip->crc32func = real_crc32; 2813 zip->ignore_crc32 = 0; 2814 } else { 2815 zip->crc32func = fake_crc32; 2816 zip->ignore_crc32 = 1; 2817 } 2818 return (ARCHIVE_OK); 2819 } else if (strcmp(key, "mac-ext") == 0) { 2820 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2821 return (ARCHIVE_OK); 2822 } 2823 2824 /* Note: The "warn" return is just to inform the options 2825 * supervisor that we didn't handle it. It will generate 2826 * a suitable error if no one used this option. */ 2827 return (ARCHIVE_WARN); 2828 } 2829 2830 int 2831 archive_read_support_format_zip(struct archive *a) 2832 { 2833 int r; 2834 r = archive_read_support_format_zip_streamable(a); 2835 if (r != ARCHIVE_OK) 2836 return r; 2837 return (archive_read_support_format_zip_seekable(a)); 2838 } 2839 2840 /* ------------------------------------------------------------------------ */ 2841 2842 /* 2843 * Streaming-mode support 2844 */ 2845 2846 2847 static int 2848 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2849 { 2850 (void)a; /* UNUSED */ 2851 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2852 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2853 } 2854 2855 static int 2856 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2857 { 2858 const char *p; 2859 2860 (void)best_bid; /* UNUSED */ 2861 2862 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2863 return (-1); 2864 2865 /* 2866 * Bid of 29 here comes from: 2867 * + 16 bits for "PK", 2868 * + next 16-bit field has 6 options so contributes 2869 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2870 * 2871 * So we've effectively verified ~29 total bits of check data. 2872 */ 2873 if (p[0] == 'P' && p[1] == 'K') { 2874 if ((p[2] == '\001' && p[3] == '\002') 2875 || (p[2] == '\003' && p[3] == '\004') 2876 || (p[2] == '\005' && p[3] == '\006') 2877 || (p[2] == '\006' && p[3] == '\006') 2878 || (p[2] == '\007' && p[3] == '\010') 2879 || (p[2] == '0' && p[3] == '0')) 2880 return (29); 2881 } 2882 2883 /* TODO: It's worth looking ahead a little bit for a valid 2884 * PK signature. In particular, that would make it possible 2885 * to read some UUEncoded SFX files or SFX files coming from 2886 * a network socket. */ 2887 2888 return (0); 2889 } 2890 2891 static int 2892 archive_read_format_zip_streamable_read_header(struct archive_read *a, 2893 struct archive_entry *entry) 2894 { 2895 struct zip *zip; 2896 2897 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2898 if (a->archive.archive_format_name == NULL) 2899 a->archive.archive_format_name = "ZIP"; 2900 2901 zip = (struct zip *)(a->format->data); 2902 2903 /* 2904 * It should be sufficient to call archive_read_next_header() for 2905 * a reader to determine if an entry is encrypted or not. If the 2906 * encryption of an entry is only detectable when calling 2907 * archive_read_data(), so be it. We'll do the same check there 2908 * as well. 2909 */ 2910 if (zip->has_encrypted_entries == 2911 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2912 zip->has_encrypted_entries = 0; 2913 2914 /* Make sure we have a zip_entry structure to use. */ 2915 if (zip->zip_entries == NULL) { 2916 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2917 if (zip->zip_entries == NULL) { 2918 archive_set_error(&a->archive, ENOMEM, 2919 "Out of memory"); 2920 return ARCHIVE_FATAL; 2921 } 2922 } 2923 zip->entry = zip->zip_entries; 2924 memset(zip->entry, 0, sizeof(struct zip_entry)); 2925 2926 if (zip->cctx_valid) 2927 archive_decrypto_aes_ctr_release(&zip->cctx); 2928 if (zip->hctx_valid) 2929 archive_hmac_sha1_cleanup(&zip->hctx); 2930 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2931 __archive_read_reset_passphrase(a); 2932 2933 /* Search ahead for the next local file header. */ 2934 __archive_read_consume(a, zip->unconsumed); 2935 zip->unconsumed = 0; 2936 for (;;) { 2937 int64_t skipped = 0; 2938 const char *p, *end; 2939 ssize_t bytes; 2940 2941 p = __archive_read_ahead(a, 4, &bytes); 2942 if (p == NULL) 2943 return (ARCHIVE_FATAL); 2944 end = p + bytes; 2945 2946 while (p + 4 <= end) { 2947 if (p[0] == 'P' && p[1] == 'K') { 2948 if (p[2] == '\003' && p[3] == '\004') { 2949 /* Regular file entry. */ 2950 __archive_read_consume(a, skipped); 2951 return zip_read_local_file_header(a, 2952 entry, zip); 2953 } 2954 2955 /* 2956 * TODO: We cannot restore permissions 2957 * based only on the local file headers. 2958 * Consider scanning the central 2959 * directory and returning additional 2960 * entries for at least directories. 2961 * This would allow us to properly set 2962 * directory permissions. 2963 * 2964 * This won't help us fix symlinks 2965 * and may not help with regular file 2966 * permissions, either. <sigh> 2967 */ 2968 if (p[2] == '\001' && p[3] == '\002') { 2969 return (ARCHIVE_EOF); 2970 } 2971 2972 /* End of central directory? Must be an 2973 * empty archive. */ 2974 if ((p[2] == '\005' && p[3] == '\006') 2975 || (p[2] == '\006' && p[3] == '\006')) 2976 return (ARCHIVE_EOF); 2977 } 2978 ++p; 2979 ++skipped; 2980 } 2981 __archive_read_consume(a, skipped); 2982 } 2983 } 2984 2985 static int 2986 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2987 { 2988 struct zip *zip; 2989 int64_t bytes_skipped; 2990 2991 zip = (struct zip *)(a->format->data); 2992 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2993 zip->unconsumed = 0; 2994 if (bytes_skipped < 0) 2995 return (ARCHIVE_FATAL); 2996 2997 /* If we've already read to end of data, we're done. */ 2998 if (zip->end_of_entry) 2999 return (ARCHIVE_OK); 3000 3001 /* So we know we're streaming... */ 3002 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 3003 || zip->entry->compressed_size > 0) { 3004 /* We know the compressed length, so we can just skip. */ 3005 bytes_skipped = __archive_read_consume(a, 3006 zip->entry_bytes_remaining); 3007 if (bytes_skipped < 0) 3008 return (ARCHIVE_FATAL); 3009 return (ARCHIVE_OK); 3010 } 3011 3012 if (zip->init_decryption) { 3013 int r; 3014 3015 zip->has_encrypted_entries = 1; 3016 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 3017 r = read_decryption_header(a); 3018 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 3019 r = init_WinZip_AES_decryption(a); 3020 else 3021 r = init_traditional_PKWARE_decryption(a); 3022 if (r != ARCHIVE_OK) 3023 return (r); 3024 zip->init_decryption = 0; 3025 } 3026 3027 /* We're streaming and we don't know the length. */ 3028 /* If the body is compressed and we know the format, we can 3029 * find an exact end-of-entry by decompressing it. */ 3030 switch (zip->entry->compression) { 3031 #ifdef HAVE_ZLIB_H 3032 case 8: /* Deflate compression. */ 3033 while (!zip->end_of_entry) { 3034 int64_t offset = 0; 3035 const void *buff = NULL; 3036 size_t size = 0; 3037 int r; 3038 r = zip_read_data_deflate(a, &buff, &size, &offset); 3039 if (r != ARCHIVE_OK) 3040 return (r); 3041 } 3042 return ARCHIVE_OK; 3043 #endif 3044 default: /* Uncompressed or unknown. */ 3045 /* Scan for a PK\007\010 signature. */ 3046 for (;;) { 3047 const char *p, *buff; 3048 ssize_t bytes_avail; 3049 buff = __archive_read_ahead(a, 16, &bytes_avail); 3050 if (bytes_avail < 16) { 3051 archive_set_error(&a->archive, 3052 ARCHIVE_ERRNO_FILE_FORMAT, 3053 "Truncated ZIP file data"); 3054 return (ARCHIVE_FATAL); 3055 } 3056 p = buff; 3057 while (p <= buff + bytes_avail - 16) { 3058 if (p[3] == 'P') { p += 3; } 3059 else if (p[3] == 'K') { p += 2; } 3060 else if (p[3] == '\007') { p += 1; } 3061 else if (p[3] == '\010' && p[2] == '\007' 3062 && p[1] == 'K' && p[0] == 'P') { 3063 if (zip->entry->flags & LA_USED_ZIP64) 3064 __archive_read_consume(a, 3065 p - buff + 24); 3066 else 3067 __archive_read_consume(a, 3068 p - buff + 16); 3069 return ARCHIVE_OK; 3070 } else { p += 4; } 3071 } 3072 __archive_read_consume(a, p - buff); 3073 } 3074 } 3075 } 3076 3077 int 3078 archive_read_support_format_zip_streamable(struct archive *_a) 3079 { 3080 struct archive_read *a = (struct archive_read *)_a; 3081 struct zip *zip; 3082 int r; 3083 3084 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3085 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 3086 3087 zip = (struct zip *)calloc(1, sizeof(*zip)); 3088 if (zip == NULL) { 3089 archive_set_error(&a->archive, ENOMEM, 3090 "Can't allocate zip data"); 3091 return (ARCHIVE_FATAL); 3092 } 3093 3094 /* Streamable reader doesn't support mac extensions. */ 3095 zip->process_mac_extensions = 0; 3096 3097 /* 3098 * Until enough data has been read, we cannot tell about 3099 * any encrypted entries yet. 3100 */ 3101 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3102 zip->crc32func = real_crc32; 3103 3104 r = __archive_read_register_format(a, 3105 zip, 3106 "zip", 3107 archive_read_format_zip_streamable_bid, 3108 archive_read_format_zip_options, 3109 archive_read_format_zip_streamable_read_header, 3110 archive_read_format_zip_read_data, 3111 archive_read_format_zip_read_data_skip_streamable, 3112 NULL, 3113 archive_read_format_zip_cleanup, 3114 archive_read_support_format_zip_capabilities_streamable, 3115 archive_read_format_zip_has_encrypted_entries); 3116 3117 if (r != ARCHIVE_OK) 3118 free(zip); 3119 return (ARCHIVE_OK); 3120 } 3121 3122 /* ------------------------------------------------------------------------ */ 3123 3124 /* 3125 * Seeking-mode support 3126 */ 3127 3128 static int 3129 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 3130 { 3131 (void)a; /* UNUSED */ 3132 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 3133 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 3134 } 3135 3136 /* 3137 * TODO: This is a performance sink because it forces the read core to 3138 * drop buffered data from the start of file, which will then have to 3139 * be re-read again if this bidder loses. 3140 * 3141 * We workaround this a little by passing in the best bid so far so 3142 * that later bidders can do nothing if they know they'll never 3143 * outbid. But we can certainly do better... 3144 */ 3145 static int 3146 read_eocd(struct zip *zip, const char *p, int64_t current_offset) 3147 { 3148 /* Sanity-check the EOCD we've found. */ 3149 3150 /* This must be the first volume. */ 3151 if (archive_le16dec(p + 4) != 0) 3152 return 0; 3153 /* Central directory must be on this volume. */ 3154 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 3155 return 0; 3156 /* All central directory entries must be on this volume. */ 3157 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 3158 return 0; 3159 /* Central directory can't extend beyond start of EOCD record. */ 3160 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 3161 > current_offset) 3162 return 0; 3163 3164 /* Save the central directory location for later use. */ 3165 zip->central_directory_offset = archive_le32dec(p + 16); 3166 3167 /* This is just a tiny bit higher than the maximum 3168 returned by the streaming Zip bidder. This ensures 3169 that the more accurate seeking Zip parser wins 3170 whenever seek is available. */ 3171 return 32; 3172 } 3173 3174 /* 3175 * Examine Zip64 EOCD locator: If it's valid, store the information 3176 * from it. 3177 */ 3178 static int 3179 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 3180 { 3181 int64_t eocd64_offset; 3182 int64_t eocd64_size; 3183 3184 /* Sanity-check the locator record. */ 3185 3186 /* Central dir must be on first volume. */ 3187 if (archive_le32dec(p + 4) != 0) 3188 return 0; 3189 /* Must be only a single volume. */ 3190 if (archive_le32dec(p + 16) != 1) 3191 return 0; 3192 3193 /* Find the Zip64 EOCD record. */ 3194 eocd64_offset = archive_le64dec(p + 8); 3195 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 3196 return 0; 3197 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 3198 return 0; 3199 /* Make sure we can read all of it. */ 3200 eocd64_size = archive_le64dec(p + 4) + 12; 3201 if (eocd64_size < 56 || eocd64_size > 16384) 3202 return 0; 3203 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 3204 return 0; 3205 3206 /* Sanity-check the EOCD64 */ 3207 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 3208 return 0; 3209 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 3210 return 0; 3211 /* CD can't be split. */ 3212 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 3213 return 0; 3214 3215 /* Save the central directory offset for later use. */ 3216 zip->central_directory_offset = archive_le64dec(p + 48); 3217 3218 return 32; 3219 } 3220 3221 static int 3222 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 3223 { 3224 struct zip *zip = (struct zip *)a->format->data; 3225 int64_t file_size, current_offset; 3226 const char *p; 3227 int i, tail; 3228 3229 /* If someone has already bid more than 32, then avoid 3230 trashing the look-ahead buffers with a seek. */ 3231 if (best_bid > 32) 3232 return (-1); 3233 3234 file_size = __archive_read_seek(a, 0, SEEK_END); 3235 if (file_size <= 0) 3236 return 0; 3237 3238 /* Search last 16k of file for end-of-central-directory 3239 * record (which starts with PK\005\006) */ 3240 tail = (int)zipmin(1024 * 16, file_size); 3241 current_offset = __archive_read_seek(a, -tail, SEEK_END); 3242 if (current_offset < 0) 3243 return 0; 3244 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 3245 return 0; 3246 /* Boyer-Moore search backwards from the end, since we want 3247 * to match the last EOCD in the file (there can be more than 3248 * one if there is an uncompressed Zip archive as a member 3249 * within this Zip archive). */ 3250 for (i = tail - 22; i > 0;) { 3251 switch (p[i]) { 3252 case 'P': 3253 if (memcmp(p + i, "PK\005\006", 4) == 0) { 3254 int ret = read_eocd(zip, p + i, 3255 current_offset + i); 3256 /* Zip64 EOCD locator precedes 3257 * regular EOCD if present. */ 3258 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 3259 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); 3260 if (ret_zip64 > ret) 3261 ret = ret_zip64; 3262 } 3263 return (ret); 3264 } 3265 i -= 4; 3266 break; 3267 case 'K': i -= 1; break; 3268 case 005: i -= 2; break; 3269 case 006: i -= 3; break; 3270 default: i -= 4; break; 3271 } 3272 } 3273 return 0; 3274 } 3275 3276 /* The red-black trees are only used in seeking mode to manage 3277 * the in-memory copy of the central directory. */ 3278 3279 static int 3280 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 3281 { 3282 const struct zip_entry *e1 = (const struct zip_entry *)n1; 3283 const struct zip_entry *e2 = (const struct zip_entry *)n2; 3284 3285 if (e1->local_header_offset > e2->local_header_offset) 3286 return -1; 3287 if (e1->local_header_offset < e2->local_header_offset) 3288 return 1; 3289 return 0; 3290 } 3291 3292 static int 3293 cmp_key(const struct archive_rb_node *n, const void *key) 3294 { 3295 /* This function won't be called */ 3296 (void)n; /* UNUSED */ 3297 (void)key; /* UNUSED */ 3298 return 1; 3299 } 3300 3301 static const struct archive_rb_tree_ops rb_ops = { 3302 &cmp_node, &cmp_key 3303 }; 3304 3305 static int 3306 rsrc_cmp_node(const struct archive_rb_node *n1, 3307 const struct archive_rb_node *n2) 3308 { 3309 const struct zip_entry *e1 = (const struct zip_entry *)n1; 3310 const struct zip_entry *e2 = (const struct zip_entry *)n2; 3311 3312 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 3313 } 3314 3315 static int 3316 rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 3317 { 3318 const struct zip_entry *e = (const struct zip_entry *)n; 3319 return (strcmp((const char *)key, e->rsrcname.s)); 3320 } 3321 3322 static const struct archive_rb_tree_ops rb_rsrc_ops = { 3323 &rsrc_cmp_node, &rsrc_cmp_key 3324 }; 3325 3326 static const char * 3327 rsrc_basename(const char *name, size_t name_length) 3328 { 3329 const char *s, *r; 3330 3331 r = s = name; 3332 for (;;) { 3333 s = memchr(s, '/', name_length - (s - name)); 3334 if (s == NULL) 3335 break; 3336 r = ++s; 3337 } 3338 return (r); 3339 } 3340 3341 static void 3342 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 3343 { 3344 struct archive_string str; 3345 struct zip_entry *dir; 3346 char *s; 3347 3348 archive_string_init(&str); 3349 archive_strncpy(&str, name, name_length); 3350 for (;;) { 3351 s = strrchr(str.s, '/'); 3352 if (s == NULL) 3353 break; 3354 *s = '\0'; 3355 /* Transfer the parent directory from zip->tree_rsrc RB 3356 * tree to zip->tree RB tree to expose. */ 3357 dir = (struct zip_entry *) 3358 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 3359 if (dir == NULL) 3360 break; 3361 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 3362 archive_string_free(&dir->rsrcname); 3363 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 3364 } 3365 archive_string_free(&str); 3366 } 3367 3368 static int 3369 slurp_central_directory(struct archive_read *a, struct zip *zip) 3370 { 3371 ssize_t i; 3372 unsigned found; 3373 int64_t correction; 3374 ssize_t bytes_avail; 3375 const char *p; 3376 3377 /* 3378 * Find the start of the central directory. The end-of-CD 3379 * record has our starting point, but there are lots of 3380 * Zip archives which have had other data prepended to the 3381 * file, which makes the recorded offsets all too small. 3382 * So we search forward from the specified offset until we 3383 * find the real start of the central directory. Then we 3384 * know the correction we need to apply to account for leading 3385 * padding. 3386 */ 3387 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 3388 return ARCHIVE_FATAL; 3389 3390 found = 0; 3391 while (!found) { 3392 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 3393 return ARCHIVE_FATAL; 3394 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 3395 switch (p[i + 3]) { 3396 case 'P': i += 3; break; 3397 case 'K': i += 2; break; 3398 case 001: i += 1; break; 3399 case 002: 3400 if (memcmp(p + i, "PK\001\002", 4) == 0) { 3401 p += i; 3402 found = 1; 3403 } else 3404 i += 4; 3405 break; 3406 case 005: i += 1; break; 3407 case 006: 3408 if (memcmp(p + i, "PK\005\006", 4) == 0) { 3409 p += i; 3410 found = 1; 3411 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 3412 p += i; 3413 found = 1; 3414 } else 3415 i += 1; 3416 break; 3417 default: i += 4; break; 3418 } 3419 } 3420 __archive_read_consume(a, i); 3421 } 3422 correction = archive_filter_bytes(&a->archive, 0) 3423 - zip->central_directory_offset; 3424 3425 __archive_rb_tree_init(&zip->tree, &rb_ops); 3426 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 3427 3428 zip->central_directory_entries_total = 0; 3429 while (1) { 3430 struct zip_entry *zip_entry; 3431 size_t filename_length, extra_length, comment_length; 3432 uint32_t external_attributes; 3433 const char *name, *r; 3434 3435 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 3436 return ARCHIVE_FATAL; 3437 if (memcmp(p, "PK\006\006", 4) == 0 3438 || memcmp(p, "PK\005\006", 4) == 0) { 3439 break; 3440 } else if (memcmp(p, "PK\001\002", 4) != 0) { 3441 archive_set_error(&a->archive, 3442 -1, "Invalid central directory signature"); 3443 return ARCHIVE_FATAL; 3444 } 3445 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 3446 return ARCHIVE_FATAL; 3447 3448 zip_entry = calloc(1, sizeof(struct zip_entry)); 3449 if (zip_entry == NULL) { 3450 archive_set_error(&a->archive, ENOMEM, 3451 "Can't allocate zip entry"); 3452 return ARCHIVE_FATAL; 3453 } 3454 zip_entry->next = zip->zip_entries; 3455 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 3456 zip->zip_entries = zip_entry; 3457 zip->central_directory_entries_total++; 3458 3459 /* version = p[4]; */ 3460 zip_entry->system = p[5]; 3461 /* version_required = archive_le16dec(p + 6); */ 3462 zip_entry->zip_flags = archive_le16dec(p + 8); 3463 if (zip_entry->zip_flags 3464 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 3465 zip->has_encrypted_entries = 1; 3466 } 3467 zip_entry->compression = (char)archive_le16dec(p + 10); 3468 zip_entry->mtime = zip_time(p + 12); 3469 zip_entry->crc32 = archive_le32dec(p + 16); 3470 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 3471 zip_entry->decdat = p[13]; 3472 else 3473 zip_entry->decdat = p[19]; 3474 zip_entry->compressed_size = archive_le32dec(p + 20); 3475 zip_entry->uncompressed_size = archive_le32dec(p + 24); 3476 filename_length = archive_le16dec(p + 28); 3477 extra_length = archive_le16dec(p + 30); 3478 comment_length = archive_le16dec(p + 32); 3479 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 3480 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 3481 external_attributes = archive_le32dec(p + 38); 3482 zip_entry->local_header_offset = 3483 archive_le32dec(p + 42) + correction; 3484 3485 /* If we can't guess the mode, leave it zero here; 3486 when we read the local file header we might get 3487 more information. */ 3488 if (zip_entry->system == 3) { 3489 zip_entry->mode = external_attributes >> 16; 3490 } else if (zip_entry->system == 0) { 3491 // Interpret MSDOS directory bit 3492 if (0x10 == (external_attributes & 0x10)) { 3493 zip_entry->mode = AE_IFDIR | 0775; 3494 } else { 3495 zip_entry->mode = AE_IFREG | 0664; 3496 } 3497 if (0x01 == (external_attributes & 0x01)) { 3498 // Read-only bit; strip write permissions 3499 zip_entry->mode &= 0555; 3500 } 3501 } else { 3502 zip_entry->mode = 0; 3503 } 3504 3505 /* We're done with the regular data; get the filename and 3506 * extra data. */ 3507 __archive_read_consume(a, 46); 3508 p = __archive_read_ahead(a, filename_length + extra_length, 3509 NULL); 3510 if (p == NULL) { 3511 archive_set_error(&a->archive, 3512 ARCHIVE_ERRNO_FILE_FORMAT, 3513 "Truncated ZIP file header"); 3514 return ARCHIVE_FATAL; 3515 } 3516 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 3517 return ARCHIVE_FATAL; 3518 } 3519 3520 /* 3521 * Mac resource fork files are stored under the 3522 * "__MACOSX/" directory, so we should check if 3523 * it is. 3524 */ 3525 if (!zip->process_mac_extensions) { 3526 /* Treat every entry as a regular entry. */ 3527 __archive_rb_tree_insert_node(&zip->tree, 3528 &zip_entry->node); 3529 } else { 3530 name = p; 3531 r = rsrc_basename(name, filename_length); 3532 if (filename_length >= 9 && 3533 strncmp("__MACOSX/", name, 9) == 0) { 3534 /* If this file is not a resource fork nor 3535 * a directory. We should treat it as a non 3536 * resource fork file to expose it. */ 3537 if (name[filename_length-1] != '/' && 3538 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 3539 __archive_rb_tree_insert_node( 3540 &zip->tree, &zip_entry->node); 3541 /* Expose its parent directories. */ 3542 expose_parent_dirs(zip, name, 3543 filename_length); 3544 } else { 3545 /* This file is a resource fork file or 3546 * a directory. */ 3547 archive_strncpy(&(zip_entry->rsrcname), 3548 name, filename_length); 3549 __archive_rb_tree_insert_node( 3550 &zip->tree_rsrc, &zip_entry->node); 3551 } 3552 } else { 3553 /* Generate resource fork name to find its 3554 * resource file at zip->tree_rsrc. */ 3555 archive_strcpy(&(zip_entry->rsrcname), 3556 "__MACOSX/"); 3557 archive_strncat(&(zip_entry->rsrcname), 3558 name, r - name); 3559 archive_strcat(&(zip_entry->rsrcname), "._"); 3560 archive_strncat(&(zip_entry->rsrcname), 3561 name + (r - name), 3562 filename_length - (r - name)); 3563 /* Register an entry to RB tree to sort it by 3564 * file offset. */ 3565 __archive_rb_tree_insert_node(&zip->tree, 3566 &zip_entry->node); 3567 } 3568 } 3569 3570 /* Skip the comment too ... */ 3571 __archive_read_consume(a, 3572 filename_length + extra_length + comment_length); 3573 } 3574 3575 return ARCHIVE_OK; 3576 } 3577 3578 static ssize_t 3579 zip_get_local_file_header_size(struct archive_read *a, size_t extra) 3580 { 3581 const char *p; 3582 ssize_t filename_length, extra_length; 3583 3584 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 3585 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3586 "Truncated ZIP file header"); 3587 return (ARCHIVE_WARN); 3588 } 3589 p += extra; 3590 3591 if (memcmp(p, "PK\003\004", 4) != 0) { 3592 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 3593 return ARCHIVE_WARN; 3594 } 3595 filename_length = archive_le16dec(p + 26); 3596 extra_length = archive_le16dec(p + 28); 3597 3598 return (30 + filename_length + extra_length); 3599 } 3600 3601 static int 3602 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 3603 struct zip_entry *rsrc) 3604 { 3605 struct zip *zip = (struct zip *)a->format->data; 3606 unsigned char *metadata, *mp; 3607 int64_t offset = archive_filter_bytes(&a->archive, 0); 3608 size_t remaining_bytes, metadata_bytes; 3609 ssize_t hsize; 3610 int ret = ARCHIVE_OK, eof; 3611 3612 switch(rsrc->compression) { 3613 case 0: /* No compression. */ 3614 if (rsrc->uncompressed_size != rsrc->compressed_size) { 3615 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3616 "Malformed OS X metadata entry: inconsistent size"); 3617 return (ARCHIVE_FATAL); 3618 } 3619 #ifdef HAVE_ZLIB_H 3620 case 8: /* Deflate compression. */ 3621 #endif 3622 break; 3623 default: /* Unsupported compression. */ 3624 /* Return a warning. */ 3625 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3626 "Unsupported ZIP compression method (%s)", 3627 compression_name(rsrc->compression)); 3628 /* We can't decompress this entry, but we will 3629 * be able to skip() it and try the next entry. */ 3630 return (ARCHIVE_WARN); 3631 } 3632 3633 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 3634 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3635 "Mac metadata is too large: %jd > 4M bytes", 3636 (intmax_t)rsrc->uncompressed_size); 3637 return (ARCHIVE_WARN); 3638 } 3639 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 3640 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 3641 "Mac metadata is too large: %jd > 4M bytes", 3642 (intmax_t)rsrc->compressed_size); 3643 return (ARCHIVE_WARN); 3644 } 3645 3646 metadata = malloc((size_t)rsrc->uncompressed_size); 3647 if (metadata == NULL) { 3648 archive_set_error(&a->archive, ENOMEM, 3649 "Can't allocate memory for Mac metadata"); 3650 return (ARCHIVE_FATAL); 3651 } 3652 3653 if (offset < rsrc->local_header_offset) 3654 __archive_read_consume(a, rsrc->local_header_offset - offset); 3655 else if (offset != rsrc->local_header_offset) { 3656 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 3657 } 3658 3659 hsize = zip_get_local_file_header_size(a, 0); 3660 __archive_read_consume(a, hsize); 3661 3662 remaining_bytes = (size_t)rsrc->compressed_size; 3663 metadata_bytes = (size_t)rsrc->uncompressed_size; 3664 mp = metadata; 3665 eof = 0; 3666 while (!eof && remaining_bytes) { 3667 const unsigned char *p; 3668 ssize_t bytes_avail; 3669 size_t bytes_used; 3670 3671 p = __archive_read_ahead(a, 1, &bytes_avail); 3672 if (p == NULL) { 3673 archive_set_error(&a->archive, 3674 ARCHIVE_ERRNO_FILE_FORMAT, 3675 "Truncated ZIP file header"); 3676 ret = ARCHIVE_WARN; 3677 goto exit_mac_metadata; 3678 } 3679 if ((size_t)bytes_avail > remaining_bytes) 3680 bytes_avail = remaining_bytes; 3681 switch(rsrc->compression) { 3682 case 0: /* No compression. */ 3683 if ((size_t)bytes_avail > metadata_bytes) 3684 bytes_avail = metadata_bytes; 3685 memcpy(mp, p, bytes_avail); 3686 bytes_used = (size_t)bytes_avail; 3687 metadata_bytes -= bytes_used; 3688 mp += bytes_used; 3689 if (metadata_bytes == 0) 3690 eof = 1; 3691 break; 3692 #ifdef HAVE_ZLIB_H 3693 case 8: /* Deflate compression. */ 3694 { 3695 int r; 3696 3697 ret = zip_deflate_init(a, zip); 3698 if (ret != ARCHIVE_OK) 3699 goto exit_mac_metadata; 3700 zip->stream.next_in = 3701 (Bytef *)(uintptr_t)(const void *)p; 3702 zip->stream.avail_in = (uInt)bytes_avail; 3703 zip->stream.total_in = 0; 3704 zip->stream.next_out = mp; 3705 zip->stream.avail_out = (uInt)metadata_bytes; 3706 zip->stream.total_out = 0; 3707 3708 r = inflate(&zip->stream, 0); 3709 switch (r) { 3710 case Z_OK: 3711 break; 3712 case Z_STREAM_END: 3713 eof = 1; 3714 break; 3715 case Z_MEM_ERROR: 3716 archive_set_error(&a->archive, ENOMEM, 3717 "Out of memory for ZIP decompression"); 3718 ret = ARCHIVE_FATAL; 3719 goto exit_mac_metadata; 3720 default: 3721 archive_set_error(&a->archive, 3722 ARCHIVE_ERRNO_MISC, 3723 "ZIP decompression failed (%d)", r); 3724 ret = ARCHIVE_FATAL; 3725 goto exit_mac_metadata; 3726 } 3727 bytes_used = zip->stream.total_in; 3728 metadata_bytes -= zip->stream.total_out; 3729 mp += zip->stream.total_out; 3730 break; 3731 } 3732 #endif 3733 default: 3734 bytes_used = 0; 3735 break; 3736 } 3737 __archive_read_consume(a, bytes_used); 3738 remaining_bytes -= bytes_used; 3739 } 3740 archive_entry_copy_mac_metadata(entry, metadata, 3741 (size_t)rsrc->uncompressed_size - metadata_bytes); 3742 3743 exit_mac_metadata: 3744 __archive_read_seek(a, offset, SEEK_SET); 3745 zip->decompress_init = 0; 3746 free(metadata); 3747 return (ret); 3748 } 3749 3750 static int 3751 archive_read_format_zip_seekable_read_header(struct archive_read *a, 3752 struct archive_entry *entry) 3753 { 3754 struct zip *zip = (struct zip *)a->format->data; 3755 struct zip_entry *rsrc; 3756 int64_t offset; 3757 int r, ret = ARCHIVE_OK; 3758 3759 /* 3760 * It should be sufficient to call archive_read_next_header() for 3761 * a reader to determine if an entry is encrypted or not. If the 3762 * encryption of an entry is only detectable when calling 3763 * archive_read_data(), so be it. We'll do the same check there 3764 * as well. 3765 */ 3766 if (zip->has_encrypted_entries == 3767 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 3768 zip->has_encrypted_entries = 0; 3769 3770 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 3771 if (a->archive.archive_format_name == NULL) 3772 a->archive.archive_format_name = "ZIP"; 3773 3774 if (zip->zip_entries == NULL) { 3775 r = slurp_central_directory(a, zip); 3776 if (r != ARCHIVE_OK) 3777 return r; 3778 /* Get first entry whose local header offset is lower than 3779 * other entries in the archive file. */ 3780 zip->entry = 3781 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 3782 } else if (zip->entry != NULL) { 3783 /* Get next entry in local header offset order. */ 3784 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 3785 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 3786 } 3787 3788 if (zip->entry == NULL) 3789 return ARCHIVE_EOF; 3790 3791 if (zip->entry->rsrcname.s) 3792 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 3793 &zip->tree_rsrc, zip->entry->rsrcname.s); 3794 else 3795 rsrc = NULL; 3796 3797 if (zip->cctx_valid) 3798 archive_decrypto_aes_ctr_release(&zip->cctx); 3799 if (zip->hctx_valid) 3800 archive_hmac_sha1_cleanup(&zip->hctx); 3801 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3802 __archive_read_reset_passphrase(a); 3803 3804 /* File entries are sorted by the header offset, we should mostly 3805 * use __archive_read_consume to advance a read point to avoid redundant 3806 * data reading. */ 3807 offset = archive_filter_bytes(&a->archive, 0); 3808 if (offset < zip->entry->local_header_offset) 3809 __archive_read_consume(a, 3810 zip->entry->local_header_offset - offset); 3811 else if (offset != zip->entry->local_header_offset) { 3812 __archive_read_seek(a, zip->entry->local_header_offset, 3813 SEEK_SET); 3814 } 3815 zip->unconsumed = 0; 3816 r = zip_read_local_file_header(a, entry, zip); 3817 if (r != ARCHIVE_OK) 3818 return r; 3819 if (rsrc) { 3820 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3821 if (ret2 < ret) 3822 ret = ret2; 3823 } 3824 return (ret); 3825 } 3826 3827 /* 3828 * We're going to seek for the next header anyway, so we don't 3829 * need to bother doing anything here. 3830 */ 3831 static int 3832 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3833 { 3834 struct zip *zip; 3835 zip = (struct zip *)(a->format->data); 3836 3837 zip->unconsumed = 0; 3838 return (ARCHIVE_OK); 3839 } 3840 3841 int 3842 archive_read_support_format_zip_seekable(struct archive *_a) 3843 { 3844 struct archive_read *a = (struct archive_read *)_a; 3845 struct zip *zip; 3846 int r; 3847 3848 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3849 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3850 3851 zip = (struct zip *)calloc(1, sizeof(*zip)); 3852 if (zip == NULL) { 3853 archive_set_error(&a->archive, ENOMEM, 3854 "Can't allocate zip data"); 3855 return (ARCHIVE_FATAL); 3856 } 3857 3858 #ifdef HAVE_COPYFILE_H 3859 /* Set this by default on Mac OS. */ 3860 zip->process_mac_extensions = 1; 3861 #endif 3862 3863 /* 3864 * Until enough data has been read, we cannot tell about 3865 * any encrypted entries yet. 3866 */ 3867 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3868 zip->crc32func = real_crc32; 3869 3870 r = __archive_read_register_format(a, 3871 zip, 3872 "zip", 3873 archive_read_format_zip_seekable_bid, 3874 archive_read_format_zip_options, 3875 archive_read_format_zip_seekable_read_header, 3876 archive_read_format_zip_read_data, 3877 archive_read_format_zip_read_data_skip_seekable, 3878 NULL, 3879 archive_read_format_zip_cleanup, 3880 archive_read_support_format_zip_capabilities_seekable, 3881 archive_read_format_zip_has_encrypted_entries); 3882 3883 if (r != ARCHIVE_OK) 3884 free(zip); 3885 return (ARCHIVE_OK); 3886 } 3887 3888 /*# vim:set noet:*/ 3889