archive_read_support_format_zip.c revision 337351
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: stable/11/contrib/libarchive/libarchive/archive_read_support_format_zip.c 337351 2018-08-05 14:35:30Z mm $"); 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 56#include "archive.h" 57#include "archive_digest_private.h" 58#include "archive_cryptor_private.h" 59#include "archive_endian.h" 60#include "archive_entry.h" 61#include "archive_entry_locale.h" 62#include "archive_hmac_private.h" 63#include "archive_private.h" 64#include "archive_rb.h" 65#include "archive_read_private.h" 66 67#ifndef HAVE_ZLIB_H 68#include "archive_crc32.h" 69#endif 70 71struct zip_entry { 72 struct archive_rb_node node; 73 struct zip_entry *next; 74 int64_t local_header_offset; 75 int64_t compressed_size; 76 int64_t uncompressed_size; 77 int64_t gid; 78 int64_t uid; 79 struct archive_string rsrcname; 80 time_t mtime; 81 time_t atime; 82 time_t ctime; 83 uint32_t crc32; 84 uint16_t mode; 85 uint16_t zip_flags; /* From GP Flags Field */ 86 unsigned char compression; 87 unsigned char system; /* From "version written by" */ 88 unsigned char flags; /* Our extra markers. */ 89 unsigned char decdat;/* Used for Decryption check */ 90 91 /* WinZip AES encryption extra field should be available 92 * when compression is 99. */ 93 struct { 94 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 95 unsigned vendor; 96#define AES_VENDOR_AE_1 0x0001 97#define AES_VENDOR_AE_2 0x0002 98 /* AES encryption strength: 99 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 100 unsigned strength; 101 /* Actual compression method. */ 102 unsigned char compression; 103 } aes_extra; 104}; 105 106struct trad_enc_ctx { 107 uint32_t keys[3]; 108}; 109 110/* Bits used in zip_flags. */ 111#define ZIP_ENCRYPTED (1 << 0) 112#define ZIP_LENGTH_AT_END (1 << 3) 113#define ZIP_STRONG_ENCRYPTED (1 << 6) 114#define ZIP_UTF8_NAME (1 << 11) 115/* See "7.2 Single Password Symmetric Encryption Method" 116 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 117#define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 118 119/* Bits used in flags. */ 120#define LA_USED_ZIP64 (1 << 0) 121#define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 122 123/* 124 * See "WinZip - AES Encryption Information" 125 * http://www.winzip.com/aes_info.htm 126 */ 127/* Value used in compression method. */ 128#define WINZIP_AES_ENCRYPTION 99 129/* Authentication code size. */ 130#define AUTH_CODE_SIZE 10 131/**/ 132#define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 133 134struct zip { 135 /* Structural information about the archive. */ 136 struct archive_string format_name; 137 int64_t central_directory_offset; 138 size_t central_directory_entries_total; 139 size_t central_directory_entries_on_this_disk; 140 int has_encrypted_entries; 141 142 /* List of entries (seekable Zip only) */ 143 struct zip_entry *zip_entries; 144 struct archive_rb_tree tree; 145 struct archive_rb_tree tree_rsrc; 146 147 /* Bytes read but not yet consumed via __archive_read_consume() */ 148 size_t unconsumed; 149 150 /* Information about entry we're currently reading. */ 151 struct zip_entry *entry; 152 int64_t entry_bytes_remaining; 153 154 /* These count the number of bytes actually read for the entry. */ 155 int64_t entry_compressed_bytes_read; 156 int64_t entry_uncompressed_bytes_read; 157 158 /* Running CRC32 of the decompressed data */ 159 unsigned long entry_crc32; 160 unsigned long (*crc32func)(unsigned long, const void *, 161 size_t); 162 char ignore_crc32; 163 164 /* Flags to mark progress of decompression. */ 165 char decompress_init; 166 char end_of_entry; 167 168#ifdef HAVE_ZLIB_H 169 unsigned char *uncompressed_buffer; 170 size_t uncompressed_buffer_size; 171 z_stream stream; 172 char stream_valid; 173#endif 174 175 struct archive_string_conv *sconv; 176 struct archive_string_conv *sconv_default; 177 struct archive_string_conv *sconv_utf8; 178 int init_default_conversion; 179 int process_mac_extensions; 180 181 char init_decryption; 182 183 /* Decryption buffer. */ 184 /* 185 * The decrypted data starts at decrypted_ptr and 186 * extends for decrypted_bytes_remaining. Decryption 187 * adds new data to the end of this block, data is returned 188 * to clients from the beginning. When the block hits the 189 * end of decrypted_buffer, it has to be shuffled back to 190 * the beginning of the buffer. 191 */ 192 unsigned char *decrypted_buffer; 193 unsigned char *decrypted_ptr; 194 size_t decrypted_buffer_size; 195 size_t decrypted_bytes_remaining; 196 size_t decrypted_unconsumed_bytes; 197 198 /* Traditional PKWARE decryption. */ 199 struct trad_enc_ctx tctx; 200 char tctx_valid; 201 202 /* WinZip AES decryption. */ 203 /* Contexts used for AES decryption. */ 204 archive_crypto_ctx cctx; 205 char cctx_valid; 206 archive_hmac_sha1_ctx hctx; 207 char hctx_valid; 208 209 /* Strong encryption's decryption header information. */ 210 unsigned iv_size; 211 unsigned alg_id; 212 unsigned bit_len; 213 unsigned flags; 214 unsigned erd_size; 215 unsigned v_size; 216 unsigned v_crc32; 217 uint8_t *iv; 218 uint8_t *erd; 219 uint8_t *v_data; 220}; 221 222/* Many systems define min or MIN, but not all. */ 223#define zipmin(a,b) ((a) < (b) ? (a) : (b)) 224 225/* ------------------------------------------------------------------------ */ 226 227/* 228 Traditional PKWARE Decryption functions. 229 */ 230 231static void 232trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 233{ 234 uint8_t t; 235#define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 236 237 ctx->keys[0] = CRC32(ctx->keys[0], c); 238 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 239 t = (ctx->keys[1] >> 24) & 0xff; 240 ctx->keys[2] = CRC32(ctx->keys[2], t); 241#undef CRC32 242} 243 244static uint8_t 245trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) 246{ 247 unsigned temp = ctx->keys[2] | 2; 248 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 249} 250 251static void 252trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 253 size_t in_len, uint8_t *out, size_t out_len) 254{ 255 unsigned i, max; 256 257 max = (unsigned)((in_len < out_len)? in_len: out_len); 258 259 for (i = 0; i < max; i++) { 260 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); 261 out[i] = t; 262 trad_enc_update_keys(ctx, t); 263 } 264} 265 266static int 267trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 268 const uint8_t *key, size_t key_len, uint8_t *crcchk) 269{ 270 uint8_t header[12]; 271 272 if (key_len < 12) { 273 *crcchk = 0xff; 274 return -1; 275 } 276 277 ctx->keys[0] = 305419896L; 278 ctx->keys[1] = 591751049L; 279 ctx->keys[2] = 878082192L; 280 281 for (;pw_len; --pw_len) 282 trad_enc_update_keys(ctx, *pw++); 283 284 trad_enc_decrypt_update(ctx, key, 12, header, 12); 285 /* Return the last byte for CRC check. */ 286 *crcchk = header[11]; 287 return 0; 288} 289 290#if 0 291static void 292crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 293 int key_size) 294{ 295#define MD_SIZE 20 296 archive_sha1_ctx ctx; 297 unsigned char md1[MD_SIZE]; 298 unsigned char md2[MD_SIZE * 2]; 299 unsigned char mkb[64]; 300 int i; 301 302 archive_sha1_init(&ctx); 303 archive_sha1_update(&ctx, p, size); 304 archive_sha1_final(&ctx, md1); 305 306 memset(mkb, 0x36, sizeof(mkb)); 307 for (i = 0; i < MD_SIZE; i++) 308 mkb[i] ^= md1[i]; 309 archive_sha1_init(&ctx); 310 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 311 archive_sha1_final(&ctx, md2); 312 313 memset(mkb, 0x5C, sizeof(mkb)); 314 for (i = 0; i < MD_SIZE; i++) 315 mkb[i] ^= md1[i]; 316 archive_sha1_init(&ctx); 317 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 318 archive_sha1_final(&ctx, md2 + MD_SIZE); 319 320 if (key_size > 32) 321 key_size = 32; 322 memcpy(key, md2, key_size); 323#undef MD_SIZE 324} 325#endif 326 327/* 328 * Common code for streaming or seeking modes. 329 * 330 * Includes code to read local file headers, decompress data 331 * from entry bodies, and common API. 332 */ 333 334static unsigned long 335real_crc32(unsigned long crc, const void *buff, size_t len) 336{ 337 return crc32(crc, buff, (unsigned int)len); 338} 339 340/* Used by "ignorecrc32" option to speed up tests. */ 341static unsigned long 342fake_crc32(unsigned long crc, const void *buff, size_t len) 343{ 344 (void)crc; /* UNUSED */ 345 (void)buff; /* UNUSED */ 346 (void)len; /* UNUSED */ 347 return 0; 348} 349 350static const struct { 351 int id; 352 const char * name; 353} compression_methods[] = { 354 {0, "uncompressed"}, /* The file is stored (no compression) */ 355 {1, "shrinking"}, /* The file is Shrunk */ 356 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 357 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 358 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 359 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 360 {6, "imploded"}, /* The file is Imploded */ 361 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 362 {8, "deflation"}, /* The file is Deflated */ 363 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 364 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 365 * (old IBM TERSE) */ 366 {11, "reserved"}, /* Reserved by PKWARE */ 367 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 368 {13, "reserved"}, /* Reserved by PKWARE */ 369 {14, "lzma"}, /* LZMA (EFS) */ 370 {15, "reserved"}, /* Reserved by PKWARE */ 371 {16, "reserved"}, /* Reserved by PKWARE */ 372 {17, "reserved"}, /* Reserved by PKWARE */ 373 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 374 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 375 {97, "wav-pack"}, /* WavPack compressed data */ 376 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 377 {99, "aes"} /* WinZip AES encryption */ 378}; 379 380static const char * 381compression_name(const int compression) 382{ 383 static const int num_compression_methods = 384 sizeof(compression_methods)/sizeof(compression_methods[0]); 385 int i=0; 386 387 while(compression >= 0 && i < num_compression_methods) { 388 if (compression_methods[i].id == compression) 389 return compression_methods[i].name; 390 i++; 391 } 392 return "??"; 393} 394 395/* Convert an MSDOS-style date/time into Unix-style time. */ 396static time_t 397zip_time(const char *p) 398{ 399 int msTime, msDate; 400 struct tm ts; 401 402 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 403 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 404 405 memset(&ts, 0, sizeof(ts)); 406 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 407 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 408 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 409 ts.tm_hour = (msTime >> 11) & 0x1f; 410 ts.tm_min = (msTime >> 5) & 0x3f; 411 ts.tm_sec = (msTime << 1) & 0x3e; 412 ts.tm_isdst = -1; 413 return mktime(&ts); 414} 415 416/* 417 * The extra data is stored as a list of 418 * id1+size1+data1 + id2+size2+data2 ... 419 * triplets. id and size are 2 bytes each. 420 */ 421static int 422process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry) 423{ 424 unsigned offset = 0; 425 426 if (extra_length == 0) { 427 return ARCHIVE_OK; 428 } 429 430 if (extra_length < 4) { 431 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 432 "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length); 433 return ARCHIVE_FAILED; 434 } 435 while (offset <= extra_length - 4) { 436 unsigned short headerid = archive_le16dec(p + offset); 437 unsigned short datasize = archive_le16dec(p + offset + 2); 438 439 offset += 4; 440 if (offset + datasize > extra_length) { 441 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 442 "Extra data overflow: Need %d bytes but only found %d bytes", 443 (int)datasize, (int)(extra_length - offset)); 444 return ARCHIVE_FAILED; 445 } 446#ifdef DEBUG 447 fprintf(stderr, "Header id 0x%04x, length %d\n", 448 headerid, datasize); 449#endif 450 switch (headerid) { 451 case 0x0001: 452 /* Zip64 extended information extra field. */ 453 zip_entry->flags |= LA_USED_ZIP64; 454 if (zip_entry->uncompressed_size == 0xffffffff) { 455 uint64_t t = 0; 456 if (datasize < 8 457 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 458 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 459 "Malformed 64-bit uncompressed size"); 460 return ARCHIVE_FAILED; 461 } 462 zip_entry->uncompressed_size = t; 463 offset += 8; 464 datasize -= 8; 465 } 466 if (zip_entry->compressed_size == 0xffffffff) { 467 uint64_t t = 0; 468 if (datasize < 8 469 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 470 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 471 "Malformed 64-bit compressed size"); 472 return ARCHIVE_FAILED; 473 } 474 zip_entry->compressed_size = t; 475 offset += 8; 476 datasize -= 8; 477 } 478 if (zip_entry->local_header_offset == 0xffffffff) { 479 uint64_t t = 0; 480 if (datasize < 8 481 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 482 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 483 "Malformed 64-bit local header offset"); 484 return ARCHIVE_FAILED; 485 } 486 zip_entry->local_header_offset = t; 487 offset += 8; 488 datasize -= 8; 489 } 490 /* archive_le32dec(p + offset) gives disk 491 * on which file starts, but we don't handle 492 * multi-volume Zip files. */ 493 break; 494#ifdef DEBUG 495 case 0x0017: 496 { 497 /* Strong encryption field. */ 498 if (archive_le16dec(p + offset) == 2) { 499 unsigned algId = 500 archive_le16dec(p + offset + 2); 501 unsigned bitLen = 502 archive_le16dec(p + offset + 4); 503 int flags = 504 archive_le16dec(p + offset + 6); 505 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 506 "flgas=%d\n", algId, bitLen,flags); 507 } 508 break; 509 } 510#endif 511 case 0x5455: 512 { 513 /* Extended time field "UT". */ 514 int flags; 515 if (datasize == 0) { 516 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 517 "Incomplete extended time field"); 518 return ARCHIVE_FAILED; 519 } 520 flags = p[offset]; 521 offset++; 522 datasize--; 523 /* Flag bits indicate which dates are present. */ 524 if (flags & 0x01) 525 { 526#ifdef DEBUG 527 fprintf(stderr, "mtime: %lld -> %d\n", 528 (long long)zip_entry->mtime, 529 archive_le32dec(p + offset)); 530#endif 531 if (datasize < 4) 532 break; 533 zip_entry->mtime = archive_le32dec(p + offset); 534 offset += 4; 535 datasize -= 4; 536 } 537 if (flags & 0x02) 538 { 539 if (datasize < 4) 540 break; 541 zip_entry->atime = archive_le32dec(p + offset); 542 offset += 4; 543 datasize -= 4; 544 } 545 if (flags & 0x04) 546 { 547 if (datasize < 4) 548 break; 549 zip_entry->ctime = archive_le32dec(p + offset); 550 offset += 4; 551 datasize -= 4; 552 } 553 break; 554 } 555 case 0x5855: 556 { 557 /* Info-ZIP Unix Extra Field (old version) "UX". */ 558 if (datasize >= 8) { 559 zip_entry->atime = archive_le32dec(p + offset); 560 zip_entry->mtime = 561 archive_le32dec(p + offset + 4); 562 } 563 if (datasize >= 12) { 564 zip_entry->uid = 565 archive_le16dec(p + offset + 8); 566 zip_entry->gid = 567 archive_le16dec(p + offset + 10); 568 } 569 break; 570 } 571 case 0x6c78: 572 { 573 /* Experimental 'xl' field */ 574 /* 575 * Introduced Dec 2013 to provide a way to 576 * include external file attributes (and other 577 * fields that ordinarily appear only in 578 * central directory) in local file header. 579 * This provides file type and permission 580 * information necessary to support full 581 * streaming extraction. Currently being 582 * discussed with other Zip developers 583 * ... subject to change. 584 * 585 * Format: 586 * The field starts with a bitmap that specifies 587 * which additional fields are included. The 588 * bitmap is variable length and can be extended in 589 * the future. 590 * 591 * n bytes - feature bitmap: first byte has low-order 592 * 7 bits. If high-order bit is set, a subsequent 593 * byte holds the next 7 bits, etc. 594 * 595 * if bitmap & 1, 2 byte "version made by" 596 * if bitmap & 2, 2 byte "internal file attributes" 597 * if bitmap & 4, 4 byte "external file attributes" 598 * if bitmap & 8, 2 byte comment length + n byte comment 599 */ 600 int bitmap, bitmap_last; 601 602 if (datasize < 1) 603 break; 604 bitmap_last = bitmap = 0xff & p[offset]; 605 offset += 1; 606 datasize -= 1; 607 608 /* We only support first 7 bits of bitmap; skip rest. */ 609 while ((bitmap_last & 0x80) != 0 610 && datasize >= 1) { 611 bitmap_last = p[offset]; 612 offset += 1; 613 datasize -= 1; 614 } 615 616 if (bitmap & 1) { 617 /* 2 byte "version made by" */ 618 if (datasize < 2) 619 break; 620 zip_entry->system 621 = archive_le16dec(p + offset) >> 8; 622 offset += 2; 623 datasize -= 2; 624 } 625 if (bitmap & 2) { 626 /* 2 byte "internal file attributes" */ 627 uint32_t internal_attributes; 628 if (datasize < 2) 629 break; 630 internal_attributes 631 = archive_le16dec(p + offset); 632 /* Not used by libarchive at present. */ 633 (void)internal_attributes; /* UNUSED */ 634 offset += 2; 635 datasize -= 2; 636 } 637 if (bitmap & 4) { 638 /* 4 byte "external file attributes" */ 639 uint32_t external_attributes; 640 if (datasize < 4) 641 break; 642 external_attributes 643 = archive_le32dec(p + offset); 644 if (zip_entry->system == 3) { 645 zip_entry->mode 646 = external_attributes >> 16; 647 } else if (zip_entry->system == 0) { 648 // Interpret MSDOS directory bit 649 if (0x10 == (external_attributes & 0x10)) { 650 zip_entry->mode = AE_IFDIR | 0775; 651 } else { 652 zip_entry->mode = AE_IFREG | 0664; 653 } 654 if (0x01 == (external_attributes & 0x01)) { 655 // Read-only bit; strip write permissions 656 zip_entry->mode &= 0555; 657 } 658 } else { 659 zip_entry->mode = 0; 660 } 661 offset += 4; 662 datasize -= 4; 663 } 664 if (bitmap & 8) { 665 /* 2 byte comment length + comment */ 666 uint32_t comment_length; 667 if (datasize < 2) 668 break; 669 comment_length 670 = archive_le16dec(p + offset); 671 offset += 2; 672 datasize -= 2; 673 674 if (datasize < comment_length) 675 break; 676 /* Comment is not supported by libarchive */ 677 offset += comment_length; 678 datasize -= comment_length; 679 } 680 break; 681 } 682 case 0x7855: 683 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 684#ifdef DEBUG 685 fprintf(stderr, "uid %d gid %d\n", 686 archive_le16dec(p + offset), 687 archive_le16dec(p + offset + 2)); 688#endif 689 if (datasize >= 2) 690 zip_entry->uid = archive_le16dec(p + offset); 691 if (datasize >= 4) 692 zip_entry->gid = 693 archive_le16dec(p + offset + 2); 694 break; 695 case 0x7875: 696 { 697 /* Info-Zip Unix Extra Field (type 3) "ux". */ 698 int uidsize = 0, gidsize = 0; 699 700 /* TODO: support arbitrary uidsize/gidsize. */ 701 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 702 if (datasize >= 4) { 703 /* get a uid size. */ 704 uidsize = 0xff & (int)p[offset+1]; 705 if (uidsize == 2) 706 zip_entry->uid = 707 archive_le16dec( 708 p + offset + 2); 709 else if (uidsize == 4 && datasize >= 6) 710 zip_entry->uid = 711 archive_le32dec( 712 p + offset + 2); 713 } 714 if (datasize >= (2 + uidsize + 3)) { 715 /* get a gid size. */ 716 gidsize = 0xff & (int)p[offset+2+uidsize]; 717 if (gidsize == 2) 718 zip_entry->gid = 719 archive_le16dec( 720 p+offset+2+uidsize+1); 721 else if (gidsize == 4 && 722 datasize >= (2 + uidsize + 5)) 723 zip_entry->gid = 724 archive_le32dec( 725 p+offset+2+uidsize+1); 726 } 727 } 728 break; 729 } 730 case 0x9901: 731 /* WinZip AES extra data field. */ 732 if (datasize < 6) { 733 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 734 "Incomplete AES field"); 735 return ARCHIVE_FAILED; 736 } 737 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 738 /* Vendor version. */ 739 zip_entry->aes_extra.vendor = 740 archive_le16dec(p + offset); 741 /* AES encryption strength. */ 742 zip_entry->aes_extra.strength = p[offset + 4]; 743 /* Actual compression method. */ 744 zip_entry->aes_extra.compression = 745 p[offset + 5]; 746 } 747 break; 748 default: 749 break; 750 } 751 offset += datasize; 752 } 753 if (offset != extra_length) { 754 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 755 "Malformed extra data: Consumed %d bytes of %d bytes", 756 (int)offset, (int)extra_length); 757 return ARCHIVE_FAILED; 758 } 759 return ARCHIVE_OK; 760} 761 762/* 763 * Assumes file pointer is at beginning of local file header. 764 */ 765static int 766zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 767 struct zip *zip) 768{ 769 const char *p; 770 const void *h; 771 const wchar_t *wp; 772 const char *cp; 773 size_t len, filename_length, extra_length; 774 struct archive_string_conv *sconv; 775 struct zip_entry *zip_entry = zip->entry; 776 struct zip_entry zip_entry_central_dir; 777 int ret = ARCHIVE_OK; 778 char version; 779 780 /* Save a copy of the original for consistency checks. */ 781 zip_entry_central_dir = *zip_entry; 782 783 zip->decompress_init = 0; 784 zip->end_of_entry = 0; 785 zip->entry_uncompressed_bytes_read = 0; 786 zip->entry_compressed_bytes_read = 0; 787 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 788 789 /* Setup default conversion. */ 790 if (zip->sconv == NULL && !zip->init_default_conversion) { 791 zip->sconv_default = 792 archive_string_default_conversion_for_read(&(a->archive)); 793 zip->init_default_conversion = 1; 794 } 795 796 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 797 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 798 "Truncated ZIP file header"); 799 return (ARCHIVE_FATAL); 800 } 801 802 if (memcmp(p, "PK\003\004", 4) != 0) { 803 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 804 return ARCHIVE_FATAL; 805 } 806 version = p[4]; 807 zip_entry->system = p[5]; 808 zip_entry->zip_flags = archive_le16dec(p + 6); 809 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 810 zip->has_encrypted_entries = 1; 811 archive_entry_set_is_data_encrypted(entry, 1); 812 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 813 zip_entry->zip_flags & ZIP_ENCRYPTED && 814 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 815 archive_entry_set_is_metadata_encrypted(entry, 1); 816 return ARCHIVE_FATAL; 817 } 818 } 819 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 820 zip_entry->compression = (char)archive_le16dec(p + 8); 821 zip_entry->mtime = zip_time(p + 10); 822 zip_entry->crc32 = archive_le32dec(p + 14); 823 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 824 zip_entry->decdat = p[11]; 825 else 826 zip_entry->decdat = p[17]; 827 zip_entry->compressed_size = archive_le32dec(p + 18); 828 zip_entry->uncompressed_size = archive_le32dec(p + 22); 829 filename_length = archive_le16dec(p + 26); 830 extra_length = archive_le16dec(p + 28); 831 832 __archive_read_consume(a, 30); 833 834 /* Read the filename. */ 835 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 836 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 837 "Truncated ZIP file header"); 838 return (ARCHIVE_FATAL); 839 } 840 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 841 /* The filename is stored to be UTF-8. */ 842 if (zip->sconv_utf8 == NULL) { 843 zip->sconv_utf8 = 844 archive_string_conversion_from_charset( 845 &a->archive, "UTF-8", 1); 846 if (zip->sconv_utf8 == NULL) 847 return (ARCHIVE_FATAL); 848 } 849 sconv = zip->sconv_utf8; 850 } else if (zip->sconv != NULL) 851 sconv = zip->sconv; 852 else 853 sconv = zip->sconv_default; 854 855 if (archive_entry_copy_pathname_l(entry, 856 h, filename_length, sconv) != 0) { 857 if (errno == ENOMEM) { 858 archive_set_error(&a->archive, ENOMEM, 859 "Can't allocate memory for Pathname"); 860 return (ARCHIVE_FATAL); 861 } 862 archive_set_error(&a->archive, 863 ARCHIVE_ERRNO_FILE_FORMAT, 864 "Pathname cannot be converted " 865 "from %s to current locale.", 866 archive_string_conversion_charset_name(sconv)); 867 ret = ARCHIVE_WARN; 868 } 869 __archive_read_consume(a, filename_length); 870 871 /* Read the extra data. */ 872 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 873 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 874 "Truncated ZIP file header"); 875 return (ARCHIVE_FATAL); 876 } 877 878 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 879 return ARCHIVE_FATAL; 880 } 881 __archive_read_consume(a, extra_length); 882 883 /* Work around a bug in Info-Zip: When reading from a pipe, it 884 * stats the pipe instead of synthesizing a file entry. */ 885 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 886 zip_entry->mode &= ~ AE_IFMT; 887 zip_entry->mode |= AE_IFREG; 888 } 889 890 /* If the mode is totally empty, set some sane default. */ 891 if (zip_entry->mode == 0) { 892 zip_entry->mode |= 0664; 893 } 894 895 /* Windows archivers sometimes use backslash as the directory separator. 896 Normalize to slash. */ 897 if (zip_entry->system == 0 && 898 (wp = archive_entry_pathname_w(entry)) != NULL) { 899 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { 900 size_t i; 901 struct archive_wstring s; 902 archive_string_init(&s); 903 archive_wstrcpy(&s, wp); 904 for (i = 0; i < archive_strlen(&s); i++) { 905 if (s.s[i] == '\\') 906 s.s[i] = '/'; 907 } 908 archive_entry_copy_pathname_w(entry, s.s); 909 archive_wstring_free(&s); 910 } 911 } 912 913 /* Make sure that entries with a trailing '/' are marked as directories 914 * even if the External File Attributes contains bogus values. If this 915 * is not a directory and there is no type, assume regularfile. */ 916 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 917 int has_slash; 918 919 wp = archive_entry_pathname_w(entry); 920 if (wp != NULL) { 921 len = wcslen(wp); 922 has_slash = len > 0 && wp[len - 1] == L'/'; 923 } else { 924 cp = archive_entry_pathname(entry); 925 len = (cp != NULL)?strlen(cp):0; 926 has_slash = len > 0 && cp[len - 1] == '/'; 927 } 928 /* Correct file type as needed. */ 929 if (has_slash) { 930 zip_entry->mode &= ~AE_IFMT; 931 zip_entry->mode |= AE_IFDIR; 932 zip_entry->mode |= 0111; 933 } else if ((zip_entry->mode & AE_IFMT) == 0) { 934 zip_entry->mode |= AE_IFREG; 935 } 936 } 937 938 /* Make sure directories end in '/' */ 939 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 940 wp = archive_entry_pathname_w(entry); 941 if (wp != NULL) { 942 len = wcslen(wp); 943 if (len > 0 && wp[len - 1] != L'/') { 944 struct archive_wstring s; 945 archive_string_init(&s); 946 archive_wstrcat(&s, wp); 947 archive_wstrappend_wchar(&s, L'/'); 948 archive_entry_copy_pathname_w(entry, s.s); 949 archive_wstring_free(&s); 950 } 951 } else { 952 cp = archive_entry_pathname(entry); 953 len = (cp != NULL)?strlen(cp):0; 954 if (len > 0 && cp[len - 1] != '/') { 955 struct archive_string s; 956 archive_string_init(&s); 957 archive_strcat(&s, cp); 958 archive_strappend_char(&s, '/'); 959 archive_entry_set_pathname(entry, s.s); 960 archive_string_free(&s); 961 } 962 } 963 } 964 965 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 966 /* If this came from the central dir, it's size info 967 * is definitive, so ignore the length-at-end flag. */ 968 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 969 /* If local header is missing a value, use the one from 970 the central directory. If both have it, warn about 971 mismatches. */ 972 if (zip_entry->crc32 == 0) { 973 zip_entry->crc32 = zip_entry_central_dir.crc32; 974 } else if (!zip->ignore_crc32 975 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 976 archive_set_error(&a->archive, 977 ARCHIVE_ERRNO_FILE_FORMAT, 978 "Inconsistent CRC32 values"); 979 ret = ARCHIVE_WARN; 980 } 981 if (zip_entry->compressed_size == 0) { 982 zip_entry->compressed_size 983 = zip_entry_central_dir.compressed_size; 984 } else if (zip_entry->compressed_size 985 != zip_entry_central_dir.compressed_size) { 986 archive_set_error(&a->archive, 987 ARCHIVE_ERRNO_FILE_FORMAT, 988 "Inconsistent compressed size: " 989 "%jd in central directory, %jd in local header", 990 (intmax_t)zip_entry_central_dir.compressed_size, 991 (intmax_t)zip_entry->compressed_size); 992 ret = ARCHIVE_WARN; 993 } 994 if (zip_entry->uncompressed_size == 0) { 995 zip_entry->uncompressed_size 996 = zip_entry_central_dir.uncompressed_size; 997 } else if (zip_entry->uncompressed_size 998 != zip_entry_central_dir.uncompressed_size) { 999 archive_set_error(&a->archive, 1000 ARCHIVE_ERRNO_FILE_FORMAT, 1001 "Inconsistent uncompressed size: " 1002 "%jd in central directory, %jd in local header", 1003 (intmax_t)zip_entry_central_dir.uncompressed_size, 1004 (intmax_t)zip_entry->uncompressed_size); 1005 ret = ARCHIVE_WARN; 1006 } 1007 } 1008 1009 /* Populate some additional entry fields: */ 1010 archive_entry_set_mode(entry, zip_entry->mode); 1011 archive_entry_set_uid(entry, zip_entry->uid); 1012 archive_entry_set_gid(entry, zip_entry->gid); 1013 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 1014 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 1015 archive_entry_set_atime(entry, zip_entry->atime, 0); 1016 1017 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 1018 size_t linkname_length; 1019 1020 if (zip_entry->compressed_size > 64 * 1024) { 1021 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1022 "Zip file with oversized link entry"); 1023 return ARCHIVE_FATAL; 1024 } 1025 1026 linkname_length = (size_t)zip_entry->compressed_size; 1027 1028 archive_entry_set_size(entry, 0); 1029 p = __archive_read_ahead(a, linkname_length, NULL); 1030 if (p == NULL) { 1031 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1032 "Truncated Zip file"); 1033 return ARCHIVE_FATAL; 1034 } 1035 1036 sconv = zip->sconv; 1037 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1038 sconv = zip->sconv_utf8; 1039 if (sconv == NULL) 1040 sconv = zip->sconv_default; 1041 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 1042 sconv) != 0) { 1043 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1044 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1045 archive_entry_copy_symlink_l(entry, p, 1046 linkname_length, NULL); 1047 if (errno == ENOMEM) { 1048 archive_set_error(&a->archive, ENOMEM, 1049 "Can't allocate memory for Symlink"); 1050 return (ARCHIVE_FATAL); 1051 } 1052 /* 1053 * Since there is no character-set regulation for 1054 * symlink name, do not report the conversion error 1055 * in an automatic conversion. 1056 */ 1057 if (sconv != zip->sconv_utf8 || 1058 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1059 archive_set_error(&a->archive, 1060 ARCHIVE_ERRNO_FILE_FORMAT, 1061 "Symlink cannot be converted " 1062 "from %s to current locale.", 1063 archive_string_conversion_charset_name( 1064 sconv)); 1065 ret = ARCHIVE_WARN; 1066 } 1067 } 1068 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1069 1070 if (__archive_read_consume(a, linkname_length) < 0) { 1071 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1072 "Read error skipping symlink target name"); 1073 return ARCHIVE_FATAL; 1074 } 1075 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1076 || zip_entry->uncompressed_size > 0) { 1077 /* Set the size only if it's meaningful. */ 1078 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1079 } 1080 zip->entry_bytes_remaining = zip_entry->compressed_size; 1081 1082 /* If there's no body, force read_data() to return EOF immediately. */ 1083 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1084 && zip->entry_bytes_remaining < 1) 1085 zip->end_of_entry = 1; 1086 1087 /* Set up a more descriptive format name. */ 1088 archive_string_empty(&zip->format_name); 1089 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1090 version / 10, version % 10, 1091 compression_name(zip->entry->compression)); 1092 a->archive.archive_format_name = zip->format_name.s; 1093 1094 return (ret); 1095} 1096 1097static int 1098check_authentication_code(struct archive_read *a, const void *_p) 1099{ 1100 struct zip *zip = (struct zip *)(a->format->data); 1101 1102 /* Check authentication code. */ 1103 if (zip->hctx_valid) { 1104 const void *p; 1105 uint8_t hmac[20]; 1106 size_t hmac_len = 20; 1107 int cmp; 1108 1109 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1110 if (_p == NULL) { 1111 /* Read authentication code. */ 1112 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1113 if (p == NULL) { 1114 archive_set_error(&a->archive, 1115 ARCHIVE_ERRNO_FILE_FORMAT, 1116 "Truncated ZIP file data"); 1117 return (ARCHIVE_FATAL); 1118 } 1119 } else { 1120 p = _p; 1121 } 1122 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1123 __archive_read_consume(a, AUTH_CODE_SIZE); 1124 if (cmp != 0) { 1125 archive_set_error(&a->archive, 1126 ARCHIVE_ERRNO_MISC, 1127 "ZIP bad Authentication code"); 1128 return (ARCHIVE_WARN); 1129 } 1130 } 1131 return (ARCHIVE_OK); 1132} 1133 1134/* 1135 * Read "uncompressed" data. There are three cases: 1136 * 1) We know the size of the data. This is always true for the 1137 * seeking reader (we've examined the Central Directory already). 1138 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1139 * Info-ZIP seems to do this; we know the size but have to grab 1140 * the CRC from the data descriptor afterwards. 1141 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1142 * we have no size information. In this case, we can do pretty 1143 * well by watching for the data descriptor record. The data 1144 * descriptor is 16 bytes and includes a computed CRC that should 1145 * provide a strong check. 1146 * 1147 * TODO: Technically, the PK\007\010 signature is optional. 1148 * In the original spec, the data descriptor contained CRC 1149 * and size fields but had no leading signature. In practice, 1150 * newer writers seem to provide the signature pretty consistently. 1151 * 1152 * For uncompressed data, the PK\007\010 marker seems essential 1153 * to be sure we've actually seen the end of the entry. 1154 * 1155 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1156 * zip->end_of_entry if it consumes all of the data. 1157 */ 1158static int 1159zip_read_data_none(struct archive_read *a, const void **_buff, 1160 size_t *size, int64_t *offset) 1161{ 1162 struct zip *zip; 1163 const char *buff; 1164 ssize_t bytes_avail; 1165 int r; 1166 1167 (void)offset; /* UNUSED */ 1168 1169 zip = (struct zip *)(a->format->data); 1170 1171 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1172 const char *p; 1173 ssize_t grabbing_bytes = 24; 1174 1175 if (zip->hctx_valid) 1176 grabbing_bytes += AUTH_CODE_SIZE; 1177 /* Grab at least 24 bytes. */ 1178 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1179 if (bytes_avail < grabbing_bytes) { 1180 /* Zip archives have end-of-archive markers 1181 that are longer than this, so a failure to get at 1182 least 24 bytes really does indicate a truncated 1183 file. */ 1184 archive_set_error(&a->archive, 1185 ARCHIVE_ERRNO_FILE_FORMAT, 1186 "Truncated ZIP file data"); 1187 return (ARCHIVE_FATAL); 1188 } 1189 /* Check for a complete PK\007\010 signature, followed 1190 * by the correct 4-byte CRC. */ 1191 p = buff; 1192 if (zip->hctx_valid) 1193 p += AUTH_CODE_SIZE; 1194 if (p[0] == 'P' && p[1] == 'K' 1195 && p[2] == '\007' && p[3] == '\010' 1196 && (archive_le32dec(p + 4) == zip->entry_crc32 1197 || zip->ignore_crc32 1198 || (zip->hctx_valid 1199 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1200 if (zip->entry->flags & LA_USED_ZIP64) { 1201 uint64_t compressed, uncompressed; 1202 zip->entry->crc32 = archive_le32dec(p + 4); 1203 compressed = archive_le64dec(p + 8); 1204 uncompressed = archive_le64dec(p + 16); 1205 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1206 archive_set_error(&a->archive, 1207 ARCHIVE_ERRNO_FILE_FORMAT, 1208 "Overflow of 64-bit file sizes"); 1209 return ARCHIVE_FAILED; 1210 } 1211 zip->entry->compressed_size = compressed; 1212 zip->entry->uncompressed_size = uncompressed; 1213 zip->unconsumed = 24; 1214 } else { 1215 zip->entry->crc32 = archive_le32dec(p + 4); 1216 zip->entry->compressed_size = 1217 archive_le32dec(p + 8); 1218 zip->entry->uncompressed_size = 1219 archive_le32dec(p + 12); 1220 zip->unconsumed = 16; 1221 } 1222 if (zip->hctx_valid) { 1223 r = check_authentication_code(a, buff); 1224 if (r != ARCHIVE_OK) 1225 return (r); 1226 } 1227 zip->end_of_entry = 1; 1228 return (ARCHIVE_OK); 1229 } 1230 /* If not at EOF, ensure we consume at least one byte. */ 1231 ++p; 1232 1233 /* Scan forward until we see where a PK\007\010 signature 1234 * might be. */ 1235 /* Return bytes up until that point. On the next call, 1236 * the code above will verify the data descriptor. */ 1237 while (p < buff + bytes_avail - 4) { 1238 if (p[3] == 'P') { p += 3; } 1239 else if (p[3] == 'K') { p += 2; } 1240 else if (p[3] == '\007') { p += 1; } 1241 else if (p[3] == '\010' && p[2] == '\007' 1242 && p[1] == 'K' && p[0] == 'P') { 1243 if (zip->hctx_valid) 1244 p -= AUTH_CODE_SIZE; 1245 break; 1246 } else { p += 4; } 1247 } 1248 bytes_avail = p - buff; 1249 } else { 1250 if (zip->entry_bytes_remaining == 0) { 1251 zip->end_of_entry = 1; 1252 if (zip->hctx_valid) { 1253 r = check_authentication_code(a, NULL); 1254 if (r != ARCHIVE_OK) 1255 return (r); 1256 } 1257 return (ARCHIVE_OK); 1258 } 1259 /* Grab a bunch of bytes. */ 1260 buff = __archive_read_ahead(a, 1, &bytes_avail); 1261 if (bytes_avail <= 0) { 1262 archive_set_error(&a->archive, 1263 ARCHIVE_ERRNO_FILE_FORMAT, 1264 "Truncated ZIP file data"); 1265 return (ARCHIVE_FATAL); 1266 } 1267 if (bytes_avail > zip->entry_bytes_remaining) 1268 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1269 } 1270 if (zip->tctx_valid || zip->cctx_valid) { 1271 size_t dec_size = bytes_avail; 1272 1273 if (dec_size > zip->decrypted_buffer_size) 1274 dec_size = zip->decrypted_buffer_size; 1275 if (zip->tctx_valid) { 1276 trad_enc_decrypt_update(&zip->tctx, 1277 (const uint8_t *)buff, dec_size, 1278 zip->decrypted_buffer, dec_size); 1279 } else { 1280 size_t dsize = dec_size; 1281 archive_hmac_sha1_update(&zip->hctx, 1282 (const uint8_t *)buff, dec_size); 1283 archive_decrypto_aes_ctr_update(&zip->cctx, 1284 (const uint8_t *)buff, dec_size, 1285 zip->decrypted_buffer, &dsize); 1286 } 1287 bytes_avail = dec_size; 1288 buff = (const char *)zip->decrypted_buffer; 1289 } 1290 *size = bytes_avail; 1291 zip->entry_bytes_remaining -= bytes_avail; 1292 zip->entry_uncompressed_bytes_read += bytes_avail; 1293 zip->entry_compressed_bytes_read += bytes_avail; 1294 zip->unconsumed += bytes_avail; 1295 *_buff = buff; 1296 return (ARCHIVE_OK); 1297} 1298 1299#ifdef HAVE_ZLIB_H 1300static int 1301zip_deflate_init(struct archive_read *a, struct zip *zip) 1302{ 1303 int r; 1304 1305 /* If we haven't yet read any data, initialize the decompressor. */ 1306 if (!zip->decompress_init) { 1307 if (zip->stream_valid) 1308 r = inflateReset(&zip->stream); 1309 else 1310 r = inflateInit2(&zip->stream, 1311 -15 /* Don't check for zlib header */); 1312 if (r != Z_OK) { 1313 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1314 "Can't initialize ZIP decompression."); 1315 return (ARCHIVE_FATAL); 1316 } 1317 /* Stream structure has been set up. */ 1318 zip->stream_valid = 1; 1319 /* We've initialized decompression for this stream. */ 1320 zip->decompress_init = 1; 1321 } 1322 return (ARCHIVE_OK); 1323} 1324 1325static int 1326zip_read_data_deflate(struct archive_read *a, const void **buff, 1327 size_t *size, int64_t *offset) 1328{ 1329 struct zip *zip; 1330 ssize_t bytes_avail; 1331 const void *compressed_buff, *sp; 1332 int r; 1333 1334 (void)offset; /* UNUSED */ 1335 1336 zip = (struct zip *)(a->format->data); 1337 1338 /* If the buffer hasn't been allocated, allocate it now. */ 1339 if (zip->uncompressed_buffer == NULL) { 1340 zip->uncompressed_buffer_size = 256 * 1024; 1341 zip->uncompressed_buffer 1342 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1343 if (zip->uncompressed_buffer == NULL) { 1344 archive_set_error(&a->archive, ENOMEM, 1345 "No memory for ZIP decompression"); 1346 return (ARCHIVE_FATAL); 1347 } 1348 } 1349 1350 r = zip_deflate_init(a, zip); 1351 if (r != ARCHIVE_OK) 1352 return (r); 1353 1354 /* 1355 * Note: '1' here is a performance optimization. 1356 * Recall that the decompression layer returns a count of 1357 * available bytes; asking for more than that forces the 1358 * decompressor to combine reads by copying data. 1359 */ 1360 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1361 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1362 && bytes_avail > zip->entry_bytes_remaining) { 1363 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1364 } 1365 if (bytes_avail < 0) { 1366 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1367 "Truncated ZIP file body"); 1368 return (ARCHIVE_FATAL); 1369 } 1370 1371 if (zip->tctx_valid || zip->cctx_valid) { 1372 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1373 size_t buff_remaining = 1374 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1375 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1376 1377 if (buff_remaining > (size_t)bytes_avail) 1378 buff_remaining = (size_t)bytes_avail; 1379 1380 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1381 zip->entry_bytes_remaining > 0) { 1382 if ((int64_t)(zip->decrypted_bytes_remaining 1383 + buff_remaining) 1384 > zip->entry_bytes_remaining) { 1385 if (zip->entry_bytes_remaining < 1386 (int64_t)zip->decrypted_bytes_remaining) 1387 buff_remaining = 0; 1388 else 1389 buff_remaining = 1390 (size_t)zip->entry_bytes_remaining 1391 - zip->decrypted_bytes_remaining; 1392 } 1393 } 1394 if (buff_remaining > 0) { 1395 if (zip->tctx_valid) { 1396 trad_enc_decrypt_update(&zip->tctx, 1397 compressed_buff, buff_remaining, 1398 zip->decrypted_ptr 1399 + zip->decrypted_bytes_remaining, 1400 buff_remaining); 1401 } else { 1402 size_t dsize = buff_remaining; 1403 archive_decrypto_aes_ctr_update( 1404 &zip->cctx, 1405 compressed_buff, buff_remaining, 1406 zip->decrypted_ptr 1407 + zip->decrypted_bytes_remaining, 1408 &dsize); 1409 } 1410 zip->decrypted_bytes_remaining += buff_remaining; 1411 } 1412 } 1413 bytes_avail = zip->decrypted_bytes_remaining; 1414 compressed_buff = (const char *)zip->decrypted_ptr; 1415 } 1416 1417 /* 1418 * A bug in zlib.h: stream.next_in should be marked 'const' 1419 * but isn't (the library never alters data through the 1420 * next_in pointer, only reads it). The result: this ugly 1421 * cast to remove 'const'. 1422 */ 1423 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1424 zip->stream.avail_in = (uInt)bytes_avail; 1425 zip->stream.total_in = 0; 1426 zip->stream.next_out = zip->uncompressed_buffer; 1427 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1428 zip->stream.total_out = 0; 1429 1430 r = inflate(&zip->stream, 0); 1431 switch (r) { 1432 case Z_OK: 1433 break; 1434 case Z_STREAM_END: 1435 zip->end_of_entry = 1; 1436 break; 1437 case Z_MEM_ERROR: 1438 archive_set_error(&a->archive, ENOMEM, 1439 "Out of memory for ZIP decompression"); 1440 return (ARCHIVE_FATAL); 1441 default: 1442 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1443 "ZIP decompression failed (%d)", r); 1444 return (ARCHIVE_FATAL); 1445 } 1446 1447 /* Consume as much as the compressor actually used. */ 1448 bytes_avail = zip->stream.total_in; 1449 if (zip->tctx_valid || zip->cctx_valid) { 1450 zip->decrypted_bytes_remaining -= bytes_avail; 1451 if (zip->decrypted_bytes_remaining == 0) 1452 zip->decrypted_ptr = zip->decrypted_buffer; 1453 else 1454 zip->decrypted_ptr += bytes_avail; 1455 } 1456 /* Calculate compressed data as much as we used.*/ 1457 if (zip->hctx_valid) 1458 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1459 __archive_read_consume(a, bytes_avail); 1460 zip->entry_bytes_remaining -= bytes_avail; 1461 zip->entry_compressed_bytes_read += bytes_avail; 1462 1463 *size = zip->stream.total_out; 1464 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1465 *buff = zip->uncompressed_buffer; 1466 1467 if (zip->end_of_entry && zip->hctx_valid) { 1468 r = check_authentication_code(a, NULL); 1469 if (r != ARCHIVE_OK) 1470 return (r); 1471 } 1472 1473 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1474 const char *p; 1475 1476 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1477 archive_set_error(&a->archive, 1478 ARCHIVE_ERRNO_FILE_FORMAT, 1479 "Truncated ZIP end-of-file record"); 1480 return (ARCHIVE_FATAL); 1481 } 1482 /* Consume the optional PK\007\010 marker. */ 1483 if (p[0] == 'P' && p[1] == 'K' && 1484 p[2] == '\007' && p[3] == '\010') { 1485 p += 4; 1486 zip->unconsumed = 4; 1487 } 1488 if (zip->entry->flags & LA_USED_ZIP64) { 1489 uint64_t compressed, uncompressed; 1490 zip->entry->crc32 = archive_le32dec(p); 1491 compressed = archive_le64dec(p + 4); 1492 uncompressed = archive_le64dec(p + 12); 1493 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1494 archive_set_error(&a->archive, 1495 ARCHIVE_ERRNO_FILE_FORMAT, 1496 "Overflow of 64-bit file sizes"); 1497 return ARCHIVE_FAILED; 1498 } 1499 zip->entry->compressed_size = compressed; 1500 zip->entry->uncompressed_size = uncompressed; 1501 zip->unconsumed += 20; 1502 } else { 1503 zip->entry->crc32 = archive_le32dec(p); 1504 zip->entry->compressed_size = archive_le32dec(p + 4); 1505 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1506 zip->unconsumed += 12; 1507 } 1508 } 1509 1510 return (ARCHIVE_OK); 1511} 1512#endif 1513 1514static int 1515read_decryption_header(struct archive_read *a) 1516{ 1517 struct zip *zip = (struct zip *)(a->format->data); 1518 const char *p; 1519 unsigned int remaining_size; 1520 unsigned int ts; 1521 1522 /* 1523 * Read an initialization vector data field. 1524 */ 1525 p = __archive_read_ahead(a, 2, NULL); 1526 if (p == NULL) 1527 goto truncated; 1528 ts = zip->iv_size; 1529 zip->iv_size = archive_le16dec(p); 1530 __archive_read_consume(a, 2); 1531 if (ts < zip->iv_size) { 1532 free(zip->iv); 1533 zip->iv = NULL; 1534 } 1535 p = __archive_read_ahead(a, zip->iv_size, NULL); 1536 if (p == NULL) 1537 goto truncated; 1538 if (zip->iv == NULL) { 1539 zip->iv = malloc(zip->iv_size); 1540 if (zip->iv == NULL) 1541 goto nomem; 1542 } 1543 memcpy(zip->iv, p, zip->iv_size); 1544 __archive_read_consume(a, zip->iv_size); 1545 1546 /* 1547 * Read a size of remaining decryption header field. 1548 */ 1549 p = __archive_read_ahead(a, 14, NULL); 1550 if (p == NULL) 1551 goto truncated; 1552 remaining_size = archive_le32dec(p); 1553 if (remaining_size < 16 || remaining_size > (1 << 18)) 1554 goto corrupted; 1555 1556 /* Check if format version is supported. */ 1557 if (archive_le16dec(p+4) != 3) { 1558 archive_set_error(&a->archive, 1559 ARCHIVE_ERRNO_FILE_FORMAT, 1560 "Unsupported encryption format version: %u", 1561 archive_le16dec(p+4)); 1562 return (ARCHIVE_FAILED); 1563 } 1564 1565 /* 1566 * Read an encryption algorithm field. 1567 */ 1568 zip->alg_id = archive_le16dec(p+6); 1569 switch (zip->alg_id) { 1570 case 0x6601:/* DES */ 1571 case 0x6602:/* RC2 */ 1572 case 0x6603:/* 3DES 168 */ 1573 case 0x6609:/* 3DES 112 */ 1574 case 0x660E:/* AES 128 */ 1575 case 0x660F:/* AES 192 */ 1576 case 0x6610:/* AES 256 */ 1577 case 0x6702:/* RC2 (version >= 5.2) */ 1578 case 0x6720:/* Blowfish */ 1579 case 0x6721:/* Twofish */ 1580 case 0x6801:/* RC4 */ 1581 /* Supported encryption algorithm. */ 1582 break; 1583 default: 1584 archive_set_error(&a->archive, 1585 ARCHIVE_ERRNO_FILE_FORMAT, 1586 "Unknown encryption algorithm: %u", zip->alg_id); 1587 return (ARCHIVE_FAILED); 1588 } 1589 1590 /* 1591 * Read a bit length field. 1592 */ 1593 zip->bit_len = archive_le16dec(p+8); 1594 1595 /* 1596 * Read a flags field. 1597 */ 1598 zip->flags = archive_le16dec(p+10); 1599 switch (zip->flags & 0xf000) { 1600 case 0x0001: /* Password is required to decrypt. */ 1601 case 0x0002: /* Certificates only. */ 1602 case 0x0003: /* Password or certificate required to decrypt. */ 1603 break; 1604 default: 1605 archive_set_error(&a->archive, 1606 ARCHIVE_ERRNO_FILE_FORMAT, 1607 "Unknown encryption flag: %u", zip->flags); 1608 return (ARCHIVE_FAILED); 1609 } 1610 if ((zip->flags & 0xf000) == 0 || 1611 (zip->flags & 0xf000) == 0x4000) { 1612 archive_set_error(&a->archive, 1613 ARCHIVE_ERRNO_FILE_FORMAT, 1614 "Unknown encryption flag: %u", zip->flags); 1615 return (ARCHIVE_FAILED); 1616 } 1617 1618 /* 1619 * Read an encrypted random data field. 1620 */ 1621 ts = zip->erd_size; 1622 zip->erd_size = archive_le16dec(p+12); 1623 __archive_read_consume(a, 14); 1624 if ((zip->erd_size & 0xf) != 0 || 1625 (zip->erd_size + 16) > remaining_size || 1626 (zip->erd_size + 16) < zip->erd_size) 1627 goto corrupted; 1628 1629 if (ts < zip->erd_size) { 1630 free(zip->erd); 1631 zip->erd = NULL; 1632 } 1633 p = __archive_read_ahead(a, zip->erd_size, NULL); 1634 if (p == NULL) 1635 goto truncated; 1636 if (zip->erd == NULL) { 1637 zip->erd = malloc(zip->erd_size); 1638 if (zip->erd == NULL) 1639 goto nomem; 1640 } 1641 memcpy(zip->erd, p, zip->erd_size); 1642 __archive_read_consume(a, zip->erd_size); 1643 1644 /* 1645 * Read a reserved data field. 1646 */ 1647 p = __archive_read_ahead(a, 4, NULL); 1648 if (p == NULL) 1649 goto truncated; 1650 /* Reserved data size should be zero. */ 1651 if (archive_le32dec(p) != 0) 1652 goto corrupted; 1653 __archive_read_consume(a, 4); 1654 1655 /* 1656 * Read a password validation data field. 1657 */ 1658 p = __archive_read_ahead(a, 2, NULL); 1659 if (p == NULL) 1660 goto truncated; 1661 ts = zip->v_size; 1662 zip->v_size = archive_le16dec(p); 1663 __archive_read_consume(a, 2); 1664 if ((zip->v_size & 0x0f) != 0 || 1665 (zip->erd_size + zip->v_size + 16) > remaining_size || 1666 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1667 goto corrupted; 1668 if (ts < zip->v_size) { 1669 free(zip->v_data); 1670 zip->v_data = NULL; 1671 } 1672 p = __archive_read_ahead(a, zip->v_size, NULL); 1673 if (p == NULL) 1674 goto truncated; 1675 if (zip->v_data == NULL) { 1676 zip->v_data = malloc(zip->v_size); 1677 if (zip->v_data == NULL) 1678 goto nomem; 1679 } 1680 memcpy(zip->v_data, p, zip->v_size); 1681 __archive_read_consume(a, zip->v_size); 1682 1683 p = __archive_read_ahead(a, 4, NULL); 1684 if (p == NULL) 1685 goto truncated; 1686 zip->v_crc32 = archive_le32dec(p); 1687 __archive_read_consume(a, 4); 1688 1689 /*return (ARCHIVE_OK); 1690 * This is not fully implemented yet.*/ 1691 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1692 "Encrypted file is unsupported"); 1693 return (ARCHIVE_FAILED); 1694truncated: 1695 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1696 "Truncated ZIP file data"); 1697 return (ARCHIVE_FATAL); 1698corrupted: 1699 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1700 "Corrupted ZIP file data"); 1701 return (ARCHIVE_FATAL); 1702nomem: 1703 archive_set_error(&a->archive, ENOMEM, 1704 "No memory for ZIP decryption"); 1705 return (ARCHIVE_FATAL); 1706} 1707 1708static int 1709zip_alloc_decryption_buffer(struct archive_read *a) 1710{ 1711 struct zip *zip = (struct zip *)(a->format->data); 1712 size_t bs = 256 * 1024; 1713 1714 if (zip->decrypted_buffer == NULL) { 1715 zip->decrypted_buffer_size = bs; 1716 zip->decrypted_buffer = malloc(bs); 1717 if (zip->decrypted_buffer == NULL) { 1718 archive_set_error(&a->archive, ENOMEM, 1719 "No memory for ZIP decryption"); 1720 return (ARCHIVE_FATAL); 1721 } 1722 } 1723 zip->decrypted_ptr = zip->decrypted_buffer; 1724 return (ARCHIVE_OK); 1725} 1726 1727static int 1728init_traditional_PKWARE_decryption(struct archive_read *a) 1729{ 1730 struct zip *zip = (struct zip *)(a->format->data); 1731 const void *p; 1732 int retry; 1733 int r; 1734 1735 if (zip->tctx_valid) 1736 return (ARCHIVE_OK); 1737 1738 /* 1739 Read the 12 bytes encryption header stored at 1740 the start of the data area. 1741 */ 1742#define ENC_HEADER_SIZE 12 1743 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1744 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1745 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1746 "Truncated Zip encrypted body: only %jd bytes available", 1747 (intmax_t)zip->entry_bytes_remaining); 1748 return (ARCHIVE_FATAL); 1749 } 1750 1751 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1752 if (p == NULL) { 1753 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1754 "Truncated ZIP file data"); 1755 return (ARCHIVE_FATAL); 1756 } 1757 1758 for (retry = 0;; retry++) { 1759 const char *passphrase; 1760 uint8_t crcchk; 1761 1762 passphrase = __archive_read_next_passphrase(a); 1763 if (passphrase == NULL) { 1764 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1765 (retry > 0)? 1766 "Incorrect passphrase": 1767 "Passphrase required for this entry"); 1768 return (ARCHIVE_FAILED); 1769 } 1770 1771 /* 1772 * Initialize ctx for Traditional PKWARE Decryption. 1773 */ 1774 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1775 p, ENC_HEADER_SIZE, &crcchk); 1776 if (r == 0 && crcchk == zip->entry->decdat) 1777 break;/* The passphrase is OK. */ 1778 if (retry > 10000) { 1779 /* Avoid infinity loop. */ 1780 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1781 "Too many incorrect passphrases"); 1782 return (ARCHIVE_FAILED); 1783 } 1784 } 1785 1786 __archive_read_consume(a, ENC_HEADER_SIZE); 1787 zip->tctx_valid = 1; 1788 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1789 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1790 } 1791 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1792 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1793 zip->decrypted_bytes_remaining = 0; 1794 1795 return (zip_alloc_decryption_buffer(a)); 1796#undef ENC_HEADER_SIZE 1797} 1798 1799static int 1800init_WinZip_AES_decryption(struct archive_read *a) 1801{ 1802 struct zip *zip = (struct zip *)(a->format->data); 1803 const void *p; 1804 const uint8_t *pv; 1805 size_t key_len, salt_len; 1806 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1807 int retry; 1808 int r; 1809 1810 if (zip->cctx_valid || zip->hctx_valid) 1811 return (ARCHIVE_OK); 1812 1813 switch (zip->entry->aes_extra.strength) { 1814 case 1: salt_len = 8; key_len = 16; break; 1815 case 2: salt_len = 12; key_len = 24; break; 1816 case 3: salt_len = 16; key_len = 32; break; 1817 default: goto corrupted; 1818 } 1819 p = __archive_read_ahead(a, salt_len + 2, NULL); 1820 if (p == NULL) 1821 goto truncated; 1822 1823 for (retry = 0;; retry++) { 1824 const char *passphrase; 1825 1826 passphrase = __archive_read_next_passphrase(a); 1827 if (passphrase == NULL) { 1828 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1829 (retry > 0)? 1830 "Incorrect passphrase": 1831 "Passphrase required for this entry"); 1832 return (ARCHIVE_FAILED); 1833 } 1834 memset(derived_key, 0, sizeof(derived_key)); 1835 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1836 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1837 if (r != 0) { 1838 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1839 "Decryption is unsupported due to lack of " 1840 "crypto library"); 1841 return (ARCHIVE_FAILED); 1842 } 1843 1844 /* Check password verification value. */ 1845 pv = ((const uint8_t *)p) + salt_len; 1846 if (derived_key[key_len * 2] == pv[0] && 1847 derived_key[key_len * 2 + 1] == pv[1]) 1848 break;/* The passphrase is OK. */ 1849 if (retry > 10000) { 1850 /* Avoid infinity loop. */ 1851 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1852 "Too many incorrect passphrases"); 1853 return (ARCHIVE_FAILED); 1854 } 1855 } 1856 1857 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1858 if (r != 0) { 1859 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1860 "Decryption is unsupported due to lack of crypto library"); 1861 return (ARCHIVE_FAILED); 1862 } 1863 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1864 if (r != 0) { 1865 archive_decrypto_aes_ctr_release(&zip->cctx); 1866 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1867 "Failed to initialize HMAC-SHA1"); 1868 return (ARCHIVE_FAILED); 1869 } 1870 zip->cctx_valid = zip->hctx_valid = 1; 1871 __archive_read_consume(a, salt_len + 2); 1872 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1873 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1874 && zip->entry_bytes_remaining < 0) 1875 goto corrupted; 1876 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1877 zip->decrypted_bytes_remaining = 0; 1878 1879 zip->entry->compression = zip->entry->aes_extra.compression; 1880 return (zip_alloc_decryption_buffer(a)); 1881 1882truncated: 1883 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1884 "Truncated ZIP file data"); 1885 return (ARCHIVE_FATAL); 1886corrupted: 1887 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1888 "Corrupted ZIP file data"); 1889 return (ARCHIVE_FATAL); 1890} 1891 1892static int 1893archive_read_format_zip_read_data(struct archive_read *a, 1894 const void **buff, size_t *size, int64_t *offset) 1895{ 1896 int r; 1897 struct zip *zip = (struct zip *)(a->format->data); 1898 1899 if (zip->has_encrypted_entries == 1900 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1901 zip->has_encrypted_entries = 0; 1902 } 1903 1904 *offset = zip->entry_uncompressed_bytes_read; 1905 *size = 0; 1906 *buff = NULL; 1907 1908 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1909 if (zip->end_of_entry) 1910 return (ARCHIVE_EOF); 1911 1912 /* Return EOF immediately if this is a non-regular file. */ 1913 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1914 return (ARCHIVE_EOF); 1915 1916 __archive_read_consume(a, zip->unconsumed); 1917 zip->unconsumed = 0; 1918 1919 if (zip->init_decryption) { 1920 zip->has_encrypted_entries = 1; 1921 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1922 r = read_decryption_header(a); 1923 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1924 r = init_WinZip_AES_decryption(a); 1925 else 1926 r = init_traditional_PKWARE_decryption(a); 1927 if (r != ARCHIVE_OK) 1928 return (r); 1929 zip->init_decryption = 0; 1930 } 1931 1932 switch(zip->entry->compression) { 1933 case 0: /* No compression. */ 1934 r = zip_read_data_none(a, buff, size, offset); 1935 break; 1936#ifdef HAVE_ZLIB_H 1937 case 8: /* Deflate compression. */ 1938 r = zip_read_data_deflate(a, buff, size, offset); 1939 break; 1940#endif 1941 default: /* Unsupported compression. */ 1942 /* Return a warning. */ 1943 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1944 "Unsupported ZIP compression method (%s)", 1945 compression_name(zip->entry->compression)); 1946 /* We can't decompress this entry, but we will 1947 * be able to skip() it and try the next entry. */ 1948 return (ARCHIVE_FAILED); 1949 break; 1950 } 1951 if (r != ARCHIVE_OK) 1952 return (r); 1953 /* Update checksum */ 1954 if (*size) 1955 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1956 (unsigned)*size); 1957 /* If we hit the end, swallow any end-of-data marker. */ 1958 if (zip->end_of_entry) { 1959 /* Check file size, CRC against these values. */ 1960 if (zip->entry->compressed_size != 1961 zip->entry_compressed_bytes_read) { 1962 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1963 "ZIP compressed data is wrong size " 1964 "(read %jd, expected %jd)", 1965 (intmax_t)zip->entry_compressed_bytes_read, 1966 (intmax_t)zip->entry->compressed_size); 1967 return (ARCHIVE_WARN); 1968 } 1969 /* Size field only stores the lower 32 bits of the actual 1970 * size. */ 1971 if ((zip->entry->uncompressed_size & UINT32_MAX) 1972 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1973 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1974 "ZIP uncompressed data is wrong size " 1975 "(read %jd, expected %jd)\n", 1976 (intmax_t)zip->entry_uncompressed_bytes_read, 1977 (intmax_t)zip->entry->uncompressed_size); 1978 return (ARCHIVE_WARN); 1979 } 1980 /* Check computed CRC against header */ 1981 if ((!zip->hctx_valid || 1982 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1983 zip->entry->crc32 != zip->entry_crc32 1984 && !zip->ignore_crc32) { 1985 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1986 "ZIP bad CRC: 0x%lx should be 0x%lx", 1987 (unsigned long)zip->entry_crc32, 1988 (unsigned long)zip->entry->crc32); 1989 return (ARCHIVE_WARN); 1990 } 1991 } 1992 1993 return (ARCHIVE_OK); 1994} 1995 1996static int 1997archive_read_format_zip_cleanup(struct archive_read *a) 1998{ 1999 struct zip *zip; 2000 struct zip_entry *zip_entry, *next_zip_entry; 2001 2002 zip = (struct zip *)(a->format->data); 2003#ifdef HAVE_ZLIB_H 2004 if (zip->stream_valid) 2005 inflateEnd(&zip->stream); 2006 free(zip->uncompressed_buffer); 2007#endif 2008 if (zip->zip_entries) { 2009 zip_entry = zip->zip_entries; 2010 while (zip_entry != NULL) { 2011 next_zip_entry = zip_entry->next; 2012 archive_string_free(&zip_entry->rsrcname); 2013 free(zip_entry); 2014 zip_entry = next_zip_entry; 2015 } 2016 } 2017 free(zip->decrypted_buffer); 2018 if (zip->cctx_valid) 2019 archive_decrypto_aes_ctr_release(&zip->cctx); 2020 if (zip->hctx_valid) 2021 archive_hmac_sha1_cleanup(&zip->hctx); 2022 free(zip->iv); 2023 free(zip->erd); 2024 free(zip->v_data); 2025 archive_string_free(&zip->format_name); 2026 free(zip); 2027 (a->format->data) = NULL; 2028 return (ARCHIVE_OK); 2029} 2030 2031static int 2032archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 2033{ 2034 if (_a && _a->format) { 2035 struct zip * zip = (struct zip *)_a->format->data; 2036 if (zip) { 2037 return zip->has_encrypted_entries; 2038 } 2039 } 2040 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2041} 2042 2043static int 2044archive_read_format_zip_options(struct archive_read *a, 2045 const char *key, const char *val) 2046{ 2047 struct zip *zip; 2048 int ret = ARCHIVE_FAILED; 2049 2050 zip = (struct zip *)(a->format->data); 2051 if (strcmp(key, "compat-2x") == 0) { 2052 /* Handle filenames as libarchive 2.x */ 2053 zip->init_default_conversion = (val != NULL) ? 1 : 0; 2054 return (ARCHIVE_OK); 2055 } else if (strcmp(key, "hdrcharset") == 0) { 2056 if (val == NULL || val[0] == 0) 2057 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2058 "zip: hdrcharset option needs a character-set name" 2059 ); 2060 else { 2061 zip->sconv = archive_string_conversion_from_charset( 2062 &a->archive, val, 0); 2063 if (zip->sconv != NULL) { 2064 if (strcmp(val, "UTF-8") == 0) 2065 zip->sconv_utf8 = zip->sconv; 2066 ret = ARCHIVE_OK; 2067 } else 2068 ret = ARCHIVE_FATAL; 2069 } 2070 return (ret); 2071 } else if (strcmp(key, "ignorecrc32") == 0) { 2072 /* Mostly useful for testing. */ 2073 if (val == NULL || val[0] == 0) { 2074 zip->crc32func = real_crc32; 2075 zip->ignore_crc32 = 0; 2076 } else { 2077 zip->crc32func = fake_crc32; 2078 zip->ignore_crc32 = 1; 2079 } 2080 return (ARCHIVE_OK); 2081 } else if (strcmp(key, "mac-ext") == 0) { 2082 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2083 return (ARCHIVE_OK); 2084 } 2085 2086 /* Note: The "warn" return is just to inform the options 2087 * supervisor that we didn't handle it. It will generate 2088 * a suitable error if no one used this option. */ 2089 return (ARCHIVE_WARN); 2090} 2091 2092int 2093archive_read_support_format_zip(struct archive *a) 2094{ 2095 int r; 2096 r = archive_read_support_format_zip_streamable(a); 2097 if (r != ARCHIVE_OK) 2098 return r; 2099 return (archive_read_support_format_zip_seekable(a)); 2100} 2101 2102/* ------------------------------------------------------------------------ */ 2103 2104/* 2105 * Streaming-mode support 2106 */ 2107 2108 2109static int 2110archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2111{ 2112 (void)a; /* UNUSED */ 2113 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2114 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2115} 2116 2117static int 2118archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2119{ 2120 const char *p; 2121 2122 (void)best_bid; /* UNUSED */ 2123 2124 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2125 return (-1); 2126 2127 /* 2128 * Bid of 29 here comes from: 2129 * + 16 bits for "PK", 2130 * + next 16-bit field has 6 options so contributes 2131 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2132 * 2133 * So we've effectively verified ~29 total bits of check data. 2134 */ 2135 if (p[0] == 'P' && p[1] == 'K') { 2136 if ((p[2] == '\001' && p[3] == '\002') 2137 || (p[2] == '\003' && p[3] == '\004') 2138 || (p[2] == '\005' && p[3] == '\006') 2139 || (p[2] == '\006' && p[3] == '\006') 2140 || (p[2] == '\007' && p[3] == '\010') 2141 || (p[2] == '0' && p[3] == '0')) 2142 return (29); 2143 } 2144 2145 /* TODO: It's worth looking ahead a little bit for a valid 2146 * PK signature. In particular, that would make it possible 2147 * to read some UUEncoded SFX files or SFX files coming from 2148 * a network socket. */ 2149 2150 return (0); 2151} 2152 2153static int 2154archive_read_format_zip_streamable_read_header(struct archive_read *a, 2155 struct archive_entry *entry) 2156{ 2157 struct zip *zip; 2158 2159 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2160 if (a->archive.archive_format_name == NULL) 2161 a->archive.archive_format_name = "ZIP"; 2162 2163 zip = (struct zip *)(a->format->data); 2164 2165 /* 2166 * It should be sufficient to call archive_read_next_header() for 2167 * a reader to determine if an entry is encrypted or not. If the 2168 * encryption of an entry is only detectable when calling 2169 * archive_read_data(), so be it. We'll do the same check there 2170 * as well. 2171 */ 2172 if (zip->has_encrypted_entries == 2173 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2174 zip->has_encrypted_entries = 0; 2175 2176 /* Make sure we have a zip_entry structure to use. */ 2177 if (zip->zip_entries == NULL) { 2178 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2179 if (zip->zip_entries == NULL) { 2180 archive_set_error(&a->archive, ENOMEM, 2181 "Out of memory"); 2182 return ARCHIVE_FATAL; 2183 } 2184 } 2185 zip->entry = zip->zip_entries; 2186 memset(zip->entry, 0, sizeof(struct zip_entry)); 2187 2188 if (zip->cctx_valid) 2189 archive_decrypto_aes_ctr_release(&zip->cctx); 2190 if (zip->hctx_valid) 2191 archive_hmac_sha1_cleanup(&zip->hctx); 2192 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2193 __archive_read_reset_passphrase(a); 2194 2195 /* Search ahead for the next local file header. */ 2196 __archive_read_consume(a, zip->unconsumed); 2197 zip->unconsumed = 0; 2198 for (;;) { 2199 int64_t skipped = 0; 2200 const char *p, *end; 2201 ssize_t bytes; 2202 2203 p = __archive_read_ahead(a, 4, &bytes); 2204 if (p == NULL) 2205 return (ARCHIVE_FATAL); 2206 end = p + bytes; 2207 2208 while (p + 4 <= end) { 2209 if (p[0] == 'P' && p[1] == 'K') { 2210 if (p[2] == '\003' && p[3] == '\004') { 2211 /* Regular file entry. */ 2212 __archive_read_consume(a, skipped); 2213 return zip_read_local_file_header(a, 2214 entry, zip); 2215 } 2216 2217 /* 2218 * TODO: We cannot restore permissions 2219 * based only on the local file headers. 2220 * Consider scanning the central 2221 * directory and returning additional 2222 * entries for at least directories. 2223 * This would allow us to properly set 2224 * directory permissions. 2225 * 2226 * This won't help us fix symlinks 2227 * and may not help with regular file 2228 * permissions, either. <sigh> 2229 */ 2230 if (p[2] == '\001' && p[3] == '\002') { 2231 return (ARCHIVE_EOF); 2232 } 2233 2234 /* End of central directory? Must be an 2235 * empty archive. */ 2236 if ((p[2] == '\005' && p[3] == '\006') 2237 || (p[2] == '\006' && p[3] == '\006')) 2238 return (ARCHIVE_EOF); 2239 } 2240 ++p; 2241 ++skipped; 2242 } 2243 __archive_read_consume(a, skipped); 2244 } 2245} 2246 2247static int 2248archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2249{ 2250 struct zip *zip; 2251 int64_t bytes_skipped; 2252 2253 zip = (struct zip *)(a->format->data); 2254 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2255 zip->unconsumed = 0; 2256 if (bytes_skipped < 0) 2257 return (ARCHIVE_FATAL); 2258 2259 /* If we've already read to end of data, we're done. */ 2260 if (zip->end_of_entry) 2261 return (ARCHIVE_OK); 2262 2263 /* So we know we're streaming... */ 2264 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2265 || zip->entry->compressed_size > 0) { 2266 /* We know the compressed length, so we can just skip. */ 2267 bytes_skipped = __archive_read_consume(a, 2268 zip->entry_bytes_remaining); 2269 if (bytes_skipped < 0) 2270 return (ARCHIVE_FATAL); 2271 return (ARCHIVE_OK); 2272 } 2273 2274 if (zip->init_decryption) { 2275 int r; 2276 2277 zip->has_encrypted_entries = 1; 2278 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2279 r = read_decryption_header(a); 2280 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2281 r = init_WinZip_AES_decryption(a); 2282 else 2283 r = init_traditional_PKWARE_decryption(a); 2284 if (r != ARCHIVE_OK) 2285 return (r); 2286 zip->init_decryption = 0; 2287 } 2288 2289 /* We're streaming and we don't know the length. */ 2290 /* If the body is compressed and we know the format, we can 2291 * find an exact end-of-entry by decompressing it. */ 2292 switch (zip->entry->compression) { 2293#ifdef HAVE_ZLIB_H 2294 case 8: /* Deflate compression. */ 2295 while (!zip->end_of_entry) { 2296 int64_t offset = 0; 2297 const void *buff = NULL; 2298 size_t size = 0; 2299 int r; 2300 r = zip_read_data_deflate(a, &buff, &size, &offset); 2301 if (r != ARCHIVE_OK) 2302 return (r); 2303 } 2304 return ARCHIVE_OK; 2305#endif 2306 default: /* Uncompressed or unknown. */ 2307 /* Scan for a PK\007\010 signature. */ 2308 for (;;) { 2309 const char *p, *buff; 2310 ssize_t bytes_avail; 2311 buff = __archive_read_ahead(a, 16, &bytes_avail); 2312 if (bytes_avail < 16) { 2313 archive_set_error(&a->archive, 2314 ARCHIVE_ERRNO_FILE_FORMAT, 2315 "Truncated ZIP file data"); 2316 return (ARCHIVE_FATAL); 2317 } 2318 p = buff; 2319 while (p <= buff + bytes_avail - 16) { 2320 if (p[3] == 'P') { p += 3; } 2321 else if (p[3] == 'K') { p += 2; } 2322 else if (p[3] == '\007') { p += 1; } 2323 else if (p[3] == '\010' && p[2] == '\007' 2324 && p[1] == 'K' && p[0] == 'P') { 2325 if (zip->entry->flags & LA_USED_ZIP64) 2326 __archive_read_consume(a, 2327 p - buff + 24); 2328 else 2329 __archive_read_consume(a, 2330 p - buff + 16); 2331 return ARCHIVE_OK; 2332 } else { p += 4; } 2333 } 2334 __archive_read_consume(a, p - buff); 2335 } 2336 } 2337} 2338 2339int 2340archive_read_support_format_zip_streamable(struct archive *_a) 2341{ 2342 struct archive_read *a = (struct archive_read *)_a; 2343 struct zip *zip; 2344 int r; 2345 2346 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2347 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2348 2349 zip = (struct zip *)calloc(1, sizeof(*zip)); 2350 if (zip == NULL) { 2351 archive_set_error(&a->archive, ENOMEM, 2352 "Can't allocate zip data"); 2353 return (ARCHIVE_FATAL); 2354 } 2355 2356 /* Streamable reader doesn't support mac extensions. */ 2357 zip->process_mac_extensions = 0; 2358 2359 /* 2360 * Until enough data has been read, we cannot tell about 2361 * any encrypted entries yet. 2362 */ 2363 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2364 zip->crc32func = real_crc32; 2365 2366 r = __archive_read_register_format(a, 2367 zip, 2368 "zip", 2369 archive_read_format_zip_streamable_bid, 2370 archive_read_format_zip_options, 2371 archive_read_format_zip_streamable_read_header, 2372 archive_read_format_zip_read_data, 2373 archive_read_format_zip_read_data_skip_streamable, 2374 NULL, 2375 archive_read_format_zip_cleanup, 2376 archive_read_support_format_zip_capabilities_streamable, 2377 archive_read_format_zip_has_encrypted_entries); 2378 2379 if (r != ARCHIVE_OK) 2380 free(zip); 2381 return (ARCHIVE_OK); 2382} 2383 2384/* ------------------------------------------------------------------------ */ 2385 2386/* 2387 * Seeking-mode support 2388 */ 2389 2390static int 2391archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2392{ 2393 (void)a; /* UNUSED */ 2394 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2395 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2396} 2397 2398/* 2399 * TODO: This is a performance sink because it forces the read core to 2400 * drop buffered data from the start of file, which will then have to 2401 * be re-read again if this bidder loses. 2402 * 2403 * We workaround this a little by passing in the best bid so far so 2404 * that later bidders can do nothing if they know they'll never 2405 * outbid. But we can certainly do better... 2406 */ 2407static int 2408read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2409{ 2410 /* Sanity-check the EOCD we've found. */ 2411 2412 /* This must be the first volume. */ 2413 if (archive_le16dec(p + 4) != 0) 2414 return 0; 2415 /* Central directory must be on this volume. */ 2416 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2417 return 0; 2418 /* All central directory entries must be on this volume. */ 2419 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2420 return 0; 2421 /* Central directory can't extend beyond start of EOCD record. */ 2422 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2423 > current_offset) 2424 return 0; 2425 2426 /* Save the central directory location for later use. */ 2427 zip->central_directory_offset = archive_le32dec(p + 16); 2428 2429 /* This is just a tiny bit higher than the maximum 2430 returned by the streaming Zip bidder. This ensures 2431 that the more accurate seeking Zip parser wins 2432 whenever seek is available. */ 2433 return 32; 2434} 2435 2436/* 2437 * Examine Zip64 EOCD locator: If it's valid, store the information 2438 * from it. 2439 */ 2440static int 2441read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2442{ 2443 int64_t eocd64_offset; 2444 int64_t eocd64_size; 2445 2446 /* Sanity-check the locator record. */ 2447 2448 /* Central dir must be on first volume. */ 2449 if (archive_le32dec(p + 4) != 0) 2450 return 0; 2451 /* Must be only a single volume. */ 2452 if (archive_le32dec(p + 16) != 1) 2453 return 0; 2454 2455 /* Find the Zip64 EOCD record. */ 2456 eocd64_offset = archive_le64dec(p + 8); 2457 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2458 return 0; 2459 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2460 return 0; 2461 /* Make sure we can read all of it. */ 2462 eocd64_size = archive_le64dec(p + 4) + 12; 2463 if (eocd64_size < 56 || eocd64_size > 16384) 2464 return 0; 2465 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2466 return 0; 2467 2468 /* Sanity-check the EOCD64 */ 2469 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2470 return 0; 2471 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2472 return 0; 2473 /* CD can't be split. */ 2474 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2475 return 0; 2476 2477 /* Save the central directory offset for later use. */ 2478 zip->central_directory_offset = archive_le64dec(p + 48); 2479 2480 return 32; 2481} 2482 2483static int 2484archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2485{ 2486 struct zip *zip = (struct zip *)a->format->data; 2487 int64_t file_size, current_offset; 2488 const char *p; 2489 int i, tail; 2490 2491 /* If someone has already bid more than 32, then avoid 2492 trashing the look-ahead buffers with a seek. */ 2493 if (best_bid > 32) 2494 return (-1); 2495 2496 file_size = __archive_read_seek(a, 0, SEEK_END); 2497 if (file_size <= 0) 2498 return 0; 2499 2500 /* Search last 16k of file for end-of-central-directory 2501 * record (which starts with PK\005\006) */ 2502 tail = (int)zipmin(1024 * 16, file_size); 2503 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2504 if (current_offset < 0) 2505 return 0; 2506 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2507 return 0; 2508 /* Boyer-Moore search backwards from the end, since we want 2509 * to match the last EOCD in the file (there can be more than 2510 * one if there is an uncompressed Zip archive as a member 2511 * within this Zip archive). */ 2512 for (i = tail - 22; i > 0;) { 2513 switch (p[i]) { 2514 case 'P': 2515 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2516 int ret = read_eocd(zip, p + i, 2517 current_offset + i); 2518 /* Zip64 EOCD locator precedes 2519 * regular EOCD if present. */ 2520 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2521 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); 2522 if (ret_zip64 > ret) 2523 ret = ret_zip64; 2524 } 2525 return (ret); 2526 } 2527 i -= 4; 2528 break; 2529 case 'K': i -= 1; break; 2530 case 005: i -= 2; break; 2531 case 006: i -= 3; break; 2532 default: i -= 4; break; 2533 } 2534 } 2535 return 0; 2536} 2537 2538/* The red-black trees are only used in seeking mode to manage 2539 * the in-memory copy of the central directory. */ 2540 2541static int 2542cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2543{ 2544 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2545 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2546 2547 if (e1->local_header_offset > e2->local_header_offset) 2548 return -1; 2549 if (e1->local_header_offset < e2->local_header_offset) 2550 return 1; 2551 return 0; 2552} 2553 2554static int 2555cmp_key(const struct archive_rb_node *n, const void *key) 2556{ 2557 /* This function won't be called */ 2558 (void)n; /* UNUSED */ 2559 (void)key; /* UNUSED */ 2560 return 1; 2561} 2562 2563static const struct archive_rb_tree_ops rb_ops = { 2564 &cmp_node, &cmp_key 2565}; 2566 2567static int 2568rsrc_cmp_node(const struct archive_rb_node *n1, 2569 const struct archive_rb_node *n2) 2570{ 2571 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2572 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2573 2574 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2575} 2576 2577static int 2578rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2579{ 2580 const struct zip_entry *e = (const struct zip_entry *)n; 2581 return (strcmp((const char *)key, e->rsrcname.s)); 2582} 2583 2584static const struct archive_rb_tree_ops rb_rsrc_ops = { 2585 &rsrc_cmp_node, &rsrc_cmp_key 2586}; 2587 2588static const char * 2589rsrc_basename(const char *name, size_t name_length) 2590{ 2591 const char *s, *r; 2592 2593 r = s = name; 2594 for (;;) { 2595 s = memchr(s, '/', name_length - (s - name)); 2596 if (s == NULL) 2597 break; 2598 r = ++s; 2599 } 2600 return (r); 2601} 2602 2603static void 2604expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2605{ 2606 struct archive_string str; 2607 struct zip_entry *dir; 2608 char *s; 2609 2610 archive_string_init(&str); 2611 archive_strncpy(&str, name, name_length); 2612 for (;;) { 2613 s = strrchr(str.s, '/'); 2614 if (s == NULL) 2615 break; 2616 *s = '\0'; 2617 /* Transfer the parent directory from zip->tree_rsrc RB 2618 * tree to zip->tree RB tree to expose. */ 2619 dir = (struct zip_entry *) 2620 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2621 if (dir == NULL) 2622 break; 2623 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2624 archive_string_free(&dir->rsrcname); 2625 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2626 } 2627 archive_string_free(&str); 2628} 2629 2630static int 2631slurp_central_directory(struct archive_read *a, struct zip *zip) 2632{ 2633 ssize_t i; 2634 unsigned found; 2635 int64_t correction; 2636 ssize_t bytes_avail; 2637 const char *p; 2638 2639 /* 2640 * Find the start of the central directory. The end-of-CD 2641 * record has our starting point, but there are lots of 2642 * Zip archives which have had other data prepended to the 2643 * file, which makes the recorded offsets all too small. 2644 * So we search forward from the specified offset until we 2645 * find the real start of the central directory. Then we 2646 * know the correction we need to apply to account for leading 2647 * padding. 2648 */ 2649 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2650 return ARCHIVE_FATAL; 2651 2652 found = 0; 2653 while (!found) { 2654 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2655 return ARCHIVE_FATAL; 2656 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2657 switch (p[i + 3]) { 2658 case 'P': i += 3; break; 2659 case 'K': i += 2; break; 2660 case 001: i += 1; break; 2661 case 002: 2662 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2663 p += i; 2664 found = 1; 2665 } else 2666 i += 4; 2667 break; 2668 case 005: i += 1; break; 2669 case 006: 2670 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2671 p += i; 2672 found = 1; 2673 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2674 p += i; 2675 found = 1; 2676 } else 2677 i += 1; 2678 break; 2679 default: i += 4; break; 2680 } 2681 } 2682 __archive_read_consume(a, i); 2683 } 2684 correction = archive_filter_bytes(&a->archive, 0) 2685 - zip->central_directory_offset; 2686 2687 __archive_rb_tree_init(&zip->tree, &rb_ops); 2688 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2689 2690 zip->central_directory_entries_total = 0; 2691 while (1) { 2692 struct zip_entry *zip_entry; 2693 size_t filename_length, extra_length, comment_length; 2694 uint32_t external_attributes; 2695 const char *name, *r; 2696 2697 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2698 return ARCHIVE_FATAL; 2699 if (memcmp(p, "PK\006\006", 4) == 0 2700 || memcmp(p, "PK\005\006", 4) == 0) { 2701 break; 2702 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2703 archive_set_error(&a->archive, 2704 -1, "Invalid central directory signature"); 2705 return ARCHIVE_FATAL; 2706 } 2707 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2708 return ARCHIVE_FATAL; 2709 2710 zip_entry = calloc(1, sizeof(struct zip_entry)); 2711 zip_entry->next = zip->zip_entries; 2712 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2713 zip->zip_entries = zip_entry; 2714 zip->central_directory_entries_total++; 2715 2716 /* version = p[4]; */ 2717 zip_entry->system = p[5]; 2718 /* version_required = archive_le16dec(p + 6); */ 2719 zip_entry->zip_flags = archive_le16dec(p + 8); 2720 if (zip_entry->zip_flags 2721 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2722 zip->has_encrypted_entries = 1; 2723 } 2724 zip_entry->compression = (char)archive_le16dec(p + 10); 2725 zip_entry->mtime = zip_time(p + 12); 2726 zip_entry->crc32 = archive_le32dec(p + 16); 2727 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2728 zip_entry->decdat = p[13]; 2729 else 2730 zip_entry->decdat = p[19]; 2731 zip_entry->compressed_size = archive_le32dec(p + 20); 2732 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2733 filename_length = archive_le16dec(p + 28); 2734 extra_length = archive_le16dec(p + 30); 2735 comment_length = archive_le16dec(p + 32); 2736 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2737 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2738 external_attributes = archive_le32dec(p + 38); 2739 zip_entry->local_header_offset = 2740 archive_le32dec(p + 42) + correction; 2741 2742 /* If we can't guess the mode, leave it zero here; 2743 when we read the local file header we might get 2744 more information. */ 2745 if (zip_entry->system == 3) { 2746 zip_entry->mode = external_attributes >> 16; 2747 } else if (zip_entry->system == 0) { 2748 // Interpret MSDOS directory bit 2749 if (0x10 == (external_attributes & 0x10)) { 2750 zip_entry->mode = AE_IFDIR | 0775; 2751 } else { 2752 zip_entry->mode = AE_IFREG | 0664; 2753 } 2754 if (0x01 == (external_attributes & 0x01)) { 2755 // Read-only bit; strip write permissions 2756 zip_entry->mode &= 0555; 2757 } 2758 } else { 2759 zip_entry->mode = 0; 2760 } 2761 2762 /* We're done with the regular data; get the filename and 2763 * extra data. */ 2764 __archive_read_consume(a, 46); 2765 p = __archive_read_ahead(a, filename_length + extra_length, 2766 NULL); 2767 if (p == NULL) { 2768 archive_set_error(&a->archive, 2769 ARCHIVE_ERRNO_FILE_FORMAT, 2770 "Truncated ZIP file header"); 2771 return ARCHIVE_FATAL; 2772 } 2773 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2774 return ARCHIVE_FATAL; 2775 } 2776 2777 /* 2778 * Mac resource fork files are stored under the 2779 * "__MACOSX/" directory, so we should check if 2780 * it is. 2781 */ 2782 if (!zip->process_mac_extensions) { 2783 /* Treat every entry as a regular entry. */ 2784 __archive_rb_tree_insert_node(&zip->tree, 2785 &zip_entry->node); 2786 } else { 2787 name = p; 2788 r = rsrc_basename(name, filename_length); 2789 if (filename_length >= 9 && 2790 strncmp("__MACOSX/", name, 9) == 0) { 2791 /* If this file is not a resource fork nor 2792 * a directory. We should treat it as a non 2793 * resource fork file to expose it. */ 2794 if (name[filename_length-1] != '/' && 2795 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2796 __archive_rb_tree_insert_node( 2797 &zip->tree, &zip_entry->node); 2798 /* Expose its parent directories. */ 2799 expose_parent_dirs(zip, name, 2800 filename_length); 2801 } else { 2802 /* This file is a resource fork file or 2803 * a directory. */ 2804 archive_strncpy(&(zip_entry->rsrcname), 2805 name, filename_length); 2806 __archive_rb_tree_insert_node( 2807 &zip->tree_rsrc, &zip_entry->node); 2808 } 2809 } else { 2810 /* Generate resource fork name to find its 2811 * resource file at zip->tree_rsrc. */ 2812 archive_strcpy(&(zip_entry->rsrcname), 2813 "__MACOSX/"); 2814 archive_strncat(&(zip_entry->rsrcname), 2815 name, r - name); 2816 archive_strcat(&(zip_entry->rsrcname), "._"); 2817 archive_strncat(&(zip_entry->rsrcname), 2818 name + (r - name), 2819 filename_length - (r - name)); 2820 /* Register an entry to RB tree to sort it by 2821 * file offset. */ 2822 __archive_rb_tree_insert_node(&zip->tree, 2823 &zip_entry->node); 2824 } 2825 } 2826 2827 /* Skip the comment too ... */ 2828 __archive_read_consume(a, 2829 filename_length + extra_length + comment_length); 2830 } 2831 2832 return ARCHIVE_OK; 2833} 2834 2835static ssize_t 2836zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2837{ 2838 const char *p; 2839 ssize_t filename_length, extra_length; 2840 2841 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2842 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2843 "Truncated ZIP file header"); 2844 return (ARCHIVE_WARN); 2845 } 2846 p += extra; 2847 2848 if (memcmp(p, "PK\003\004", 4) != 0) { 2849 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2850 return ARCHIVE_WARN; 2851 } 2852 filename_length = archive_le16dec(p + 26); 2853 extra_length = archive_le16dec(p + 28); 2854 2855 return (30 + filename_length + extra_length); 2856} 2857 2858static int 2859zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2860 struct zip_entry *rsrc) 2861{ 2862 struct zip *zip = (struct zip *)a->format->data; 2863 unsigned char *metadata, *mp; 2864 int64_t offset = archive_filter_bytes(&a->archive, 0); 2865 size_t remaining_bytes, metadata_bytes; 2866 ssize_t hsize; 2867 int ret = ARCHIVE_OK, eof; 2868 2869 switch(rsrc->compression) { 2870 case 0: /* No compression. */ 2871 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2872 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2873 "Malformed OS X metadata entry: inconsistent size"); 2874 return (ARCHIVE_FATAL); 2875 } 2876#ifdef HAVE_ZLIB_H 2877 case 8: /* Deflate compression. */ 2878#endif 2879 break; 2880 default: /* Unsupported compression. */ 2881 /* Return a warning. */ 2882 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2883 "Unsupported ZIP compression method (%s)", 2884 compression_name(rsrc->compression)); 2885 /* We can't decompress this entry, but we will 2886 * be able to skip() it and try the next entry. */ 2887 return (ARCHIVE_WARN); 2888 } 2889 2890 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2891 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2892 "Mac metadata is too large: %jd > 4M bytes", 2893 (intmax_t)rsrc->uncompressed_size); 2894 return (ARCHIVE_WARN); 2895 } 2896 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2897 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2898 "Mac metadata is too large: %jd > 4M bytes", 2899 (intmax_t)rsrc->compressed_size); 2900 return (ARCHIVE_WARN); 2901 } 2902 2903 metadata = malloc((size_t)rsrc->uncompressed_size); 2904 if (metadata == NULL) { 2905 archive_set_error(&a->archive, ENOMEM, 2906 "Can't allocate memory for Mac metadata"); 2907 return (ARCHIVE_FATAL); 2908 } 2909 2910 if (offset < rsrc->local_header_offset) 2911 __archive_read_consume(a, rsrc->local_header_offset - offset); 2912 else if (offset != rsrc->local_header_offset) { 2913 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2914 } 2915 2916 hsize = zip_get_local_file_header_size(a, 0); 2917 __archive_read_consume(a, hsize); 2918 2919 remaining_bytes = (size_t)rsrc->compressed_size; 2920 metadata_bytes = (size_t)rsrc->uncompressed_size; 2921 mp = metadata; 2922 eof = 0; 2923 while (!eof && remaining_bytes) { 2924 const unsigned char *p; 2925 ssize_t bytes_avail; 2926 size_t bytes_used; 2927 2928 p = __archive_read_ahead(a, 1, &bytes_avail); 2929 if (p == NULL) { 2930 archive_set_error(&a->archive, 2931 ARCHIVE_ERRNO_FILE_FORMAT, 2932 "Truncated ZIP file header"); 2933 ret = ARCHIVE_WARN; 2934 goto exit_mac_metadata; 2935 } 2936 if ((size_t)bytes_avail > remaining_bytes) 2937 bytes_avail = remaining_bytes; 2938 switch(rsrc->compression) { 2939 case 0: /* No compression. */ 2940 if ((size_t)bytes_avail > metadata_bytes) 2941 bytes_avail = metadata_bytes; 2942 memcpy(mp, p, bytes_avail); 2943 bytes_used = (size_t)bytes_avail; 2944 metadata_bytes -= bytes_used; 2945 mp += bytes_used; 2946 if (metadata_bytes == 0) 2947 eof = 1; 2948 break; 2949#ifdef HAVE_ZLIB_H 2950 case 8: /* Deflate compression. */ 2951 { 2952 int r; 2953 2954 ret = zip_deflate_init(a, zip); 2955 if (ret != ARCHIVE_OK) 2956 goto exit_mac_metadata; 2957 zip->stream.next_in = 2958 (Bytef *)(uintptr_t)(const void *)p; 2959 zip->stream.avail_in = (uInt)bytes_avail; 2960 zip->stream.total_in = 0; 2961 zip->stream.next_out = mp; 2962 zip->stream.avail_out = (uInt)metadata_bytes; 2963 zip->stream.total_out = 0; 2964 2965 r = inflate(&zip->stream, 0); 2966 switch (r) { 2967 case Z_OK: 2968 break; 2969 case Z_STREAM_END: 2970 eof = 1; 2971 break; 2972 case Z_MEM_ERROR: 2973 archive_set_error(&a->archive, ENOMEM, 2974 "Out of memory for ZIP decompression"); 2975 ret = ARCHIVE_FATAL; 2976 goto exit_mac_metadata; 2977 default: 2978 archive_set_error(&a->archive, 2979 ARCHIVE_ERRNO_MISC, 2980 "ZIP decompression failed (%d)", r); 2981 ret = ARCHIVE_FATAL; 2982 goto exit_mac_metadata; 2983 } 2984 bytes_used = zip->stream.total_in; 2985 metadata_bytes -= zip->stream.total_out; 2986 mp += zip->stream.total_out; 2987 break; 2988 } 2989#endif 2990 default: 2991 bytes_used = 0; 2992 break; 2993 } 2994 __archive_read_consume(a, bytes_used); 2995 remaining_bytes -= bytes_used; 2996 } 2997 archive_entry_copy_mac_metadata(entry, metadata, 2998 (size_t)rsrc->uncompressed_size - metadata_bytes); 2999 3000exit_mac_metadata: 3001 __archive_read_seek(a, offset, SEEK_SET); 3002 zip->decompress_init = 0; 3003 free(metadata); 3004 return (ret); 3005} 3006 3007static int 3008archive_read_format_zip_seekable_read_header(struct archive_read *a, 3009 struct archive_entry *entry) 3010{ 3011 struct zip *zip = (struct zip *)a->format->data; 3012 struct zip_entry *rsrc; 3013 int64_t offset; 3014 int r, ret = ARCHIVE_OK; 3015 3016 /* 3017 * It should be sufficient to call archive_read_next_header() for 3018 * a reader to determine if an entry is encrypted or not. If the 3019 * encryption of an entry is only detectable when calling 3020 * archive_read_data(), so be it. We'll do the same check there 3021 * as well. 3022 */ 3023 if (zip->has_encrypted_entries == 3024 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 3025 zip->has_encrypted_entries = 0; 3026 3027 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 3028 if (a->archive.archive_format_name == NULL) 3029 a->archive.archive_format_name = "ZIP"; 3030 3031 if (zip->zip_entries == NULL) { 3032 r = slurp_central_directory(a, zip); 3033 if (r != ARCHIVE_OK) 3034 return r; 3035 /* Get first entry whose local header offset is lower than 3036 * other entries in the archive file. */ 3037 zip->entry = 3038 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 3039 } else if (zip->entry != NULL) { 3040 /* Get next entry in local header offset order. */ 3041 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 3042 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 3043 } 3044 3045 if (zip->entry == NULL) 3046 return ARCHIVE_EOF; 3047 3048 if (zip->entry->rsrcname.s) 3049 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 3050 &zip->tree_rsrc, zip->entry->rsrcname.s); 3051 else 3052 rsrc = NULL; 3053 3054 if (zip->cctx_valid) 3055 archive_decrypto_aes_ctr_release(&zip->cctx); 3056 if (zip->hctx_valid) 3057 archive_hmac_sha1_cleanup(&zip->hctx); 3058 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3059 __archive_read_reset_passphrase(a); 3060 3061 /* File entries are sorted by the header offset, we should mostly 3062 * use __archive_read_consume to advance a read point to avoid redundant 3063 * data reading. */ 3064 offset = archive_filter_bytes(&a->archive, 0); 3065 if (offset < zip->entry->local_header_offset) 3066 __archive_read_consume(a, 3067 zip->entry->local_header_offset - offset); 3068 else if (offset != zip->entry->local_header_offset) { 3069 __archive_read_seek(a, zip->entry->local_header_offset, 3070 SEEK_SET); 3071 } 3072 zip->unconsumed = 0; 3073 r = zip_read_local_file_header(a, entry, zip); 3074 if (r != ARCHIVE_OK) 3075 return r; 3076 if (rsrc) { 3077 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3078 if (ret2 < ret) 3079 ret = ret2; 3080 } 3081 return (ret); 3082} 3083 3084/* 3085 * We're going to seek for the next header anyway, so we don't 3086 * need to bother doing anything here. 3087 */ 3088static int 3089archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3090{ 3091 struct zip *zip; 3092 zip = (struct zip *)(a->format->data); 3093 3094 zip->unconsumed = 0; 3095 return (ARCHIVE_OK); 3096} 3097 3098int 3099archive_read_support_format_zip_seekable(struct archive *_a) 3100{ 3101 struct archive_read *a = (struct archive_read *)_a; 3102 struct zip *zip; 3103 int r; 3104 3105 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3106 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3107 3108 zip = (struct zip *)calloc(1, sizeof(*zip)); 3109 if (zip == NULL) { 3110 archive_set_error(&a->archive, ENOMEM, 3111 "Can't allocate zip data"); 3112 return (ARCHIVE_FATAL); 3113 } 3114 3115#ifdef HAVE_COPYFILE_H 3116 /* Set this by default on Mac OS. */ 3117 zip->process_mac_extensions = 1; 3118#endif 3119 3120 /* 3121 * Until enough data has been read, we cannot tell about 3122 * any encrypted entries yet. 3123 */ 3124 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3125 zip->crc32func = real_crc32; 3126 3127 r = __archive_read_register_format(a, 3128 zip, 3129 "zip", 3130 archive_read_format_zip_seekable_bid, 3131 archive_read_format_zip_options, 3132 archive_read_format_zip_seekable_read_header, 3133 archive_read_format_zip_read_data, 3134 archive_read_format_zip_read_data_skip_seekable, 3135 NULL, 3136 archive_read_format_zip_cleanup, 3137 archive_read_support_format_zip_capabilities_seekable, 3138 archive_read_format_zip_has_encrypted_entries); 3139 3140 if (r != ARCHIVE_OK) 3141 free(zip); 3142 return (ARCHIVE_OK); 3143} 3144