archive_read_support_format_zip.c revision 328828
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/10/contrib/libarchive/libarchive/archive_read_support_format_zip.c 328828 2018-02-03 02:17:25Z 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 = p[offset]; 515 offset++; 516 datasize--; 517 /* Flag bits indicate which dates are present. */ 518 if (flags & 0x01) 519 { 520#ifdef DEBUG 521 fprintf(stderr, "mtime: %lld -> %d\n", 522 (long long)zip_entry->mtime, 523 archive_le32dec(p + offset)); 524#endif 525 if (datasize < 4) 526 break; 527 zip_entry->mtime = archive_le32dec(p + offset); 528 offset += 4; 529 datasize -= 4; 530 } 531 if (flags & 0x02) 532 { 533 if (datasize < 4) 534 break; 535 zip_entry->atime = archive_le32dec(p + offset); 536 offset += 4; 537 datasize -= 4; 538 } 539 if (flags & 0x04) 540 { 541 if (datasize < 4) 542 break; 543 zip_entry->ctime = archive_le32dec(p + offset); 544 offset += 4; 545 datasize -= 4; 546 } 547 break; 548 } 549 case 0x5855: 550 { 551 /* Info-ZIP Unix Extra Field (old version) "UX". */ 552 if (datasize >= 8) { 553 zip_entry->atime = archive_le32dec(p + offset); 554 zip_entry->mtime = 555 archive_le32dec(p + offset + 4); 556 } 557 if (datasize >= 12) { 558 zip_entry->uid = 559 archive_le16dec(p + offset + 8); 560 zip_entry->gid = 561 archive_le16dec(p + offset + 10); 562 } 563 break; 564 } 565 case 0x6c78: 566 { 567 /* Experimental 'xl' field */ 568 /* 569 * Introduced Dec 2013 to provide a way to 570 * include external file attributes (and other 571 * fields that ordinarily appear only in 572 * central directory) in local file header. 573 * This provides file type and permission 574 * information necessary to support full 575 * streaming extraction. Currently being 576 * discussed with other Zip developers 577 * ... subject to change. 578 * 579 * Format: 580 * The field starts with a bitmap that specifies 581 * which additional fields are included. The 582 * bitmap is variable length and can be extended in 583 * the future. 584 * 585 * n bytes - feature bitmap: first byte has low-order 586 * 7 bits. If high-order bit is set, a subsequent 587 * byte holds the next 7 bits, etc. 588 * 589 * if bitmap & 1, 2 byte "version made by" 590 * if bitmap & 2, 2 byte "internal file attributes" 591 * if bitmap & 4, 4 byte "external file attributes" 592 * if bitmap & 8, 2 byte comment length + n byte comment 593 */ 594 int bitmap, bitmap_last; 595 596 if (datasize < 1) 597 break; 598 bitmap_last = bitmap = 0xff & p[offset]; 599 offset += 1; 600 datasize -= 1; 601 602 /* We only support first 7 bits of bitmap; skip rest. */ 603 while ((bitmap_last & 0x80) != 0 604 && datasize >= 1) { 605 bitmap_last = p[offset]; 606 offset += 1; 607 datasize -= 1; 608 } 609 610 if (bitmap & 1) { 611 /* 2 byte "version made by" */ 612 if (datasize < 2) 613 break; 614 zip_entry->system 615 = archive_le16dec(p + offset) >> 8; 616 offset += 2; 617 datasize -= 2; 618 } 619 if (bitmap & 2) { 620 /* 2 byte "internal file attributes" */ 621 uint32_t internal_attributes; 622 if (datasize < 2) 623 break; 624 internal_attributes 625 = archive_le16dec(p + offset); 626 /* Not used by libarchive at present. */ 627 (void)internal_attributes; /* UNUSED */ 628 offset += 2; 629 datasize -= 2; 630 } 631 if (bitmap & 4) { 632 /* 4 byte "external file attributes" */ 633 uint32_t external_attributes; 634 if (datasize < 4) 635 break; 636 external_attributes 637 = archive_le32dec(p + offset); 638 if (zip_entry->system == 3) { 639 zip_entry->mode 640 = external_attributes >> 16; 641 } else if (zip_entry->system == 0) { 642 // Interpret MSDOS directory bit 643 if (0x10 == (external_attributes & 0x10)) { 644 zip_entry->mode = AE_IFDIR | 0775; 645 } else { 646 zip_entry->mode = AE_IFREG | 0664; 647 } 648 if (0x01 == (external_attributes & 0x01)) { 649 // Read-only bit; strip write permissions 650 zip_entry->mode &= 0555; 651 } 652 } else { 653 zip_entry->mode = 0; 654 } 655 offset += 4; 656 datasize -= 4; 657 } 658 if (bitmap & 8) { 659 /* 2 byte comment length + comment */ 660 uint32_t comment_length; 661 if (datasize < 2) 662 break; 663 comment_length 664 = archive_le16dec(p + offset); 665 offset += 2; 666 datasize -= 2; 667 668 if (datasize < comment_length) 669 break; 670 /* Comment is not supported by libarchive */ 671 offset += comment_length; 672 datasize -= comment_length; 673 } 674 break; 675 } 676 case 0x7855: 677 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 678#ifdef DEBUG 679 fprintf(stderr, "uid %d gid %d\n", 680 archive_le16dec(p + offset), 681 archive_le16dec(p + offset + 2)); 682#endif 683 if (datasize >= 2) 684 zip_entry->uid = archive_le16dec(p + offset); 685 if (datasize >= 4) 686 zip_entry->gid = 687 archive_le16dec(p + offset + 2); 688 break; 689 case 0x7875: 690 { 691 /* Info-Zip Unix Extra Field (type 3) "ux". */ 692 int uidsize = 0, gidsize = 0; 693 694 /* TODO: support arbitrary uidsize/gidsize. */ 695 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 696 if (datasize >= 4) { 697 /* get a uid size. */ 698 uidsize = 0xff & (int)p[offset+1]; 699 if (uidsize == 2) 700 zip_entry->uid = 701 archive_le16dec( 702 p + offset + 2); 703 else if (uidsize == 4 && datasize >= 6) 704 zip_entry->uid = 705 archive_le32dec( 706 p + offset + 2); 707 } 708 if (datasize >= (2 + uidsize + 3)) { 709 /* get a gid size. */ 710 gidsize = 0xff & (int)p[offset+2+uidsize]; 711 if (gidsize == 2) 712 zip_entry->gid = 713 archive_le16dec( 714 p+offset+2+uidsize+1); 715 else if (gidsize == 4 && 716 datasize >= (2 + uidsize + 5)) 717 zip_entry->gid = 718 archive_le32dec( 719 p+offset+2+uidsize+1); 720 } 721 } 722 break; 723 } 724 case 0x9901: 725 /* WinZip AES extra data field. */ 726 if (datasize < 6) { 727 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 728 "Incomplete AES field"); 729 return ARCHIVE_FAILED; 730 } 731 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 732 /* Vendor version. */ 733 zip_entry->aes_extra.vendor = 734 archive_le16dec(p + offset); 735 /* AES encryption strength. */ 736 zip_entry->aes_extra.strength = p[offset + 4]; 737 /* Actual compression method. */ 738 zip_entry->aes_extra.compression = 739 p[offset + 5]; 740 } 741 break; 742 default: 743 break; 744 } 745 offset += datasize; 746 } 747 if (offset != extra_length) { 748 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 749 "Malformed extra data: Consumed %d bytes of %d bytes", 750 (int)offset, (int)extra_length); 751 return ARCHIVE_FAILED; 752 } 753 return ARCHIVE_OK; 754} 755 756/* 757 * Assumes file pointer is at beginning of local file header. 758 */ 759static int 760zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 761 struct zip *zip) 762{ 763 const char *p; 764 const void *h; 765 const wchar_t *wp; 766 const char *cp; 767 size_t len, filename_length, extra_length; 768 struct archive_string_conv *sconv; 769 struct zip_entry *zip_entry = zip->entry; 770 struct zip_entry zip_entry_central_dir; 771 int ret = ARCHIVE_OK; 772 char version; 773 774 /* Save a copy of the original for consistency checks. */ 775 zip_entry_central_dir = *zip_entry; 776 777 zip->decompress_init = 0; 778 zip->end_of_entry = 0; 779 zip->entry_uncompressed_bytes_read = 0; 780 zip->entry_compressed_bytes_read = 0; 781 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 782 783 /* Setup default conversion. */ 784 if (zip->sconv == NULL && !zip->init_default_conversion) { 785 zip->sconv_default = 786 archive_string_default_conversion_for_read(&(a->archive)); 787 zip->init_default_conversion = 1; 788 } 789 790 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 791 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 792 "Truncated ZIP file header"); 793 return (ARCHIVE_FATAL); 794 } 795 796 if (memcmp(p, "PK\003\004", 4) != 0) { 797 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 798 return ARCHIVE_FATAL; 799 } 800 version = p[4]; 801 zip_entry->system = p[5]; 802 zip_entry->zip_flags = archive_le16dec(p + 6); 803 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 804 zip->has_encrypted_entries = 1; 805 archive_entry_set_is_data_encrypted(entry, 1); 806 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 807 zip_entry->zip_flags & ZIP_ENCRYPTED && 808 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 809 archive_entry_set_is_metadata_encrypted(entry, 1); 810 return ARCHIVE_FATAL; 811 } 812 } 813 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 814 zip_entry->compression = (char)archive_le16dec(p + 8); 815 zip_entry->mtime = zip_time(p + 10); 816 zip_entry->crc32 = archive_le32dec(p + 14); 817 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 818 zip_entry->decdat = p[11]; 819 else 820 zip_entry->decdat = p[17]; 821 zip_entry->compressed_size = archive_le32dec(p + 18); 822 zip_entry->uncompressed_size = archive_le32dec(p + 22); 823 filename_length = archive_le16dec(p + 26); 824 extra_length = archive_le16dec(p + 28); 825 826 __archive_read_consume(a, 30); 827 828 /* Read the filename. */ 829 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 830 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 831 "Truncated ZIP file header"); 832 return (ARCHIVE_FATAL); 833 } 834 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 835 /* The filename is stored to be UTF-8. */ 836 if (zip->sconv_utf8 == NULL) { 837 zip->sconv_utf8 = 838 archive_string_conversion_from_charset( 839 &a->archive, "UTF-8", 1); 840 if (zip->sconv_utf8 == NULL) 841 return (ARCHIVE_FATAL); 842 } 843 sconv = zip->sconv_utf8; 844 } else if (zip->sconv != NULL) 845 sconv = zip->sconv; 846 else 847 sconv = zip->sconv_default; 848 849 if (archive_entry_copy_pathname_l(entry, 850 h, filename_length, sconv) != 0) { 851 if (errno == ENOMEM) { 852 archive_set_error(&a->archive, ENOMEM, 853 "Can't allocate memory for Pathname"); 854 return (ARCHIVE_FATAL); 855 } 856 archive_set_error(&a->archive, 857 ARCHIVE_ERRNO_FILE_FORMAT, 858 "Pathname cannot be converted " 859 "from %s to current locale.", 860 archive_string_conversion_charset_name(sconv)); 861 ret = ARCHIVE_WARN; 862 } 863 __archive_read_consume(a, filename_length); 864 865 /* Read the extra data. */ 866 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 867 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 868 "Truncated ZIP file header"); 869 return (ARCHIVE_FATAL); 870 } 871 872 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 873 return ARCHIVE_FATAL; 874 } 875 __archive_read_consume(a, extra_length); 876 877 /* Work around a bug in Info-Zip: When reading from a pipe, it 878 * stats the pipe instead of synthesizing a file entry. */ 879 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 880 zip_entry->mode &= ~ AE_IFMT; 881 zip_entry->mode |= AE_IFREG; 882 } 883 884 /* If the mode is totally empty, set some sane default. */ 885 if (zip_entry->mode == 0) { 886 zip_entry->mode |= 0664; 887 } 888 889 /* Windows archivers sometimes use backslash as the directory separator. 890 Normalize to slash. */ 891 if (zip_entry->system == 0 && 892 (wp = archive_entry_pathname_w(entry)) != NULL) { 893 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { 894 size_t i; 895 struct archive_wstring s; 896 archive_string_init(&s); 897 archive_wstrcpy(&s, wp); 898 for (i = 0; i < archive_strlen(&s); i++) { 899 if (s.s[i] == '\\') 900 s.s[i] = '/'; 901 } 902 archive_entry_copy_pathname_w(entry, s.s); 903 archive_wstring_free(&s); 904 } 905 } 906 907 /* Make sure that entries with a trailing '/' are marked as directories 908 * even if the External File Attributes contains bogus values. If this 909 * is not a directory and there is no type, assume regularfile. */ 910 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 911 int has_slash; 912 913 wp = archive_entry_pathname_w(entry); 914 if (wp != NULL) { 915 len = wcslen(wp); 916 has_slash = len > 0 && wp[len - 1] == L'/'; 917 } else { 918 cp = archive_entry_pathname(entry); 919 len = (cp != NULL)?strlen(cp):0; 920 has_slash = len > 0 && cp[len - 1] == '/'; 921 } 922 /* Correct file type as needed. */ 923 if (has_slash) { 924 zip_entry->mode &= ~AE_IFMT; 925 zip_entry->mode |= AE_IFDIR; 926 zip_entry->mode |= 0111; 927 } else if ((zip_entry->mode & AE_IFMT) == 0) { 928 zip_entry->mode |= AE_IFREG; 929 } 930 } 931 932 /* Make sure directories end in '/' */ 933 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 934 wp = archive_entry_pathname_w(entry); 935 if (wp != NULL) { 936 len = wcslen(wp); 937 if (len > 0 && wp[len - 1] != L'/') { 938 struct archive_wstring s; 939 archive_string_init(&s); 940 archive_wstrcat(&s, wp); 941 archive_wstrappend_wchar(&s, L'/'); 942 archive_entry_copy_pathname_w(entry, s.s); 943 archive_wstring_free(&s); 944 } 945 } else { 946 cp = archive_entry_pathname(entry); 947 len = (cp != NULL)?strlen(cp):0; 948 if (len > 0 && cp[len - 1] != '/') { 949 struct archive_string s; 950 archive_string_init(&s); 951 archive_strcat(&s, cp); 952 archive_strappend_char(&s, '/'); 953 archive_entry_set_pathname(entry, s.s); 954 archive_string_free(&s); 955 } 956 } 957 } 958 959 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 960 /* If this came from the central dir, it's size info 961 * is definitive, so ignore the length-at-end flag. */ 962 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 963 /* If local header is missing a value, use the one from 964 the central directory. If both have it, warn about 965 mismatches. */ 966 if (zip_entry->crc32 == 0) { 967 zip_entry->crc32 = zip_entry_central_dir.crc32; 968 } else if (!zip->ignore_crc32 969 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 970 archive_set_error(&a->archive, 971 ARCHIVE_ERRNO_FILE_FORMAT, 972 "Inconsistent CRC32 values"); 973 ret = ARCHIVE_WARN; 974 } 975 if (zip_entry->compressed_size == 0) { 976 zip_entry->compressed_size 977 = zip_entry_central_dir.compressed_size; 978 } else if (zip_entry->compressed_size 979 != zip_entry_central_dir.compressed_size) { 980 archive_set_error(&a->archive, 981 ARCHIVE_ERRNO_FILE_FORMAT, 982 "Inconsistent compressed size: " 983 "%jd in central directory, %jd in local header", 984 (intmax_t)zip_entry_central_dir.compressed_size, 985 (intmax_t)zip_entry->compressed_size); 986 ret = ARCHIVE_WARN; 987 } 988 if (zip_entry->uncompressed_size == 0) { 989 zip_entry->uncompressed_size 990 = zip_entry_central_dir.uncompressed_size; 991 } else if (zip_entry->uncompressed_size 992 != zip_entry_central_dir.uncompressed_size) { 993 archive_set_error(&a->archive, 994 ARCHIVE_ERRNO_FILE_FORMAT, 995 "Inconsistent uncompressed size: " 996 "%jd in central directory, %jd in local header", 997 (intmax_t)zip_entry_central_dir.uncompressed_size, 998 (intmax_t)zip_entry->uncompressed_size); 999 ret = ARCHIVE_WARN; 1000 } 1001 } 1002 1003 /* Populate some additional entry fields: */ 1004 archive_entry_set_mode(entry, zip_entry->mode); 1005 archive_entry_set_uid(entry, zip_entry->uid); 1006 archive_entry_set_gid(entry, zip_entry->gid); 1007 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 1008 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 1009 archive_entry_set_atime(entry, zip_entry->atime, 0); 1010 1011 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 1012 size_t linkname_length; 1013 1014 if (zip_entry->compressed_size > 64 * 1024) { 1015 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1016 "Zip file with oversized link entry"); 1017 return ARCHIVE_FATAL; 1018 } 1019 1020 linkname_length = (size_t)zip_entry->compressed_size; 1021 1022 archive_entry_set_size(entry, 0); 1023 p = __archive_read_ahead(a, linkname_length, NULL); 1024 if (p == NULL) { 1025 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1026 "Truncated Zip file"); 1027 return ARCHIVE_FATAL; 1028 } 1029 1030 sconv = zip->sconv; 1031 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1032 sconv = zip->sconv_utf8; 1033 if (sconv == NULL) 1034 sconv = zip->sconv_default; 1035 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 1036 sconv) != 0) { 1037 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1038 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1039 archive_entry_copy_symlink_l(entry, p, 1040 linkname_length, NULL); 1041 if (errno == ENOMEM) { 1042 archive_set_error(&a->archive, ENOMEM, 1043 "Can't allocate memory for Symlink"); 1044 return (ARCHIVE_FATAL); 1045 } 1046 /* 1047 * Since there is no character-set regulation for 1048 * symlink name, do not report the conversion error 1049 * in an automatic conversion. 1050 */ 1051 if (sconv != zip->sconv_utf8 || 1052 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1053 archive_set_error(&a->archive, 1054 ARCHIVE_ERRNO_FILE_FORMAT, 1055 "Symlink cannot be converted " 1056 "from %s to current locale.", 1057 archive_string_conversion_charset_name( 1058 sconv)); 1059 ret = ARCHIVE_WARN; 1060 } 1061 } 1062 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1063 1064 if (__archive_read_consume(a, linkname_length) < 0) { 1065 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1066 "Read error skipping symlink target name"); 1067 return ARCHIVE_FATAL; 1068 } 1069 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1070 || zip_entry->uncompressed_size > 0) { 1071 /* Set the size only if it's meaningful. */ 1072 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1073 } 1074 zip->entry_bytes_remaining = zip_entry->compressed_size; 1075 1076 /* If there's no body, force read_data() to return EOF immediately. */ 1077 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1078 && zip->entry_bytes_remaining < 1) 1079 zip->end_of_entry = 1; 1080 1081 /* Set up a more descriptive format name. */ 1082 archive_string_empty(&zip->format_name); 1083 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1084 version / 10, version % 10, 1085 compression_name(zip->entry->compression)); 1086 a->archive.archive_format_name = zip->format_name.s; 1087 1088 return (ret); 1089} 1090 1091static int 1092check_authentication_code(struct archive_read *a, const void *_p) 1093{ 1094 struct zip *zip = (struct zip *)(a->format->data); 1095 1096 /* Check authentication code. */ 1097 if (zip->hctx_valid) { 1098 const void *p; 1099 uint8_t hmac[20]; 1100 size_t hmac_len = 20; 1101 int cmp; 1102 1103 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1104 if (_p == NULL) { 1105 /* Read authentication code. */ 1106 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1107 if (p == NULL) { 1108 archive_set_error(&a->archive, 1109 ARCHIVE_ERRNO_FILE_FORMAT, 1110 "Truncated ZIP file data"); 1111 return (ARCHIVE_FATAL); 1112 } 1113 } else { 1114 p = _p; 1115 } 1116 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1117 __archive_read_consume(a, AUTH_CODE_SIZE); 1118 if (cmp != 0) { 1119 archive_set_error(&a->archive, 1120 ARCHIVE_ERRNO_MISC, 1121 "ZIP bad Authentication code"); 1122 return (ARCHIVE_WARN); 1123 } 1124 } 1125 return (ARCHIVE_OK); 1126} 1127 1128/* 1129 * Read "uncompressed" data. There are three cases: 1130 * 1) We know the size of the data. This is always true for the 1131 * seeking reader (we've examined the Central Directory already). 1132 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1133 * Info-ZIP seems to do this; we know the size but have to grab 1134 * the CRC from the data descriptor afterwards. 1135 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1136 * we have no size information. In this case, we can do pretty 1137 * well by watching for the data descriptor record. The data 1138 * descriptor is 16 bytes and includes a computed CRC that should 1139 * provide a strong check. 1140 * 1141 * TODO: Technically, the PK\007\010 signature is optional. 1142 * In the original spec, the data descriptor contained CRC 1143 * and size fields but had no leading signature. In practice, 1144 * newer writers seem to provide the signature pretty consistently. 1145 * 1146 * For uncompressed data, the PK\007\010 marker seems essential 1147 * to be sure we've actually seen the end of the entry. 1148 * 1149 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1150 * zip->end_of_entry if it consumes all of the data. 1151 */ 1152static int 1153zip_read_data_none(struct archive_read *a, const void **_buff, 1154 size_t *size, int64_t *offset) 1155{ 1156 struct zip *zip; 1157 const char *buff; 1158 ssize_t bytes_avail; 1159 int r; 1160 1161 (void)offset; /* UNUSED */ 1162 1163 zip = (struct zip *)(a->format->data); 1164 1165 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1166 const char *p; 1167 ssize_t grabbing_bytes = 24; 1168 1169 if (zip->hctx_valid) 1170 grabbing_bytes += AUTH_CODE_SIZE; 1171 /* Grab at least 24 bytes. */ 1172 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1173 if (bytes_avail < grabbing_bytes) { 1174 /* Zip archives have end-of-archive markers 1175 that are longer than this, so a failure to get at 1176 least 24 bytes really does indicate a truncated 1177 file. */ 1178 archive_set_error(&a->archive, 1179 ARCHIVE_ERRNO_FILE_FORMAT, 1180 "Truncated ZIP file data"); 1181 return (ARCHIVE_FATAL); 1182 } 1183 /* Check for a complete PK\007\010 signature, followed 1184 * by the correct 4-byte CRC. */ 1185 p = buff; 1186 if (zip->hctx_valid) 1187 p += AUTH_CODE_SIZE; 1188 if (p[0] == 'P' && p[1] == 'K' 1189 && p[2] == '\007' && p[3] == '\010' 1190 && (archive_le32dec(p + 4) == zip->entry_crc32 1191 || zip->ignore_crc32 1192 || (zip->hctx_valid 1193 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1194 if (zip->entry->flags & LA_USED_ZIP64) { 1195 uint64_t compressed, uncompressed; 1196 zip->entry->crc32 = archive_le32dec(p + 4); 1197 compressed = archive_le64dec(p + 8); 1198 uncompressed = archive_le64dec(p + 16); 1199 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1200 archive_set_error(&a->archive, 1201 ARCHIVE_ERRNO_FILE_FORMAT, 1202 "Overflow of 64-bit file sizes"); 1203 return ARCHIVE_FAILED; 1204 } 1205 zip->entry->compressed_size = compressed; 1206 zip->entry->uncompressed_size = uncompressed; 1207 zip->unconsumed = 24; 1208 } else { 1209 zip->entry->crc32 = archive_le32dec(p + 4); 1210 zip->entry->compressed_size = 1211 archive_le32dec(p + 8); 1212 zip->entry->uncompressed_size = 1213 archive_le32dec(p + 12); 1214 zip->unconsumed = 16; 1215 } 1216 if (zip->hctx_valid) { 1217 r = check_authentication_code(a, buff); 1218 if (r != ARCHIVE_OK) 1219 return (r); 1220 } 1221 zip->end_of_entry = 1; 1222 return (ARCHIVE_OK); 1223 } 1224 /* If not at EOF, ensure we consume at least one byte. */ 1225 ++p; 1226 1227 /* Scan forward until we see where a PK\007\010 signature 1228 * might be. */ 1229 /* Return bytes up until that point. On the next call, 1230 * the code above will verify the data descriptor. */ 1231 while (p < buff + bytes_avail - 4) { 1232 if (p[3] == 'P') { p += 3; } 1233 else if (p[3] == 'K') { p += 2; } 1234 else if (p[3] == '\007') { p += 1; } 1235 else if (p[3] == '\010' && p[2] == '\007' 1236 && p[1] == 'K' && p[0] == 'P') { 1237 if (zip->hctx_valid) 1238 p -= AUTH_CODE_SIZE; 1239 break; 1240 } else { p += 4; } 1241 } 1242 bytes_avail = p - buff; 1243 } else { 1244 if (zip->entry_bytes_remaining == 0) { 1245 zip->end_of_entry = 1; 1246 if (zip->hctx_valid) { 1247 r = check_authentication_code(a, NULL); 1248 if (r != ARCHIVE_OK) 1249 return (r); 1250 } 1251 return (ARCHIVE_OK); 1252 } 1253 /* Grab a bunch of bytes. */ 1254 buff = __archive_read_ahead(a, 1, &bytes_avail); 1255 if (bytes_avail <= 0) { 1256 archive_set_error(&a->archive, 1257 ARCHIVE_ERRNO_FILE_FORMAT, 1258 "Truncated ZIP file data"); 1259 return (ARCHIVE_FATAL); 1260 } 1261 if (bytes_avail > zip->entry_bytes_remaining) 1262 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1263 } 1264 if (zip->tctx_valid || zip->cctx_valid) { 1265 size_t dec_size = bytes_avail; 1266 1267 if (dec_size > zip->decrypted_buffer_size) 1268 dec_size = zip->decrypted_buffer_size; 1269 if (zip->tctx_valid) { 1270 trad_enc_decrypt_update(&zip->tctx, 1271 (const uint8_t *)buff, dec_size, 1272 zip->decrypted_buffer, dec_size); 1273 } else { 1274 size_t dsize = dec_size; 1275 archive_hmac_sha1_update(&zip->hctx, 1276 (const uint8_t *)buff, dec_size); 1277 archive_decrypto_aes_ctr_update(&zip->cctx, 1278 (const uint8_t *)buff, dec_size, 1279 zip->decrypted_buffer, &dsize); 1280 } 1281 bytes_avail = dec_size; 1282 buff = (const char *)zip->decrypted_buffer; 1283 } 1284 *size = bytes_avail; 1285 zip->entry_bytes_remaining -= bytes_avail; 1286 zip->entry_uncompressed_bytes_read += bytes_avail; 1287 zip->entry_compressed_bytes_read += bytes_avail; 1288 zip->unconsumed += bytes_avail; 1289 *_buff = buff; 1290 return (ARCHIVE_OK); 1291} 1292 1293#ifdef HAVE_ZLIB_H 1294static int 1295zip_deflate_init(struct archive_read *a, struct zip *zip) 1296{ 1297 int r; 1298 1299 /* If we haven't yet read any data, initialize the decompressor. */ 1300 if (!zip->decompress_init) { 1301 if (zip->stream_valid) 1302 r = inflateReset(&zip->stream); 1303 else 1304 r = inflateInit2(&zip->stream, 1305 -15 /* Don't check for zlib header */); 1306 if (r != Z_OK) { 1307 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1308 "Can't initialize ZIP decompression."); 1309 return (ARCHIVE_FATAL); 1310 } 1311 /* Stream structure has been set up. */ 1312 zip->stream_valid = 1; 1313 /* We've initialized decompression for this stream. */ 1314 zip->decompress_init = 1; 1315 } 1316 return (ARCHIVE_OK); 1317} 1318 1319static int 1320zip_read_data_deflate(struct archive_read *a, const void **buff, 1321 size_t *size, int64_t *offset) 1322{ 1323 struct zip *zip; 1324 ssize_t bytes_avail; 1325 const void *compressed_buff, *sp; 1326 int r; 1327 1328 (void)offset; /* UNUSED */ 1329 1330 zip = (struct zip *)(a->format->data); 1331 1332 /* If the buffer hasn't been allocated, allocate it now. */ 1333 if (zip->uncompressed_buffer == NULL) { 1334 zip->uncompressed_buffer_size = 256 * 1024; 1335 zip->uncompressed_buffer 1336 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1337 if (zip->uncompressed_buffer == NULL) { 1338 archive_set_error(&a->archive, ENOMEM, 1339 "No memory for ZIP decompression"); 1340 return (ARCHIVE_FATAL); 1341 } 1342 } 1343 1344 r = zip_deflate_init(a, zip); 1345 if (r != ARCHIVE_OK) 1346 return (r); 1347 1348 /* 1349 * Note: '1' here is a performance optimization. 1350 * Recall that the decompression layer returns a count of 1351 * available bytes; asking for more than that forces the 1352 * decompressor to combine reads by copying data. 1353 */ 1354 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1355 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1356 && bytes_avail > zip->entry_bytes_remaining) { 1357 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1358 } 1359 if (bytes_avail < 0) { 1360 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1361 "Truncated ZIP file body"); 1362 return (ARCHIVE_FATAL); 1363 } 1364 1365 if (zip->tctx_valid || zip->cctx_valid) { 1366 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1367 size_t buff_remaining = 1368 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1369 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1370 1371 if (buff_remaining > (size_t)bytes_avail) 1372 buff_remaining = (size_t)bytes_avail; 1373 1374 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1375 zip->entry_bytes_remaining > 0) { 1376 if ((int64_t)(zip->decrypted_bytes_remaining 1377 + buff_remaining) 1378 > zip->entry_bytes_remaining) { 1379 if (zip->entry_bytes_remaining < 1380 (int64_t)zip->decrypted_bytes_remaining) 1381 buff_remaining = 0; 1382 else 1383 buff_remaining = 1384 (size_t)zip->entry_bytes_remaining 1385 - zip->decrypted_bytes_remaining; 1386 } 1387 } 1388 if (buff_remaining > 0) { 1389 if (zip->tctx_valid) { 1390 trad_enc_decrypt_update(&zip->tctx, 1391 compressed_buff, buff_remaining, 1392 zip->decrypted_ptr 1393 + zip->decrypted_bytes_remaining, 1394 buff_remaining); 1395 } else { 1396 size_t dsize = buff_remaining; 1397 archive_decrypto_aes_ctr_update( 1398 &zip->cctx, 1399 compressed_buff, buff_remaining, 1400 zip->decrypted_ptr 1401 + zip->decrypted_bytes_remaining, 1402 &dsize); 1403 } 1404 zip->decrypted_bytes_remaining += buff_remaining; 1405 } 1406 } 1407 bytes_avail = zip->decrypted_bytes_remaining; 1408 compressed_buff = (const char *)zip->decrypted_ptr; 1409 } 1410 1411 /* 1412 * A bug in zlib.h: stream.next_in should be marked 'const' 1413 * but isn't (the library never alters data through the 1414 * next_in pointer, only reads it). The result: this ugly 1415 * cast to remove 'const'. 1416 */ 1417 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1418 zip->stream.avail_in = (uInt)bytes_avail; 1419 zip->stream.total_in = 0; 1420 zip->stream.next_out = zip->uncompressed_buffer; 1421 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1422 zip->stream.total_out = 0; 1423 1424 r = inflate(&zip->stream, 0); 1425 switch (r) { 1426 case Z_OK: 1427 break; 1428 case Z_STREAM_END: 1429 zip->end_of_entry = 1; 1430 break; 1431 case Z_MEM_ERROR: 1432 archive_set_error(&a->archive, ENOMEM, 1433 "Out of memory for ZIP decompression"); 1434 return (ARCHIVE_FATAL); 1435 default: 1436 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1437 "ZIP decompression failed (%d)", r); 1438 return (ARCHIVE_FATAL); 1439 } 1440 1441 /* Consume as much as the compressor actually used. */ 1442 bytes_avail = zip->stream.total_in; 1443 if (zip->tctx_valid || zip->cctx_valid) { 1444 zip->decrypted_bytes_remaining -= bytes_avail; 1445 if (zip->decrypted_bytes_remaining == 0) 1446 zip->decrypted_ptr = zip->decrypted_buffer; 1447 else 1448 zip->decrypted_ptr += bytes_avail; 1449 } 1450 /* Calculate compressed data as much as we used.*/ 1451 if (zip->hctx_valid) 1452 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1453 __archive_read_consume(a, bytes_avail); 1454 zip->entry_bytes_remaining -= bytes_avail; 1455 zip->entry_compressed_bytes_read += bytes_avail; 1456 1457 *size = zip->stream.total_out; 1458 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1459 *buff = zip->uncompressed_buffer; 1460 1461 if (zip->end_of_entry && zip->hctx_valid) { 1462 r = check_authentication_code(a, NULL); 1463 if (r != ARCHIVE_OK) 1464 return (r); 1465 } 1466 1467 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1468 const char *p; 1469 1470 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1471 archive_set_error(&a->archive, 1472 ARCHIVE_ERRNO_FILE_FORMAT, 1473 "Truncated ZIP end-of-file record"); 1474 return (ARCHIVE_FATAL); 1475 } 1476 /* Consume the optional PK\007\010 marker. */ 1477 if (p[0] == 'P' && p[1] == 'K' && 1478 p[2] == '\007' && p[3] == '\010') { 1479 p += 4; 1480 zip->unconsumed = 4; 1481 } 1482 if (zip->entry->flags & LA_USED_ZIP64) { 1483 uint64_t compressed, uncompressed; 1484 zip->entry->crc32 = archive_le32dec(p); 1485 compressed = archive_le64dec(p + 4); 1486 uncompressed = archive_le64dec(p + 12); 1487 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1488 archive_set_error(&a->archive, 1489 ARCHIVE_ERRNO_FILE_FORMAT, 1490 "Overflow of 64-bit file sizes"); 1491 return ARCHIVE_FAILED; 1492 } 1493 zip->entry->compressed_size = compressed; 1494 zip->entry->uncompressed_size = uncompressed; 1495 zip->unconsumed += 20; 1496 } else { 1497 zip->entry->crc32 = archive_le32dec(p); 1498 zip->entry->compressed_size = archive_le32dec(p + 4); 1499 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1500 zip->unconsumed += 12; 1501 } 1502 } 1503 1504 return (ARCHIVE_OK); 1505} 1506#endif 1507 1508static int 1509read_decryption_header(struct archive_read *a) 1510{ 1511 struct zip *zip = (struct zip *)(a->format->data); 1512 const char *p; 1513 unsigned int remaining_size; 1514 unsigned int ts; 1515 1516 /* 1517 * Read an initialization vector data field. 1518 */ 1519 p = __archive_read_ahead(a, 2, NULL); 1520 if (p == NULL) 1521 goto truncated; 1522 ts = zip->iv_size; 1523 zip->iv_size = archive_le16dec(p); 1524 __archive_read_consume(a, 2); 1525 if (ts < zip->iv_size) { 1526 free(zip->iv); 1527 zip->iv = NULL; 1528 } 1529 p = __archive_read_ahead(a, zip->iv_size, NULL); 1530 if (p == NULL) 1531 goto truncated; 1532 if (zip->iv == NULL) { 1533 zip->iv = malloc(zip->iv_size); 1534 if (zip->iv == NULL) 1535 goto nomem; 1536 } 1537 memcpy(zip->iv, p, zip->iv_size); 1538 __archive_read_consume(a, zip->iv_size); 1539 1540 /* 1541 * Read a size of remaining decryption header field. 1542 */ 1543 p = __archive_read_ahead(a, 14, NULL); 1544 if (p == NULL) 1545 goto truncated; 1546 remaining_size = archive_le32dec(p); 1547 if (remaining_size < 16 || remaining_size > (1 << 18)) 1548 goto corrupted; 1549 1550 /* Check if format version is supported. */ 1551 if (archive_le16dec(p+4) != 3) { 1552 archive_set_error(&a->archive, 1553 ARCHIVE_ERRNO_FILE_FORMAT, 1554 "Unsupported encryption format version: %u", 1555 archive_le16dec(p+4)); 1556 return (ARCHIVE_FAILED); 1557 } 1558 1559 /* 1560 * Read an encryption algorithm field. 1561 */ 1562 zip->alg_id = archive_le16dec(p+6); 1563 switch (zip->alg_id) { 1564 case 0x6601:/* DES */ 1565 case 0x6602:/* RC2 */ 1566 case 0x6603:/* 3DES 168 */ 1567 case 0x6609:/* 3DES 112 */ 1568 case 0x660E:/* AES 128 */ 1569 case 0x660F:/* AES 192 */ 1570 case 0x6610:/* AES 256 */ 1571 case 0x6702:/* RC2 (version >= 5.2) */ 1572 case 0x6720:/* Blowfish */ 1573 case 0x6721:/* Twofish */ 1574 case 0x6801:/* RC4 */ 1575 /* Supported encryption algorithm. */ 1576 break; 1577 default: 1578 archive_set_error(&a->archive, 1579 ARCHIVE_ERRNO_FILE_FORMAT, 1580 "Unknown encryption algorithm: %u", zip->alg_id); 1581 return (ARCHIVE_FAILED); 1582 } 1583 1584 /* 1585 * Read a bit length field. 1586 */ 1587 zip->bit_len = archive_le16dec(p+8); 1588 1589 /* 1590 * Read a flags field. 1591 */ 1592 zip->flags = archive_le16dec(p+10); 1593 switch (zip->flags & 0xf000) { 1594 case 0x0001: /* Password is required to decrypt. */ 1595 case 0x0002: /* Certificates only. */ 1596 case 0x0003: /* Password or certificate required to decrypt. */ 1597 break; 1598 default: 1599 archive_set_error(&a->archive, 1600 ARCHIVE_ERRNO_FILE_FORMAT, 1601 "Unknown encryption flag: %u", zip->flags); 1602 return (ARCHIVE_FAILED); 1603 } 1604 if ((zip->flags & 0xf000) == 0 || 1605 (zip->flags & 0xf000) == 0x4000) { 1606 archive_set_error(&a->archive, 1607 ARCHIVE_ERRNO_FILE_FORMAT, 1608 "Unknown encryption flag: %u", zip->flags); 1609 return (ARCHIVE_FAILED); 1610 } 1611 1612 /* 1613 * Read an encrypted random data field. 1614 */ 1615 ts = zip->erd_size; 1616 zip->erd_size = archive_le16dec(p+12); 1617 __archive_read_consume(a, 14); 1618 if ((zip->erd_size & 0xf) != 0 || 1619 (zip->erd_size + 16) > remaining_size || 1620 (zip->erd_size + 16) < zip->erd_size) 1621 goto corrupted; 1622 1623 if (ts < zip->erd_size) { 1624 free(zip->erd); 1625 zip->erd = NULL; 1626 } 1627 p = __archive_read_ahead(a, zip->erd_size, NULL); 1628 if (p == NULL) 1629 goto truncated; 1630 if (zip->erd == NULL) { 1631 zip->erd = malloc(zip->erd_size); 1632 if (zip->erd == NULL) 1633 goto nomem; 1634 } 1635 memcpy(zip->erd, p, zip->erd_size); 1636 __archive_read_consume(a, zip->erd_size); 1637 1638 /* 1639 * Read a reserved data field. 1640 */ 1641 p = __archive_read_ahead(a, 4, NULL); 1642 if (p == NULL) 1643 goto truncated; 1644 /* Reserved data size should be zero. */ 1645 if (archive_le32dec(p) != 0) 1646 goto corrupted; 1647 __archive_read_consume(a, 4); 1648 1649 /* 1650 * Read a password validation data field. 1651 */ 1652 p = __archive_read_ahead(a, 2, NULL); 1653 if (p == NULL) 1654 goto truncated; 1655 ts = zip->v_size; 1656 zip->v_size = archive_le16dec(p); 1657 __archive_read_consume(a, 2); 1658 if ((zip->v_size & 0x0f) != 0 || 1659 (zip->erd_size + zip->v_size + 16) > remaining_size || 1660 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1661 goto corrupted; 1662 if (ts < zip->v_size) { 1663 free(zip->v_data); 1664 zip->v_data = NULL; 1665 } 1666 p = __archive_read_ahead(a, zip->v_size, NULL); 1667 if (p == NULL) 1668 goto truncated; 1669 if (zip->v_data == NULL) { 1670 zip->v_data = malloc(zip->v_size); 1671 if (zip->v_data == NULL) 1672 goto nomem; 1673 } 1674 memcpy(zip->v_data, p, zip->v_size); 1675 __archive_read_consume(a, zip->v_size); 1676 1677 p = __archive_read_ahead(a, 4, NULL); 1678 if (p == NULL) 1679 goto truncated; 1680 zip->v_crc32 = archive_le32dec(p); 1681 __archive_read_consume(a, 4); 1682 1683 /*return (ARCHIVE_OK); 1684 * This is not fully implemented yet.*/ 1685 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1686 "Encrypted file is unsupported"); 1687 return (ARCHIVE_FAILED); 1688truncated: 1689 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1690 "Truncated ZIP file data"); 1691 return (ARCHIVE_FATAL); 1692corrupted: 1693 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1694 "Corrupted ZIP file data"); 1695 return (ARCHIVE_FATAL); 1696nomem: 1697 archive_set_error(&a->archive, ENOMEM, 1698 "No memory for ZIP decryption"); 1699 return (ARCHIVE_FATAL); 1700} 1701 1702static int 1703zip_alloc_decryption_buffer(struct archive_read *a) 1704{ 1705 struct zip *zip = (struct zip *)(a->format->data); 1706 size_t bs = 256 * 1024; 1707 1708 if (zip->decrypted_buffer == NULL) { 1709 zip->decrypted_buffer_size = bs; 1710 zip->decrypted_buffer = malloc(bs); 1711 if (zip->decrypted_buffer == NULL) { 1712 archive_set_error(&a->archive, ENOMEM, 1713 "No memory for ZIP decryption"); 1714 return (ARCHIVE_FATAL); 1715 } 1716 } 1717 zip->decrypted_ptr = zip->decrypted_buffer; 1718 return (ARCHIVE_OK); 1719} 1720 1721static int 1722init_traditional_PKWARE_decryption(struct archive_read *a) 1723{ 1724 struct zip *zip = (struct zip *)(a->format->data); 1725 const void *p; 1726 int retry; 1727 int r; 1728 1729 if (zip->tctx_valid) 1730 return (ARCHIVE_OK); 1731 1732 /* 1733 Read the 12 bytes encryption header stored at 1734 the start of the data area. 1735 */ 1736#define ENC_HEADER_SIZE 12 1737 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1738 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1739 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1740 "Truncated Zip encrypted body: only %jd bytes available", 1741 (intmax_t)zip->entry_bytes_remaining); 1742 return (ARCHIVE_FATAL); 1743 } 1744 1745 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1746 if (p == NULL) { 1747 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1748 "Truncated ZIP file data"); 1749 return (ARCHIVE_FATAL); 1750 } 1751 1752 for (retry = 0;; retry++) { 1753 const char *passphrase; 1754 uint8_t crcchk; 1755 1756 passphrase = __archive_read_next_passphrase(a); 1757 if (passphrase == NULL) { 1758 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1759 (retry > 0)? 1760 "Incorrect passphrase": 1761 "Passphrase required for this entry"); 1762 return (ARCHIVE_FAILED); 1763 } 1764 1765 /* 1766 * Initialize ctx for Traditional PKWARE Decryption. 1767 */ 1768 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1769 p, ENC_HEADER_SIZE, &crcchk); 1770 if (r == 0 && crcchk == zip->entry->decdat) 1771 break;/* The passphrase is OK. */ 1772 if (retry > 10000) { 1773 /* Avoid infinity loop. */ 1774 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1775 "Too many incorrect passphrases"); 1776 return (ARCHIVE_FAILED); 1777 } 1778 } 1779 1780 __archive_read_consume(a, ENC_HEADER_SIZE); 1781 zip->tctx_valid = 1; 1782 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1783 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1784 } 1785 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1786 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1787 zip->decrypted_bytes_remaining = 0; 1788 1789 return (zip_alloc_decryption_buffer(a)); 1790#undef ENC_HEADER_SIZE 1791} 1792 1793static int 1794init_WinZip_AES_decryption(struct archive_read *a) 1795{ 1796 struct zip *zip = (struct zip *)(a->format->data); 1797 const void *p; 1798 const uint8_t *pv; 1799 size_t key_len, salt_len; 1800 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1801 int retry; 1802 int r; 1803 1804 if (zip->cctx_valid || zip->hctx_valid) 1805 return (ARCHIVE_OK); 1806 1807 switch (zip->entry->aes_extra.strength) { 1808 case 1: salt_len = 8; key_len = 16; break; 1809 case 2: salt_len = 12; key_len = 24; break; 1810 case 3: salt_len = 16; key_len = 32; break; 1811 default: goto corrupted; 1812 } 1813 p = __archive_read_ahead(a, salt_len + 2, NULL); 1814 if (p == NULL) 1815 goto truncated; 1816 1817 for (retry = 0;; retry++) { 1818 const char *passphrase; 1819 1820 passphrase = __archive_read_next_passphrase(a); 1821 if (passphrase == NULL) { 1822 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1823 (retry > 0)? 1824 "Incorrect passphrase": 1825 "Passphrase required for this entry"); 1826 return (ARCHIVE_FAILED); 1827 } 1828 memset(derived_key, 0, sizeof(derived_key)); 1829 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1830 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1831 if (r != 0) { 1832 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1833 "Decryption is unsupported due to lack of " 1834 "crypto library"); 1835 return (ARCHIVE_FAILED); 1836 } 1837 1838 /* Check password verification value. */ 1839 pv = ((const uint8_t *)p) + salt_len; 1840 if (derived_key[key_len * 2] == pv[0] && 1841 derived_key[key_len * 2 + 1] == pv[1]) 1842 break;/* The passphrase is OK. */ 1843 if (retry > 10000) { 1844 /* Avoid infinity loop. */ 1845 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1846 "Too many incorrect passphrases"); 1847 return (ARCHIVE_FAILED); 1848 } 1849 } 1850 1851 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1852 if (r != 0) { 1853 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1854 "Decryption is unsupported due to lack of crypto library"); 1855 return (ARCHIVE_FAILED); 1856 } 1857 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1858 if (r != 0) { 1859 archive_decrypto_aes_ctr_release(&zip->cctx); 1860 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1861 "Failed to initialize HMAC-SHA1"); 1862 return (ARCHIVE_FAILED); 1863 } 1864 zip->cctx_valid = zip->hctx_valid = 1; 1865 __archive_read_consume(a, salt_len + 2); 1866 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1867 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1868 && zip->entry_bytes_remaining < 0) 1869 goto corrupted; 1870 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1871 zip->decrypted_bytes_remaining = 0; 1872 1873 zip->entry->compression = zip->entry->aes_extra.compression; 1874 return (zip_alloc_decryption_buffer(a)); 1875 1876truncated: 1877 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1878 "Truncated ZIP file data"); 1879 return (ARCHIVE_FATAL); 1880corrupted: 1881 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1882 "Corrupted ZIP file data"); 1883 return (ARCHIVE_FATAL); 1884} 1885 1886static int 1887archive_read_format_zip_read_data(struct archive_read *a, 1888 const void **buff, size_t *size, int64_t *offset) 1889{ 1890 int r; 1891 struct zip *zip = (struct zip *)(a->format->data); 1892 1893 if (zip->has_encrypted_entries == 1894 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1895 zip->has_encrypted_entries = 0; 1896 } 1897 1898 *offset = zip->entry_uncompressed_bytes_read; 1899 *size = 0; 1900 *buff = NULL; 1901 1902 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1903 if (zip->end_of_entry) 1904 return (ARCHIVE_EOF); 1905 1906 /* Return EOF immediately if this is a non-regular file. */ 1907 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1908 return (ARCHIVE_EOF); 1909 1910 __archive_read_consume(a, zip->unconsumed); 1911 zip->unconsumed = 0; 1912 1913 if (zip->init_decryption) { 1914 zip->has_encrypted_entries = 1; 1915 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1916 r = read_decryption_header(a); 1917 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1918 r = init_WinZip_AES_decryption(a); 1919 else 1920 r = init_traditional_PKWARE_decryption(a); 1921 if (r != ARCHIVE_OK) 1922 return (r); 1923 zip->init_decryption = 0; 1924 } 1925 1926 switch(zip->entry->compression) { 1927 case 0: /* No compression. */ 1928 r = zip_read_data_none(a, buff, size, offset); 1929 break; 1930#ifdef HAVE_ZLIB_H 1931 case 8: /* Deflate compression. */ 1932 r = zip_read_data_deflate(a, buff, size, offset); 1933 break; 1934#endif 1935 default: /* Unsupported compression. */ 1936 /* Return a warning. */ 1937 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1938 "Unsupported ZIP compression method (%s)", 1939 compression_name(zip->entry->compression)); 1940 /* We can't decompress this entry, but we will 1941 * be able to skip() it and try the next entry. */ 1942 return (ARCHIVE_FAILED); 1943 break; 1944 } 1945 if (r != ARCHIVE_OK) 1946 return (r); 1947 /* Update checksum */ 1948 if (*size) 1949 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1950 (unsigned)*size); 1951 /* If we hit the end, swallow any end-of-data marker. */ 1952 if (zip->end_of_entry) { 1953 /* Check file size, CRC against these values. */ 1954 if (zip->entry->compressed_size != 1955 zip->entry_compressed_bytes_read) { 1956 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1957 "ZIP compressed data is wrong size " 1958 "(read %jd, expected %jd)", 1959 (intmax_t)zip->entry_compressed_bytes_read, 1960 (intmax_t)zip->entry->compressed_size); 1961 return (ARCHIVE_WARN); 1962 } 1963 /* Size field only stores the lower 32 bits of the actual 1964 * size. */ 1965 if ((zip->entry->uncompressed_size & UINT32_MAX) 1966 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1967 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1968 "ZIP uncompressed data is wrong size " 1969 "(read %jd, expected %jd)\n", 1970 (intmax_t)zip->entry_uncompressed_bytes_read, 1971 (intmax_t)zip->entry->uncompressed_size); 1972 return (ARCHIVE_WARN); 1973 } 1974 /* Check computed CRC against header */ 1975 if ((!zip->hctx_valid || 1976 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1977 zip->entry->crc32 != zip->entry_crc32 1978 && !zip->ignore_crc32) { 1979 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1980 "ZIP bad CRC: 0x%lx should be 0x%lx", 1981 (unsigned long)zip->entry_crc32, 1982 (unsigned long)zip->entry->crc32); 1983 return (ARCHIVE_WARN); 1984 } 1985 } 1986 1987 return (ARCHIVE_OK); 1988} 1989 1990static int 1991archive_read_format_zip_cleanup(struct archive_read *a) 1992{ 1993 struct zip *zip; 1994 struct zip_entry *zip_entry, *next_zip_entry; 1995 1996 zip = (struct zip *)(a->format->data); 1997#ifdef HAVE_ZLIB_H 1998 if (zip->stream_valid) 1999 inflateEnd(&zip->stream); 2000 free(zip->uncompressed_buffer); 2001#endif 2002 if (zip->zip_entries) { 2003 zip_entry = zip->zip_entries; 2004 while (zip_entry != NULL) { 2005 next_zip_entry = zip_entry->next; 2006 archive_string_free(&zip_entry->rsrcname); 2007 free(zip_entry); 2008 zip_entry = next_zip_entry; 2009 } 2010 } 2011 free(zip->decrypted_buffer); 2012 if (zip->cctx_valid) 2013 archive_decrypto_aes_ctr_release(&zip->cctx); 2014 if (zip->hctx_valid) 2015 archive_hmac_sha1_cleanup(&zip->hctx); 2016 free(zip->iv); 2017 free(zip->erd); 2018 free(zip->v_data); 2019 archive_string_free(&zip->format_name); 2020 free(zip); 2021 (a->format->data) = NULL; 2022 return (ARCHIVE_OK); 2023} 2024 2025static int 2026archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 2027{ 2028 if (_a && _a->format) { 2029 struct zip * zip = (struct zip *)_a->format->data; 2030 if (zip) { 2031 return zip->has_encrypted_entries; 2032 } 2033 } 2034 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2035} 2036 2037static int 2038archive_read_format_zip_options(struct archive_read *a, 2039 const char *key, const char *val) 2040{ 2041 struct zip *zip; 2042 int ret = ARCHIVE_FAILED; 2043 2044 zip = (struct zip *)(a->format->data); 2045 if (strcmp(key, "compat-2x") == 0) { 2046 /* Handle filenames as libarchive 2.x */ 2047 zip->init_default_conversion = (val != NULL) ? 1 : 0; 2048 return (ARCHIVE_OK); 2049 } else if (strcmp(key, "hdrcharset") == 0) { 2050 if (val == NULL || val[0] == 0) 2051 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2052 "zip: hdrcharset option needs a character-set name" 2053 ); 2054 else { 2055 zip->sconv = archive_string_conversion_from_charset( 2056 &a->archive, val, 0); 2057 if (zip->sconv != NULL) { 2058 if (strcmp(val, "UTF-8") == 0) 2059 zip->sconv_utf8 = zip->sconv; 2060 ret = ARCHIVE_OK; 2061 } else 2062 ret = ARCHIVE_FATAL; 2063 } 2064 return (ret); 2065 } else if (strcmp(key, "ignorecrc32") == 0) { 2066 /* Mostly useful for testing. */ 2067 if (val == NULL || val[0] == 0) { 2068 zip->crc32func = real_crc32; 2069 zip->ignore_crc32 = 0; 2070 } else { 2071 zip->crc32func = fake_crc32; 2072 zip->ignore_crc32 = 1; 2073 } 2074 return (ARCHIVE_OK); 2075 } else if (strcmp(key, "mac-ext") == 0) { 2076 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2077 return (ARCHIVE_OK); 2078 } 2079 2080 /* Note: The "warn" return is just to inform the options 2081 * supervisor that we didn't handle it. It will generate 2082 * a suitable error if no one used this option. */ 2083 return (ARCHIVE_WARN); 2084} 2085 2086int 2087archive_read_support_format_zip(struct archive *a) 2088{ 2089 int r; 2090 r = archive_read_support_format_zip_streamable(a); 2091 if (r != ARCHIVE_OK) 2092 return r; 2093 return (archive_read_support_format_zip_seekable(a)); 2094} 2095 2096/* ------------------------------------------------------------------------ */ 2097 2098/* 2099 * Streaming-mode support 2100 */ 2101 2102 2103static int 2104archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2105{ 2106 (void)a; /* UNUSED */ 2107 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2108 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2109} 2110 2111static int 2112archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2113{ 2114 const char *p; 2115 2116 (void)best_bid; /* UNUSED */ 2117 2118 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2119 return (-1); 2120 2121 /* 2122 * Bid of 29 here comes from: 2123 * + 16 bits for "PK", 2124 * + next 16-bit field has 6 options so contributes 2125 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2126 * 2127 * So we've effectively verified ~29 total bits of check data. 2128 */ 2129 if (p[0] == 'P' && p[1] == 'K') { 2130 if ((p[2] == '\001' && p[3] == '\002') 2131 || (p[2] == '\003' && p[3] == '\004') 2132 || (p[2] == '\005' && p[3] == '\006') 2133 || (p[2] == '\006' && p[3] == '\006') 2134 || (p[2] == '\007' && p[3] == '\010') 2135 || (p[2] == '0' && p[3] == '0')) 2136 return (29); 2137 } 2138 2139 /* TODO: It's worth looking ahead a little bit for a valid 2140 * PK signature. In particular, that would make it possible 2141 * to read some UUEncoded SFX files or SFX files coming from 2142 * a network socket. */ 2143 2144 return (0); 2145} 2146 2147static int 2148archive_read_format_zip_streamable_read_header(struct archive_read *a, 2149 struct archive_entry *entry) 2150{ 2151 struct zip *zip; 2152 2153 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2154 if (a->archive.archive_format_name == NULL) 2155 a->archive.archive_format_name = "ZIP"; 2156 2157 zip = (struct zip *)(a->format->data); 2158 2159 /* 2160 * It should be sufficient to call archive_read_next_header() for 2161 * a reader to determine if an entry is encrypted or not. If the 2162 * encryption of an entry is only detectable when calling 2163 * archive_read_data(), so be it. We'll do the same check there 2164 * as well. 2165 */ 2166 if (zip->has_encrypted_entries == 2167 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2168 zip->has_encrypted_entries = 0; 2169 2170 /* Make sure we have a zip_entry structure to use. */ 2171 if (zip->zip_entries == NULL) { 2172 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2173 if (zip->zip_entries == NULL) { 2174 archive_set_error(&a->archive, ENOMEM, 2175 "Out of memory"); 2176 return ARCHIVE_FATAL; 2177 } 2178 } 2179 zip->entry = zip->zip_entries; 2180 memset(zip->entry, 0, sizeof(struct zip_entry)); 2181 2182 if (zip->cctx_valid) 2183 archive_decrypto_aes_ctr_release(&zip->cctx); 2184 if (zip->hctx_valid) 2185 archive_hmac_sha1_cleanup(&zip->hctx); 2186 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2187 __archive_read_reset_passphrase(a); 2188 2189 /* Search ahead for the next local file header. */ 2190 __archive_read_consume(a, zip->unconsumed); 2191 zip->unconsumed = 0; 2192 for (;;) { 2193 int64_t skipped = 0; 2194 const char *p, *end; 2195 ssize_t bytes; 2196 2197 p = __archive_read_ahead(a, 4, &bytes); 2198 if (p == NULL) 2199 return (ARCHIVE_FATAL); 2200 end = p + bytes; 2201 2202 while (p + 4 <= end) { 2203 if (p[0] == 'P' && p[1] == 'K') { 2204 if (p[2] == '\003' && p[3] == '\004') { 2205 /* Regular file entry. */ 2206 __archive_read_consume(a, skipped); 2207 return zip_read_local_file_header(a, 2208 entry, zip); 2209 } 2210 2211 /* 2212 * TODO: We cannot restore permissions 2213 * based only on the local file headers. 2214 * Consider scanning the central 2215 * directory and returning additional 2216 * entries for at least directories. 2217 * This would allow us to properly set 2218 * directory permissions. 2219 * 2220 * This won't help us fix symlinks 2221 * and may not help with regular file 2222 * permissions, either. <sigh> 2223 */ 2224 if (p[2] == '\001' && p[3] == '\002') { 2225 return (ARCHIVE_EOF); 2226 } 2227 2228 /* End of central directory? Must be an 2229 * empty archive. */ 2230 if ((p[2] == '\005' && p[3] == '\006') 2231 || (p[2] == '\006' && p[3] == '\006')) 2232 return (ARCHIVE_EOF); 2233 } 2234 ++p; 2235 ++skipped; 2236 } 2237 __archive_read_consume(a, skipped); 2238 } 2239} 2240 2241static int 2242archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2243{ 2244 struct zip *zip; 2245 int64_t bytes_skipped; 2246 2247 zip = (struct zip *)(a->format->data); 2248 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2249 zip->unconsumed = 0; 2250 if (bytes_skipped < 0) 2251 return (ARCHIVE_FATAL); 2252 2253 /* If we've already read to end of data, we're done. */ 2254 if (zip->end_of_entry) 2255 return (ARCHIVE_OK); 2256 2257 /* So we know we're streaming... */ 2258 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2259 || zip->entry->compressed_size > 0) { 2260 /* We know the compressed length, so we can just skip. */ 2261 bytes_skipped = __archive_read_consume(a, 2262 zip->entry_bytes_remaining); 2263 if (bytes_skipped < 0) 2264 return (ARCHIVE_FATAL); 2265 return (ARCHIVE_OK); 2266 } 2267 2268 if (zip->init_decryption) { 2269 int r; 2270 2271 zip->has_encrypted_entries = 1; 2272 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2273 r = read_decryption_header(a); 2274 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2275 r = init_WinZip_AES_decryption(a); 2276 else 2277 r = init_traditional_PKWARE_decryption(a); 2278 if (r != ARCHIVE_OK) 2279 return (r); 2280 zip->init_decryption = 0; 2281 } 2282 2283 /* We're streaming and we don't know the length. */ 2284 /* If the body is compressed and we know the format, we can 2285 * find an exact end-of-entry by decompressing it. */ 2286 switch (zip->entry->compression) { 2287#ifdef HAVE_ZLIB_H 2288 case 8: /* Deflate compression. */ 2289 while (!zip->end_of_entry) { 2290 int64_t offset = 0; 2291 const void *buff = NULL; 2292 size_t size = 0; 2293 int r; 2294 r = zip_read_data_deflate(a, &buff, &size, &offset); 2295 if (r != ARCHIVE_OK) 2296 return (r); 2297 } 2298 return ARCHIVE_OK; 2299#endif 2300 default: /* Uncompressed or unknown. */ 2301 /* Scan for a PK\007\010 signature. */ 2302 for (;;) { 2303 const char *p, *buff; 2304 ssize_t bytes_avail; 2305 buff = __archive_read_ahead(a, 16, &bytes_avail); 2306 if (bytes_avail < 16) { 2307 archive_set_error(&a->archive, 2308 ARCHIVE_ERRNO_FILE_FORMAT, 2309 "Truncated ZIP file data"); 2310 return (ARCHIVE_FATAL); 2311 } 2312 p = buff; 2313 while (p <= buff + bytes_avail - 16) { 2314 if (p[3] == 'P') { p += 3; } 2315 else if (p[3] == 'K') { p += 2; } 2316 else if (p[3] == '\007') { p += 1; } 2317 else if (p[3] == '\010' && p[2] == '\007' 2318 && p[1] == 'K' && p[0] == 'P') { 2319 if (zip->entry->flags & LA_USED_ZIP64) 2320 __archive_read_consume(a, 2321 p - buff + 24); 2322 else 2323 __archive_read_consume(a, 2324 p - buff + 16); 2325 return ARCHIVE_OK; 2326 } else { p += 4; } 2327 } 2328 __archive_read_consume(a, p - buff); 2329 } 2330 } 2331} 2332 2333int 2334archive_read_support_format_zip_streamable(struct archive *_a) 2335{ 2336 struct archive_read *a = (struct archive_read *)_a; 2337 struct zip *zip; 2338 int r; 2339 2340 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2341 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2342 2343 zip = (struct zip *)calloc(1, sizeof(*zip)); 2344 if (zip == NULL) { 2345 archive_set_error(&a->archive, ENOMEM, 2346 "Can't allocate zip data"); 2347 return (ARCHIVE_FATAL); 2348 } 2349 2350 /* Streamable reader doesn't support mac extensions. */ 2351 zip->process_mac_extensions = 0; 2352 2353 /* 2354 * Until enough data has been read, we cannot tell about 2355 * any encrypted entries yet. 2356 */ 2357 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2358 zip->crc32func = real_crc32; 2359 2360 r = __archive_read_register_format(a, 2361 zip, 2362 "zip", 2363 archive_read_format_zip_streamable_bid, 2364 archive_read_format_zip_options, 2365 archive_read_format_zip_streamable_read_header, 2366 archive_read_format_zip_read_data, 2367 archive_read_format_zip_read_data_skip_streamable, 2368 NULL, 2369 archive_read_format_zip_cleanup, 2370 archive_read_support_format_zip_capabilities_streamable, 2371 archive_read_format_zip_has_encrypted_entries); 2372 2373 if (r != ARCHIVE_OK) 2374 free(zip); 2375 return (ARCHIVE_OK); 2376} 2377 2378/* ------------------------------------------------------------------------ */ 2379 2380/* 2381 * Seeking-mode support 2382 */ 2383 2384static int 2385archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2386{ 2387 (void)a; /* UNUSED */ 2388 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2389 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2390} 2391 2392/* 2393 * TODO: This is a performance sink because it forces the read core to 2394 * drop buffered data from the start of file, which will then have to 2395 * be re-read again if this bidder loses. 2396 * 2397 * We workaround this a little by passing in the best bid so far so 2398 * that later bidders can do nothing if they know they'll never 2399 * outbid. But we can certainly do better... 2400 */ 2401static int 2402read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2403{ 2404 /* Sanity-check the EOCD we've found. */ 2405 2406 /* This must be the first volume. */ 2407 if (archive_le16dec(p + 4) != 0) 2408 return 0; 2409 /* Central directory must be on this volume. */ 2410 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2411 return 0; 2412 /* All central directory entries must be on this volume. */ 2413 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2414 return 0; 2415 /* Central directory can't extend beyond start of EOCD record. */ 2416 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2417 > current_offset) 2418 return 0; 2419 2420 /* Save the central directory location for later use. */ 2421 zip->central_directory_offset = archive_le32dec(p + 16); 2422 2423 /* This is just a tiny bit higher than the maximum 2424 returned by the streaming Zip bidder. This ensures 2425 that the more accurate seeking Zip parser wins 2426 whenever seek is available. */ 2427 return 32; 2428} 2429 2430/* 2431 * Examine Zip64 EOCD locator: If it's valid, store the information 2432 * from it. 2433 */ 2434static int 2435read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2436{ 2437 int64_t eocd64_offset; 2438 int64_t eocd64_size; 2439 2440 /* Sanity-check the locator record. */ 2441 2442 /* Central dir must be on first volume. */ 2443 if (archive_le32dec(p + 4) != 0) 2444 return 0; 2445 /* Must be only a single volume. */ 2446 if (archive_le32dec(p + 16) != 1) 2447 return 0; 2448 2449 /* Find the Zip64 EOCD record. */ 2450 eocd64_offset = archive_le64dec(p + 8); 2451 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2452 return 0; 2453 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2454 return 0; 2455 /* Make sure we can read all of it. */ 2456 eocd64_size = archive_le64dec(p + 4) + 12; 2457 if (eocd64_size < 56 || eocd64_size > 16384) 2458 return 0; 2459 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2460 return 0; 2461 2462 /* Sanity-check the EOCD64 */ 2463 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2464 return 0; 2465 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2466 return 0; 2467 /* CD can't be split. */ 2468 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2469 return 0; 2470 2471 /* Save the central directory offset for later use. */ 2472 zip->central_directory_offset = archive_le64dec(p + 48); 2473 2474 return 32; 2475} 2476 2477static int 2478archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2479{ 2480 struct zip *zip = (struct zip *)a->format->data; 2481 int64_t file_size, current_offset; 2482 const char *p; 2483 int i, tail; 2484 2485 /* If someone has already bid more than 32, then avoid 2486 trashing the look-ahead buffers with a seek. */ 2487 if (best_bid > 32) 2488 return (-1); 2489 2490 file_size = __archive_read_seek(a, 0, SEEK_END); 2491 if (file_size <= 0) 2492 return 0; 2493 2494 /* Search last 16k of file for end-of-central-directory 2495 * record (which starts with PK\005\006) */ 2496 tail = (int)zipmin(1024 * 16, file_size); 2497 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2498 if (current_offset < 0) 2499 return 0; 2500 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2501 return 0; 2502 /* Boyer-Moore search backwards from the end, since we want 2503 * to match the last EOCD in the file (there can be more than 2504 * one if there is an uncompressed Zip archive as a member 2505 * within this Zip archive). */ 2506 for (i = tail - 22; i > 0;) { 2507 switch (p[i]) { 2508 case 'P': 2509 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2510 int ret = read_eocd(zip, p + i, 2511 current_offset + i); 2512 /* Zip64 EOCD locator precedes 2513 * regular EOCD if present. */ 2514 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2515 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); 2516 if (ret_zip64 > ret) 2517 ret = ret_zip64; 2518 } 2519 return (ret); 2520 } 2521 i -= 4; 2522 break; 2523 case 'K': i -= 1; break; 2524 case 005: i -= 2; break; 2525 case 006: i -= 3; break; 2526 default: i -= 4; break; 2527 } 2528 } 2529 return 0; 2530} 2531 2532/* The red-black trees are only used in seeking mode to manage 2533 * the in-memory copy of the central directory. */ 2534 2535static int 2536cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2537{ 2538 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2539 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2540 2541 if (e1->local_header_offset > e2->local_header_offset) 2542 return -1; 2543 if (e1->local_header_offset < e2->local_header_offset) 2544 return 1; 2545 return 0; 2546} 2547 2548static int 2549cmp_key(const struct archive_rb_node *n, const void *key) 2550{ 2551 /* This function won't be called */ 2552 (void)n; /* UNUSED */ 2553 (void)key; /* UNUSED */ 2554 return 1; 2555} 2556 2557static const struct archive_rb_tree_ops rb_ops = { 2558 &cmp_node, &cmp_key 2559}; 2560 2561static int 2562rsrc_cmp_node(const struct archive_rb_node *n1, 2563 const struct archive_rb_node *n2) 2564{ 2565 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2566 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2567 2568 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2569} 2570 2571static int 2572rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2573{ 2574 const struct zip_entry *e = (const struct zip_entry *)n; 2575 return (strcmp((const char *)key, e->rsrcname.s)); 2576} 2577 2578static const struct archive_rb_tree_ops rb_rsrc_ops = { 2579 &rsrc_cmp_node, &rsrc_cmp_key 2580}; 2581 2582static const char * 2583rsrc_basename(const char *name, size_t name_length) 2584{ 2585 const char *s, *r; 2586 2587 r = s = name; 2588 for (;;) { 2589 s = memchr(s, '/', name_length - (s - name)); 2590 if (s == NULL) 2591 break; 2592 r = ++s; 2593 } 2594 return (r); 2595} 2596 2597static void 2598expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2599{ 2600 struct archive_string str; 2601 struct zip_entry *dir; 2602 char *s; 2603 2604 archive_string_init(&str); 2605 archive_strncpy(&str, name, name_length); 2606 for (;;) { 2607 s = strrchr(str.s, '/'); 2608 if (s == NULL) 2609 break; 2610 *s = '\0'; 2611 /* Transfer the parent directory from zip->tree_rsrc RB 2612 * tree to zip->tree RB tree to expose. */ 2613 dir = (struct zip_entry *) 2614 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2615 if (dir == NULL) 2616 break; 2617 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2618 archive_string_free(&dir->rsrcname); 2619 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2620 } 2621 archive_string_free(&str); 2622} 2623 2624static int 2625slurp_central_directory(struct archive_read *a, struct zip *zip) 2626{ 2627 ssize_t i; 2628 unsigned found; 2629 int64_t correction; 2630 ssize_t bytes_avail; 2631 const char *p; 2632 2633 /* 2634 * Find the start of the central directory. The end-of-CD 2635 * record has our starting point, but there are lots of 2636 * Zip archives which have had other data prepended to the 2637 * file, which makes the recorded offsets all too small. 2638 * So we search forward from the specified offset until we 2639 * find the real start of the central directory. Then we 2640 * know the correction we need to apply to account for leading 2641 * padding. 2642 */ 2643 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2644 return ARCHIVE_FATAL; 2645 2646 found = 0; 2647 while (!found) { 2648 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2649 return ARCHIVE_FATAL; 2650 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2651 switch (p[i + 3]) { 2652 case 'P': i += 3; break; 2653 case 'K': i += 2; break; 2654 case 001: i += 1; break; 2655 case 002: 2656 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2657 p += i; 2658 found = 1; 2659 } else 2660 i += 4; 2661 break; 2662 case 005: i += 1; break; 2663 case 006: 2664 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2665 p += i; 2666 found = 1; 2667 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2668 p += i; 2669 found = 1; 2670 } else 2671 i += 1; 2672 break; 2673 default: i += 4; break; 2674 } 2675 } 2676 __archive_read_consume(a, i); 2677 } 2678 correction = archive_filter_bytes(&a->archive, 0) 2679 - zip->central_directory_offset; 2680 2681 __archive_rb_tree_init(&zip->tree, &rb_ops); 2682 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2683 2684 zip->central_directory_entries_total = 0; 2685 while (1) { 2686 struct zip_entry *zip_entry; 2687 size_t filename_length, extra_length, comment_length; 2688 uint32_t external_attributes; 2689 const char *name, *r; 2690 2691 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2692 return ARCHIVE_FATAL; 2693 if (memcmp(p, "PK\006\006", 4) == 0 2694 || memcmp(p, "PK\005\006", 4) == 0) { 2695 break; 2696 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2697 archive_set_error(&a->archive, 2698 -1, "Invalid central directory signature"); 2699 return ARCHIVE_FATAL; 2700 } 2701 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2702 return ARCHIVE_FATAL; 2703 2704 zip_entry = calloc(1, sizeof(struct zip_entry)); 2705 zip_entry->next = zip->zip_entries; 2706 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2707 zip->zip_entries = zip_entry; 2708 zip->central_directory_entries_total++; 2709 2710 /* version = p[4]; */ 2711 zip_entry->system = p[5]; 2712 /* version_required = archive_le16dec(p + 6); */ 2713 zip_entry->zip_flags = archive_le16dec(p + 8); 2714 if (zip_entry->zip_flags 2715 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2716 zip->has_encrypted_entries = 1; 2717 } 2718 zip_entry->compression = (char)archive_le16dec(p + 10); 2719 zip_entry->mtime = zip_time(p + 12); 2720 zip_entry->crc32 = archive_le32dec(p + 16); 2721 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2722 zip_entry->decdat = p[13]; 2723 else 2724 zip_entry->decdat = p[19]; 2725 zip_entry->compressed_size = archive_le32dec(p + 20); 2726 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2727 filename_length = archive_le16dec(p + 28); 2728 extra_length = archive_le16dec(p + 30); 2729 comment_length = archive_le16dec(p + 32); 2730 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2731 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2732 external_attributes = archive_le32dec(p + 38); 2733 zip_entry->local_header_offset = 2734 archive_le32dec(p + 42) + correction; 2735 2736 /* If we can't guess the mode, leave it zero here; 2737 when we read the local file header we might get 2738 more information. */ 2739 if (zip_entry->system == 3) { 2740 zip_entry->mode = external_attributes >> 16; 2741 } else if (zip_entry->system == 0) { 2742 // Interpret MSDOS directory bit 2743 if (0x10 == (external_attributes & 0x10)) { 2744 zip_entry->mode = AE_IFDIR | 0775; 2745 } else { 2746 zip_entry->mode = AE_IFREG | 0664; 2747 } 2748 if (0x01 == (external_attributes & 0x01)) { 2749 // Read-only bit; strip write permissions 2750 zip_entry->mode &= 0555; 2751 } 2752 } else { 2753 zip_entry->mode = 0; 2754 } 2755 2756 /* We're done with the regular data; get the filename and 2757 * extra data. */ 2758 __archive_read_consume(a, 46); 2759 p = __archive_read_ahead(a, filename_length + extra_length, 2760 NULL); 2761 if (p == NULL) { 2762 archive_set_error(&a->archive, 2763 ARCHIVE_ERRNO_FILE_FORMAT, 2764 "Truncated ZIP file header"); 2765 return ARCHIVE_FATAL; 2766 } 2767 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2768 return ARCHIVE_FATAL; 2769 } 2770 2771 /* 2772 * Mac resource fork files are stored under the 2773 * "__MACOSX/" directory, so we should check if 2774 * it is. 2775 */ 2776 if (!zip->process_mac_extensions) { 2777 /* Treat every entry as a regular entry. */ 2778 __archive_rb_tree_insert_node(&zip->tree, 2779 &zip_entry->node); 2780 } else { 2781 name = p; 2782 r = rsrc_basename(name, filename_length); 2783 if (filename_length >= 9 && 2784 strncmp("__MACOSX/", name, 9) == 0) { 2785 /* If this file is not a resource fork nor 2786 * a directory. We should treat it as a non 2787 * resource fork file to expose it. */ 2788 if (name[filename_length-1] != '/' && 2789 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2790 __archive_rb_tree_insert_node( 2791 &zip->tree, &zip_entry->node); 2792 /* Expose its parent directories. */ 2793 expose_parent_dirs(zip, name, 2794 filename_length); 2795 } else { 2796 /* This file is a resource fork file or 2797 * a directory. */ 2798 archive_strncpy(&(zip_entry->rsrcname), 2799 name, filename_length); 2800 __archive_rb_tree_insert_node( 2801 &zip->tree_rsrc, &zip_entry->node); 2802 } 2803 } else { 2804 /* Generate resource fork name to find its 2805 * resource file at zip->tree_rsrc. */ 2806 archive_strcpy(&(zip_entry->rsrcname), 2807 "__MACOSX/"); 2808 archive_strncat(&(zip_entry->rsrcname), 2809 name, r - name); 2810 archive_strcat(&(zip_entry->rsrcname), "._"); 2811 archive_strncat(&(zip_entry->rsrcname), 2812 name + (r - name), 2813 filename_length - (r - name)); 2814 /* Register an entry to RB tree to sort it by 2815 * file offset. */ 2816 __archive_rb_tree_insert_node(&zip->tree, 2817 &zip_entry->node); 2818 } 2819 } 2820 2821 /* Skip the comment too ... */ 2822 __archive_read_consume(a, 2823 filename_length + extra_length + comment_length); 2824 } 2825 2826 return ARCHIVE_OK; 2827} 2828 2829static ssize_t 2830zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2831{ 2832 const char *p; 2833 ssize_t filename_length, extra_length; 2834 2835 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2836 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2837 "Truncated ZIP file header"); 2838 return (ARCHIVE_WARN); 2839 } 2840 p += extra; 2841 2842 if (memcmp(p, "PK\003\004", 4) != 0) { 2843 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2844 return ARCHIVE_WARN; 2845 } 2846 filename_length = archive_le16dec(p + 26); 2847 extra_length = archive_le16dec(p + 28); 2848 2849 return (30 + filename_length + extra_length); 2850} 2851 2852static int 2853zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2854 struct zip_entry *rsrc) 2855{ 2856 struct zip *zip = (struct zip *)a->format->data; 2857 unsigned char *metadata, *mp; 2858 int64_t offset = archive_filter_bytes(&a->archive, 0); 2859 size_t remaining_bytes, metadata_bytes; 2860 ssize_t hsize; 2861 int ret = ARCHIVE_OK, eof; 2862 2863 switch(rsrc->compression) { 2864 case 0: /* No compression. */ 2865 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2866 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2867 "Malformed OS X metadata entry: inconsistent size"); 2868 return (ARCHIVE_FATAL); 2869 } 2870#ifdef HAVE_ZLIB_H 2871 case 8: /* Deflate compression. */ 2872#endif 2873 break; 2874 default: /* Unsupported compression. */ 2875 /* Return a warning. */ 2876 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2877 "Unsupported ZIP compression method (%s)", 2878 compression_name(rsrc->compression)); 2879 /* We can't decompress this entry, but we will 2880 * be able to skip() it and try the next entry. */ 2881 return (ARCHIVE_WARN); 2882 } 2883 2884 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2885 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2886 "Mac metadata is too large: %jd > 4M bytes", 2887 (intmax_t)rsrc->uncompressed_size); 2888 return (ARCHIVE_WARN); 2889 } 2890 if (rsrc->compressed_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->compressed_size); 2894 return (ARCHIVE_WARN); 2895 } 2896 2897 metadata = malloc((size_t)rsrc->uncompressed_size); 2898 if (metadata == NULL) { 2899 archive_set_error(&a->archive, ENOMEM, 2900 "Can't allocate memory for Mac metadata"); 2901 return (ARCHIVE_FATAL); 2902 } 2903 2904 if (offset < rsrc->local_header_offset) 2905 __archive_read_consume(a, rsrc->local_header_offset - offset); 2906 else if (offset != rsrc->local_header_offset) { 2907 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2908 } 2909 2910 hsize = zip_get_local_file_header_size(a, 0); 2911 __archive_read_consume(a, hsize); 2912 2913 remaining_bytes = (size_t)rsrc->compressed_size; 2914 metadata_bytes = (size_t)rsrc->uncompressed_size; 2915 mp = metadata; 2916 eof = 0; 2917 while (!eof && remaining_bytes) { 2918 const unsigned char *p; 2919 ssize_t bytes_avail; 2920 size_t bytes_used; 2921 2922 p = __archive_read_ahead(a, 1, &bytes_avail); 2923 if (p == NULL) { 2924 archive_set_error(&a->archive, 2925 ARCHIVE_ERRNO_FILE_FORMAT, 2926 "Truncated ZIP file header"); 2927 ret = ARCHIVE_WARN; 2928 goto exit_mac_metadata; 2929 } 2930 if ((size_t)bytes_avail > remaining_bytes) 2931 bytes_avail = remaining_bytes; 2932 switch(rsrc->compression) { 2933 case 0: /* No compression. */ 2934 if ((size_t)bytes_avail > metadata_bytes) 2935 bytes_avail = metadata_bytes; 2936 memcpy(mp, p, bytes_avail); 2937 bytes_used = (size_t)bytes_avail; 2938 metadata_bytes -= bytes_used; 2939 mp += bytes_used; 2940 if (metadata_bytes == 0) 2941 eof = 1; 2942 break; 2943#ifdef HAVE_ZLIB_H 2944 case 8: /* Deflate compression. */ 2945 { 2946 int r; 2947 2948 ret = zip_deflate_init(a, zip); 2949 if (ret != ARCHIVE_OK) 2950 goto exit_mac_metadata; 2951 zip->stream.next_in = 2952 (Bytef *)(uintptr_t)(const void *)p; 2953 zip->stream.avail_in = (uInt)bytes_avail; 2954 zip->stream.total_in = 0; 2955 zip->stream.next_out = mp; 2956 zip->stream.avail_out = (uInt)metadata_bytes; 2957 zip->stream.total_out = 0; 2958 2959 r = inflate(&zip->stream, 0); 2960 switch (r) { 2961 case Z_OK: 2962 break; 2963 case Z_STREAM_END: 2964 eof = 1; 2965 break; 2966 case Z_MEM_ERROR: 2967 archive_set_error(&a->archive, ENOMEM, 2968 "Out of memory for ZIP decompression"); 2969 ret = ARCHIVE_FATAL; 2970 goto exit_mac_metadata; 2971 default: 2972 archive_set_error(&a->archive, 2973 ARCHIVE_ERRNO_MISC, 2974 "ZIP decompression failed (%d)", r); 2975 ret = ARCHIVE_FATAL; 2976 goto exit_mac_metadata; 2977 } 2978 bytes_used = zip->stream.total_in; 2979 metadata_bytes -= zip->stream.total_out; 2980 mp += zip->stream.total_out; 2981 break; 2982 } 2983#endif 2984 default: 2985 bytes_used = 0; 2986 break; 2987 } 2988 __archive_read_consume(a, bytes_used); 2989 remaining_bytes -= bytes_used; 2990 } 2991 archive_entry_copy_mac_metadata(entry, metadata, 2992 (size_t)rsrc->uncompressed_size - metadata_bytes); 2993 2994exit_mac_metadata: 2995 __archive_read_seek(a, offset, SEEK_SET); 2996 zip->decompress_init = 0; 2997 free(metadata); 2998 return (ret); 2999} 3000 3001static int 3002archive_read_format_zip_seekable_read_header(struct archive_read *a, 3003 struct archive_entry *entry) 3004{ 3005 struct zip *zip = (struct zip *)a->format->data; 3006 struct zip_entry *rsrc; 3007 int64_t offset; 3008 int r, ret = ARCHIVE_OK; 3009 3010 /* 3011 * It should be sufficient to call archive_read_next_header() for 3012 * a reader to determine if an entry is encrypted or not. If the 3013 * encryption of an entry is only detectable when calling 3014 * archive_read_data(), so be it. We'll do the same check there 3015 * as well. 3016 */ 3017 if (zip->has_encrypted_entries == 3018 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 3019 zip->has_encrypted_entries = 0; 3020 3021 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 3022 if (a->archive.archive_format_name == NULL) 3023 a->archive.archive_format_name = "ZIP"; 3024 3025 if (zip->zip_entries == NULL) { 3026 r = slurp_central_directory(a, zip); 3027 if (r != ARCHIVE_OK) 3028 return r; 3029 /* Get first entry whose local header offset is lower than 3030 * other entries in the archive file. */ 3031 zip->entry = 3032 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 3033 } else if (zip->entry != NULL) { 3034 /* Get next entry in local header offset order. */ 3035 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 3036 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 3037 } 3038 3039 if (zip->entry == NULL) 3040 return ARCHIVE_EOF; 3041 3042 if (zip->entry->rsrcname.s) 3043 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 3044 &zip->tree_rsrc, zip->entry->rsrcname.s); 3045 else 3046 rsrc = NULL; 3047 3048 if (zip->cctx_valid) 3049 archive_decrypto_aes_ctr_release(&zip->cctx); 3050 if (zip->hctx_valid) 3051 archive_hmac_sha1_cleanup(&zip->hctx); 3052 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3053 __archive_read_reset_passphrase(a); 3054 3055 /* File entries are sorted by the header offset, we should mostly 3056 * use __archive_read_consume to advance a read point to avoid redundant 3057 * data reading. */ 3058 offset = archive_filter_bytes(&a->archive, 0); 3059 if (offset < zip->entry->local_header_offset) 3060 __archive_read_consume(a, 3061 zip->entry->local_header_offset - offset); 3062 else if (offset != zip->entry->local_header_offset) { 3063 __archive_read_seek(a, zip->entry->local_header_offset, 3064 SEEK_SET); 3065 } 3066 zip->unconsumed = 0; 3067 r = zip_read_local_file_header(a, entry, zip); 3068 if (r != ARCHIVE_OK) 3069 return r; 3070 if (rsrc) { 3071 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3072 if (ret2 < ret) 3073 ret = ret2; 3074 } 3075 return (ret); 3076} 3077 3078/* 3079 * We're going to seek for the next header anyway, so we don't 3080 * need to bother doing anything here. 3081 */ 3082static int 3083archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3084{ 3085 struct zip *zip; 3086 zip = (struct zip *)(a->format->data); 3087 3088 zip->unconsumed = 0; 3089 return (ARCHIVE_OK); 3090} 3091 3092int 3093archive_read_support_format_zip_seekable(struct archive *_a) 3094{ 3095 struct archive_read *a = (struct archive_read *)_a; 3096 struct zip *zip; 3097 int r; 3098 3099 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3100 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3101 3102 zip = (struct zip *)calloc(1, sizeof(*zip)); 3103 if (zip == NULL) { 3104 archive_set_error(&a->archive, ENOMEM, 3105 "Can't allocate zip data"); 3106 return (ARCHIVE_FATAL); 3107 } 3108 3109#ifdef HAVE_COPYFILE_H 3110 /* Set this by default on Mac OS. */ 3111 zip->process_mac_extensions = 1; 3112#endif 3113 3114 /* 3115 * Until enough data has been read, we cannot tell about 3116 * any encrypted entries yet. 3117 */ 3118 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3119 zip->crc32func = real_crc32; 3120 3121 r = __archive_read_register_format(a, 3122 zip, 3123 "zip", 3124 archive_read_format_zip_seekable_bid, 3125 archive_read_format_zip_options, 3126 archive_read_format_zip_seekable_read_header, 3127 archive_read_format_zip_read_data, 3128 archive_read_format_zip_read_data_skip_seekable, 3129 NULL, 3130 archive_read_format_zip_cleanup, 3131 archive_read_support_format_zip_capabilities_seekable, 3132 archive_read_format_zip_has_encrypted_entries); 3133 3134 if (r != ARCHIVE_OK) 3135 free(zip); 3136 return (ARCHIVE_OK); 3137} 3138