archive_read_support_format_zip.c revision 313926
1/*- 2 * Copyright (c) 2004-2013 Tim Kientzle 3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA 4 * Copyright (c) 2013 Konrad Kleine 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28#include "archive_platform.h" 29__FBSDID("$FreeBSD: stable/11/contrib/libarchive/libarchive/archive_read_support_format_zip.c 313926 2017-02-18 21:58:56Z 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 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 (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 727 /* Vendor version. */ 728 zip_entry->aes_extra.vendor = 729 archive_le16dec(p + offset); 730 /* AES encryption strength. */ 731 zip_entry->aes_extra.strength = p[offset + 4]; 732 /* Actual compression method. */ 733 zip_entry->aes_extra.compression = 734 p[offset + 5]; 735 } 736 break; 737 default: 738 break; 739 } 740 offset += datasize; 741 } 742 if (offset != extra_length) { 743 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 744 "Malformed extra data: Consumed %d bytes of %d bytes", 745 (int)offset, (int)extra_length); 746 return ARCHIVE_FAILED; 747 } 748 return ARCHIVE_OK; 749} 750 751/* 752 * Assumes file pointer is at beginning of local file header. 753 */ 754static int 755zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 756 struct zip *zip) 757{ 758 const char *p; 759 const void *h; 760 const wchar_t *wp; 761 const char *cp; 762 size_t len, filename_length, extra_length; 763 struct archive_string_conv *sconv; 764 struct zip_entry *zip_entry = zip->entry; 765 struct zip_entry zip_entry_central_dir; 766 int ret = ARCHIVE_OK; 767 char version; 768 769 /* Save a copy of the original for consistency checks. */ 770 zip_entry_central_dir = *zip_entry; 771 772 zip->decompress_init = 0; 773 zip->end_of_entry = 0; 774 zip->entry_uncompressed_bytes_read = 0; 775 zip->entry_compressed_bytes_read = 0; 776 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 777 778 /* Setup default conversion. */ 779 if (zip->sconv == NULL && !zip->init_default_conversion) { 780 zip->sconv_default = 781 archive_string_default_conversion_for_read(&(a->archive)); 782 zip->init_default_conversion = 1; 783 } 784 785 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 786 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 787 "Truncated ZIP file header"); 788 return (ARCHIVE_FATAL); 789 } 790 791 if (memcmp(p, "PK\003\004", 4) != 0) { 792 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 793 return ARCHIVE_FATAL; 794 } 795 version = p[4]; 796 zip_entry->system = p[5]; 797 zip_entry->zip_flags = archive_le16dec(p + 6); 798 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 799 zip->has_encrypted_entries = 1; 800 archive_entry_set_is_data_encrypted(entry, 1); 801 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 802 zip_entry->zip_flags & ZIP_ENCRYPTED && 803 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 804 archive_entry_set_is_metadata_encrypted(entry, 1); 805 return ARCHIVE_FATAL; 806 } 807 } 808 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 809 zip_entry->compression = (char)archive_le16dec(p + 8); 810 zip_entry->mtime = zip_time(p + 10); 811 zip_entry->crc32 = archive_le32dec(p + 14); 812 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 813 zip_entry->decdat = p[11]; 814 else 815 zip_entry->decdat = p[17]; 816 zip_entry->compressed_size = archive_le32dec(p + 18); 817 zip_entry->uncompressed_size = archive_le32dec(p + 22); 818 filename_length = archive_le16dec(p + 26); 819 extra_length = archive_le16dec(p + 28); 820 821 __archive_read_consume(a, 30); 822 823 /* Read the filename. */ 824 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 825 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 826 "Truncated ZIP file header"); 827 return (ARCHIVE_FATAL); 828 } 829 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 830 /* The filename is stored to be UTF-8. */ 831 if (zip->sconv_utf8 == NULL) { 832 zip->sconv_utf8 = 833 archive_string_conversion_from_charset( 834 &a->archive, "UTF-8", 1); 835 if (zip->sconv_utf8 == NULL) 836 return (ARCHIVE_FATAL); 837 } 838 sconv = zip->sconv_utf8; 839 } else if (zip->sconv != NULL) 840 sconv = zip->sconv; 841 else 842 sconv = zip->sconv_default; 843 844 if (archive_entry_copy_pathname_l(entry, 845 h, filename_length, sconv) != 0) { 846 if (errno == ENOMEM) { 847 archive_set_error(&a->archive, ENOMEM, 848 "Can't allocate memory for Pathname"); 849 return (ARCHIVE_FATAL); 850 } 851 archive_set_error(&a->archive, 852 ARCHIVE_ERRNO_FILE_FORMAT, 853 "Pathname cannot be converted " 854 "from %s to current locale.", 855 archive_string_conversion_charset_name(sconv)); 856 ret = ARCHIVE_WARN; 857 } 858 __archive_read_consume(a, filename_length); 859 860 /* Read the extra data. */ 861 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 862 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 863 "Truncated ZIP file header"); 864 return (ARCHIVE_FATAL); 865 } 866 867 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 868 return ARCHIVE_FATAL; 869 } 870 __archive_read_consume(a, extra_length); 871 872 /* Work around a bug in Info-Zip: When reading from a pipe, it 873 * stats the pipe instead of synthesizing a file entry. */ 874 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 875 zip_entry->mode &= ~ AE_IFMT; 876 zip_entry->mode |= AE_IFREG; 877 } 878 879 /* If the mode is totally empty, set some sane default. */ 880 if (zip_entry->mode == 0) { 881 zip_entry->mode |= 0664; 882 } 883 884 /* Make sure that entries with a trailing '/' are marked as directories 885 * even if the External File Attributes contains bogus values. If this 886 * is not a directory and there is no type, assume regularfile. */ 887 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 888 int has_slash; 889 890 wp = archive_entry_pathname_w(entry); 891 if (wp != NULL) { 892 len = wcslen(wp); 893 has_slash = len > 0 && wp[len - 1] == L'/'; 894 } else { 895 cp = archive_entry_pathname(entry); 896 len = (cp != NULL)?strlen(cp):0; 897 has_slash = len > 0 && cp[len - 1] == '/'; 898 } 899 /* Correct file type as needed. */ 900 if (has_slash) { 901 zip_entry->mode &= ~AE_IFMT; 902 zip_entry->mode |= AE_IFDIR; 903 zip_entry->mode |= 0111; 904 } else if ((zip_entry->mode & AE_IFMT) == 0) { 905 zip_entry->mode |= AE_IFREG; 906 } 907 } 908 909 /* Make sure directories end in '/' */ 910 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 911 wp = archive_entry_pathname_w(entry); 912 if (wp != NULL) { 913 len = wcslen(wp); 914 if (len > 0 && wp[len - 1] != L'/') { 915 struct archive_wstring s; 916 archive_string_init(&s); 917 archive_wstrcat(&s, wp); 918 archive_wstrappend_wchar(&s, L'/'); 919 archive_entry_copy_pathname_w(entry, s.s); 920 archive_wstring_free(&s); 921 } 922 } else { 923 cp = archive_entry_pathname(entry); 924 len = (cp != NULL)?strlen(cp):0; 925 if (len > 0 && cp[len - 1] != '/') { 926 struct archive_string s; 927 archive_string_init(&s); 928 archive_strcat(&s, cp); 929 archive_strappend_char(&s, '/'); 930 archive_entry_set_pathname(entry, s.s); 931 archive_string_free(&s); 932 } 933 } 934 } 935 936 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 937 /* If this came from the central dir, it's size info 938 * is definitive, so ignore the length-at-end flag. */ 939 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 940 /* If local header is missing a value, use the one from 941 the central directory. If both have it, warn about 942 mismatches. */ 943 if (zip_entry->crc32 == 0) { 944 zip_entry->crc32 = zip_entry_central_dir.crc32; 945 } else if (!zip->ignore_crc32 946 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 947 archive_set_error(&a->archive, 948 ARCHIVE_ERRNO_FILE_FORMAT, 949 "Inconsistent CRC32 values"); 950 ret = ARCHIVE_WARN; 951 } 952 if (zip_entry->compressed_size == 0) { 953 zip_entry->compressed_size 954 = zip_entry_central_dir.compressed_size; 955 } else if (zip_entry->compressed_size 956 != zip_entry_central_dir.compressed_size) { 957 archive_set_error(&a->archive, 958 ARCHIVE_ERRNO_FILE_FORMAT, 959 "Inconsistent compressed size: " 960 "%jd in central directory, %jd in local header", 961 (intmax_t)zip_entry_central_dir.compressed_size, 962 (intmax_t)zip_entry->compressed_size); 963 ret = ARCHIVE_WARN; 964 } 965 if (zip_entry->uncompressed_size == 0) { 966 zip_entry->uncompressed_size 967 = zip_entry_central_dir.uncompressed_size; 968 } else if (zip_entry->uncompressed_size 969 != zip_entry_central_dir.uncompressed_size) { 970 archive_set_error(&a->archive, 971 ARCHIVE_ERRNO_FILE_FORMAT, 972 "Inconsistent uncompressed size: " 973 "%jd in central directory, %jd in local header", 974 (intmax_t)zip_entry_central_dir.uncompressed_size, 975 (intmax_t)zip_entry->uncompressed_size); 976 ret = ARCHIVE_WARN; 977 } 978 } 979 980 /* Populate some additional entry fields: */ 981 archive_entry_set_mode(entry, zip_entry->mode); 982 archive_entry_set_uid(entry, zip_entry->uid); 983 archive_entry_set_gid(entry, zip_entry->gid); 984 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 985 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 986 archive_entry_set_atime(entry, zip_entry->atime, 0); 987 988 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 989 size_t linkname_length; 990 991 if (zip_entry->compressed_size > 64 * 1024) { 992 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 993 "Zip file with oversized link entry"); 994 return ARCHIVE_FATAL; 995 } 996 997 linkname_length = (size_t)zip_entry->compressed_size; 998 999 archive_entry_set_size(entry, 0); 1000 p = __archive_read_ahead(a, linkname_length, NULL); 1001 if (p == NULL) { 1002 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1003 "Truncated Zip file"); 1004 return ARCHIVE_FATAL; 1005 } 1006 1007 sconv = zip->sconv; 1008 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1009 sconv = zip->sconv_utf8; 1010 if (sconv == NULL) 1011 sconv = zip->sconv_default; 1012 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 1013 sconv) != 0) { 1014 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1015 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1016 archive_entry_copy_symlink_l(entry, p, 1017 linkname_length, NULL); 1018 if (errno == ENOMEM) { 1019 archive_set_error(&a->archive, ENOMEM, 1020 "Can't allocate memory for Symlink"); 1021 return (ARCHIVE_FATAL); 1022 } 1023 /* 1024 * Since there is no character-set regulation for 1025 * symlink name, do not report the conversion error 1026 * in an automatic conversion. 1027 */ 1028 if (sconv != zip->sconv_utf8 || 1029 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1030 archive_set_error(&a->archive, 1031 ARCHIVE_ERRNO_FILE_FORMAT, 1032 "Symlink cannot be converted " 1033 "from %s to current locale.", 1034 archive_string_conversion_charset_name( 1035 sconv)); 1036 ret = ARCHIVE_WARN; 1037 } 1038 } 1039 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1040 1041 if (__archive_read_consume(a, linkname_length) < 0) { 1042 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1043 "Read error skipping symlink target name"); 1044 return ARCHIVE_FATAL; 1045 } 1046 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1047 || zip_entry->uncompressed_size > 0) { 1048 /* Set the size only if it's meaningful. */ 1049 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1050 } 1051 zip->entry_bytes_remaining = zip_entry->compressed_size; 1052 1053 /* If there's no body, force read_data() to return EOF immediately. */ 1054 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1055 && zip->entry_bytes_remaining < 1) 1056 zip->end_of_entry = 1; 1057 1058 /* Set up a more descriptive format name. */ 1059 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1060 version / 10, version % 10, 1061 compression_name(zip->entry->compression)); 1062 a->archive.archive_format_name = zip->format_name.s; 1063 1064 return (ret); 1065} 1066 1067static int 1068check_authentication_code(struct archive_read *a, const void *_p) 1069{ 1070 struct zip *zip = (struct zip *)(a->format->data); 1071 1072 /* Check authentication code. */ 1073 if (zip->hctx_valid) { 1074 const void *p; 1075 uint8_t hmac[20]; 1076 size_t hmac_len = 20; 1077 int cmp; 1078 1079 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1080 if (_p == NULL) { 1081 /* Read authentication code. */ 1082 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1083 if (p == NULL) { 1084 archive_set_error(&a->archive, 1085 ARCHIVE_ERRNO_FILE_FORMAT, 1086 "Truncated ZIP file data"); 1087 return (ARCHIVE_FATAL); 1088 } 1089 } else { 1090 p = _p; 1091 } 1092 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1093 __archive_read_consume(a, AUTH_CODE_SIZE); 1094 if (cmp != 0) { 1095 archive_set_error(&a->archive, 1096 ARCHIVE_ERRNO_MISC, 1097 "ZIP bad Authentication code"); 1098 return (ARCHIVE_WARN); 1099 } 1100 } 1101 return (ARCHIVE_OK); 1102} 1103 1104/* 1105 * Read "uncompressed" data. There are three cases: 1106 * 1) We know the size of the data. This is always true for the 1107 * seeking reader (we've examined the Central Directory already). 1108 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1109 * Info-ZIP seems to do this; we know the size but have to grab 1110 * the CRC from the data descriptor afterwards. 1111 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1112 * we have no size information. In this case, we can do pretty 1113 * well by watching for the data descriptor record. The data 1114 * descriptor is 16 bytes and includes a computed CRC that should 1115 * provide a strong check. 1116 * 1117 * TODO: Technically, the PK\007\010 signature is optional. 1118 * In the original spec, the data descriptor contained CRC 1119 * and size fields but had no leading signature. In practice, 1120 * newer writers seem to provide the signature pretty consistently. 1121 * 1122 * For uncompressed data, the PK\007\010 marker seems essential 1123 * to be sure we've actually seen the end of the entry. 1124 * 1125 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1126 * zip->end_of_entry if it consumes all of the data. 1127 */ 1128static int 1129zip_read_data_none(struct archive_read *a, const void **_buff, 1130 size_t *size, int64_t *offset) 1131{ 1132 struct zip *zip; 1133 const char *buff; 1134 ssize_t bytes_avail; 1135 int r; 1136 1137 (void)offset; /* UNUSED */ 1138 1139 zip = (struct zip *)(a->format->data); 1140 1141 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1142 const char *p; 1143 ssize_t grabbing_bytes = 24; 1144 1145 if (zip->hctx_valid) 1146 grabbing_bytes += AUTH_CODE_SIZE; 1147 /* Grab at least 24 bytes. */ 1148 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1149 if (bytes_avail < grabbing_bytes) { 1150 /* Zip archives have end-of-archive markers 1151 that are longer than this, so a failure to get at 1152 least 24 bytes really does indicate a truncated 1153 file. */ 1154 archive_set_error(&a->archive, 1155 ARCHIVE_ERRNO_FILE_FORMAT, 1156 "Truncated ZIP file data"); 1157 return (ARCHIVE_FATAL); 1158 } 1159 /* Check for a complete PK\007\010 signature, followed 1160 * by the correct 4-byte CRC. */ 1161 p = buff; 1162 if (zip->hctx_valid) 1163 p += AUTH_CODE_SIZE; 1164 if (p[0] == 'P' && p[1] == 'K' 1165 && p[2] == '\007' && p[3] == '\010' 1166 && (archive_le32dec(p + 4) == zip->entry_crc32 1167 || zip->ignore_crc32 1168 || (zip->hctx_valid 1169 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1170 if (zip->entry->flags & LA_USED_ZIP64) { 1171 uint64_t compressed, uncompressed; 1172 zip->entry->crc32 = archive_le32dec(p + 4); 1173 compressed = archive_le64dec(p + 8); 1174 uncompressed = archive_le64dec(p + 16); 1175 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1176 archive_set_error(&a->archive, 1177 ARCHIVE_ERRNO_FILE_FORMAT, 1178 "Overflow of 64-bit file sizes"); 1179 return ARCHIVE_FAILED; 1180 } 1181 zip->entry->compressed_size = compressed; 1182 zip->entry->uncompressed_size = uncompressed; 1183 zip->unconsumed = 24; 1184 } else { 1185 zip->entry->crc32 = archive_le32dec(p + 4); 1186 zip->entry->compressed_size = 1187 archive_le32dec(p + 8); 1188 zip->entry->uncompressed_size = 1189 archive_le32dec(p + 12); 1190 zip->unconsumed = 16; 1191 } 1192 if (zip->hctx_valid) { 1193 r = check_authentication_code(a, buff); 1194 if (r != ARCHIVE_OK) 1195 return (r); 1196 } 1197 zip->end_of_entry = 1; 1198 return (ARCHIVE_OK); 1199 } 1200 /* If not at EOF, ensure we consume at least one byte. */ 1201 ++p; 1202 1203 /* Scan forward until we see where a PK\007\010 signature 1204 * might be. */ 1205 /* Return bytes up until that point. On the next call, 1206 * the code above will verify the data descriptor. */ 1207 while (p < buff + bytes_avail - 4) { 1208 if (p[3] == 'P') { p += 3; } 1209 else if (p[3] == 'K') { p += 2; } 1210 else if (p[3] == '\007') { p += 1; } 1211 else if (p[3] == '\010' && p[2] == '\007' 1212 && p[1] == 'K' && p[0] == 'P') { 1213 if (zip->hctx_valid) 1214 p -= AUTH_CODE_SIZE; 1215 break; 1216 } else { p += 4; } 1217 } 1218 bytes_avail = p - buff; 1219 } else { 1220 if (zip->entry_bytes_remaining == 0) { 1221 zip->end_of_entry = 1; 1222 if (zip->hctx_valid) { 1223 r = check_authentication_code(a, NULL); 1224 if (r != ARCHIVE_OK) 1225 return (r); 1226 } 1227 return (ARCHIVE_OK); 1228 } 1229 /* Grab a bunch of bytes. */ 1230 buff = __archive_read_ahead(a, 1, &bytes_avail); 1231 if (bytes_avail <= 0) { 1232 archive_set_error(&a->archive, 1233 ARCHIVE_ERRNO_FILE_FORMAT, 1234 "Truncated ZIP file data"); 1235 return (ARCHIVE_FATAL); 1236 } 1237 if (bytes_avail > zip->entry_bytes_remaining) 1238 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1239 } 1240 if (zip->tctx_valid || zip->cctx_valid) { 1241 size_t dec_size = bytes_avail; 1242 1243 if (dec_size > zip->decrypted_buffer_size) 1244 dec_size = zip->decrypted_buffer_size; 1245 if (zip->tctx_valid) { 1246 trad_enc_decrypt_update(&zip->tctx, 1247 (const uint8_t *)buff, dec_size, 1248 zip->decrypted_buffer, dec_size); 1249 } else { 1250 size_t dsize = dec_size; 1251 archive_hmac_sha1_update(&zip->hctx, 1252 (const uint8_t *)buff, dec_size); 1253 archive_decrypto_aes_ctr_update(&zip->cctx, 1254 (const uint8_t *)buff, dec_size, 1255 zip->decrypted_buffer, &dsize); 1256 } 1257 bytes_avail = dec_size; 1258 buff = (const char *)zip->decrypted_buffer; 1259 } 1260 *size = bytes_avail; 1261 zip->entry_bytes_remaining -= bytes_avail; 1262 zip->entry_uncompressed_bytes_read += bytes_avail; 1263 zip->entry_compressed_bytes_read += bytes_avail; 1264 zip->unconsumed += bytes_avail; 1265 *_buff = buff; 1266 return (ARCHIVE_OK); 1267} 1268 1269#ifdef HAVE_ZLIB_H 1270static int 1271zip_deflate_init(struct archive_read *a, struct zip *zip) 1272{ 1273 int r; 1274 1275 /* If we haven't yet read any data, initialize the decompressor. */ 1276 if (!zip->decompress_init) { 1277 if (zip->stream_valid) 1278 r = inflateReset(&zip->stream); 1279 else 1280 r = inflateInit2(&zip->stream, 1281 -15 /* Don't check for zlib header */); 1282 if (r != Z_OK) { 1283 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1284 "Can't initialize ZIP decompression."); 1285 return (ARCHIVE_FATAL); 1286 } 1287 /* Stream structure has been set up. */ 1288 zip->stream_valid = 1; 1289 /* We've initialized decompression for this stream. */ 1290 zip->decompress_init = 1; 1291 } 1292 return (ARCHIVE_OK); 1293} 1294 1295static int 1296zip_read_data_deflate(struct archive_read *a, const void **buff, 1297 size_t *size, int64_t *offset) 1298{ 1299 struct zip *zip; 1300 ssize_t bytes_avail; 1301 const void *compressed_buff, *sp; 1302 int r; 1303 1304 (void)offset; /* UNUSED */ 1305 1306 zip = (struct zip *)(a->format->data); 1307 1308 /* If the buffer hasn't been allocated, allocate it now. */ 1309 if (zip->uncompressed_buffer == NULL) { 1310 zip->uncompressed_buffer_size = 256 * 1024; 1311 zip->uncompressed_buffer 1312 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1313 if (zip->uncompressed_buffer == NULL) { 1314 archive_set_error(&a->archive, ENOMEM, 1315 "No memory for ZIP decompression"); 1316 return (ARCHIVE_FATAL); 1317 } 1318 } 1319 1320 r = zip_deflate_init(a, zip); 1321 if (r != ARCHIVE_OK) 1322 return (r); 1323 1324 /* 1325 * Note: '1' here is a performance optimization. 1326 * Recall that the decompression layer returns a count of 1327 * available bytes; asking for more than that forces the 1328 * decompressor to combine reads by copying data. 1329 */ 1330 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1331 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1332 && bytes_avail > zip->entry_bytes_remaining) { 1333 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1334 } 1335 if (bytes_avail < 0) { 1336 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1337 "Truncated ZIP file body"); 1338 return (ARCHIVE_FATAL); 1339 } 1340 1341 if (zip->tctx_valid || zip->cctx_valid) { 1342 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1343 size_t buff_remaining = 1344 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1345 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1346 1347 if (buff_remaining > (size_t)bytes_avail) 1348 buff_remaining = (size_t)bytes_avail; 1349 1350 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1351 zip->entry_bytes_remaining > 0) { 1352 if ((int64_t)(zip->decrypted_bytes_remaining 1353 + buff_remaining) 1354 > zip->entry_bytes_remaining) { 1355 if (zip->entry_bytes_remaining < 1356 (int64_t)zip->decrypted_bytes_remaining) 1357 buff_remaining = 0; 1358 else 1359 buff_remaining = 1360 (size_t)zip->entry_bytes_remaining 1361 - zip->decrypted_bytes_remaining; 1362 } 1363 } 1364 if (buff_remaining > 0) { 1365 if (zip->tctx_valid) { 1366 trad_enc_decrypt_update(&zip->tctx, 1367 compressed_buff, buff_remaining, 1368 zip->decrypted_ptr 1369 + zip->decrypted_bytes_remaining, 1370 buff_remaining); 1371 } else { 1372 size_t dsize = buff_remaining; 1373 archive_decrypto_aes_ctr_update( 1374 &zip->cctx, 1375 compressed_buff, buff_remaining, 1376 zip->decrypted_ptr 1377 + zip->decrypted_bytes_remaining, 1378 &dsize); 1379 } 1380 zip->decrypted_bytes_remaining += buff_remaining; 1381 } 1382 } 1383 bytes_avail = zip->decrypted_bytes_remaining; 1384 compressed_buff = (const char *)zip->decrypted_ptr; 1385 } 1386 1387 /* 1388 * A bug in zlib.h: stream.next_in should be marked 'const' 1389 * but isn't (the library never alters data through the 1390 * next_in pointer, only reads it). The result: this ugly 1391 * cast to remove 'const'. 1392 */ 1393 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1394 zip->stream.avail_in = (uInt)bytes_avail; 1395 zip->stream.total_in = 0; 1396 zip->stream.next_out = zip->uncompressed_buffer; 1397 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1398 zip->stream.total_out = 0; 1399 1400 r = inflate(&zip->stream, 0); 1401 switch (r) { 1402 case Z_OK: 1403 break; 1404 case Z_STREAM_END: 1405 zip->end_of_entry = 1; 1406 break; 1407 case Z_MEM_ERROR: 1408 archive_set_error(&a->archive, ENOMEM, 1409 "Out of memory for ZIP decompression"); 1410 return (ARCHIVE_FATAL); 1411 default: 1412 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1413 "ZIP decompression failed (%d)", r); 1414 return (ARCHIVE_FATAL); 1415 } 1416 1417 /* Consume as much as the compressor actually used. */ 1418 bytes_avail = zip->stream.total_in; 1419 if (zip->tctx_valid || zip->cctx_valid) { 1420 zip->decrypted_bytes_remaining -= bytes_avail; 1421 if (zip->decrypted_bytes_remaining == 0) 1422 zip->decrypted_ptr = zip->decrypted_buffer; 1423 else 1424 zip->decrypted_ptr += bytes_avail; 1425 } 1426 /* Calculate compressed data as much as we used.*/ 1427 if (zip->hctx_valid) 1428 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1429 __archive_read_consume(a, bytes_avail); 1430 zip->entry_bytes_remaining -= bytes_avail; 1431 zip->entry_compressed_bytes_read += bytes_avail; 1432 1433 *size = zip->stream.total_out; 1434 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1435 *buff = zip->uncompressed_buffer; 1436 1437 if (zip->end_of_entry && zip->hctx_valid) { 1438 r = check_authentication_code(a, NULL); 1439 if (r != ARCHIVE_OK) 1440 return (r); 1441 } 1442 1443 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1444 const char *p; 1445 1446 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1447 archive_set_error(&a->archive, 1448 ARCHIVE_ERRNO_FILE_FORMAT, 1449 "Truncated ZIP end-of-file record"); 1450 return (ARCHIVE_FATAL); 1451 } 1452 /* Consume the optional PK\007\010 marker. */ 1453 if (p[0] == 'P' && p[1] == 'K' && 1454 p[2] == '\007' && p[3] == '\010') { 1455 p += 4; 1456 zip->unconsumed = 4; 1457 } 1458 if (zip->entry->flags & LA_USED_ZIP64) { 1459 uint64_t compressed, uncompressed; 1460 zip->entry->crc32 = archive_le32dec(p); 1461 compressed = archive_le64dec(p + 4); 1462 uncompressed = archive_le64dec(p + 12); 1463 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1464 archive_set_error(&a->archive, 1465 ARCHIVE_ERRNO_FILE_FORMAT, 1466 "Overflow of 64-bit file sizes"); 1467 return ARCHIVE_FAILED; 1468 } 1469 zip->entry->compressed_size = compressed; 1470 zip->entry->uncompressed_size = uncompressed; 1471 zip->unconsumed += 20; 1472 } else { 1473 zip->entry->crc32 = archive_le32dec(p); 1474 zip->entry->compressed_size = archive_le32dec(p + 4); 1475 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1476 zip->unconsumed += 12; 1477 } 1478 } 1479 1480 return (ARCHIVE_OK); 1481} 1482#endif 1483 1484static int 1485read_decryption_header(struct archive_read *a) 1486{ 1487 struct zip *zip = (struct zip *)(a->format->data); 1488 const char *p; 1489 unsigned int remaining_size; 1490 unsigned int ts; 1491 1492 /* 1493 * Read an initialization vector data field. 1494 */ 1495 p = __archive_read_ahead(a, 2, NULL); 1496 if (p == NULL) 1497 goto truncated; 1498 ts = zip->iv_size; 1499 zip->iv_size = archive_le16dec(p); 1500 __archive_read_consume(a, 2); 1501 if (ts < zip->iv_size) { 1502 free(zip->iv); 1503 zip->iv = NULL; 1504 } 1505 p = __archive_read_ahead(a, zip->iv_size, NULL); 1506 if (p == NULL) 1507 goto truncated; 1508 if (zip->iv == NULL) { 1509 zip->iv = malloc(zip->iv_size); 1510 if (zip->iv == NULL) 1511 goto nomem; 1512 } 1513 memcpy(zip->iv, p, zip->iv_size); 1514 __archive_read_consume(a, zip->iv_size); 1515 1516 /* 1517 * Read a size of remaining decryption header field. 1518 */ 1519 p = __archive_read_ahead(a, 14, NULL); 1520 if (p == NULL) 1521 goto truncated; 1522 remaining_size = archive_le32dec(p); 1523 if (remaining_size < 16 || remaining_size > (1 << 18)) 1524 goto corrupted; 1525 1526 /* Check if format version is supported. */ 1527 if (archive_le16dec(p+4) != 3) { 1528 archive_set_error(&a->archive, 1529 ARCHIVE_ERRNO_FILE_FORMAT, 1530 "Unsupported encryption format version: %u", 1531 archive_le16dec(p+4)); 1532 return (ARCHIVE_FAILED); 1533 } 1534 1535 /* 1536 * Read an encryption algorithm field. 1537 */ 1538 zip->alg_id = archive_le16dec(p+6); 1539 switch (zip->alg_id) { 1540 case 0x6601:/* DES */ 1541 case 0x6602:/* RC2 */ 1542 case 0x6603:/* 3DES 168 */ 1543 case 0x6609:/* 3DES 112 */ 1544 case 0x660E:/* AES 128 */ 1545 case 0x660F:/* AES 192 */ 1546 case 0x6610:/* AES 256 */ 1547 case 0x6702:/* RC2 (version >= 5.2) */ 1548 case 0x6720:/* Blowfish */ 1549 case 0x6721:/* Twofish */ 1550 case 0x6801:/* RC4 */ 1551 /* Supported encryption algorithm. */ 1552 break; 1553 default: 1554 archive_set_error(&a->archive, 1555 ARCHIVE_ERRNO_FILE_FORMAT, 1556 "Unknown encryption algorithm: %u", zip->alg_id); 1557 return (ARCHIVE_FAILED); 1558 } 1559 1560 /* 1561 * Read a bit length field. 1562 */ 1563 zip->bit_len = archive_le16dec(p+8); 1564 1565 /* 1566 * Read a flags field. 1567 */ 1568 zip->flags = archive_le16dec(p+10); 1569 switch (zip->flags & 0xf000) { 1570 case 0x0001: /* Password is required to decrypt. */ 1571 case 0x0002: /* Certificates only. */ 1572 case 0x0003: /* Password or certificate required to decrypt. */ 1573 break; 1574 default: 1575 archive_set_error(&a->archive, 1576 ARCHIVE_ERRNO_FILE_FORMAT, 1577 "Unknown encryption flag: %u", zip->flags); 1578 return (ARCHIVE_FAILED); 1579 } 1580 if ((zip->flags & 0xf000) == 0 || 1581 (zip->flags & 0xf000) == 0x4000) { 1582 archive_set_error(&a->archive, 1583 ARCHIVE_ERRNO_FILE_FORMAT, 1584 "Unknown encryption flag: %u", zip->flags); 1585 return (ARCHIVE_FAILED); 1586 } 1587 1588 /* 1589 * Read an encrypted random data field. 1590 */ 1591 ts = zip->erd_size; 1592 zip->erd_size = archive_le16dec(p+12); 1593 __archive_read_consume(a, 14); 1594 if ((zip->erd_size & 0xf) != 0 || 1595 (zip->erd_size + 16) > remaining_size || 1596 (zip->erd_size + 16) < zip->erd_size) 1597 goto corrupted; 1598 1599 if (ts < zip->erd_size) { 1600 free(zip->erd); 1601 zip->erd = NULL; 1602 } 1603 p = __archive_read_ahead(a, zip->erd_size, NULL); 1604 if (p == NULL) 1605 goto truncated; 1606 if (zip->erd == NULL) { 1607 zip->erd = malloc(zip->erd_size); 1608 if (zip->erd == NULL) 1609 goto nomem; 1610 } 1611 memcpy(zip->erd, p, zip->erd_size); 1612 __archive_read_consume(a, zip->erd_size); 1613 1614 /* 1615 * Read a reserved data field. 1616 */ 1617 p = __archive_read_ahead(a, 4, NULL); 1618 if (p == NULL) 1619 goto truncated; 1620 /* Reserved data size should be zero. */ 1621 if (archive_le32dec(p) != 0) 1622 goto corrupted; 1623 __archive_read_consume(a, 4); 1624 1625 /* 1626 * Read a password validation data field. 1627 */ 1628 p = __archive_read_ahead(a, 2, NULL); 1629 if (p == NULL) 1630 goto truncated; 1631 ts = zip->v_size; 1632 zip->v_size = archive_le16dec(p); 1633 __archive_read_consume(a, 2); 1634 if ((zip->v_size & 0x0f) != 0 || 1635 (zip->erd_size + zip->v_size + 16) > remaining_size || 1636 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1637 goto corrupted; 1638 if (ts < zip->v_size) { 1639 free(zip->v_data); 1640 zip->v_data = NULL; 1641 } 1642 p = __archive_read_ahead(a, zip->v_size, NULL); 1643 if (p == NULL) 1644 goto truncated; 1645 if (zip->v_data == NULL) { 1646 zip->v_data = malloc(zip->v_size); 1647 if (zip->v_data == NULL) 1648 goto nomem; 1649 } 1650 memcpy(zip->v_data, p, zip->v_size); 1651 __archive_read_consume(a, zip->v_size); 1652 1653 p = __archive_read_ahead(a, 4, NULL); 1654 if (p == NULL) 1655 goto truncated; 1656 zip->v_crc32 = archive_le32dec(p); 1657 __archive_read_consume(a, 4); 1658 1659 /*return (ARCHIVE_OK); 1660 * This is not fully implemented yet.*/ 1661 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1662 "Encrypted file is unsupported"); 1663 return (ARCHIVE_FAILED); 1664truncated: 1665 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1666 "Truncated ZIP file data"); 1667 return (ARCHIVE_FATAL); 1668corrupted: 1669 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1670 "Corrupted ZIP file data"); 1671 return (ARCHIVE_FATAL); 1672nomem: 1673 archive_set_error(&a->archive, ENOMEM, 1674 "No memory for ZIP decryption"); 1675 return (ARCHIVE_FATAL); 1676} 1677 1678static int 1679zip_alloc_decryption_buffer(struct archive_read *a) 1680{ 1681 struct zip *zip = (struct zip *)(a->format->data); 1682 size_t bs = 256 * 1024; 1683 1684 if (zip->decrypted_buffer == NULL) { 1685 zip->decrypted_buffer_size = bs; 1686 zip->decrypted_buffer = malloc(bs); 1687 if (zip->decrypted_buffer == NULL) { 1688 archive_set_error(&a->archive, ENOMEM, 1689 "No memory for ZIP decryption"); 1690 return (ARCHIVE_FATAL); 1691 } 1692 } 1693 zip->decrypted_ptr = zip->decrypted_buffer; 1694 return (ARCHIVE_OK); 1695} 1696 1697static int 1698init_traditional_PKWARE_decryption(struct archive_read *a) 1699{ 1700 struct zip *zip = (struct zip *)(a->format->data); 1701 const void *p; 1702 int retry; 1703 int r; 1704 1705 if (zip->tctx_valid) 1706 return (ARCHIVE_OK); 1707 1708 /* 1709 Read the 12 bytes encryption header stored at 1710 the start of the data area. 1711 */ 1712#define ENC_HEADER_SIZE 12 1713 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1714 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1715 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1716 "Truncated Zip encrypted body: only %jd bytes available", 1717 (intmax_t)zip->entry_bytes_remaining); 1718 return (ARCHIVE_FATAL); 1719 } 1720 1721 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1722 if (p == NULL) { 1723 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1724 "Truncated ZIP file data"); 1725 return (ARCHIVE_FATAL); 1726 } 1727 1728 for (retry = 0;; retry++) { 1729 const char *passphrase; 1730 uint8_t crcchk; 1731 1732 passphrase = __archive_read_next_passphrase(a); 1733 if (passphrase == NULL) { 1734 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1735 (retry > 0)? 1736 "Incorrect passphrase": 1737 "Passphrase required for this entry"); 1738 return (ARCHIVE_FAILED); 1739 } 1740 1741 /* 1742 * Initialize ctx for Traditional PKWARE Decryption. 1743 */ 1744 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1745 p, ENC_HEADER_SIZE, &crcchk); 1746 if (r == 0 && crcchk == zip->entry->decdat) 1747 break;/* The passphrase is OK. */ 1748 if (retry > 10000) { 1749 /* Avoid infinity loop. */ 1750 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1751 "Too many incorrect passphrases"); 1752 return (ARCHIVE_FAILED); 1753 } 1754 } 1755 1756 __archive_read_consume(a, ENC_HEADER_SIZE); 1757 zip->tctx_valid = 1; 1758 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1759 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1760 } 1761 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1762 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1763 zip->decrypted_bytes_remaining = 0; 1764 1765 return (zip_alloc_decryption_buffer(a)); 1766#undef ENC_HEADER_SIZE 1767} 1768 1769static int 1770init_WinZip_AES_decryption(struct archive_read *a) 1771{ 1772 struct zip *zip = (struct zip *)(a->format->data); 1773 const void *p; 1774 const uint8_t *pv; 1775 size_t key_len, salt_len; 1776 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1777 int retry; 1778 int r; 1779 1780 if (zip->cctx_valid || zip->hctx_valid) 1781 return (ARCHIVE_OK); 1782 1783 switch (zip->entry->aes_extra.strength) { 1784 case 1: salt_len = 8; key_len = 16; break; 1785 case 2: salt_len = 12; key_len = 24; break; 1786 case 3: salt_len = 16; key_len = 32; break; 1787 default: goto corrupted; 1788 } 1789 p = __archive_read_ahead(a, salt_len + 2, NULL); 1790 if (p == NULL) 1791 goto truncated; 1792 1793 for (retry = 0;; retry++) { 1794 const char *passphrase; 1795 1796 passphrase = __archive_read_next_passphrase(a); 1797 if (passphrase == NULL) { 1798 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1799 (retry > 0)? 1800 "Incorrect passphrase": 1801 "Passphrase required for this entry"); 1802 return (ARCHIVE_FAILED); 1803 } 1804 memset(derived_key, 0, sizeof(derived_key)); 1805 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1806 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1807 if (r != 0) { 1808 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1809 "Decryption is unsupported due to lack of " 1810 "crypto library"); 1811 return (ARCHIVE_FAILED); 1812 } 1813 1814 /* Check password verification value. */ 1815 pv = ((const uint8_t *)p) + salt_len; 1816 if (derived_key[key_len * 2] == pv[0] && 1817 derived_key[key_len * 2 + 1] == pv[1]) 1818 break;/* The passphrase is OK. */ 1819 if (retry > 10000) { 1820 /* Avoid infinity loop. */ 1821 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1822 "Too many incorrect passphrases"); 1823 return (ARCHIVE_FAILED); 1824 } 1825 } 1826 1827 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1828 if (r != 0) { 1829 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1830 "Decryption is unsupported due to lack of crypto library"); 1831 return (ARCHIVE_FAILED); 1832 } 1833 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1834 if (r != 0) { 1835 archive_decrypto_aes_ctr_release(&zip->cctx); 1836 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1837 "Failed to initialize HMAC-SHA1"); 1838 return (ARCHIVE_FAILED); 1839 } 1840 zip->cctx_valid = zip->hctx_valid = 1; 1841 __archive_read_consume(a, salt_len + 2); 1842 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1843 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1844 && zip->entry_bytes_remaining < 0) 1845 goto corrupted; 1846 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1847 zip->decrypted_bytes_remaining = 0; 1848 1849 zip->entry->compression = zip->entry->aes_extra.compression; 1850 return (zip_alloc_decryption_buffer(a)); 1851 1852truncated: 1853 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1854 "Truncated ZIP file data"); 1855 return (ARCHIVE_FATAL); 1856corrupted: 1857 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1858 "Corrupted ZIP file data"); 1859 return (ARCHIVE_FATAL); 1860} 1861 1862static int 1863archive_read_format_zip_read_data(struct archive_read *a, 1864 const void **buff, size_t *size, int64_t *offset) 1865{ 1866 int r; 1867 struct zip *zip = (struct zip *)(a->format->data); 1868 1869 if (zip->has_encrypted_entries == 1870 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1871 zip->has_encrypted_entries = 0; 1872 } 1873 1874 *offset = zip->entry_uncompressed_bytes_read; 1875 *size = 0; 1876 *buff = NULL; 1877 1878 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1879 if (zip->end_of_entry) 1880 return (ARCHIVE_EOF); 1881 1882 /* Return EOF immediately if this is a non-regular file. */ 1883 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1884 return (ARCHIVE_EOF); 1885 1886 __archive_read_consume(a, zip->unconsumed); 1887 zip->unconsumed = 0; 1888 1889 if (zip->init_decryption) { 1890 zip->has_encrypted_entries = 1; 1891 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1892 r = read_decryption_header(a); 1893 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1894 r = init_WinZip_AES_decryption(a); 1895 else 1896 r = init_traditional_PKWARE_decryption(a); 1897 if (r != ARCHIVE_OK) 1898 return (r); 1899 zip->init_decryption = 0; 1900 } 1901 1902 switch(zip->entry->compression) { 1903 case 0: /* No compression. */ 1904 r = zip_read_data_none(a, buff, size, offset); 1905 break; 1906#ifdef HAVE_ZLIB_H 1907 case 8: /* Deflate compression. */ 1908 r = zip_read_data_deflate(a, buff, size, offset); 1909 break; 1910#endif 1911 default: /* Unsupported compression. */ 1912 /* Return a warning. */ 1913 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1914 "Unsupported ZIP compression method (%s)", 1915 compression_name(zip->entry->compression)); 1916 /* We can't decompress this entry, but we will 1917 * be able to skip() it and try the next entry. */ 1918 return (ARCHIVE_FAILED); 1919 break; 1920 } 1921 if (r != ARCHIVE_OK) 1922 return (r); 1923 /* Update checksum */ 1924 if (*size) 1925 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1926 (unsigned)*size); 1927 /* If we hit the end, swallow any end-of-data marker. */ 1928 if (zip->end_of_entry) { 1929 /* Check file size, CRC against these values. */ 1930 if (zip->entry->compressed_size != 1931 zip->entry_compressed_bytes_read) { 1932 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1933 "ZIP compressed data is wrong size " 1934 "(read %jd, expected %jd)", 1935 (intmax_t)zip->entry_compressed_bytes_read, 1936 (intmax_t)zip->entry->compressed_size); 1937 return (ARCHIVE_WARN); 1938 } 1939 /* Size field only stores the lower 32 bits of the actual 1940 * size. */ 1941 if ((zip->entry->uncompressed_size & UINT32_MAX) 1942 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1943 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1944 "ZIP uncompressed data is wrong size " 1945 "(read %jd, expected %jd)\n", 1946 (intmax_t)zip->entry_uncompressed_bytes_read, 1947 (intmax_t)zip->entry->uncompressed_size); 1948 return (ARCHIVE_WARN); 1949 } 1950 /* Check computed CRC against header */ 1951 if ((!zip->hctx_valid || 1952 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1953 zip->entry->crc32 != zip->entry_crc32 1954 && !zip->ignore_crc32) { 1955 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1956 "ZIP bad CRC: 0x%lx should be 0x%lx", 1957 (unsigned long)zip->entry_crc32, 1958 (unsigned long)zip->entry->crc32); 1959 return (ARCHIVE_WARN); 1960 } 1961 } 1962 1963 return (ARCHIVE_OK); 1964} 1965 1966static int 1967archive_read_format_zip_cleanup(struct archive_read *a) 1968{ 1969 struct zip *zip; 1970 struct zip_entry *zip_entry, *next_zip_entry; 1971 1972 zip = (struct zip *)(a->format->data); 1973#ifdef HAVE_ZLIB_H 1974 if (zip->stream_valid) 1975 inflateEnd(&zip->stream); 1976 free(zip->uncompressed_buffer); 1977#endif 1978 if (zip->zip_entries) { 1979 zip_entry = zip->zip_entries; 1980 while (zip_entry != NULL) { 1981 next_zip_entry = zip_entry->next; 1982 archive_string_free(&zip_entry->rsrcname); 1983 free(zip_entry); 1984 zip_entry = next_zip_entry; 1985 } 1986 } 1987 free(zip->decrypted_buffer); 1988 if (zip->cctx_valid) 1989 archive_decrypto_aes_ctr_release(&zip->cctx); 1990 if (zip->hctx_valid) 1991 archive_hmac_sha1_cleanup(&zip->hctx); 1992 free(zip->iv); 1993 free(zip->erd); 1994 free(zip->v_data); 1995 archive_string_free(&zip->format_name); 1996 free(zip); 1997 (a->format->data) = NULL; 1998 return (ARCHIVE_OK); 1999} 2000 2001static int 2002archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 2003{ 2004 if (_a && _a->format) { 2005 struct zip * zip = (struct zip *)_a->format->data; 2006 if (zip) { 2007 return zip->has_encrypted_entries; 2008 } 2009 } 2010 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2011} 2012 2013static int 2014archive_read_format_zip_options(struct archive_read *a, 2015 const char *key, const char *val) 2016{ 2017 struct zip *zip; 2018 int ret = ARCHIVE_FAILED; 2019 2020 zip = (struct zip *)(a->format->data); 2021 if (strcmp(key, "compat-2x") == 0) { 2022 /* Handle filenames as libarchive 2.x */ 2023 zip->init_default_conversion = (val != NULL) ? 1 : 0; 2024 return (ARCHIVE_OK); 2025 } else if (strcmp(key, "hdrcharset") == 0) { 2026 if (val == NULL || val[0] == 0) 2027 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2028 "zip: hdrcharset option needs a character-set name" 2029 ); 2030 else { 2031 zip->sconv = archive_string_conversion_from_charset( 2032 &a->archive, val, 0); 2033 if (zip->sconv != NULL) { 2034 if (strcmp(val, "UTF-8") == 0) 2035 zip->sconv_utf8 = zip->sconv; 2036 ret = ARCHIVE_OK; 2037 } else 2038 ret = ARCHIVE_FATAL; 2039 } 2040 return (ret); 2041 } else if (strcmp(key, "ignorecrc32") == 0) { 2042 /* Mostly useful for testing. */ 2043 if (val == NULL || val[0] == 0) { 2044 zip->crc32func = real_crc32; 2045 zip->ignore_crc32 = 0; 2046 } else { 2047 zip->crc32func = fake_crc32; 2048 zip->ignore_crc32 = 1; 2049 } 2050 return (ARCHIVE_OK); 2051 } else if (strcmp(key, "mac-ext") == 0) { 2052 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2053 return (ARCHIVE_OK); 2054 } 2055 2056 /* Note: The "warn" return is just to inform the options 2057 * supervisor that we didn't handle it. It will generate 2058 * a suitable error if no one used this option. */ 2059 return (ARCHIVE_WARN); 2060} 2061 2062int 2063archive_read_support_format_zip(struct archive *a) 2064{ 2065 int r; 2066 r = archive_read_support_format_zip_streamable(a); 2067 if (r != ARCHIVE_OK) 2068 return r; 2069 return (archive_read_support_format_zip_seekable(a)); 2070} 2071 2072/* ------------------------------------------------------------------------ */ 2073 2074/* 2075 * Streaming-mode support 2076 */ 2077 2078 2079static int 2080archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2081{ 2082 (void)a; /* UNUSED */ 2083 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2084 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2085} 2086 2087static int 2088archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2089{ 2090 const char *p; 2091 2092 (void)best_bid; /* UNUSED */ 2093 2094 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2095 return (-1); 2096 2097 /* 2098 * Bid of 29 here comes from: 2099 * + 16 bits for "PK", 2100 * + next 16-bit field has 6 options so contributes 2101 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2102 * 2103 * So we've effectively verified ~29 total bits of check data. 2104 */ 2105 if (p[0] == 'P' && p[1] == 'K') { 2106 if ((p[2] == '\001' && p[3] == '\002') 2107 || (p[2] == '\003' && p[3] == '\004') 2108 || (p[2] == '\005' && p[3] == '\006') 2109 || (p[2] == '\006' && p[3] == '\006') 2110 || (p[2] == '\007' && p[3] == '\010') 2111 || (p[2] == '0' && p[3] == '0')) 2112 return (29); 2113 } 2114 2115 /* TODO: It's worth looking ahead a little bit for a valid 2116 * PK signature. In particular, that would make it possible 2117 * to read some UUEncoded SFX files or SFX files coming from 2118 * a network socket. */ 2119 2120 return (0); 2121} 2122 2123static int 2124archive_read_format_zip_streamable_read_header(struct archive_read *a, 2125 struct archive_entry *entry) 2126{ 2127 struct zip *zip; 2128 2129 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2130 if (a->archive.archive_format_name == NULL) 2131 a->archive.archive_format_name = "ZIP"; 2132 2133 zip = (struct zip *)(a->format->data); 2134 2135 /* 2136 * It should be sufficient to call archive_read_next_header() for 2137 * a reader to determine if an entry is encrypted or not. If the 2138 * encryption of an entry is only detectable when calling 2139 * archive_read_data(), so be it. We'll do the same check there 2140 * as well. 2141 */ 2142 if (zip->has_encrypted_entries == 2143 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2144 zip->has_encrypted_entries = 0; 2145 2146 /* Make sure we have a zip_entry structure to use. */ 2147 if (zip->zip_entries == NULL) { 2148 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2149 if (zip->zip_entries == NULL) { 2150 archive_set_error(&a->archive, ENOMEM, 2151 "Out of memory"); 2152 return ARCHIVE_FATAL; 2153 } 2154 } 2155 zip->entry = zip->zip_entries; 2156 memset(zip->entry, 0, sizeof(struct zip_entry)); 2157 2158 if (zip->cctx_valid) 2159 archive_decrypto_aes_ctr_release(&zip->cctx); 2160 if (zip->hctx_valid) 2161 archive_hmac_sha1_cleanup(&zip->hctx); 2162 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2163 __archive_read_reset_passphrase(a); 2164 2165 /* Search ahead for the next local file header. */ 2166 __archive_read_consume(a, zip->unconsumed); 2167 zip->unconsumed = 0; 2168 for (;;) { 2169 int64_t skipped = 0; 2170 const char *p, *end; 2171 ssize_t bytes; 2172 2173 p = __archive_read_ahead(a, 4, &bytes); 2174 if (p == NULL) 2175 return (ARCHIVE_FATAL); 2176 end = p + bytes; 2177 2178 while (p + 4 <= end) { 2179 if (p[0] == 'P' && p[1] == 'K') { 2180 if (p[2] == '\003' && p[3] == '\004') { 2181 /* Regular file entry. */ 2182 __archive_read_consume(a, skipped); 2183 return zip_read_local_file_header(a, 2184 entry, zip); 2185 } 2186 2187 /* 2188 * TODO: We cannot restore permissions 2189 * based only on the local file headers. 2190 * Consider scanning the central 2191 * directory and returning additional 2192 * entries for at least directories. 2193 * This would allow us to properly set 2194 * directory permissions. 2195 * 2196 * This won't help us fix symlinks 2197 * and may not help with regular file 2198 * permissions, either. <sigh> 2199 */ 2200 if (p[2] == '\001' && p[3] == '\002') { 2201 return (ARCHIVE_EOF); 2202 } 2203 2204 /* End of central directory? Must be an 2205 * empty archive. */ 2206 if ((p[2] == '\005' && p[3] == '\006') 2207 || (p[2] == '\006' && p[3] == '\006')) 2208 return (ARCHIVE_EOF); 2209 } 2210 ++p; 2211 ++skipped; 2212 } 2213 __archive_read_consume(a, skipped); 2214 } 2215} 2216 2217static int 2218archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2219{ 2220 struct zip *zip; 2221 int64_t bytes_skipped; 2222 2223 zip = (struct zip *)(a->format->data); 2224 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2225 zip->unconsumed = 0; 2226 if (bytes_skipped < 0) 2227 return (ARCHIVE_FATAL); 2228 2229 /* If we've already read to end of data, we're done. */ 2230 if (zip->end_of_entry) 2231 return (ARCHIVE_OK); 2232 2233 /* So we know we're streaming... */ 2234 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2235 || zip->entry->compressed_size > 0) { 2236 /* We know the compressed length, so we can just skip. */ 2237 bytes_skipped = __archive_read_consume(a, 2238 zip->entry_bytes_remaining); 2239 if (bytes_skipped < 0) 2240 return (ARCHIVE_FATAL); 2241 return (ARCHIVE_OK); 2242 } 2243 2244 if (zip->init_decryption) { 2245 int r; 2246 2247 zip->has_encrypted_entries = 1; 2248 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2249 r = read_decryption_header(a); 2250 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2251 r = init_WinZip_AES_decryption(a); 2252 else 2253 r = init_traditional_PKWARE_decryption(a); 2254 if (r != ARCHIVE_OK) 2255 return (r); 2256 zip->init_decryption = 0; 2257 } 2258 2259 /* We're streaming and we don't know the length. */ 2260 /* If the body is compressed and we know the format, we can 2261 * find an exact end-of-entry by decompressing it. */ 2262 switch (zip->entry->compression) { 2263#ifdef HAVE_ZLIB_H 2264 case 8: /* Deflate compression. */ 2265 while (!zip->end_of_entry) { 2266 int64_t offset = 0; 2267 const void *buff = NULL; 2268 size_t size = 0; 2269 int r; 2270 r = zip_read_data_deflate(a, &buff, &size, &offset); 2271 if (r != ARCHIVE_OK) 2272 return (r); 2273 } 2274 return ARCHIVE_OK; 2275#endif 2276 default: /* Uncompressed or unknown. */ 2277 /* Scan for a PK\007\010 signature. */ 2278 for (;;) { 2279 const char *p, *buff; 2280 ssize_t bytes_avail; 2281 buff = __archive_read_ahead(a, 16, &bytes_avail); 2282 if (bytes_avail < 16) { 2283 archive_set_error(&a->archive, 2284 ARCHIVE_ERRNO_FILE_FORMAT, 2285 "Truncated ZIP file data"); 2286 return (ARCHIVE_FATAL); 2287 } 2288 p = buff; 2289 while (p <= buff + bytes_avail - 16) { 2290 if (p[3] == 'P') { p += 3; } 2291 else if (p[3] == 'K') { p += 2; } 2292 else if (p[3] == '\007') { p += 1; } 2293 else if (p[3] == '\010' && p[2] == '\007' 2294 && p[1] == 'K' && p[0] == 'P') { 2295 if (zip->entry->flags & LA_USED_ZIP64) 2296 __archive_read_consume(a, 2297 p - buff + 24); 2298 else 2299 __archive_read_consume(a, 2300 p - buff + 16); 2301 return ARCHIVE_OK; 2302 } else { p += 4; } 2303 } 2304 __archive_read_consume(a, p - buff); 2305 } 2306 } 2307} 2308 2309int 2310archive_read_support_format_zip_streamable(struct archive *_a) 2311{ 2312 struct archive_read *a = (struct archive_read *)_a; 2313 struct zip *zip; 2314 int r; 2315 2316 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2317 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2318 2319 zip = (struct zip *)calloc(1, sizeof(*zip)); 2320 if (zip == NULL) { 2321 archive_set_error(&a->archive, ENOMEM, 2322 "Can't allocate zip data"); 2323 return (ARCHIVE_FATAL); 2324 } 2325 2326 /* Streamable reader doesn't support mac extensions. */ 2327 zip->process_mac_extensions = 0; 2328 2329 /* 2330 * Until enough data has been read, we cannot tell about 2331 * any encrypted entries yet. 2332 */ 2333 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2334 zip->crc32func = real_crc32; 2335 2336 r = __archive_read_register_format(a, 2337 zip, 2338 "zip", 2339 archive_read_format_zip_streamable_bid, 2340 archive_read_format_zip_options, 2341 archive_read_format_zip_streamable_read_header, 2342 archive_read_format_zip_read_data, 2343 archive_read_format_zip_read_data_skip_streamable, 2344 NULL, 2345 archive_read_format_zip_cleanup, 2346 archive_read_support_format_zip_capabilities_streamable, 2347 archive_read_format_zip_has_encrypted_entries); 2348 2349 if (r != ARCHIVE_OK) 2350 free(zip); 2351 return (ARCHIVE_OK); 2352} 2353 2354/* ------------------------------------------------------------------------ */ 2355 2356/* 2357 * Seeking-mode support 2358 */ 2359 2360static int 2361archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2362{ 2363 (void)a; /* UNUSED */ 2364 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2365 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2366} 2367 2368/* 2369 * TODO: This is a performance sink because it forces the read core to 2370 * drop buffered data from the start of file, which will then have to 2371 * be re-read again if this bidder loses. 2372 * 2373 * We workaround this a little by passing in the best bid so far so 2374 * that later bidders can do nothing if they know they'll never 2375 * outbid. But we can certainly do better... 2376 */ 2377static int 2378read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2379{ 2380 /* Sanity-check the EOCD we've found. */ 2381 2382 /* This must be the first volume. */ 2383 if (archive_le16dec(p + 4) != 0) 2384 return 0; 2385 /* Central directory must be on this volume. */ 2386 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2387 return 0; 2388 /* All central directory entries must be on this volume. */ 2389 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2390 return 0; 2391 /* Central directory can't extend beyond start of EOCD record. */ 2392 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2393 > current_offset) 2394 return 0; 2395 2396 /* Save the central directory location for later use. */ 2397 zip->central_directory_offset = archive_le32dec(p + 16); 2398 2399 /* This is just a tiny bit higher than the maximum 2400 returned by the streaming Zip bidder. This ensures 2401 that the more accurate seeking Zip parser wins 2402 whenever seek is available. */ 2403 return 32; 2404} 2405 2406/* 2407 * Examine Zip64 EOCD locator: If it's valid, store the information 2408 * from it. 2409 */ 2410static void 2411read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2412{ 2413 int64_t eocd64_offset; 2414 int64_t eocd64_size; 2415 2416 /* Sanity-check the locator record. */ 2417 2418 /* Central dir must be on first volume. */ 2419 if (archive_le32dec(p + 4) != 0) 2420 return; 2421 /* Must be only a single volume. */ 2422 if (archive_le32dec(p + 16) != 1) 2423 return; 2424 2425 /* Find the Zip64 EOCD record. */ 2426 eocd64_offset = archive_le64dec(p + 8); 2427 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2428 return; 2429 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2430 return; 2431 /* Make sure we can read all of it. */ 2432 eocd64_size = archive_le64dec(p + 4) + 12; 2433 if (eocd64_size < 56 || eocd64_size > 16384) 2434 return; 2435 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2436 return; 2437 2438 /* Sanity-check the EOCD64 */ 2439 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2440 return; 2441 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2442 return; 2443 /* CD can't be split. */ 2444 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2445 return; 2446 2447 /* Save the central directory offset for later use. */ 2448 zip->central_directory_offset = archive_le64dec(p + 48); 2449} 2450 2451static int 2452archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2453{ 2454 struct zip *zip = (struct zip *)a->format->data; 2455 int64_t file_size, current_offset; 2456 const char *p; 2457 int i, tail; 2458 2459 /* If someone has already bid more than 32, then avoid 2460 trashing the look-ahead buffers with a seek. */ 2461 if (best_bid > 32) 2462 return (-1); 2463 2464 file_size = __archive_read_seek(a, 0, SEEK_END); 2465 if (file_size <= 0) 2466 return 0; 2467 2468 /* Search last 16k of file for end-of-central-directory 2469 * record (which starts with PK\005\006) */ 2470 tail = (int)zipmin(1024 * 16, file_size); 2471 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2472 if (current_offset < 0) 2473 return 0; 2474 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2475 return 0; 2476 /* Boyer-Moore search backwards from the end, since we want 2477 * to match the last EOCD in the file (there can be more than 2478 * one if there is an uncompressed Zip archive as a member 2479 * within this Zip archive). */ 2480 for (i = tail - 22; i > 0;) { 2481 switch (p[i]) { 2482 case 'P': 2483 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2484 int ret = read_eocd(zip, p + i, 2485 current_offset + i); 2486 if (ret > 0) { 2487 /* Zip64 EOCD locator precedes 2488 * regular EOCD if present. */ 2489 if (i >= 20 2490 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2491 read_zip64_eocd(a, zip, p + i - 20); 2492 } 2493 return (ret); 2494 } 2495 } 2496 i -= 4; 2497 break; 2498 case 'K': i -= 1; break; 2499 case 005: i -= 2; break; 2500 case 006: i -= 3; break; 2501 default: i -= 4; break; 2502 } 2503 } 2504 return 0; 2505} 2506 2507/* The red-black trees are only used in seeking mode to manage 2508 * the in-memory copy of the central directory. */ 2509 2510static int 2511cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2512{ 2513 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2514 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2515 2516 if (e1->local_header_offset > e2->local_header_offset) 2517 return -1; 2518 if (e1->local_header_offset < e2->local_header_offset) 2519 return 1; 2520 return 0; 2521} 2522 2523static int 2524cmp_key(const struct archive_rb_node *n, const void *key) 2525{ 2526 /* This function won't be called */ 2527 (void)n; /* UNUSED */ 2528 (void)key; /* UNUSED */ 2529 return 1; 2530} 2531 2532static const struct archive_rb_tree_ops rb_ops = { 2533 &cmp_node, &cmp_key 2534}; 2535 2536static int 2537rsrc_cmp_node(const struct archive_rb_node *n1, 2538 const struct archive_rb_node *n2) 2539{ 2540 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2541 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2542 2543 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2544} 2545 2546static int 2547rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2548{ 2549 const struct zip_entry *e = (const struct zip_entry *)n; 2550 return (strcmp((const char *)key, e->rsrcname.s)); 2551} 2552 2553static const struct archive_rb_tree_ops rb_rsrc_ops = { 2554 &rsrc_cmp_node, &rsrc_cmp_key 2555}; 2556 2557static const char * 2558rsrc_basename(const char *name, size_t name_length) 2559{ 2560 const char *s, *r; 2561 2562 r = s = name; 2563 for (;;) { 2564 s = memchr(s, '/', name_length - (s - name)); 2565 if (s == NULL) 2566 break; 2567 r = ++s; 2568 } 2569 return (r); 2570} 2571 2572static void 2573expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2574{ 2575 struct archive_string str; 2576 struct zip_entry *dir; 2577 char *s; 2578 2579 archive_string_init(&str); 2580 archive_strncpy(&str, name, name_length); 2581 for (;;) { 2582 s = strrchr(str.s, '/'); 2583 if (s == NULL) 2584 break; 2585 *s = '\0'; 2586 /* Transfer the parent directory from zip->tree_rsrc RB 2587 * tree to zip->tree RB tree to expose. */ 2588 dir = (struct zip_entry *) 2589 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2590 if (dir == NULL) 2591 break; 2592 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2593 archive_string_free(&dir->rsrcname); 2594 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2595 } 2596 archive_string_free(&str); 2597} 2598 2599static int 2600slurp_central_directory(struct archive_read *a, struct zip *zip) 2601{ 2602 ssize_t i; 2603 unsigned found; 2604 int64_t correction; 2605 ssize_t bytes_avail; 2606 const char *p; 2607 2608 /* 2609 * Find the start of the central directory. The end-of-CD 2610 * record has our starting point, but there are lots of 2611 * Zip archives which have had other data prepended to the 2612 * file, which makes the recorded offsets all too small. 2613 * So we search forward from the specified offset until we 2614 * find the real start of the central directory. Then we 2615 * know the correction we need to apply to account for leading 2616 * padding. 2617 */ 2618 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2619 return ARCHIVE_FATAL; 2620 2621 found = 0; 2622 while (!found) { 2623 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2624 return ARCHIVE_FATAL; 2625 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2626 switch (p[i + 3]) { 2627 case 'P': i += 3; break; 2628 case 'K': i += 2; break; 2629 case 001: i += 1; break; 2630 case 002: 2631 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2632 p += i; 2633 found = 1; 2634 } else 2635 i += 4; 2636 break; 2637 case 005: i += 1; break; 2638 case 006: 2639 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2640 p += i; 2641 found = 1; 2642 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2643 p += i; 2644 found = 1; 2645 } else 2646 i += 1; 2647 break; 2648 default: i += 4; break; 2649 } 2650 } 2651 __archive_read_consume(a, i); 2652 } 2653 correction = archive_filter_bytes(&a->archive, 0) 2654 - zip->central_directory_offset; 2655 2656 __archive_rb_tree_init(&zip->tree, &rb_ops); 2657 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2658 2659 zip->central_directory_entries_total = 0; 2660 while (1) { 2661 struct zip_entry *zip_entry; 2662 size_t filename_length, extra_length, comment_length; 2663 uint32_t external_attributes; 2664 const char *name, *r; 2665 2666 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2667 return ARCHIVE_FATAL; 2668 if (memcmp(p, "PK\006\006", 4) == 0 2669 || memcmp(p, "PK\005\006", 4) == 0) { 2670 break; 2671 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2672 archive_set_error(&a->archive, 2673 -1, "Invalid central directory signature"); 2674 return ARCHIVE_FATAL; 2675 } 2676 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2677 return ARCHIVE_FATAL; 2678 2679 zip_entry = calloc(1, sizeof(struct zip_entry)); 2680 zip_entry->next = zip->zip_entries; 2681 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2682 zip->zip_entries = zip_entry; 2683 zip->central_directory_entries_total++; 2684 2685 /* version = p[4]; */ 2686 zip_entry->system = p[5]; 2687 /* version_required = archive_le16dec(p + 6); */ 2688 zip_entry->zip_flags = archive_le16dec(p + 8); 2689 if (zip_entry->zip_flags 2690 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2691 zip->has_encrypted_entries = 1; 2692 } 2693 zip_entry->compression = (char)archive_le16dec(p + 10); 2694 zip_entry->mtime = zip_time(p + 12); 2695 zip_entry->crc32 = archive_le32dec(p + 16); 2696 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2697 zip_entry->decdat = p[13]; 2698 else 2699 zip_entry->decdat = p[19]; 2700 zip_entry->compressed_size = archive_le32dec(p + 20); 2701 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2702 filename_length = archive_le16dec(p + 28); 2703 extra_length = archive_le16dec(p + 30); 2704 comment_length = archive_le16dec(p + 32); 2705 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2706 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2707 external_attributes = archive_le32dec(p + 38); 2708 zip_entry->local_header_offset = 2709 archive_le32dec(p + 42) + correction; 2710 2711 /* If we can't guess the mode, leave it zero here; 2712 when we read the local file header we might get 2713 more information. */ 2714 if (zip_entry->system == 3) { 2715 zip_entry->mode = external_attributes >> 16; 2716 } else if (zip_entry->system == 0) { 2717 // Interpret MSDOS directory bit 2718 if (0x10 == (external_attributes & 0x10)) { 2719 zip_entry->mode = AE_IFDIR | 0775; 2720 } else { 2721 zip_entry->mode = AE_IFREG | 0664; 2722 } 2723 if (0x01 == (external_attributes & 0x01)) { 2724 // Read-only bit; strip write permissions 2725 zip_entry->mode &= 0555; 2726 } 2727 } else { 2728 zip_entry->mode = 0; 2729 } 2730 2731 /* We're done with the regular data; get the filename and 2732 * extra data. */ 2733 __archive_read_consume(a, 46); 2734 p = __archive_read_ahead(a, filename_length + extra_length, 2735 NULL); 2736 if (p == NULL) { 2737 archive_set_error(&a->archive, 2738 ARCHIVE_ERRNO_FILE_FORMAT, 2739 "Truncated ZIP file header"); 2740 return ARCHIVE_FATAL; 2741 } 2742 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2743 return ARCHIVE_FATAL; 2744 } 2745 2746 /* 2747 * Mac resource fork files are stored under the 2748 * "__MACOSX/" directory, so we should check if 2749 * it is. 2750 */ 2751 if (!zip->process_mac_extensions) { 2752 /* Treat every entry as a regular entry. */ 2753 __archive_rb_tree_insert_node(&zip->tree, 2754 &zip_entry->node); 2755 } else { 2756 name = p; 2757 r = rsrc_basename(name, filename_length); 2758 if (filename_length >= 9 && 2759 strncmp("__MACOSX/", name, 9) == 0) { 2760 /* If this file is not a resource fork nor 2761 * a directory. We should treat it as a non 2762 * resource fork file to expose it. */ 2763 if (name[filename_length-1] != '/' && 2764 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2765 __archive_rb_tree_insert_node( 2766 &zip->tree, &zip_entry->node); 2767 /* Expose its parent directories. */ 2768 expose_parent_dirs(zip, name, 2769 filename_length); 2770 } else { 2771 /* This file is a resource fork file or 2772 * a directory. */ 2773 archive_strncpy(&(zip_entry->rsrcname), 2774 name, filename_length); 2775 __archive_rb_tree_insert_node( 2776 &zip->tree_rsrc, &zip_entry->node); 2777 } 2778 } else { 2779 /* Generate resource fork name to find its 2780 * resource file at zip->tree_rsrc. */ 2781 archive_strcpy(&(zip_entry->rsrcname), 2782 "__MACOSX/"); 2783 archive_strncat(&(zip_entry->rsrcname), 2784 name, r - name); 2785 archive_strcat(&(zip_entry->rsrcname), "._"); 2786 archive_strncat(&(zip_entry->rsrcname), 2787 name + (r - name), 2788 filename_length - (r - name)); 2789 /* Register an entry to RB tree to sort it by 2790 * file offset. */ 2791 __archive_rb_tree_insert_node(&zip->tree, 2792 &zip_entry->node); 2793 } 2794 } 2795 2796 /* Skip the comment too ... */ 2797 __archive_read_consume(a, 2798 filename_length + extra_length + comment_length); 2799 } 2800 2801 return ARCHIVE_OK; 2802} 2803 2804static ssize_t 2805zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2806{ 2807 const char *p; 2808 ssize_t filename_length, extra_length; 2809 2810 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2811 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2812 "Truncated ZIP file header"); 2813 return (ARCHIVE_WARN); 2814 } 2815 p += extra; 2816 2817 if (memcmp(p, "PK\003\004", 4) != 0) { 2818 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2819 return ARCHIVE_WARN; 2820 } 2821 filename_length = archive_le16dec(p + 26); 2822 extra_length = archive_le16dec(p + 28); 2823 2824 return (30 + filename_length + extra_length); 2825} 2826 2827static int 2828zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2829 struct zip_entry *rsrc) 2830{ 2831 struct zip *zip = (struct zip *)a->format->data; 2832 unsigned char *metadata, *mp; 2833 int64_t offset = archive_filter_bytes(&a->archive, 0); 2834 size_t remaining_bytes, metadata_bytes; 2835 ssize_t hsize; 2836 int ret = ARCHIVE_OK, eof; 2837 2838 switch(rsrc->compression) { 2839 case 0: /* No compression. */ 2840 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2841 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2842 "Malformed OS X metadata entry: inconsistent size"); 2843 return (ARCHIVE_FATAL); 2844 } 2845#ifdef HAVE_ZLIB_H 2846 case 8: /* Deflate compression. */ 2847#endif 2848 break; 2849 default: /* Unsupported compression. */ 2850 /* Return a warning. */ 2851 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2852 "Unsupported ZIP compression method (%s)", 2853 compression_name(rsrc->compression)); 2854 /* We can't decompress this entry, but we will 2855 * be able to skip() it and try the next entry. */ 2856 return (ARCHIVE_WARN); 2857 } 2858 2859 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2860 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2861 "Mac metadata is too large: %jd > 4M bytes", 2862 (intmax_t)rsrc->uncompressed_size); 2863 return (ARCHIVE_WARN); 2864 } 2865 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2866 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2867 "Mac metadata is too large: %jd > 4M bytes", 2868 (intmax_t)rsrc->compressed_size); 2869 return (ARCHIVE_WARN); 2870 } 2871 2872 metadata = malloc((size_t)rsrc->uncompressed_size); 2873 if (metadata == NULL) { 2874 archive_set_error(&a->archive, ENOMEM, 2875 "Can't allocate memory for Mac metadata"); 2876 return (ARCHIVE_FATAL); 2877 } 2878 2879 if (offset < rsrc->local_header_offset) 2880 __archive_read_consume(a, rsrc->local_header_offset - offset); 2881 else if (offset != rsrc->local_header_offset) { 2882 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2883 } 2884 2885 hsize = zip_get_local_file_header_size(a, 0); 2886 __archive_read_consume(a, hsize); 2887 2888 remaining_bytes = (size_t)rsrc->compressed_size; 2889 metadata_bytes = (size_t)rsrc->uncompressed_size; 2890 mp = metadata; 2891 eof = 0; 2892 while (!eof && remaining_bytes) { 2893 const unsigned char *p; 2894 ssize_t bytes_avail; 2895 size_t bytes_used; 2896 2897 p = __archive_read_ahead(a, 1, &bytes_avail); 2898 if (p == NULL) { 2899 archive_set_error(&a->archive, 2900 ARCHIVE_ERRNO_FILE_FORMAT, 2901 "Truncated ZIP file header"); 2902 ret = ARCHIVE_WARN; 2903 goto exit_mac_metadata; 2904 } 2905 if ((size_t)bytes_avail > remaining_bytes) 2906 bytes_avail = remaining_bytes; 2907 switch(rsrc->compression) { 2908 case 0: /* No compression. */ 2909 if ((size_t)bytes_avail > metadata_bytes) 2910 bytes_avail = metadata_bytes; 2911 memcpy(mp, p, bytes_avail); 2912 bytes_used = (size_t)bytes_avail; 2913 metadata_bytes -= bytes_used; 2914 mp += bytes_used; 2915 if (metadata_bytes == 0) 2916 eof = 1; 2917 break; 2918#ifdef HAVE_ZLIB_H 2919 case 8: /* Deflate compression. */ 2920 { 2921 int r; 2922 2923 ret = zip_deflate_init(a, zip); 2924 if (ret != ARCHIVE_OK) 2925 goto exit_mac_metadata; 2926 zip->stream.next_in = 2927 (Bytef *)(uintptr_t)(const void *)p; 2928 zip->stream.avail_in = (uInt)bytes_avail; 2929 zip->stream.total_in = 0; 2930 zip->stream.next_out = mp; 2931 zip->stream.avail_out = (uInt)metadata_bytes; 2932 zip->stream.total_out = 0; 2933 2934 r = inflate(&zip->stream, 0); 2935 switch (r) { 2936 case Z_OK: 2937 break; 2938 case Z_STREAM_END: 2939 eof = 1; 2940 break; 2941 case Z_MEM_ERROR: 2942 archive_set_error(&a->archive, ENOMEM, 2943 "Out of memory for ZIP decompression"); 2944 ret = ARCHIVE_FATAL; 2945 goto exit_mac_metadata; 2946 default: 2947 archive_set_error(&a->archive, 2948 ARCHIVE_ERRNO_MISC, 2949 "ZIP decompression failed (%d)", r); 2950 ret = ARCHIVE_FATAL; 2951 goto exit_mac_metadata; 2952 } 2953 bytes_used = zip->stream.total_in; 2954 metadata_bytes -= zip->stream.total_out; 2955 mp += zip->stream.total_out; 2956 break; 2957 } 2958#endif 2959 default: 2960 bytes_used = 0; 2961 break; 2962 } 2963 __archive_read_consume(a, bytes_used); 2964 remaining_bytes -= bytes_used; 2965 } 2966 archive_entry_copy_mac_metadata(entry, metadata, 2967 (size_t)rsrc->uncompressed_size - metadata_bytes); 2968 2969exit_mac_metadata: 2970 __archive_read_seek(a, offset, SEEK_SET); 2971 zip->decompress_init = 0; 2972 free(metadata); 2973 return (ret); 2974} 2975 2976static int 2977archive_read_format_zip_seekable_read_header(struct archive_read *a, 2978 struct archive_entry *entry) 2979{ 2980 struct zip *zip = (struct zip *)a->format->data; 2981 struct zip_entry *rsrc; 2982 int64_t offset; 2983 int r, ret = ARCHIVE_OK; 2984 2985 /* 2986 * It should be sufficient to call archive_read_next_header() for 2987 * a reader to determine if an entry is encrypted or not. If the 2988 * encryption of an entry is only detectable when calling 2989 * archive_read_data(), so be it. We'll do the same check there 2990 * as well. 2991 */ 2992 if (zip->has_encrypted_entries == 2993 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2994 zip->has_encrypted_entries = 0; 2995 2996 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2997 if (a->archive.archive_format_name == NULL) 2998 a->archive.archive_format_name = "ZIP"; 2999 3000 if (zip->zip_entries == NULL) { 3001 r = slurp_central_directory(a, zip); 3002 if (r != ARCHIVE_OK) 3003 return r; 3004 /* Get first entry whose local header offset is lower than 3005 * other entries in the archive file. */ 3006 zip->entry = 3007 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 3008 } else if (zip->entry != NULL) { 3009 /* Get next entry in local header offset order. */ 3010 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 3011 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 3012 } 3013 3014 if (zip->entry == NULL) 3015 return ARCHIVE_EOF; 3016 3017 if (zip->entry->rsrcname.s) 3018 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 3019 &zip->tree_rsrc, zip->entry->rsrcname.s); 3020 else 3021 rsrc = NULL; 3022 3023 if (zip->cctx_valid) 3024 archive_decrypto_aes_ctr_release(&zip->cctx); 3025 if (zip->hctx_valid) 3026 archive_hmac_sha1_cleanup(&zip->hctx); 3027 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3028 __archive_read_reset_passphrase(a); 3029 3030 /* File entries are sorted by the header offset, we should mostly 3031 * use __archive_read_consume to advance a read point to avoid redundant 3032 * data reading. */ 3033 offset = archive_filter_bytes(&a->archive, 0); 3034 if (offset < zip->entry->local_header_offset) 3035 __archive_read_consume(a, 3036 zip->entry->local_header_offset - offset); 3037 else if (offset != zip->entry->local_header_offset) { 3038 __archive_read_seek(a, zip->entry->local_header_offset, 3039 SEEK_SET); 3040 } 3041 zip->unconsumed = 0; 3042 r = zip_read_local_file_header(a, entry, zip); 3043 if (r != ARCHIVE_OK) 3044 return r; 3045 if (rsrc) { 3046 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3047 if (ret2 < ret) 3048 ret = ret2; 3049 } 3050 return (ret); 3051} 3052 3053/* 3054 * We're going to seek for the next header anyway, so we don't 3055 * need to bother doing anything here. 3056 */ 3057static int 3058archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3059{ 3060 struct zip *zip; 3061 zip = (struct zip *)(a->format->data); 3062 3063 zip->unconsumed = 0; 3064 return (ARCHIVE_OK); 3065} 3066 3067int 3068archive_read_support_format_zip_seekable(struct archive *_a) 3069{ 3070 struct archive_read *a = (struct archive_read *)_a; 3071 struct zip *zip; 3072 int r; 3073 3074 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3075 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3076 3077 zip = (struct zip *)calloc(1, sizeof(*zip)); 3078 if (zip == NULL) { 3079 archive_set_error(&a->archive, ENOMEM, 3080 "Can't allocate zip data"); 3081 return (ARCHIVE_FATAL); 3082 } 3083 3084#ifdef HAVE_COPYFILE_H 3085 /* Set this by default on Mac OS. */ 3086 zip->process_mac_extensions = 1; 3087#endif 3088 3089 /* 3090 * Until enough data has been read, we cannot tell about 3091 * any encrypted entries yet. 3092 */ 3093 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3094 zip->crc32func = real_crc32; 3095 3096 r = __archive_read_register_format(a, 3097 zip, 3098 "zip", 3099 archive_read_format_zip_seekable_bid, 3100 archive_read_format_zip_options, 3101 archive_read_format_zip_seekable_read_header, 3102 archive_read_format_zip_read_data, 3103 archive_read_format_zip_read_data_skip_seekable, 3104 NULL, 3105 archive_read_format_zip_cleanup, 3106 archive_read_support_format_zip_capabilities_seekable, 3107 archive_read_format_zip_has_encrypted_entries); 3108 3109 if (r != ARCHIVE_OK) 3110 free(zip); 3111 return (ARCHIVE_OK); 3112} 3113