archive_read_support_format_zip.c revision 305192
1/*- 2 * Copyright (c) 2004-2013 Tim Kientzle 3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA 4 * Copyright (c) 2013 Konrad Kleine 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28#include "archive_platform.h" 29__FBSDID("$FreeBSD: stable/10/contrib/libarchive/libarchive/archive_read_support_format_zip.c 305192 2016-09-01 12:01:23Z 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 decyption. */ 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_decypt_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_decypt_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 if (datasize < 8) 456 break; 457 zip_entry->uncompressed_size = 458 archive_le64dec(p + offset); 459 offset += 8; 460 datasize -= 8; 461 } 462 if (zip_entry->compressed_size == 0xffffffff) { 463 if (datasize < 8) 464 break; 465 zip_entry->compressed_size = 466 archive_le64dec(p + offset); 467 offset += 8; 468 datasize -= 8; 469 } 470 if (zip_entry->local_header_offset == 0xffffffff) { 471 if (datasize < 8) 472 break; 473 zip_entry->local_header_offset = 474 archive_le64dec(p + offset); 475 offset += 8; 476 datasize -= 8; 477 } 478 /* archive_le32dec(p + offset) gives disk 479 * on which file starts, but we don't handle 480 * multi-volume Zip files. */ 481 break; 482#ifdef DEBUG 483 case 0x0017: 484 { 485 /* Strong encryption field. */ 486 if (archive_le16dec(p + offset) == 2) { 487 unsigned algId = 488 archive_le16dec(p + offset + 2); 489 unsigned bitLen = 490 archive_le16dec(p + offset + 4); 491 int flags = 492 archive_le16dec(p + offset + 6); 493 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 494 "flgas=%d\n", algId, bitLen,flags); 495 } 496 break; 497 } 498#endif 499 case 0x5455: 500 { 501 /* Extended time field "UT". */ 502 int flags = p[offset]; 503 offset++; 504 datasize--; 505 /* Flag bits indicate which dates are present. */ 506 if (flags & 0x01) 507 { 508#ifdef DEBUG 509 fprintf(stderr, "mtime: %lld -> %d\n", 510 (long long)zip_entry->mtime, 511 archive_le32dec(p + offset)); 512#endif 513 if (datasize < 4) 514 break; 515 zip_entry->mtime = archive_le32dec(p + offset); 516 offset += 4; 517 datasize -= 4; 518 } 519 if (flags & 0x02) 520 { 521 if (datasize < 4) 522 break; 523 zip_entry->atime = archive_le32dec(p + offset); 524 offset += 4; 525 datasize -= 4; 526 } 527 if (flags & 0x04) 528 { 529 if (datasize < 4) 530 break; 531 zip_entry->ctime = archive_le32dec(p + offset); 532 offset += 4; 533 datasize -= 4; 534 } 535 break; 536 } 537 case 0x5855: 538 { 539 /* Info-ZIP Unix Extra Field (old version) "UX". */ 540 if (datasize >= 8) { 541 zip_entry->atime = archive_le32dec(p + offset); 542 zip_entry->mtime = 543 archive_le32dec(p + offset + 4); 544 } 545 if (datasize >= 12) { 546 zip_entry->uid = 547 archive_le16dec(p + offset + 8); 548 zip_entry->gid = 549 archive_le16dec(p + offset + 10); 550 } 551 break; 552 } 553 case 0x6c78: 554 { 555 /* Experimental 'xl' field */ 556 /* 557 * Introduced Dec 2013 to provide a way to 558 * include external file attributes (and other 559 * fields that ordinarily appear only in 560 * central directory) in local file header. 561 * This provides file type and permission 562 * information necessary to support full 563 * streaming extraction. Currently being 564 * discussed with other Zip developers 565 * ... subject to change. 566 * 567 * Format: 568 * The field starts with a bitmap that specifies 569 * which additional fields are included. The 570 * bitmap is variable length and can be extended in 571 * the future. 572 * 573 * n bytes - feature bitmap: first byte has low-order 574 * 7 bits. If high-order bit is set, a subsequent 575 * byte holds the next 7 bits, etc. 576 * 577 * if bitmap & 1, 2 byte "version made by" 578 * if bitmap & 2, 2 byte "internal file attributes" 579 * if bitmap & 4, 4 byte "external file attributes" 580 * if bitmap & 8, 2 byte comment length + n byte comment 581 */ 582 int bitmap, bitmap_last; 583 584 if (datasize < 1) 585 break; 586 bitmap_last = bitmap = 0xff & p[offset]; 587 offset += 1; 588 datasize -= 1; 589 590 /* We only support first 7 bits of bitmap; skip rest. */ 591 while ((bitmap_last & 0x80) != 0 592 && datasize >= 1) { 593 bitmap_last = p[offset]; 594 offset += 1; 595 datasize -= 1; 596 } 597 598 if (bitmap & 1) { 599 /* 2 byte "version made by" */ 600 if (datasize < 2) 601 break; 602 zip_entry->system 603 = archive_le16dec(p + offset) >> 8; 604 offset += 2; 605 datasize -= 2; 606 } 607 if (bitmap & 2) { 608 /* 2 byte "internal file attributes" */ 609 uint32_t internal_attributes; 610 if (datasize < 2) 611 break; 612 internal_attributes 613 = archive_le16dec(p + offset); 614 /* Not used by libarchive at present. */ 615 (void)internal_attributes; /* UNUSED */ 616 offset += 2; 617 datasize -= 2; 618 } 619 if (bitmap & 4) { 620 /* 4 byte "external file attributes" */ 621 uint32_t external_attributes; 622 if (datasize < 4) 623 break; 624 external_attributes 625 = archive_le32dec(p + offset); 626 if (zip_entry->system == 3) { 627 zip_entry->mode 628 = external_attributes >> 16; 629 } else if (zip_entry->system == 0) { 630 // Interpret MSDOS directory bit 631 if (0x10 == (external_attributes & 0x10)) { 632 zip_entry->mode = AE_IFDIR | 0775; 633 } else { 634 zip_entry->mode = AE_IFREG | 0664; 635 } 636 if (0x01 == (external_attributes & 0x01)) { 637 // Read-only bit; strip write permissions 638 zip_entry->mode &= 0555; 639 } 640 } else { 641 zip_entry->mode = 0; 642 } 643 offset += 4; 644 datasize -= 4; 645 } 646 if (bitmap & 8) { 647 /* 2 byte comment length + comment */ 648 uint32_t comment_length; 649 if (datasize < 2) 650 break; 651 comment_length 652 = archive_le16dec(p + offset); 653 offset += 2; 654 datasize -= 2; 655 656 if (datasize < comment_length) 657 break; 658 /* Comment is not supported by libarchive */ 659 offset += comment_length; 660 datasize -= comment_length; 661 } 662 break; 663 } 664 case 0x7855: 665 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 666#ifdef DEBUG 667 fprintf(stderr, "uid %d gid %d\n", 668 archive_le16dec(p + offset), 669 archive_le16dec(p + offset + 2)); 670#endif 671 if (datasize >= 2) 672 zip_entry->uid = archive_le16dec(p + offset); 673 if (datasize >= 4) 674 zip_entry->gid = 675 archive_le16dec(p + offset + 2); 676 break; 677 case 0x7875: 678 { 679 /* Info-Zip Unix Extra Field (type 3) "ux". */ 680 int uidsize = 0, gidsize = 0; 681 682 /* TODO: support arbitrary uidsize/gidsize. */ 683 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 684 if (datasize >= 4) { 685 /* get a uid size. */ 686 uidsize = 0xff & (int)p[offset+1]; 687 if (uidsize == 2) 688 zip_entry->uid = 689 archive_le16dec( 690 p + offset + 2); 691 else if (uidsize == 4 && datasize >= 6) 692 zip_entry->uid = 693 archive_le32dec( 694 p + offset + 2); 695 } 696 if (datasize >= (2 + uidsize + 3)) { 697 /* get a gid size. */ 698 gidsize = 0xff & (int)p[offset+2+uidsize]; 699 if (gidsize == 2) 700 zip_entry->gid = 701 archive_le16dec( 702 p+offset+2+uidsize+1); 703 else if (gidsize == 4 && 704 datasize >= (2 + uidsize + 5)) 705 zip_entry->gid = 706 archive_le32dec( 707 p+offset+2+uidsize+1); 708 } 709 } 710 break; 711 } 712 case 0x9901: 713 /* WinZIp AES extra data field. */ 714 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 715 /* Vendor version. */ 716 zip_entry->aes_extra.vendor = 717 archive_le16dec(p + offset); 718 /* AES encryption strength. */ 719 zip_entry->aes_extra.strength = p[offset + 4]; 720 /* Actual compression method. */ 721 zip_entry->aes_extra.compression = 722 p[offset + 5]; 723 } 724 break; 725 default: 726 break; 727 } 728 offset += datasize; 729 } 730 if (offset != extra_length) { 731 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 732 "Malformed extra data: Consumed %d bytes of %d bytes", 733 (int)offset, (int)extra_length); 734 return ARCHIVE_FAILED; 735 } 736 return ARCHIVE_OK; 737} 738 739/* 740 * Assumes file pointer is at beginning of local file header. 741 */ 742static int 743zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 744 struct zip *zip) 745{ 746 const char *p; 747 const void *h; 748 const wchar_t *wp; 749 const char *cp; 750 size_t len, filename_length, extra_length; 751 struct archive_string_conv *sconv; 752 struct zip_entry *zip_entry = zip->entry; 753 struct zip_entry zip_entry_central_dir; 754 int ret = ARCHIVE_OK; 755 char version; 756 757 /* Save a copy of the original for consistency checks. */ 758 zip_entry_central_dir = *zip_entry; 759 760 zip->decompress_init = 0; 761 zip->end_of_entry = 0; 762 zip->entry_uncompressed_bytes_read = 0; 763 zip->entry_compressed_bytes_read = 0; 764 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 765 766 /* Setup default conversion. */ 767 if (zip->sconv == NULL && !zip->init_default_conversion) { 768 zip->sconv_default = 769 archive_string_default_conversion_for_read(&(a->archive)); 770 zip->init_default_conversion = 1; 771 } 772 773 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 774 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 775 "Truncated ZIP file header"); 776 return (ARCHIVE_FATAL); 777 } 778 779 if (memcmp(p, "PK\003\004", 4) != 0) { 780 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 781 return ARCHIVE_FATAL; 782 } 783 version = p[4]; 784 zip_entry->system = p[5]; 785 zip_entry->zip_flags = archive_le16dec(p + 6); 786 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 787 zip->has_encrypted_entries = 1; 788 archive_entry_set_is_data_encrypted(entry, 1); 789 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 790 zip_entry->zip_flags & ZIP_ENCRYPTED && 791 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 792 archive_entry_set_is_metadata_encrypted(entry, 1); 793 return ARCHIVE_FATAL; 794 } 795 } 796 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 797 zip_entry->compression = (char)archive_le16dec(p + 8); 798 zip_entry->mtime = zip_time(p + 10); 799 zip_entry->crc32 = archive_le32dec(p + 14); 800 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 801 zip_entry->decdat = p[11]; 802 else 803 zip_entry->decdat = p[17]; 804 zip_entry->compressed_size = archive_le32dec(p + 18); 805 zip_entry->uncompressed_size = archive_le32dec(p + 22); 806 filename_length = archive_le16dec(p + 26); 807 extra_length = archive_le16dec(p + 28); 808 809 __archive_read_consume(a, 30); 810 811 /* Read the filename. */ 812 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 813 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 814 "Truncated ZIP file header"); 815 return (ARCHIVE_FATAL); 816 } 817 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 818 /* The filename is stored to be UTF-8. */ 819 if (zip->sconv_utf8 == NULL) { 820 zip->sconv_utf8 = 821 archive_string_conversion_from_charset( 822 &a->archive, "UTF-8", 1); 823 if (zip->sconv_utf8 == NULL) 824 return (ARCHIVE_FATAL); 825 } 826 sconv = zip->sconv_utf8; 827 } else if (zip->sconv != NULL) 828 sconv = zip->sconv; 829 else 830 sconv = zip->sconv_default; 831 832 if (archive_entry_copy_pathname_l(entry, 833 h, filename_length, sconv) != 0) { 834 if (errno == ENOMEM) { 835 archive_set_error(&a->archive, ENOMEM, 836 "Can't allocate memory for Pathname"); 837 return (ARCHIVE_FATAL); 838 } 839 archive_set_error(&a->archive, 840 ARCHIVE_ERRNO_FILE_FORMAT, 841 "Pathname cannot be converted " 842 "from %s to current locale.", 843 archive_string_conversion_charset_name(sconv)); 844 ret = ARCHIVE_WARN; 845 } 846 __archive_read_consume(a, filename_length); 847 848 /* Read the extra data. */ 849 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 850 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 851 "Truncated ZIP file header"); 852 return (ARCHIVE_FATAL); 853 } 854 855 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 856 return ARCHIVE_FATAL; 857 } 858 __archive_read_consume(a, extra_length); 859 860 /* Work around a bug in Info-Zip: When reading from a pipe, it 861 * stats the pipe instead of synthesizing a file entry. */ 862 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 863 zip_entry->mode &= ~ AE_IFMT; 864 zip_entry->mode |= AE_IFREG; 865 } 866 867 if ((zip_entry->mode & AE_IFMT) == 0) { 868 /* Especially in streaming mode, we can end up 869 here without having seen proper mode information. 870 Guess from the filename. */ 871 wp = archive_entry_pathname_w(entry); 872 if (wp != NULL) { 873 len = wcslen(wp); 874 if (len > 0 && wp[len - 1] == L'/') 875 zip_entry->mode |= AE_IFDIR; 876 else 877 zip_entry->mode |= AE_IFREG; 878 } else { 879 cp = archive_entry_pathname(entry); 880 len = (cp != NULL)?strlen(cp):0; 881 if (len > 0 && cp[len - 1] == '/') 882 zip_entry->mode |= AE_IFDIR; 883 else 884 zip_entry->mode |= AE_IFREG; 885 } 886 if (zip_entry->mode == AE_IFDIR) { 887 zip_entry->mode |= 0775; 888 } else if (zip_entry->mode == AE_IFREG) { 889 zip_entry->mode |= 0664; 890 } 891 } 892 893 /* Make sure directories end in '/' */ 894 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 895 wp = archive_entry_pathname_w(entry); 896 if (wp != NULL) { 897 len = wcslen(wp); 898 if (len > 0 && wp[len - 1] != L'/') { 899 struct archive_wstring s; 900 archive_string_init(&s); 901 archive_wstrcat(&s, wp); 902 archive_wstrappend_wchar(&s, L'/'); 903 archive_entry_copy_pathname_w(entry, s.s); 904 } 905 } else { 906 cp = archive_entry_pathname(entry); 907 len = (cp != NULL)?strlen(cp):0; 908 if (len > 0 && cp[len - 1] != '/') { 909 struct archive_string s; 910 archive_string_init(&s); 911 archive_strcat(&s, cp); 912 archive_strappend_char(&s, '/'); 913 archive_entry_set_pathname(entry, s.s); 914 } 915 } 916 } 917 918 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 919 /* If this came from the central dir, it's size info 920 * is definitive, so ignore the length-at-end flag. */ 921 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 922 /* If local header is missing a value, use the one from 923 the central directory. If both have it, warn about 924 mismatches. */ 925 if (zip_entry->crc32 == 0) { 926 zip_entry->crc32 = zip_entry_central_dir.crc32; 927 } else if (!zip->ignore_crc32 928 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 929 archive_set_error(&a->archive, 930 ARCHIVE_ERRNO_FILE_FORMAT, 931 "Inconsistent CRC32 values"); 932 ret = ARCHIVE_WARN; 933 } 934 if (zip_entry->compressed_size == 0) { 935 zip_entry->compressed_size 936 = zip_entry_central_dir.compressed_size; 937 } else if (zip_entry->compressed_size 938 != zip_entry_central_dir.compressed_size) { 939 archive_set_error(&a->archive, 940 ARCHIVE_ERRNO_FILE_FORMAT, 941 "Inconsistent compressed size: " 942 "%jd in central directory, %jd in local header", 943 (intmax_t)zip_entry_central_dir.compressed_size, 944 (intmax_t)zip_entry->compressed_size); 945 ret = ARCHIVE_WARN; 946 } 947 if (zip_entry->uncompressed_size == 0) { 948 zip_entry->uncompressed_size 949 = zip_entry_central_dir.uncompressed_size; 950 } else if (zip_entry->uncompressed_size 951 != zip_entry_central_dir.uncompressed_size) { 952 archive_set_error(&a->archive, 953 ARCHIVE_ERRNO_FILE_FORMAT, 954 "Inconsistent uncompressed size: " 955 "%jd in central directory, %jd in local header", 956 (intmax_t)zip_entry_central_dir.uncompressed_size, 957 (intmax_t)zip_entry->uncompressed_size); 958 ret = ARCHIVE_WARN; 959 } 960 } 961 962 /* Populate some additional entry fields: */ 963 archive_entry_set_mode(entry, zip_entry->mode); 964 archive_entry_set_uid(entry, zip_entry->uid); 965 archive_entry_set_gid(entry, zip_entry->gid); 966 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 967 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 968 archive_entry_set_atime(entry, zip_entry->atime, 0); 969 970 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 971 size_t linkname_length; 972 973 if (zip_entry->compressed_size > 64 * 1024) { 974 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 975 "Zip file with oversized link entry"); 976 return ARCHIVE_FATAL; 977 } 978 979 linkname_length = (size_t)zip_entry->compressed_size; 980 981 archive_entry_set_size(entry, 0); 982 p = __archive_read_ahead(a, linkname_length, NULL); 983 if (p == NULL) { 984 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 985 "Truncated Zip file"); 986 return ARCHIVE_FATAL; 987 } 988 989 sconv = zip->sconv; 990 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 991 sconv = zip->sconv_utf8; 992 if (sconv == NULL) 993 sconv = zip->sconv_default; 994 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 995 sconv) != 0) { 996 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 997 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 998 archive_entry_copy_symlink_l(entry, p, 999 linkname_length, NULL); 1000 if (errno == ENOMEM) { 1001 archive_set_error(&a->archive, ENOMEM, 1002 "Can't allocate memory for Symlink"); 1003 return (ARCHIVE_FATAL); 1004 } 1005 /* 1006 * Since there is no character-set regulation for 1007 * symlink name, do not report the conversion error 1008 * in an automatic conversion. 1009 */ 1010 if (sconv != zip->sconv_utf8 || 1011 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1012 archive_set_error(&a->archive, 1013 ARCHIVE_ERRNO_FILE_FORMAT, 1014 "Symlink cannot be converted " 1015 "from %s to current locale.", 1016 archive_string_conversion_charset_name( 1017 sconv)); 1018 ret = ARCHIVE_WARN; 1019 } 1020 } 1021 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1022 1023 if (__archive_read_consume(a, linkname_length) < 0) { 1024 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1025 "Read error skipping symlink target name"); 1026 return ARCHIVE_FATAL; 1027 } 1028 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1029 || zip_entry->uncompressed_size > 0) { 1030 /* Set the size only if it's meaningful. */ 1031 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1032 } 1033 zip->entry_bytes_remaining = zip_entry->compressed_size; 1034 1035 /* If there's no body, force read_data() to return EOF immediately. */ 1036 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1037 && zip->entry_bytes_remaining < 1) 1038 zip->end_of_entry = 1; 1039 1040 /* Set up a more descriptive format name. */ 1041 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1042 version / 10, version % 10, 1043 compression_name(zip->entry->compression)); 1044 a->archive.archive_format_name = zip->format_name.s; 1045 1046 return (ret); 1047} 1048 1049static int 1050check_authentication_code(struct archive_read *a, const void *_p) 1051{ 1052 struct zip *zip = (struct zip *)(a->format->data); 1053 1054 /* Check authentication code. */ 1055 if (zip->hctx_valid) { 1056 const void *p; 1057 uint8_t hmac[20]; 1058 size_t hmac_len = 20; 1059 int cmp; 1060 1061 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1062 if (_p == NULL) { 1063 /* Read authentication code. */ 1064 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1065 if (p == NULL) { 1066 archive_set_error(&a->archive, 1067 ARCHIVE_ERRNO_FILE_FORMAT, 1068 "Truncated ZIP file data"); 1069 return (ARCHIVE_FATAL); 1070 } 1071 } else { 1072 p = _p; 1073 } 1074 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1075 __archive_read_consume(a, AUTH_CODE_SIZE); 1076 if (cmp != 0) { 1077 archive_set_error(&a->archive, 1078 ARCHIVE_ERRNO_MISC, 1079 "ZIP bad Authentication code"); 1080 return (ARCHIVE_WARN); 1081 } 1082 } 1083 return (ARCHIVE_OK); 1084} 1085 1086/* 1087 * Read "uncompressed" data. There are three cases: 1088 * 1) We know the size of the data. This is always true for the 1089 * seeking reader (we've examined the Central Directory already). 1090 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1091 * Info-ZIP seems to do this; we know the size but have to grab 1092 * the CRC from the data descriptor afterwards. 1093 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1094 * we have no size information. In this case, we can do pretty 1095 * well by watching for the data descriptor record. The data 1096 * descriptor is 16 bytes and includes a computed CRC that should 1097 * provide a strong check. 1098 * 1099 * TODO: Technically, the PK\007\010 signature is optional. 1100 * In the original spec, the data descriptor contained CRC 1101 * and size fields but had no leading signature. In practice, 1102 * newer writers seem to provide the signature pretty consistently. 1103 * 1104 * For uncompressed data, the PK\007\010 marker seems essential 1105 * to be sure we've actually seen the end of the entry. 1106 * 1107 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1108 * zip->end_of_entry if it consumes all of the data. 1109 */ 1110static int 1111zip_read_data_none(struct archive_read *a, const void **_buff, 1112 size_t *size, int64_t *offset) 1113{ 1114 struct zip *zip; 1115 const char *buff; 1116 ssize_t bytes_avail; 1117 int r; 1118 1119 (void)offset; /* UNUSED */ 1120 1121 zip = (struct zip *)(a->format->data); 1122 1123 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1124 const char *p; 1125 ssize_t grabbing_bytes = 24; 1126 1127 if (zip->hctx_valid) 1128 grabbing_bytes += AUTH_CODE_SIZE; 1129 /* Grab at least 24 bytes. */ 1130 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1131 if (bytes_avail < grabbing_bytes) { 1132 /* Zip archives have end-of-archive markers 1133 that are longer than this, so a failure to get at 1134 least 24 bytes really does indicate a truncated 1135 file. */ 1136 archive_set_error(&a->archive, 1137 ARCHIVE_ERRNO_FILE_FORMAT, 1138 "Truncated ZIP file data"); 1139 return (ARCHIVE_FATAL); 1140 } 1141 /* Check for a complete PK\007\010 signature, followed 1142 * by the correct 4-byte CRC. */ 1143 p = buff; 1144 if (zip->hctx_valid) 1145 p += AUTH_CODE_SIZE; 1146 if (p[0] == 'P' && p[1] == 'K' 1147 && p[2] == '\007' && p[3] == '\010' 1148 && (archive_le32dec(p + 4) == zip->entry_crc32 1149 || zip->ignore_crc32 1150 || (zip->hctx_valid 1151 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1152 if (zip->entry->flags & LA_USED_ZIP64) { 1153 zip->entry->crc32 = archive_le32dec(p + 4); 1154 zip->entry->compressed_size = 1155 archive_le64dec(p + 8); 1156 zip->entry->uncompressed_size = 1157 archive_le64dec(p + 16); 1158 zip->unconsumed = 24; 1159 } else { 1160 zip->entry->crc32 = archive_le32dec(p + 4); 1161 zip->entry->compressed_size = 1162 archive_le32dec(p + 8); 1163 zip->entry->uncompressed_size = 1164 archive_le32dec(p + 12); 1165 zip->unconsumed = 16; 1166 } 1167 if (zip->hctx_valid) { 1168 r = check_authentication_code(a, buff); 1169 if (r != ARCHIVE_OK) 1170 return (r); 1171 } 1172 zip->end_of_entry = 1; 1173 return (ARCHIVE_OK); 1174 } 1175 /* If not at EOF, ensure we consume at least one byte. */ 1176 ++p; 1177 1178 /* Scan forward until we see where a PK\007\010 signature 1179 * might be. */ 1180 /* Return bytes up until that point. On the next call, 1181 * the code above will verify the data descriptor. */ 1182 while (p < buff + bytes_avail - 4) { 1183 if (p[3] == 'P') { p += 3; } 1184 else if (p[3] == 'K') { p += 2; } 1185 else if (p[3] == '\007') { p += 1; } 1186 else if (p[3] == '\010' && p[2] == '\007' 1187 && p[1] == 'K' && p[0] == 'P') { 1188 if (zip->hctx_valid) 1189 p -= AUTH_CODE_SIZE; 1190 break; 1191 } else { p += 4; } 1192 } 1193 bytes_avail = p - buff; 1194 } else { 1195 if (zip->entry_bytes_remaining == 0) { 1196 zip->end_of_entry = 1; 1197 if (zip->hctx_valid) { 1198 r = check_authentication_code(a, NULL); 1199 if (r != ARCHIVE_OK) 1200 return (r); 1201 } 1202 return (ARCHIVE_OK); 1203 } 1204 /* Grab a bunch of bytes. */ 1205 buff = __archive_read_ahead(a, 1, &bytes_avail); 1206 if (bytes_avail <= 0) { 1207 archive_set_error(&a->archive, 1208 ARCHIVE_ERRNO_FILE_FORMAT, 1209 "Truncated ZIP file data"); 1210 return (ARCHIVE_FATAL); 1211 } 1212 if (bytes_avail > zip->entry_bytes_remaining) 1213 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1214 } 1215 if (zip->tctx_valid || zip->cctx_valid) { 1216 size_t dec_size = bytes_avail; 1217 1218 if (dec_size > zip->decrypted_buffer_size) 1219 dec_size = zip->decrypted_buffer_size; 1220 if (zip->tctx_valid) { 1221 trad_enc_decrypt_update(&zip->tctx, 1222 (const uint8_t *)buff, dec_size, 1223 zip->decrypted_buffer, dec_size); 1224 } else { 1225 size_t dsize = dec_size; 1226 archive_hmac_sha1_update(&zip->hctx, 1227 (const uint8_t *)buff, dec_size); 1228 archive_decrypto_aes_ctr_update(&zip->cctx, 1229 (const uint8_t *)buff, dec_size, 1230 zip->decrypted_buffer, &dsize); 1231 } 1232 bytes_avail = dec_size; 1233 buff = (const char *)zip->decrypted_buffer; 1234 } 1235 *size = bytes_avail; 1236 zip->entry_bytes_remaining -= bytes_avail; 1237 zip->entry_uncompressed_bytes_read += bytes_avail; 1238 zip->entry_compressed_bytes_read += bytes_avail; 1239 zip->unconsumed += bytes_avail; 1240 *_buff = buff; 1241 return (ARCHIVE_OK); 1242} 1243 1244#ifdef HAVE_ZLIB_H 1245static int 1246zip_deflate_init(struct archive_read *a, struct zip *zip) 1247{ 1248 int r; 1249 1250 /* If we haven't yet read any data, initialize the decompressor. */ 1251 if (!zip->decompress_init) { 1252 if (zip->stream_valid) 1253 r = inflateReset(&zip->stream); 1254 else 1255 r = inflateInit2(&zip->stream, 1256 -15 /* Don't check for zlib header */); 1257 if (r != Z_OK) { 1258 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1259 "Can't initialize ZIP decompression."); 1260 return (ARCHIVE_FATAL); 1261 } 1262 /* Stream structure has been set up. */ 1263 zip->stream_valid = 1; 1264 /* We've initialized decompression for this stream. */ 1265 zip->decompress_init = 1; 1266 } 1267 return (ARCHIVE_OK); 1268} 1269 1270static int 1271zip_read_data_deflate(struct archive_read *a, const void **buff, 1272 size_t *size, int64_t *offset) 1273{ 1274 struct zip *zip; 1275 ssize_t bytes_avail; 1276 const void *compressed_buff, *sp; 1277 int r; 1278 1279 (void)offset; /* UNUSED */ 1280 1281 zip = (struct zip *)(a->format->data); 1282 1283 /* If the buffer hasn't been allocated, allocate it now. */ 1284 if (zip->uncompressed_buffer == NULL) { 1285 zip->uncompressed_buffer_size = 256 * 1024; 1286 zip->uncompressed_buffer 1287 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1288 if (zip->uncompressed_buffer == NULL) { 1289 archive_set_error(&a->archive, ENOMEM, 1290 "No memory for ZIP decompression"); 1291 return (ARCHIVE_FATAL); 1292 } 1293 } 1294 1295 r = zip_deflate_init(a, zip); 1296 if (r != ARCHIVE_OK) 1297 return (r); 1298 1299 /* 1300 * Note: '1' here is a performance optimization. 1301 * Recall that the decompression layer returns a count of 1302 * available bytes; asking for more than that forces the 1303 * decompressor to combine reads by copying data. 1304 */ 1305 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1306 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1307 && bytes_avail > zip->entry_bytes_remaining) { 1308 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1309 } 1310 if (bytes_avail < 0) { 1311 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1312 "Truncated ZIP file body"); 1313 return (ARCHIVE_FATAL); 1314 } 1315 1316 if (zip->tctx_valid || zip->cctx_valid) { 1317 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1318 size_t buff_remaining = 1319 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1320 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1321 1322 if (buff_remaining > (size_t)bytes_avail) 1323 buff_remaining = (size_t)bytes_avail; 1324 1325 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1326 zip->entry_bytes_remaining > 0) { 1327 if ((int64_t)(zip->decrypted_bytes_remaining 1328 + buff_remaining) 1329 > zip->entry_bytes_remaining) { 1330 if (zip->entry_bytes_remaining < 1331 (int64_t)zip->decrypted_bytes_remaining) 1332 buff_remaining = 0; 1333 else 1334 buff_remaining = 1335 (size_t)zip->entry_bytes_remaining 1336 - zip->decrypted_bytes_remaining; 1337 } 1338 } 1339 if (buff_remaining > 0) { 1340 if (zip->tctx_valid) { 1341 trad_enc_decrypt_update(&zip->tctx, 1342 compressed_buff, buff_remaining, 1343 zip->decrypted_ptr 1344 + zip->decrypted_bytes_remaining, 1345 buff_remaining); 1346 } else { 1347 size_t dsize = buff_remaining; 1348 archive_decrypto_aes_ctr_update( 1349 &zip->cctx, 1350 compressed_buff, buff_remaining, 1351 zip->decrypted_ptr 1352 + zip->decrypted_bytes_remaining, 1353 &dsize); 1354 } 1355 zip->decrypted_bytes_remaining += buff_remaining; 1356 } 1357 } 1358 bytes_avail = zip->decrypted_bytes_remaining; 1359 compressed_buff = (const char *)zip->decrypted_ptr; 1360 } 1361 1362 /* 1363 * A bug in zlib.h: stream.next_in should be marked 'const' 1364 * but isn't (the library never alters data through the 1365 * next_in pointer, only reads it). The result: this ugly 1366 * cast to remove 'const'. 1367 */ 1368 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1369 zip->stream.avail_in = (uInt)bytes_avail; 1370 zip->stream.total_in = 0; 1371 zip->stream.next_out = zip->uncompressed_buffer; 1372 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1373 zip->stream.total_out = 0; 1374 1375 r = inflate(&zip->stream, 0); 1376 switch (r) { 1377 case Z_OK: 1378 break; 1379 case Z_STREAM_END: 1380 zip->end_of_entry = 1; 1381 break; 1382 case Z_MEM_ERROR: 1383 archive_set_error(&a->archive, ENOMEM, 1384 "Out of memory for ZIP decompression"); 1385 return (ARCHIVE_FATAL); 1386 default: 1387 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1388 "ZIP decompression failed (%d)", r); 1389 return (ARCHIVE_FATAL); 1390 } 1391 1392 /* Consume as much as the compressor actually used. */ 1393 bytes_avail = zip->stream.total_in; 1394 if (zip->tctx_valid || zip->cctx_valid) { 1395 zip->decrypted_bytes_remaining -= bytes_avail; 1396 if (zip->decrypted_bytes_remaining == 0) 1397 zip->decrypted_ptr = zip->decrypted_buffer; 1398 else 1399 zip->decrypted_ptr += bytes_avail; 1400 } 1401 /* Calculate compressed data as much as we used.*/ 1402 if (zip->hctx_valid) 1403 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1404 __archive_read_consume(a, bytes_avail); 1405 zip->entry_bytes_remaining -= bytes_avail; 1406 zip->entry_compressed_bytes_read += bytes_avail; 1407 1408 *size = zip->stream.total_out; 1409 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1410 *buff = zip->uncompressed_buffer; 1411 1412 if (zip->end_of_entry && zip->hctx_valid) { 1413 r = check_authentication_code(a, NULL); 1414 if (r != ARCHIVE_OK) 1415 return (r); 1416 } 1417 1418 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1419 const char *p; 1420 1421 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1422 archive_set_error(&a->archive, 1423 ARCHIVE_ERRNO_FILE_FORMAT, 1424 "Truncated ZIP end-of-file record"); 1425 return (ARCHIVE_FATAL); 1426 } 1427 /* Consume the optional PK\007\010 marker. */ 1428 if (p[0] == 'P' && p[1] == 'K' && 1429 p[2] == '\007' && p[3] == '\010') { 1430 p += 4; 1431 zip->unconsumed = 4; 1432 } 1433 if (zip->entry->flags & LA_USED_ZIP64) { 1434 zip->entry->crc32 = archive_le32dec(p); 1435 zip->entry->compressed_size = archive_le64dec(p + 4); 1436 zip->entry->uncompressed_size = archive_le64dec(p + 12); 1437 zip->unconsumed += 20; 1438 } else { 1439 zip->entry->crc32 = archive_le32dec(p); 1440 zip->entry->compressed_size = archive_le32dec(p + 4); 1441 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1442 zip->unconsumed += 12; 1443 } 1444 } 1445 1446 return (ARCHIVE_OK); 1447} 1448#endif 1449 1450static int 1451read_decryption_header(struct archive_read *a) 1452{ 1453 struct zip *zip = (struct zip *)(a->format->data); 1454 const char *p; 1455 unsigned int remaining_size; 1456 unsigned int ts; 1457 1458 /* 1459 * Read an initialization vector data field. 1460 */ 1461 p = __archive_read_ahead(a, 2, NULL); 1462 if (p == NULL) 1463 goto truncated; 1464 ts = zip->iv_size; 1465 zip->iv_size = archive_le16dec(p); 1466 __archive_read_consume(a, 2); 1467 if (ts < zip->iv_size) { 1468 free(zip->iv); 1469 zip->iv = NULL; 1470 } 1471 p = __archive_read_ahead(a, zip->iv_size, NULL); 1472 if (p == NULL) 1473 goto truncated; 1474 if (zip->iv == NULL) { 1475 zip->iv = malloc(zip->iv_size); 1476 if (zip->iv == NULL) 1477 goto nomem; 1478 } 1479 memcpy(zip->iv, p, zip->iv_size); 1480 __archive_read_consume(a, zip->iv_size); 1481 1482 /* 1483 * Read a size of remaining decryption header field. 1484 */ 1485 p = __archive_read_ahead(a, 14, NULL); 1486 if (p == NULL) 1487 goto truncated; 1488 remaining_size = archive_le32dec(p); 1489 if (remaining_size < 16 || remaining_size > (1 << 18)) 1490 goto corrupted; 1491 1492 /* Check if format version is supported. */ 1493 if (archive_le16dec(p+4) != 3) { 1494 archive_set_error(&a->archive, 1495 ARCHIVE_ERRNO_FILE_FORMAT, 1496 "Unsupported encryption format version: %u", 1497 archive_le16dec(p+4)); 1498 return (ARCHIVE_FAILED); 1499 } 1500 1501 /* 1502 * Read an encryption algorithm field. 1503 */ 1504 zip->alg_id = archive_le16dec(p+6); 1505 switch (zip->alg_id) { 1506 case 0x6601:/* DES */ 1507 case 0x6602:/* RC2 */ 1508 case 0x6603:/* 3DES 168 */ 1509 case 0x6609:/* 3DES 112 */ 1510 case 0x660E:/* AES 128 */ 1511 case 0x660F:/* AES 192 */ 1512 case 0x6610:/* AES 256 */ 1513 case 0x6702:/* RC2 (version >= 5.2) */ 1514 case 0x6720:/* Blowfish */ 1515 case 0x6721:/* Twofish */ 1516 case 0x6801:/* RC4 */ 1517 /* Suuported encryption algorithm. */ 1518 break; 1519 default: 1520 archive_set_error(&a->archive, 1521 ARCHIVE_ERRNO_FILE_FORMAT, 1522 "Unknown encryption algorithm: %u", zip->alg_id); 1523 return (ARCHIVE_FAILED); 1524 } 1525 1526 /* 1527 * Read a bit length field. 1528 */ 1529 zip->bit_len = archive_le16dec(p+8); 1530 1531 /* 1532 * Read a flags field. 1533 */ 1534 zip->flags = archive_le16dec(p+10); 1535 switch (zip->flags & 0xf000) { 1536 case 0x0001: /* Password is required to decrypt. */ 1537 case 0x0002: /* Certificates only. */ 1538 case 0x0003: /* Password or certificate required to decrypt. */ 1539 break; 1540 default: 1541 archive_set_error(&a->archive, 1542 ARCHIVE_ERRNO_FILE_FORMAT, 1543 "Unknown encryption flag: %u", zip->flags); 1544 return (ARCHIVE_FAILED); 1545 } 1546 if ((zip->flags & 0xf000) == 0 || 1547 (zip->flags & 0xf000) == 0x4000) { 1548 archive_set_error(&a->archive, 1549 ARCHIVE_ERRNO_FILE_FORMAT, 1550 "Unknown encryption flag: %u", zip->flags); 1551 return (ARCHIVE_FAILED); 1552 } 1553 1554 /* 1555 * Read an encrypted random data field. 1556 */ 1557 ts = zip->erd_size; 1558 zip->erd_size = archive_le16dec(p+12); 1559 __archive_read_consume(a, 14); 1560 if ((zip->erd_size & 0xf) != 0 || 1561 (zip->erd_size + 16) > remaining_size || 1562 (zip->erd_size + 16) < zip->erd_size) 1563 goto corrupted; 1564 1565 if (ts < zip->erd_size) { 1566 free(zip->erd); 1567 zip->erd = NULL; 1568 } 1569 p = __archive_read_ahead(a, zip->erd_size, NULL); 1570 if (p == NULL) 1571 goto truncated; 1572 if (zip->erd == NULL) { 1573 zip->erd = malloc(zip->erd_size); 1574 if (zip->erd == NULL) 1575 goto nomem; 1576 } 1577 memcpy(zip->erd, p, zip->erd_size); 1578 __archive_read_consume(a, zip->erd_size); 1579 1580 /* 1581 * Read a reserved data field. 1582 */ 1583 p = __archive_read_ahead(a, 4, NULL); 1584 if (p == NULL) 1585 goto truncated; 1586 /* Reserved data size should be zero. */ 1587 if (archive_le32dec(p) != 0) 1588 goto corrupted; 1589 __archive_read_consume(a, 4); 1590 1591 /* 1592 * Read a password validation data field. 1593 */ 1594 p = __archive_read_ahead(a, 2, NULL); 1595 if (p == NULL) 1596 goto truncated; 1597 ts = zip->v_size; 1598 zip->v_size = archive_le16dec(p); 1599 __archive_read_consume(a, 2); 1600 if ((zip->v_size & 0x0f) != 0 || 1601 (zip->erd_size + zip->v_size + 16) > remaining_size || 1602 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1603 goto corrupted; 1604 if (ts < zip->v_size) { 1605 free(zip->v_data); 1606 zip->v_data = NULL; 1607 } 1608 p = __archive_read_ahead(a, zip->v_size, NULL); 1609 if (p == NULL) 1610 goto truncated; 1611 if (zip->v_data == NULL) { 1612 zip->v_data = malloc(zip->v_size); 1613 if (zip->v_data == NULL) 1614 goto nomem; 1615 } 1616 memcpy(zip->v_data, p, zip->v_size); 1617 __archive_read_consume(a, zip->v_size); 1618 1619 p = __archive_read_ahead(a, 4, NULL); 1620 if (p == NULL) 1621 goto truncated; 1622 zip->v_crc32 = archive_le32dec(p); 1623 __archive_read_consume(a, 4); 1624 1625 /*return (ARCHIVE_OK); 1626 * This is not fully implemnted yet.*/ 1627 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1628 "Encrypted file is unsupported"); 1629 return (ARCHIVE_FAILED); 1630truncated: 1631 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1632 "Truncated ZIP file data"); 1633 return (ARCHIVE_FATAL); 1634corrupted: 1635 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1636 "Corrupted ZIP file data"); 1637 return (ARCHIVE_FATAL); 1638nomem: 1639 archive_set_error(&a->archive, ENOMEM, 1640 "No memory for ZIP decryption"); 1641 return (ARCHIVE_FATAL); 1642} 1643 1644static int 1645zip_alloc_decryption_buffer(struct archive_read *a) 1646{ 1647 struct zip *zip = (struct zip *)(a->format->data); 1648 size_t bs = 256 * 1024; 1649 1650 if (zip->decrypted_buffer == NULL) { 1651 zip->decrypted_buffer_size = bs; 1652 zip->decrypted_buffer = malloc(bs); 1653 if (zip->decrypted_buffer == NULL) { 1654 archive_set_error(&a->archive, ENOMEM, 1655 "No memory for ZIP decryption"); 1656 return (ARCHIVE_FATAL); 1657 } 1658 } 1659 zip->decrypted_ptr = zip->decrypted_buffer; 1660 return (ARCHIVE_OK); 1661} 1662 1663static int 1664init_traditional_PKWARE_decryption(struct archive_read *a) 1665{ 1666 struct zip *zip = (struct zip *)(a->format->data); 1667 const void *p; 1668 int retry; 1669 int r; 1670 1671 if (zip->tctx_valid) 1672 return (ARCHIVE_OK); 1673 1674 /* 1675 Read the 12 bytes encryption header stored at 1676 the start of the data area. 1677 */ 1678#define ENC_HEADER_SIZE 12 1679 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1680 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1681 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1682 "Truncated Zip encrypted body: only %jd bytes available", 1683 (intmax_t)zip->entry_bytes_remaining); 1684 return (ARCHIVE_FATAL); 1685 } 1686 1687 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1688 if (p == NULL) { 1689 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1690 "Truncated ZIP file data"); 1691 return (ARCHIVE_FATAL); 1692 } 1693 1694 for (retry = 0;; retry++) { 1695 const char *passphrase; 1696 uint8_t crcchk; 1697 1698 passphrase = __archive_read_next_passphrase(a); 1699 if (passphrase == NULL) { 1700 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1701 (retry > 0)? 1702 "Incorrect passphrase": 1703 "Passphrase required for this entry"); 1704 return (ARCHIVE_FAILED); 1705 } 1706 1707 /* 1708 * Initialize ctx for Traditional PKWARE Decyption. 1709 */ 1710 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1711 p, ENC_HEADER_SIZE, &crcchk); 1712 if (r == 0 && crcchk == zip->entry->decdat) 1713 break;/* The passphrase is OK. */ 1714 if (retry > 10000) { 1715 /* Avoid infinity loop. */ 1716 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1717 "Too many incorrect passphrases"); 1718 return (ARCHIVE_FAILED); 1719 } 1720 } 1721 1722 __archive_read_consume(a, ENC_HEADER_SIZE); 1723 zip->tctx_valid = 1; 1724 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1725 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1726 } 1727 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1728 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1729 zip->decrypted_bytes_remaining = 0; 1730 1731 return (zip_alloc_decryption_buffer(a)); 1732#undef ENC_HEADER_SIZE 1733} 1734 1735static int 1736init_WinZip_AES_decryption(struct archive_read *a) 1737{ 1738 struct zip *zip = (struct zip *)(a->format->data); 1739 const void *p; 1740 const uint8_t *pv; 1741 size_t key_len, salt_len; 1742 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1743 int retry; 1744 int r; 1745 1746 if (zip->cctx_valid || zip->hctx_valid) 1747 return (ARCHIVE_OK); 1748 1749 switch (zip->entry->aes_extra.strength) { 1750 case 1: salt_len = 8; key_len = 16; break; 1751 case 2: salt_len = 12; key_len = 24; break; 1752 case 3: salt_len = 16; key_len = 32; break; 1753 default: goto corrupted; 1754 } 1755 p = __archive_read_ahead(a, salt_len + 2, NULL); 1756 if (p == NULL) 1757 goto truncated; 1758 1759 for (retry = 0;; retry++) { 1760 const char *passphrase; 1761 1762 passphrase = __archive_read_next_passphrase(a); 1763 if (passphrase == NULL) { 1764 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1765 (retry > 0)? 1766 "Incorrect passphrase": 1767 "Passphrase required for this entry"); 1768 return (ARCHIVE_FAILED); 1769 } 1770 memset(derived_key, 0, sizeof(derived_key)); 1771 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1772 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1773 if (r != 0) { 1774 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1775 "Decryption is unsupported due to lack of " 1776 "crypto library"); 1777 return (ARCHIVE_FAILED); 1778 } 1779 1780 /* Check password verification value. */ 1781 pv = ((const uint8_t *)p) + salt_len; 1782 if (derived_key[key_len * 2] == pv[0] && 1783 derived_key[key_len * 2 + 1] == pv[1]) 1784 break;/* The passphrase is OK. */ 1785 if (retry > 10000) { 1786 /* Avoid infinity loop. */ 1787 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1788 "Too many incorrect passphrases"); 1789 return (ARCHIVE_FAILED); 1790 } 1791 } 1792 1793 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1794 if (r != 0) { 1795 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1796 "Decryption is unsupported due to lack of crypto library"); 1797 return (ARCHIVE_FAILED); 1798 } 1799 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1800 if (r != 0) { 1801 archive_decrypto_aes_ctr_release(&zip->cctx); 1802 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1803 "Failed to initialize HMAC-SHA1"); 1804 return (ARCHIVE_FAILED); 1805 } 1806 zip->cctx_valid = zip->hctx_valid = 1; 1807 __archive_read_consume(a, salt_len + 2); 1808 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1809 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1810 && zip->entry_bytes_remaining < 0) 1811 goto corrupted; 1812 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1813 zip->decrypted_bytes_remaining = 0; 1814 1815 zip->entry->compression = zip->entry->aes_extra.compression; 1816 return (zip_alloc_decryption_buffer(a)); 1817 1818truncated: 1819 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1820 "Truncated ZIP file data"); 1821 return (ARCHIVE_FATAL); 1822corrupted: 1823 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1824 "Corrupted ZIP file data"); 1825 return (ARCHIVE_FATAL); 1826} 1827 1828static int 1829archive_read_format_zip_read_data(struct archive_read *a, 1830 const void **buff, size_t *size, int64_t *offset) 1831{ 1832 int r; 1833 struct zip *zip = (struct zip *)(a->format->data); 1834 1835 if (zip->has_encrypted_entries == 1836 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1837 zip->has_encrypted_entries = 0; 1838 } 1839 1840 *offset = zip->entry_uncompressed_bytes_read; 1841 *size = 0; 1842 *buff = NULL; 1843 1844 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1845 if (zip->end_of_entry) 1846 return (ARCHIVE_EOF); 1847 1848 /* Return EOF immediately if this is a non-regular file. */ 1849 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1850 return (ARCHIVE_EOF); 1851 1852 __archive_read_consume(a, zip->unconsumed); 1853 zip->unconsumed = 0; 1854 1855 if (zip->init_decryption) { 1856 zip->has_encrypted_entries = 1; 1857 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1858 r = read_decryption_header(a); 1859 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1860 r = init_WinZip_AES_decryption(a); 1861 else 1862 r = init_traditional_PKWARE_decryption(a); 1863 if (r != ARCHIVE_OK) 1864 return (r); 1865 zip->init_decryption = 0; 1866 } 1867 1868 switch(zip->entry->compression) { 1869 case 0: /* No compression. */ 1870 r = zip_read_data_none(a, buff, size, offset); 1871 break; 1872#ifdef HAVE_ZLIB_H 1873 case 8: /* Deflate compression. */ 1874 r = zip_read_data_deflate(a, buff, size, offset); 1875 break; 1876#endif 1877 default: /* Unsupported compression. */ 1878 /* Return a warning. */ 1879 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1880 "Unsupported ZIP compression method (%s)", 1881 compression_name(zip->entry->compression)); 1882 /* We can't decompress this entry, but we will 1883 * be able to skip() it and try the next entry. */ 1884 return (ARCHIVE_FAILED); 1885 break; 1886 } 1887 if (r != ARCHIVE_OK) 1888 return (r); 1889 /* Update checksum */ 1890 if (*size) 1891 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1892 (unsigned)*size); 1893 /* If we hit the end, swallow any end-of-data marker. */ 1894 if (zip->end_of_entry) { 1895 /* Check file size, CRC against these values. */ 1896 if (zip->entry->compressed_size != 1897 zip->entry_compressed_bytes_read) { 1898 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1899 "ZIP compressed data is wrong size " 1900 "(read %jd, expected %jd)", 1901 (intmax_t)zip->entry_compressed_bytes_read, 1902 (intmax_t)zip->entry->compressed_size); 1903 return (ARCHIVE_WARN); 1904 } 1905 /* Size field only stores the lower 32 bits of the actual 1906 * size. */ 1907 if ((zip->entry->uncompressed_size & UINT32_MAX) 1908 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1909 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1910 "ZIP uncompressed data is wrong size " 1911 "(read %jd, expected %jd)\n", 1912 (intmax_t)zip->entry_uncompressed_bytes_read, 1913 (intmax_t)zip->entry->uncompressed_size); 1914 return (ARCHIVE_WARN); 1915 } 1916 /* Check computed CRC against header */ 1917 if ((!zip->hctx_valid || 1918 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1919 zip->entry->crc32 != zip->entry_crc32 1920 && !zip->ignore_crc32) { 1921 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1922 "ZIP bad CRC: 0x%lx should be 0x%lx", 1923 (unsigned long)zip->entry_crc32, 1924 (unsigned long)zip->entry->crc32); 1925 return (ARCHIVE_WARN); 1926 } 1927 } 1928 1929 return (ARCHIVE_OK); 1930} 1931 1932static int 1933archive_read_format_zip_cleanup(struct archive_read *a) 1934{ 1935 struct zip *zip; 1936 struct zip_entry *zip_entry, *next_zip_entry; 1937 1938 zip = (struct zip *)(a->format->data); 1939#ifdef HAVE_ZLIB_H 1940 if (zip->stream_valid) 1941 inflateEnd(&zip->stream); 1942 free(zip->uncompressed_buffer); 1943#endif 1944 if (zip->zip_entries) { 1945 zip_entry = zip->zip_entries; 1946 while (zip_entry != NULL) { 1947 next_zip_entry = zip_entry->next; 1948 archive_string_free(&zip_entry->rsrcname); 1949 free(zip_entry); 1950 zip_entry = next_zip_entry; 1951 } 1952 } 1953 free(zip->decrypted_buffer); 1954 if (zip->cctx_valid) 1955 archive_decrypto_aes_ctr_release(&zip->cctx); 1956 if (zip->hctx_valid) 1957 archive_hmac_sha1_cleanup(&zip->hctx); 1958 free(zip->iv); 1959 free(zip->erd); 1960 free(zip->v_data); 1961 archive_string_free(&zip->format_name); 1962 free(zip); 1963 (a->format->data) = NULL; 1964 return (ARCHIVE_OK); 1965} 1966 1967static int 1968archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 1969{ 1970 if (_a && _a->format) { 1971 struct zip * zip = (struct zip *)_a->format->data; 1972 if (zip) { 1973 return zip->has_encrypted_entries; 1974 } 1975 } 1976 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 1977} 1978 1979static int 1980archive_read_format_zip_options(struct archive_read *a, 1981 const char *key, const char *val) 1982{ 1983 struct zip *zip; 1984 int ret = ARCHIVE_FAILED; 1985 1986 zip = (struct zip *)(a->format->data); 1987 if (strcmp(key, "compat-2x") == 0) { 1988 /* Handle filenames as libarchive 2.x */ 1989 zip->init_default_conversion = (val != NULL) ? 1 : 0; 1990 return (ARCHIVE_OK); 1991 } else if (strcmp(key, "hdrcharset") == 0) { 1992 if (val == NULL || val[0] == 0) 1993 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1994 "zip: hdrcharset option needs a character-set name" 1995 ); 1996 else { 1997 zip->sconv = archive_string_conversion_from_charset( 1998 &a->archive, val, 0); 1999 if (zip->sconv != NULL) { 2000 if (strcmp(val, "UTF-8") == 0) 2001 zip->sconv_utf8 = zip->sconv; 2002 ret = ARCHIVE_OK; 2003 } else 2004 ret = ARCHIVE_FATAL; 2005 } 2006 return (ret); 2007 } else if (strcmp(key, "ignorecrc32") == 0) { 2008 /* Mostly useful for testing. */ 2009 if (val == NULL || val[0] == 0) { 2010 zip->crc32func = real_crc32; 2011 zip->ignore_crc32 = 0; 2012 } else { 2013 zip->crc32func = fake_crc32; 2014 zip->ignore_crc32 = 1; 2015 } 2016 return (ARCHIVE_OK); 2017 } else if (strcmp(key, "mac-ext") == 0) { 2018 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2019 return (ARCHIVE_OK); 2020 } 2021 2022 /* Note: The "warn" return is just to inform the options 2023 * supervisor that we didn't handle it. It will generate 2024 * a suitable error if no one used this option. */ 2025 return (ARCHIVE_WARN); 2026} 2027 2028int 2029archive_read_support_format_zip(struct archive *a) 2030{ 2031 int r; 2032 r = archive_read_support_format_zip_streamable(a); 2033 if (r != ARCHIVE_OK) 2034 return r; 2035 return (archive_read_support_format_zip_seekable(a)); 2036} 2037 2038/* ------------------------------------------------------------------------ */ 2039 2040/* 2041 * Streaming-mode support 2042 */ 2043 2044 2045static int 2046archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2047{ 2048 (void)a; /* UNUSED */ 2049 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2050 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2051} 2052 2053static int 2054archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2055{ 2056 const char *p; 2057 2058 (void)best_bid; /* UNUSED */ 2059 2060 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2061 return (-1); 2062 2063 /* 2064 * Bid of 29 here comes from: 2065 * + 16 bits for "PK", 2066 * + next 16-bit field has 6 options so contributes 2067 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2068 * 2069 * So we've effectively verified ~29 total bits of check data. 2070 */ 2071 if (p[0] == 'P' && p[1] == 'K') { 2072 if ((p[2] == '\001' && p[3] == '\002') 2073 || (p[2] == '\003' && p[3] == '\004') 2074 || (p[2] == '\005' && p[3] == '\006') 2075 || (p[2] == '\006' && p[3] == '\006') 2076 || (p[2] == '\007' && p[3] == '\010') 2077 || (p[2] == '0' && p[3] == '0')) 2078 return (29); 2079 } 2080 2081 /* TODO: It's worth looking ahead a little bit for a valid 2082 * PK signature. In particular, that would make it possible 2083 * to read some UUEncoded SFX files or SFX files coming from 2084 * a network socket. */ 2085 2086 return (0); 2087} 2088 2089static int 2090archive_read_format_zip_streamable_read_header(struct archive_read *a, 2091 struct archive_entry *entry) 2092{ 2093 struct zip *zip; 2094 2095 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2096 if (a->archive.archive_format_name == NULL) 2097 a->archive.archive_format_name = "ZIP"; 2098 2099 zip = (struct zip *)(a->format->data); 2100 2101 /* 2102 * It should be sufficient to call archive_read_next_header() for 2103 * a reader to determine if an entry is encrypted or not. If the 2104 * encryption of an entry is only detectable when calling 2105 * archive_read_data(), so be it. We'll do the same check there 2106 * as well. 2107 */ 2108 if (zip->has_encrypted_entries == 2109 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2110 zip->has_encrypted_entries = 0; 2111 2112 /* Make sure we have a zip_entry structure to use. */ 2113 if (zip->zip_entries == NULL) { 2114 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2115 if (zip->zip_entries == NULL) { 2116 archive_set_error(&a->archive, ENOMEM, 2117 "Out of memory"); 2118 return ARCHIVE_FATAL; 2119 } 2120 } 2121 zip->entry = zip->zip_entries; 2122 memset(zip->entry, 0, sizeof(struct zip_entry)); 2123 2124 if (zip->cctx_valid) 2125 archive_decrypto_aes_ctr_release(&zip->cctx); 2126 if (zip->hctx_valid) 2127 archive_hmac_sha1_cleanup(&zip->hctx); 2128 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2129 __archive_read_reset_passphrase(a); 2130 2131 /* Search ahead for the next local file header. */ 2132 __archive_read_consume(a, zip->unconsumed); 2133 zip->unconsumed = 0; 2134 for (;;) { 2135 int64_t skipped = 0; 2136 const char *p, *end; 2137 ssize_t bytes; 2138 2139 p = __archive_read_ahead(a, 4, &bytes); 2140 if (p == NULL) 2141 return (ARCHIVE_FATAL); 2142 end = p + bytes; 2143 2144 while (p + 4 <= end) { 2145 if (p[0] == 'P' && p[1] == 'K') { 2146 if (p[2] == '\003' && p[3] == '\004') { 2147 /* Regular file entry. */ 2148 __archive_read_consume(a, skipped); 2149 return zip_read_local_file_header(a, 2150 entry, zip); 2151 } 2152 2153 /* 2154 * TODO: We cannot restore permissions 2155 * based only on the local file headers. 2156 * Consider scanning the central 2157 * directory and returning additional 2158 * entries for at least directories. 2159 * This would allow us to properly set 2160 * directory permissions. 2161 * 2162 * This won't help us fix symlinks 2163 * and may not help with regular file 2164 * permissions, either. <sigh> 2165 */ 2166 if (p[2] == '\001' && p[3] == '\002') { 2167 return (ARCHIVE_EOF); 2168 } 2169 2170 /* End of central directory? Must be an 2171 * empty archive. */ 2172 if ((p[2] == '\005' && p[3] == '\006') 2173 || (p[2] == '\006' && p[3] == '\006')) 2174 return (ARCHIVE_EOF); 2175 } 2176 ++p; 2177 ++skipped; 2178 } 2179 __archive_read_consume(a, skipped); 2180 } 2181} 2182 2183static int 2184archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2185{ 2186 struct zip *zip; 2187 int64_t bytes_skipped; 2188 2189 zip = (struct zip *)(a->format->data); 2190 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2191 zip->unconsumed = 0; 2192 if (bytes_skipped < 0) 2193 return (ARCHIVE_FATAL); 2194 2195 /* If we've already read to end of data, we're done. */ 2196 if (zip->end_of_entry) 2197 return (ARCHIVE_OK); 2198 2199 /* So we know we're streaming... */ 2200 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2201 || zip->entry->compressed_size > 0) { 2202 /* We know the compressed length, so we can just skip. */ 2203 bytes_skipped = __archive_read_consume(a, 2204 zip->entry_bytes_remaining); 2205 if (bytes_skipped < 0) 2206 return (ARCHIVE_FATAL); 2207 return (ARCHIVE_OK); 2208 } 2209 2210 if (zip->init_decryption) { 2211 int r; 2212 2213 zip->has_encrypted_entries = 1; 2214 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2215 r = read_decryption_header(a); 2216 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2217 r = init_WinZip_AES_decryption(a); 2218 else 2219 r = init_traditional_PKWARE_decryption(a); 2220 if (r != ARCHIVE_OK) 2221 return (r); 2222 zip->init_decryption = 0; 2223 } 2224 2225 /* We're streaming and we don't know the length. */ 2226 /* If the body is compressed and we know the format, we can 2227 * find an exact end-of-entry by decompressing it. */ 2228 switch (zip->entry->compression) { 2229#ifdef HAVE_ZLIB_H 2230 case 8: /* Deflate compression. */ 2231 while (!zip->end_of_entry) { 2232 int64_t offset = 0; 2233 const void *buff = NULL; 2234 size_t size = 0; 2235 int r; 2236 r = zip_read_data_deflate(a, &buff, &size, &offset); 2237 if (r != ARCHIVE_OK) 2238 return (r); 2239 } 2240 return ARCHIVE_OK; 2241#endif 2242 default: /* Uncompressed or unknown. */ 2243 /* Scan for a PK\007\010 signature. */ 2244 for (;;) { 2245 const char *p, *buff; 2246 ssize_t bytes_avail; 2247 buff = __archive_read_ahead(a, 16, &bytes_avail); 2248 if (bytes_avail < 16) { 2249 archive_set_error(&a->archive, 2250 ARCHIVE_ERRNO_FILE_FORMAT, 2251 "Truncated ZIP file data"); 2252 return (ARCHIVE_FATAL); 2253 } 2254 p = buff; 2255 while (p <= buff + bytes_avail - 16) { 2256 if (p[3] == 'P') { p += 3; } 2257 else if (p[3] == 'K') { p += 2; } 2258 else if (p[3] == '\007') { p += 1; } 2259 else if (p[3] == '\010' && p[2] == '\007' 2260 && p[1] == 'K' && p[0] == 'P') { 2261 if (zip->entry->flags & LA_USED_ZIP64) 2262 __archive_read_consume(a, 2263 p - buff + 24); 2264 else 2265 __archive_read_consume(a, 2266 p - buff + 16); 2267 return ARCHIVE_OK; 2268 } else { p += 4; } 2269 } 2270 __archive_read_consume(a, p - buff); 2271 } 2272 } 2273} 2274 2275int 2276archive_read_support_format_zip_streamable(struct archive *_a) 2277{ 2278 struct archive_read *a = (struct archive_read *)_a; 2279 struct zip *zip; 2280 int r; 2281 2282 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2283 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2284 2285 zip = (struct zip *)calloc(1, sizeof(*zip)); 2286 if (zip == NULL) { 2287 archive_set_error(&a->archive, ENOMEM, 2288 "Can't allocate zip data"); 2289 return (ARCHIVE_FATAL); 2290 } 2291 2292 /* Streamable reader doesn't support mac extensions. */ 2293 zip->process_mac_extensions = 0; 2294 2295 /* 2296 * Until enough data has been read, we cannot tell about 2297 * any encrypted entries yet. 2298 */ 2299 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2300 zip->crc32func = real_crc32; 2301 2302 r = __archive_read_register_format(a, 2303 zip, 2304 "zip", 2305 archive_read_format_zip_streamable_bid, 2306 archive_read_format_zip_options, 2307 archive_read_format_zip_streamable_read_header, 2308 archive_read_format_zip_read_data, 2309 archive_read_format_zip_read_data_skip_streamable, 2310 NULL, 2311 archive_read_format_zip_cleanup, 2312 archive_read_support_format_zip_capabilities_streamable, 2313 archive_read_format_zip_has_encrypted_entries); 2314 2315 if (r != ARCHIVE_OK) 2316 free(zip); 2317 return (ARCHIVE_OK); 2318} 2319 2320/* ------------------------------------------------------------------------ */ 2321 2322/* 2323 * Seeking-mode support 2324 */ 2325 2326static int 2327archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2328{ 2329 (void)a; /* UNUSED */ 2330 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2331 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2332} 2333 2334/* 2335 * TODO: This is a performance sink because it forces the read core to 2336 * drop buffered data from the start of file, which will then have to 2337 * be re-read again if this bidder loses. 2338 * 2339 * We workaround this a little by passing in the best bid so far so 2340 * that later bidders can do nothing if they know they'll never 2341 * outbid. But we can certainly do better... 2342 */ 2343static int 2344read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2345{ 2346 /* Sanity-check the EOCD we've found. */ 2347 2348 /* This must be the first volume. */ 2349 if (archive_le16dec(p + 4) != 0) 2350 return 0; 2351 /* Central directory must be on this volume. */ 2352 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2353 return 0; 2354 /* All central directory entries must be on this volume. */ 2355 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2356 return 0; 2357 /* Central directory can't extend beyond start of EOCD record. */ 2358 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2359 > current_offset) 2360 return 0; 2361 2362 /* Save the central directory location for later use. */ 2363 zip->central_directory_offset = archive_le32dec(p + 16); 2364 2365 /* This is just a tiny bit higher than the maximum 2366 returned by the streaming Zip bidder. This ensures 2367 that the more accurate seeking Zip parser wins 2368 whenever seek is available. */ 2369 return 32; 2370} 2371 2372/* 2373 * Examine Zip64 EOCD locator: If it's valid, store the information 2374 * from it. 2375 */ 2376static void 2377read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2378{ 2379 int64_t eocd64_offset; 2380 int64_t eocd64_size; 2381 2382 /* Sanity-check the locator record. */ 2383 2384 /* Central dir must be on first volume. */ 2385 if (archive_le32dec(p + 4) != 0) 2386 return; 2387 /* Must be only a single volume. */ 2388 if (archive_le32dec(p + 16) != 1) 2389 return; 2390 2391 /* Find the Zip64 EOCD record. */ 2392 eocd64_offset = archive_le64dec(p + 8); 2393 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2394 return; 2395 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2396 return; 2397 /* Make sure we can read all of it. */ 2398 eocd64_size = archive_le64dec(p + 4) + 12; 2399 if (eocd64_size < 56 || eocd64_size > 16384) 2400 return; 2401 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2402 return; 2403 2404 /* Sanity-check the EOCD64 */ 2405 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2406 return; 2407 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2408 return; 2409 /* CD can't be split. */ 2410 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2411 return; 2412 2413 /* Save the central directory offset for later use. */ 2414 zip->central_directory_offset = archive_le64dec(p + 48); 2415} 2416 2417static int 2418archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2419{ 2420 struct zip *zip = (struct zip *)a->format->data; 2421 int64_t file_size, current_offset; 2422 const char *p; 2423 int i, tail; 2424 2425 /* If someone has already bid more than 32, then avoid 2426 trashing the look-ahead buffers with a seek. */ 2427 if (best_bid > 32) 2428 return (-1); 2429 2430 file_size = __archive_read_seek(a, 0, SEEK_END); 2431 if (file_size <= 0) 2432 return 0; 2433 2434 /* Search last 16k of file for end-of-central-directory 2435 * record (which starts with PK\005\006) */ 2436 tail = (int)zipmin(1024 * 16, file_size); 2437 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2438 if (current_offset < 0) 2439 return 0; 2440 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2441 return 0; 2442 /* Boyer-Moore search backwards from the end, since we want 2443 * to match the last EOCD in the file (there can be more than 2444 * one if there is an uncompressed Zip archive as a member 2445 * within this Zip archive). */ 2446 for (i = tail - 22; i > 0;) { 2447 switch (p[i]) { 2448 case 'P': 2449 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2450 int ret = read_eocd(zip, p + i, 2451 current_offset + i); 2452 if (ret > 0) { 2453 /* Zip64 EOCD locator precedes 2454 * regular EOCD if present. */ 2455 if (i >= 20 2456 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2457 read_zip64_eocd(a, zip, p + i - 20); 2458 } 2459 return (ret); 2460 } 2461 } 2462 i -= 4; 2463 break; 2464 case 'K': i -= 1; break; 2465 case 005: i -= 2; break; 2466 case 006: i -= 3; break; 2467 default: i -= 4; break; 2468 } 2469 } 2470 return 0; 2471} 2472 2473/* The red-black trees are only used in seeking mode to manage 2474 * the in-memory copy of the central directory. */ 2475 2476static int 2477cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2478{ 2479 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2480 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2481 2482 if (e1->local_header_offset > e2->local_header_offset) 2483 return -1; 2484 if (e1->local_header_offset < e2->local_header_offset) 2485 return 1; 2486 return 0; 2487} 2488 2489static int 2490cmp_key(const struct archive_rb_node *n, const void *key) 2491{ 2492 /* This function won't be called */ 2493 (void)n; /* UNUSED */ 2494 (void)key; /* UNUSED */ 2495 return 1; 2496} 2497 2498static const struct archive_rb_tree_ops rb_ops = { 2499 &cmp_node, &cmp_key 2500}; 2501 2502static int 2503rsrc_cmp_node(const struct archive_rb_node *n1, 2504 const struct archive_rb_node *n2) 2505{ 2506 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2507 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2508 2509 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2510} 2511 2512static int 2513rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2514{ 2515 const struct zip_entry *e = (const struct zip_entry *)n; 2516 return (strcmp((const char *)key, e->rsrcname.s)); 2517} 2518 2519static const struct archive_rb_tree_ops rb_rsrc_ops = { 2520 &rsrc_cmp_node, &rsrc_cmp_key 2521}; 2522 2523static const char * 2524rsrc_basename(const char *name, size_t name_length) 2525{ 2526 const char *s, *r; 2527 2528 r = s = name; 2529 for (;;) { 2530 s = memchr(s, '/', name_length - (s - name)); 2531 if (s == NULL) 2532 break; 2533 r = ++s; 2534 } 2535 return (r); 2536} 2537 2538static void 2539expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2540{ 2541 struct archive_string str; 2542 struct zip_entry *dir; 2543 char *s; 2544 2545 archive_string_init(&str); 2546 archive_strncpy(&str, name, name_length); 2547 for (;;) { 2548 s = strrchr(str.s, '/'); 2549 if (s == NULL) 2550 break; 2551 *s = '\0'; 2552 /* Transfer the parent directory from zip->tree_rsrc RB 2553 * tree to zip->tree RB tree to expose. */ 2554 dir = (struct zip_entry *) 2555 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2556 if (dir == NULL) 2557 break; 2558 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2559 archive_string_free(&dir->rsrcname); 2560 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2561 } 2562 archive_string_free(&str); 2563} 2564 2565static int 2566slurp_central_directory(struct archive_read *a, struct zip *zip) 2567{ 2568 ssize_t i; 2569 unsigned found; 2570 int64_t correction; 2571 ssize_t bytes_avail; 2572 const char *p; 2573 2574 /* 2575 * Find the start of the central directory. The end-of-CD 2576 * record has our starting point, but there are lots of 2577 * Zip archives which have had other data prepended to the 2578 * file, which makes the recorded offsets all too small. 2579 * So we search forward from the specified offset until we 2580 * find the real start of the central directory. Then we 2581 * know the correction we need to apply to account for leading 2582 * padding. 2583 */ 2584 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2585 return ARCHIVE_FATAL; 2586 2587 found = 0; 2588 while (!found) { 2589 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2590 return ARCHIVE_FATAL; 2591 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2592 switch (p[i + 3]) { 2593 case 'P': i += 3; break; 2594 case 'K': i += 2; break; 2595 case 001: i += 1; break; 2596 case 002: 2597 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2598 p += i; 2599 found = 1; 2600 } else 2601 i += 4; 2602 break; 2603 case 005: i += 1; break; 2604 case 006: 2605 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2606 p += i; 2607 found = 1; 2608 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2609 p += i; 2610 found = 1; 2611 } else 2612 i += 1; 2613 break; 2614 default: i += 4; break; 2615 } 2616 } 2617 __archive_read_consume(a, i); 2618 } 2619 correction = archive_filter_bytes(&a->archive, 0) 2620 - zip->central_directory_offset; 2621 2622 __archive_rb_tree_init(&zip->tree, &rb_ops); 2623 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2624 2625 zip->central_directory_entries_total = 0; 2626 while (1) { 2627 struct zip_entry *zip_entry; 2628 size_t filename_length, extra_length, comment_length; 2629 uint32_t external_attributes; 2630 const char *name, *r; 2631 2632 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2633 return ARCHIVE_FATAL; 2634 if (memcmp(p, "PK\006\006", 4) == 0 2635 || memcmp(p, "PK\005\006", 4) == 0) { 2636 break; 2637 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2638 archive_set_error(&a->archive, 2639 -1, "Invalid central directory signature"); 2640 return ARCHIVE_FATAL; 2641 } 2642 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2643 return ARCHIVE_FATAL; 2644 2645 zip_entry = calloc(1, sizeof(struct zip_entry)); 2646 zip_entry->next = zip->zip_entries; 2647 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2648 zip->zip_entries = zip_entry; 2649 zip->central_directory_entries_total++; 2650 2651 /* version = p[4]; */ 2652 zip_entry->system = p[5]; 2653 /* version_required = archive_le16dec(p + 6); */ 2654 zip_entry->zip_flags = archive_le16dec(p + 8); 2655 if (zip_entry->zip_flags 2656 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2657 zip->has_encrypted_entries = 1; 2658 } 2659 zip_entry->compression = (char)archive_le16dec(p + 10); 2660 zip_entry->mtime = zip_time(p + 12); 2661 zip_entry->crc32 = archive_le32dec(p + 16); 2662 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2663 zip_entry->decdat = p[13]; 2664 else 2665 zip_entry->decdat = p[19]; 2666 zip_entry->compressed_size = archive_le32dec(p + 20); 2667 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2668 filename_length = archive_le16dec(p + 28); 2669 extra_length = archive_le16dec(p + 30); 2670 comment_length = archive_le16dec(p + 32); 2671 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2672 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2673 external_attributes = archive_le32dec(p + 38); 2674 zip_entry->local_header_offset = 2675 archive_le32dec(p + 42) + correction; 2676 2677 /* If we can't guess the mode, leave it zero here; 2678 when we read the local file header we might get 2679 more information. */ 2680 if (zip_entry->system == 3) { 2681 zip_entry->mode = external_attributes >> 16; 2682 } else if (zip_entry->system == 0) { 2683 // Interpret MSDOS directory bit 2684 if (0x10 == (external_attributes & 0x10)) { 2685 zip_entry->mode = AE_IFDIR | 0775; 2686 } else { 2687 zip_entry->mode = AE_IFREG | 0664; 2688 } 2689 if (0x01 == (external_attributes & 0x01)) { 2690 // Read-only bit; strip write permissions 2691 zip_entry->mode &= 0555; 2692 } 2693 } else { 2694 zip_entry->mode = 0; 2695 } 2696 2697 /* We're done with the regular data; get the filename and 2698 * extra data. */ 2699 __archive_read_consume(a, 46); 2700 p = __archive_read_ahead(a, filename_length + extra_length, 2701 NULL); 2702 if (p == NULL) { 2703 archive_set_error(&a->archive, 2704 ARCHIVE_ERRNO_FILE_FORMAT, 2705 "Truncated ZIP file header"); 2706 return ARCHIVE_FATAL; 2707 } 2708 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2709 return ARCHIVE_FATAL; 2710 } 2711 2712 /* 2713 * Mac resource fork files are stored under the 2714 * "__MACOSX/" directory, so we should check if 2715 * it is. 2716 */ 2717 if (!zip->process_mac_extensions) { 2718 /* Treat every entry as a regular entry. */ 2719 __archive_rb_tree_insert_node(&zip->tree, 2720 &zip_entry->node); 2721 } else { 2722 name = p; 2723 r = rsrc_basename(name, filename_length); 2724 if (filename_length >= 9 && 2725 strncmp("__MACOSX/", name, 9) == 0) { 2726 /* If this file is not a resource fork nor 2727 * a directory. We should treat it as a non 2728 * resource fork file to expose it. */ 2729 if (name[filename_length-1] != '/' && 2730 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2731 __archive_rb_tree_insert_node( 2732 &zip->tree, &zip_entry->node); 2733 /* Expose its parent directories. */ 2734 expose_parent_dirs(zip, name, 2735 filename_length); 2736 } else { 2737 /* This file is a resource fork file or 2738 * a directory. */ 2739 archive_strncpy(&(zip_entry->rsrcname), 2740 name, filename_length); 2741 __archive_rb_tree_insert_node( 2742 &zip->tree_rsrc, &zip_entry->node); 2743 } 2744 } else { 2745 /* Generate resource fork name to find its 2746 * resource file at zip->tree_rsrc. */ 2747 archive_strcpy(&(zip_entry->rsrcname), 2748 "__MACOSX/"); 2749 archive_strncat(&(zip_entry->rsrcname), 2750 name, r - name); 2751 archive_strcat(&(zip_entry->rsrcname), "._"); 2752 archive_strncat(&(zip_entry->rsrcname), 2753 name + (r - name), 2754 filename_length - (r - name)); 2755 /* Register an entry to RB tree to sort it by 2756 * file offset. */ 2757 __archive_rb_tree_insert_node(&zip->tree, 2758 &zip_entry->node); 2759 } 2760 } 2761 2762 /* Skip the comment too ... */ 2763 __archive_read_consume(a, 2764 filename_length + extra_length + comment_length); 2765 } 2766 2767 return ARCHIVE_OK; 2768} 2769 2770static ssize_t 2771zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2772{ 2773 const char *p; 2774 ssize_t filename_length, extra_length; 2775 2776 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2777 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2778 "Truncated ZIP file header"); 2779 return (ARCHIVE_WARN); 2780 } 2781 p += extra; 2782 2783 if (memcmp(p, "PK\003\004", 4) != 0) { 2784 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2785 return ARCHIVE_WARN; 2786 } 2787 filename_length = archive_le16dec(p + 26); 2788 extra_length = archive_le16dec(p + 28); 2789 2790 return (30 + filename_length + extra_length); 2791} 2792 2793static int 2794zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2795 struct zip_entry *rsrc) 2796{ 2797 struct zip *zip = (struct zip *)a->format->data; 2798 unsigned char *metadata, *mp; 2799 int64_t offset = archive_filter_bytes(&a->archive, 0); 2800 size_t remaining_bytes, metadata_bytes; 2801 ssize_t hsize; 2802 int ret = ARCHIVE_OK, eof; 2803 2804 switch(rsrc->compression) { 2805 case 0: /* No compression. */ 2806 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2807 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2808 "Malformed OS X metadata entry: inconsistent size"); 2809 return (ARCHIVE_FATAL); 2810 } 2811#ifdef HAVE_ZLIB_H 2812 case 8: /* Deflate compression. */ 2813#endif 2814 break; 2815 default: /* Unsupported compression. */ 2816 /* Return a warning. */ 2817 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2818 "Unsupported ZIP compression method (%s)", 2819 compression_name(rsrc->compression)); 2820 /* We can't decompress this entry, but we will 2821 * be able to skip() it and try the next entry. */ 2822 return (ARCHIVE_WARN); 2823 } 2824 2825 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2826 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2827 "Mac metadata is too large: %jd > 4M bytes", 2828 (intmax_t)rsrc->uncompressed_size); 2829 return (ARCHIVE_WARN); 2830 } 2831 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2832 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2833 "Mac metadata is too large: %jd > 4M bytes", 2834 (intmax_t)rsrc->compressed_size); 2835 return (ARCHIVE_WARN); 2836 } 2837 2838 metadata = malloc((size_t)rsrc->uncompressed_size); 2839 if (metadata == NULL) { 2840 archive_set_error(&a->archive, ENOMEM, 2841 "Can't allocate memory for Mac metadata"); 2842 return (ARCHIVE_FATAL); 2843 } 2844 2845 if (offset < rsrc->local_header_offset) 2846 __archive_read_consume(a, rsrc->local_header_offset - offset); 2847 else if (offset != rsrc->local_header_offset) { 2848 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2849 } 2850 2851 hsize = zip_get_local_file_header_size(a, 0); 2852 __archive_read_consume(a, hsize); 2853 2854 remaining_bytes = (size_t)rsrc->compressed_size; 2855 metadata_bytes = (size_t)rsrc->uncompressed_size; 2856 mp = metadata; 2857 eof = 0; 2858 while (!eof && remaining_bytes) { 2859 const unsigned char *p; 2860 ssize_t bytes_avail; 2861 size_t bytes_used; 2862 2863 p = __archive_read_ahead(a, 1, &bytes_avail); 2864 if (p == NULL) { 2865 archive_set_error(&a->archive, 2866 ARCHIVE_ERRNO_FILE_FORMAT, 2867 "Truncated ZIP file header"); 2868 ret = ARCHIVE_WARN; 2869 goto exit_mac_metadata; 2870 } 2871 if ((size_t)bytes_avail > remaining_bytes) 2872 bytes_avail = remaining_bytes; 2873 switch(rsrc->compression) { 2874 case 0: /* No compression. */ 2875 if ((size_t)bytes_avail > metadata_bytes) 2876 bytes_avail = metadata_bytes; 2877 memcpy(mp, p, bytes_avail); 2878 bytes_used = (size_t)bytes_avail; 2879 metadata_bytes -= bytes_used; 2880 mp += bytes_used; 2881 if (metadata_bytes == 0) 2882 eof = 1; 2883 break; 2884#ifdef HAVE_ZLIB_H 2885 case 8: /* Deflate compression. */ 2886 { 2887 int r; 2888 2889 ret = zip_deflate_init(a, zip); 2890 if (ret != ARCHIVE_OK) 2891 goto exit_mac_metadata; 2892 zip->stream.next_in = 2893 (Bytef *)(uintptr_t)(const void *)p; 2894 zip->stream.avail_in = (uInt)bytes_avail; 2895 zip->stream.total_in = 0; 2896 zip->stream.next_out = mp; 2897 zip->stream.avail_out = (uInt)metadata_bytes; 2898 zip->stream.total_out = 0; 2899 2900 r = inflate(&zip->stream, 0); 2901 switch (r) { 2902 case Z_OK: 2903 break; 2904 case Z_STREAM_END: 2905 eof = 1; 2906 break; 2907 case Z_MEM_ERROR: 2908 archive_set_error(&a->archive, ENOMEM, 2909 "Out of memory for ZIP decompression"); 2910 ret = ARCHIVE_FATAL; 2911 goto exit_mac_metadata; 2912 default: 2913 archive_set_error(&a->archive, 2914 ARCHIVE_ERRNO_MISC, 2915 "ZIP decompression failed (%d)", r); 2916 ret = ARCHIVE_FATAL; 2917 goto exit_mac_metadata; 2918 } 2919 bytes_used = zip->stream.total_in; 2920 metadata_bytes -= zip->stream.total_out; 2921 mp += zip->stream.total_out; 2922 break; 2923 } 2924#endif 2925 default: 2926 bytes_used = 0; 2927 break; 2928 } 2929 __archive_read_consume(a, bytes_used); 2930 remaining_bytes -= bytes_used; 2931 } 2932 archive_entry_copy_mac_metadata(entry, metadata, 2933 (size_t)rsrc->uncompressed_size - metadata_bytes); 2934 2935exit_mac_metadata: 2936 __archive_read_seek(a, offset, SEEK_SET); 2937 zip->decompress_init = 0; 2938 free(metadata); 2939 return (ret); 2940} 2941 2942static int 2943archive_read_format_zip_seekable_read_header(struct archive_read *a, 2944 struct archive_entry *entry) 2945{ 2946 struct zip *zip = (struct zip *)a->format->data; 2947 struct zip_entry *rsrc; 2948 int64_t offset; 2949 int r, ret = ARCHIVE_OK; 2950 2951 /* 2952 * It should be sufficient to call archive_read_next_header() for 2953 * a reader to determine if an entry is encrypted or not. If the 2954 * encryption of an entry is only detectable when calling 2955 * archive_read_data(), so be it. We'll do the same check there 2956 * as well. 2957 */ 2958 if (zip->has_encrypted_entries == 2959 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2960 zip->has_encrypted_entries = 0; 2961 2962 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2963 if (a->archive.archive_format_name == NULL) 2964 a->archive.archive_format_name = "ZIP"; 2965 2966 if (zip->zip_entries == NULL) { 2967 r = slurp_central_directory(a, zip); 2968 if (r != ARCHIVE_OK) 2969 return r; 2970 /* Get first entry whose local header offset is lower than 2971 * other entries in the archive file. */ 2972 zip->entry = 2973 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 2974 } else if (zip->entry != NULL) { 2975 /* Get next entry in local header offset order. */ 2976 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 2977 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 2978 } 2979 2980 if (zip->entry == NULL) 2981 return ARCHIVE_EOF; 2982 2983 if (zip->entry->rsrcname.s) 2984 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 2985 &zip->tree_rsrc, zip->entry->rsrcname.s); 2986 else 2987 rsrc = NULL; 2988 2989 if (zip->cctx_valid) 2990 archive_decrypto_aes_ctr_release(&zip->cctx); 2991 if (zip->hctx_valid) 2992 archive_hmac_sha1_cleanup(&zip->hctx); 2993 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2994 __archive_read_reset_passphrase(a); 2995 2996 /* File entries are sorted by the header offset, we should mostly 2997 * use __archive_read_consume to advance a read point to avoid redundant 2998 * data reading. */ 2999 offset = archive_filter_bytes(&a->archive, 0); 3000 if (offset < zip->entry->local_header_offset) 3001 __archive_read_consume(a, 3002 zip->entry->local_header_offset - offset); 3003 else if (offset != zip->entry->local_header_offset) { 3004 __archive_read_seek(a, zip->entry->local_header_offset, 3005 SEEK_SET); 3006 } 3007 zip->unconsumed = 0; 3008 r = zip_read_local_file_header(a, entry, zip); 3009 if (r != ARCHIVE_OK) 3010 return r; 3011 if (rsrc) { 3012 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3013 if (ret2 < ret) 3014 ret = ret2; 3015 } 3016 return (ret); 3017} 3018 3019/* 3020 * We're going to seek for the next header anyway, so we don't 3021 * need to bother doing anything here. 3022 */ 3023static int 3024archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3025{ 3026 struct zip *zip; 3027 zip = (struct zip *)(a->format->data); 3028 3029 zip->unconsumed = 0; 3030 return (ARCHIVE_OK); 3031} 3032 3033int 3034archive_read_support_format_zip_seekable(struct archive *_a) 3035{ 3036 struct archive_read *a = (struct archive_read *)_a; 3037 struct zip *zip; 3038 int r; 3039 3040 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3041 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3042 3043 zip = (struct zip *)calloc(1, sizeof(*zip)); 3044 if (zip == NULL) { 3045 archive_set_error(&a->archive, ENOMEM, 3046 "Can't allocate zip data"); 3047 return (ARCHIVE_FATAL); 3048 } 3049 3050#ifdef HAVE_COPYFILE_H 3051 /* Set this by default on Mac OS. */ 3052 zip->process_mac_extensions = 1; 3053#endif 3054 3055 /* 3056 * Until enough data has been read, we cannot tell about 3057 * any encrypted entries yet. 3058 */ 3059 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3060 zip->crc32func = real_crc32; 3061 3062 r = __archive_read_register_format(a, 3063 zip, 3064 "zip", 3065 archive_read_format_zip_seekable_bid, 3066 archive_read_format_zip_options, 3067 archive_read_format_zip_seekable_read_header, 3068 archive_read_format_zip_read_data, 3069 archive_read_format_zip_read_data_skip_seekable, 3070 NULL, 3071 archive_read_format_zip_cleanup, 3072 archive_read_support_format_zip_capabilities_seekable, 3073 archive_read_format_zip_has_encrypted_entries); 3074 3075 if (r != ARCHIVE_OK) 3076 free(zip); 3077 return (ARCHIVE_OK); 3078} 3079