archive_read_support_format_zip.c revision 313570
1/*- 2 * Copyright (c) 2004-2013 Tim Kientzle 3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA 4 * Copyright (c) 2013 Konrad Kleine 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28#include "archive_platform.h" 29__FBSDID("$FreeBSD: stable/11/contrib/libarchive/libarchive/archive_read_support_format_zip.c 313570 2017-02-11 00:54:16Z mm $"); 30 31/* 32 * The definitive documentation of the Zip file format is: 33 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT 34 * 35 * The Info-Zip project has pioneered various extensions to better 36 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 37 * "Ux", and 0x7875 "ux" extensions for time and ownership 38 * information. 39 * 40 * History of this code: The streaming Zip reader was first added to 41 * libarchive in January 2005. Support for seekable input sources was 42 * added in Nov 2011. Zip64 support (including a significant code 43 * refactoring) was added in 2014. 44 */ 45 46#ifdef HAVE_ERRNO_H 47#include <errno.h> 48#endif 49#ifdef HAVE_STDLIB_H 50#include <stdlib.h> 51#endif 52#ifdef HAVE_ZLIB_H 53#include <zlib.h> 54#endif 55 56#include "archive.h" 57#include "archive_digest_private.h" 58#include "archive_cryptor_private.h" 59#include "archive_endian.h" 60#include "archive_entry.h" 61#include "archive_entry_locale.h" 62#include "archive_hmac_private.h" 63#include "archive_private.h" 64#include "archive_rb.h" 65#include "archive_read_private.h" 66 67#ifndef HAVE_ZLIB_H 68#include "archive_crc32.h" 69#endif 70 71struct zip_entry { 72 struct archive_rb_node node; 73 struct zip_entry *next; 74 int64_t local_header_offset; 75 int64_t compressed_size; 76 int64_t uncompressed_size; 77 int64_t gid; 78 int64_t uid; 79 struct archive_string rsrcname; 80 time_t mtime; 81 time_t atime; 82 time_t ctime; 83 uint32_t crc32; 84 uint16_t mode; 85 uint16_t zip_flags; /* From GP Flags Field */ 86 unsigned char compression; 87 unsigned char system; /* From "version written by" */ 88 unsigned char flags; /* Our extra markers. */ 89 unsigned char decdat;/* Used for Decryption check */ 90 91 /* WinZip AES encryption extra field should be available 92 * when compression is 99. */ 93 struct { 94 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 95 unsigned vendor; 96#define AES_VENDOR_AE_1 0x0001 97#define AES_VENDOR_AE_2 0x0002 98 /* AES encryption strength: 99 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 100 unsigned strength; 101 /* Actual compression method. */ 102 unsigned char compression; 103 } aes_extra; 104}; 105 106struct trad_enc_ctx { 107 uint32_t keys[3]; 108}; 109 110/* Bits used in zip_flags. */ 111#define ZIP_ENCRYPTED (1 << 0) 112#define ZIP_LENGTH_AT_END (1 << 3) 113#define ZIP_STRONG_ENCRYPTED (1 << 6) 114#define ZIP_UTF8_NAME (1 << 11) 115/* See "7.2 Single Password Symmetric Encryption Method" 116 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 117#define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 118 119/* Bits used in flags. */ 120#define LA_USED_ZIP64 (1 << 0) 121#define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 122 123/* 124 * See "WinZip - AES Encryption Information" 125 * http://www.winzip.com/aes_info.htm 126 */ 127/* Value used in compression method. */ 128#define WINZIP_AES_ENCRYPTION 99 129/* Authentication code size. */ 130#define AUTH_CODE_SIZE 10 131/**/ 132#define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 133 134struct zip { 135 /* Structural information about the archive. */ 136 struct archive_string format_name; 137 int64_t central_directory_offset; 138 size_t central_directory_entries_total; 139 size_t central_directory_entries_on_this_disk; 140 int has_encrypted_entries; 141 142 /* List of entries (seekable Zip only) */ 143 struct zip_entry *zip_entries; 144 struct archive_rb_tree tree; 145 struct archive_rb_tree tree_rsrc; 146 147 /* Bytes read but not yet consumed via __archive_read_consume() */ 148 size_t unconsumed; 149 150 /* Information about entry we're currently reading. */ 151 struct zip_entry *entry; 152 int64_t entry_bytes_remaining; 153 154 /* These count the number of bytes actually read for the entry. */ 155 int64_t entry_compressed_bytes_read; 156 int64_t entry_uncompressed_bytes_read; 157 158 /* Running CRC32 of the decompressed data */ 159 unsigned long entry_crc32; 160 unsigned long (*crc32func)(unsigned long, const void *, 161 size_t); 162 char ignore_crc32; 163 164 /* Flags to mark progress of decompression. */ 165 char decompress_init; 166 char end_of_entry; 167 168#ifdef HAVE_ZLIB_H 169 unsigned char *uncompressed_buffer; 170 size_t uncompressed_buffer_size; 171 z_stream stream; 172 char stream_valid; 173#endif 174 175 struct archive_string_conv *sconv; 176 struct archive_string_conv *sconv_default; 177 struct archive_string_conv *sconv_utf8; 178 int init_default_conversion; 179 int process_mac_extensions; 180 181 char init_decryption; 182 183 /* Decryption buffer. */ 184 /* 185 * The decrypted data starts at decrypted_ptr and 186 * extends for decrypted_bytes_remaining. Decryption 187 * adds new data to the end of this block, data is returned 188 * to clients from the beginning. When the block hits the 189 * end of decrypted_buffer, it has to be shuffled back to 190 * the beginning of the buffer. 191 */ 192 unsigned char *decrypted_buffer; 193 unsigned char *decrypted_ptr; 194 size_t decrypted_buffer_size; 195 size_t decrypted_bytes_remaining; 196 size_t decrypted_unconsumed_bytes; 197 198 /* Traditional PKWARE decryption. */ 199 struct trad_enc_ctx tctx; 200 char tctx_valid; 201 202 /* WinZip AES decryption. */ 203 /* Contexts used for AES decryption. */ 204 archive_crypto_ctx cctx; 205 char cctx_valid; 206 archive_hmac_sha1_ctx hctx; 207 char hctx_valid; 208 209 /* Strong encryption's decryption header information. */ 210 unsigned iv_size; 211 unsigned alg_id; 212 unsigned bit_len; 213 unsigned flags; 214 unsigned erd_size; 215 unsigned v_size; 216 unsigned v_crc32; 217 uint8_t *iv; 218 uint8_t *erd; 219 uint8_t *v_data; 220}; 221 222/* Many systems define min or MIN, but not all. */ 223#define zipmin(a,b) ((a) < (b) ? (a) : (b)) 224 225/* ------------------------------------------------------------------------ */ 226 227/* 228 Traditional PKWARE Decryption functions. 229 */ 230 231static void 232trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 233{ 234 uint8_t t; 235#define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 236 237 ctx->keys[0] = CRC32(ctx->keys[0], c); 238 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 239 t = (ctx->keys[1] >> 24) & 0xff; 240 ctx->keys[2] = CRC32(ctx->keys[2], t); 241#undef CRC32 242} 243 244static uint8_t 245trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) 246{ 247 unsigned temp = ctx->keys[2] | 2; 248 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 249} 250 251static void 252trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 253 size_t in_len, uint8_t *out, size_t out_len) 254{ 255 unsigned i, max; 256 257 max = (unsigned)((in_len < out_len)? in_len: out_len); 258 259 for (i = 0; i < max; i++) { 260 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); 261 out[i] = t; 262 trad_enc_update_keys(ctx, t); 263 } 264} 265 266static int 267trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 268 const uint8_t *key, size_t key_len, uint8_t *crcchk) 269{ 270 uint8_t header[12]; 271 272 if (key_len < 12) { 273 *crcchk = 0xff; 274 return -1; 275 } 276 277 ctx->keys[0] = 305419896L; 278 ctx->keys[1] = 591751049L; 279 ctx->keys[2] = 878082192L; 280 281 for (;pw_len; --pw_len) 282 trad_enc_update_keys(ctx, *pw++); 283 284 trad_enc_decrypt_update(ctx, key, 12, header, 12); 285 /* Return the last byte for CRC check. */ 286 *crcchk = header[11]; 287 return 0; 288} 289 290#if 0 291static void 292crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 293 int key_size) 294{ 295#define MD_SIZE 20 296 archive_sha1_ctx ctx; 297 unsigned char md1[MD_SIZE]; 298 unsigned char md2[MD_SIZE * 2]; 299 unsigned char mkb[64]; 300 int i; 301 302 archive_sha1_init(&ctx); 303 archive_sha1_update(&ctx, p, size); 304 archive_sha1_final(&ctx, md1); 305 306 memset(mkb, 0x36, sizeof(mkb)); 307 for (i = 0; i < MD_SIZE; i++) 308 mkb[i] ^= md1[i]; 309 archive_sha1_init(&ctx); 310 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 311 archive_sha1_final(&ctx, md2); 312 313 memset(mkb, 0x5C, sizeof(mkb)); 314 for (i = 0; i < MD_SIZE; i++) 315 mkb[i] ^= md1[i]; 316 archive_sha1_init(&ctx); 317 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 318 archive_sha1_final(&ctx, md2 + MD_SIZE); 319 320 if (key_size > 32) 321 key_size = 32; 322 memcpy(key, md2, key_size); 323#undef MD_SIZE 324} 325#endif 326 327/* 328 * Common code for streaming or seeking modes. 329 * 330 * Includes code to read local file headers, decompress data 331 * from entry bodies, and common API. 332 */ 333 334static unsigned long 335real_crc32(unsigned long crc, const void *buff, size_t len) 336{ 337 return crc32(crc, buff, (unsigned int)len); 338} 339 340/* Used by "ignorecrc32" option to speed up tests. */ 341static unsigned long 342fake_crc32(unsigned long crc, const void *buff, size_t len) 343{ 344 (void)crc; /* UNUSED */ 345 (void)buff; /* UNUSED */ 346 (void)len; /* UNUSED */ 347 return 0; 348} 349 350static struct { 351 int id; 352 const char * name; 353} compression_methods[] = { 354 {0, "uncompressed"}, /* The file is stored (no compression) */ 355 {1, "shrinking"}, /* The file is Shrunk */ 356 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 357 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 358 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 359 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 360 {6, "imploded"}, /* The file is Imploded */ 361 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 362 {8, "deflation"}, /* The file is Deflated */ 363 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 364 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 365 * (old IBM TERSE) */ 366 {11, "reserved"}, /* Reserved by PKWARE */ 367 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 368 {13, "reserved"}, /* Reserved by PKWARE */ 369 {14, "lzma"}, /* LZMA (EFS) */ 370 {15, "reserved"}, /* Reserved by PKWARE */ 371 {16, "reserved"}, /* Reserved by PKWARE */ 372 {17, "reserved"}, /* Reserved by PKWARE */ 373 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 374 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 375 {97, "wav-pack"}, /* WavPack compressed data */ 376 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 377 {99, "aes"} /* WinZip AES encryption */ 378}; 379 380static const char * 381compression_name(const int compression) 382{ 383 static const int num_compression_methods = 384 sizeof(compression_methods)/sizeof(compression_methods[0]); 385 int i=0; 386 387 while(compression >= 0 && i < num_compression_methods) { 388 if (compression_methods[i].id == compression) 389 return compression_methods[i].name; 390 i++; 391 } 392 return "??"; 393} 394 395/* Convert an MSDOS-style date/time into Unix-style time. */ 396static time_t 397zip_time(const char *p) 398{ 399 int msTime, msDate; 400 struct tm ts; 401 402 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 403 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 404 405 memset(&ts, 0, sizeof(ts)); 406 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 407 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 408 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 409 ts.tm_hour = (msTime >> 11) & 0x1f; 410 ts.tm_min = (msTime >> 5) & 0x3f; 411 ts.tm_sec = (msTime << 1) & 0x3e; 412 ts.tm_isdst = -1; 413 return mktime(&ts); 414} 415 416/* 417 * The extra data is stored as a list of 418 * id1+size1+data1 + id2+size2+data2 ... 419 * triplets. id and size are 2 bytes each. 420 */ 421static int 422process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry) 423{ 424 unsigned offset = 0; 425 426 if (extra_length == 0) { 427 return ARCHIVE_OK; 428 } 429 430 if (extra_length < 4) { 431 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 432 "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length); 433 return ARCHIVE_FAILED; 434 } 435 while (offset <= extra_length - 4) { 436 unsigned short headerid = archive_le16dec(p + offset); 437 unsigned short datasize = archive_le16dec(p + offset + 2); 438 439 offset += 4; 440 if (offset + datasize > extra_length) { 441 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 442 "Extra data overflow: Need %d bytes but only found %d bytes", 443 (int)datasize, (int)(extra_length - offset)); 444 return ARCHIVE_FAILED; 445 } 446#ifdef DEBUG 447 fprintf(stderr, "Header id 0x%04x, length %d\n", 448 headerid, datasize); 449#endif 450 switch (headerid) { 451 case 0x0001: 452 /* Zip64 extended information extra field. */ 453 zip_entry->flags |= LA_USED_ZIP64; 454 if (zip_entry->uncompressed_size == 0xffffffff) { 455 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 the mode is totally empty, set some sane default. */ 868 if (zip_entry->mode == 0) { 869 zip_entry->mode |= 0664; 870 } 871 872 /* Make sure that entries with a trailing '/' are marked as directories 873 * even if the External File Attributes contains bogus values. If this 874 * is not a directory and there is no type, assume regularfile. */ 875 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 876 int has_slash; 877 878 wp = archive_entry_pathname_w(entry); 879 if (wp != NULL) { 880 len = wcslen(wp); 881 has_slash = len > 0 && wp[len - 1] == L'/'; 882 } else { 883 cp = archive_entry_pathname(entry); 884 len = (cp != NULL)?strlen(cp):0; 885 has_slash = len > 0 && cp[len - 1] == '/'; 886 } 887 /* Correct file type as needed. */ 888 if (has_slash) { 889 zip_entry->mode &= ~AE_IFMT; 890 zip_entry->mode |= AE_IFDIR; 891 zip_entry->mode |= 0111; 892 } else if ((zip_entry->mode & AE_IFMT) == 0) { 893 zip_entry->mode |= AE_IFREG; 894 } 895 } 896 897 /* Make sure directories end in '/' */ 898 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 899 wp = archive_entry_pathname_w(entry); 900 if (wp != NULL) { 901 len = wcslen(wp); 902 if (len > 0 && wp[len - 1] != L'/') { 903 struct archive_wstring s; 904 archive_string_init(&s); 905 archive_wstrcat(&s, wp); 906 archive_wstrappend_wchar(&s, L'/'); 907 archive_entry_copy_pathname_w(entry, s.s); 908 archive_wstring_free(&s); 909 } 910 } else { 911 cp = archive_entry_pathname(entry); 912 len = (cp != NULL)?strlen(cp):0; 913 if (len > 0 && cp[len - 1] != '/') { 914 struct archive_string s; 915 archive_string_init(&s); 916 archive_strcat(&s, cp); 917 archive_strappend_char(&s, '/'); 918 archive_entry_set_pathname(entry, s.s); 919 archive_string_free(&s); 920 } 921 } 922 } 923 924 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 925 /* If this came from the central dir, it's size info 926 * is definitive, so ignore the length-at-end flag. */ 927 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 928 /* If local header is missing a value, use the one from 929 the central directory. If both have it, warn about 930 mismatches. */ 931 if (zip_entry->crc32 == 0) { 932 zip_entry->crc32 = zip_entry_central_dir.crc32; 933 } else if (!zip->ignore_crc32 934 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 935 archive_set_error(&a->archive, 936 ARCHIVE_ERRNO_FILE_FORMAT, 937 "Inconsistent CRC32 values"); 938 ret = ARCHIVE_WARN; 939 } 940 if (zip_entry->compressed_size == 0) { 941 zip_entry->compressed_size 942 = zip_entry_central_dir.compressed_size; 943 } else if (zip_entry->compressed_size 944 != zip_entry_central_dir.compressed_size) { 945 archive_set_error(&a->archive, 946 ARCHIVE_ERRNO_FILE_FORMAT, 947 "Inconsistent compressed size: " 948 "%jd in central directory, %jd in local header", 949 (intmax_t)zip_entry_central_dir.compressed_size, 950 (intmax_t)zip_entry->compressed_size); 951 ret = ARCHIVE_WARN; 952 } 953 if (zip_entry->uncompressed_size == 0) { 954 zip_entry->uncompressed_size 955 = zip_entry_central_dir.uncompressed_size; 956 } else if (zip_entry->uncompressed_size 957 != zip_entry_central_dir.uncompressed_size) { 958 archive_set_error(&a->archive, 959 ARCHIVE_ERRNO_FILE_FORMAT, 960 "Inconsistent uncompressed size: " 961 "%jd in central directory, %jd in local header", 962 (intmax_t)zip_entry_central_dir.uncompressed_size, 963 (intmax_t)zip_entry->uncompressed_size); 964 ret = ARCHIVE_WARN; 965 } 966 } 967 968 /* Populate some additional entry fields: */ 969 archive_entry_set_mode(entry, zip_entry->mode); 970 archive_entry_set_uid(entry, zip_entry->uid); 971 archive_entry_set_gid(entry, zip_entry->gid); 972 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 973 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 974 archive_entry_set_atime(entry, zip_entry->atime, 0); 975 976 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 977 size_t linkname_length; 978 979 if (zip_entry->compressed_size > 64 * 1024) { 980 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 981 "Zip file with oversized link entry"); 982 return ARCHIVE_FATAL; 983 } 984 985 linkname_length = (size_t)zip_entry->compressed_size; 986 987 archive_entry_set_size(entry, 0); 988 p = __archive_read_ahead(a, linkname_length, NULL); 989 if (p == NULL) { 990 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 991 "Truncated Zip file"); 992 return ARCHIVE_FATAL; 993 } 994 995 sconv = zip->sconv; 996 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 997 sconv = zip->sconv_utf8; 998 if (sconv == NULL) 999 sconv = zip->sconv_default; 1000 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 1001 sconv) != 0) { 1002 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1003 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1004 archive_entry_copy_symlink_l(entry, p, 1005 linkname_length, NULL); 1006 if (errno == ENOMEM) { 1007 archive_set_error(&a->archive, ENOMEM, 1008 "Can't allocate memory for Symlink"); 1009 return (ARCHIVE_FATAL); 1010 } 1011 /* 1012 * Since there is no character-set regulation for 1013 * symlink name, do not report the conversion error 1014 * in an automatic conversion. 1015 */ 1016 if (sconv != zip->sconv_utf8 || 1017 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1018 archive_set_error(&a->archive, 1019 ARCHIVE_ERRNO_FILE_FORMAT, 1020 "Symlink cannot be converted " 1021 "from %s to current locale.", 1022 archive_string_conversion_charset_name( 1023 sconv)); 1024 ret = ARCHIVE_WARN; 1025 } 1026 } 1027 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1028 1029 if (__archive_read_consume(a, linkname_length) < 0) { 1030 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1031 "Read error skipping symlink target name"); 1032 return ARCHIVE_FATAL; 1033 } 1034 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1035 || zip_entry->uncompressed_size > 0) { 1036 /* Set the size only if it's meaningful. */ 1037 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1038 } 1039 zip->entry_bytes_remaining = zip_entry->compressed_size; 1040 1041 /* If there's no body, force read_data() to return EOF immediately. */ 1042 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1043 && zip->entry_bytes_remaining < 1) 1044 zip->end_of_entry = 1; 1045 1046 /* Set up a more descriptive format name. */ 1047 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1048 version / 10, version % 10, 1049 compression_name(zip->entry->compression)); 1050 a->archive.archive_format_name = zip->format_name.s; 1051 1052 return (ret); 1053} 1054 1055static int 1056check_authentication_code(struct archive_read *a, const void *_p) 1057{ 1058 struct zip *zip = (struct zip *)(a->format->data); 1059 1060 /* Check authentication code. */ 1061 if (zip->hctx_valid) { 1062 const void *p; 1063 uint8_t hmac[20]; 1064 size_t hmac_len = 20; 1065 int cmp; 1066 1067 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1068 if (_p == NULL) { 1069 /* Read authentication code. */ 1070 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1071 if (p == NULL) { 1072 archive_set_error(&a->archive, 1073 ARCHIVE_ERRNO_FILE_FORMAT, 1074 "Truncated ZIP file data"); 1075 return (ARCHIVE_FATAL); 1076 } 1077 } else { 1078 p = _p; 1079 } 1080 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1081 __archive_read_consume(a, AUTH_CODE_SIZE); 1082 if (cmp != 0) { 1083 archive_set_error(&a->archive, 1084 ARCHIVE_ERRNO_MISC, 1085 "ZIP bad Authentication code"); 1086 return (ARCHIVE_WARN); 1087 } 1088 } 1089 return (ARCHIVE_OK); 1090} 1091 1092/* 1093 * Read "uncompressed" data. There are three cases: 1094 * 1) We know the size of the data. This is always true for the 1095 * seeking reader (we've examined the Central Directory already). 1096 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1097 * Info-ZIP seems to do this; we know the size but have to grab 1098 * the CRC from the data descriptor afterwards. 1099 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1100 * we have no size information. In this case, we can do pretty 1101 * well by watching for the data descriptor record. The data 1102 * descriptor is 16 bytes and includes a computed CRC that should 1103 * provide a strong check. 1104 * 1105 * TODO: Technically, the PK\007\010 signature is optional. 1106 * In the original spec, the data descriptor contained CRC 1107 * and size fields but had no leading signature. In practice, 1108 * newer writers seem to provide the signature pretty consistently. 1109 * 1110 * For uncompressed data, the PK\007\010 marker seems essential 1111 * to be sure we've actually seen the end of the entry. 1112 * 1113 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1114 * zip->end_of_entry if it consumes all of the data. 1115 */ 1116static int 1117zip_read_data_none(struct archive_read *a, const void **_buff, 1118 size_t *size, int64_t *offset) 1119{ 1120 struct zip *zip; 1121 const char *buff; 1122 ssize_t bytes_avail; 1123 int r; 1124 1125 (void)offset; /* UNUSED */ 1126 1127 zip = (struct zip *)(a->format->data); 1128 1129 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1130 const char *p; 1131 ssize_t grabbing_bytes = 24; 1132 1133 if (zip->hctx_valid) 1134 grabbing_bytes += AUTH_CODE_SIZE; 1135 /* Grab at least 24 bytes. */ 1136 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1137 if (bytes_avail < grabbing_bytes) { 1138 /* Zip archives have end-of-archive markers 1139 that are longer than this, so a failure to get at 1140 least 24 bytes really does indicate a truncated 1141 file. */ 1142 archive_set_error(&a->archive, 1143 ARCHIVE_ERRNO_FILE_FORMAT, 1144 "Truncated ZIP file data"); 1145 return (ARCHIVE_FATAL); 1146 } 1147 /* Check for a complete PK\007\010 signature, followed 1148 * by the correct 4-byte CRC. */ 1149 p = buff; 1150 if (zip->hctx_valid) 1151 p += AUTH_CODE_SIZE; 1152 if (p[0] == 'P' && p[1] == 'K' 1153 && p[2] == '\007' && p[3] == '\010' 1154 && (archive_le32dec(p + 4) == zip->entry_crc32 1155 || zip->ignore_crc32 1156 || (zip->hctx_valid 1157 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1158 if (zip->entry->flags & LA_USED_ZIP64) { 1159 zip->entry->crc32 = archive_le32dec(p + 4); 1160 zip->entry->compressed_size = 1161 archive_le64dec(p + 8); 1162 zip->entry->uncompressed_size = 1163 archive_le64dec(p + 16); 1164 zip->unconsumed = 24; 1165 } else { 1166 zip->entry->crc32 = archive_le32dec(p + 4); 1167 zip->entry->compressed_size = 1168 archive_le32dec(p + 8); 1169 zip->entry->uncompressed_size = 1170 archive_le32dec(p + 12); 1171 zip->unconsumed = 16; 1172 } 1173 if (zip->hctx_valid) { 1174 r = check_authentication_code(a, buff); 1175 if (r != ARCHIVE_OK) 1176 return (r); 1177 } 1178 zip->end_of_entry = 1; 1179 return (ARCHIVE_OK); 1180 } 1181 /* If not at EOF, ensure we consume at least one byte. */ 1182 ++p; 1183 1184 /* Scan forward until we see where a PK\007\010 signature 1185 * might be. */ 1186 /* Return bytes up until that point. On the next call, 1187 * the code above will verify the data descriptor. */ 1188 while (p < buff + bytes_avail - 4) { 1189 if (p[3] == 'P') { p += 3; } 1190 else if (p[3] == 'K') { p += 2; } 1191 else if (p[3] == '\007') { p += 1; } 1192 else if (p[3] == '\010' && p[2] == '\007' 1193 && p[1] == 'K' && p[0] == 'P') { 1194 if (zip->hctx_valid) 1195 p -= AUTH_CODE_SIZE; 1196 break; 1197 } else { p += 4; } 1198 } 1199 bytes_avail = p - buff; 1200 } else { 1201 if (zip->entry_bytes_remaining == 0) { 1202 zip->end_of_entry = 1; 1203 if (zip->hctx_valid) { 1204 r = check_authentication_code(a, NULL); 1205 if (r != ARCHIVE_OK) 1206 return (r); 1207 } 1208 return (ARCHIVE_OK); 1209 } 1210 /* Grab a bunch of bytes. */ 1211 buff = __archive_read_ahead(a, 1, &bytes_avail); 1212 if (bytes_avail <= 0) { 1213 archive_set_error(&a->archive, 1214 ARCHIVE_ERRNO_FILE_FORMAT, 1215 "Truncated ZIP file data"); 1216 return (ARCHIVE_FATAL); 1217 } 1218 if (bytes_avail > zip->entry_bytes_remaining) 1219 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1220 } 1221 if (zip->tctx_valid || zip->cctx_valid) { 1222 size_t dec_size = bytes_avail; 1223 1224 if (dec_size > zip->decrypted_buffer_size) 1225 dec_size = zip->decrypted_buffer_size; 1226 if (zip->tctx_valid) { 1227 trad_enc_decrypt_update(&zip->tctx, 1228 (const uint8_t *)buff, dec_size, 1229 zip->decrypted_buffer, dec_size); 1230 } else { 1231 size_t dsize = dec_size; 1232 archive_hmac_sha1_update(&zip->hctx, 1233 (const uint8_t *)buff, dec_size); 1234 archive_decrypto_aes_ctr_update(&zip->cctx, 1235 (const uint8_t *)buff, dec_size, 1236 zip->decrypted_buffer, &dsize); 1237 } 1238 bytes_avail = dec_size; 1239 buff = (const char *)zip->decrypted_buffer; 1240 } 1241 *size = bytes_avail; 1242 zip->entry_bytes_remaining -= bytes_avail; 1243 zip->entry_uncompressed_bytes_read += bytes_avail; 1244 zip->entry_compressed_bytes_read += bytes_avail; 1245 zip->unconsumed += bytes_avail; 1246 *_buff = buff; 1247 return (ARCHIVE_OK); 1248} 1249 1250#ifdef HAVE_ZLIB_H 1251static int 1252zip_deflate_init(struct archive_read *a, struct zip *zip) 1253{ 1254 int r; 1255 1256 /* If we haven't yet read any data, initialize the decompressor. */ 1257 if (!zip->decompress_init) { 1258 if (zip->stream_valid) 1259 r = inflateReset(&zip->stream); 1260 else 1261 r = inflateInit2(&zip->stream, 1262 -15 /* Don't check for zlib header */); 1263 if (r != Z_OK) { 1264 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1265 "Can't initialize ZIP decompression."); 1266 return (ARCHIVE_FATAL); 1267 } 1268 /* Stream structure has been set up. */ 1269 zip->stream_valid = 1; 1270 /* We've initialized decompression for this stream. */ 1271 zip->decompress_init = 1; 1272 } 1273 return (ARCHIVE_OK); 1274} 1275 1276static int 1277zip_read_data_deflate(struct archive_read *a, const void **buff, 1278 size_t *size, int64_t *offset) 1279{ 1280 struct zip *zip; 1281 ssize_t bytes_avail; 1282 const void *compressed_buff, *sp; 1283 int r; 1284 1285 (void)offset; /* UNUSED */ 1286 1287 zip = (struct zip *)(a->format->data); 1288 1289 /* If the buffer hasn't been allocated, allocate it now. */ 1290 if (zip->uncompressed_buffer == NULL) { 1291 zip->uncompressed_buffer_size = 256 * 1024; 1292 zip->uncompressed_buffer 1293 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1294 if (zip->uncompressed_buffer == NULL) { 1295 archive_set_error(&a->archive, ENOMEM, 1296 "No memory for ZIP decompression"); 1297 return (ARCHIVE_FATAL); 1298 } 1299 } 1300 1301 r = zip_deflate_init(a, zip); 1302 if (r != ARCHIVE_OK) 1303 return (r); 1304 1305 /* 1306 * Note: '1' here is a performance optimization. 1307 * Recall that the decompression layer returns a count of 1308 * available bytes; asking for more than that forces the 1309 * decompressor to combine reads by copying data. 1310 */ 1311 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1312 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1313 && bytes_avail > zip->entry_bytes_remaining) { 1314 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1315 } 1316 if (bytes_avail < 0) { 1317 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1318 "Truncated ZIP file body"); 1319 return (ARCHIVE_FATAL); 1320 } 1321 1322 if (zip->tctx_valid || zip->cctx_valid) { 1323 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1324 size_t buff_remaining = 1325 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1326 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1327 1328 if (buff_remaining > (size_t)bytes_avail) 1329 buff_remaining = (size_t)bytes_avail; 1330 1331 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1332 zip->entry_bytes_remaining > 0) { 1333 if ((int64_t)(zip->decrypted_bytes_remaining 1334 + buff_remaining) 1335 > zip->entry_bytes_remaining) { 1336 if (zip->entry_bytes_remaining < 1337 (int64_t)zip->decrypted_bytes_remaining) 1338 buff_remaining = 0; 1339 else 1340 buff_remaining = 1341 (size_t)zip->entry_bytes_remaining 1342 - zip->decrypted_bytes_remaining; 1343 } 1344 } 1345 if (buff_remaining > 0) { 1346 if (zip->tctx_valid) { 1347 trad_enc_decrypt_update(&zip->tctx, 1348 compressed_buff, buff_remaining, 1349 zip->decrypted_ptr 1350 + zip->decrypted_bytes_remaining, 1351 buff_remaining); 1352 } else { 1353 size_t dsize = buff_remaining; 1354 archive_decrypto_aes_ctr_update( 1355 &zip->cctx, 1356 compressed_buff, buff_remaining, 1357 zip->decrypted_ptr 1358 + zip->decrypted_bytes_remaining, 1359 &dsize); 1360 } 1361 zip->decrypted_bytes_remaining += buff_remaining; 1362 } 1363 } 1364 bytes_avail = zip->decrypted_bytes_remaining; 1365 compressed_buff = (const char *)zip->decrypted_ptr; 1366 } 1367 1368 /* 1369 * A bug in zlib.h: stream.next_in should be marked 'const' 1370 * but isn't (the library never alters data through the 1371 * next_in pointer, only reads it). The result: this ugly 1372 * cast to remove 'const'. 1373 */ 1374 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1375 zip->stream.avail_in = (uInt)bytes_avail; 1376 zip->stream.total_in = 0; 1377 zip->stream.next_out = zip->uncompressed_buffer; 1378 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1379 zip->stream.total_out = 0; 1380 1381 r = inflate(&zip->stream, 0); 1382 switch (r) { 1383 case Z_OK: 1384 break; 1385 case Z_STREAM_END: 1386 zip->end_of_entry = 1; 1387 break; 1388 case Z_MEM_ERROR: 1389 archive_set_error(&a->archive, ENOMEM, 1390 "Out of memory for ZIP decompression"); 1391 return (ARCHIVE_FATAL); 1392 default: 1393 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1394 "ZIP decompression failed (%d)", r); 1395 return (ARCHIVE_FATAL); 1396 } 1397 1398 /* Consume as much as the compressor actually used. */ 1399 bytes_avail = zip->stream.total_in; 1400 if (zip->tctx_valid || zip->cctx_valid) { 1401 zip->decrypted_bytes_remaining -= bytes_avail; 1402 if (zip->decrypted_bytes_remaining == 0) 1403 zip->decrypted_ptr = zip->decrypted_buffer; 1404 else 1405 zip->decrypted_ptr += bytes_avail; 1406 } 1407 /* Calculate compressed data as much as we used.*/ 1408 if (zip->hctx_valid) 1409 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1410 __archive_read_consume(a, bytes_avail); 1411 zip->entry_bytes_remaining -= bytes_avail; 1412 zip->entry_compressed_bytes_read += bytes_avail; 1413 1414 *size = zip->stream.total_out; 1415 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1416 *buff = zip->uncompressed_buffer; 1417 1418 if (zip->end_of_entry && zip->hctx_valid) { 1419 r = check_authentication_code(a, NULL); 1420 if (r != ARCHIVE_OK) 1421 return (r); 1422 } 1423 1424 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1425 const char *p; 1426 1427 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1428 archive_set_error(&a->archive, 1429 ARCHIVE_ERRNO_FILE_FORMAT, 1430 "Truncated ZIP end-of-file record"); 1431 return (ARCHIVE_FATAL); 1432 } 1433 /* Consume the optional PK\007\010 marker. */ 1434 if (p[0] == 'P' && p[1] == 'K' && 1435 p[2] == '\007' && p[3] == '\010') { 1436 p += 4; 1437 zip->unconsumed = 4; 1438 } 1439 if (zip->entry->flags & LA_USED_ZIP64) { 1440 zip->entry->crc32 = archive_le32dec(p); 1441 zip->entry->compressed_size = archive_le64dec(p + 4); 1442 zip->entry->uncompressed_size = archive_le64dec(p + 12); 1443 zip->unconsumed += 20; 1444 } else { 1445 zip->entry->crc32 = archive_le32dec(p); 1446 zip->entry->compressed_size = archive_le32dec(p + 4); 1447 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1448 zip->unconsumed += 12; 1449 } 1450 } 1451 1452 return (ARCHIVE_OK); 1453} 1454#endif 1455 1456static int 1457read_decryption_header(struct archive_read *a) 1458{ 1459 struct zip *zip = (struct zip *)(a->format->data); 1460 const char *p; 1461 unsigned int remaining_size; 1462 unsigned int ts; 1463 1464 /* 1465 * Read an initialization vector data field. 1466 */ 1467 p = __archive_read_ahead(a, 2, NULL); 1468 if (p == NULL) 1469 goto truncated; 1470 ts = zip->iv_size; 1471 zip->iv_size = archive_le16dec(p); 1472 __archive_read_consume(a, 2); 1473 if (ts < zip->iv_size) { 1474 free(zip->iv); 1475 zip->iv = NULL; 1476 } 1477 p = __archive_read_ahead(a, zip->iv_size, NULL); 1478 if (p == NULL) 1479 goto truncated; 1480 if (zip->iv == NULL) { 1481 zip->iv = malloc(zip->iv_size); 1482 if (zip->iv == NULL) 1483 goto nomem; 1484 } 1485 memcpy(zip->iv, p, zip->iv_size); 1486 __archive_read_consume(a, zip->iv_size); 1487 1488 /* 1489 * Read a size of remaining decryption header field. 1490 */ 1491 p = __archive_read_ahead(a, 14, NULL); 1492 if (p == NULL) 1493 goto truncated; 1494 remaining_size = archive_le32dec(p); 1495 if (remaining_size < 16 || remaining_size > (1 << 18)) 1496 goto corrupted; 1497 1498 /* Check if format version is supported. */ 1499 if (archive_le16dec(p+4) != 3) { 1500 archive_set_error(&a->archive, 1501 ARCHIVE_ERRNO_FILE_FORMAT, 1502 "Unsupported encryption format version: %u", 1503 archive_le16dec(p+4)); 1504 return (ARCHIVE_FAILED); 1505 } 1506 1507 /* 1508 * Read an encryption algorithm field. 1509 */ 1510 zip->alg_id = archive_le16dec(p+6); 1511 switch (zip->alg_id) { 1512 case 0x6601:/* DES */ 1513 case 0x6602:/* RC2 */ 1514 case 0x6603:/* 3DES 168 */ 1515 case 0x6609:/* 3DES 112 */ 1516 case 0x660E:/* AES 128 */ 1517 case 0x660F:/* AES 192 */ 1518 case 0x6610:/* AES 256 */ 1519 case 0x6702:/* RC2 (version >= 5.2) */ 1520 case 0x6720:/* Blowfish */ 1521 case 0x6721:/* Twofish */ 1522 case 0x6801:/* RC4 */ 1523 /* Supported encryption algorithm. */ 1524 break; 1525 default: 1526 archive_set_error(&a->archive, 1527 ARCHIVE_ERRNO_FILE_FORMAT, 1528 "Unknown encryption algorithm: %u", zip->alg_id); 1529 return (ARCHIVE_FAILED); 1530 } 1531 1532 /* 1533 * Read a bit length field. 1534 */ 1535 zip->bit_len = archive_le16dec(p+8); 1536 1537 /* 1538 * Read a flags field. 1539 */ 1540 zip->flags = archive_le16dec(p+10); 1541 switch (zip->flags & 0xf000) { 1542 case 0x0001: /* Password is required to decrypt. */ 1543 case 0x0002: /* Certificates only. */ 1544 case 0x0003: /* Password or certificate required to decrypt. */ 1545 break; 1546 default: 1547 archive_set_error(&a->archive, 1548 ARCHIVE_ERRNO_FILE_FORMAT, 1549 "Unknown encryption flag: %u", zip->flags); 1550 return (ARCHIVE_FAILED); 1551 } 1552 if ((zip->flags & 0xf000) == 0 || 1553 (zip->flags & 0xf000) == 0x4000) { 1554 archive_set_error(&a->archive, 1555 ARCHIVE_ERRNO_FILE_FORMAT, 1556 "Unknown encryption flag: %u", zip->flags); 1557 return (ARCHIVE_FAILED); 1558 } 1559 1560 /* 1561 * Read an encrypted random data field. 1562 */ 1563 ts = zip->erd_size; 1564 zip->erd_size = archive_le16dec(p+12); 1565 __archive_read_consume(a, 14); 1566 if ((zip->erd_size & 0xf) != 0 || 1567 (zip->erd_size + 16) > remaining_size || 1568 (zip->erd_size + 16) < zip->erd_size) 1569 goto corrupted; 1570 1571 if (ts < zip->erd_size) { 1572 free(zip->erd); 1573 zip->erd = NULL; 1574 } 1575 p = __archive_read_ahead(a, zip->erd_size, NULL); 1576 if (p == NULL) 1577 goto truncated; 1578 if (zip->erd == NULL) { 1579 zip->erd = malloc(zip->erd_size); 1580 if (zip->erd == NULL) 1581 goto nomem; 1582 } 1583 memcpy(zip->erd, p, zip->erd_size); 1584 __archive_read_consume(a, zip->erd_size); 1585 1586 /* 1587 * Read a reserved data field. 1588 */ 1589 p = __archive_read_ahead(a, 4, NULL); 1590 if (p == NULL) 1591 goto truncated; 1592 /* Reserved data size should be zero. */ 1593 if (archive_le32dec(p) != 0) 1594 goto corrupted; 1595 __archive_read_consume(a, 4); 1596 1597 /* 1598 * Read a password validation data field. 1599 */ 1600 p = __archive_read_ahead(a, 2, NULL); 1601 if (p == NULL) 1602 goto truncated; 1603 ts = zip->v_size; 1604 zip->v_size = archive_le16dec(p); 1605 __archive_read_consume(a, 2); 1606 if ((zip->v_size & 0x0f) != 0 || 1607 (zip->erd_size + zip->v_size + 16) > remaining_size || 1608 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1609 goto corrupted; 1610 if (ts < zip->v_size) { 1611 free(zip->v_data); 1612 zip->v_data = NULL; 1613 } 1614 p = __archive_read_ahead(a, zip->v_size, NULL); 1615 if (p == NULL) 1616 goto truncated; 1617 if (zip->v_data == NULL) { 1618 zip->v_data = malloc(zip->v_size); 1619 if (zip->v_data == NULL) 1620 goto nomem; 1621 } 1622 memcpy(zip->v_data, p, zip->v_size); 1623 __archive_read_consume(a, zip->v_size); 1624 1625 p = __archive_read_ahead(a, 4, NULL); 1626 if (p == NULL) 1627 goto truncated; 1628 zip->v_crc32 = archive_le32dec(p); 1629 __archive_read_consume(a, 4); 1630 1631 /*return (ARCHIVE_OK); 1632 * This is not fully implemented yet.*/ 1633 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1634 "Encrypted file is unsupported"); 1635 return (ARCHIVE_FAILED); 1636truncated: 1637 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1638 "Truncated ZIP file data"); 1639 return (ARCHIVE_FATAL); 1640corrupted: 1641 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1642 "Corrupted ZIP file data"); 1643 return (ARCHIVE_FATAL); 1644nomem: 1645 archive_set_error(&a->archive, ENOMEM, 1646 "No memory for ZIP decryption"); 1647 return (ARCHIVE_FATAL); 1648} 1649 1650static int 1651zip_alloc_decryption_buffer(struct archive_read *a) 1652{ 1653 struct zip *zip = (struct zip *)(a->format->data); 1654 size_t bs = 256 * 1024; 1655 1656 if (zip->decrypted_buffer == NULL) { 1657 zip->decrypted_buffer_size = bs; 1658 zip->decrypted_buffer = malloc(bs); 1659 if (zip->decrypted_buffer == NULL) { 1660 archive_set_error(&a->archive, ENOMEM, 1661 "No memory for ZIP decryption"); 1662 return (ARCHIVE_FATAL); 1663 } 1664 } 1665 zip->decrypted_ptr = zip->decrypted_buffer; 1666 return (ARCHIVE_OK); 1667} 1668 1669static int 1670init_traditional_PKWARE_decryption(struct archive_read *a) 1671{ 1672 struct zip *zip = (struct zip *)(a->format->data); 1673 const void *p; 1674 int retry; 1675 int r; 1676 1677 if (zip->tctx_valid) 1678 return (ARCHIVE_OK); 1679 1680 /* 1681 Read the 12 bytes encryption header stored at 1682 the start of the data area. 1683 */ 1684#define ENC_HEADER_SIZE 12 1685 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1686 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1687 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1688 "Truncated Zip encrypted body: only %jd bytes available", 1689 (intmax_t)zip->entry_bytes_remaining); 1690 return (ARCHIVE_FATAL); 1691 } 1692 1693 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1694 if (p == NULL) { 1695 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1696 "Truncated ZIP file data"); 1697 return (ARCHIVE_FATAL); 1698 } 1699 1700 for (retry = 0;; retry++) { 1701 const char *passphrase; 1702 uint8_t crcchk; 1703 1704 passphrase = __archive_read_next_passphrase(a); 1705 if (passphrase == NULL) { 1706 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1707 (retry > 0)? 1708 "Incorrect passphrase": 1709 "Passphrase required for this entry"); 1710 return (ARCHIVE_FAILED); 1711 } 1712 1713 /* 1714 * Initialize ctx for Traditional PKWARE Decryption. 1715 */ 1716 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1717 p, ENC_HEADER_SIZE, &crcchk); 1718 if (r == 0 && crcchk == zip->entry->decdat) 1719 break;/* The passphrase is OK. */ 1720 if (retry > 10000) { 1721 /* Avoid infinity loop. */ 1722 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1723 "Too many incorrect passphrases"); 1724 return (ARCHIVE_FAILED); 1725 } 1726 } 1727 1728 __archive_read_consume(a, ENC_HEADER_SIZE); 1729 zip->tctx_valid = 1; 1730 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1731 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1732 } 1733 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1734 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1735 zip->decrypted_bytes_remaining = 0; 1736 1737 return (zip_alloc_decryption_buffer(a)); 1738#undef ENC_HEADER_SIZE 1739} 1740 1741static int 1742init_WinZip_AES_decryption(struct archive_read *a) 1743{ 1744 struct zip *zip = (struct zip *)(a->format->data); 1745 const void *p; 1746 const uint8_t *pv; 1747 size_t key_len, salt_len; 1748 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1749 int retry; 1750 int r; 1751 1752 if (zip->cctx_valid || zip->hctx_valid) 1753 return (ARCHIVE_OK); 1754 1755 switch (zip->entry->aes_extra.strength) { 1756 case 1: salt_len = 8; key_len = 16; break; 1757 case 2: salt_len = 12; key_len = 24; break; 1758 case 3: salt_len = 16; key_len = 32; break; 1759 default: goto corrupted; 1760 } 1761 p = __archive_read_ahead(a, salt_len + 2, NULL); 1762 if (p == NULL) 1763 goto truncated; 1764 1765 for (retry = 0;; retry++) { 1766 const char *passphrase; 1767 1768 passphrase = __archive_read_next_passphrase(a); 1769 if (passphrase == NULL) { 1770 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1771 (retry > 0)? 1772 "Incorrect passphrase": 1773 "Passphrase required for this entry"); 1774 return (ARCHIVE_FAILED); 1775 } 1776 memset(derived_key, 0, sizeof(derived_key)); 1777 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1778 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1779 if (r != 0) { 1780 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1781 "Decryption is unsupported due to lack of " 1782 "crypto library"); 1783 return (ARCHIVE_FAILED); 1784 } 1785 1786 /* Check password verification value. */ 1787 pv = ((const uint8_t *)p) + salt_len; 1788 if (derived_key[key_len * 2] == pv[0] && 1789 derived_key[key_len * 2 + 1] == pv[1]) 1790 break;/* The passphrase is OK. */ 1791 if (retry > 10000) { 1792 /* Avoid infinity loop. */ 1793 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1794 "Too many incorrect passphrases"); 1795 return (ARCHIVE_FAILED); 1796 } 1797 } 1798 1799 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1800 if (r != 0) { 1801 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1802 "Decryption is unsupported due to lack of crypto library"); 1803 return (ARCHIVE_FAILED); 1804 } 1805 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1806 if (r != 0) { 1807 archive_decrypto_aes_ctr_release(&zip->cctx); 1808 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1809 "Failed to initialize HMAC-SHA1"); 1810 return (ARCHIVE_FAILED); 1811 } 1812 zip->cctx_valid = zip->hctx_valid = 1; 1813 __archive_read_consume(a, salt_len + 2); 1814 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1815 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1816 && zip->entry_bytes_remaining < 0) 1817 goto corrupted; 1818 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1819 zip->decrypted_bytes_remaining = 0; 1820 1821 zip->entry->compression = zip->entry->aes_extra.compression; 1822 return (zip_alloc_decryption_buffer(a)); 1823 1824truncated: 1825 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1826 "Truncated ZIP file data"); 1827 return (ARCHIVE_FATAL); 1828corrupted: 1829 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1830 "Corrupted ZIP file data"); 1831 return (ARCHIVE_FATAL); 1832} 1833 1834static int 1835archive_read_format_zip_read_data(struct archive_read *a, 1836 const void **buff, size_t *size, int64_t *offset) 1837{ 1838 int r; 1839 struct zip *zip = (struct zip *)(a->format->data); 1840 1841 if (zip->has_encrypted_entries == 1842 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1843 zip->has_encrypted_entries = 0; 1844 } 1845 1846 *offset = zip->entry_uncompressed_bytes_read; 1847 *size = 0; 1848 *buff = NULL; 1849 1850 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1851 if (zip->end_of_entry) 1852 return (ARCHIVE_EOF); 1853 1854 /* Return EOF immediately if this is a non-regular file. */ 1855 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1856 return (ARCHIVE_EOF); 1857 1858 __archive_read_consume(a, zip->unconsumed); 1859 zip->unconsumed = 0; 1860 1861 if (zip->init_decryption) { 1862 zip->has_encrypted_entries = 1; 1863 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1864 r = read_decryption_header(a); 1865 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1866 r = init_WinZip_AES_decryption(a); 1867 else 1868 r = init_traditional_PKWARE_decryption(a); 1869 if (r != ARCHIVE_OK) 1870 return (r); 1871 zip->init_decryption = 0; 1872 } 1873 1874 switch(zip->entry->compression) { 1875 case 0: /* No compression. */ 1876 r = zip_read_data_none(a, buff, size, offset); 1877 break; 1878#ifdef HAVE_ZLIB_H 1879 case 8: /* Deflate compression. */ 1880 r = zip_read_data_deflate(a, buff, size, offset); 1881 break; 1882#endif 1883 default: /* Unsupported compression. */ 1884 /* Return a warning. */ 1885 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1886 "Unsupported ZIP compression method (%s)", 1887 compression_name(zip->entry->compression)); 1888 /* We can't decompress this entry, but we will 1889 * be able to skip() it and try the next entry. */ 1890 return (ARCHIVE_FAILED); 1891 break; 1892 } 1893 if (r != ARCHIVE_OK) 1894 return (r); 1895 /* Update checksum */ 1896 if (*size) 1897 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1898 (unsigned)*size); 1899 /* If we hit the end, swallow any end-of-data marker. */ 1900 if (zip->end_of_entry) { 1901 /* Check file size, CRC against these values. */ 1902 if (zip->entry->compressed_size != 1903 zip->entry_compressed_bytes_read) { 1904 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1905 "ZIP compressed data is wrong size " 1906 "(read %jd, expected %jd)", 1907 (intmax_t)zip->entry_compressed_bytes_read, 1908 (intmax_t)zip->entry->compressed_size); 1909 return (ARCHIVE_WARN); 1910 } 1911 /* Size field only stores the lower 32 bits of the actual 1912 * size. */ 1913 if ((zip->entry->uncompressed_size & UINT32_MAX) 1914 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1915 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1916 "ZIP uncompressed data is wrong size " 1917 "(read %jd, expected %jd)\n", 1918 (intmax_t)zip->entry_uncompressed_bytes_read, 1919 (intmax_t)zip->entry->uncompressed_size); 1920 return (ARCHIVE_WARN); 1921 } 1922 /* Check computed CRC against header */ 1923 if ((!zip->hctx_valid || 1924 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1925 zip->entry->crc32 != zip->entry_crc32 1926 && !zip->ignore_crc32) { 1927 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1928 "ZIP bad CRC: 0x%lx should be 0x%lx", 1929 (unsigned long)zip->entry_crc32, 1930 (unsigned long)zip->entry->crc32); 1931 return (ARCHIVE_WARN); 1932 } 1933 } 1934 1935 return (ARCHIVE_OK); 1936} 1937 1938static int 1939archive_read_format_zip_cleanup(struct archive_read *a) 1940{ 1941 struct zip *zip; 1942 struct zip_entry *zip_entry, *next_zip_entry; 1943 1944 zip = (struct zip *)(a->format->data); 1945#ifdef HAVE_ZLIB_H 1946 if (zip->stream_valid) 1947 inflateEnd(&zip->stream); 1948 free(zip->uncompressed_buffer); 1949#endif 1950 if (zip->zip_entries) { 1951 zip_entry = zip->zip_entries; 1952 while (zip_entry != NULL) { 1953 next_zip_entry = zip_entry->next; 1954 archive_string_free(&zip_entry->rsrcname); 1955 free(zip_entry); 1956 zip_entry = next_zip_entry; 1957 } 1958 } 1959 free(zip->decrypted_buffer); 1960 if (zip->cctx_valid) 1961 archive_decrypto_aes_ctr_release(&zip->cctx); 1962 if (zip->hctx_valid) 1963 archive_hmac_sha1_cleanup(&zip->hctx); 1964 free(zip->iv); 1965 free(zip->erd); 1966 free(zip->v_data); 1967 archive_string_free(&zip->format_name); 1968 free(zip); 1969 (a->format->data) = NULL; 1970 return (ARCHIVE_OK); 1971} 1972 1973static int 1974archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 1975{ 1976 if (_a && _a->format) { 1977 struct zip * zip = (struct zip *)_a->format->data; 1978 if (zip) { 1979 return zip->has_encrypted_entries; 1980 } 1981 } 1982 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 1983} 1984 1985static int 1986archive_read_format_zip_options(struct archive_read *a, 1987 const char *key, const char *val) 1988{ 1989 struct zip *zip; 1990 int ret = ARCHIVE_FAILED; 1991 1992 zip = (struct zip *)(a->format->data); 1993 if (strcmp(key, "compat-2x") == 0) { 1994 /* Handle filenames as libarchive 2.x */ 1995 zip->init_default_conversion = (val != NULL) ? 1 : 0; 1996 return (ARCHIVE_OK); 1997 } else if (strcmp(key, "hdrcharset") == 0) { 1998 if (val == NULL || val[0] == 0) 1999 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2000 "zip: hdrcharset option needs a character-set name" 2001 ); 2002 else { 2003 zip->sconv = archive_string_conversion_from_charset( 2004 &a->archive, val, 0); 2005 if (zip->sconv != NULL) { 2006 if (strcmp(val, "UTF-8") == 0) 2007 zip->sconv_utf8 = zip->sconv; 2008 ret = ARCHIVE_OK; 2009 } else 2010 ret = ARCHIVE_FATAL; 2011 } 2012 return (ret); 2013 } else if (strcmp(key, "ignorecrc32") == 0) { 2014 /* Mostly useful for testing. */ 2015 if (val == NULL || val[0] == 0) { 2016 zip->crc32func = real_crc32; 2017 zip->ignore_crc32 = 0; 2018 } else { 2019 zip->crc32func = fake_crc32; 2020 zip->ignore_crc32 = 1; 2021 } 2022 return (ARCHIVE_OK); 2023 } else if (strcmp(key, "mac-ext") == 0) { 2024 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2025 return (ARCHIVE_OK); 2026 } 2027 2028 /* Note: The "warn" return is just to inform the options 2029 * supervisor that we didn't handle it. It will generate 2030 * a suitable error if no one used this option. */ 2031 return (ARCHIVE_WARN); 2032} 2033 2034int 2035archive_read_support_format_zip(struct archive *a) 2036{ 2037 int r; 2038 r = archive_read_support_format_zip_streamable(a); 2039 if (r != ARCHIVE_OK) 2040 return r; 2041 return (archive_read_support_format_zip_seekable(a)); 2042} 2043 2044/* ------------------------------------------------------------------------ */ 2045 2046/* 2047 * Streaming-mode support 2048 */ 2049 2050 2051static int 2052archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2053{ 2054 (void)a; /* UNUSED */ 2055 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2056 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2057} 2058 2059static int 2060archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2061{ 2062 const char *p; 2063 2064 (void)best_bid; /* UNUSED */ 2065 2066 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2067 return (-1); 2068 2069 /* 2070 * Bid of 29 here comes from: 2071 * + 16 bits for "PK", 2072 * + next 16-bit field has 6 options so contributes 2073 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2074 * 2075 * So we've effectively verified ~29 total bits of check data. 2076 */ 2077 if (p[0] == 'P' && p[1] == 'K') { 2078 if ((p[2] == '\001' && p[3] == '\002') 2079 || (p[2] == '\003' && p[3] == '\004') 2080 || (p[2] == '\005' && p[3] == '\006') 2081 || (p[2] == '\006' && p[3] == '\006') 2082 || (p[2] == '\007' && p[3] == '\010') 2083 || (p[2] == '0' && p[3] == '0')) 2084 return (29); 2085 } 2086 2087 /* TODO: It's worth looking ahead a little bit for a valid 2088 * PK signature. In particular, that would make it possible 2089 * to read some UUEncoded SFX files or SFX files coming from 2090 * a network socket. */ 2091 2092 return (0); 2093} 2094 2095static int 2096archive_read_format_zip_streamable_read_header(struct archive_read *a, 2097 struct archive_entry *entry) 2098{ 2099 struct zip *zip; 2100 2101 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2102 if (a->archive.archive_format_name == NULL) 2103 a->archive.archive_format_name = "ZIP"; 2104 2105 zip = (struct zip *)(a->format->data); 2106 2107 /* 2108 * It should be sufficient to call archive_read_next_header() for 2109 * a reader to determine if an entry is encrypted or not. If the 2110 * encryption of an entry is only detectable when calling 2111 * archive_read_data(), so be it. We'll do the same check there 2112 * as well. 2113 */ 2114 if (zip->has_encrypted_entries == 2115 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2116 zip->has_encrypted_entries = 0; 2117 2118 /* Make sure we have a zip_entry structure to use. */ 2119 if (zip->zip_entries == NULL) { 2120 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2121 if (zip->zip_entries == NULL) { 2122 archive_set_error(&a->archive, ENOMEM, 2123 "Out of memory"); 2124 return ARCHIVE_FATAL; 2125 } 2126 } 2127 zip->entry = zip->zip_entries; 2128 memset(zip->entry, 0, sizeof(struct zip_entry)); 2129 2130 if (zip->cctx_valid) 2131 archive_decrypto_aes_ctr_release(&zip->cctx); 2132 if (zip->hctx_valid) 2133 archive_hmac_sha1_cleanup(&zip->hctx); 2134 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2135 __archive_read_reset_passphrase(a); 2136 2137 /* Search ahead for the next local file header. */ 2138 __archive_read_consume(a, zip->unconsumed); 2139 zip->unconsumed = 0; 2140 for (;;) { 2141 int64_t skipped = 0; 2142 const char *p, *end; 2143 ssize_t bytes; 2144 2145 p = __archive_read_ahead(a, 4, &bytes); 2146 if (p == NULL) 2147 return (ARCHIVE_FATAL); 2148 end = p + bytes; 2149 2150 while (p + 4 <= end) { 2151 if (p[0] == 'P' && p[1] == 'K') { 2152 if (p[2] == '\003' && p[3] == '\004') { 2153 /* Regular file entry. */ 2154 __archive_read_consume(a, skipped); 2155 return zip_read_local_file_header(a, 2156 entry, zip); 2157 } 2158 2159 /* 2160 * TODO: We cannot restore permissions 2161 * based only on the local file headers. 2162 * Consider scanning the central 2163 * directory and returning additional 2164 * entries for at least directories. 2165 * This would allow us to properly set 2166 * directory permissions. 2167 * 2168 * This won't help us fix symlinks 2169 * and may not help with regular file 2170 * permissions, either. <sigh> 2171 */ 2172 if (p[2] == '\001' && p[3] == '\002') { 2173 return (ARCHIVE_EOF); 2174 } 2175 2176 /* End of central directory? Must be an 2177 * empty archive. */ 2178 if ((p[2] == '\005' && p[3] == '\006') 2179 || (p[2] == '\006' && p[3] == '\006')) 2180 return (ARCHIVE_EOF); 2181 } 2182 ++p; 2183 ++skipped; 2184 } 2185 __archive_read_consume(a, skipped); 2186 } 2187} 2188 2189static int 2190archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2191{ 2192 struct zip *zip; 2193 int64_t bytes_skipped; 2194 2195 zip = (struct zip *)(a->format->data); 2196 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2197 zip->unconsumed = 0; 2198 if (bytes_skipped < 0) 2199 return (ARCHIVE_FATAL); 2200 2201 /* If we've already read to end of data, we're done. */ 2202 if (zip->end_of_entry) 2203 return (ARCHIVE_OK); 2204 2205 /* So we know we're streaming... */ 2206 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2207 || zip->entry->compressed_size > 0) { 2208 /* We know the compressed length, so we can just skip. */ 2209 bytes_skipped = __archive_read_consume(a, 2210 zip->entry_bytes_remaining); 2211 if (bytes_skipped < 0) 2212 return (ARCHIVE_FATAL); 2213 return (ARCHIVE_OK); 2214 } 2215 2216 if (zip->init_decryption) { 2217 int r; 2218 2219 zip->has_encrypted_entries = 1; 2220 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2221 r = read_decryption_header(a); 2222 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2223 r = init_WinZip_AES_decryption(a); 2224 else 2225 r = init_traditional_PKWARE_decryption(a); 2226 if (r != ARCHIVE_OK) 2227 return (r); 2228 zip->init_decryption = 0; 2229 } 2230 2231 /* We're streaming and we don't know the length. */ 2232 /* If the body is compressed and we know the format, we can 2233 * find an exact end-of-entry by decompressing it. */ 2234 switch (zip->entry->compression) { 2235#ifdef HAVE_ZLIB_H 2236 case 8: /* Deflate compression. */ 2237 while (!zip->end_of_entry) { 2238 int64_t offset = 0; 2239 const void *buff = NULL; 2240 size_t size = 0; 2241 int r; 2242 r = zip_read_data_deflate(a, &buff, &size, &offset); 2243 if (r != ARCHIVE_OK) 2244 return (r); 2245 } 2246 return ARCHIVE_OK; 2247#endif 2248 default: /* Uncompressed or unknown. */ 2249 /* Scan for a PK\007\010 signature. */ 2250 for (;;) { 2251 const char *p, *buff; 2252 ssize_t bytes_avail; 2253 buff = __archive_read_ahead(a, 16, &bytes_avail); 2254 if (bytes_avail < 16) { 2255 archive_set_error(&a->archive, 2256 ARCHIVE_ERRNO_FILE_FORMAT, 2257 "Truncated ZIP file data"); 2258 return (ARCHIVE_FATAL); 2259 } 2260 p = buff; 2261 while (p <= buff + bytes_avail - 16) { 2262 if (p[3] == 'P') { p += 3; } 2263 else if (p[3] == 'K') { p += 2; } 2264 else if (p[3] == '\007') { p += 1; } 2265 else if (p[3] == '\010' && p[2] == '\007' 2266 && p[1] == 'K' && p[0] == 'P') { 2267 if (zip->entry->flags & LA_USED_ZIP64) 2268 __archive_read_consume(a, 2269 p - buff + 24); 2270 else 2271 __archive_read_consume(a, 2272 p - buff + 16); 2273 return ARCHIVE_OK; 2274 } else { p += 4; } 2275 } 2276 __archive_read_consume(a, p - buff); 2277 } 2278 } 2279} 2280 2281int 2282archive_read_support_format_zip_streamable(struct archive *_a) 2283{ 2284 struct archive_read *a = (struct archive_read *)_a; 2285 struct zip *zip; 2286 int r; 2287 2288 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2289 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2290 2291 zip = (struct zip *)calloc(1, sizeof(*zip)); 2292 if (zip == NULL) { 2293 archive_set_error(&a->archive, ENOMEM, 2294 "Can't allocate zip data"); 2295 return (ARCHIVE_FATAL); 2296 } 2297 2298 /* Streamable reader doesn't support mac extensions. */ 2299 zip->process_mac_extensions = 0; 2300 2301 /* 2302 * Until enough data has been read, we cannot tell about 2303 * any encrypted entries yet. 2304 */ 2305 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2306 zip->crc32func = real_crc32; 2307 2308 r = __archive_read_register_format(a, 2309 zip, 2310 "zip", 2311 archive_read_format_zip_streamable_bid, 2312 archive_read_format_zip_options, 2313 archive_read_format_zip_streamable_read_header, 2314 archive_read_format_zip_read_data, 2315 archive_read_format_zip_read_data_skip_streamable, 2316 NULL, 2317 archive_read_format_zip_cleanup, 2318 archive_read_support_format_zip_capabilities_streamable, 2319 archive_read_format_zip_has_encrypted_entries); 2320 2321 if (r != ARCHIVE_OK) 2322 free(zip); 2323 return (ARCHIVE_OK); 2324} 2325 2326/* ------------------------------------------------------------------------ */ 2327 2328/* 2329 * Seeking-mode support 2330 */ 2331 2332static int 2333archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2334{ 2335 (void)a; /* UNUSED */ 2336 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2337 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2338} 2339 2340/* 2341 * TODO: This is a performance sink because it forces the read core to 2342 * drop buffered data from the start of file, which will then have to 2343 * be re-read again if this bidder loses. 2344 * 2345 * We workaround this a little by passing in the best bid so far so 2346 * that later bidders can do nothing if they know they'll never 2347 * outbid. But we can certainly do better... 2348 */ 2349static int 2350read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2351{ 2352 /* Sanity-check the EOCD we've found. */ 2353 2354 /* This must be the first volume. */ 2355 if (archive_le16dec(p + 4) != 0) 2356 return 0; 2357 /* Central directory must be on this volume. */ 2358 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2359 return 0; 2360 /* All central directory entries must be on this volume. */ 2361 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2362 return 0; 2363 /* Central directory can't extend beyond start of EOCD record. */ 2364 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2365 > current_offset) 2366 return 0; 2367 2368 /* Save the central directory location for later use. */ 2369 zip->central_directory_offset = archive_le32dec(p + 16); 2370 2371 /* This is just a tiny bit higher than the maximum 2372 returned by the streaming Zip bidder. This ensures 2373 that the more accurate seeking Zip parser wins 2374 whenever seek is available. */ 2375 return 32; 2376} 2377 2378/* 2379 * Examine Zip64 EOCD locator: If it's valid, store the information 2380 * from it. 2381 */ 2382static void 2383read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2384{ 2385 int64_t eocd64_offset; 2386 int64_t eocd64_size; 2387 2388 /* Sanity-check the locator record. */ 2389 2390 /* Central dir must be on first volume. */ 2391 if (archive_le32dec(p + 4) != 0) 2392 return; 2393 /* Must be only a single volume. */ 2394 if (archive_le32dec(p + 16) != 1) 2395 return; 2396 2397 /* Find the Zip64 EOCD record. */ 2398 eocd64_offset = archive_le64dec(p + 8); 2399 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2400 return; 2401 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2402 return; 2403 /* Make sure we can read all of it. */ 2404 eocd64_size = archive_le64dec(p + 4) + 12; 2405 if (eocd64_size < 56 || eocd64_size > 16384) 2406 return; 2407 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2408 return; 2409 2410 /* Sanity-check the EOCD64 */ 2411 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2412 return; 2413 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2414 return; 2415 /* CD can't be split. */ 2416 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2417 return; 2418 2419 /* Save the central directory offset for later use. */ 2420 zip->central_directory_offset = archive_le64dec(p + 48); 2421} 2422 2423static int 2424archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2425{ 2426 struct zip *zip = (struct zip *)a->format->data; 2427 int64_t file_size, current_offset; 2428 const char *p; 2429 int i, tail; 2430 2431 /* If someone has already bid more than 32, then avoid 2432 trashing the look-ahead buffers with a seek. */ 2433 if (best_bid > 32) 2434 return (-1); 2435 2436 file_size = __archive_read_seek(a, 0, SEEK_END); 2437 if (file_size <= 0) 2438 return 0; 2439 2440 /* Search last 16k of file for end-of-central-directory 2441 * record (which starts with PK\005\006) */ 2442 tail = (int)zipmin(1024 * 16, file_size); 2443 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2444 if (current_offset < 0) 2445 return 0; 2446 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2447 return 0; 2448 /* Boyer-Moore search backwards from the end, since we want 2449 * to match the last EOCD in the file (there can be more than 2450 * one if there is an uncompressed Zip archive as a member 2451 * within this Zip archive). */ 2452 for (i = tail - 22; i > 0;) { 2453 switch (p[i]) { 2454 case 'P': 2455 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2456 int ret = read_eocd(zip, p + i, 2457 current_offset + i); 2458 if (ret > 0) { 2459 /* Zip64 EOCD locator precedes 2460 * regular EOCD if present. */ 2461 if (i >= 20 2462 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2463 read_zip64_eocd(a, zip, p + i - 20); 2464 } 2465 return (ret); 2466 } 2467 } 2468 i -= 4; 2469 break; 2470 case 'K': i -= 1; break; 2471 case 005: i -= 2; break; 2472 case 006: i -= 3; break; 2473 default: i -= 4; break; 2474 } 2475 } 2476 return 0; 2477} 2478 2479/* The red-black trees are only used in seeking mode to manage 2480 * the in-memory copy of the central directory. */ 2481 2482static int 2483cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2484{ 2485 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2486 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2487 2488 if (e1->local_header_offset > e2->local_header_offset) 2489 return -1; 2490 if (e1->local_header_offset < e2->local_header_offset) 2491 return 1; 2492 return 0; 2493} 2494 2495static int 2496cmp_key(const struct archive_rb_node *n, const void *key) 2497{ 2498 /* This function won't be called */ 2499 (void)n; /* UNUSED */ 2500 (void)key; /* UNUSED */ 2501 return 1; 2502} 2503 2504static const struct archive_rb_tree_ops rb_ops = { 2505 &cmp_node, &cmp_key 2506}; 2507 2508static int 2509rsrc_cmp_node(const struct archive_rb_node *n1, 2510 const struct archive_rb_node *n2) 2511{ 2512 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2513 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2514 2515 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2516} 2517 2518static int 2519rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2520{ 2521 const struct zip_entry *e = (const struct zip_entry *)n; 2522 return (strcmp((const char *)key, e->rsrcname.s)); 2523} 2524 2525static const struct archive_rb_tree_ops rb_rsrc_ops = { 2526 &rsrc_cmp_node, &rsrc_cmp_key 2527}; 2528 2529static const char * 2530rsrc_basename(const char *name, size_t name_length) 2531{ 2532 const char *s, *r; 2533 2534 r = s = name; 2535 for (;;) { 2536 s = memchr(s, '/', name_length - (s - name)); 2537 if (s == NULL) 2538 break; 2539 r = ++s; 2540 } 2541 return (r); 2542} 2543 2544static void 2545expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2546{ 2547 struct archive_string str; 2548 struct zip_entry *dir; 2549 char *s; 2550 2551 archive_string_init(&str); 2552 archive_strncpy(&str, name, name_length); 2553 for (;;) { 2554 s = strrchr(str.s, '/'); 2555 if (s == NULL) 2556 break; 2557 *s = '\0'; 2558 /* Transfer the parent directory from zip->tree_rsrc RB 2559 * tree to zip->tree RB tree to expose. */ 2560 dir = (struct zip_entry *) 2561 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2562 if (dir == NULL) 2563 break; 2564 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2565 archive_string_free(&dir->rsrcname); 2566 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2567 } 2568 archive_string_free(&str); 2569} 2570 2571static int 2572slurp_central_directory(struct archive_read *a, struct zip *zip) 2573{ 2574 ssize_t i; 2575 unsigned found; 2576 int64_t correction; 2577 ssize_t bytes_avail; 2578 const char *p; 2579 2580 /* 2581 * Find the start of the central directory. The end-of-CD 2582 * record has our starting point, but there are lots of 2583 * Zip archives which have had other data prepended to the 2584 * file, which makes the recorded offsets all too small. 2585 * So we search forward from the specified offset until we 2586 * find the real start of the central directory. Then we 2587 * know the correction we need to apply to account for leading 2588 * padding. 2589 */ 2590 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2591 return ARCHIVE_FATAL; 2592 2593 found = 0; 2594 while (!found) { 2595 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2596 return ARCHIVE_FATAL; 2597 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2598 switch (p[i + 3]) { 2599 case 'P': i += 3; break; 2600 case 'K': i += 2; break; 2601 case 001: i += 1; break; 2602 case 002: 2603 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2604 p += i; 2605 found = 1; 2606 } else 2607 i += 4; 2608 break; 2609 case 005: i += 1; break; 2610 case 006: 2611 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2612 p += i; 2613 found = 1; 2614 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2615 p += i; 2616 found = 1; 2617 } else 2618 i += 1; 2619 break; 2620 default: i += 4; break; 2621 } 2622 } 2623 __archive_read_consume(a, i); 2624 } 2625 correction = archive_filter_bytes(&a->archive, 0) 2626 - zip->central_directory_offset; 2627 2628 __archive_rb_tree_init(&zip->tree, &rb_ops); 2629 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2630 2631 zip->central_directory_entries_total = 0; 2632 while (1) { 2633 struct zip_entry *zip_entry; 2634 size_t filename_length, extra_length, comment_length; 2635 uint32_t external_attributes; 2636 const char *name, *r; 2637 2638 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2639 return ARCHIVE_FATAL; 2640 if (memcmp(p, "PK\006\006", 4) == 0 2641 || memcmp(p, "PK\005\006", 4) == 0) { 2642 break; 2643 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2644 archive_set_error(&a->archive, 2645 -1, "Invalid central directory signature"); 2646 return ARCHIVE_FATAL; 2647 } 2648 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2649 return ARCHIVE_FATAL; 2650 2651 zip_entry = calloc(1, sizeof(struct zip_entry)); 2652 zip_entry->next = zip->zip_entries; 2653 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2654 zip->zip_entries = zip_entry; 2655 zip->central_directory_entries_total++; 2656 2657 /* version = p[4]; */ 2658 zip_entry->system = p[5]; 2659 /* version_required = archive_le16dec(p + 6); */ 2660 zip_entry->zip_flags = archive_le16dec(p + 8); 2661 if (zip_entry->zip_flags 2662 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2663 zip->has_encrypted_entries = 1; 2664 } 2665 zip_entry->compression = (char)archive_le16dec(p + 10); 2666 zip_entry->mtime = zip_time(p + 12); 2667 zip_entry->crc32 = archive_le32dec(p + 16); 2668 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2669 zip_entry->decdat = p[13]; 2670 else 2671 zip_entry->decdat = p[19]; 2672 zip_entry->compressed_size = archive_le32dec(p + 20); 2673 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2674 filename_length = archive_le16dec(p + 28); 2675 extra_length = archive_le16dec(p + 30); 2676 comment_length = archive_le16dec(p + 32); 2677 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2678 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2679 external_attributes = archive_le32dec(p + 38); 2680 zip_entry->local_header_offset = 2681 archive_le32dec(p + 42) + correction; 2682 2683 /* If we can't guess the mode, leave it zero here; 2684 when we read the local file header we might get 2685 more information. */ 2686 if (zip_entry->system == 3) { 2687 zip_entry->mode = external_attributes >> 16; 2688 } else if (zip_entry->system == 0) { 2689 // Interpret MSDOS directory bit 2690 if (0x10 == (external_attributes & 0x10)) { 2691 zip_entry->mode = AE_IFDIR | 0775; 2692 } else { 2693 zip_entry->mode = AE_IFREG | 0664; 2694 } 2695 if (0x01 == (external_attributes & 0x01)) { 2696 // Read-only bit; strip write permissions 2697 zip_entry->mode &= 0555; 2698 } 2699 } else { 2700 zip_entry->mode = 0; 2701 } 2702 2703 /* We're done with the regular data; get the filename and 2704 * extra data. */ 2705 __archive_read_consume(a, 46); 2706 p = __archive_read_ahead(a, filename_length + extra_length, 2707 NULL); 2708 if (p == NULL) { 2709 archive_set_error(&a->archive, 2710 ARCHIVE_ERRNO_FILE_FORMAT, 2711 "Truncated ZIP file header"); 2712 return ARCHIVE_FATAL; 2713 } 2714 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2715 return ARCHIVE_FATAL; 2716 } 2717 2718 /* 2719 * Mac resource fork files are stored under the 2720 * "__MACOSX/" directory, so we should check if 2721 * it is. 2722 */ 2723 if (!zip->process_mac_extensions) { 2724 /* Treat every entry as a regular entry. */ 2725 __archive_rb_tree_insert_node(&zip->tree, 2726 &zip_entry->node); 2727 } else { 2728 name = p; 2729 r = rsrc_basename(name, filename_length); 2730 if (filename_length >= 9 && 2731 strncmp("__MACOSX/", name, 9) == 0) { 2732 /* If this file is not a resource fork nor 2733 * a directory. We should treat it as a non 2734 * resource fork file to expose it. */ 2735 if (name[filename_length-1] != '/' && 2736 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2737 __archive_rb_tree_insert_node( 2738 &zip->tree, &zip_entry->node); 2739 /* Expose its parent directories. */ 2740 expose_parent_dirs(zip, name, 2741 filename_length); 2742 } else { 2743 /* This file is a resource fork file or 2744 * a directory. */ 2745 archive_strncpy(&(zip_entry->rsrcname), 2746 name, filename_length); 2747 __archive_rb_tree_insert_node( 2748 &zip->tree_rsrc, &zip_entry->node); 2749 } 2750 } else { 2751 /* Generate resource fork name to find its 2752 * resource file at zip->tree_rsrc. */ 2753 archive_strcpy(&(zip_entry->rsrcname), 2754 "__MACOSX/"); 2755 archive_strncat(&(zip_entry->rsrcname), 2756 name, r - name); 2757 archive_strcat(&(zip_entry->rsrcname), "._"); 2758 archive_strncat(&(zip_entry->rsrcname), 2759 name + (r - name), 2760 filename_length - (r - name)); 2761 /* Register an entry to RB tree to sort it by 2762 * file offset. */ 2763 __archive_rb_tree_insert_node(&zip->tree, 2764 &zip_entry->node); 2765 } 2766 } 2767 2768 /* Skip the comment too ... */ 2769 __archive_read_consume(a, 2770 filename_length + extra_length + comment_length); 2771 } 2772 2773 return ARCHIVE_OK; 2774} 2775 2776static ssize_t 2777zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2778{ 2779 const char *p; 2780 ssize_t filename_length, extra_length; 2781 2782 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2783 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2784 "Truncated ZIP file header"); 2785 return (ARCHIVE_WARN); 2786 } 2787 p += extra; 2788 2789 if (memcmp(p, "PK\003\004", 4) != 0) { 2790 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2791 return ARCHIVE_WARN; 2792 } 2793 filename_length = archive_le16dec(p + 26); 2794 extra_length = archive_le16dec(p + 28); 2795 2796 return (30 + filename_length + extra_length); 2797} 2798 2799static int 2800zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2801 struct zip_entry *rsrc) 2802{ 2803 struct zip *zip = (struct zip *)a->format->data; 2804 unsigned char *metadata, *mp; 2805 int64_t offset = archive_filter_bytes(&a->archive, 0); 2806 size_t remaining_bytes, metadata_bytes; 2807 ssize_t hsize; 2808 int ret = ARCHIVE_OK, eof; 2809 2810 switch(rsrc->compression) { 2811 case 0: /* No compression. */ 2812 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2813 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2814 "Malformed OS X metadata entry: inconsistent size"); 2815 return (ARCHIVE_FATAL); 2816 } 2817#ifdef HAVE_ZLIB_H 2818 case 8: /* Deflate compression. */ 2819#endif 2820 break; 2821 default: /* Unsupported compression. */ 2822 /* Return a warning. */ 2823 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2824 "Unsupported ZIP compression method (%s)", 2825 compression_name(rsrc->compression)); 2826 /* We can't decompress this entry, but we will 2827 * be able to skip() it and try the next entry. */ 2828 return (ARCHIVE_WARN); 2829 } 2830 2831 if (rsrc->uncompressed_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->uncompressed_size); 2835 return (ARCHIVE_WARN); 2836 } 2837 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2838 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2839 "Mac metadata is too large: %jd > 4M bytes", 2840 (intmax_t)rsrc->compressed_size); 2841 return (ARCHIVE_WARN); 2842 } 2843 2844 metadata = malloc((size_t)rsrc->uncompressed_size); 2845 if (metadata == NULL) { 2846 archive_set_error(&a->archive, ENOMEM, 2847 "Can't allocate memory for Mac metadata"); 2848 return (ARCHIVE_FATAL); 2849 } 2850 2851 if (offset < rsrc->local_header_offset) 2852 __archive_read_consume(a, rsrc->local_header_offset - offset); 2853 else if (offset != rsrc->local_header_offset) { 2854 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2855 } 2856 2857 hsize = zip_get_local_file_header_size(a, 0); 2858 __archive_read_consume(a, hsize); 2859 2860 remaining_bytes = (size_t)rsrc->compressed_size; 2861 metadata_bytes = (size_t)rsrc->uncompressed_size; 2862 mp = metadata; 2863 eof = 0; 2864 while (!eof && remaining_bytes) { 2865 const unsigned char *p; 2866 ssize_t bytes_avail; 2867 size_t bytes_used; 2868 2869 p = __archive_read_ahead(a, 1, &bytes_avail); 2870 if (p == NULL) { 2871 archive_set_error(&a->archive, 2872 ARCHIVE_ERRNO_FILE_FORMAT, 2873 "Truncated ZIP file header"); 2874 ret = ARCHIVE_WARN; 2875 goto exit_mac_metadata; 2876 } 2877 if ((size_t)bytes_avail > remaining_bytes) 2878 bytes_avail = remaining_bytes; 2879 switch(rsrc->compression) { 2880 case 0: /* No compression. */ 2881 if ((size_t)bytes_avail > metadata_bytes) 2882 bytes_avail = metadata_bytes; 2883 memcpy(mp, p, bytes_avail); 2884 bytes_used = (size_t)bytes_avail; 2885 metadata_bytes -= bytes_used; 2886 mp += bytes_used; 2887 if (metadata_bytes == 0) 2888 eof = 1; 2889 break; 2890#ifdef HAVE_ZLIB_H 2891 case 8: /* Deflate compression. */ 2892 { 2893 int r; 2894 2895 ret = zip_deflate_init(a, zip); 2896 if (ret != ARCHIVE_OK) 2897 goto exit_mac_metadata; 2898 zip->stream.next_in = 2899 (Bytef *)(uintptr_t)(const void *)p; 2900 zip->stream.avail_in = (uInt)bytes_avail; 2901 zip->stream.total_in = 0; 2902 zip->stream.next_out = mp; 2903 zip->stream.avail_out = (uInt)metadata_bytes; 2904 zip->stream.total_out = 0; 2905 2906 r = inflate(&zip->stream, 0); 2907 switch (r) { 2908 case Z_OK: 2909 break; 2910 case Z_STREAM_END: 2911 eof = 1; 2912 break; 2913 case Z_MEM_ERROR: 2914 archive_set_error(&a->archive, ENOMEM, 2915 "Out of memory for ZIP decompression"); 2916 ret = ARCHIVE_FATAL; 2917 goto exit_mac_metadata; 2918 default: 2919 archive_set_error(&a->archive, 2920 ARCHIVE_ERRNO_MISC, 2921 "ZIP decompression failed (%d)", r); 2922 ret = ARCHIVE_FATAL; 2923 goto exit_mac_metadata; 2924 } 2925 bytes_used = zip->stream.total_in; 2926 metadata_bytes -= zip->stream.total_out; 2927 mp += zip->stream.total_out; 2928 break; 2929 } 2930#endif 2931 default: 2932 bytes_used = 0; 2933 break; 2934 } 2935 __archive_read_consume(a, bytes_used); 2936 remaining_bytes -= bytes_used; 2937 } 2938 archive_entry_copy_mac_metadata(entry, metadata, 2939 (size_t)rsrc->uncompressed_size - metadata_bytes); 2940 2941exit_mac_metadata: 2942 __archive_read_seek(a, offset, SEEK_SET); 2943 zip->decompress_init = 0; 2944 free(metadata); 2945 return (ret); 2946} 2947 2948static int 2949archive_read_format_zip_seekable_read_header(struct archive_read *a, 2950 struct archive_entry *entry) 2951{ 2952 struct zip *zip = (struct zip *)a->format->data; 2953 struct zip_entry *rsrc; 2954 int64_t offset; 2955 int r, ret = ARCHIVE_OK; 2956 2957 /* 2958 * It should be sufficient to call archive_read_next_header() for 2959 * a reader to determine if an entry is encrypted or not. If the 2960 * encryption of an entry is only detectable when calling 2961 * archive_read_data(), so be it. We'll do the same check there 2962 * as well. 2963 */ 2964 if (zip->has_encrypted_entries == 2965 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2966 zip->has_encrypted_entries = 0; 2967 2968 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2969 if (a->archive.archive_format_name == NULL) 2970 a->archive.archive_format_name = "ZIP"; 2971 2972 if (zip->zip_entries == NULL) { 2973 r = slurp_central_directory(a, zip); 2974 if (r != ARCHIVE_OK) 2975 return r; 2976 /* Get first entry whose local header offset is lower than 2977 * other entries in the archive file. */ 2978 zip->entry = 2979 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 2980 } else if (zip->entry != NULL) { 2981 /* Get next entry in local header offset order. */ 2982 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 2983 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 2984 } 2985 2986 if (zip->entry == NULL) 2987 return ARCHIVE_EOF; 2988 2989 if (zip->entry->rsrcname.s) 2990 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 2991 &zip->tree_rsrc, zip->entry->rsrcname.s); 2992 else 2993 rsrc = NULL; 2994 2995 if (zip->cctx_valid) 2996 archive_decrypto_aes_ctr_release(&zip->cctx); 2997 if (zip->hctx_valid) 2998 archive_hmac_sha1_cleanup(&zip->hctx); 2999 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3000 __archive_read_reset_passphrase(a); 3001 3002 /* File entries are sorted by the header offset, we should mostly 3003 * use __archive_read_consume to advance a read point to avoid redundant 3004 * data reading. */ 3005 offset = archive_filter_bytes(&a->archive, 0); 3006 if (offset < zip->entry->local_header_offset) 3007 __archive_read_consume(a, 3008 zip->entry->local_header_offset - offset); 3009 else if (offset != zip->entry->local_header_offset) { 3010 __archive_read_seek(a, zip->entry->local_header_offset, 3011 SEEK_SET); 3012 } 3013 zip->unconsumed = 0; 3014 r = zip_read_local_file_header(a, entry, zip); 3015 if (r != ARCHIVE_OK) 3016 return r; 3017 if (rsrc) { 3018 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3019 if (ret2 < ret) 3020 ret = ret2; 3021 } 3022 return (ret); 3023} 3024 3025/* 3026 * We're going to seek for the next header anyway, so we don't 3027 * need to bother doing anything here. 3028 */ 3029static int 3030archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3031{ 3032 struct zip *zip; 3033 zip = (struct zip *)(a->format->data); 3034 3035 zip->unconsumed = 0; 3036 return (ARCHIVE_OK); 3037} 3038 3039int 3040archive_read_support_format_zip_seekable(struct archive *_a) 3041{ 3042 struct archive_read *a = (struct archive_read *)_a; 3043 struct zip *zip; 3044 int r; 3045 3046 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3047 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3048 3049 zip = (struct zip *)calloc(1, sizeof(*zip)); 3050 if (zip == NULL) { 3051 archive_set_error(&a->archive, ENOMEM, 3052 "Can't allocate zip data"); 3053 return (ARCHIVE_FATAL); 3054 } 3055 3056#ifdef HAVE_COPYFILE_H 3057 /* Set this by default on Mac OS. */ 3058 zip->process_mac_extensions = 1; 3059#endif 3060 3061 /* 3062 * Until enough data has been read, we cannot tell about 3063 * any encrypted entries yet. 3064 */ 3065 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3066 zip->crc32func = real_crc32; 3067 3068 r = __archive_read_register_format(a, 3069 zip, 3070 "zip", 3071 archive_read_format_zip_seekable_bid, 3072 archive_read_format_zip_options, 3073 archive_read_format_zip_seekable_read_header, 3074 archive_read_format_zip_read_data, 3075 archive_read_format_zip_read_data_skip_seekable, 3076 NULL, 3077 archive_read_format_zip_cleanup, 3078 archive_read_support_format_zip_capabilities_seekable, 3079 archive_read_format_zip_has_encrypted_entries); 3080 3081 if (r != ARCHIVE_OK) 3082 free(zip); 3083 return (ARCHIVE_OK); 3084} 3085