archive_read_support_format_zip.c revision 324417
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 324417 2017-10-08 20:54:53Z mm $"); 30 31/* 32 * The definitive documentation of the Zip file format is: 33 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT 34 * 35 * The Info-Zip project has pioneered various extensions to better 36 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 37 * "Ux", and 0x7875 "ux" extensions for time and ownership 38 * information. 39 * 40 * History of this code: The streaming Zip reader was first added to 41 * libarchive in January 2005. Support for seekable input sources was 42 * added in Nov 2011. Zip64 support (including a significant code 43 * refactoring) was added in 2014. 44 */ 45 46#ifdef HAVE_ERRNO_H 47#include <errno.h> 48#endif 49#ifdef HAVE_STDLIB_H 50#include <stdlib.h> 51#endif 52#ifdef HAVE_ZLIB_H 53#include <zlib.h> 54#endif 55 56#include "archive.h" 57#include "archive_digest_private.h" 58#include "archive_cryptor_private.h" 59#include "archive_endian.h" 60#include "archive_entry.h" 61#include "archive_entry_locale.h" 62#include "archive_hmac_private.h" 63#include "archive_private.h" 64#include "archive_rb.h" 65#include "archive_read_private.h" 66 67#ifndef HAVE_ZLIB_H 68#include "archive_crc32.h" 69#endif 70 71struct zip_entry { 72 struct archive_rb_node node; 73 struct zip_entry *next; 74 int64_t local_header_offset; 75 int64_t compressed_size; 76 int64_t uncompressed_size; 77 int64_t gid; 78 int64_t uid; 79 struct archive_string rsrcname; 80 time_t mtime; 81 time_t atime; 82 time_t ctime; 83 uint32_t crc32; 84 uint16_t mode; 85 uint16_t zip_flags; /* From GP Flags Field */ 86 unsigned char compression; 87 unsigned char system; /* From "version written by" */ 88 unsigned char flags; /* Our extra markers. */ 89 unsigned char decdat;/* Used for Decryption check */ 90 91 /* WinZip AES encryption extra field should be available 92 * when compression is 99. */ 93 struct { 94 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 95 unsigned vendor; 96#define AES_VENDOR_AE_1 0x0001 97#define AES_VENDOR_AE_2 0x0002 98 /* AES encryption strength: 99 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 100 unsigned strength; 101 /* Actual compression method. */ 102 unsigned char compression; 103 } aes_extra; 104}; 105 106struct trad_enc_ctx { 107 uint32_t keys[3]; 108}; 109 110/* Bits used in zip_flags. */ 111#define ZIP_ENCRYPTED (1 << 0) 112#define ZIP_LENGTH_AT_END (1 << 3) 113#define ZIP_STRONG_ENCRYPTED (1 << 6) 114#define ZIP_UTF8_NAME (1 << 11) 115/* See "7.2 Single Password Symmetric Encryption Method" 116 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 117#define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 118 119/* Bits used in flags. */ 120#define LA_USED_ZIP64 (1 << 0) 121#define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 122 123/* 124 * See "WinZip - AES Encryption Information" 125 * http://www.winzip.com/aes_info.htm 126 */ 127/* Value used in compression method. */ 128#define WINZIP_AES_ENCRYPTION 99 129/* Authentication code size. */ 130#define AUTH_CODE_SIZE 10 131/**/ 132#define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 133 134struct zip { 135 /* Structural information about the archive. */ 136 struct archive_string format_name; 137 int64_t central_directory_offset; 138 size_t central_directory_entries_total; 139 size_t central_directory_entries_on_this_disk; 140 int has_encrypted_entries; 141 142 /* List of entries (seekable Zip only) */ 143 struct zip_entry *zip_entries; 144 struct archive_rb_tree tree; 145 struct archive_rb_tree tree_rsrc; 146 147 /* Bytes read but not yet consumed via __archive_read_consume() */ 148 size_t unconsumed; 149 150 /* Information about entry we're currently reading. */ 151 struct zip_entry *entry; 152 int64_t entry_bytes_remaining; 153 154 /* These count the number of bytes actually read for the entry. */ 155 int64_t entry_compressed_bytes_read; 156 int64_t entry_uncompressed_bytes_read; 157 158 /* Running CRC32 of the decompressed data */ 159 unsigned long entry_crc32; 160 unsigned long (*crc32func)(unsigned long, const void *, 161 size_t); 162 char ignore_crc32; 163 164 /* Flags to mark progress of decompression. */ 165 char decompress_init; 166 char end_of_entry; 167 168#ifdef HAVE_ZLIB_H 169 unsigned char *uncompressed_buffer; 170 size_t uncompressed_buffer_size; 171 z_stream stream; 172 char stream_valid; 173#endif 174 175 struct archive_string_conv *sconv; 176 struct archive_string_conv *sconv_default; 177 struct archive_string_conv *sconv_utf8; 178 int init_default_conversion; 179 int process_mac_extensions; 180 181 char init_decryption; 182 183 /* Decryption buffer. */ 184 /* 185 * The decrypted data starts at decrypted_ptr and 186 * extends for decrypted_bytes_remaining. Decryption 187 * adds new data to the end of this block, data is returned 188 * to clients from the beginning. When the block hits the 189 * end of decrypted_buffer, it has to be shuffled back to 190 * the beginning of the buffer. 191 */ 192 unsigned char *decrypted_buffer; 193 unsigned char *decrypted_ptr; 194 size_t decrypted_buffer_size; 195 size_t decrypted_bytes_remaining; 196 size_t decrypted_unconsumed_bytes; 197 198 /* Traditional PKWARE decryption. */ 199 struct trad_enc_ctx tctx; 200 char tctx_valid; 201 202 /* WinZip AES decryption. */ 203 /* Contexts used for AES decryption. */ 204 archive_crypto_ctx cctx; 205 char cctx_valid; 206 archive_hmac_sha1_ctx hctx; 207 char hctx_valid; 208 209 /* Strong encryption's decryption header information. */ 210 unsigned iv_size; 211 unsigned alg_id; 212 unsigned bit_len; 213 unsigned flags; 214 unsigned erd_size; 215 unsigned v_size; 216 unsigned v_crc32; 217 uint8_t *iv; 218 uint8_t *erd; 219 uint8_t *v_data; 220}; 221 222/* Many systems define min or MIN, but not all. */ 223#define zipmin(a,b) ((a) < (b) ? (a) : (b)) 224 225/* ------------------------------------------------------------------------ */ 226 227/* 228 Traditional PKWARE Decryption functions. 229 */ 230 231static void 232trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 233{ 234 uint8_t t; 235#define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 236 237 ctx->keys[0] = CRC32(ctx->keys[0], c); 238 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 239 t = (ctx->keys[1] >> 24) & 0xff; 240 ctx->keys[2] = CRC32(ctx->keys[2], t); 241#undef CRC32 242} 243 244static uint8_t 245trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) 246{ 247 unsigned temp = ctx->keys[2] | 2; 248 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 249} 250 251static void 252trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 253 size_t in_len, uint8_t *out, size_t out_len) 254{ 255 unsigned i, max; 256 257 max = (unsigned)((in_len < out_len)? in_len: out_len); 258 259 for (i = 0; i < max; i++) { 260 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); 261 out[i] = t; 262 trad_enc_update_keys(ctx, t); 263 } 264} 265 266static int 267trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 268 const uint8_t *key, size_t key_len, uint8_t *crcchk) 269{ 270 uint8_t header[12]; 271 272 if (key_len < 12) { 273 *crcchk = 0xff; 274 return -1; 275 } 276 277 ctx->keys[0] = 305419896L; 278 ctx->keys[1] = 591751049L; 279 ctx->keys[2] = 878082192L; 280 281 for (;pw_len; --pw_len) 282 trad_enc_update_keys(ctx, *pw++); 283 284 trad_enc_decrypt_update(ctx, key, 12, header, 12); 285 /* Return the last byte for CRC check. */ 286 *crcchk = header[11]; 287 return 0; 288} 289 290#if 0 291static void 292crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 293 int key_size) 294{ 295#define MD_SIZE 20 296 archive_sha1_ctx ctx; 297 unsigned char md1[MD_SIZE]; 298 unsigned char md2[MD_SIZE * 2]; 299 unsigned char mkb[64]; 300 int i; 301 302 archive_sha1_init(&ctx); 303 archive_sha1_update(&ctx, p, size); 304 archive_sha1_final(&ctx, md1); 305 306 memset(mkb, 0x36, sizeof(mkb)); 307 for (i = 0; i < MD_SIZE; i++) 308 mkb[i] ^= md1[i]; 309 archive_sha1_init(&ctx); 310 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 311 archive_sha1_final(&ctx, md2); 312 313 memset(mkb, 0x5C, sizeof(mkb)); 314 for (i = 0; i < MD_SIZE; i++) 315 mkb[i] ^= md1[i]; 316 archive_sha1_init(&ctx); 317 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 318 archive_sha1_final(&ctx, md2 + MD_SIZE); 319 320 if (key_size > 32) 321 key_size = 32; 322 memcpy(key, md2, key_size); 323#undef MD_SIZE 324} 325#endif 326 327/* 328 * Common code for streaming or seeking modes. 329 * 330 * Includes code to read local file headers, decompress data 331 * from entry bodies, and common API. 332 */ 333 334static unsigned long 335real_crc32(unsigned long crc, const void *buff, size_t len) 336{ 337 return crc32(crc, buff, (unsigned int)len); 338} 339 340/* Used by "ignorecrc32" option to speed up tests. */ 341static unsigned long 342fake_crc32(unsigned long crc, const void *buff, size_t len) 343{ 344 (void)crc; /* UNUSED */ 345 (void)buff; /* UNUSED */ 346 (void)len; /* UNUSED */ 347 return 0; 348} 349 350static const struct { 351 int id; 352 const char * name; 353} compression_methods[] = { 354 {0, "uncompressed"}, /* The file is stored (no compression) */ 355 {1, "shrinking"}, /* The file is Shrunk */ 356 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 357 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 358 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 359 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 360 {6, "imploded"}, /* The file is Imploded */ 361 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 362 {8, "deflation"}, /* The file is Deflated */ 363 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 364 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 365 * (old IBM TERSE) */ 366 {11, "reserved"}, /* Reserved by PKWARE */ 367 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 368 {13, "reserved"}, /* Reserved by PKWARE */ 369 {14, "lzma"}, /* LZMA (EFS) */ 370 {15, "reserved"}, /* Reserved by PKWARE */ 371 {16, "reserved"}, /* Reserved by PKWARE */ 372 {17, "reserved"}, /* Reserved by PKWARE */ 373 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 374 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 375 {97, "wav-pack"}, /* WavPack compressed data */ 376 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 377 {99, "aes"} /* WinZip AES encryption */ 378}; 379 380static const char * 381compression_name(const int compression) 382{ 383 static const int num_compression_methods = 384 sizeof(compression_methods)/sizeof(compression_methods[0]); 385 int i=0; 386 387 while(compression >= 0 && i < num_compression_methods) { 388 if (compression_methods[i].id == compression) 389 return compression_methods[i].name; 390 i++; 391 } 392 return "??"; 393} 394 395/* Convert an MSDOS-style date/time into Unix-style time. */ 396static time_t 397zip_time(const char *p) 398{ 399 int msTime, msDate; 400 struct tm ts; 401 402 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 403 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 404 405 memset(&ts, 0, sizeof(ts)); 406 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 407 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 408 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 409 ts.tm_hour = (msTime >> 11) & 0x1f; 410 ts.tm_min = (msTime >> 5) & 0x3f; 411 ts.tm_sec = (msTime << 1) & 0x3e; 412 ts.tm_isdst = -1; 413 return mktime(&ts); 414} 415 416/* 417 * The extra data is stored as a list of 418 * id1+size1+data1 + id2+size2+data2 ... 419 * triplets. id and size are 2 bytes each. 420 */ 421static int 422process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry) 423{ 424 unsigned offset = 0; 425 426 if (extra_length == 0) { 427 return ARCHIVE_OK; 428 } 429 430 if (extra_length < 4) { 431 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 432 "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length); 433 return ARCHIVE_FAILED; 434 } 435 while (offset <= extra_length - 4) { 436 unsigned short headerid = archive_le16dec(p + offset); 437 unsigned short datasize = archive_le16dec(p + offset + 2); 438 439 offset += 4; 440 if (offset + datasize > extra_length) { 441 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 442 "Extra data overflow: Need %d bytes but only found %d bytes", 443 (int)datasize, (int)(extra_length - offset)); 444 return ARCHIVE_FAILED; 445 } 446#ifdef DEBUG 447 fprintf(stderr, "Header id 0x%04x, length %d\n", 448 headerid, datasize); 449#endif 450 switch (headerid) { 451 case 0x0001: 452 /* Zip64 extended information extra field. */ 453 zip_entry->flags |= LA_USED_ZIP64; 454 if (zip_entry->uncompressed_size == 0xffffffff) { 455 uint64_t t = 0; 456 if (datasize < 8 457 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 458 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 459 "Malformed 64-bit uncompressed size"); 460 return ARCHIVE_FAILED; 461 } 462 zip_entry->uncompressed_size = t; 463 offset += 8; 464 datasize -= 8; 465 } 466 if (zip_entry->compressed_size == 0xffffffff) { 467 uint64_t t = 0; 468 if (datasize < 8 469 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 470 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 471 "Malformed 64-bit compressed size"); 472 return ARCHIVE_FAILED; 473 } 474 zip_entry->compressed_size = t; 475 offset += 8; 476 datasize -= 8; 477 } 478 if (zip_entry->local_header_offset == 0xffffffff) { 479 uint64_t t = 0; 480 if (datasize < 8 481 || (t = archive_le64dec(p + offset)) > INT64_MAX) { 482 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 483 "Malformed 64-bit local header offset"); 484 return ARCHIVE_FAILED; 485 } 486 zip_entry->local_header_offset = t; 487 offset += 8; 488 datasize -= 8; 489 } 490 /* archive_le32dec(p + offset) gives disk 491 * on which file starts, but we don't handle 492 * multi-volume Zip files. */ 493 break; 494#ifdef DEBUG 495 case 0x0017: 496 { 497 /* Strong encryption field. */ 498 if (archive_le16dec(p + offset) == 2) { 499 unsigned algId = 500 archive_le16dec(p + offset + 2); 501 unsigned bitLen = 502 archive_le16dec(p + offset + 4); 503 int flags = 504 archive_le16dec(p + offset + 6); 505 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 506 "flgas=%d\n", algId, bitLen,flags); 507 } 508 break; 509 } 510#endif 511 case 0x5455: 512 { 513 /* Extended time field "UT". */ 514 int flags = p[offset]; 515 offset++; 516 datasize--; 517 /* Flag bits indicate which dates are present. */ 518 if (flags & 0x01) 519 { 520#ifdef DEBUG 521 fprintf(stderr, "mtime: %lld -> %d\n", 522 (long long)zip_entry->mtime, 523 archive_le32dec(p + offset)); 524#endif 525 if (datasize < 4) 526 break; 527 zip_entry->mtime = archive_le32dec(p + offset); 528 offset += 4; 529 datasize -= 4; 530 } 531 if (flags & 0x02) 532 { 533 if (datasize < 4) 534 break; 535 zip_entry->atime = archive_le32dec(p + offset); 536 offset += 4; 537 datasize -= 4; 538 } 539 if (flags & 0x04) 540 { 541 if (datasize < 4) 542 break; 543 zip_entry->ctime = archive_le32dec(p + offset); 544 offset += 4; 545 datasize -= 4; 546 } 547 break; 548 } 549 case 0x5855: 550 { 551 /* Info-ZIP Unix Extra Field (old version) "UX". */ 552 if (datasize >= 8) { 553 zip_entry->atime = archive_le32dec(p + offset); 554 zip_entry->mtime = 555 archive_le32dec(p + offset + 4); 556 } 557 if (datasize >= 12) { 558 zip_entry->uid = 559 archive_le16dec(p + offset + 8); 560 zip_entry->gid = 561 archive_le16dec(p + offset + 10); 562 } 563 break; 564 } 565 case 0x6c78: 566 { 567 /* Experimental 'xl' field */ 568 /* 569 * Introduced Dec 2013 to provide a way to 570 * include external file attributes (and other 571 * fields that ordinarily appear only in 572 * central directory) in local file header. 573 * This provides file type and permission 574 * information necessary to support full 575 * streaming extraction. Currently being 576 * discussed with other Zip developers 577 * ... subject to change. 578 * 579 * Format: 580 * The field starts with a bitmap that specifies 581 * which additional fields are included. The 582 * bitmap is variable length and can be extended in 583 * the future. 584 * 585 * n bytes - feature bitmap: first byte has low-order 586 * 7 bits. If high-order bit is set, a subsequent 587 * byte holds the next 7 bits, etc. 588 * 589 * if bitmap & 1, 2 byte "version made by" 590 * if bitmap & 2, 2 byte "internal file attributes" 591 * if bitmap & 4, 4 byte "external file attributes" 592 * if bitmap & 8, 2 byte comment length + n byte comment 593 */ 594 int bitmap, bitmap_last; 595 596 if (datasize < 1) 597 break; 598 bitmap_last = bitmap = 0xff & p[offset]; 599 offset += 1; 600 datasize -= 1; 601 602 /* We only support first 7 bits of bitmap; skip rest. */ 603 while ((bitmap_last & 0x80) != 0 604 && datasize >= 1) { 605 bitmap_last = p[offset]; 606 offset += 1; 607 datasize -= 1; 608 } 609 610 if (bitmap & 1) { 611 /* 2 byte "version made by" */ 612 if (datasize < 2) 613 break; 614 zip_entry->system 615 = archive_le16dec(p + offset) >> 8; 616 offset += 2; 617 datasize -= 2; 618 } 619 if (bitmap & 2) { 620 /* 2 byte "internal file attributes" */ 621 uint32_t internal_attributes; 622 if (datasize < 2) 623 break; 624 internal_attributes 625 = archive_le16dec(p + offset); 626 /* Not used by libarchive at present. */ 627 (void)internal_attributes; /* UNUSED */ 628 offset += 2; 629 datasize -= 2; 630 } 631 if (bitmap & 4) { 632 /* 4 byte "external file attributes" */ 633 uint32_t external_attributes; 634 if (datasize < 4) 635 break; 636 external_attributes 637 = archive_le32dec(p + offset); 638 if (zip_entry->system == 3) { 639 zip_entry->mode 640 = external_attributes >> 16; 641 } else if (zip_entry->system == 0) { 642 // Interpret MSDOS directory bit 643 if (0x10 == (external_attributes & 0x10)) { 644 zip_entry->mode = AE_IFDIR | 0775; 645 } else { 646 zip_entry->mode = AE_IFREG | 0664; 647 } 648 if (0x01 == (external_attributes & 0x01)) { 649 // Read-only bit; strip write permissions 650 zip_entry->mode &= 0555; 651 } 652 } else { 653 zip_entry->mode = 0; 654 } 655 offset += 4; 656 datasize -= 4; 657 } 658 if (bitmap & 8) { 659 /* 2 byte comment length + comment */ 660 uint32_t comment_length; 661 if (datasize < 2) 662 break; 663 comment_length 664 = archive_le16dec(p + offset); 665 offset += 2; 666 datasize -= 2; 667 668 if (datasize < comment_length) 669 break; 670 /* Comment is not supported by libarchive */ 671 offset += comment_length; 672 datasize -= comment_length; 673 } 674 break; 675 } 676 case 0x7855: 677 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 678#ifdef DEBUG 679 fprintf(stderr, "uid %d gid %d\n", 680 archive_le16dec(p + offset), 681 archive_le16dec(p + offset + 2)); 682#endif 683 if (datasize >= 2) 684 zip_entry->uid = archive_le16dec(p + offset); 685 if (datasize >= 4) 686 zip_entry->gid = 687 archive_le16dec(p + offset + 2); 688 break; 689 case 0x7875: 690 { 691 /* Info-Zip Unix Extra Field (type 3) "ux". */ 692 int uidsize = 0, gidsize = 0; 693 694 /* TODO: support arbitrary uidsize/gidsize. */ 695 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 696 if (datasize >= 4) { 697 /* get a uid size. */ 698 uidsize = 0xff & (int)p[offset+1]; 699 if (uidsize == 2) 700 zip_entry->uid = 701 archive_le16dec( 702 p + offset + 2); 703 else if (uidsize == 4 && datasize >= 6) 704 zip_entry->uid = 705 archive_le32dec( 706 p + offset + 2); 707 } 708 if (datasize >= (2 + uidsize + 3)) { 709 /* get a gid size. */ 710 gidsize = 0xff & (int)p[offset+2+uidsize]; 711 if (gidsize == 2) 712 zip_entry->gid = 713 archive_le16dec( 714 p+offset+2+uidsize+1); 715 else if (gidsize == 4 && 716 datasize >= (2 + uidsize + 5)) 717 zip_entry->gid = 718 archive_le32dec( 719 p+offset+2+uidsize+1); 720 } 721 } 722 break; 723 } 724 case 0x9901: 725 /* WinZip AES extra data field. */ 726 if (datasize < 6) { 727 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 728 "Incomplete AES field"); 729 return ARCHIVE_FAILED; 730 } 731 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 732 /* Vendor version. */ 733 zip_entry->aes_extra.vendor = 734 archive_le16dec(p + offset); 735 /* AES encryption strength. */ 736 zip_entry->aes_extra.strength = p[offset + 4]; 737 /* Actual compression method. */ 738 zip_entry->aes_extra.compression = 739 p[offset + 5]; 740 } 741 break; 742 default: 743 break; 744 } 745 offset += datasize; 746 } 747 if (offset != extra_length) { 748 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 749 "Malformed extra data: Consumed %d bytes of %d bytes", 750 (int)offset, (int)extra_length); 751 return ARCHIVE_FAILED; 752 } 753 return ARCHIVE_OK; 754} 755 756/* 757 * Assumes file pointer is at beginning of local file header. 758 */ 759static int 760zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 761 struct zip *zip) 762{ 763 const char *p; 764 const void *h; 765 const wchar_t *wp; 766 const char *cp; 767 size_t len, filename_length, extra_length; 768 struct archive_string_conv *sconv; 769 struct zip_entry *zip_entry = zip->entry; 770 struct zip_entry zip_entry_central_dir; 771 int ret = ARCHIVE_OK; 772 char version; 773 774 /* Save a copy of the original for consistency checks. */ 775 zip_entry_central_dir = *zip_entry; 776 777 zip->decompress_init = 0; 778 zip->end_of_entry = 0; 779 zip->entry_uncompressed_bytes_read = 0; 780 zip->entry_compressed_bytes_read = 0; 781 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 782 783 /* Setup default conversion. */ 784 if (zip->sconv == NULL && !zip->init_default_conversion) { 785 zip->sconv_default = 786 archive_string_default_conversion_for_read(&(a->archive)); 787 zip->init_default_conversion = 1; 788 } 789 790 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 791 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 792 "Truncated ZIP file header"); 793 return (ARCHIVE_FATAL); 794 } 795 796 if (memcmp(p, "PK\003\004", 4) != 0) { 797 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 798 return ARCHIVE_FATAL; 799 } 800 version = p[4]; 801 zip_entry->system = p[5]; 802 zip_entry->zip_flags = archive_le16dec(p + 6); 803 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 804 zip->has_encrypted_entries = 1; 805 archive_entry_set_is_data_encrypted(entry, 1); 806 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 807 zip_entry->zip_flags & ZIP_ENCRYPTED && 808 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 809 archive_entry_set_is_metadata_encrypted(entry, 1); 810 return ARCHIVE_FATAL; 811 } 812 } 813 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 814 zip_entry->compression = (char)archive_le16dec(p + 8); 815 zip_entry->mtime = zip_time(p + 10); 816 zip_entry->crc32 = archive_le32dec(p + 14); 817 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 818 zip_entry->decdat = p[11]; 819 else 820 zip_entry->decdat = p[17]; 821 zip_entry->compressed_size = archive_le32dec(p + 18); 822 zip_entry->uncompressed_size = archive_le32dec(p + 22); 823 filename_length = archive_le16dec(p + 26); 824 extra_length = archive_le16dec(p + 28); 825 826 __archive_read_consume(a, 30); 827 828 /* Read the filename. */ 829 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 830 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 831 "Truncated ZIP file header"); 832 return (ARCHIVE_FATAL); 833 } 834 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 835 /* The filename is stored to be UTF-8. */ 836 if (zip->sconv_utf8 == NULL) { 837 zip->sconv_utf8 = 838 archive_string_conversion_from_charset( 839 &a->archive, "UTF-8", 1); 840 if (zip->sconv_utf8 == NULL) 841 return (ARCHIVE_FATAL); 842 } 843 sconv = zip->sconv_utf8; 844 } else if (zip->sconv != NULL) 845 sconv = zip->sconv; 846 else 847 sconv = zip->sconv_default; 848 849 if (archive_entry_copy_pathname_l(entry, 850 h, filename_length, sconv) != 0) { 851 if (errno == ENOMEM) { 852 archive_set_error(&a->archive, ENOMEM, 853 "Can't allocate memory for Pathname"); 854 return (ARCHIVE_FATAL); 855 } 856 archive_set_error(&a->archive, 857 ARCHIVE_ERRNO_FILE_FORMAT, 858 "Pathname cannot be converted " 859 "from %s to current locale.", 860 archive_string_conversion_charset_name(sconv)); 861 ret = ARCHIVE_WARN; 862 } 863 __archive_read_consume(a, filename_length); 864 865 /* Read the extra data. */ 866 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 867 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 868 "Truncated ZIP file header"); 869 return (ARCHIVE_FATAL); 870 } 871 872 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 873 return ARCHIVE_FATAL; 874 } 875 __archive_read_consume(a, extra_length); 876 877 /* Work around a bug in Info-Zip: When reading from a pipe, it 878 * stats the pipe instead of synthesizing a file entry. */ 879 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 880 zip_entry->mode &= ~ AE_IFMT; 881 zip_entry->mode |= AE_IFREG; 882 } 883 884 /* If the mode is totally empty, set some sane default. */ 885 if (zip_entry->mode == 0) { 886 zip_entry->mode |= 0664; 887 } 888 889 /* Make sure that entries with a trailing '/' are marked as directories 890 * even if the External File Attributes contains bogus values. If this 891 * is not a directory and there is no type, assume regularfile. */ 892 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 893 int has_slash; 894 895 wp = archive_entry_pathname_w(entry); 896 if (wp != NULL) { 897 len = wcslen(wp); 898 has_slash = len > 0 && wp[len - 1] == L'/'; 899 } else { 900 cp = archive_entry_pathname(entry); 901 len = (cp != NULL)?strlen(cp):0; 902 has_slash = len > 0 && cp[len - 1] == '/'; 903 } 904 /* Correct file type as needed. */ 905 if (has_slash) { 906 zip_entry->mode &= ~AE_IFMT; 907 zip_entry->mode |= AE_IFDIR; 908 zip_entry->mode |= 0111; 909 } else if ((zip_entry->mode & AE_IFMT) == 0) { 910 zip_entry->mode |= AE_IFREG; 911 } 912 } 913 914 /* Make sure directories end in '/' */ 915 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 916 wp = archive_entry_pathname_w(entry); 917 if (wp != NULL) { 918 len = wcslen(wp); 919 if (len > 0 && wp[len - 1] != L'/') { 920 struct archive_wstring s; 921 archive_string_init(&s); 922 archive_wstrcat(&s, wp); 923 archive_wstrappend_wchar(&s, L'/'); 924 archive_entry_copy_pathname_w(entry, s.s); 925 archive_wstring_free(&s); 926 } 927 } else { 928 cp = archive_entry_pathname(entry); 929 len = (cp != NULL)?strlen(cp):0; 930 if (len > 0 && cp[len - 1] != '/') { 931 struct archive_string s; 932 archive_string_init(&s); 933 archive_strcat(&s, cp); 934 archive_strappend_char(&s, '/'); 935 archive_entry_set_pathname(entry, s.s); 936 archive_string_free(&s); 937 } 938 } 939 } 940 941 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 942 /* If this came from the central dir, it's size info 943 * is definitive, so ignore the length-at-end flag. */ 944 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 945 /* If local header is missing a value, use the one from 946 the central directory. If both have it, warn about 947 mismatches. */ 948 if (zip_entry->crc32 == 0) { 949 zip_entry->crc32 = zip_entry_central_dir.crc32; 950 } else if (!zip->ignore_crc32 951 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 952 archive_set_error(&a->archive, 953 ARCHIVE_ERRNO_FILE_FORMAT, 954 "Inconsistent CRC32 values"); 955 ret = ARCHIVE_WARN; 956 } 957 if (zip_entry->compressed_size == 0) { 958 zip_entry->compressed_size 959 = zip_entry_central_dir.compressed_size; 960 } else if (zip_entry->compressed_size 961 != zip_entry_central_dir.compressed_size) { 962 archive_set_error(&a->archive, 963 ARCHIVE_ERRNO_FILE_FORMAT, 964 "Inconsistent compressed size: " 965 "%jd in central directory, %jd in local header", 966 (intmax_t)zip_entry_central_dir.compressed_size, 967 (intmax_t)zip_entry->compressed_size); 968 ret = ARCHIVE_WARN; 969 } 970 if (zip_entry->uncompressed_size == 0) { 971 zip_entry->uncompressed_size 972 = zip_entry_central_dir.uncompressed_size; 973 } else if (zip_entry->uncompressed_size 974 != zip_entry_central_dir.uncompressed_size) { 975 archive_set_error(&a->archive, 976 ARCHIVE_ERRNO_FILE_FORMAT, 977 "Inconsistent uncompressed size: " 978 "%jd in central directory, %jd in local header", 979 (intmax_t)zip_entry_central_dir.uncompressed_size, 980 (intmax_t)zip_entry->uncompressed_size); 981 ret = ARCHIVE_WARN; 982 } 983 } 984 985 /* Populate some additional entry fields: */ 986 archive_entry_set_mode(entry, zip_entry->mode); 987 archive_entry_set_uid(entry, zip_entry->uid); 988 archive_entry_set_gid(entry, zip_entry->gid); 989 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 990 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 991 archive_entry_set_atime(entry, zip_entry->atime, 0); 992 993 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 994 size_t linkname_length; 995 996 if (zip_entry->compressed_size > 64 * 1024) { 997 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 998 "Zip file with oversized link entry"); 999 return ARCHIVE_FATAL; 1000 } 1001 1002 linkname_length = (size_t)zip_entry->compressed_size; 1003 1004 archive_entry_set_size(entry, 0); 1005 p = __archive_read_ahead(a, linkname_length, NULL); 1006 if (p == NULL) { 1007 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1008 "Truncated Zip file"); 1009 return ARCHIVE_FATAL; 1010 } 1011 1012 sconv = zip->sconv; 1013 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1014 sconv = zip->sconv_utf8; 1015 if (sconv == NULL) 1016 sconv = zip->sconv_default; 1017 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 1018 sconv) != 0) { 1019 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1020 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1021 archive_entry_copy_symlink_l(entry, p, 1022 linkname_length, NULL); 1023 if (errno == ENOMEM) { 1024 archive_set_error(&a->archive, ENOMEM, 1025 "Can't allocate memory for Symlink"); 1026 return (ARCHIVE_FATAL); 1027 } 1028 /* 1029 * Since there is no character-set regulation for 1030 * symlink name, do not report the conversion error 1031 * in an automatic conversion. 1032 */ 1033 if (sconv != zip->sconv_utf8 || 1034 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1035 archive_set_error(&a->archive, 1036 ARCHIVE_ERRNO_FILE_FORMAT, 1037 "Symlink cannot be converted " 1038 "from %s to current locale.", 1039 archive_string_conversion_charset_name( 1040 sconv)); 1041 ret = ARCHIVE_WARN; 1042 } 1043 } 1044 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1045 1046 if (__archive_read_consume(a, linkname_length) < 0) { 1047 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1048 "Read error skipping symlink target name"); 1049 return ARCHIVE_FATAL; 1050 } 1051 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1052 || zip_entry->uncompressed_size > 0) { 1053 /* Set the size only if it's meaningful. */ 1054 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1055 } 1056 zip->entry_bytes_remaining = zip_entry->compressed_size; 1057 1058 /* If there's no body, force read_data() to return EOF immediately. */ 1059 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1060 && zip->entry_bytes_remaining < 1) 1061 zip->end_of_entry = 1; 1062 1063 /* Set up a more descriptive format name. */ 1064 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1065 version / 10, version % 10, 1066 compression_name(zip->entry->compression)); 1067 a->archive.archive_format_name = zip->format_name.s; 1068 1069 return (ret); 1070} 1071 1072static int 1073check_authentication_code(struct archive_read *a, const void *_p) 1074{ 1075 struct zip *zip = (struct zip *)(a->format->data); 1076 1077 /* Check authentication code. */ 1078 if (zip->hctx_valid) { 1079 const void *p; 1080 uint8_t hmac[20]; 1081 size_t hmac_len = 20; 1082 int cmp; 1083 1084 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1085 if (_p == NULL) { 1086 /* Read authentication code. */ 1087 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1088 if (p == NULL) { 1089 archive_set_error(&a->archive, 1090 ARCHIVE_ERRNO_FILE_FORMAT, 1091 "Truncated ZIP file data"); 1092 return (ARCHIVE_FATAL); 1093 } 1094 } else { 1095 p = _p; 1096 } 1097 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1098 __archive_read_consume(a, AUTH_CODE_SIZE); 1099 if (cmp != 0) { 1100 archive_set_error(&a->archive, 1101 ARCHIVE_ERRNO_MISC, 1102 "ZIP bad Authentication code"); 1103 return (ARCHIVE_WARN); 1104 } 1105 } 1106 return (ARCHIVE_OK); 1107} 1108 1109/* 1110 * Read "uncompressed" data. There are three cases: 1111 * 1) We know the size of the data. This is always true for the 1112 * seeking reader (we've examined the Central Directory already). 1113 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1114 * Info-ZIP seems to do this; we know the size but have to grab 1115 * the CRC from the data descriptor afterwards. 1116 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1117 * we have no size information. In this case, we can do pretty 1118 * well by watching for the data descriptor record. The data 1119 * descriptor is 16 bytes and includes a computed CRC that should 1120 * provide a strong check. 1121 * 1122 * TODO: Technically, the PK\007\010 signature is optional. 1123 * In the original spec, the data descriptor contained CRC 1124 * and size fields but had no leading signature. In practice, 1125 * newer writers seem to provide the signature pretty consistently. 1126 * 1127 * For uncompressed data, the PK\007\010 marker seems essential 1128 * to be sure we've actually seen the end of the entry. 1129 * 1130 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1131 * zip->end_of_entry if it consumes all of the data. 1132 */ 1133static int 1134zip_read_data_none(struct archive_read *a, const void **_buff, 1135 size_t *size, int64_t *offset) 1136{ 1137 struct zip *zip; 1138 const char *buff; 1139 ssize_t bytes_avail; 1140 int r; 1141 1142 (void)offset; /* UNUSED */ 1143 1144 zip = (struct zip *)(a->format->data); 1145 1146 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1147 const char *p; 1148 ssize_t grabbing_bytes = 24; 1149 1150 if (zip->hctx_valid) 1151 grabbing_bytes += AUTH_CODE_SIZE; 1152 /* Grab at least 24 bytes. */ 1153 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1154 if (bytes_avail < grabbing_bytes) { 1155 /* Zip archives have end-of-archive markers 1156 that are longer than this, so a failure to get at 1157 least 24 bytes really does indicate a truncated 1158 file. */ 1159 archive_set_error(&a->archive, 1160 ARCHIVE_ERRNO_FILE_FORMAT, 1161 "Truncated ZIP file data"); 1162 return (ARCHIVE_FATAL); 1163 } 1164 /* Check for a complete PK\007\010 signature, followed 1165 * by the correct 4-byte CRC. */ 1166 p = buff; 1167 if (zip->hctx_valid) 1168 p += AUTH_CODE_SIZE; 1169 if (p[0] == 'P' && p[1] == 'K' 1170 && p[2] == '\007' && p[3] == '\010' 1171 && (archive_le32dec(p + 4) == zip->entry_crc32 1172 || zip->ignore_crc32 1173 || (zip->hctx_valid 1174 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1175 if (zip->entry->flags & LA_USED_ZIP64) { 1176 uint64_t compressed, uncompressed; 1177 zip->entry->crc32 = archive_le32dec(p + 4); 1178 compressed = archive_le64dec(p + 8); 1179 uncompressed = archive_le64dec(p + 16); 1180 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1181 archive_set_error(&a->archive, 1182 ARCHIVE_ERRNO_FILE_FORMAT, 1183 "Overflow of 64-bit file sizes"); 1184 return ARCHIVE_FAILED; 1185 } 1186 zip->entry->compressed_size = compressed; 1187 zip->entry->uncompressed_size = uncompressed; 1188 zip->unconsumed = 24; 1189 } else { 1190 zip->entry->crc32 = archive_le32dec(p + 4); 1191 zip->entry->compressed_size = 1192 archive_le32dec(p + 8); 1193 zip->entry->uncompressed_size = 1194 archive_le32dec(p + 12); 1195 zip->unconsumed = 16; 1196 } 1197 if (zip->hctx_valid) { 1198 r = check_authentication_code(a, buff); 1199 if (r != ARCHIVE_OK) 1200 return (r); 1201 } 1202 zip->end_of_entry = 1; 1203 return (ARCHIVE_OK); 1204 } 1205 /* If not at EOF, ensure we consume at least one byte. */ 1206 ++p; 1207 1208 /* Scan forward until we see where a PK\007\010 signature 1209 * might be. */ 1210 /* Return bytes up until that point. On the next call, 1211 * the code above will verify the data descriptor. */ 1212 while (p < buff + bytes_avail - 4) { 1213 if (p[3] == 'P') { p += 3; } 1214 else if (p[3] == 'K') { p += 2; } 1215 else if (p[3] == '\007') { p += 1; } 1216 else if (p[3] == '\010' && p[2] == '\007' 1217 && p[1] == 'K' && p[0] == 'P') { 1218 if (zip->hctx_valid) 1219 p -= AUTH_CODE_SIZE; 1220 break; 1221 } else { p += 4; } 1222 } 1223 bytes_avail = p - buff; 1224 } else { 1225 if (zip->entry_bytes_remaining == 0) { 1226 zip->end_of_entry = 1; 1227 if (zip->hctx_valid) { 1228 r = check_authentication_code(a, NULL); 1229 if (r != ARCHIVE_OK) 1230 return (r); 1231 } 1232 return (ARCHIVE_OK); 1233 } 1234 /* Grab a bunch of bytes. */ 1235 buff = __archive_read_ahead(a, 1, &bytes_avail); 1236 if (bytes_avail <= 0) { 1237 archive_set_error(&a->archive, 1238 ARCHIVE_ERRNO_FILE_FORMAT, 1239 "Truncated ZIP file data"); 1240 return (ARCHIVE_FATAL); 1241 } 1242 if (bytes_avail > zip->entry_bytes_remaining) 1243 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1244 } 1245 if (zip->tctx_valid || zip->cctx_valid) { 1246 size_t dec_size = bytes_avail; 1247 1248 if (dec_size > zip->decrypted_buffer_size) 1249 dec_size = zip->decrypted_buffer_size; 1250 if (zip->tctx_valid) { 1251 trad_enc_decrypt_update(&zip->tctx, 1252 (const uint8_t *)buff, dec_size, 1253 zip->decrypted_buffer, dec_size); 1254 } else { 1255 size_t dsize = dec_size; 1256 archive_hmac_sha1_update(&zip->hctx, 1257 (const uint8_t *)buff, dec_size); 1258 archive_decrypto_aes_ctr_update(&zip->cctx, 1259 (const uint8_t *)buff, dec_size, 1260 zip->decrypted_buffer, &dsize); 1261 } 1262 bytes_avail = dec_size; 1263 buff = (const char *)zip->decrypted_buffer; 1264 } 1265 *size = bytes_avail; 1266 zip->entry_bytes_remaining -= bytes_avail; 1267 zip->entry_uncompressed_bytes_read += bytes_avail; 1268 zip->entry_compressed_bytes_read += bytes_avail; 1269 zip->unconsumed += bytes_avail; 1270 *_buff = buff; 1271 return (ARCHIVE_OK); 1272} 1273 1274#ifdef HAVE_ZLIB_H 1275static int 1276zip_deflate_init(struct archive_read *a, struct zip *zip) 1277{ 1278 int r; 1279 1280 /* If we haven't yet read any data, initialize the decompressor. */ 1281 if (!zip->decompress_init) { 1282 if (zip->stream_valid) 1283 r = inflateReset(&zip->stream); 1284 else 1285 r = inflateInit2(&zip->stream, 1286 -15 /* Don't check for zlib header */); 1287 if (r != Z_OK) { 1288 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1289 "Can't initialize ZIP decompression."); 1290 return (ARCHIVE_FATAL); 1291 } 1292 /* Stream structure has been set up. */ 1293 zip->stream_valid = 1; 1294 /* We've initialized decompression for this stream. */ 1295 zip->decompress_init = 1; 1296 } 1297 return (ARCHIVE_OK); 1298} 1299 1300static int 1301zip_read_data_deflate(struct archive_read *a, const void **buff, 1302 size_t *size, int64_t *offset) 1303{ 1304 struct zip *zip; 1305 ssize_t bytes_avail; 1306 const void *compressed_buff, *sp; 1307 int r; 1308 1309 (void)offset; /* UNUSED */ 1310 1311 zip = (struct zip *)(a->format->data); 1312 1313 /* If the buffer hasn't been allocated, allocate it now. */ 1314 if (zip->uncompressed_buffer == NULL) { 1315 zip->uncompressed_buffer_size = 256 * 1024; 1316 zip->uncompressed_buffer 1317 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1318 if (zip->uncompressed_buffer == NULL) { 1319 archive_set_error(&a->archive, ENOMEM, 1320 "No memory for ZIP decompression"); 1321 return (ARCHIVE_FATAL); 1322 } 1323 } 1324 1325 r = zip_deflate_init(a, zip); 1326 if (r != ARCHIVE_OK) 1327 return (r); 1328 1329 /* 1330 * Note: '1' here is a performance optimization. 1331 * Recall that the decompression layer returns a count of 1332 * available bytes; asking for more than that forces the 1333 * decompressor to combine reads by copying data. 1334 */ 1335 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1336 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1337 && bytes_avail > zip->entry_bytes_remaining) { 1338 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1339 } 1340 if (bytes_avail < 0) { 1341 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1342 "Truncated ZIP file body"); 1343 return (ARCHIVE_FATAL); 1344 } 1345 1346 if (zip->tctx_valid || zip->cctx_valid) { 1347 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1348 size_t buff_remaining = 1349 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1350 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1351 1352 if (buff_remaining > (size_t)bytes_avail) 1353 buff_remaining = (size_t)bytes_avail; 1354 1355 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1356 zip->entry_bytes_remaining > 0) { 1357 if ((int64_t)(zip->decrypted_bytes_remaining 1358 + buff_remaining) 1359 > zip->entry_bytes_remaining) { 1360 if (zip->entry_bytes_remaining < 1361 (int64_t)zip->decrypted_bytes_remaining) 1362 buff_remaining = 0; 1363 else 1364 buff_remaining = 1365 (size_t)zip->entry_bytes_remaining 1366 - zip->decrypted_bytes_remaining; 1367 } 1368 } 1369 if (buff_remaining > 0) { 1370 if (zip->tctx_valid) { 1371 trad_enc_decrypt_update(&zip->tctx, 1372 compressed_buff, buff_remaining, 1373 zip->decrypted_ptr 1374 + zip->decrypted_bytes_remaining, 1375 buff_remaining); 1376 } else { 1377 size_t dsize = buff_remaining; 1378 archive_decrypto_aes_ctr_update( 1379 &zip->cctx, 1380 compressed_buff, buff_remaining, 1381 zip->decrypted_ptr 1382 + zip->decrypted_bytes_remaining, 1383 &dsize); 1384 } 1385 zip->decrypted_bytes_remaining += buff_remaining; 1386 } 1387 } 1388 bytes_avail = zip->decrypted_bytes_remaining; 1389 compressed_buff = (const char *)zip->decrypted_ptr; 1390 } 1391 1392 /* 1393 * A bug in zlib.h: stream.next_in should be marked 'const' 1394 * but isn't (the library never alters data through the 1395 * next_in pointer, only reads it). The result: this ugly 1396 * cast to remove 'const'. 1397 */ 1398 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1399 zip->stream.avail_in = (uInt)bytes_avail; 1400 zip->stream.total_in = 0; 1401 zip->stream.next_out = zip->uncompressed_buffer; 1402 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1403 zip->stream.total_out = 0; 1404 1405 r = inflate(&zip->stream, 0); 1406 switch (r) { 1407 case Z_OK: 1408 break; 1409 case Z_STREAM_END: 1410 zip->end_of_entry = 1; 1411 break; 1412 case Z_MEM_ERROR: 1413 archive_set_error(&a->archive, ENOMEM, 1414 "Out of memory for ZIP decompression"); 1415 return (ARCHIVE_FATAL); 1416 default: 1417 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1418 "ZIP decompression failed (%d)", r); 1419 return (ARCHIVE_FATAL); 1420 } 1421 1422 /* Consume as much as the compressor actually used. */ 1423 bytes_avail = zip->stream.total_in; 1424 if (zip->tctx_valid || zip->cctx_valid) { 1425 zip->decrypted_bytes_remaining -= bytes_avail; 1426 if (zip->decrypted_bytes_remaining == 0) 1427 zip->decrypted_ptr = zip->decrypted_buffer; 1428 else 1429 zip->decrypted_ptr += bytes_avail; 1430 } 1431 /* Calculate compressed data as much as we used.*/ 1432 if (zip->hctx_valid) 1433 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1434 __archive_read_consume(a, bytes_avail); 1435 zip->entry_bytes_remaining -= bytes_avail; 1436 zip->entry_compressed_bytes_read += bytes_avail; 1437 1438 *size = zip->stream.total_out; 1439 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1440 *buff = zip->uncompressed_buffer; 1441 1442 if (zip->end_of_entry && zip->hctx_valid) { 1443 r = check_authentication_code(a, NULL); 1444 if (r != ARCHIVE_OK) 1445 return (r); 1446 } 1447 1448 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1449 const char *p; 1450 1451 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1452 archive_set_error(&a->archive, 1453 ARCHIVE_ERRNO_FILE_FORMAT, 1454 "Truncated ZIP end-of-file record"); 1455 return (ARCHIVE_FATAL); 1456 } 1457 /* Consume the optional PK\007\010 marker. */ 1458 if (p[0] == 'P' && p[1] == 'K' && 1459 p[2] == '\007' && p[3] == '\010') { 1460 p += 4; 1461 zip->unconsumed = 4; 1462 } 1463 if (zip->entry->flags & LA_USED_ZIP64) { 1464 uint64_t compressed, uncompressed; 1465 zip->entry->crc32 = archive_le32dec(p); 1466 compressed = archive_le64dec(p + 4); 1467 uncompressed = archive_le64dec(p + 12); 1468 if (compressed > INT64_MAX || uncompressed > INT64_MAX) { 1469 archive_set_error(&a->archive, 1470 ARCHIVE_ERRNO_FILE_FORMAT, 1471 "Overflow of 64-bit file sizes"); 1472 return ARCHIVE_FAILED; 1473 } 1474 zip->entry->compressed_size = compressed; 1475 zip->entry->uncompressed_size = uncompressed; 1476 zip->unconsumed += 20; 1477 } else { 1478 zip->entry->crc32 = archive_le32dec(p); 1479 zip->entry->compressed_size = archive_le32dec(p + 4); 1480 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1481 zip->unconsumed += 12; 1482 } 1483 } 1484 1485 return (ARCHIVE_OK); 1486} 1487#endif 1488 1489static int 1490read_decryption_header(struct archive_read *a) 1491{ 1492 struct zip *zip = (struct zip *)(a->format->data); 1493 const char *p; 1494 unsigned int remaining_size; 1495 unsigned int ts; 1496 1497 /* 1498 * Read an initialization vector data field. 1499 */ 1500 p = __archive_read_ahead(a, 2, NULL); 1501 if (p == NULL) 1502 goto truncated; 1503 ts = zip->iv_size; 1504 zip->iv_size = archive_le16dec(p); 1505 __archive_read_consume(a, 2); 1506 if (ts < zip->iv_size) { 1507 free(zip->iv); 1508 zip->iv = NULL; 1509 } 1510 p = __archive_read_ahead(a, zip->iv_size, NULL); 1511 if (p == NULL) 1512 goto truncated; 1513 if (zip->iv == NULL) { 1514 zip->iv = malloc(zip->iv_size); 1515 if (zip->iv == NULL) 1516 goto nomem; 1517 } 1518 memcpy(zip->iv, p, zip->iv_size); 1519 __archive_read_consume(a, zip->iv_size); 1520 1521 /* 1522 * Read a size of remaining decryption header field. 1523 */ 1524 p = __archive_read_ahead(a, 14, NULL); 1525 if (p == NULL) 1526 goto truncated; 1527 remaining_size = archive_le32dec(p); 1528 if (remaining_size < 16 || remaining_size > (1 << 18)) 1529 goto corrupted; 1530 1531 /* Check if format version is supported. */ 1532 if (archive_le16dec(p+4) != 3) { 1533 archive_set_error(&a->archive, 1534 ARCHIVE_ERRNO_FILE_FORMAT, 1535 "Unsupported encryption format version: %u", 1536 archive_le16dec(p+4)); 1537 return (ARCHIVE_FAILED); 1538 } 1539 1540 /* 1541 * Read an encryption algorithm field. 1542 */ 1543 zip->alg_id = archive_le16dec(p+6); 1544 switch (zip->alg_id) { 1545 case 0x6601:/* DES */ 1546 case 0x6602:/* RC2 */ 1547 case 0x6603:/* 3DES 168 */ 1548 case 0x6609:/* 3DES 112 */ 1549 case 0x660E:/* AES 128 */ 1550 case 0x660F:/* AES 192 */ 1551 case 0x6610:/* AES 256 */ 1552 case 0x6702:/* RC2 (version >= 5.2) */ 1553 case 0x6720:/* Blowfish */ 1554 case 0x6721:/* Twofish */ 1555 case 0x6801:/* RC4 */ 1556 /* Supported encryption algorithm. */ 1557 break; 1558 default: 1559 archive_set_error(&a->archive, 1560 ARCHIVE_ERRNO_FILE_FORMAT, 1561 "Unknown encryption algorithm: %u", zip->alg_id); 1562 return (ARCHIVE_FAILED); 1563 } 1564 1565 /* 1566 * Read a bit length field. 1567 */ 1568 zip->bit_len = archive_le16dec(p+8); 1569 1570 /* 1571 * Read a flags field. 1572 */ 1573 zip->flags = archive_le16dec(p+10); 1574 switch (zip->flags & 0xf000) { 1575 case 0x0001: /* Password is required to decrypt. */ 1576 case 0x0002: /* Certificates only. */ 1577 case 0x0003: /* Password or certificate required to decrypt. */ 1578 break; 1579 default: 1580 archive_set_error(&a->archive, 1581 ARCHIVE_ERRNO_FILE_FORMAT, 1582 "Unknown encryption flag: %u", zip->flags); 1583 return (ARCHIVE_FAILED); 1584 } 1585 if ((zip->flags & 0xf000) == 0 || 1586 (zip->flags & 0xf000) == 0x4000) { 1587 archive_set_error(&a->archive, 1588 ARCHIVE_ERRNO_FILE_FORMAT, 1589 "Unknown encryption flag: %u", zip->flags); 1590 return (ARCHIVE_FAILED); 1591 } 1592 1593 /* 1594 * Read an encrypted random data field. 1595 */ 1596 ts = zip->erd_size; 1597 zip->erd_size = archive_le16dec(p+12); 1598 __archive_read_consume(a, 14); 1599 if ((zip->erd_size & 0xf) != 0 || 1600 (zip->erd_size + 16) > remaining_size || 1601 (zip->erd_size + 16) < zip->erd_size) 1602 goto corrupted; 1603 1604 if (ts < zip->erd_size) { 1605 free(zip->erd); 1606 zip->erd = NULL; 1607 } 1608 p = __archive_read_ahead(a, zip->erd_size, NULL); 1609 if (p == NULL) 1610 goto truncated; 1611 if (zip->erd == NULL) { 1612 zip->erd = malloc(zip->erd_size); 1613 if (zip->erd == NULL) 1614 goto nomem; 1615 } 1616 memcpy(zip->erd, p, zip->erd_size); 1617 __archive_read_consume(a, zip->erd_size); 1618 1619 /* 1620 * Read a reserved data field. 1621 */ 1622 p = __archive_read_ahead(a, 4, NULL); 1623 if (p == NULL) 1624 goto truncated; 1625 /* Reserved data size should be zero. */ 1626 if (archive_le32dec(p) != 0) 1627 goto corrupted; 1628 __archive_read_consume(a, 4); 1629 1630 /* 1631 * Read a password validation data field. 1632 */ 1633 p = __archive_read_ahead(a, 2, NULL); 1634 if (p == NULL) 1635 goto truncated; 1636 ts = zip->v_size; 1637 zip->v_size = archive_le16dec(p); 1638 __archive_read_consume(a, 2); 1639 if ((zip->v_size & 0x0f) != 0 || 1640 (zip->erd_size + zip->v_size + 16) > remaining_size || 1641 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1642 goto corrupted; 1643 if (ts < zip->v_size) { 1644 free(zip->v_data); 1645 zip->v_data = NULL; 1646 } 1647 p = __archive_read_ahead(a, zip->v_size, NULL); 1648 if (p == NULL) 1649 goto truncated; 1650 if (zip->v_data == NULL) { 1651 zip->v_data = malloc(zip->v_size); 1652 if (zip->v_data == NULL) 1653 goto nomem; 1654 } 1655 memcpy(zip->v_data, p, zip->v_size); 1656 __archive_read_consume(a, zip->v_size); 1657 1658 p = __archive_read_ahead(a, 4, NULL); 1659 if (p == NULL) 1660 goto truncated; 1661 zip->v_crc32 = archive_le32dec(p); 1662 __archive_read_consume(a, 4); 1663 1664 /*return (ARCHIVE_OK); 1665 * This is not fully implemented yet.*/ 1666 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1667 "Encrypted file is unsupported"); 1668 return (ARCHIVE_FAILED); 1669truncated: 1670 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1671 "Truncated ZIP file data"); 1672 return (ARCHIVE_FATAL); 1673corrupted: 1674 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1675 "Corrupted ZIP file data"); 1676 return (ARCHIVE_FATAL); 1677nomem: 1678 archive_set_error(&a->archive, ENOMEM, 1679 "No memory for ZIP decryption"); 1680 return (ARCHIVE_FATAL); 1681} 1682 1683static int 1684zip_alloc_decryption_buffer(struct archive_read *a) 1685{ 1686 struct zip *zip = (struct zip *)(a->format->data); 1687 size_t bs = 256 * 1024; 1688 1689 if (zip->decrypted_buffer == NULL) { 1690 zip->decrypted_buffer_size = bs; 1691 zip->decrypted_buffer = malloc(bs); 1692 if (zip->decrypted_buffer == NULL) { 1693 archive_set_error(&a->archive, ENOMEM, 1694 "No memory for ZIP decryption"); 1695 return (ARCHIVE_FATAL); 1696 } 1697 } 1698 zip->decrypted_ptr = zip->decrypted_buffer; 1699 return (ARCHIVE_OK); 1700} 1701 1702static int 1703init_traditional_PKWARE_decryption(struct archive_read *a) 1704{ 1705 struct zip *zip = (struct zip *)(a->format->data); 1706 const void *p; 1707 int retry; 1708 int r; 1709 1710 if (zip->tctx_valid) 1711 return (ARCHIVE_OK); 1712 1713 /* 1714 Read the 12 bytes encryption header stored at 1715 the start of the data area. 1716 */ 1717#define ENC_HEADER_SIZE 12 1718 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1719 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1720 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1721 "Truncated Zip encrypted body: only %jd bytes available", 1722 (intmax_t)zip->entry_bytes_remaining); 1723 return (ARCHIVE_FATAL); 1724 } 1725 1726 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1727 if (p == NULL) { 1728 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1729 "Truncated ZIP file data"); 1730 return (ARCHIVE_FATAL); 1731 } 1732 1733 for (retry = 0;; retry++) { 1734 const char *passphrase; 1735 uint8_t crcchk; 1736 1737 passphrase = __archive_read_next_passphrase(a); 1738 if (passphrase == NULL) { 1739 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1740 (retry > 0)? 1741 "Incorrect passphrase": 1742 "Passphrase required for this entry"); 1743 return (ARCHIVE_FAILED); 1744 } 1745 1746 /* 1747 * Initialize ctx for Traditional PKWARE Decryption. 1748 */ 1749 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1750 p, ENC_HEADER_SIZE, &crcchk); 1751 if (r == 0 && crcchk == zip->entry->decdat) 1752 break;/* The passphrase is OK. */ 1753 if (retry > 10000) { 1754 /* Avoid infinity loop. */ 1755 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1756 "Too many incorrect passphrases"); 1757 return (ARCHIVE_FAILED); 1758 } 1759 } 1760 1761 __archive_read_consume(a, ENC_HEADER_SIZE); 1762 zip->tctx_valid = 1; 1763 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1764 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1765 } 1766 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1767 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1768 zip->decrypted_bytes_remaining = 0; 1769 1770 return (zip_alloc_decryption_buffer(a)); 1771#undef ENC_HEADER_SIZE 1772} 1773 1774static int 1775init_WinZip_AES_decryption(struct archive_read *a) 1776{ 1777 struct zip *zip = (struct zip *)(a->format->data); 1778 const void *p; 1779 const uint8_t *pv; 1780 size_t key_len, salt_len; 1781 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1782 int retry; 1783 int r; 1784 1785 if (zip->cctx_valid || zip->hctx_valid) 1786 return (ARCHIVE_OK); 1787 1788 switch (zip->entry->aes_extra.strength) { 1789 case 1: salt_len = 8; key_len = 16; break; 1790 case 2: salt_len = 12; key_len = 24; break; 1791 case 3: salt_len = 16; key_len = 32; break; 1792 default: goto corrupted; 1793 } 1794 p = __archive_read_ahead(a, salt_len + 2, NULL); 1795 if (p == NULL) 1796 goto truncated; 1797 1798 for (retry = 0;; retry++) { 1799 const char *passphrase; 1800 1801 passphrase = __archive_read_next_passphrase(a); 1802 if (passphrase == NULL) { 1803 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1804 (retry > 0)? 1805 "Incorrect passphrase": 1806 "Passphrase required for this entry"); 1807 return (ARCHIVE_FAILED); 1808 } 1809 memset(derived_key, 0, sizeof(derived_key)); 1810 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1811 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1812 if (r != 0) { 1813 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1814 "Decryption is unsupported due to lack of " 1815 "crypto library"); 1816 return (ARCHIVE_FAILED); 1817 } 1818 1819 /* Check password verification value. */ 1820 pv = ((const uint8_t *)p) + salt_len; 1821 if (derived_key[key_len * 2] == pv[0] && 1822 derived_key[key_len * 2 + 1] == pv[1]) 1823 break;/* The passphrase is OK. */ 1824 if (retry > 10000) { 1825 /* Avoid infinity loop. */ 1826 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1827 "Too many incorrect passphrases"); 1828 return (ARCHIVE_FAILED); 1829 } 1830 } 1831 1832 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1833 if (r != 0) { 1834 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1835 "Decryption is unsupported due to lack of crypto library"); 1836 return (ARCHIVE_FAILED); 1837 } 1838 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1839 if (r != 0) { 1840 archive_decrypto_aes_ctr_release(&zip->cctx); 1841 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1842 "Failed to initialize HMAC-SHA1"); 1843 return (ARCHIVE_FAILED); 1844 } 1845 zip->cctx_valid = zip->hctx_valid = 1; 1846 __archive_read_consume(a, salt_len + 2); 1847 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1848 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1849 && zip->entry_bytes_remaining < 0) 1850 goto corrupted; 1851 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1852 zip->decrypted_bytes_remaining = 0; 1853 1854 zip->entry->compression = zip->entry->aes_extra.compression; 1855 return (zip_alloc_decryption_buffer(a)); 1856 1857truncated: 1858 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1859 "Truncated ZIP file data"); 1860 return (ARCHIVE_FATAL); 1861corrupted: 1862 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1863 "Corrupted ZIP file data"); 1864 return (ARCHIVE_FATAL); 1865} 1866 1867static int 1868archive_read_format_zip_read_data(struct archive_read *a, 1869 const void **buff, size_t *size, int64_t *offset) 1870{ 1871 int r; 1872 struct zip *zip = (struct zip *)(a->format->data); 1873 1874 if (zip->has_encrypted_entries == 1875 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1876 zip->has_encrypted_entries = 0; 1877 } 1878 1879 *offset = zip->entry_uncompressed_bytes_read; 1880 *size = 0; 1881 *buff = NULL; 1882 1883 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1884 if (zip->end_of_entry) 1885 return (ARCHIVE_EOF); 1886 1887 /* Return EOF immediately if this is a non-regular file. */ 1888 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1889 return (ARCHIVE_EOF); 1890 1891 __archive_read_consume(a, zip->unconsumed); 1892 zip->unconsumed = 0; 1893 1894 if (zip->init_decryption) { 1895 zip->has_encrypted_entries = 1; 1896 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1897 r = read_decryption_header(a); 1898 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1899 r = init_WinZip_AES_decryption(a); 1900 else 1901 r = init_traditional_PKWARE_decryption(a); 1902 if (r != ARCHIVE_OK) 1903 return (r); 1904 zip->init_decryption = 0; 1905 } 1906 1907 switch(zip->entry->compression) { 1908 case 0: /* No compression. */ 1909 r = zip_read_data_none(a, buff, size, offset); 1910 break; 1911#ifdef HAVE_ZLIB_H 1912 case 8: /* Deflate compression. */ 1913 r = zip_read_data_deflate(a, buff, size, offset); 1914 break; 1915#endif 1916 default: /* Unsupported compression. */ 1917 /* Return a warning. */ 1918 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1919 "Unsupported ZIP compression method (%s)", 1920 compression_name(zip->entry->compression)); 1921 /* We can't decompress this entry, but we will 1922 * be able to skip() it and try the next entry. */ 1923 return (ARCHIVE_FAILED); 1924 break; 1925 } 1926 if (r != ARCHIVE_OK) 1927 return (r); 1928 /* Update checksum */ 1929 if (*size) 1930 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1931 (unsigned)*size); 1932 /* If we hit the end, swallow any end-of-data marker. */ 1933 if (zip->end_of_entry) { 1934 /* Check file size, CRC against these values. */ 1935 if (zip->entry->compressed_size != 1936 zip->entry_compressed_bytes_read) { 1937 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1938 "ZIP compressed data is wrong size " 1939 "(read %jd, expected %jd)", 1940 (intmax_t)zip->entry_compressed_bytes_read, 1941 (intmax_t)zip->entry->compressed_size); 1942 return (ARCHIVE_WARN); 1943 } 1944 /* Size field only stores the lower 32 bits of the actual 1945 * size. */ 1946 if ((zip->entry->uncompressed_size & UINT32_MAX) 1947 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1948 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1949 "ZIP uncompressed data is wrong size " 1950 "(read %jd, expected %jd)\n", 1951 (intmax_t)zip->entry_uncompressed_bytes_read, 1952 (intmax_t)zip->entry->uncompressed_size); 1953 return (ARCHIVE_WARN); 1954 } 1955 /* Check computed CRC against header */ 1956 if ((!zip->hctx_valid || 1957 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1958 zip->entry->crc32 != zip->entry_crc32 1959 && !zip->ignore_crc32) { 1960 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1961 "ZIP bad CRC: 0x%lx should be 0x%lx", 1962 (unsigned long)zip->entry_crc32, 1963 (unsigned long)zip->entry->crc32); 1964 return (ARCHIVE_WARN); 1965 } 1966 } 1967 1968 return (ARCHIVE_OK); 1969} 1970 1971static int 1972archive_read_format_zip_cleanup(struct archive_read *a) 1973{ 1974 struct zip *zip; 1975 struct zip_entry *zip_entry, *next_zip_entry; 1976 1977 zip = (struct zip *)(a->format->data); 1978#ifdef HAVE_ZLIB_H 1979 if (zip->stream_valid) 1980 inflateEnd(&zip->stream); 1981 free(zip->uncompressed_buffer); 1982#endif 1983 if (zip->zip_entries) { 1984 zip_entry = zip->zip_entries; 1985 while (zip_entry != NULL) { 1986 next_zip_entry = zip_entry->next; 1987 archive_string_free(&zip_entry->rsrcname); 1988 free(zip_entry); 1989 zip_entry = next_zip_entry; 1990 } 1991 } 1992 free(zip->decrypted_buffer); 1993 if (zip->cctx_valid) 1994 archive_decrypto_aes_ctr_release(&zip->cctx); 1995 if (zip->hctx_valid) 1996 archive_hmac_sha1_cleanup(&zip->hctx); 1997 free(zip->iv); 1998 free(zip->erd); 1999 free(zip->v_data); 2000 archive_string_free(&zip->format_name); 2001 free(zip); 2002 (a->format->data) = NULL; 2003 return (ARCHIVE_OK); 2004} 2005 2006static int 2007archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 2008{ 2009 if (_a && _a->format) { 2010 struct zip * zip = (struct zip *)_a->format->data; 2011 if (zip) { 2012 return zip->has_encrypted_entries; 2013 } 2014 } 2015 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2016} 2017 2018static int 2019archive_read_format_zip_options(struct archive_read *a, 2020 const char *key, const char *val) 2021{ 2022 struct zip *zip; 2023 int ret = ARCHIVE_FAILED; 2024 2025 zip = (struct zip *)(a->format->data); 2026 if (strcmp(key, "compat-2x") == 0) { 2027 /* Handle filenames as libarchive 2.x */ 2028 zip->init_default_conversion = (val != NULL) ? 1 : 0; 2029 return (ARCHIVE_OK); 2030 } else if (strcmp(key, "hdrcharset") == 0) { 2031 if (val == NULL || val[0] == 0) 2032 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2033 "zip: hdrcharset option needs a character-set name" 2034 ); 2035 else { 2036 zip->sconv = archive_string_conversion_from_charset( 2037 &a->archive, val, 0); 2038 if (zip->sconv != NULL) { 2039 if (strcmp(val, "UTF-8") == 0) 2040 zip->sconv_utf8 = zip->sconv; 2041 ret = ARCHIVE_OK; 2042 } else 2043 ret = ARCHIVE_FATAL; 2044 } 2045 return (ret); 2046 } else if (strcmp(key, "ignorecrc32") == 0) { 2047 /* Mostly useful for testing. */ 2048 if (val == NULL || val[0] == 0) { 2049 zip->crc32func = real_crc32; 2050 zip->ignore_crc32 = 0; 2051 } else { 2052 zip->crc32func = fake_crc32; 2053 zip->ignore_crc32 = 1; 2054 } 2055 return (ARCHIVE_OK); 2056 } else if (strcmp(key, "mac-ext") == 0) { 2057 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2058 return (ARCHIVE_OK); 2059 } 2060 2061 /* Note: The "warn" return is just to inform the options 2062 * supervisor that we didn't handle it. It will generate 2063 * a suitable error if no one used this option. */ 2064 return (ARCHIVE_WARN); 2065} 2066 2067int 2068archive_read_support_format_zip(struct archive *a) 2069{ 2070 int r; 2071 r = archive_read_support_format_zip_streamable(a); 2072 if (r != ARCHIVE_OK) 2073 return r; 2074 return (archive_read_support_format_zip_seekable(a)); 2075} 2076 2077/* ------------------------------------------------------------------------ */ 2078 2079/* 2080 * Streaming-mode support 2081 */ 2082 2083 2084static int 2085archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2086{ 2087 (void)a; /* UNUSED */ 2088 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2089 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2090} 2091 2092static int 2093archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2094{ 2095 const char *p; 2096 2097 (void)best_bid; /* UNUSED */ 2098 2099 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2100 return (-1); 2101 2102 /* 2103 * Bid of 29 here comes from: 2104 * + 16 bits for "PK", 2105 * + next 16-bit field has 6 options so contributes 2106 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2107 * 2108 * So we've effectively verified ~29 total bits of check data. 2109 */ 2110 if (p[0] == 'P' && p[1] == 'K') { 2111 if ((p[2] == '\001' && p[3] == '\002') 2112 || (p[2] == '\003' && p[3] == '\004') 2113 || (p[2] == '\005' && p[3] == '\006') 2114 || (p[2] == '\006' && p[3] == '\006') 2115 || (p[2] == '\007' && p[3] == '\010') 2116 || (p[2] == '0' && p[3] == '0')) 2117 return (29); 2118 } 2119 2120 /* TODO: It's worth looking ahead a little bit for a valid 2121 * PK signature. In particular, that would make it possible 2122 * to read some UUEncoded SFX files or SFX files coming from 2123 * a network socket. */ 2124 2125 return (0); 2126} 2127 2128static int 2129archive_read_format_zip_streamable_read_header(struct archive_read *a, 2130 struct archive_entry *entry) 2131{ 2132 struct zip *zip; 2133 2134 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2135 if (a->archive.archive_format_name == NULL) 2136 a->archive.archive_format_name = "ZIP"; 2137 2138 zip = (struct zip *)(a->format->data); 2139 2140 /* 2141 * It should be sufficient to call archive_read_next_header() for 2142 * a reader to determine if an entry is encrypted or not. If the 2143 * encryption of an entry is only detectable when calling 2144 * archive_read_data(), so be it. We'll do the same check there 2145 * as well. 2146 */ 2147 if (zip->has_encrypted_entries == 2148 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2149 zip->has_encrypted_entries = 0; 2150 2151 /* Make sure we have a zip_entry structure to use. */ 2152 if (zip->zip_entries == NULL) { 2153 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2154 if (zip->zip_entries == NULL) { 2155 archive_set_error(&a->archive, ENOMEM, 2156 "Out of memory"); 2157 return ARCHIVE_FATAL; 2158 } 2159 } 2160 zip->entry = zip->zip_entries; 2161 memset(zip->entry, 0, sizeof(struct zip_entry)); 2162 2163 if (zip->cctx_valid) 2164 archive_decrypto_aes_ctr_release(&zip->cctx); 2165 if (zip->hctx_valid) 2166 archive_hmac_sha1_cleanup(&zip->hctx); 2167 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2168 __archive_read_reset_passphrase(a); 2169 2170 /* Search ahead for the next local file header. */ 2171 __archive_read_consume(a, zip->unconsumed); 2172 zip->unconsumed = 0; 2173 for (;;) { 2174 int64_t skipped = 0; 2175 const char *p, *end; 2176 ssize_t bytes; 2177 2178 p = __archive_read_ahead(a, 4, &bytes); 2179 if (p == NULL) 2180 return (ARCHIVE_FATAL); 2181 end = p + bytes; 2182 2183 while (p + 4 <= end) { 2184 if (p[0] == 'P' && p[1] == 'K') { 2185 if (p[2] == '\003' && p[3] == '\004') { 2186 /* Regular file entry. */ 2187 __archive_read_consume(a, skipped); 2188 return zip_read_local_file_header(a, 2189 entry, zip); 2190 } 2191 2192 /* 2193 * TODO: We cannot restore permissions 2194 * based only on the local file headers. 2195 * Consider scanning the central 2196 * directory and returning additional 2197 * entries for at least directories. 2198 * This would allow us to properly set 2199 * directory permissions. 2200 * 2201 * This won't help us fix symlinks 2202 * and may not help with regular file 2203 * permissions, either. <sigh> 2204 */ 2205 if (p[2] == '\001' && p[3] == '\002') { 2206 return (ARCHIVE_EOF); 2207 } 2208 2209 /* End of central directory? Must be an 2210 * empty archive. */ 2211 if ((p[2] == '\005' && p[3] == '\006') 2212 || (p[2] == '\006' && p[3] == '\006')) 2213 return (ARCHIVE_EOF); 2214 } 2215 ++p; 2216 ++skipped; 2217 } 2218 __archive_read_consume(a, skipped); 2219 } 2220} 2221 2222static int 2223archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2224{ 2225 struct zip *zip; 2226 int64_t bytes_skipped; 2227 2228 zip = (struct zip *)(a->format->data); 2229 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2230 zip->unconsumed = 0; 2231 if (bytes_skipped < 0) 2232 return (ARCHIVE_FATAL); 2233 2234 /* If we've already read to end of data, we're done. */ 2235 if (zip->end_of_entry) 2236 return (ARCHIVE_OK); 2237 2238 /* So we know we're streaming... */ 2239 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2240 || zip->entry->compressed_size > 0) { 2241 /* We know the compressed length, so we can just skip. */ 2242 bytes_skipped = __archive_read_consume(a, 2243 zip->entry_bytes_remaining); 2244 if (bytes_skipped < 0) 2245 return (ARCHIVE_FATAL); 2246 return (ARCHIVE_OK); 2247 } 2248 2249 if (zip->init_decryption) { 2250 int r; 2251 2252 zip->has_encrypted_entries = 1; 2253 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2254 r = read_decryption_header(a); 2255 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2256 r = init_WinZip_AES_decryption(a); 2257 else 2258 r = init_traditional_PKWARE_decryption(a); 2259 if (r != ARCHIVE_OK) 2260 return (r); 2261 zip->init_decryption = 0; 2262 } 2263 2264 /* We're streaming and we don't know the length. */ 2265 /* If the body is compressed and we know the format, we can 2266 * find an exact end-of-entry by decompressing it. */ 2267 switch (zip->entry->compression) { 2268#ifdef HAVE_ZLIB_H 2269 case 8: /* Deflate compression. */ 2270 while (!zip->end_of_entry) { 2271 int64_t offset = 0; 2272 const void *buff = NULL; 2273 size_t size = 0; 2274 int r; 2275 r = zip_read_data_deflate(a, &buff, &size, &offset); 2276 if (r != ARCHIVE_OK) 2277 return (r); 2278 } 2279 return ARCHIVE_OK; 2280#endif 2281 default: /* Uncompressed or unknown. */ 2282 /* Scan for a PK\007\010 signature. */ 2283 for (;;) { 2284 const char *p, *buff; 2285 ssize_t bytes_avail; 2286 buff = __archive_read_ahead(a, 16, &bytes_avail); 2287 if (bytes_avail < 16) { 2288 archive_set_error(&a->archive, 2289 ARCHIVE_ERRNO_FILE_FORMAT, 2290 "Truncated ZIP file data"); 2291 return (ARCHIVE_FATAL); 2292 } 2293 p = buff; 2294 while (p <= buff + bytes_avail - 16) { 2295 if (p[3] == 'P') { p += 3; } 2296 else if (p[3] == 'K') { p += 2; } 2297 else if (p[3] == '\007') { p += 1; } 2298 else if (p[3] == '\010' && p[2] == '\007' 2299 && p[1] == 'K' && p[0] == 'P') { 2300 if (zip->entry->flags & LA_USED_ZIP64) 2301 __archive_read_consume(a, 2302 p - buff + 24); 2303 else 2304 __archive_read_consume(a, 2305 p - buff + 16); 2306 return ARCHIVE_OK; 2307 } else { p += 4; } 2308 } 2309 __archive_read_consume(a, p - buff); 2310 } 2311 } 2312} 2313 2314int 2315archive_read_support_format_zip_streamable(struct archive *_a) 2316{ 2317 struct archive_read *a = (struct archive_read *)_a; 2318 struct zip *zip; 2319 int r; 2320 2321 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2322 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2323 2324 zip = (struct zip *)calloc(1, sizeof(*zip)); 2325 if (zip == NULL) { 2326 archive_set_error(&a->archive, ENOMEM, 2327 "Can't allocate zip data"); 2328 return (ARCHIVE_FATAL); 2329 } 2330 2331 /* Streamable reader doesn't support mac extensions. */ 2332 zip->process_mac_extensions = 0; 2333 2334 /* 2335 * Until enough data has been read, we cannot tell about 2336 * any encrypted entries yet. 2337 */ 2338 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2339 zip->crc32func = real_crc32; 2340 2341 r = __archive_read_register_format(a, 2342 zip, 2343 "zip", 2344 archive_read_format_zip_streamable_bid, 2345 archive_read_format_zip_options, 2346 archive_read_format_zip_streamable_read_header, 2347 archive_read_format_zip_read_data, 2348 archive_read_format_zip_read_data_skip_streamable, 2349 NULL, 2350 archive_read_format_zip_cleanup, 2351 archive_read_support_format_zip_capabilities_streamable, 2352 archive_read_format_zip_has_encrypted_entries); 2353 2354 if (r != ARCHIVE_OK) 2355 free(zip); 2356 return (ARCHIVE_OK); 2357} 2358 2359/* ------------------------------------------------------------------------ */ 2360 2361/* 2362 * Seeking-mode support 2363 */ 2364 2365static int 2366archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2367{ 2368 (void)a; /* UNUSED */ 2369 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2370 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2371} 2372 2373/* 2374 * TODO: This is a performance sink because it forces the read core to 2375 * drop buffered data from the start of file, which will then have to 2376 * be re-read again if this bidder loses. 2377 * 2378 * We workaround this a little by passing in the best bid so far so 2379 * that later bidders can do nothing if they know they'll never 2380 * outbid. But we can certainly do better... 2381 */ 2382static int 2383read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2384{ 2385 /* Sanity-check the EOCD we've found. */ 2386 2387 /* This must be the first volume. */ 2388 if (archive_le16dec(p + 4) != 0) 2389 return 0; 2390 /* Central directory must be on this volume. */ 2391 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2392 return 0; 2393 /* All central directory entries must be on this volume. */ 2394 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2395 return 0; 2396 /* Central directory can't extend beyond start of EOCD record. */ 2397 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2398 > current_offset) 2399 return 0; 2400 2401 /* Save the central directory location for later use. */ 2402 zip->central_directory_offset = archive_le32dec(p + 16); 2403 2404 /* This is just a tiny bit higher than the maximum 2405 returned by the streaming Zip bidder. This ensures 2406 that the more accurate seeking Zip parser wins 2407 whenever seek is available. */ 2408 return 32; 2409} 2410 2411/* 2412 * Examine Zip64 EOCD locator: If it's valid, store the information 2413 * from it. 2414 */ 2415static int 2416read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2417{ 2418 int64_t eocd64_offset; 2419 int64_t eocd64_size; 2420 2421 /* Sanity-check the locator record. */ 2422 2423 /* Central dir must be on first volume. */ 2424 if (archive_le32dec(p + 4) != 0) 2425 return 0; 2426 /* Must be only a single volume. */ 2427 if (archive_le32dec(p + 16) != 1) 2428 return 0; 2429 2430 /* Find the Zip64 EOCD record. */ 2431 eocd64_offset = archive_le64dec(p + 8); 2432 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2433 return 0; 2434 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2435 return 0; 2436 /* Make sure we can read all of it. */ 2437 eocd64_size = archive_le64dec(p + 4) + 12; 2438 if (eocd64_size < 56 || eocd64_size > 16384) 2439 return 0; 2440 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2441 return 0; 2442 2443 /* Sanity-check the EOCD64 */ 2444 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2445 return 0; 2446 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2447 return 0; 2448 /* CD can't be split. */ 2449 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2450 return 0; 2451 2452 /* Save the central directory offset for later use. */ 2453 zip->central_directory_offset = archive_le64dec(p + 48); 2454 2455 return 32; 2456} 2457 2458static int 2459archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2460{ 2461 struct zip *zip = (struct zip *)a->format->data; 2462 int64_t file_size, current_offset; 2463 const char *p; 2464 int i, tail; 2465 2466 /* If someone has already bid more than 32, then avoid 2467 trashing the look-ahead buffers with a seek. */ 2468 if (best_bid > 32) 2469 return (-1); 2470 2471 file_size = __archive_read_seek(a, 0, SEEK_END); 2472 if (file_size <= 0) 2473 return 0; 2474 2475 /* Search last 16k of file for end-of-central-directory 2476 * record (which starts with PK\005\006) */ 2477 tail = (int)zipmin(1024 * 16, file_size); 2478 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2479 if (current_offset < 0) 2480 return 0; 2481 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2482 return 0; 2483 /* Boyer-Moore search backwards from the end, since we want 2484 * to match the last EOCD in the file (there can be more than 2485 * one if there is an uncompressed Zip archive as a member 2486 * within this Zip archive). */ 2487 for (i = tail - 22; i > 0;) { 2488 switch (p[i]) { 2489 case 'P': 2490 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2491 int ret = read_eocd(zip, p + i, 2492 current_offset + i); 2493 /* Zip64 EOCD locator precedes 2494 * regular EOCD if present. */ 2495 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2496 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); 2497 if (ret_zip64 > ret) 2498 ret = ret_zip64; 2499 } 2500 return (ret); 2501 } 2502 i -= 4; 2503 break; 2504 case 'K': i -= 1; break; 2505 case 005: i -= 2; break; 2506 case 006: i -= 3; break; 2507 default: i -= 4; break; 2508 } 2509 } 2510 return 0; 2511} 2512 2513/* The red-black trees are only used in seeking mode to manage 2514 * the in-memory copy of the central directory. */ 2515 2516static int 2517cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2518{ 2519 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2520 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2521 2522 if (e1->local_header_offset > e2->local_header_offset) 2523 return -1; 2524 if (e1->local_header_offset < e2->local_header_offset) 2525 return 1; 2526 return 0; 2527} 2528 2529static int 2530cmp_key(const struct archive_rb_node *n, const void *key) 2531{ 2532 /* This function won't be called */ 2533 (void)n; /* UNUSED */ 2534 (void)key; /* UNUSED */ 2535 return 1; 2536} 2537 2538static const struct archive_rb_tree_ops rb_ops = { 2539 &cmp_node, &cmp_key 2540}; 2541 2542static int 2543rsrc_cmp_node(const struct archive_rb_node *n1, 2544 const struct archive_rb_node *n2) 2545{ 2546 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2547 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2548 2549 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2550} 2551 2552static int 2553rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2554{ 2555 const struct zip_entry *e = (const struct zip_entry *)n; 2556 return (strcmp((const char *)key, e->rsrcname.s)); 2557} 2558 2559static const struct archive_rb_tree_ops rb_rsrc_ops = { 2560 &rsrc_cmp_node, &rsrc_cmp_key 2561}; 2562 2563static const char * 2564rsrc_basename(const char *name, size_t name_length) 2565{ 2566 const char *s, *r; 2567 2568 r = s = name; 2569 for (;;) { 2570 s = memchr(s, '/', name_length - (s - name)); 2571 if (s == NULL) 2572 break; 2573 r = ++s; 2574 } 2575 return (r); 2576} 2577 2578static void 2579expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2580{ 2581 struct archive_string str; 2582 struct zip_entry *dir; 2583 char *s; 2584 2585 archive_string_init(&str); 2586 archive_strncpy(&str, name, name_length); 2587 for (;;) { 2588 s = strrchr(str.s, '/'); 2589 if (s == NULL) 2590 break; 2591 *s = '\0'; 2592 /* Transfer the parent directory from zip->tree_rsrc RB 2593 * tree to zip->tree RB tree to expose. */ 2594 dir = (struct zip_entry *) 2595 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2596 if (dir == NULL) 2597 break; 2598 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2599 archive_string_free(&dir->rsrcname); 2600 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2601 } 2602 archive_string_free(&str); 2603} 2604 2605static int 2606slurp_central_directory(struct archive_read *a, struct zip *zip) 2607{ 2608 ssize_t i; 2609 unsigned found; 2610 int64_t correction; 2611 ssize_t bytes_avail; 2612 const char *p; 2613 2614 /* 2615 * Find the start of the central directory. The end-of-CD 2616 * record has our starting point, but there are lots of 2617 * Zip archives which have had other data prepended to the 2618 * file, which makes the recorded offsets all too small. 2619 * So we search forward from the specified offset until we 2620 * find the real start of the central directory. Then we 2621 * know the correction we need to apply to account for leading 2622 * padding. 2623 */ 2624 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2625 return ARCHIVE_FATAL; 2626 2627 found = 0; 2628 while (!found) { 2629 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2630 return ARCHIVE_FATAL; 2631 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2632 switch (p[i + 3]) { 2633 case 'P': i += 3; break; 2634 case 'K': i += 2; break; 2635 case 001: i += 1; break; 2636 case 002: 2637 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2638 p += i; 2639 found = 1; 2640 } else 2641 i += 4; 2642 break; 2643 case 005: i += 1; break; 2644 case 006: 2645 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2646 p += i; 2647 found = 1; 2648 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2649 p += i; 2650 found = 1; 2651 } else 2652 i += 1; 2653 break; 2654 default: i += 4; break; 2655 } 2656 } 2657 __archive_read_consume(a, i); 2658 } 2659 correction = archive_filter_bytes(&a->archive, 0) 2660 - zip->central_directory_offset; 2661 2662 __archive_rb_tree_init(&zip->tree, &rb_ops); 2663 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2664 2665 zip->central_directory_entries_total = 0; 2666 while (1) { 2667 struct zip_entry *zip_entry; 2668 size_t filename_length, extra_length, comment_length; 2669 uint32_t external_attributes; 2670 const char *name, *r; 2671 2672 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2673 return ARCHIVE_FATAL; 2674 if (memcmp(p, "PK\006\006", 4) == 0 2675 || memcmp(p, "PK\005\006", 4) == 0) { 2676 break; 2677 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2678 archive_set_error(&a->archive, 2679 -1, "Invalid central directory signature"); 2680 return ARCHIVE_FATAL; 2681 } 2682 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2683 return ARCHIVE_FATAL; 2684 2685 zip_entry = calloc(1, sizeof(struct zip_entry)); 2686 zip_entry->next = zip->zip_entries; 2687 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2688 zip->zip_entries = zip_entry; 2689 zip->central_directory_entries_total++; 2690 2691 /* version = p[4]; */ 2692 zip_entry->system = p[5]; 2693 /* version_required = archive_le16dec(p + 6); */ 2694 zip_entry->zip_flags = archive_le16dec(p + 8); 2695 if (zip_entry->zip_flags 2696 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2697 zip->has_encrypted_entries = 1; 2698 } 2699 zip_entry->compression = (char)archive_le16dec(p + 10); 2700 zip_entry->mtime = zip_time(p + 12); 2701 zip_entry->crc32 = archive_le32dec(p + 16); 2702 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2703 zip_entry->decdat = p[13]; 2704 else 2705 zip_entry->decdat = p[19]; 2706 zip_entry->compressed_size = archive_le32dec(p + 20); 2707 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2708 filename_length = archive_le16dec(p + 28); 2709 extra_length = archive_le16dec(p + 30); 2710 comment_length = archive_le16dec(p + 32); 2711 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2712 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2713 external_attributes = archive_le32dec(p + 38); 2714 zip_entry->local_header_offset = 2715 archive_le32dec(p + 42) + correction; 2716 2717 /* If we can't guess the mode, leave it zero here; 2718 when we read the local file header we might get 2719 more information. */ 2720 if (zip_entry->system == 3) { 2721 zip_entry->mode = external_attributes >> 16; 2722 } else if (zip_entry->system == 0) { 2723 // Interpret MSDOS directory bit 2724 if (0x10 == (external_attributes & 0x10)) { 2725 zip_entry->mode = AE_IFDIR | 0775; 2726 } else { 2727 zip_entry->mode = AE_IFREG | 0664; 2728 } 2729 if (0x01 == (external_attributes & 0x01)) { 2730 // Read-only bit; strip write permissions 2731 zip_entry->mode &= 0555; 2732 } 2733 } else { 2734 zip_entry->mode = 0; 2735 } 2736 2737 /* We're done with the regular data; get the filename and 2738 * extra data. */ 2739 __archive_read_consume(a, 46); 2740 p = __archive_read_ahead(a, filename_length + extra_length, 2741 NULL); 2742 if (p == NULL) { 2743 archive_set_error(&a->archive, 2744 ARCHIVE_ERRNO_FILE_FORMAT, 2745 "Truncated ZIP file header"); 2746 return ARCHIVE_FATAL; 2747 } 2748 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2749 return ARCHIVE_FATAL; 2750 } 2751 2752 /* 2753 * Mac resource fork files are stored under the 2754 * "__MACOSX/" directory, so we should check if 2755 * it is. 2756 */ 2757 if (!zip->process_mac_extensions) { 2758 /* Treat every entry as a regular entry. */ 2759 __archive_rb_tree_insert_node(&zip->tree, 2760 &zip_entry->node); 2761 } else { 2762 name = p; 2763 r = rsrc_basename(name, filename_length); 2764 if (filename_length >= 9 && 2765 strncmp("__MACOSX/", name, 9) == 0) { 2766 /* If this file is not a resource fork nor 2767 * a directory. We should treat it as a non 2768 * resource fork file to expose it. */ 2769 if (name[filename_length-1] != '/' && 2770 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2771 __archive_rb_tree_insert_node( 2772 &zip->tree, &zip_entry->node); 2773 /* Expose its parent directories. */ 2774 expose_parent_dirs(zip, name, 2775 filename_length); 2776 } else { 2777 /* This file is a resource fork file or 2778 * a directory. */ 2779 archive_strncpy(&(zip_entry->rsrcname), 2780 name, filename_length); 2781 __archive_rb_tree_insert_node( 2782 &zip->tree_rsrc, &zip_entry->node); 2783 } 2784 } else { 2785 /* Generate resource fork name to find its 2786 * resource file at zip->tree_rsrc. */ 2787 archive_strcpy(&(zip_entry->rsrcname), 2788 "__MACOSX/"); 2789 archive_strncat(&(zip_entry->rsrcname), 2790 name, r - name); 2791 archive_strcat(&(zip_entry->rsrcname), "._"); 2792 archive_strncat(&(zip_entry->rsrcname), 2793 name + (r - name), 2794 filename_length - (r - name)); 2795 /* Register an entry to RB tree to sort it by 2796 * file offset. */ 2797 __archive_rb_tree_insert_node(&zip->tree, 2798 &zip_entry->node); 2799 } 2800 } 2801 2802 /* Skip the comment too ... */ 2803 __archive_read_consume(a, 2804 filename_length + extra_length + comment_length); 2805 } 2806 2807 return ARCHIVE_OK; 2808} 2809 2810static ssize_t 2811zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2812{ 2813 const char *p; 2814 ssize_t filename_length, extra_length; 2815 2816 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2817 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2818 "Truncated ZIP file header"); 2819 return (ARCHIVE_WARN); 2820 } 2821 p += extra; 2822 2823 if (memcmp(p, "PK\003\004", 4) != 0) { 2824 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2825 return ARCHIVE_WARN; 2826 } 2827 filename_length = archive_le16dec(p + 26); 2828 extra_length = archive_le16dec(p + 28); 2829 2830 return (30 + filename_length + extra_length); 2831} 2832 2833static int 2834zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2835 struct zip_entry *rsrc) 2836{ 2837 struct zip *zip = (struct zip *)a->format->data; 2838 unsigned char *metadata, *mp; 2839 int64_t offset = archive_filter_bytes(&a->archive, 0); 2840 size_t remaining_bytes, metadata_bytes; 2841 ssize_t hsize; 2842 int ret = ARCHIVE_OK, eof; 2843 2844 switch(rsrc->compression) { 2845 case 0: /* No compression. */ 2846 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2847 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2848 "Malformed OS X metadata entry: inconsistent size"); 2849 return (ARCHIVE_FATAL); 2850 } 2851#ifdef HAVE_ZLIB_H 2852 case 8: /* Deflate compression. */ 2853#endif 2854 break; 2855 default: /* Unsupported compression. */ 2856 /* Return a warning. */ 2857 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2858 "Unsupported ZIP compression method (%s)", 2859 compression_name(rsrc->compression)); 2860 /* We can't decompress this entry, but we will 2861 * be able to skip() it and try the next entry. */ 2862 return (ARCHIVE_WARN); 2863 } 2864 2865 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2866 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2867 "Mac metadata is too large: %jd > 4M bytes", 2868 (intmax_t)rsrc->uncompressed_size); 2869 return (ARCHIVE_WARN); 2870 } 2871 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2872 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2873 "Mac metadata is too large: %jd > 4M bytes", 2874 (intmax_t)rsrc->compressed_size); 2875 return (ARCHIVE_WARN); 2876 } 2877 2878 metadata = malloc((size_t)rsrc->uncompressed_size); 2879 if (metadata == NULL) { 2880 archive_set_error(&a->archive, ENOMEM, 2881 "Can't allocate memory for Mac metadata"); 2882 return (ARCHIVE_FATAL); 2883 } 2884 2885 if (offset < rsrc->local_header_offset) 2886 __archive_read_consume(a, rsrc->local_header_offset - offset); 2887 else if (offset != rsrc->local_header_offset) { 2888 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2889 } 2890 2891 hsize = zip_get_local_file_header_size(a, 0); 2892 __archive_read_consume(a, hsize); 2893 2894 remaining_bytes = (size_t)rsrc->compressed_size; 2895 metadata_bytes = (size_t)rsrc->uncompressed_size; 2896 mp = metadata; 2897 eof = 0; 2898 while (!eof && remaining_bytes) { 2899 const unsigned char *p; 2900 ssize_t bytes_avail; 2901 size_t bytes_used; 2902 2903 p = __archive_read_ahead(a, 1, &bytes_avail); 2904 if (p == NULL) { 2905 archive_set_error(&a->archive, 2906 ARCHIVE_ERRNO_FILE_FORMAT, 2907 "Truncated ZIP file header"); 2908 ret = ARCHIVE_WARN; 2909 goto exit_mac_metadata; 2910 } 2911 if ((size_t)bytes_avail > remaining_bytes) 2912 bytes_avail = remaining_bytes; 2913 switch(rsrc->compression) { 2914 case 0: /* No compression. */ 2915 if ((size_t)bytes_avail > metadata_bytes) 2916 bytes_avail = metadata_bytes; 2917 memcpy(mp, p, bytes_avail); 2918 bytes_used = (size_t)bytes_avail; 2919 metadata_bytes -= bytes_used; 2920 mp += bytes_used; 2921 if (metadata_bytes == 0) 2922 eof = 1; 2923 break; 2924#ifdef HAVE_ZLIB_H 2925 case 8: /* Deflate compression. */ 2926 { 2927 int r; 2928 2929 ret = zip_deflate_init(a, zip); 2930 if (ret != ARCHIVE_OK) 2931 goto exit_mac_metadata; 2932 zip->stream.next_in = 2933 (Bytef *)(uintptr_t)(const void *)p; 2934 zip->stream.avail_in = (uInt)bytes_avail; 2935 zip->stream.total_in = 0; 2936 zip->stream.next_out = mp; 2937 zip->stream.avail_out = (uInt)metadata_bytes; 2938 zip->stream.total_out = 0; 2939 2940 r = inflate(&zip->stream, 0); 2941 switch (r) { 2942 case Z_OK: 2943 break; 2944 case Z_STREAM_END: 2945 eof = 1; 2946 break; 2947 case Z_MEM_ERROR: 2948 archive_set_error(&a->archive, ENOMEM, 2949 "Out of memory for ZIP decompression"); 2950 ret = ARCHIVE_FATAL; 2951 goto exit_mac_metadata; 2952 default: 2953 archive_set_error(&a->archive, 2954 ARCHIVE_ERRNO_MISC, 2955 "ZIP decompression failed (%d)", r); 2956 ret = ARCHIVE_FATAL; 2957 goto exit_mac_metadata; 2958 } 2959 bytes_used = zip->stream.total_in; 2960 metadata_bytes -= zip->stream.total_out; 2961 mp += zip->stream.total_out; 2962 break; 2963 } 2964#endif 2965 default: 2966 bytes_used = 0; 2967 break; 2968 } 2969 __archive_read_consume(a, bytes_used); 2970 remaining_bytes -= bytes_used; 2971 } 2972 archive_entry_copy_mac_metadata(entry, metadata, 2973 (size_t)rsrc->uncompressed_size - metadata_bytes); 2974 2975exit_mac_metadata: 2976 __archive_read_seek(a, offset, SEEK_SET); 2977 zip->decompress_init = 0; 2978 free(metadata); 2979 return (ret); 2980} 2981 2982static int 2983archive_read_format_zip_seekable_read_header(struct archive_read *a, 2984 struct archive_entry *entry) 2985{ 2986 struct zip *zip = (struct zip *)a->format->data; 2987 struct zip_entry *rsrc; 2988 int64_t offset; 2989 int r, ret = ARCHIVE_OK; 2990 2991 /* 2992 * It should be sufficient to call archive_read_next_header() for 2993 * a reader to determine if an entry is encrypted or not. If the 2994 * encryption of an entry is only detectable when calling 2995 * archive_read_data(), so be it. We'll do the same check there 2996 * as well. 2997 */ 2998 if (zip->has_encrypted_entries == 2999 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 3000 zip->has_encrypted_entries = 0; 3001 3002 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 3003 if (a->archive.archive_format_name == NULL) 3004 a->archive.archive_format_name = "ZIP"; 3005 3006 if (zip->zip_entries == NULL) { 3007 r = slurp_central_directory(a, zip); 3008 if (r != ARCHIVE_OK) 3009 return r; 3010 /* Get first entry whose local header offset is lower than 3011 * other entries in the archive file. */ 3012 zip->entry = 3013 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 3014 } else if (zip->entry != NULL) { 3015 /* Get next entry in local header offset order. */ 3016 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 3017 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 3018 } 3019 3020 if (zip->entry == NULL) 3021 return ARCHIVE_EOF; 3022 3023 if (zip->entry->rsrcname.s) 3024 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 3025 &zip->tree_rsrc, zip->entry->rsrcname.s); 3026 else 3027 rsrc = NULL; 3028 3029 if (zip->cctx_valid) 3030 archive_decrypto_aes_ctr_release(&zip->cctx); 3031 if (zip->hctx_valid) 3032 archive_hmac_sha1_cleanup(&zip->hctx); 3033 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 3034 __archive_read_reset_passphrase(a); 3035 3036 /* File entries are sorted by the header offset, we should mostly 3037 * use __archive_read_consume to advance a read point to avoid redundant 3038 * data reading. */ 3039 offset = archive_filter_bytes(&a->archive, 0); 3040 if (offset < zip->entry->local_header_offset) 3041 __archive_read_consume(a, 3042 zip->entry->local_header_offset - offset); 3043 else if (offset != zip->entry->local_header_offset) { 3044 __archive_read_seek(a, zip->entry->local_header_offset, 3045 SEEK_SET); 3046 } 3047 zip->unconsumed = 0; 3048 r = zip_read_local_file_header(a, entry, zip); 3049 if (r != ARCHIVE_OK) 3050 return r; 3051 if (rsrc) { 3052 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3053 if (ret2 < ret) 3054 ret = ret2; 3055 } 3056 return (ret); 3057} 3058 3059/* 3060 * We're going to seek for the next header anyway, so we don't 3061 * need to bother doing anything here. 3062 */ 3063static int 3064archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3065{ 3066 struct zip *zip; 3067 zip = (struct zip *)(a->format->data); 3068 3069 zip->unconsumed = 0; 3070 return (ARCHIVE_OK); 3071} 3072 3073int 3074archive_read_support_format_zip_seekable(struct archive *_a) 3075{ 3076 struct archive_read *a = (struct archive_read *)_a; 3077 struct zip *zip; 3078 int r; 3079 3080 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3081 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3082 3083 zip = (struct zip *)calloc(1, sizeof(*zip)); 3084 if (zip == NULL) { 3085 archive_set_error(&a->archive, ENOMEM, 3086 "Can't allocate zip data"); 3087 return (ARCHIVE_FATAL); 3088 } 3089 3090#ifdef HAVE_COPYFILE_H 3091 /* Set this by default on Mac OS. */ 3092 zip->process_mac_extensions = 1; 3093#endif 3094 3095 /* 3096 * Until enough data has been read, we cannot tell about 3097 * any encrypted entries yet. 3098 */ 3099 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3100 zip->crc32func = real_crc32; 3101 3102 r = __archive_read_register_format(a, 3103 zip, 3104 "zip", 3105 archive_read_format_zip_seekable_bid, 3106 archive_read_format_zip_options, 3107 archive_read_format_zip_seekable_read_header, 3108 archive_read_format_zip_read_data, 3109 archive_read_format_zip_read_data_skip_seekable, 3110 NULL, 3111 archive_read_format_zip_cleanup, 3112 archive_read_support_format_zip_capabilities_seekable, 3113 archive_read_format_zip_has_encrypted_entries); 3114 3115 if (r != ARCHIVE_OK) 3116 free(zip); 3117 return (ARCHIVE_OK); 3118} 3119