archive_read_support_format_zip.c revision 311042
168651Skris/*- 2280304Sjkim * Copyright (c) 2004-2013 Tim Kientzle 3280304Sjkim * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA 4280304Sjkim * Copyright (c) 2013 Konrad Kleine 568651Skris * All rights reserved. 668651Skris * 768651Skris * Redistribution and use in source and binary forms, with or without 868651Skris * modification, are permitted provided that the following conditions 968651Skris * are met: 1068651Skris * 1. Redistributions of source code must retain the above copyright 1168651Skris * notice, this list of conditions and the following disclaimer. 1268651Skris * 2. Redistributions in binary form must reproduce the above copyright 1368651Skris * notice, this list of conditions and the following disclaimer in the 14280304Sjkim * documentation and/or other materials provided with the distribution. 1568651Skris * 1668651Skris * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 1768651Skris * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 1868651Skris * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 1968651Skris * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 2068651Skris * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 2168651Skris * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 2268651Skris * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 2368651Skris * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 2468651Skris * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 2568651Skris * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 2668651Skris */ 2768651Skris 2868651Skris#include "archive_platform.h" 2968651Skris__FBSDID("$FreeBSD: stable/10/contrib/libarchive/libarchive/archive_read_support_format_zip.c 311042 2017-01-02 01:43:11Z mm $"); 3068651Skris 3168651Skris/* 3268651Skris * The definitive documentation of the Zip file format is: 3368651Skris * http://www.pkware.com/documents/casestudies/APPNOTE.TXT 3468651Skris * 3568651Skris * The Info-Zip project has pioneered various extensions to better 3668651Skris * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 3768651Skris * "Ux", and 0x7875 "ux" extensions for time and ownership 3868651Skris * information. 3968651Skris * 4068651Skris * History of this code: The streaming Zip reader was first added to 4168651Skris * libarchive in January 2005. Support for seekable input sources was 4268651Skris * added in Nov 2011. Zip64 support (including a significant code 4368651Skris * refactoring) was added in 2014. 4468651Skris */ 4568651Skris 4668651Skris#ifdef HAVE_ERRNO_H 4768651Skris#include <errno.h> 4868651Skris#endif 4968651Skris#ifdef HAVE_STDLIB_H 5068651Skris#include <stdlib.h> 5168651Skris#endif 5268651Skris#ifdef HAVE_ZLIB_H 5368651Skris#include <zlib.h> 5468651Skris#endif 5568651Skris 5668651Skris#include "archive.h" 5768651Skris#include "archive_digest_private.h" 5868651Skris#include "archive_cryptor_private.h" 5968651Skris#include "archive_endian.h" 60280304Sjkim#include "archive_entry.h" 61280304Sjkim#include "archive_entry_locale.h" 62280304Sjkim#include "archive_hmac_private.h" 63280304Sjkim#include "archive_private.h" 6468651Skris#include "archive_rb.h" 6568651Skris#include "archive_read_private.h" 6668651Skris 6768651Skris#ifndef HAVE_ZLIB_H 6868651Skris#include "archive_crc32.h" 6968651Skris#endif 70280304Sjkim 71280304Sjkimstruct zip_entry { 72280304Sjkim struct archive_rb_node node; 73280304Sjkim struct zip_entry *next; 74280304Sjkim int64_t local_header_offset; 7568651Skris int64_t compressed_size; 7668651Skris int64_t uncompressed_size; 77280304Sjkim int64_t gid; 78280304Sjkim int64_t uid; 7968651Skris struct archive_string rsrcname; 80280304Sjkim time_t mtime; 81280304Sjkim time_t atime; 82280304Sjkim time_t ctime; 83280304Sjkim uint32_t crc32; 84280304Sjkim uint16_t mode; 85280304Sjkim uint16_t zip_flags; /* From GP Flags Field */ 86280304Sjkim unsigned char compression; 87280304Sjkim unsigned char system; /* From "version written by" */ 8868651Skris unsigned char flags; /* Our extra markers. */ 8968651Skris unsigned char decdat;/* Used for Decryption check */ 90280304Sjkim 91280304Sjkim /* WinZip AES encryption extra field should be available 92280304Sjkim * when compression is 99. */ 93 struct { 94 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 95 unsigned vendor; 96#define AES_VENDOR_AE_1 0x0001 97#define AES_VENDOR_AE_2 0x0002 98 /* AES encryption strength: 99 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 100 unsigned strength; 101 /* Actual compression method. */ 102 unsigned char compression; 103 } aes_extra; 104}; 105 106struct trad_enc_ctx { 107 uint32_t keys[3]; 108}; 109 110/* Bits used in zip_flags. */ 111#define ZIP_ENCRYPTED (1 << 0) 112#define ZIP_LENGTH_AT_END (1 << 3) 113#define ZIP_STRONG_ENCRYPTED (1 << 6) 114#define ZIP_UTF8_NAME (1 << 11) 115/* See "7.2 Single Password Symmetric Encryption Method" 116 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ 117#define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) 118 119/* Bits used in flags. */ 120#define LA_USED_ZIP64 (1 << 0) 121#define LA_FROM_CENTRAL_DIRECTORY (1 << 1) 122 123/* 124 * See "WinZip - AES Encryption Information" 125 * http://www.winzip.com/aes_info.htm 126 */ 127/* Value used in compression method. */ 128#define WINZIP_AES_ENCRYPTION 99 129/* Authentication code size. */ 130#define AUTH_CODE_SIZE 10 131/**/ 132#define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) 133 134struct zip { 135 /* Structural information about the archive. */ 136 struct archive_string format_name; 137 int64_t central_directory_offset; 138 size_t central_directory_entries_total; 139 size_t central_directory_entries_on_this_disk; 140 int has_encrypted_entries; 141 142 /* List of entries (seekable Zip only) */ 143 struct zip_entry *zip_entries; 144 struct archive_rb_tree tree; 145 struct archive_rb_tree tree_rsrc; 146 147 /* Bytes read but not yet consumed via __archive_read_consume() */ 148 size_t unconsumed; 149 150 /* Information about entry we're currently reading. */ 151 struct zip_entry *entry; 152 int64_t entry_bytes_remaining; 153 154 /* These count the number of bytes actually read for the entry. */ 155 int64_t entry_compressed_bytes_read; 156 int64_t entry_uncompressed_bytes_read; 157 158 /* Running CRC32 of the decompressed data */ 159 unsigned long entry_crc32; 160 unsigned long (*crc32func)(unsigned long, const void *, 161 size_t); 162 char ignore_crc32; 163 164 /* Flags to mark progress of decompression. */ 165 char decompress_init; 166 char end_of_entry; 167 168#ifdef HAVE_ZLIB_H 169 unsigned char *uncompressed_buffer; 170 size_t uncompressed_buffer_size; 171 z_stream stream; 172 char stream_valid; 173#endif 174 175 struct archive_string_conv *sconv; 176 struct archive_string_conv *sconv_default; 177 struct archive_string_conv *sconv_utf8; 178 int init_default_conversion; 179 int process_mac_extensions; 180 181 char init_decryption; 182 183 /* Decryption buffer. */ 184 /* 185 * The decrypted data starts at decrypted_ptr and 186 * extends for decrypted_bytes_remaining. Decryption 187 * adds new data to the end of this block, data is returned 188 * to clients from the beginning. When the block hits the 189 * end of decrypted_buffer, it has to be shuffled back to 190 * the beginning of the buffer. 191 */ 192 unsigned char *decrypted_buffer; 193 unsigned char *decrypted_ptr; 194 size_t decrypted_buffer_size; 195 size_t decrypted_bytes_remaining; 196 size_t decrypted_unconsumed_bytes; 197 198 /* Traditional PKWARE decryption. */ 199 struct trad_enc_ctx tctx; 200 char tctx_valid; 201 202 /* WinZip AES decyption. */ 203 /* Contexts used for AES decryption. */ 204 archive_crypto_ctx cctx; 205 char cctx_valid; 206 archive_hmac_sha1_ctx hctx; 207 char hctx_valid; 208 209 /* Strong encryption's decryption header information. */ 210 unsigned iv_size; 211 unsigned alg_id; 212 unsigned bit_len; 213 unsigned flags; 214 unsigned erd_size; 215 unsigned v_size; 216 unsigned v_crc32; 217 uint8_t *iv; 218 uint8_t *erd; 219 uint8_t *v_data; 220}; 221 222/* Many systems define min or MIN, but not all. */ 223#define zipmin(a,b) ((a) < (b) ? (a) : (b)) 224 225/* ------------------------------------------------------------------------ */ 226 227/* 228 Traditional PKWARE Decryption functions. 229 */ 230 231static void 232trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 233{ 234 uint8_t t; 235#define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 236 237 ctx->keys[0] = CRC32(ctx->keys[0], c); 238 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 239 t = (ctx->keys[1] >> 24) & 0xff; 240 ctx->keys[2] = CRC32(ctx->keys[2], t); 241#undef CRC32 242} 243 244static uint8_t 245trad_enc_decypt_byte(struct trad_enc_ctx *ctx) 246{ 247 unsigned temp = ctx->keys[2] | 2; 248 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 249} 250 251static void 252trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 253 size_t in_len, uint8_t *out, size_t out_len) 254{ 255 unsigned i, max; 256 257 max = (unsigned)((in_len < out_len)? in_len: out_len); 258 259 for (i = 0; i < max; i++) { 260 uint8_t t = in[i] ^ trad_enc_decypt_byte(ctx); 261 out[i] = t; 262 trad_enc_update_keys(ctx, t); 263 } 264} 265 266static int 267trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 268 const uint8_t *key, size_t key_len, uint8_t *crcchk) 269{ 270 uint8_t header[12]; 271 272 if (key_len < 12) { 273 *crcchk = 0xff; 274 return -1; 275 } 276 277 ctx->keys[0] = 305419896L; 278 ctx->keys[1] = 591751049L; 279 ctx->keys[2] = 878082192L; 280 281 for (;pw_len; --pw_len) 282 trad_enc_update_keys(ctx, *pw++); 283 284 trad_enc_decrypt_update(ctx, key, 12, header, 12); 285 /* Return the last byte for CRC check. */ 286 *crcchk = header[11]; 287 return 0; 288} 289 290#if 0 291static void 292crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 293 int key_size) 294{ 295#define MD_SIZE 20 296 archive_sha1_ctx ctx; 297 unsigned char md1[MD_SIZE]; 298 unsigned char md2[MD_SIZE * 2]; 299 unsigned char mkb[64]; 300 int i; 301 302 archive_sha1_init(&ctx); 303 archive_sha1_update(&ctx, p, size); 304 archive_sha1_final(&ctx, md1); 305 306 memset(mkb, 0x36, sizeof(mkb)); 307 for (i = 0; i < MD_SIZE; i++) 308 mkb[i] ^= md1[i]; 309 archive_sha1_init(&ctx); 310 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 311 archive_sha1_final(&ctx, md2); 312 313 memset(mkb, 0x5C, sizeof(mkb)); 314 for (i = 0; i < MD_SIZE; i++) 315 mkb[i] ^= md1[i]; 316 archive_sha1_init(&ctx); 317 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 318 archive_sha1_final(&ctx, md2 + MD_SIZE); 319 320 if (key_size > 32) 321 key_size = 32; 322 memcpy(key, md2, key_size); 323#undef MD_SIZE 324} 325#endif 326 327/* 328 * Common code for streaming or seeking modes. 329 * 330 * Includes code to read local file headers, decompress data 331 * from entry bodies, and common API. 332 */ 333 334static unsigned long 335real_crc32(unsigned long crc, const void *buff, size_t len) 336{ 337 return crc32(crc, buff, (unsigned int)len); 338} 339 340/* Used by "ignorecrc32" option to speed up tests. */ 341static unsigned long 342fake_crc32(unsigned long crc, const void *buff, size_t len) 343{ 344 (void)crc; /* UNUSED */ 345 (void)buff; /* UNUSED */ 346 (void)len; /* UNUSED */ 347 return 0; 348} 349 350static struct { 351 int id; 352 const char * name; 353} compression_methods[] = { 354 {0, "uncompressed"}, /* The file is stored (no compression) */ 355 {1, "shrinking"}, /* The file is Shrunk */ 356 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 357 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 358 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 359 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 360 {6, "imploded"}, /* The file is Imploded */ 361 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 362 {8, "deflation"}, /* The file is Deflated */ 363 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 364 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 365 * (old IBM TERSE) */ 366 {11, "reserved"}, /* Reserved by PKWARE */ 367 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 368 {13, "reserved"}, /* Reserved by PKWARE */ 369 {14, "lzma"}, /* LZMA (EFS) */ 370 {15, "reserved"}, /* Reserved by PKWARE */ 371 {16, "reserved"}, /* Reserved by PKWARE */ 372 {17, "reserved"}, /* Reserved by PKWARE */ 373 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 374 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 375 {97, "wav-pack"}, /* WavPack compressed data */ 376 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 377 {99, "aes"} /* WinZip AES encryption */ 378}; 379 380static const char * 381compression_name(const int compression) 382{ 383 static const int num_compression_methods = 384 sizeof(compression_methods)/sizeof(compression_methods[0]); 385 int i=0; 386 387 while(compression >= 0 && i < num_compression_methods) { 388 if (compression_methods[i].id == compression) 389 return compression_methods[i].name; 390 i++; 391 } 392 return "??"; 393} 394 395/* Convert an MSDOS-style date/time into Unix-style time. */ 396static time_t 397zip_time(const char *p) 398{ 399 int msTime, msDate; 400 struct tm ts; 401 402 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 403 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 404 405 memset(&ts, 0, sizeof(ts)); 406 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 407 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 408 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 409 ts.tm_hour = (msTime >> 11) & 0x1f; 410 ts.tm_min = (msTime >> 5) & 0x3f; 411 ts.tm_sec = (msTime << 1) & 0x3e; 412 ts.tm_isdst = -1; 413 return mktime(&ts); 414} 415 416/* 417 * The extra data is stored as a list of 418 * id1+size1+data1 + id2+size2+data2 ... 419 * triplets. id and size are 2 bytes each. 420 */ 421static int 422process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry) 423{ 424 unsigned offset = 0; 425 426 if (extra_length == 0) { 427 return ARCHIVE_OK; 428 } 429 430 if (extra_length < 4) { 431 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 432 "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length); 433 return ARCHIVE_FAILED; 434 } 435 while (offset <= extra_length - 4) { 436 unsigned short headerid = archive_le16dec(p + offset); 437 unsigned short datasize = archive_le16dec(p + offset + 2); 438 439 offset += 4; 440 if (offset + datasize > extra_length) { 441 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 442 "Extra data overflow: Need %d bytes but only found %d bytes", 443 (int)datasize, (int)(extra_length - offset)); 444 return ARCHIVE_FAILED; 445 } 446#ifdef DEBUG 447 fprintf(stderr, "Header id 0x%04x, length %d\n", 448 headerid, datasize); 449#endif 450 switch (headerid) { 451 case 0x0001: 452 /* Zip64 extended information extra field. */ 453 zip_entry->flags |= LA_USED_ZIP64; 454 if (zip_entry->uncompressed_size == 0xffffffff) { 455 if (datasize < 8) 456 break; 457 zip_entry->uncompressed_size = 458 archive_le64dec(p + offset); 459 offset += 8; 460 datasize -= 8; 461 } 462 if (zip_entry->compressed_size == 0xffffffff) { 463 if (datasize < 8) 464 break; 465 zip_entry->compressed_size = 466 archive_le64dec(p + offset); 467 offset += 8; 468 datasize -= 8; 469 } 470 if (zip_entry->local_header_offset == 0xffffffff) { 471 if (datasize < 8) 472 break; 473 zip_entry->local_header_offset = 474 archive_le64dec(p + offset); 475 offset += 8; 476 datasize -= 8; 477 } 478 /* archive_le32dec(p + offset) gives disk 479 * on which file starts, but we don't handle 480 * multi-volume Zip files. */ 481 break; 482#ifdef DEBUG 483 case 0x0017: 484 { 485 /* Strong encryption field. */ 486 if (archive_le16dec(p + offset) == 2) { 487 unsigned algId = 488 archive_le16dec(p + offset + 2); 489 unsigned bitLen = 490 archive_le16dec(p + offset + 4); 491 int flags = 492 archive_le16dec(p + offset + 6); 493 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 494 "flgas=%d\n", algId, bitLen,flags); 495 } 496 break; 497 } 498#endif 499 case 0x5455: 500 { 501 /* Extended time field "UT". */ 502 int flags = p[offset]; 503 offset++; 504 datasize--; 505 /* Flag bits indicate which dates are present. */ 506 if (flags & 0x01) 507 { 508#ifdef DEBUG 509 fprintf(stderr, "mtime: %lld -> %d\n", 510 (long long)zip_entry->mtime, 511 archive_le32dec(p + offset)); 512#endif 513 if (datasize < 4) 514 break; 515 zip_entry->mtime = archive_le32dec(p + offset); 516 offset += 4; 517 datasize -= 4; 518 } 519 if (flags & 0x02) 520 { 521 if (datasize < 4) 522 break; 523 zip_entry->atime = archive_le32dec(p + offset); 524 offset += 4; 525 datasize -= 4; 526 } 527 if (flags & 0x04) 528 { 529 if (datasize < 4) 530 break; 531 zip_entry->ctime = archive_le32dec(p + offset); 532 offset += 4; 533 datasize -= 4; 534 } 535 break; 536 } 537 case 0x5855: 538 { 539 /* Info-ZIP Unix Extra Field (old version) "UX". */ 540 if (datasize >= 8) { 541 zip_entry->atime = archive_le32dec(p + offset); 542 zip_entry->mtime = 543 archive_le32dec(p + offset + 4); 544 } 545 if (datasize >= 12) { 546 zip_entry->uid = 547 archive_le16dec(p + offset + 8); 548 zip_entry->gid = 549 archive_le16dec(p + offset + 10); 550 } 551 break; 552 } 553 case 0x6c78: 554 { 555 /* Experimental 'xl' field */ 556 /* 557 * Introduced Dec 2013 to provide a way to 558 * include external file attributes (and other 559 * fields that ordinarily appear only in 560 * central directory) in local file header. 561 * This provides file type and permission 562 * information necessary to support full 563 * streaming extraction. Currently being 564 * discussed with other Zip developers 565 * ... subject to change. 566 * 567 * Format: 568 * The field starts with a bitmap that specifies 569 * which additional fields are included. The 570 * bitmap is variable length and can be extended in 571 * the future. 572 * 573 * n bytes - feature bitmap: first byte has low-order 574 * 7 bits. If high-order bit is set, a subsequent 575 * byte holds the next 7 bits, etc. 576 * 577 * if bitmap & 1, 2 byte "version made by" 578 * if bitmap & 2, 2 byte "internal file attributes" 579 * if bitmap & 4, 4 byte "external file attributes" 580 * if bitmap & 8, 2 byte comment length + n byte comment 581 */ 582 int bitmap, bitmap_last; 583 584 if (datasize < 1) 585 break; 586 bitmap_last = bitmap = 0xff & p[offset]; 587 offset += 1; 588 datasize -= 1; 589 590 /* We only support first 7 bits of bitmap; skip rest. */ 591 while ((bitmap_last & 0x80) != 0 592 && datasize >= 1) { 593 bitmap_last = p[offset]; 594 offset += 1; 595 datasize -= 1; 596 } 597 598 if (bitmap & 1) { 599 /* 2 byte "version made by" */ 600 if (datasize < 2) 601 break; 602 zip_entry->system 603 = archive_le16dec(p + offset) >> 8; 604 offset += 2; 605 datasize -= 2; 606 } 607 if (bitmap & 2) { 608 /* 2 byte "internal file attributes" */ 609 uint32_t internal_attributes; 610 if (datasize < 2) 611 break; 612 internal_attributes 613 = archive_le16dec(p + offset); 614 /* Not used by libarchive at present. */ 615 (void)internal_attributes; /* UNUSED */ 616 offset += 2; 617 datasize -= 2; 618 } 619 if (bitmap & 4) { 620 /* 4 byte "external file attributes" */ 621 uint32_t external_attributes; 622 if (datasize < 4) 623 break; 624 external_attributes 625 = archive_le32dec(p + offset); 626 if (zip_entry->system == 3) { 627 zip_entry->mode 628 = external_attributes >> 16; 629 } else if (zip_entry->system == 0) { 630 // Interpret MSDOS directory bit 631 if (0x10 == (external_attributes & 0x10)) { 632 zip_entry->mode = AE_IFDIR | 0775; 633 } else { 634 zip_entry->mode = AE_IFREG | 0664; 635 } 636 if (0x01 == (external_attributes & 0x01)) { 637 // Read-only bit; strip write permissions 638 zip_entry->mode &= 0555; 639 } 640 } else { 641 zip_entry->mode = 0; 642 } 643 offset += 4; 644 datasize -= 4; 645 } 646 if (bitmap & 8) { 647 /* 2 byte comment length + comment */ 648 uint32_t comment_length; 649 if (datasize < 2) 650 break; 651 comment_length 652 = archive_le16dec(p + offset); 653 offset += 2; 654 datasize -= 2; 655 656 if (datasize < comment_length) 657 break; 658 /* Comment is not supported by libarchive */ 659 offset += comment_length; 660 datasize -= comment_length; 661 } 662 break; 663 } 664 case 0x7855: 665 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 666#ifdef DEBUG 667 fprintf(stderr, "uid %d gid %d\n", 668 archive_le16dec(p + offset), 669 archive_le16dec(p + offset + 2)); 670#endif 671 if (datasize >= 2) 672 zip_entry->uid = archive_le16dec(p + offset); 673 if (datasize >= 4) 674 zip_entry->gid = 675 archive_le16dec(p + offset + 2); 676 break; 677 case 0x7875: 678 { 679 /* Info-Zip Unix Extra Field (type 3) "ux". */ 680 int uidsize = 0, gidsize = 0; 681 682 /* TODO: support arbitrary uidsize/gidsize. */ 683 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 684 if (datasize >= 4) { 685 /* get a uid size. */ 686 uidsize = 0xff & (int)p[offset+1]; 687 if (uidsize == 2) 688 zip_entry->uid = 689 archive_le16dec( 690 p + offset + 2); 691 else if (uidsize == 4 && datasize >= 6) 692 zip_entry->uid = 693 archive_le32dec( 694 p + offset + 2); 695 } 696 if (datasize >= (2 + uidsize + 3)) { 697 /* get a gid size. */ 698 gidsize = 0xff & (int)p[offset+2+uidsize]; 699 if (gidsize == 2) 700 zip_entry->gid = 701 archive_le16dec( 702 p+offset+2+uidsize+1); 703 else if (gidsize == 4 && 704 datasize >= (2 + uidsize + 5)) 705 zip_entry->gid = 706 archive_le32dec( 707 p+offset+2+uidsize+1); 708 } 709 } 710 break; 711 } 712 case 0x9901: 713 /* WinZIp AES extra data field. */ 714 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 715 /* Vendor version. */ 716 zip_entry->aes_extra.vendor = 717 archive_le16dec(p + offset); 718 /* AES encryption strength. */ 719 zip_entry->aes_extra.strength = p[offset + 4]; 720 /* Actual compression method. */ 721 zip_entry->aes_extra.compression = 722 p[offset + 5]; 723 } 724 break; 725 default: 726 break; 727 } 728 offset += datasize; 729 } 730 if (offset != extra_length) { 731 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 732 "Malformed extra data: Consumed %d bytes of %d bytes", 733 (int)offset, (int)extra_length); 734 return ARCHIVE_FAILED; 735 } 736 return ARCHIVE_OK; 737} 738 739/* 740 * Assumes file pointer is at beginning of local file header. 741 */ 742static int 743zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 744 struct zip *zip) 745{ 746 const char *p; 747 const void *h; 748 const wchar_t *wp; 749 const char *cp; 750 size_t len, filename_length, extra_length; 751 struct archive_string_conv *sconv; 752 struct zip_entry *zip_entry = zip->entry; 753 struct zip_entry zip_entry_central_dir; 754 int ret = ARCHIVE_OK; 755 char version; 756 757 /* Save a copy of the original for consistency checks. */ 758 zip_entry_central_dir = *zip_entry; 759 760 zip->decompress_init = 0; 761 zip->end_of_entry = 0; 762 zip->entry_uncompressed_bytes_read = 0; 763 zip->entry_compressed_bytes_read = 0; 764 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 765 766 /* Setup default conversion. */ 767 if (zip->sconv == NULL && !zip->init_default_conversion) { 768 zip->sconv_default = 769 archive_string_default_conversion_for_read(&(a->archive)); 770 zip->init_default_conversion = 1; 771 } 772 773 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 774 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 775 "Truncated ZIP file header"); 776 return (ARCHIVE_FATAL); 777 } 778 779 if (memcmp(p, "PK\003\004", 4) != 0) { 780 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 781 return ARCHIVE_FATAL; 782 } 783 version = p[4]; 784 zip_entry->system = p[5]; 785 zip_entry->zip_flags = archive_le16dec(p + 6); 786 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 787 zip->has_encrypted_entries = 1; 788 archive_entry_set_is_data_encrypted(entry, 1); 789 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 790 zip_entry->zip_flags & ZIP_ENCRYPTED && 791 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 792 archive_entry_set_is_metadata_encrypted(entry, 1); 793 return ARCHIVE_FATAL; 794 } 795 } 796 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 797 zip_entry->compression = (char)archive_le16dec(p + 8); 798 zip_entry->mtime = zip_time(p + 10); 799 zip_entry->crc32 = archive_le32dec(p + 14); 800 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 801 zip_entry->decdat = p[11]; 802 else 803 zip_entry->decdat = p[17]; 804 zip_entry->compressed_size = archive_le32dec(p + 18); 805 zip_entry->uncompressed_size = archive_le32dec(p + 22); 806 filename_length = archive_le16dec(p + 26); 807 extra_length = archive_le16dec(p + 28); 808 809 __archive_read_consume(a, 30); 810 811 /* Read the filename. */ 812 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 813 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 814 "Truncated ZIP file header"); 815 return (ARCHIVE_FATAL); 816 } 817 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 818 /* The filename is stored to be UTF-8. */ 819 if (zip->sconv_utf8 == NULL) { 820 zip->sconv_utf8 = 821 archive_string_conversion_from_charset( 822 &a->archive, "UTF-8", 1); 823 if (zip->sconv_utf8 == NULL) 824 return (ARCHIVE_FATAL); 825 } 826 sconv = zip->sconv_utf8; 827 } else if (zip->sconv != NULL) 828 sconv = zip->sconv; 829 else 830 sconv = zip->sconv_default; 831 832 if (archive_entry_copy_pathname_l(entry, 833 h, filename_length, sconv) != 0) { 834 if (errno == ENOMEM) { 835 archive_set_error(&a->archive, ENOMEM, 836 "Can't allocate memory for Pathname"); 837 return (ARCHIVE_FATAL); 838 } 839 archive_set_error(&a->archive, 840 ARCHIVE_ERRNO_FILE_FORMAT, 841 "Pathname cannot be converted " 842 "from %s to current locale.", 843 archive_string_conversion_charset_name(sconv)); 844 ret = ARCHIVE_WARN; 845 } 846 __archive_read_consume(a, filename_length); 847 848 /* Read the extra data. */ 849 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 850 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 851 "Truncated ZIP file header"); 852 return (ARCHIVE_FATAL); 853 } 854 855 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { 856 return ARCHIVE_FATAL; 857 } 858 __archive_read_consume(a, extra_length); 859 860 /* Work around a bug in Info-Zip: When reading from a pipe, it 861 * stats the pipe instead of synthesizing a file entry. */ 862 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 863 zip_entry->mode &= ~ AE_IFMT; 864 zip_entry->mode |= AE_IFREG; 865 } 866 867 /* If the mode is totally empty, set some sane default. */ 868 if (zip_entry->mode == 0) { 869 zip_entry->mode |= 0664; 870 } 871 872 /* Make sure that entries with a trailing '/' are marked as directories 873 * even if the External File Attributes contains bogus values. If this 874 * is not a directory and there is no type, assume regularfile. */ 875 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { 876 int has_slash; 877 878 wp = archive_entry_pathname_w(entry); 879 if (wp != NULL) { 880 len = wcslen(wp); 881 has_slash = len > 0 && wp[len - 1] == L'/'; 882 } else { 883 cp = archive_entry_pathname(entry); 884 len = (cp != NULL)?strlen(cp):0; 885 has_slash = len > 0 && cp[len - 1] == '/'; 886 } 887 /* Correct file type as needed. */ 888 if (has_slash) { 889 zip_entry->mode &= ~AE_IFMT; 890 zip_entry->mode |= AE_IFDIR; 891 zip_entry->mode |= 0111; 892 } else if ((zip_entry->mode & AE_IFMT) == 0) { 893 zip_entry->mode |= AE_IFREG; 894 } 895 } 896 897 /* Make sure directories end in '/' */ 898 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 899 wp = archive_entry_pathname_w(entry); 900 if (wp != NULL) { 901 len = wcslen(wp); 902 if (len > 0 && wp[len - 1] != L'/') { 903 struct archive_wstring s; 904 archive_string_init(&s); 905 archive_wstrcat(&s, wp); 906 archive_wstrappend_wchar(&s, L'/'); 907 archive_entry_copy_pathname_w(entry, s.s); 908 } 909 } else { 910 cp = archive_entry_pathname(entry); 911 len = (cp != NULL)?strlen(cp):0; 912 if (len > 0 && cp[len - 1] != '/') { 913 struct archive_string s; 914 archive_string_init(&s); 915 archive_strcat(&s, cp); 916 archive_strappend_char(&s, '/'); 917 archive_entry_set_pathname(entry, s.s); 918 } 919 } 920 } 921 922 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 923 /* If this came from the central dir, it's size info 924 * is definitive, so ignore the length-at-end flag. */ 925 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 926 /* If local header is missing a value, use the one from 927 the central directory. If both have it, warn about 928 mismatches. */ 929 if (zip_entry->crc32 == 0) { 930 zip_entry->crc32 = zip_entry_central_dir.crc32; 931 } else if (!zip->ignore_crc32 932 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 933 archive_set_error(&a->archive, 934 ARCHIVE_ERRNO_FILE_FORMAT, 935 "Inconsistent CRC32 values"); 936 ret = ARCHIVE_WARN; 937 } 938 if (zip_entry->compressed_size == 0) { 939 zip_entry->compressed_size 940 = zip_entry_central_dir.compressed_size; 941 } else if (zip_entry->compressed_size 942 != zip_entry_central_dir.compressed_size) { 943 archive_set_error(&a->archive, 944 ARCHIVE_ERRNO_FILE_FORMAT, 945 "Inconsistent compressed size: " 946 "%jd in central directory, %jd in local header", 947 (intmax_t)zip_entry_central_dir.compressed_size, 948 (intmax_t)zip_entry->compressed_size); 949 ret = ARCHIVE_WARN; 950 } 951 if (zip_entry->uncompressed_size == 0) { 952 zip_entry->uncompressed_size 953 = zip_entry_central_dir.uncompressed_size; 954 } else if (zip_entry->uncompressed_size 955 != zip_entry_central_dir.uncompressed_size) { 956 archive_set_error(&a->archive, 957 ARCHIVE_ERRNO_FILE_FORMAT, 958 "Inconsistent uncompressed size: " 959 "%jd in central directory, %jd in local header", 960 (intmax_t)zip_entry_central_dir.uncompressed_size, 961 (intmax_t)zip_entry->uncompressed_size); 962 ret = ARCHIVE_WARN; 963 } 964 } 965 966 /* Populate some additional entry fields: */ 967 archive_entry_set_mode(entry, zip_entry->mode); 968 archive_entry_set_uid(entry, zip_entry->uid); 969 archive_entry_set_gid(entry, zip_entry->gid); 970 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 971 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 972 archive_entry_set_atime(entry, zip_entry->atime, 0); 973 974 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 975 size_t linkname_length; 976 977 if (zip_entry->compressed_size > 64 * 1024) { 978 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 979 "Zip file with oversized link entry"); 980 return ARCHIVE_FATAL; 981 } 982 983 linkname_length = (size_t)zip_entry->compressed_size; 984 985 archive_entry_set_size(entry, 0); 986 p = __archive_read_ahead(a, linkname_length, NULL); 987 if (p == NULL) { 988 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 989 "Truncated Zip file"); 990 return ARCHIVE_FATAL; 991 } 992 993 sconv = zip->sconv; 994 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 995 sconv = zip->sconv_utf8; 996 if (sconv == NULL) 997 sconv = zip->sconv_default; 998 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 999 sconv) != 0) { 1000 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 1001 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 1002 archive_entry_copy_symlink_l(entry, p, 1003 linkname_length, NULL); 1004 if (errno == ENOMEM) { 1005 archive_set_error(&a->archive, ENOMEM, 1006 "Can't allocate memory for Symlink"); 1007 return (ARCHIVE_FATAL); 1008 } 1009 /* 1010 * Since there is no character-set regulation for 1011 * symlink name, do not report the conversion error 1012 * in an automatic conversion. 1013 */ 1014 if (sconv != zip->sconv_utf8 || 1015 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 1016 archive_set_error(&a->archive, 1017 ARCHIVE_ERRNO_FILE_FORMAT, 1018 "Symlink cannot be converted " 1019 "from %s to current locale.", 1020 archive_string_conversion_charset_name( 1021 sconv)); 1022 ret = ARCHIVE_WARN; 1023 } 1024 } 1025 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1026 1027 if (__archive_read_consume(a, linkname_length) < 0) { 1028 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1029 "Read error skipping symlink target name"); 1030 return ARCHIVE_FATAL; 1031 } 1032 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1033 || zip_entry->uncompressed_size > 0) { 1034 /* Set the size only if it's meaningful. */ 1035 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1036 } 1037 zip->entry_bytes_remaining = zip_entry->compressed_size; 1038 1039 /* If there's no body, force read_data() to return EOF immediately. */ 1040 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1041 && zip->entry_bytes_remaining < 1) 1042 zip->end_of_entry = 1; 1043 1044 /* Set up a more descriptive format name. */ 1045 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1046 version / 10, version % 10, 1047 compression_name(zip->entry->compression)); 1048 a->archive.archive_format_name = zip->format_name.s; 1049 1050 return (ret); 1051} 1052 1053static int 1054check_authentication_code(struct archive_read *a, const void *_p) 1055{ 1056 struct zip *zip = (struct zip *)(a->format->data); 1057 1058 /* Check authentication code. */ 1059 if (zip->hctx_valid) { 1060 const void *p; 1061 uint8_t hmac[20]; 1062 size_t hmac_len = 20; 1063 int cmp; 1064 1065 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1066 if (_p == NULL) { 1067 /* Read authentication code. */ 1068 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1069 if (p == NULL) { 1070 archive_set_error(&a->archive, 1071 ARCHIVE_ERRNO_FILE_FORMAT, 1072 "Truncated ZIP file data"); 1073 return (ARCHIVE_FATAL); 1074 } 1075 } else { 1076 p = _p; 1077 } 1078 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1079 __archive_read_consume(a, AUTH_CODE_SIZE); 1080 if (cmp != 0) { 1081 archive_set_error(&a->archive, 1082 ARCHIVE_ERRNO_MISC, 1083 "ZIP bad Authentication code"); 1084 return (ARCHIVE_WARN); 1085 } 1086 } 1087 return (ARCHIVE_OK); 1088} 1089 1090/* 1091 * Read "uncompressed" data. There are three cases: 1092 * 1) We know the size of the data. This is always true for the 1093 * seeking reader (we've examined the Central Directory already). 1094 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1095 * Info-ZIP seems to do this; we know the size but have to grab 1096 * the CRC from the data descriptor afterwards. 1097 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1098 * we have no size information. In this case, we can do pretty 1099 * well by watching for the data descriptor record. The data 1100 * descriptor is 16 bytes and includes a computed CRC that should 1101 * provide a strong check. 1102 * 1103 * TODO: Technically, the PK\007\010 signature is optional. 1104 * In the original spec, the data descriptor contained CRC 1105 * and size fields but had no leading signature. In practice, 1106 * newer writers seem to provide the signature pretty consistently. 1107 * 1108 * For uncompressed data, the PK\007\010 marker seems essential 1109 * to be sure we've actually seen the end of the entry. 1110 * 1111 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1112 * zip->end_of_entry if it consumes all of the data. 1113 */ 1114static int 1115zip_read_data_none(struct archive_read *a, const void **_buff, 1116 size_t *size, int64_t *offset) 1117{ 1118 struct zip *zip; 1119 const char *buff; 1120 ssize_t bytes_avail; 1121 int r; 1122 1123 (void)offset; /* UNUSED */ 1124 1125 zip = (struct zip *)(a->format->data); 1126 1127 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1128 const char *p; 1129 ssize_t grabbing_bytes = 24; 1130 1131 if (zip->hctx_valid) 1132 grabbing_bytes += AUTH_CODE_SIZE; 1133 /* Grab at least 24 bytes. */ 1134 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1135 if (bytes_avail < grabbing_bytes) { 1136 /* Zip archives have end-of-archive markers 1137 that are longer than this, so a failure to get at 1138 least 24 bytes really does indicate a truncated 1139 file. */ 1140 archive_set_error(&a->archive, 1141 ARCHIVE_ERRNO_FILE_FORMAT, 1142 "Truncated ZIP file data"); 1143 return (ARCHIVE_FATAL); 1144 } 1145 /* Check for a complete PK\007\010 signature, followed 1146 * by the correct 4-byte CRC. */ 1147 p = buff; 1148 if (zip->hctx_valid) 1149 p += AUTH_CODE_SIZE; 1150 if (p[0] == 'P' && p[1] == 'K' 1151 && p[2] == '\007' && p[3] == '\010' 1152 && (archive_le32dec(p + 4) == zip->entry_crc32 1153 || zip->ignore_crc32 1154 || (zip->hctx_valid 1155 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1156 if (zip->entry->flags & LA_USED_ZIP64) { 1157 zip->entry->crc32 = archive_le32dec(p + 4); 1158 zip->entry->compressed_size = 1159 archive_le64dec(p + 8); 1160 zip->entry->uncompressed_size = 1161 archive_le64dec(p + 16); 1162 zip->unconsumed = 24; 1163 } else { 1164 zip->entry->crc32 = archive_le32dec(p + 4); 1165 zip->entry->compressed_size = 1166 archive_le32dec(p + 8); 1167 zip->entry->uncompressed_size = 1168 archive_le32dec(p + 12); 1169 zip->unconsumed = 16; 1170 } 1171 if (zip->hctx_valid) { 1172 r = check_authentication_code(a, buff); 1173 if (r != ARCHIVE_OK) 1174 return (r); 1175 } 1176 zip->end_of_entry = 1; 1177 return (ARCHIVE_OK); 1178 } 1179 /* If not at EOF, ensure we consume at least one byte. */ 1180 ++p; 1181 1182 /* Scan forward until we see where a PK\007\010 signature 1183 * might be. */ 1184 /* Return bytes up until that point. On the next call, 1185 * the code above will verify the data descriptor. */ 1186 while (p < buff + bytes_avail - 4) { 1187 if (p[3] == 'P') { p += 3; } 1188 else if (p[3] == 'K') { p += 2; } 1189 else if (p[3] == '\007') { p += 1; } 1190 else if (p[3] == '\010' && p[2] == '\007' 1191 && p[1] == 'K' && p[0] == 'P') { 1192 if (zip->hctx_valid) 1193 p -= AUTH_CODE_SIZE; 1194 break; 1195 } else { p += 4; } 1196 } 1197 bytes_avail = p - buff; 1198 } else { 1199 if (zip->entry_bytes_remaining == 0) { 1200 zip->end_of_entry = 1; 1201 if (zip->hctx_valid) { 1202 r = check_authentication_code(a, NULL); 1203 if (r != ARCHIVE_OK) 1204 return (r); 1205 } 1206 return (ARCHIVE_OK); 1207 } 1208 /* Grab a bunch of bytes. */ 1209 buff = __archive_read_ahead(a, 1, &bytes_avail); 1210 if (bytes_avail <= 0) { 1211 archive_set_error(&a->archive, 1212 ARCHIVE_ERRNO_FILE_FORMAT, 1213 "Truncated ZIP file data"); 1214 return (ARCHIVE_FATAL); 1215 } 1216 if (bytes_avail > zip->entry_bytes_remaining) 1217 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1218 } 1219 if (zip->tctx_valid || zip->cctx_valid) { 1220 size_t dec_size = bytes_avail; 1221 1222 if (dec_size > zip->decrypted_buffer_size) 1223 dec_size = zip->decrypted_buffer_size; 1224 if (zip->tctx_valid) { 1225 trad_enc_decrypt_update(&zip->tctx, 1226 (const uint8_t *)buff, dec_size, 1227 zip->decrypted_buffer, dec_size); 1228 } else { 1229 size_t dsize = dec_size; 1230 archive_hmac_sha1_update(&zip->hctx, 1231 (const uint8_t *)buff, dec_size); 1232 archive_decrypto_aes_ctr_update(&zip->cctx, 1233 (const uint8_t *)buff, dec_size, 1234 zip->decrypted_buffer, &dsize); 1235 } 1236 bytes_avail = dec_size; 1237 buff = (const char *)zip->decrypted_buffer; 1238 } 1239 *size = bytes_avail; 1240 zip->entry_bytes_remaining -= bytes_avail; 1241 zip->entry_uncompressed_bytes_read += bytes_avail; 1242 zip->entry_compressed_bytes_read += bytes_avail; 1243 zip->unconsumed += bytes_avail; 1244 *_buff = buff; 1245 return (ARCHIVE_OK); 1246} 1247 1248#ifdef HAVE_ZLIB_H 1249static int 1250zip_deflate_init(struct archive_read *a, struct zip *zip) 1251{ 1252 int r; 1253 1254 /* If we haven't yet read any data, initialize the decompressor. */ 1255 if (!zip->decompress_init) { 1256 if (zip->stream_valid) 1257 r = inflateReset(&zip->stream); 1258 else 1259 r = inflateInit2(&zip->stream, 1260 -15 /* Don't check for zlib header */); 1261 if (r != Z_OK) { 1262 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1263 "Can't initialize ZIP decompression."); 1264 return (ARCHIVE_FATAL); 1265 } 1266 /* Stream structure has been set up. */ 1267 zip->stream_valid = 1; 1268 /* We've initialized decompression for this stream. */ 1269 zip->decompress_init = 1; 1270 } 1271 return (ARCHIVE_OK); 1272} 1273 1274static int 1275zip_read_data_deflate(struct archive_read *a, const void **buff, 1276 size_t *size, int64_t *offset) 1277{ 1278 struct zip *zip; 1279 ssize_t bytes_avail; 1280 const void *compressed_buff, *sp; 1281 int r; 1282 1283 (void)offset; /* UNUSED */ 1284 1285 zip = (struct zip *)(a->format->data); 1286 1287 /* If the buffer hasn't been allocated, allocate it now. */ 1288 if (zip->uncompressed_buffer == NULL) { 1289 zip->uncompressed_buffer_size = 256 * 1024; 1290 zip->uncompressed_buffer 1291 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1292 if (zip->uncompressed_buffer == NULL) { 1293 archive_set_error(&a->archive, ENOMEM, 1294 "No memory for ZIP decompression"); 1295 return (ARCHIVE_FATAL); 1296 } 1297 } 1298 1299 r = zip_deflate_init(a, zip); 1300 if (r != ARCHIVE_OK) 1301 return (r); 1302 1303 /* 1304 * Note: '1' here is a performance optimization. 1305 * Recall that the decompression layer returns a count of 1306 * available bytes; asking for more than that forces the 1307 * decompressor to combine reads by copying data. 1308 */ 1309 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1310 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1311 && bytes_avail > zip->entry_bytes_remaining) { 1312 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1313 } 1314 if (bytes_avail < 0) { 1315 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1316 "Truncated ZIP file body"); 1317 return (ARCHIVE_FATAL); 1318 } 1319 1320 if (zip->tctx_valid || zip->cctx_valid) { 1321 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1322 size_t buff_remaining = 1323 (zip->decrypted_buffer + zip->decrypted_buffer_size) 1324 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); 1325 1326 if (buff_remaining > (size_t)bytes_avail) 1327 buff_remaining = (size_t)bytes_avail; 1328 1329 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1330 zip->entry_bytes_remaining > 0) { 1331 if ((int64_t)(zip->decrypted_bytes_remaining 1332 + buff_remaining) 1333 > zip->entry_bytes_remaining) { 1334 if (zip->entry_bytes_remaining < 1335 (int64_t)zip->decrypted_bytes_remaining) 1336 buff_remaining = 0; 1337 else 1338 buff_remaining = 1339 (size_t)zip->entry_bytes_remaining 1340 - zip->decrypted_bytes_remaining; 1341 } 1342 } 1343 if (buff_remaining > 0) { 1344 if (zip->tctx_valid) { 1345 trad_enc_decrypt_update(&zip->tctx, 1346 compressed_buff, buff_remaining, 1347 zip->decrypted_ptr 1348 + zip->decrypted_bytes_remaining, 1349 buff_remaining); 1350 } else { 1351 size_t dsize = buff_remaining; 1352 archive_decrypto_aes_ctr_update( 1353 &zip->cctx, 1354 compressed_buff, buff_remaining, 1355 zip->decrypted_ptr 1356 + zip->decrypted_bytes_remaining, 1357 &dsize); 1358 } 1359 zip->decrypted_bytes_remaining += buff_remaining; 1360 } 1361 } 1362 bytes_avail = zip->decrypted_bytes_remaining; 1363 compressed_buff = (const char *)zip->decrypted_ptr; 1364 } 1365 1366 /* 1367 * A bug in zlib.h: stream.next_in should be marked 'const' 1368 * but isn't (the library never alters data through the 1369 * next_in pointer, only reads it). The result: this ugly 1370 * cast to remove 'const'. 1371 */ 1372 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1373 zip->stream.avail_in = (uInt)bytes_avail; 1374 zip->stream.total_in = 0; 1375 zip->stream.next_out = zip->uncompressed_buffer; 1376 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1377 zip->stream.total_out = 0; 1378 1379 r = inflate(&zip->stream, 0); 1380 switch (r) { 1381 case Z_OK: 1382 break; 1383 case Z_STREAM_END: 1384 zip->end_of_entry = 1; 1385 break; 1386 case Z_MEM_ERROR: 1387 archive_set_error(&a->archive, ENOMEM, 1388 "Out of memory for ZIP decompression"); 1389 return (ARCHIVE_FATAL); 1390 default: 1391 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1392 "ZIP decompression failed (%d)", r); 1393 return (ARCHIVE_FATAL); 1394 } 1395 1396 /* Consume as much as the compressor actually used. */ 1397 bytes_avail = zip->stream.total_in; 1398 if (zip->tctx_valid || zip->cctx_valid) { 1399 zip->decrypted_bytes_remaining -= bytes_avail; 1400 if (zip->decrypted_bytes_remaining == 0) 1401 zip->decrypted_ptr = zip->decrypted_buffer; 1402 else 1403 zip->decrypted_ptr += bytes_avail; 1404 } 1405 /* Calculate compressed data as much as we used.*/ 1406 if (zip->hctx_valid) 1407 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1408 __archive_read_consume(a, bytes_avail); 1409 zip->entry_bytes_remaining -= bytes_avail; 1410 zip->entry_compressed_bytes_read += bytes_avail; 1411 1412 *size = zip->stream.total_out; 1413 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1414 *buff = zip->uncompressed_buffer; 1415 1416 if (zip->end_of_entry && zip->hctx_valid) { 1417 r = check_authentication_code(a, NULL); 1418 if (r != ARCHIVE_OK) 1419 return (r); 1420 } 1421 1422 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1423 const char *p; 1424 1425 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1426 archive_set_error(&a->archive, 1427 ARCHIVE_ERRNO_FILE_FORMAT, 1428 "Truncated ZIP end-of-file record"); 1429 return (ARCHIVE_FATAL); 1430 } 1431 /* Consume the optional PK\007\010 marker. */ 1432 if (p[0] == 'P' && p[1] == 'K' && 1433 p[2] == '\007' && p[3] == '\010') { 1434 p += 4; 1435 zip->unconsumed = 4; 1436 } 1437 if (zip->entry->flags & LA_USED_ZIP64) { 1438 zip->entry->crc32 = archive_le32dec(p); 1439 zip->entry->compressed_size = archive_le64dec(p + 4); 1440 zip->entry->uncompressed_size = archive_le64dec(p + 12); 1441 zip->unconsumed += 20; 1442 } else { 1443 zip->entry->crc32 = archive_le32dec(p); 1444 zip->entry->compressed_size = archive_le32dec(p + 4); 1445 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1446 zip->unconsumed += 12; 1447 } 1448 } 1449 1450 return (ARCHIVE_OK); 1451} 1452#endif 1453 1454static int 1455read_decryption_header(struct archive_read *a) 1456{ 1457 struct zip *zip = (struct zip *)(a->format->data); 1458 const char *p; 1459 unsigned int remaining_size; 1460 unsigned int ts; 1461 1462 /* 1463 * Read an initialization vector data field. 1464 */ 1465 p = __archive_read_ahead(a, 2, NULL); 1466 if (p == NULL) 1467 goto truncated; 1468 ts = zip->iv_size; 1469 zip->iv_size = archive_le16dec(p); 1470 __archive_read_consume(a, 2); 1471 if (ts < zip->iv_size) { 1472 free(zip->iv); 1473 zip->iv = NULL; 1474 } 1475 p = __archive_read_ahead(a, zip->iv_size, NULL); 1476 if (p == NULL) 1477 goto truncated; 1478 if (zip->iv == NULL) { 1479 zip->iv = malloc(zip->iv_size); 1480 if (zip->iv == NULL) 1481 goto nomem; 1482 } 1483 memcpy(zip->iv, p, zip->iv_size); 1484 __archive_read_consume(a, zip->iv_size); 1485 1486 /* 1487 * Read a size of remaining decryption header field. 1488 */ 1489 p = __archive_read_ahead(a, 14, NULL); 1490 if (p == NULL) 1491 goto truncated; 1492 remaining_size = archive_le32dec(p); 1493 if (remaining_size < 16 || remaining_size > (1 << 18)) 1494 goto corrupted; 1495 1496 /* Check if format version is supported. */ 1497 if (archive_le16dec(p+4) != 3) { 1498 archive_set_error(&a->archive, 1499 ARCHIVE_ERRNO_FILE_FORMAT, 1500 "Unsupported encryption format version: %u", 1501 archive_le16dec(p+4)); 1502 return (ARCHIVE_FAILED); 1503 } 1504 1505 /* 1506 * Read an encryption algorithm field. 1507 */ 1508 zip->alg_id = archive_le16dec(p+6); 1509 switch (zip->alg_id) { 1510 case 0x6601:/* DES */ 1511 case 0x6602:/* RC2 */ 1512 case 0x6603:/* 3DES 168 */ 1513 case 0x6609:/* 3DES 112 */ 1514 case 0x660E:/* AES 128 */ 1515 case 0x660F:/* AES 192 */ 1516 case 0x6610:/* AES 256 */ 1517 case 0x6702:/* RC2 (version >= 5.2) */ 1518 case 0x6720:/* Blowfish */ 1519 case 0x6721:/* Twofish */ 1520 case 0x6801:/* RC4 */ 1521 /* Suuported encryption algorithm. */ 1522 break; 1523 default: 1524 archive_set_error(&a->archive, 1525 ARCHIVE_ERRNO_FILE_FORMAT, 1526 "Unknown encryption algorithm: %u", zip->alg_id); 1527 return (ARCHIVE_FAILED); 1528 } 1529 1530 /* 1531 * Read a bit length field. 1532 */ 1533 zip->bit_len = archive_le16dec(p+8); 1534 1535 /* 1536 * Read a flags field. 1537 */ 1538 zip->flags = archive_le16dec(p+10); 1539 switch (zip->flags & 0xf000) { 1540 case 0x0001: /* Password is required to decrypt. */ 1541 case 0x0002: /* Certificates only. */ 1542 case 0x0003: /* Password or certificate required to decrypt. */ 1543 break; 1544 default: 1545 archive_set_error(&a->archive, 1546 ARCHIVE_ERRNO_FILE_FORMAT, 1547 "Unknown encryption flag: %u", zip->flags); 1548 return (ARCHIVE_FAILED); 1549 } 1550 if ((zip->flags & 0xf000) == 0 || 1551 (zip->flags & 0xf000) == 0x4000) { 1552 archive_set_error(&a->archive, 1553 ARCHIVE_ERRNO_FILE_FORMAT, 1554 "Unknown encryption flag: %u", zip->flags); 1555 return (ARCHIVE_FAILED); 1556 } 1557 1558 /* 1559 * Read an encrypted random data field. 1560 */ 1561 ts = zip->erd_size; 1562 zip->erd_size = archive_le16dec(p+12); 1563 __archive_read_consume(a, 14); 1564 if ((zip->erd_size & 0xf) != 0 || 1565 (zip->erd_size + 16) > remaining_size || 1566 (zip->erd_size + 16) < zip->erd_size) 1567 goto corrupted; 1568 1569 if (ts < zip->erd_size) { 1570 free(zip->erd); 1571 zip->erd = NULL; 1572 } 1573 p = __archive_read_ahead(a, zip->erd_size, NULL); 1574 if (p == NULL) 1575 goto truncated; 1576 if (zip->erd == NULL) { 1577 zip->erd = malloc(zip->erd_size); 1578 if (zip->erd == NULL) 1579 goto nomem; 1580 } 1581 memcpy(zip->erd, p, zip->erd_size); 1582 __archive_read_consume(a, zip->erd_size); 1583 1584 /* 1585 * Read a reserved data field. 1586 */ 1587 p = __archive_read_ahead(a, 4, NULL); 1588 if (p == NULL) 1589 goto truncated; 1590 /* Reserved data size should be zero. */ 1591 if (archive_le32dec(p) != 0) 1592 goto corrupted; 1593 __archive_read_consume(a, 4); 1594 1595 /* 1596 * Read a password validation data field. 1597 */ 1598 p = __archive_read_ahead(a, 2, NULL); 1599 if (p == NULL) 1600 goto truncated; 1601 ts = zip->v_size; 1602 zip->v_size = archive_le16dec(p); 1603 __archive_read_consume(a, 2); 1604 if ((zip->v_size & 0x0f) != 0 || 1605 (zip->erd_size + zip->v_size + 16) > remaining_size || 1606 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1607 goto corrupted; 1608 if (ts < zip->v_size) { 1609 free(zip->v_data); 1610 zip->v_data = NULL; 1611 } 1612 p = __archive_read_ahead(a, zip->v_size, NULL); 1613 if (p == NULL) 1614 goto truncated; 1615 if (zip->v_data == NULL) { 1616 zip->v_data = malloc(zip->v_size); 1617 if (zip->v_data == NULL) 1618 goto nomem; 1619 } 1620 memcpy(zip->v_data, p, zip->v_size); 1621 __archive_read_consume(a, zip->v_size); 1622 1623 p = __archive_read_ahead(a, 4, NULL); 1624 if (p == NULL) 1625 goto truncated; 1626 zip->v_crc32 = archive_le32dec(p); 1627 __archive_read_consume(a, 4); 1628 1629 /*return (ARCHIVE_OK); 1630 * This is not fully implemnted yet.*/ 1631 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1632 "Encrypted file is unsupported"); 1633 return (ARCHIVE_FAILED); 1634truncated: 1635 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1636 "Truncated ZIP file data"); 1637 return (ARCHIVE_FATAL); 1638corrupted: 1639 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1640 "Corrupted ZIP file data"); 1641 return (ARCHIVE_FATAL); 1642nomem: 1643 archive_set_error(&a->archive, ENOMEM, 1644 "No memory for ZIP decryption"); 1645 return (ARCHIVE_FATAL); 1646} 1647 1648static int 1649zip_alloc_decryption_buffer(struct archive_read *a) 1650{ 1651 struct zip *zip = (struct zip *)(a->format->data); 1652 size_t bs = 256 * 1024; 1653 1654 if (zip->decrypted_buffer == NULL) { 1655 zip->decrypted_buffer_size = bs; 1656 zip->decrypted_buffer = malloc(bs); 1657 if (zip->decrypted_buffer == NULL) { 1658 archive_set_error(&a->archive, ENOMEM, 1659 "No memory for ZIP decryption"); 1660 return (ARCHIVE_FATAL); 1661 } 1662 } 1663 zip->decrypted_ptr = zip->decrypted_buffer; 1664 return (ARCHIVE_OK); 1665} 1666 1667static int 1668init_traditional_PKWARE_decryption(struct archive_read *a) 1669{ 1670 struct zip *zip = (struct zip *)(a->format->data); 1671 const void *p; 1672 int retry; 1673 int r; 1674 1675 if (zip->tctx_valid) 1676 return (ARCHIVE_OK); 1677 1678 /* 1679 Read the 12 bytes encryption header stored at 1680 the start of the data area. 1681 */ 1682#define ENC_HEADER_SIZE 12 1683 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1684 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1685 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1686 "Truncated Zip encrypted body: only %jd bytes available", 1687 (intmax_t)zip->entry_bytes_remaining); 1688 return (ARCHIVE_FATAL); 1689 } 1690 1691 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1692 if (p == NULL) { 1693 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1694 "Truncated ZIP file data"); 1695 return (ARCHIVE_FATAL); 1696 } 1697 1698 for (retry = 0;; retry++) { 1699 const char *passphrase; 1700 uint8_t crcchk; 1701 1702 passphrase = __archive_read_next_passphrase(a); 1703 if (passphrase == NULL) { 1704 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1705 (retry > 0)? 1706 "Incorrect passphrase": 1707 "Passphrase required for this entry"); 1708 return (ARCHIVE_FAILED); 1709 } 1710 1711 /* 1712 * Initialize ctx for Traditional PKWARE Decyption. 1713 */ 1714 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1715 p, ENC_HEADER_SIZE, &crcchk); 1716 if (r == 0 && crcchk == zip->entry->decdat) 1717 break;/* The passphrase is OK. */ 1718 if (retry > 10000) { 1719 /* Avoid infinity loop. */ 1720 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1721 "Too many incorrect passphrases"); 1722 return (ARCHIVE_FAILED); 1723 } 1724 } 1725 1726 __archive_read_consume(a, ENC_HEADER_SIZE); 1727 zip->tctx_valid = 1; 1728 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1729 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1730 } 1731 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1732 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1733 zip->decrypted_bytes_remaining = 0; 1734 1735 return (zip_alloc_decryption_buffer(a)); 1736#undef ENC_HEADER_SIZE 1737} 1738 1739static int 1740init_WinZip_AES_decryption(struct archive_read *a) 1741{ 1742 struct zip *zip = (struct zip *)(a->format->data); 1743 const void *p; 1744 const uint8_t *pv; 1745 size_t key_len, salt_len; 1746 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1747 int retry; 1748 int r; 1749 1750 if (zip->cctx_valid || zip->hctx_valid) 1751 return (ARCHIVE_OK); 1752 1753 switch (zip->entry->aes_extra.strength) { 1754 case 1: salt_len = 8; key_len = 16; break; 1755 case 2: salt_len = 12; key_len = 24; break; 1756 case 3: salt_len = 16; key_len = 32; break; 1757 default: goto corrupted; 1758 } 1759 p = __archive_read_ahead(a, salt_len + 2, NULL); 1760 if (p == NULL) 1761 goto truncated; 1762 1763 for (retry = 0;; retry++) { 1764 const char *passphrase; 1765 1766 passphrase = __archive_read_next_passphrase(a); 1767 if (passphrase == NULL) { 1768 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1769 (retry > 0)? 1770 "Incorrect passphrase": 1771 "Passphrase required for this entry"); 1772 return (ARCHIVE_FAILED); 1773 } 1774 memset(derived_key, 0, sizeof(derived_key)); 1775 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1776 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1777 if (r != 0) { 1778 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1779 "Decryption is unsupported due to lack of " 1780 "crypto library"); 1781 return (ARCHIVE_FAILED); 1782 } 1783 1784 /* Check password verification value. */ 1785 pv = ((const uint8_t *)p) + salt_len; 1786 if (derived_key[key_len * 2] == pv[0] && 1787 derived_key[key_len * 2 + 1] == pv[1]) 1788 break;/* The passphrase is OK. */ 1789 if (retry > 10000) { 1790 /* Avoid infinity loop. */ 1791 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1792 "Too many incorrect passphrases"); 1793 return (ARCHIVE_FAILED); 1794 } 1795 } 1796 1797 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1798 if (r != 0) { 1799 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1800 "Decryption is unsupported due to lack of crypto library"); 1801 return (ARCHIVE_FAILED); 1802 } 1803 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1804 if (r != 0) { 1805 archive_decrypto_aes_ctr_release(&zip->cctx); 1806 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1807 "Failed to initialize HMAC-SHA1"); 1808 return (ARCHIVE_FAILED); 1809 } 1810 zip->cctx_valid = zip->hctx_valid = 1; 1811 __archive_read_consume(a, salt_len + 2); 1812 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1813 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1814 && zip->entry_bytes_remaining < 0) 1815 goto corrupted; 1816 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1817 zip->decrypted_bytes_remaining = 0; 1818 1819 zip->entry->compression = zip->entry->aes_extra.compression; 1820 return (zip_alloc_decryption_buffer(a)); 1821 1822truncated: 1823 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1824 "Truncated ZIP file data"); 1825 return (ARCHIVE_FATAL); 1826corrupted: 1827 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1828 "Corrupted ZIP file data"); 1829 return (ARCHIVE_FATAL); 1830} 1831 1832static int 1833archive_read_format_zip_read_data(struct archive_read *a, 1834 const void **buff, size_t *size, int64_t *offset) 1835{ 1836 int r; 1837 struct zip *zip = (struct zip *)(a->format->data); 1838 1839 if (zip->has_encrypted_entries == 1840 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1841 zip->has_encrypted_entries = 0; 1842 } 1843 1844 *offset = zip->entry_uncompressed_bytes_read; 1845 *size = 0; 1846 *buff = NULL; 1847 1848 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1849 if (zip->end_of_entry) 1850 return (ARCHIVE_EOF); 1851 1852 /* Return EOF immediately if this is a non-regular file. */ 1853 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1854 return (ARCHIVE_EOF); 1855 1856 __archive_read_consume(a, zip->unconsumed); 1857 zip->unconsumed = 0; 1858 1859 if (zip->init_decryption) { 1860 zip->has_encrypted_entries = 1; 1861 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1862 r = read_decryption_header(a); 1863 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1864 r = init_WinZip_AES_decryption(a); 1865 else 1866 r = init_traditional_PKWARE_decryption(a); 1867 if (r != ARCHIVE_OK) 1868 return (r); 1869 zip->init_decryption = 0; 1870 } 1871 1872 switch(zip->entry->compression) { 1873 case 0: /* No compression. */ 1874 r = zip_read_data_none(a, buff, size, offset); 1875 break; 1876#ifdef HAVE_ZLIB_H 1877 case 8: /* Deflate compression. */ 1878 r = zip_read_data_deflate(a, buff, size, offset); 1879 break; 1880#endif 1881 default: /* Unsupported compression. */ 1882 /* Return a warning. */ 1883 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1884 "Unsupported ZIP compression method (%s)", 1885 compression_name(zip->entry->compression)); 1886 /* We can't decompress this entry, but we will 1887 * be able to skip() it and try the next entry. */ 1888 return (ARCHIVE_FAILED); 1889 break; 1890 } 1891 if (r != ARCHIVE_OK) 1892 return (r); 1893 /* Update checksum */ 1894 if (*size) 1895 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1896 (unsigned)*size); 1897 /* If we hit the end, swallow any end-of-data marker. */ 1898 if (zip->end_of_entry) { 1899 /* Check file size, CRC against these values. */ 1900 if (zip->entry->compressed_size != 1901 zip->entry_compressed_bytes_read) { 1902 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1903 "ZIP compressed data is wrong size " 1904 "(read %jd, expected %jd)", 1905 (intmax_t)zip->entry_compressed_bytes_read, 1906 (intmax_t)zip->entry->compressed_size); 1907 return (ARCHIVE_WARN); 1908 } 1909 /* Size field only stores the lower 32 bits of the actual 1910 * size. */ 1911 if ((zip->entry->uncompressed_size & UINT32_MAX) 1912 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1913 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1914 "ZIP uncompressed data is wrong size " 1915 "(read %jd, expected %jd)\n", 1916 (intmax_t)zip->entry_uncompressed_bytes_read, 1917 (intmax_t)zip->entry->uncompressed_size); 1918 return (ARCHIVE_WARN); 1919 } 1920 /* Check computed CRC against header */ 1921 if ((!zip->hctx_valid || 1922 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1923 zip->entry->crc32 != zip->entry_crc32 1924 && !zip->ignore_crc32) { 1925 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1926 "ZIP bad CRC: 0x%lx should be 0x%lx", 1927 (unsigned long)zip->entry_crc32, 1928 (unsigned long)zip->entry->crc32); 1929 return (ARCHIVE_WARN); 1930 } 1931 } 1932 1933 return (ARCHIVE_OK); 1934} 1935 1936static int 1937archive_read_format_zip_cleanup(struct archive_read *a) 1938{ 1939 struct zip *zip; 1940 struct zip_entry *zip_entry, *next_zip_entry; 1941 1942 zip = (struct zip *)(a->format->data); 1943#ifdef HAVE_ZLIB_H 1944 if (zip->stream_valid) 1945 inflateEnd(&zip->stream); 1946 free(zip->uncompressed_buffer); 1947#endif 1948 if (zip->zip_entries) { 1949 zip_entry = zip->zip_entries; 1950 while (zip_entry != NULL) { 1951 next_zip_entry = zip_entry->next; 1952 archive_string_free(&zip_entry->rsrcname); 1953 free(zip_entry); 1954 zip_entry = next_zip_entry; 1955 } 1956 } 1957 free(zip->decrypted_buffer); 1958 if (zip->cctx_valid) 1959 archive_decrypto_aes_ctr_release(&zip->cctx); 1960 if (zip->hctx_valid) 1961 archive_hmac_sha1_cleanup(&zip->hctx); 1962 free(zip->iv); 1963 free(zip->erd); 1964 free(zip->v_data); 1965 archive_string_free(&zip->format_name); 1966 free(zip); 1967 (a->format->data) = NULL; 1968 return (ARCHIVE_OK); 1969} 1970 1971static int 1972archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 1973{ 1974 if (_a && _a->format) { 1975 struct zip * zip = (struct zip *)_a->format->data; 1976 if (zip) { 1977 return zip->has_encrypted_entries; 1978 } 1979 } 1980 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 1981} 1982 1983static int 1984archive_read_format_zip_options(struct archive_read *a, 1985 const char *key, const char *val) 1986{ 1987 struct zip *zip; 1988 int ret = ARCHIVE_FAILED; 1989 1990 zip = (struct zip *)(a->format->data); 1991 if (strcmp(key, "compat-2x") == 0) { 1992 /* Handle filenames as libarchive 2.x */ 1993 zip->init_default_conversion = (val != NULL) ? 1 : 0; 1994 return (ARCHIVE_OK); 1995 } else if (strcmp(key, "hdrcharset") == 0) { 1996 if (val == NULL || val[0] == 0) 1997 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1998 "zip: hdrcharset option needs a character-set name" 1999 ); 2000 else { 2001 zip->sconv = archive_string_conversion_from_charset( 2002 &a->archive, val, 0); 2003 if (zip->sconv != NULL) { 2004 if (strcmp(val, "UTF-8") == 0) 2005 zip->sconv_utf8 = zip->sconv; 2006 ret = ARCHIVE_OK; 2007 } else 2008 ret = ARCHIVE_FATAL; 2009 } 2010 return (ret); 2011 } else if (strcmp(key, "ignorecrc32") == 0) { 2012 /* Mostly useful for testing. */ 2013 if (val == NULL || val[0] == 0) { 2014 zip->crc32func = real_crc32; 2015 zip->ignore_crc32 = 0; 2016 } else { 2017 zip->crc32func = fake_crc32; 2018 zip->ignore_crc32 = 1; 2019 } 2020 return (ARCHIVE_OK); 2021 } else if (strcmp(key, "mac-ext") == 0) { 2022 zip->process_mac_extensions = (val != NULL && val[0] != 0); 2023 return (ARCHIVE_OK); 2024 } 2025 2026 /* Note: The "warn" return is just to inform the options 2027 * supervisor that we didn't handle it. It will generate 2028 * a suitable error if no one used this option. */ 2029 return (ARCHIVE_WARN); 2030} 2031 2032int 2033archive_read_support_format_zip(struct archive *a) 2034{ 2035 int r; 2036 r = archive_read_support_format_zip_streamable(a); 2037 if (r != ARCHIVE_OK) 2038 return r; 2039 return (archive_read_support_format_zip_seekable(a)); 2040} 2041 2042/* ------------------------------------------------------------------------ */ 2043 2044/* 2045 * Streaming-mode support 2046 */ 2047 2048 2049static int 2050archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2051{ 2052 (void)a; /* UNUSED */ 2053 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2054 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2055} 2056 2057static int 2058archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2059{ 2060 const char *p; 2061 2062 (void)best_bid; /* UNUSED */ 2063 2064 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2065 return (-1); 2066 2067 /* 2068 * Bid of 29 here comes from: 2069 * + 16 bits for "PK", 2070 * + next 16-bit field has 6 options so contributes 2071 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2072 * 2073 * So we've effectively verified ~29 total bits of check data. 2074 */ 2075 if (p[0] == 'P' && p[1] == 'K') { 2076 if ((p[2] == '\001' && p[3] == '\002') 2077 || (p[2] == '\003' && p[3] == '\004') 2078 || (p[2] == '\005' && p[3] == '\006') 2079 || (p[2] == '\006' && p[3] == '\006') 2080 || (p[2] == '\007' && p[3] == '\010') 2081 || (p[2] == '0' && p[3] == '0')) 2082 return (29); 2083 } 2084 2085 /* TODO: It's worth looking ahead a little bit for a valid 2086 * PK signature. In particular, that would make it possible 2087 * to read some UUEncoded SFX files or SFX files coming from 2088 * a network socket. */ 2089 2090 return (0); 2091} 2092 2093static int 2094archive_read_format_zip_streamable_read_header(struct archive_read *a, 2095 struct archive_entry *entry) 2096{ 2097 struct zip *zip; 2098 2099 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2100 if (a->archive.archive_format_name == NULL) 2101 a->archive.archive_format_name = "ZIP"; 2102 2103 zip = (struct zip *)(a->format->data); 2104 2105 /* 2106 * It should be sufficient to call archive_read_next_header() for 2107 * a reader to determine if an entry is encrypted or not. If the 2108 * encryption of an entry is only detectable when calling 2109 * archive_read_data(), so be it. We'll do the same check there 2110 * as well. 2111 */ 2112 if (zip->has_encrypted_entries == 2113 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2114 zip->has_encrypted_entries = 0; 2115 2116 /* Make sure we have a zip_entry structure to use. */ 2117 if (zip->zip_entries == NULL) { 2118 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2119 if (zip->zip_entries == NULL) { 2120 archive_set_error(&a->archive, ENOMEM, 2121 "Out of memory"); 2122 return ARCHIVE_FATAL; 2123 } 2124 } 2125 zip->entry = zip->zip_entries; 2126 memset(zip->entry, 0, sizeof(struct zip_entry)); 2127 2128 if (zip->cctx_valid) 2129 archive_decrypto_aes_ctr_release(&zip->cctx); 2130 if (zip->hctx_valid) 2131 archive_hmac_sha1_cleanup(&zip->hctx); 2132 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2133 __archive_read_reset_passphrase(a); 2134 2135 /* Search ahead for the next local file header. */ 2136 __archive_read_consume(a, zip->unconsumed); 2137 zip->unconsumed = 0; 2138 for (;;) { 2139 int64_t skipped = 0; 2140 const char *p, *end; 2141 ssize_t bytes; 2142 2143 p = __archive_read_ahead(a, 4, &bytes); 2144 if (p == NULL) 2145 return (ARCHIVE_FATAL); 2146 end = p + bytes; 2147 2148 while (p + 4 <= end) { 2149 if (p[0] == 'P' && p[1] == 'K') { 2150 if (p[2] == '\003' && p[3] == '\004') { 2151 /* Regular file entry. */ 2152 __archive_read_consume(a, skipped); 2153 return zip_read_local_file_header(a, 2154 entry, zip); 2155 } 2156 2157 /* 2158 * TODO: We cannot restore permissions 2159 * based only on the local file headers. 2160 * Consider scanning the central 2161 * directory and returning additional 2162 * entries for at least directories. 2163 * This would allow us to properly set 2164 * directory permissions. 2165 * 2166 * This won't help us fix symlinks 2167 * and may not help with regular file 2168 * permissions, either. <sigh> 2169 */ 2170 if (p[2] == '\001' && p[3] == '\002') { 2171 return (ARCHIVE_EOF); 2172 } 2173 2174 /* End of central directory? Must be an 2175 * empty archive. */ 2176 if ((p[2] == '\005' && p[3] == '\006') 2177 || (p[2] == '\006' && p[3] == '\006')) 2178 return (ARCHIVE_EOF); 2179 } 2180 ++p; 2181 ++skipped; 2182 } 2183 __archive_read_consume(a, skipped); 2184 } 2185} 2186 2187static int 2188archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2189{ 2190 struct zip *zip; 2191 int64_t bytes_skipped; 2192 2193 zip = (struct zip *)(a->format->data); 2194 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2195 zip->unconsumed = 0; 2196 if (bytes_skipped < 0) 2197 return (ARCHIVE_FATAL); 2198 2199 /* If we've already read to end of data, we're done. */ 2200 if (zip->end_of_entry) 2201 return (ARCHIVE_OK); 2202 2203 /* So we know we're streaming... */ 2204 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2205 || zip->entry->compressed_size > 0) { 2206 /* We know the compressed length, so we can just skip. */ 2207 bytes_skipped = __archive_read_consume(a, 2208 zip->entry_bytes_remaining); 2209 if (bytes_skipped < 0) 2210 return (ARCHIVE_FATAL); 2211 return (ARCHIVE_OK); 2212 } 2213 2214 if (zip->init_decryption) { 2215 int r; 2216 2217 zip->has_encrypted_entries = 1; 2218 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2219 r = read_decryption_header(a); 2220 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2221 r = init_WinZip_AES_decryption(a); 2222 else 2223 r = init_traditional_PKWARE_decryption(a); 2224 if (r != ARCHIVE_OK) 2225 return (r); 2226 zip->init_decryption = 0; 2227 } 2228 2229 /* We're streaming and we don't know the length. */ 2230 /* If the body is compressed and we know the format, we can 2231 * find an exact end-of-entry by decompressing it. */ 2232 switch (zip->entry->compression) { 2233#ifdef HAVE_ZLIB_H 2234 case 8: /* Deflate compression. */ 2235 while (!zip->end_of_entry) { 2236 int64_t offset = 0; 2237 const void *buff = NULL; 2238 size_t size = 0; 2239 int r; 2240 r = zip_read_data_deflate(a, &buff, &size, &offset); 2241 if (r != ARCHIVE_OK) 2242 return (r); 2243 } 2244 return ARCHIVE_OK; 2245#endif 2246 default: /* Uncompressed or unknown. */ 2247 /* Scan for a PK\007\010 signature. */ 2248 for (;;) { 2249 const char *p, *buff; 2250 ssize_t bytes_avail; 2251 buff = __archive_read_ahead(a, 16, &bytes_avail); 2252 if (bytes_avail < 16) { 2253 archive_set_error(&a->archive, 2254 ARCHIVE_ERRNO_FILE_FORMAT, 2255 "Truncated ZIP file data"); 2256 return (ARCHIVE_FATAL); 2257 } 2258 p = buff; 2259 while (p <= buff + bytes_avail - 16) { 2260 if (p[3] == 'P') { p += 3; } 2261 else if (p[3] == 'K') { p += 2; } 2262 else if (p[3] == '\007') { p += 1; } 2263 else if (p[3] == '\010' && p[2] == '\007' 2264 && p[1] == 'K' && p[0] == 'P') { 2265 if (zip->entry->flags & LA_USED_ZIP64) 2266 __archive_read_consume(a, 2267 p - buff + 24); 2268 else 2269 __archive_read_consume(a, 2270 p - buff + 16); 2271 return ARCHIVE_OK; 2272 } else { p += 4; } 2273 } 2274 __archive_read_consume(a, p - buff); 2275 } 2276 } 2277} 2278 2279int 2280archive_read_support_format_zip_streamable(struct archive *_a) 2281{ 2282 struct archive_read *a = (struct archive_read *)_a; 2283 struct zip *zip; 2284 int r; 2285 2286 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2287 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2288 2289 zip = (struct zip *)calloc(1, sizeof(*zip)); 2290 if (zip == NULL) { 2291 archive_set_error(&a->archive, ENOMEM, 2292 "Can't allocate zip data"); 2293 return (ARCHIVE_FATAL); 2294 } 2295 2296 /* Streamable reader doesn't support mac extensions. */ 2297 zip->process_mac_extensions = 0; 2298 2299 /* 2300 * Until enough data has been read, we cannot tell about 2301 * any encrypted entries yet. 2302 */ 2303 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2304 zip->crc32func = real_crc32; 2305 2306 r = __archive_read_register_format(a, 2307 zip, 2308 "zip", 2309 archive_read_format_zip_streamable_bid, 2310 archive_read_format_zip_options, 2311 archive_read_format_zip_streamable_read_header, 2312 archive_read_format_zip_read_data, 2313 archive_read_format_zip_read_data_skip_streamable, 2314 NULL, 2315 archive_read_format_zip_cleanup, 2316 archive_read_support_format_zip_capabilities_streamable, 2317 archive_read_format_zip_has_encrypted_entries); 2318 2319 if (r != ARCHIVE_OK) 2320 free(zip); 2321 return (ARCHIVE_OK); 2322} 2323 2324/* ------------------------------------------------------------------------ */ 2325 2326/* 2327 * Seeking-mode support 2328 */ 2329 2330static int 2331archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2332{ 2333 (void)a; /* UNUSED */ 2334 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2335 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2336} 2337 2338/* 2339 * TODO: This is a performance sink because it forces the read core to 2340 * drop buffered data from the start of file, which will then have to 2341 * be re-read again if this bidder loses. 2342 * 2343 * We workaround this a little by passing in the best bid so far so 2344 * that later bidders can do nothing if they know they'll never 2345 * outbid. But we can certainly do better... 2346 */ 2347static int 2348read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2349{ 2350 /* Sanity-check the EOCD we've found. */ 2351 2352 /* This must be the first volume. */ 2353 if (archive_le16dec(p + 4) != 0) 2354 return 0; 2355 /* Central directory must be on this volume. */ 2356 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2357 return 0; 2358 /* All central directory entries must be on this volume. */ 2359 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2360 return 0; 2361 /* Central directory can't extend beyond start of EOCD record. */ 2362 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2363 > current_offset) 2364 return 0; 2365 2366 /* Save the central directory location for later use. */ 2367 zip->central_directory_offset = archive_le32dec(p + 16); 2368 2369 /* This is just a tiny bit higher than the maximum 2370 returned by the streaming Zip bidder. This ensures 2371 that the more accurate seeking Zip parser wins 2372 whenever seek is available. */ 2373 return 32; 2374} 2375 2376/* 2377 * Examine Zip64 EOCD locator: If it's valid, store the information 2378 * from it. 2379 */ 2380static void 2381read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2382{ 2383 int64_t eocd64_offset; 2384 int64_t eocd64_size; 2385 2386 /* Sanity-check the locator record. */ 2387 2388 /* Central dir must be on first volume. */ 2389 if (archive_le32dec(p + 4) != 0) 2390 return; 2391 /* Must be only a single volume. */ 2392 if (archive_le32dec(p + 16) != 1) 2393 return; 2394 2395 /* Find the Zip64 EOCD record. */ 2396 eocd64_offset = archive_le64dec(p + 8); 2397 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2398 return; 2399 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2400 return; 2401 /* Make sure we can read all of it. */ 2402 eocd64_size = archive_le64dec(p + 4) + 12; 2403 if (eocd64_size < 56 || eocd64_size > 16384) 2404 return; 2405 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2406 return; 2407 2408 /* Sanity-check the EOCD64 */ 2409 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2410 return; 2411 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2412 return; 2413 /* CD can't be split. */ 2414 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2415 return; 2416 2417 /* Save the central directory offset for later use. */ 2418 zip->central_directory_offset = archive_le64dec(p + 48); 2419} 2420 2421static int 2422archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2423{ 2424 struct zip *zip = (struct zip *)a->format->data; 2425 int64_t file_size, current_offset; 2426 const char *p; 2427 int i, tail; 2428 2429 /* If someone has already bid more than 32, then avoid 2430 trashing the look-ahead buffers with a seek. */ 2431 if (best_bid > 32) 2432 return (-1); 2433 2434 file_size = __archive_read_seek(a, 0, SEEK_END); 2435 if (file_size <= 0) 2436 return 0; 2437 2438 /* Search last 16k of file for end-of-central-directory 2439 * record (which starts with PK\005\006) */ 2440 tail = (int)zipmin(1024 * 16, file_size); 2441 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2442 if (current_offset < 0) 2443 return 0; 2444 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2445 return 0; 2446 /* Boyer-Moore search backwards from the end, since we want 2447 * to match the last EOCD in the file (there can be more than 2448 * one if there is an uncompressed Zip archive as a member 2449 * within this Zip archive). */ 2450 for (i = tail - 22; i > 0;) { 2451 switch (p[i]) { 2452 case 'P': 2453 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2454 int ret = read_eocd(zip, p + i, 2455 current_offset + i); 2456 if (ret > 0) { 2457 /* Zip64 EOCD locator precedes 2458 * regular EOCD if present. */ 2459 if (i >= 20 2460 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2461 read_zip64_eocd(a, zip, p + i - 20); 2462 } 2463 return (ret); 2464 } 2465 } 2466 i -= 4; 2467 break; 2468 case 'K': i -= 1; break; 2469 case 005: i -= 2; break; 2470 case 006: i -= 3; break; 2471 default: i -= 4; break; 2472 } 2473 } 2474 return 0; 2475} 2476 2477/* The red-black trees are only used in seeking mode to manage 2478 * the in-memory copy of the central directory. */ 2479 2480static int 2481cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2482{ 2483 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2484 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2485 2486 if (e1->local_header_offset > e2->local_header_offset) 2487 return -1; 2488 if (e1->local_header_offset < e2->local_header_offset) 2489 return 1; 2490 return 0; 2491} 2492 2493static int 2494cmp_key(const struct archive_rb_node *n, const void *key) 2495{ 2496 /* This function won't be called */ 2497 (void)n; /* UNUSED */ 2498 (void)key; /* UNUSED */ 2499 return 1; 2500} 2501 2502static const struct archive_rb_tree_ops rb_ops = { 2503 &cmp_node, &cmp_key 2504}; 2505 2506static int 2507rsrc_cmp_node(const struct archive_rb_node *n1, 2508 const struct archive_rb_node *n2) 2509{ 2510 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2511 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2512 2513 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2514} 2515 2516static int 2517rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2518{ 2519 const struct zip_entry *e = (const struct zip_entry *)n; 2520 return (strcmp((const char *)key, e->rsrcname.s)); 2521} 2522 2523static const struct archive_rb_tree_ops rb_rsrc_ops = { 2524 &rsrc_cmp_node, &rsrc_cmp_key 2525}; 2526 2527static const char * 2528rsrc_basename(const char *name, size_t name_length) 2529{ 2530 const char *s, *r; 2531 2532 r = s = name; 2533 for (;;) { 2534 s = memchr(s, '/', name_length - (s - name)); 2535 if (s == NULL) 2536 break; 2537 r = ++s; 2538 } 2539 return (r); 2540} 2541 2542static void 2543expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2544{ 2545 struct archive_string str; 2546 struct zip_entry *dir; 2547 char *s; 2548 2549 archive_string_init(&str); 2550 archive_strncpy(&str, name, name_length); 2551 for (;;) { 2552 s = strrchr(str.s, '/'); 2553 if (s == NULL) 2554 break; 2555 *s = '\0'; 2556 /* Transfer the parent directory from zip->tree_rsrc RB 2557 * tree to zip->tree RB tree to expose. */ 2558 dir = (struct zip_entry *) 2559 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2560 if (dir == NULL) 2561 break; 2562 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2563 archive_string_free(&dir->rsrcname); 2564 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2565 } 2566 archive_string_free(&str); 2567} 2568 2569static int 2570slurp_central_directory(struct archive_read *a, struct zip *zip) 2571{ 2572 ssize_t i; 2573 unsigned found; 2574 int64_t correction; 2575 ssize_t bytes_avail; 2576 const char *p; 2577 2578 /* 2579 * Find the start of the central directory. The end-of-CD 2580 * record has our starting point, but there are lots of 2581 * Zip archives which have had other data prepended to the 2582 * file, which makes the recorded offsets all too small. 2583 * So we search forward from the specified offset until we 2584 * find the real start of the central directory. Then we 2585 * know the correction we need to apply to account for leading 2586 * padding. 2587 */ 2588 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2589 return ARCHIVE_FATAL; 2590 2591 found = 0; 2592 while (!found) { 2593 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2594 return ARCHIVE_FATAL; 2595 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2596 switch (p[i + 3]) { 2597 case 'P': i += 3; break; 2598 case 'K': i += 2; break; 2599 case 001: i += 1; break; 2600 case 002: 2601 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2602 p += i; 2603 found = 1; 2604 } else 2605 i += 4; 2606 break; 2607 case 005: i += 1; break; 2608 case 006: 2609 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2610 p += i; 2611 found = 1; 2612 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2613 p += i; 2614 found = 1; 2615 } else 2616 i += 1; 2617 break; 2618 default: i += 4; break; 2619 } 2620 } 2621 __archive_read_consume(a, i); 2622 } 2623 correction = archive_filter_bytes(&a->archive, 0) 2624 - zip->central_directory_offset; 2625 2626 __archive_rb_tree_init(&zip->tree, &rb_ops); 2627 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2628 2629 zip->central_directory_entries_total = 0; 2630 while (1) { 2631 struct zip_entry *zip_entry; 2632 size_t filename_length, extra_length, comment_length; 2633 uint32_t external_attributes; 2634 const char *name, *r; 2635 2636 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2637 return ARCHIVE_FATAL; 2638 if (memcmp(p, "PK\006\006", 4) == 0 2639 || memcmp(p, "PK\005\006", 4) == 0) { 2640 break; 2641 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2642 archive_set_error(&a->archive, 2643 -1, "Invalid central directory signature"); 2644 return ARCHIVE_FATAL; 2645 } 2646 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2647 return ARCHIVE_FATAL; 2648 2649 zip_entry = calloc(1, sizeof(struct zip_entry)); 2650 zip_entry->next = zip->zip_entries; 2651 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2652 zip->zip_entries = zip_entry; 2653 zip->central_directory_entries_total++; 2654 2655 /* version = p[4]; */ 2656 zip_entry->system = p[5]; 2657 /* version_required = archive_le16dec(p + 6); */ 2658 zip_entry->zip_flags = archive_le16dec(p + 8); 2659 if (zip_entry->zip_flags 2660 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2661 zip->has_encrypted_entries = 1; 2662 } 2663 zip_entry->compression = (char)archive_le16dec(p + 10); 2664 zip_entry->mtime = zip_time(p + 12); 2665 zip_entry->crc32 = archive_le32dec(p + 16); 2666 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2667 zip_entry->decdat = p[13]; 2668 else 2669 zip_entry->decdat = p[19]; 2670 zip_entry->compressed_size = archive_le32dec(p + 20); 2671 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2672 filename_length = archive_le16dec(p + 28); 2673 extra_length = archive_le16dec(p + 30); 2674 comment_length = archive_le16dec(p + 32); 2675 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2676 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2677 external_attributes = archive_le32dec(p + 38); 2678 zip_entry->local_header_offset = 2679 archive_le32dec(p + 42) + correction; 2680 2681 /* If we can't guess the mode, leave it zero here; 2682 when we read the local file header we might get 2683 more information. */ 2684 if (zip_entry->system == 3) { 2685 zip_entry->mode = external_attributes >> 16; 2686 } else if (zip_entry->system == 0) { 2687 // Interpret MSDOS directory bit 2688 if (0x10 == (external_attributes & 0x10)) { 2689 zip_entry->mode = AE_IFDIR | 0775; 2690 } else { 2691 zip_entry->mode = AE_IFREG | 0664; 2692 } 2693 if (0x01 == (external_attributes & 0x01)) { 2694 // Read-only bit; strip write permissions 2695 zip_entry->mode &= 0555; 2696 } 2697 } else { 2698 zip_entry->mode = 0; 2699 } 2700 2701 /* We're done with the regular data; get the filename and 2702 * extra data. */ 2703 __archive_read_consume(a, 46); 2704 p = __archive_read_ahead(a, filename_length + extra_length, 2705 NULL); 2706 if (p == NULL) { 2707 archive_set_error(&a->archive, 2708 ARCHIVE_ERRNO_FILE_FORMAT, 2709 "Truncated ZIP file header"); 2710 return ARCHIVE_FATAL; 2711 } 2712 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { 2713 return ARCHIVE_FATAL; 2714 } 2715 2716 /* 2717 * Mac resource fork files are stored under the 2718 * "__MACOSX/" directory, so we should check if 2719 * it is. 2720 */ 2721 if (!zip->process_mac_extensions) { 2722 /* Treat every entry as a regular entry. */ 2723 __archive_rb_tree_insert_node(&zip->tree, 2724 &zip_entry->node); 2725 } else { 2726 name = p; 2727 r = rsrc_basename(name, filename_length); 2728 if (filename_length >= 9 && 2729 strncmp("__MACOSX/", name, 9) == 0) { 2730 /* If this file is not a resource fork nor 2731 * a directory. We should treat it as a non 2732 * resource fork file to expose it. */ 2733 if (name[filename_length-1] != '/' && 2734 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2735 __archive_rb_tree_insert_node( 2736 &zip->tree, &zip_entry->node); 2737 /* Expose its parent directories. */ 2738 expose_parent_dirs(zip, name, 2739 filename_length); 2740 } else { 2741 /* This file is a resource fork file or 2742 * a directory. */ 2743 archive_strncpy(&(zip_entry->rsrcname), 2744 name, filename_length); 2745 __archive_rb_tree_insert_node( 2746 &zip->tree_rsrc, &zip_entry->node); 2747 } 2748 } else { 2749 /* Generate resource fork name to find its 2750 * resource file at zip->tree_rsrc. */ 2751 archive_strcpy(&(zip_entry->rsrcname), 2752 "__MACOSX/"); 2753 archive_strncat(&(zip_entry->rsrcname), 2754 name, r - name); 2755 archive_strcat(&(zip_entry->rsrcname), "._"); 2756 archive_strncat(&(zip_entry->rsrcname), 2757 name + (r - name), 2758 filename_length - (r - name)); 2759 /* Register an entry to RB tree to sort it by 2760 * file offset. */ 2761 __archive_rb_tree_insert_node(&zip->tree, 2762 &zip_entry->node); 2763 } 2764 } 2765 2766 /* Skip the comment too ... */ 2767 __archive_read_consume(a, 2768 filename_length + extra_length + comment_length); 2769 } 2770 2771 return ARCHIVE_OK; 2772} 2773 2774static ssize_t 2775zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2776{ 2777 const char *p; 2778 ssize_t filename_length, extra_length; 2779 2780 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2781 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2782 "Truncated ZIP file header"); 2783 return (ARCHIVE_WARN); 2784 } 2785 p += extra; 2786 2787 if (memcmp(p, "PK\003\004", 4) != 0) { 2788 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2789 return ARCHIVE_WARN; 2790 } 2791 filename_length = archive_le16dec(p + 26); 2792 extra_length = archive_le16dec(p + 28); 2793 2794 return (30 + filename_length + extra_length); 2795} 2796 2797static int 2798zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2799 struct zip_entry *rsrc) 2800{ 2801 struct zip *zip = (struct zip *)a->format->data; 2802 unsigned char *metadata, *mp; 2803 int64_t offset = archive_filter_bytes(&a->archive, 0); 2804 size_t remaining_bytes, metadata_bytes; 2805 ssize_t hsize; 2806 int ret = ARCHIVE_OK, eof; 2807 2808 switch(rsrc->compression) { 2809 case 0: /* No compression. */ 2810 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2811 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2812 "Malformed OS X metadata entry: inconsistent size"); 2813 return (ARCHIVE_FATAL); 2814 } 2815#ifdef HAVE_ZLIB_H 2816 case 8: /* Deflate compression. */ 2817#endif 2818 break; 2819 default: /* Unsupported compression. */ 2820 /* Return a warning. */ 2821 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2822 "Unsupported ZIP compression method (%s)", 2823 compression_name(rsrc->compression)); 2824 /* We can't decompress this entry, but we will 2825 * be able to skip() it and try the next entry. */ 2826 return (ARCHIVE_WARN); 2827 } 2828 2829 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2830 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2831 "Mac metadata is too large: %jd > 4M bytes", 2832 (intmax_t)rsrc->uncompressed_size); 2833 return (ARCHIVE_WARN); 2834 } 2835 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2836 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2837 "Mac metadata is too large: %jd > 4M bytes", 2838 (intmax_t)rsrc->compressed_size); 2839 return (ARCHIVE_WARN); 2840 } 2841 2842 metadata = malloc((size_t)rsrc->uncompressed_size); 2843 if (metadata == NULL) { 2844 archive_set_error(&a->archive, ENOMEM, 2845 "Can't allocate memory for Mac metadata"); 2846 return (ARCHIVE_FATAL); 2847 } 2848 2849 if (offset < rsrc->local_header_offset) 2850 __archive_read_consume(a, rsrc->local_header_offset - offset); 2851 else if (offset != rsrc->local_header_offset) { 2852 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2853 } 2854 2855 hsize = zip_get_local_file_header_size(a, 0); 2856 __archive_read_consume(a, hsize); 2857 2858 remaining_bytes = (size_t)rsrc->compressed_size; 2859 metadata_bytes = (size_t)rsrc->uncompressed_size; 2860 mp = metadata; 2861 eof = 0; 2862 while (!eof && remaining_bytes) { 2863 const unsigned char *p; 2864 ssize_t bytes_avail; 2865 size_t bytes_used; 2866 2867 p = __archive_read_ahead(a, 1, &bytes_avail); 2868 if (p == NULL) { 2869 archive_set_error(&a->archive, 2870 ARCHIVE_ERRNO_FILE_FORMAT, 2871 "Truncated ZIP file header"); 2872 ret = ARCHIVE_WARN; 2873 goto exit_mac_metadata; 2874 } 2875 if ((size_t)bytes_avail > remaining_bytes) 2876 bytes_avail = remaining_bytes; 2877 switch(rsrc->compression) { 2878 case 0: /* No compression. */ 2879 if ((size_t)bytes_avail > metadata_bytes) 2880 bytes_avail = metadata_bytes; 2881 memcpy(mp, p, bytes_avail); 2882 bytes_used = (size_t)bytes_avail; 2883 metadata_bytes -= bytes_used; 2884 mp += bytes_used; 2885 if (metadata_bytes == 0) 2886 eof = 1; 2887 break; 2888#ifdef HAVE_ZLIB_H 2889 case 8: /* Deflate compression. */ 2890 { 2891 int r; 2892 2893 ret = zip_deflate_init(a, zip); 2894 if (ret != ARCHIVE_OK) 2895 goto exit_mac_metadata; 2896 zip->stream.next_in = 2897 (Bytef *)(uintptr_t)(const void *)p; 2898 zip->stream.avail_in = (uInt)bytes_avail; 2899 zip->stream.total_in = 0; 2900 zip->stream.next_out = mp; 2901 zip->stream.avail_out = (uInt)metadata_bytes; 2902 zip->stream.total_out = 0; 2903 2904 r = inflate(&zip->stream, 0); 2905 switch (r) { 2906 case Z_OK: 2907 break; 2908 case Z_STREAM_END: 2909 eof = 1; 2910 break; 2911 case Z_MEM_ERROR: 2912 archive_set_error(&a->archive, ENOMEM, 2913 "Out of memory for ZIP decompression"); 2914 ret = ARCHIVE_FATAL; 2915 goto exit_mac_metadata; 2916 default: 2917 archive_set_error(&a->archive, 2918 ARCHIVE_ERRNO_MISC, 2919 "ZIP decompression failed (%d)", r); 2920 ret = ARCHIVE_FATAL; 2921 goto exit_mac_metadata; 2922 } 2923 bytes_used = zip->stream.total_in; 2924 metadata_bytes -= zip->stream.total_out; 2925 mp += zip->stream.total_out; 2926 break; 2927 } 2928#endif 2929 default: 2930 bytes_used = 0; 2931 break; 2932 } 2933 __archive_read_consume(a, bytes_used); 2934 remaining_bytes -= bytes_used; 2935 } 2936 archive_entry_copy_mac_metadata(entry, metadata, 2937 (size_t)rsrc->uncompressed_size - metadata_bytes); 2938 2939exit_mac_metadata: 2940 __archive_read_seek(a, offset, SEEK_SET); 2941 zip->decompress_init = 0; 2942 free(metadata); 2943 return (ret); 2944} 2945 2946static int 2947archive_read_format_zip_seekable_read_header(struct archive_read *a, 2948 struct archive_entry *entry) 2949{ 2950 struct zip *zip = (struct zip *)a->format->data; 2951 struct zip_entry *rsrc; 2952 int64_t offset; 2953 int r, ret = ARCHIVE_OK; 2954 2955 /* 2956 * It should be sufficient to call archive_read_next_header() for 2957 * a reader to determine if an entry is encrypted or not. If the 2958 * encryption of an entry is only detectable when calling 2959 * archive_read_data(), so be it. We'll do the same check there 2960 * as well. 2961 */ 2962 if (zip->has_encrypted_entries == 2963 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2964 zip->has_encrypted_entries = 0; 2965 2966 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2967 if (a->archive.archive_format_name == NULL) 2968 a->archive.archive_format_name = "ZIP"; 2969 2970 if (zip->zip_entries == NULL) { 2971 r = slurp_central_directory(a, zip); 2972 if (r != ARCHIVE_OK) 2973 return r; 2974 /* Get first entry whose local header offset is lower than 2975 * other entries in the archive file. */ 2976 zip->entry = 2977 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 2978 } else if (zip->entry != NULL) { 2979 /* Get next entry in local header offset order. */ 2980 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 2981 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 2982 } 2983 2984 if (zip->entry == NULL) 2985 return ARCHIVE_EOF; 2986 2987 if (zip->entry->rsrcname.s) 2988 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 2989 &zip->tree_rsrc, zip->entry->rsrcname.s); 2990 else 2991 rsrc = NULL; 2992 2993 if (zip->cctx_valid) 2994 archive_decrypto_aes_ctr_release(&zip->cctx); 2995 if (zip->hctx_valid) 2996 archive_hmac_sha1_cleanup(&zip->hctx); 2997 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2998 __archive_read_reset_passphrase(a); 2999 3000 /* File entries are sorted by the header offset, we should mostly 3001 * use __archive_read_consume to advance a read point to avoid redundant 3002 * data reading. */ 3003 offset = archive_filter_bytes(&a->archive, 0); 3004 if (offset < zip->entry->local_header_offset) 3005 __archive_read_consume(a, 3006 zip->entry->local_header_offset - offset); 3007 else if (offset != zip->entry->local_header_offset) { 3008 __archive_read_seek(a, zip->entry->local_header_offset, 3009 SEEK_SET); 3010 } 3011 zip->unconsumed = 0; 3012 r = zip_read_local_file_header(a, entry, zip); 3013 if (r != ARCHIVE_OK) 3014 return r; 3015 if (rsrc) { 3016 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 3017 if (ret2 < ret) 3018 ret = ret2; 3019 } 3020 return (ret); 3021} 3022 3023/* 3024 * We're going to seek for the next header anyway, so we don't 3025 * need to bother doing anything here. 3026 */ 3027static int 3028archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3029{ 3030 struct zip *zip; 3031 zip = (struct zip *)(a->format->data); 3032 3033 zip->unconsumed = 0; 3034 return (ARCHIVE_OK); 3035} 3036 3037int 3038archive_read_support_format_zip_seekable(struct archive *_a) 3039{ 3040 struct archive_read *a = (struct archive_read *)_a; 3041 struct zip *zip; 3042 int r; 3043 3044 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3045 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3046 3047 zip = (struct zip *)calloc(1, sizeof(*zip)); 3048 if (zip == NULL) { 3049 archive_set_error(&a->archive, ENOMEM, 3050 "Can't allocate zip data"); 3051 return (ARCHIVE_FATAL); 3052 } 3053 3054#ifdef HAVE_COPYFILE_H 3055 /* Set this by default on Mac OS. */ 3056 zip->process_mac_extensions = 1; 3057#endif 3058 3059 /* 3060 * Until enough data has been read, we cannot tell about 3061 * any encrypted entries yet. 3062 */ 3063 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3064 zip->crc32func = real_crc32; 3065 3066 r = __archive_read_register_format(a, 3067 zip, 3068 "zip", 3069 archive_read_format_zip_seekable_bid, 3070 archive_read_format_zip_options, 3071 archive_read_format_zip_seekable_read_header, 3072 archive_read_format_zip_read_data, 3073 archive_read_format_zip_read_data_skip_seekable, 3074 NULL, 3075 archive_read_format_zip_cleanup, 3076 archive_read_support_format_zip_capabilities_seekable, 3077 archive_read_format_zip_has_encrypted_entries); 3078 3079 if (r != ARCHIVE_OK) 3080 free(zip); 3081 return (ARCHIVE_OK); 3082} 3083