1 /*- 2 * Copyright (c) 2004-2013 Tim Kientzle 3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA 4 * Copyright (c) 2013 Konrad Kleine 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28 #include "archive_platform.h" 29 __FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_zip.c 201102 2009-12-28 03:11:36Z kientzle $"); 30 31 /* 32 * The definitive documentation of the Zip file format is: 33 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT 34 * 35 * The Info-Zip project has pioneered various extensions to better 36 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 37 * "Ux", and 0x7875 "ux" extensions for time and ownership 38 * information. 39 * 40 * History of this code: The streaming Zip reader was first added to 41 * libarchive in January 2005. Support for seekable input sources was 42 * added in Nov 2011. Zip64 support (including a significant code 43 * refactoring) was added in 2014. 44 */ 45 46 #ifdef HAVE_ERRNO_H 47 #include <errno.h> 48 #endif 49 #ifdef HAVE_STDLIB_H 50 #include <stdlib.h> 51 #endif 52 #ifdef HAVE_ZLIB_H 53 #include <zlib.h> 54 #endif 55 56 #include "archive.h" 57 #include "archive_digest_private.h" 58 #include "archive_cryptor_private.h" 59 #include "archive_endian.h" 60 #include "archive_entry.h" 61 #include "archive_entry_locale.h" 62 #include "archive_hmac_private.h" 63 #include "archive_private.h" 64 #include "archive_rb.h" 65 #include "archive_read_private.h" 66 67 #ifndef HAVE_ZLIB_H 68 #include "archive_crc32.h" 69 #endif 70 71 struct zip_entry { 72 struct archive_rb_node node; 73 struct zip_entry *next; 74 int64_t local_header_offset; 75 int64_t compressed_size; 76 int64_t uncompressed_size; 77 int64_t gid; 78 int64_t uid; 79 struct archive_string rsrcname; 80 time_t mtime; 81 time_t atime; 82 time_t ctime; 83 uint32_t crc32; 84 uint16_t mode; 85 uint16_t zip_flags; /* From GP Flags Field */ 86 unsigned char compression; 87 unsigned char system; /* From "version written by" */ 88 unsigned char flags; /* Our extra markers. */ 89 unsigned char decdat;/* Used for Decryption check */ 90 91 /* WinZip AES encryption extra field should be available 92 * when compression is 99. */ 93 struct { 94 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ 95 unsigned vendor; 96 #define AES_VENDOR_AE_1 0x0001 97 #define AES_VENDOR_AE_2 0x0002 98 /* AES encryption strength: 99 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ 100 unsigned strength; 101 /* Actual compression method. */ 102 unsigned char compression; 103 } aes_extra; 104 }; 105 106 struct 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) 133 134 struct 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 unsigned char *decrypted_buffer; 185 unsigned char *decrypted_ptr; 186 size_t decrypted_buffer_size; 187 size_t decrypted_bytes_remaining; 188 size_t decrypted_unconsumed_bytes; 189 190 /* Traditional PKWARE decryption. */ 191 struct trad_enc_ctx tctx; 192 char tctx_valid; 193 194 /* WinZip AES decyption. */ 195 /* Contexts used for AES decryption. */ 196 archive_crypto_ctx cctx; 197 char cctx_valid; 198 archive_hmac_sha1_ctx hctx; 199 char hctx_valid; 200 201 /* Strong encryption's decryption header information. */ 202 unsigned iv_size; 203 unsigned alg_id; 204 unsigned bit_len; 205 unsigned flags; 206 unsigned erd_size; 207 unsigned v_size; 208 unsigned v_crc32; 209 uint8_t *iv; 210 uint8_t *erd; 211 uint8_t *v_data; 212 }; 213 214 /* Many systems define min or MIN, but not all. */ 215 #define zipmin(a,b) ((a) < (b) ? (a) : (b)) 216 217 /* ------------------------------------------------------------------------ */ 218 219 /* 220 Traditional PKWARE Decryption functions. 221 */ 222 223 static void 224 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) 225 { 226 uint8_t t; 227 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) 228 229 ctx->keys[0] = CRC32(ctx->keys[0], c); 230 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; 231 t = (ctx->keys[1] >> 24) & 0xff; 232 ctx->keys[2] = CRC32(ctx->keys[2], t); 233 #undef CRC32 234 } 235 236 static uint8_t 237 trad_enc_decypt_byte(struct trad_enc_ctx *ctx) 238 { 239 unsigned temp = ctx->keys[2] | 2; 240 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; 241 } 242 243 static void 244 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, 245 size_t in_len, uint8_t *out, size_t out_len) 246 { 247 unsigned i, max; 248 249 max = (unsigned)((in_len < out_len)? in_len: out_len); 250 251 for (i = 0; i < max; i++) { 252 uint8_t t = in[i] ^ trad_enc_decypt_byte(ctx); 253 out[i] = t; 254 trad_enc_update_keys(ctx, t); 255 } 256 } 257 258 static int 259 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, 260 const uint8_t *key, size_t key_len, uint8_t *crcchk) 261 { 262 uint8_t header[12]; 263 264 if (key_len < 12) { 265 *crcchk = 0xff; 266 return -1; 267 } 268 269 ctx->keys[0] = 305419896L; 270 ctx->keys[1] = 591751049L; 271 ctx->keys[2] = 878082192L; 272 273 for (;pw_len; --pw_len) 274 trad_enc_update_keys(ctx, *pw++); 275 276 trad_enc_decrypt_update(ctx, key, 12, header, 12); 277 /* Return the last byte for CRC check. */ 278 *crcchk = header[11]; 279 return 0; 280 } 281 282 #if 0 283 static void 284 crypt_derive_key_sha1(const void *p, int size, unsigned char *key, 285 int key_size) 286 { 287 #define MD_SIZE 20 288 archive_sha1_ctx ctx; 289 unsigned char md1[MD_SIZE]; 290 unsigned char md2[MD_SIZE * 2]; 291 unsigned char mkb[64]; 292 int i; 293 294 archive_sha1_init(&ctx); 295 archive_sha1_update(&ctx, p, size); 296 archive_sha1_final(&ctx, md1); 297 298 memset(mkb, 0x36, sizeof(mkb)); 299 for (i = 0; i < MD_SIZE; i++) 300 mkb[i] ^= md1[i]; 301 archive_sha1_init(&ctx); 302 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 303 archive_sha1_final(&ctx, md2); 304 305 memset(mkb, 0x5C, sizeof(mkb)); 306 for (i = 0; i < MD_SIZE; i++) 307 mkb[i] ^= md1[i]; 308 archive_sha1_init(&ctx); 309 archive_sha1_update(&ctx, mkb, sizeof(mkb)); 310 archive_sha1_final(&ctx, md2 + MD_SIZE); 311 312 if (key_size > 32) 313 key_size = 32; 314 memcpy(key, md2, key_size); 315 #undef MD_SIZE 316 } 317 #endif 318 319 /* 320 * Common code for streaming or seeking modes. 321 * 322 * Includes code to read local file headers, decompress data 323 * from entry bodies, and common API. 324 */ 325 326 static unsigned long 327 real_crc32(unsigned long crc, const void *buff, size_t len) 328 { 329 return crc32(crc, buff, (unsigned int)len); 330 } 331 332 /* Used by "ignorecrc32" option to speed up tests. */ 333 static unsigned long 334 fake_crc32(unsigned long crc, const void *buff, size_t len) 335 { 336 (void)crc; /* UNUSED */ 337 (void)buff; /* UNUSED */ 338 (void)len; /* UNUSED */ 339 return 0; 340 } 341 342 static struct { 343 int id; 344 const char * name; 345 } compression_methods[] = { 346 {0, "uncompressed"}, /* The file is stored (no compression) */ 347 {1, "shrinking"}, /* The file is Shrunk */ 348 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ 349 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ 350 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ 351 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ 352 {6, "imploded"}, /* The file is Imploded */ 353 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ 354 {8, "deflation"}, /* The file is Deflated */ 355 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ 356 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding 357 * (old IBM TERSE) */ 358 {11, "reserved"}, /* Reserved by PKWARE */ 359 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ 360 {13, "reserved"}, /* Reserved by PKWARE */ 361 {14, "lzma"}, /* LZMA (EFS) */ 362 {15, "reserved"}, /* Reserved by PKWARE */ 363 {16, "reserved"}, /* Reserved by PKWARE */ 364 {17, "reserved"}, /* Reserved by PKWARE */ 365 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ 366 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ 367 {97, "wav-pack"}, /* WavPack compressed data */ 368 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ 369 {99, "aes"} /* WinZip AES encryption */ 370 }; 371 372 static const char * 373 compression_name(const int compression) 374 { 375 static const int num_compression_methods = 376 sizeof(compression_methods)/sizeof(compression_methods[0]); 377 int i=0; 378 379 while(compression >= 0 && i < num_compression_methods) { 380 if (compression_methods[i].id == compression) 381 return compression_methods[i].name; 382 i++; 383 } 384 return "??"; 385 } 386 387 /* Convert an MSDOS-style date/time into Unix-style time. */ 388 static time_t 389 zip_time(const char *p) 390 { 391 int msTime, msDate; 392 struct tm ts; 393 394 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); 395 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); 396 397 memset(&ts, 0, sizeof(ts)); 398 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ 399 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ 400 ts.tm_mday = msDate & 0x1f; /* Day of month. */ 401 ts.tm_hour = (msTime >> 11) & 0x1f; 402 ts.tm_min = (msTime >> 5) & 0x3f; 403 ts.tm_sec = (msTime << 1) & 0x3e; 404 ts.tm_isdst = -1; 405 return mktime(&ts); 406 } 407 408 /* 409 * The extra data is stored as a list of 410 * id1+size1+data1 + id2+size2+data2 ... 411 * triplets. id and size are 2 bytes each. 412 */ 413 static void 414 process_extra(const char *p, size_t extra_length, struct zip_entry* zip_entry) 415 { 416 unsigned offset = 0; 417 418 while (offset < extra_length - 4) { 419 unsigned short headerid = archive_le16dec(p + offset); 420 unsigned short datasize = archive_le16dec(p + offset + 2); 421 422 offset += 4; 423 if (offset + datasize > extra_length) { 424 break; 425 } 426 #ifdef DEBUG 427 fprintf(stderr, "Header id 0x%04x, length %d\n", 428 headerid, datasize); 429 #endif 430 switch (headerid) { 431 case 0x0001: 432 /* Zip64 extended information extra field. */ 433 zip_entry->flags |= LA_USED_ZIP64; 434 if (zip_entry->uncompressed_size == 0xffffffff) { 435 if (datasize < 8) 436 break; 437 zip_entry->uncompressed_size = 438 archive_le64dec(p + offset); 439 offset += 8; 440 datasize -= 8; 441 } 442 if (zip_entry->compressed_size == 0xffffffff) { 443 if (datasize < 8) 444 break; 445 zip_entry->compressed_size = 446 archive_le64dec(p + offset); 447 offset += 8; 448 datasize -= 8; 449 } 450 if (zip_entry->local_header_offset == 0xffffffff) { 451 if (datasize < 8) 452 break; 453 zip_entry->local_header_offset = 454 archive_le64dec(p + offset); 455 offset += 8; 456 datasize -= 8; 457 } 458 /* archive_le32dec(p + offset) gives disk 459 * on which file starts, but we don't handle 460 * multi-volume Zip files. */ 461 break; 462 #ifdef DEBUG 463 case 0x0017: 464 { 465 /* Strong encryption field. */ 466 if (archive_le16dec(p + offset) == 2) { 467 unsigned algId = 468 archive_le16dec(p + offset + 2); 469 unsigned bitLen = 470 archive_le16dec(p + offset + 4); 471 int flags = 472 archive_le16dec(p + offset + 6); 473 fprintf(stderr, "algId=0x%04x, bitLen=%u, " 474 "flgas=%d\n", algId, bitLen,flags); 475 } 476 break; 477 } 478 #endif 479 case 0x5455: 480 { 481 /* Extended time field "UT". */ 482 int flags = p[offset]; 483 offset++; 484 datasize--; 485 /* Flag bits indicate which dates are present. */ 486 if (flags & 0x01) 487 { 488 #ifdef DEBUG 489 fprintf(stderr, "mtime: %lld -> %d\n", 490 (long long)zip_entry->mtime, 491 archive_le32dec(p + offset)); 492 #endif 493 if (datasize < 4) 494 break; 495 zip_entry->mtime = archive_le32dec(p + offset); 496 offset += 4; 497 datasize -= 4; 498 } 499 if (flags & 0x02) 500 { 501 if (datasize < 4) 502 break; 503 zip_entry->atime = archive_le32dec(p + offset); 504 offset += 4; 505 datasize -= 4; 506 } 507 if (flags & 0x04) 508 { 509 if (datasize < 4) 510 break; 511 zip_entry->ctime = archive_le32dec(p + offset); 512 offset += 4; 513 datasize -= 4; 514 } 515 break; 516 } 517 case 0x5855: 518 { 519 /* Info-ZIP Unix Extra Field (old version) "UX". */ 520 if (datasize >= 8) { 521 zip_entry->atime = archive_le32dec(p + offset); 522 zip_entry->mtime = 523 archive_le32dec(p + offset + 4); 524 } 525 if (datasize >= 12) { 526 zip_entry->uid = 527 archive_le16dec(p + offset + 8); 528 zip_entry->gid = 529 archive_le16dec(p + offset + 10); 530 } 531 break; 532 } 533 case 0x6c78: 534 { 535 /* Experimental 'xl' field */ 536 /* 537 * Introduced Dec 2013 to provide a way to 538 * include external file attributes (and other 539 * fields that ordinarily appear only in 540 * central directory) in local file header. 541 * This provides file type and permission 542 * information necessary to support full 543 * streaming extraction. Currently being 544 * discussed with other Zip developers 545 * ... subject to change. 546 * 547 * Format: 548 * The field starts with a bitmap that specifies 549 * which additional fields are included. The 550 * bitmap is variable length and can be extended in 551 * the future. 552 * 553 * n bytes - feature bitmap: first byte has low-order 554 * 7 bits. If high-order bit is set, a subsequent 555 * byte holds the next 7 bits, etc. 556 * 557 * if bitmap & 1, 2 byte "version made by" 558 * if bitmap & 2, 2 byte "internal file attributes" 559 * if bitmap & 4, 4 byte "external file attributes" 560 * if bitmap & 8, 2 byte comment length + n byte comment 561 */ 562 int bitmap, bitmap_last; 563 564 if (datasize < 1) 565 break; 566 bitmap_last = bitmap = 0xff & p[offset]; 567 offset += 1; 568 datasize -= 1; 569 570 /* We only support first 7 bits of bitmap; skip rest. */ 571 while ((bitmap_last & 0x80) != 0 572 && datasize >= 1) { 573 bitmap_last = p[offset]; 574 offset += 1; 575 datasize -= 1; 576 } 577 578 if (bitmap & 1) { 579 /* 2 byte "version made by" */ 580 if (datasize < 2) 581 break; 582 zip_entry->system 583 = archive_le16dec(p + offset) >> 8; 584 offset += 2; 585 datasize -= 2; 586 } 587 if (bitmap & 2) { 588 /* 2 byte "internal file attributes" */ 589 uint32_t internal_attributes; 590 if (datasize < 2) 591 break; 592 internal_attributes 593 = archive_le16dec(p + offset); 594 /* Not used by libarchive at present. */ 595 (void)internal_attributes; /* UNUSED */ 596 offset += 2; 597 datasize -= 2; 598 } 599 if (bitmap & 4) { 600 /* 4 byte "external file attributes" */ 601 uint32_t external_attributes; 602 if (datasize < 4) 603 break; 604 external_attributes 605 = archive_le32dec(p + offset); 606 if (zip_entry->system == 3) { 607 zip_entry->mode 608 = external_attributes >> 16; 609 } else if (zip_entry->system == 0) { 610 // Interpret MSDOS directory bit 611 if (0x10 == (external_attributes & 0x10)) { 612 zip_entry->mode = AE_IFDIR | 0775; 613 } else { 614 zip_entry->mode = AE_IFREG | 0664; 615 } 616 if (0x01 == (external_attributes & 0x01)) { 617 // Read-only bit; strip write permissions 618 zip_entry->mode &= 0555; 619 } 620 } else { 621 zip_entry->mode = 0; 622 } 623 offset += 4; 624 datasize -= 4; 625 } 626 if (bitmap & 8) { 627 /* 2 byte comment length + comment */ 628 uint32_t comment_length; 629 if (datasize < 2) 630 break; 631 comment_length 632 = archive_le16dec(p + offset); 633 offset += 2; 634 datasize -= 2; 635 636 if (datasize < comment_length) 637 break; 638 /* Comment is not supported by libarchive */ 639 offset += comment_length; 640 datasize -= comment_length; 641 } 642 break; 643 } 644 case 0x7855: 645 /* Info-ZIP Unix Extra Field (type 2) "Ux". */ 646 #ifdef DEBUG 647 fprintf(stderr, "uid %d gid %d\n", 648 archive_le16dec(p + offset), 649 archive_le16dec(p + offset + 2)); 650 #endif 651 if (datasize >= 2) 652 zip_entry->uid = archive_le16dec(p + offset); 653 if (datasize >= 4) 654 zip_entry->gid = 655 archive_le16dec(p + offset + 2); 656 break; 657 case 0x7875: 658 { 659 /* Info-Zip Unix Extra Field (type 3) "ux". */ 660 int uidsize = 0, gidsize = 0; 661 662 /* TODO: support arbitrary uidsize/gidsize. */ 663 if (datasize >= 1 && p[offset] == 1) {/* version=1 */ 664 if (datasize >= 4) { 665 /* get a uid size. */ 666 uidsize = 0xff & (int)p[offset+1]; 667 if (uidsize == 2) 668 zip_entry->uid = 669 archive_le16dec( 670 p + offset + 2); 671 else if (uidsize == 4 && datasize >= 6) 672 zip_entry->uid = 673 archive_le32dec( 674 p + offset + 2); 675 } 676 if (datasize >= (2 + uidsize + 3)) { 677 /* get a gid size. */ 678 gidsize = 0xff & (int)p[offset+2+uidsize]; 679 if (gidsize == 2) 680 zip_entry->gid = 681 archive_le16dec( 682 p+offset+2+uidsize+1); 683 else if (gidsize == 4 && 684 datasize >= (2 + uidsize + 5)) 685 zip_entry->gid = 686 archive_le32dec( 687 p+offset+2+uidsize+1); 688 } 689 } 690 break; 691 } 692 case 0x9901: 693 /* WinZIp AES extra data field. */ 694 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { 695 /* Vendor version. */ 696 zip_entry->aes_extra.vendor = 697 archive_le16dec(p + offset); 698 /* AES encryption strength. */ 699 zip_entry->aes_extra.strength = p[offset + 4]; 700 /* Actual compression method. */ 701 zip_entry->aes_extra.compression = 702 p[offset + 5]; 703 } 704 break; 705 default: 706 break; 707 } 708 offset += datasize; 709 } 710 #ifdef DEBUG 711 if (offset != extra_length) 712 { 713 fprintf(stderr, 714 "Extra data field contents do not match reported size!\n"); 715 } 716 #endif 717 } 718 719 /* 720 * Assumes file pointer is at beginning of local file header. 721 */ 722 static int 723 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, 724 struct zip *zip) 725 { 726 const char *p; 727 const void *h; 728 const wchar_t *wp; 729 const char *cp; 730 size_t len, filename_length, extra_length; 731 struct archive_string_conv *sconv; 732 struct zip_entry *zip_entry = zip->entry; 733 struct zip_entry zip_entry_central_dir; 734 int ret = ARCHIVE_OK; 735 char version; 736 737 /* Save a copy of the original for consistency checks. */ 738 zip_entry_central_dir = *zip_entry; 739 740 zip->decompress_init = 0; 741 zip->end_of_entry = 0; 742 zip->entry_uncompressed_bytes_read = 0; 743 zip->entry_compressed_bytes_read = 0; 744 zip->entry_crc32 = zip->crc32func(0, NULL, 0); 745 746 /* Setup default conversion. */ 747 if (zip->sconv == NULL && !zip->init_default_conversion) { 748 zip->sconv_default = 749 archive_string_default_conversion_for_read(&(a->archive)); 750 zip->init_default_conversion = 1; 751 } 752 753 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { 754 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 755 "Truncated ZIP file header"); 756 return (ARCHIVE_FATAL); 757 } 758 759 if (memcmp(p, "PK\003\004", 4) != 0) { 760 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 761 return ARCHIVE_FATAL; 762 } 763 version = p[4]; 764 zip_entry->system = p[5]; 765 zip_entry->zip_flags = archive_le16dec(p + 6); 766 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { 767 zip->has_encrypted_entries = 1; 768 archive_entry_set_is_data_encrypted(entry, 1); 769 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && 770 zip_entry->zip_flags & ZIP_ENCRYPTED && 771 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { 772 archive_entry_set_is_metadata_encrypted(entry, 1); 773 return ARCHIVE_FATAL; 774 } 775 } 776 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); 777 zip_entry->compression = (char)archive_le16dec(p + 8); 778 zip_entry->mtime = zip_time(p + 10); 779 zip_entry->crc32 = archive_le32dec(p + 14); 780 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 781 zip_entry->decdat = p[11]; 782 else 783 zip_entry->decdat = p[17]; 784 zip_entry->compressed_size = archive_le32dec(p + 18); 785 zip_entry->uncompressed_size = archive_le32dec(p + 22); 786 filename_length = archive_le16dec(p + 26); 787 extra_length = archive_le16dec(p + 28); 788 789 __archive_read_consume(a, 30); 790 791 /* Read the filename. */ 792 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { 793 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 794 "Truncated ZIP file header"); 795 return (ARCHIVE_FATAL); 796 } 797 if (zip_entry->zip_flags & ZIP_UTF8_NAME) { 798 /* The filename is stored to be UTF-8. */ 799 if (zip->sconv_utf8 == NULL) { 800 zip->sconv_utf8 = 801 archive_string_conversion_from_charset( 802 &a->archive, "UTF-8", 1); 803 if (zip->sconv_utf8 == NULL) 804 return (ARCHIVE_FATAL); 805 } 806 sconv = zip->sconv_utf8; 807 } else if (zip->sconv != NULL) 808 sconv = zip->sconv; 809 else 810 sconv = zip->sconv_default; 811 812 if (archive_entry_copy_pathname_l(entry, 813 h, filename_length, sconv) != 0) { 814 if (errno == ENOMEM) { 815 archive_set_error(&a->archive, ENOMEM, 816 "Can't allocate memory for Pathname"); 817 return (ARCHIVE_FATAL); 818 } 819 archive_set_error(&a->archive, 820 ARCHIVE_ERRNO_FILE_FORMAT, 821 "Pathname cannot be converted " 822 "from %s to current locale.", 823 archive_string_conversion_charset_name(sconv)); 824 ret = ARCHIVE_WARN; 825 } 826 __archive_read_consume(a, filename_length); 827 828 /* Read the extra data. */ 829 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { 830 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 831 "Truncated ZIP file header"); 832 return (ARCHIVE_FATAL); 833 } 834 835 process_extra(h, extra_length, zip_entry); 836 __archive_read_consume(a, extra_length); 837 838 /* Work around a bug in Info-Zip: When reading from a pipe, it 839 * stats the pipe instead of synthesizing a file entry. */ 840 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { 841 zip_entry->mode &= ~ AE_IFMT; 842 zip_entry->mode |= AE_IFREG; 843 } 844 845 if ((zip_entry->mode & AE_IFMT) == 0) { 846 /* Especially in streaming mode, we can end up 847 here without having seen proper mode information. 848 Guess from the filename. */ 849 wp = archive_entry_pathname_w(entry); 850 if (wp != NULL) { 851 len = wcslen(wp); 852 if (len > 0 && wp[len - 1] == L'/') 853 zip_entry->mode |= AE_IFDIR; 854 else 855 zip_entry->mode |= AE_IFREG; 856 } else { 857 cp = archive_entry_pathname(entry); 858 len = (cp != NULL)?strlen(cp):0; 859 if (len > 0 && cp[len - 1] == '/') 860 zip_entry->mode |= AE_IFDIR; 861 else 862 zip_entry->mode |= AE_IFREG; 863 } 864 if (zip_entry->mode == AE_IFDIR) { 865 zip_entry->mode |= 0775; 866 } else if (zip_entry->mode == AE_IFREG) { 867 zip_entry->mode |= 0664; 868 } 869 } 870 871 /* Make sure directories end in '/' */ 872 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { 873 wp = archive_entry_pathname_w(entry); 874 if (wp != NULL) { 875 len = wcslen(wp); 876 if (len > 0 && wp[len - 1] != L'/') { 877 struct archive_wstring s; 878 archive_string_init(&s); 879 archive_wstrcat(&s, wp); 880 archive_wstrappend_wchar(&s, L'/'); 881 archive_entry_copy_pathname_w(entry, s.s); 882 } 883 } else { 884 cp = archive_entry_pathname(entry); 885 len = (cp != NULL)?strlen(cp):0; 886 if (len > 0 && cp[len - 1] != '/') { 887 struct archive_string s; 888 archive_string_init(&s); 889 archive_strcat(&s, cp); 890 archive_strappend_char(&s, '/'); 891 archive_entry_set_pathname(entry, s.s); 892 } 893 } 894 } 895 896 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { 897 /* If this came from the central dir, it's size info 898 * is definitive, so ignore the length-at-end flag. */ 899 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; 900 /* If local header is missing a value, use the one from 901 the central directory. If both have it, warn about 902 mismatches. */ 903 if (zip_entry->crc32 == 0) { 904 zip_entry->crc32 = zip_entry_central_dir.crc32; 905 } else if (!zip->ignore_crc32 906 && zip_entry->crc32 != zip_entry_central_dir.crc32) { 907 archive_set_error(&a->archive, 908 ARCHIVE_ERRNO_FILE_FORMAT, 909 "Inconsistent CRC32 values"); 910 ret = ARCHIVE_WARN; 911 } 912 if (zip_entry->compressed_size == 0) { 913 zip_entry->compressed_size 914 = zip_entry_central_dir.compressed_size; 915 } else if (zip_entry->compressed_size 916 != zip_entry_central_dir.compressed_size) { 917 archive_set_error(&a->archive, 918 ARCHIVE_ERRNO_FILE_FORMAT, 919 "Inconsistent compressed size: " 920 "%jd in central directory, %jd in local header", 921 (intmax_t)zip_entry_central_dir.compressed_size, 922 (intmax_t)zip_entry->compressed_size); 923 ret = ARCHIVE_WARN; 924 } 925 if (zip_entry->uncompressed_size == 0) { 926 zip_entry->uncompressed_size 927 = zip_entry_central_dir.uncompressed_size; 928 } else if (zip_entry->uncompressed_size 929 != zip_entry_central_dir.uncompressed_size) { 930 archive_set_error(&a->archive, 931 ARCHIVE_ERRNO_FILE_FORMAT, 932 "Inconsistent uncompressed size: " 933 "%jd in central directory, %jd in local header", 934 (intmax_t)zip_entry_central_dir.uncompressed_size, 935 (intmax_t)zip_entry->uncompressed_size); 936 ret = ARCHIVE_WARN; 937 } 938 } 939 940 /* Populate some additional entry fields: */ 941 archive_entry_set_mode(entry, zip_entry->mode); 942 archive_entry_set_uid(entry, zip_entry->uid); 943 archive_entry_set_gid(entry, zip_entry->gid); 944 archive_entry_set_mtime(entry, zip_entry->mtime, 0); 945 archive_entry_set_ctime(entry, zip_entry->ctime, 0); 946 archive_entry_set_atime(entry, zip_entry->atime, 0); 947 948 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { 949 size_t linkname_length; 950 951 if (zip_entry->compressed_size > 64 * 1024) { 952 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 953 "Zip file with oversized link entry"); 954 return ARCHIVE_FATAL; 955 } 956 957 linkname_length = (size_t)zip_entry->compressed_size; 958 959 archive_entry_set_size(entry, 0); 960 p = __archive_read_ahead(a, linkname_length, NULL); 961 if (p == NULL) { 962 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 963 "Truncated Zip file"); 964 return ARCHIVE_FATAL; 965 } 966 967 sconv = zip->sconv; 968 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) 969 sconv = zip->sconv_utf8; 970 if (sconv == NULL) 971 sconv = zip->sconv_default; 972 if (archive_entry_copy_symlink_l(entry, p, linkname_length, 973 sconv) != 0) { 974 if (errno != ENOMEM && sconv == zip->sconv_utf8 && 975 (zip->entry->zip_flags & ZIP_UTF8_NAME)) 976 archive_entry_copy_symlink_l(entry, p, 977 linkname_length, NULL); 978 if (errno == ENOMEM) { 979 archive_set_error(&a->archive, ENOMEM, 980 "Can't allocate memory for Symlink"); 981 return (ARCHIVE_FATAL); 982 } 983 /* 984 * Since there is no character-set regulation for 985 * symlink name, do not report the conversion error 986 * in an automatic conversion. 987 */ 988 if (sconv != zip->sconv_utf8 || 989 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { 990 archive_set_error(&a->archive, 991 ARCHIVE_ERRNO_FILE_FORMAT, 992 "Symlink cannot be converted " 993 "from %s to current locale.", 994 archive_string_conversion_charset_name( 995 sconv)); 996 ret = ARCHIVE_WARN; 997 } 998 } 999 zip_entry->uncompressed_size = zip_entry->compressed_size = 0; 1000 1001 if (__archive_read_consume(a, linkname_length) < 0) { 1002 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1003 "Read error skipping symlink target name"); 1004 return ARCHIVE_FATAL; 1005 } 1006 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1007 || zip_entry->uncompressed_size > 0) { 1008 /* Set the size only if it's meaningful. */ 1009 archive_entry_set_size(entry, zip_entry->uncompressed_size); 1010 } 1011 zip->entry_bytes_remaining = zip_entry->compressed_size; 1012 1013 /* If there's no body, force read_data() to return EOF immediately. */ 1014 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 1015 && zip->entry_bytes_remaining < 1) 1016 zip->end_of_entry = 1; 1017 1018 /* Set up a more descriptive format name. */ 1019 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", 1020 version / 10, version % 10, 1021 compression_name(zip->entry->compression)); 1022 a->archive.archive_format_name = zip->format_name.s; 1023 1024 return (ret); 1025 } 1026 1027 static int 1028 check_authentication_code(struct archive_read *a, const void *_p) 1029 { 1030 struct zip *zip = (struct zip *)(a->format->data); 1031 1032 /* Check authentication code. */ 1033 if (zip->hctx_valid) { 1034 const void *p; 1035 uint8_t hmac[20]; 1036 size_t hmac_len = 20; 1037 int cmp; 1038 1039 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); 1040 if (_p == NULL) { 1041 /* Read authentication code. */ 1042 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); 1043 if (p == NULL) { 1044 archive_set_error(&a->archive, 1045 ARCHIVE_ERRNO_FILE_FORMAT, 1046 "Truncated ZIP file data"); 1047 return (ARCHIVE_FATAL); 1048 } 1049 } else { 1050 p = _p; 1051 } 1052 cmp = memcmp(hmac, p, AUTH_CODE_SIZE); 1053 __archive_read_consume(a, AUTH_CODE_SIZE); 1054 if (cmp != 0) { 1055 archive_set_error(&a->archive, 1056 ARCHIVE_ERRNO_MISC, 1057 "ZIP bad Authentication code"); 1058 return (ARCHIVE_WARN); 1059 } 1060 } 1061 return (ARCHIVE_OK); 1062 } 1063 1064 /* 1065 * Read "uncompressed" data. There are three cases: 1066 * 1) We know the size of the data. This is always true for the 1067 * seeking reader (we've examined the Central Directory already). 1068 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. 1069 * Info-ZIP seems to do this; we know the size but have to grab 1070 * the CRC from the data descriptor afterwards. 1071 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and 1072 * we have no size information. In this case, we can do pretty 1073 * well by watching for the data descriptor record. The data 1074 * descriptor is 16 bytes and includes a computed CRC that should 1075 * provide a strong check. 1076 * 1077 * TODO: Technically, the PK\007\010 signature is optional. 1078 * In the original spec, the data descriptor contained CRC 1079 * and size fields but had no leading signature. In practice, 1080 * newer writers seem to provide the signature pretty consistently. 1081 * 1082 * For uncompressed data, the PK\007\010 marker seems essential 1083 * to be sure we've actually seen the end of the entry. 1084 * 1085 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets 1086 * zip->end_of_entry if it consumes all of the data. 1087 */ 1088 static int 1089 zip_read_data_none(struct archive_read *a, const void **_buff, 1090 size_t *size, int64_t *offset) 1091 { 1092 struct zip *zip; 1093 const char *buff; 1094 ssize_t bytes_avail; 1095 int r; 1096 1097 (void)offset; /* UNUSED */ 1098 1099 zip = (struct zip *)(a->format->data); 1100 1101 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { 1102 const char *p; 1103 ssize_t grabbing_bytes = 24; 1104 1105 if (zip->hctx_valid) 1106 grabbing_bytes += AUTH_CODE_SIZE; 1107 /* Grab at least 24 bytes. */ 1108 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); 1109 if (bytes_avail < grabbing_bytes) { 1110 /* Zip archives have end-of-archive markers 1111 that are longer than this, so a failure to get at 1112 least 24 bytes really does indicate a truncated 1113 file. */ 1114 archive_set_error(&a->archive, 1115 ARCHIVE_ERRNO_FILE_FORMAT, 1116 "Truncated ZIP file data"); 1117 return (ARCHIVE_FATAL); 1118 } 1119 /* Check for a complete PK\007\010 signature, followed 1120 * by the correct 4-byte CRC. */ 1121 p = buff; 1122 if (zip->hctx_valid) 1123 p += AUTH_CODE_SIZE; 1124 if (p[0] == 'P' && p[1] == 'K' 1125 && p[2] == '\007' && p[3] == '\010' 1126 && (archive_le32dec(p + 4) == zip->entry_crc32 1127 || zip->ignore_crc32 1128 || (zip->hctx_valid 1129 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { 1130 if (zip->entry->flags & LA_USED_ZIP64) { 1131 zip->entry->crc32 = archive_le32dec(p + 4); 1132 zip->entry->compressed_size = 1133 archive_le64dec(p + 8); 1134 zip->entry->uncompressed_size = 1135 archive_le64dec(p + 16); 1136 zip->unconsumed = 24; 1137 } else { 1138 zip->entry->crc32 = archive_le32dec(p + 4); 1139 zip->entry->compressed_size = 1140 archive_le32dec(p + 8); 1141 zip->entry->uncompressed_size = 1142 archive_le32dec(p + 12); 1143 zip->unconsumed = 16; 1144 } 1145 if (zip->hctx_valid) { 1146 r = check_authentication_code(a, buff); 1147 if (r != ARCHIVE_OK) 1148 return (r); 1149 } 1150 zip->end_of_entry = 1; 1151 return (ARCHIVE_OK); 1152 } 1153 /* If not at EOF, ensure we consume at least one byte. */ 1154 ++p; 1155 1156 /* Scan forward until we see where a PK\007\010 signature 1157 * might be. */ 1158 /* Return bytes up until that point. On the next call, 1159 * the code above will verify the data descriptor. */ 1160 while (p < buff + bytes_avail - 4) { 1161 if (p[3] == 'P') { p += 3; } 1162 else if (p[3] == 'K') { p += 2; } 1163 else if (p[3] == '\007') { p += 1; } 1164 else if (p[3] == '\010' && p[2] == '\007' 1165 && p[1] == 'K' && p[0] == 'P') { 1166 if (zip->hctx_valid) 1167 p -= AUTH_CODE_SIZE; 1168 break; 1169 } else { p += 4; } 1170 } 1171 bytes_avail = p - buff; 1172 } else { 1173 if (zip->entry_bytes_remaining == 0) { 1174 zip->end_of_entry = 1; 1175 if (zip->hctx_valid) { 1176 r = check_authentication_code(a, NULL); 1177 if (r != ARCHIVE_OK) 1178 return (r); 1179 } 1180 return (ARCHIVE_OK); 1181 } 1182 /* Grab a bunch of bytes. */ 1183 buff = __archive_read_ahead(a, 1, &bytes_avail); 1184 if (bytes_avail <= 0) { 1185 archive_set_error(&a->archive, 1186 ARCHIVE_ERRNO_FILE_FORMAT, 1187 "Truncated ZIP file data"); 1188 return (ARCHIVE_FATAL); 1189 } 1190 if (bytes_avail > zip->entry_bytes_remaining) 1191 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1192 } 1193 if (zip->tctx_valid || zip->cctx_valid) { 1194 size_t dec_size = bytes_avail; 1195 1196 if (dec_size > zip->decrypted_buffer_size) 1197 dec_size = zip->decrypted_buffer_size; 1198 if (zip->tctx_valid) { 1199 trad_enc_decrypt_update(&zip->tctx, 1200 (const uint8_t *)buff, dec_size, 1201 zip->decrypted_buffer, dec_size); 1202 } else { 1203 size_t dsize = dec_size; 1204 archive_hmac_sha1_update(&zip->hctx, 1205 (const uint8_t *)buff, dec_size); 1206 archive_decrypto_aes_ctr_update(&zip->cctx, 1207 (const uint8_t *)buff, dec_size, 1208 zip->decrypted_buffer, &dsize); 1209 } 1210 bytes_avail = dec_size; 1211 buff = (const char *)zip->decrypted_buffer; 1212 } 1213 *size = bytes_avail; 1214 zip->entry_bytes_remaining -= bytes_avail; 1215 zip->entry_uncompressed_bytes_read += bytes_avail; 1216 zip->entry_compressed_bytes_read += bytes_avail; 1217 zip->unconsumed += bytes_avail; 1218 *_buff = buff; 1219 return (ARCHIVE_OK); 1220 } 1221 1222 #ifdef HAVE_ZLIB_H 1223 static int 1224 zip_deflate_init(struct archive_read *a, struct zip *zip) 1225 { 1226 int r; 1227 1228 /* If we haven't yet read any data, initialize the decompressor. */ 1229 if (!zip->decompress_init) { 1230 if (zip->stream_valid) 1231 r = inflateReset(&zip->stream); 1232 else 1233 r = inflateInit2(&zip->stream, 1234 -15 /* Don't check for zlib header */); 1235 if (r != Z_OK) { 1236 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1237 "Can't initialize ZIP decompression."); 1238 return (ARCHIVE_FATAL); 1239 } 1240 /* Stream structure has been set up. */ 1241 zip->stream_valid = 1; 1242 /* We've initialized decompression for this stream. */ 1243 zip->decompress_init = 1; 1244 } 1245 return (ARCHIVE_OK); 1246 } 1247 1248 static int 1249 zip_read_data_deflate(struct archive_read *a, const void **buff, 1250 size_t *size, int64_t *offset) 1251 { 1252 struct zip *zip; 1253 ssize_t bytes_avail; 1254 const void *compressed_buff, *sp; 1255 int r; 1256 1257 (void)offset; /* UNUSED */ 1258 1259 zip = (struct zip *)(a->format->data); 1260 1261 /* If the buffer hasn't been allocated, allocate it now. */ 1262 if (zip->uncompressed_buffer == NULL) { 1263 zip->uncompressed_buffer_size = 256 * 1024; 1264 zip->uncompressed_buffer 1265 = (unsigned char *)malloc(zip->uncompressed_buffer_size); 1266 if (zip->uncompressed_buffer == NULL) { 1267 archive_set_error(&a->archive, ENOMEM, 1268 "No memory for ZIP decompression"); 1269 return (ARCHIVE_FATAL); 1270 } 1271 } 1272 1273 r = zip_deflate_init(a, zip); 1274 if (r != ARCHIVE_OK) 1275 return (r); 1276 1277 /* 1278 * Note: '1' here is a performance optimization. 1279 * Recall that the decompression layer returns a count of 1280 * available bytes; asking for more than that forces the 1281 * decompressor to combine reads by copying data. 1282 */ 1283 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); 1284 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1285 && bytes_avail > zip->entry_bytes_remaining) { 1286 bytes_avail = (ssize_t)zip->entry_bytes_remaining; 1287 } 1288 if (bytes_avail <= 0) { 1289 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1290 "Truncated ZIP file body"); 1291 return (ARCHIVE_FATAL); 1292 } 1293 1294 if (zip->tctx_valid || zip->cctx_valid) { 1295 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { 1296 size_t buff_remaining = zip->decrypted_buffer_size 1297 - (zip->decrypted_ptr - zip->decrypted_buffer); 1298 1299 if (buff_remaining > (size_t)bytes_avail) 1300 buff_remaining = (size_t)bytes_avail; 1301 1302 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && 1303 zip->entry_bytes_remaining > 0) { 1304 if ((int64_t)(zip->decrypted_bytes_remaining 1305 + buff_remaining) 1306 > zip->entry_bytes_remaining) { 1307 if (zip->entry_bytes_remaining < 1308 (int64_t)zip->decrypted_bytes_remaining) 1309 buff_remaining = 0; 1310 else 1311 buff_remaining = 1312 (size_t)zip->entry_bytes_remaining 1313 - zip->decrypted_bytes_remaining; 1314 } 1315 } 1316 if (buff_remaining > 0) { 1317 if (zip->tctx_valid) { 1318 trad_enc_decrypt_update(&zip->tctx, 1319 compressed_buff, buff_remaining, 1320 zip->decrypted_ptr 1321 + zip->decrypted_bytes_remaining, 1322 buff_remaining); 1323 } else { 1324 size_t dsize = buff_remaining; 1325 archive_decrypto_aes_ctr_update( 1326 &zip->cctx, 1327 compressed_buff, buff_remaining, 1328 zip->decrypted_ptr 1329 + zip->decrypted_bytes_remaining, 1330 &dsize); 1331 } 1332 zip->decrypted_bytes_remaining += buff_remaining; 1333 } 1334 } 1335 bytes_avail = zip->decrypted_bytes_remaining; 1336 compressed_buff = (const char *)zip->decrypted_ptr; 1337 } 1338 1339 /* 1340 * A bug in zlib.h: stream.next_in should be marked 'const' 1341 * but isn't (the library never alters data through the 1342 * next_in pointer, only reads it). The result: this ugly 1343 * cast to remove 'const'. 1344 */ 1345 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; 1346 zip->stream.avail_in = (uInt)bytes_avail; 1347 zip->stream.total_in = 0; 1348 zip->stream.next_out = zip->uncompressed_buffer; 1349 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; 1350 zip->stream.total_out = 0; 1351 1352 r = inflate(&zip->stream, 0); 1353 switch (r) { 1354 case Z_OK: 1355 break; 1356 case Z_STREAM_END: 1357 zip->end_of_entry = 1; 1358 break; 1359 case Z_MEM_ERROR: 1360 archive_set_error(&a->archive, ENOMEM, 1361 "Out of memory for ZIP decompression"); 1362 return (ARCHIVE_FATAL); 1363 default: 1364 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1365 "ZIP decompression failed (%d)", r); 1366 return (ARCHIVE_FATAL); 1367 } 1368 1369 /* Consume as much as the compressor actually used. */ 1370 bytes_avail = zip->stream.total_in; 1371 if (zip->tctx_valid || zip->cctx_valid) { 1372 zip->decrypted_bytes_remaining -= bytes_avail; 1373 if (zip->decrypted_bytes_remaining == 0) 1374 zip->decrypted_ptr = zip->decrypted_buffer; 1375 else 1376 zip->decrypted_ptr += bytes_avail; 1377 } 1378 /* Calculate compressed data as much as we used.*/ 1379 if (zip->hctx_valid) 1380 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); 1381 __archive_read_consume(a, bytes_avail); 1382 zip->entry_bytes_remaining -= bytes_avail; 1383 zip->entry_compressed_bytes_read += bytes_avail; 1384 1385 *size = zip->stream.total_out; 1386 zip->entry_uncompressed_bytes_read += zip->stream.total_out; 1387 *buff = zip->uncompressed_buffer; 1388 1389 if (zip->end_of_entry && zip->hctx_valid) { 1390 r = check_authentication_code(a, NULL); 1391 if (r != ARCHIVE_OK) 1392 return (r); 1393 } 1394 1395 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1396 const char *p; 1397 1398 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { 1399 archive_set_error(&a->archive, 1400 ARCHIVE_ERRNO_FILE_FORMAT, 1401 "Truncated ZIP end-of-file record"); 1402 return (ARCHIVE_FATAL); 1403 } 1404 /* Consume the optional PK\007\010 marker. */ 1405 if (p[0] == 'P' && p[1] == 'K' && 1406 p[2] == '\007' && p[3] == '\010') { 1407 p += 4; 1408 zip->unconsumed = 4; 1409 } 1410 if (zip->entry->flags & LA_USED_ZIP64) { 1411 zip->entry->crc32 = archive_le32dec(p); 1412 zip->entry->compressed_size = archive_le64dec(p + 4); 1413 zip->entry->uncompressed_size = archive_le64dec(p + 12); 1414 zip->unconsumed += 20; 1415 } else { 1416 zip->entry->crc32 = archive_le32dec(p); 1417 zip->entry->compressed_size = archive_le32dec(p + 4); 1418 zip->entry->uncompressed_size = archive_le32dec(p + 8); 1419 zip->unconsumed += 12; 1420 } 1421 } 1422 1423 return (ARCHIVE_OK); 1424 } 1425 #endif 1426 1427 static int 1428 read_decryption_header(struct archive_read *a) 1429 { 1430 struct zip *zip = (struct zip *)(a->format->data); 1431 const char *p; 1432 unsigned int remaining_size; 1433 unsigned int ts; 1434 1435 /* 1436 * Read an initialization vector data field. 1437 */ 1438 p = __archive_read_ahead(a, 2, NULL); 1439 if (p == NULL) 1440 goto truncated; 1441 ts = zip->iv_size; 1442 zip->iv_size = archive_le16dec(p); 1443 __archive_read_consume(a, 2); 1444 if (ts < zip->iv_size) { 1445 free(zip->iv); 1446 zip->iv = NULL; 1447 } 1448 p = __archive_read_ahead(a, zip->iv_size, NULL); 1449 if (p == NULL) 1450 goto truncated; 1451 if (zip->iv == NULL) { 1452 zip->iv = malloc(zip->iv_size); 1453 if (zip->iv == NULL) 1454 goto nomem; 1455 } 1456 memcpy(zip->iv, p, zip->iv_size); 1457 __archive_read_consume(a, zip->iv_size); 1458 1459 /* 1460 * Read a size of remaining decryption header field. 1461 */ 1462 p = __archive_read_ahead(a, 14, NULL); 1463 if (p == NULL) 1464 goto truncated; 1465 remaining_size = archive_le32dec(p); 1466 if (remaining_size < 16 || remaining_size > (1 << 18)) 1467 goto corrupted; 1468 1469 /* Check if format version is supported. */ 1470 if (archive_le16dec(p+4) != 3) { 1471 archive_set_error(&a->archive, 1472 ARCHIVE_ERRNO_FILE_FORMAT, 1473 "Unsupported encryption format version: %u", 1474 archive_le16dec(p+4)); 1475 return (ARCHIVE_FAILED); 1476 } 1477 1478 /* 1479 * Read an encryption algorithm field. 1480 */ 1481 zip->alg_id = archive_le16dec(p+6); 1482 switch (zip->alg_id) { 1483 case 0x6601:/* DES */ 1484 case 0x6602:/* RC2 */ 1485 case 0x6603:/* 3DES 168 */ 1486 case 0x6609:/* 3DES 112 */ 1487 case 0x660E:/* AES 128 */ 1488 case 0x660F:/* AES 192 */ 1489 case 0x6610:/* AES 256 */ 1490 case 0x6702:/* RC2 (version >= 5.2) */ 1491 case 0x6720:/* Blowfish */ 1492 case 0x6721:/* Twofish */ 1493 case 0x6801:/* RC4 */ 1494 /* Suuported encryption algorithm. */ 1495 break; 1496 default: 1497 archive_set_error(&a->archive, 1498 ARCHIVE_ERRNO_FILE_FORMAT, 1499 "Unknown encryption algorithm: %u", zip->alg_id); 1500 return (ARCHIVE_FAILED); 1501 } 1502 1503 /* 1504 * Read a bit length field. 1505 */ 1506 zip->bit_len = archive_le16dec(p+8); 1507 1508 /* 1509 * Read a flags field. 1510 */ 1511 zip->flags = archive_le16dec(p+10); 1512 switch (zip->flags & 0xf000) { 1513 case 0x0001: /* Password is required to decrypt. */ 1514 case 0x0002: /* Certificates only. */ 1515 case 0x0003: /* Password or certificate required to decrypt. */ 1516 break; 1517 default: 1518 archive_set_error(&a->archive, 1519 ARCHIVE_ERRNO_FILE_FORMAT, 1520 "Unknown encryption flag: %u", zip->flags); 1521 return (ARCHIVE_FAILED); 1522 } 1523 if ((zip->flags & 0xf000) == 0 || 1524 (zip->flags & 0xf000) == 0x4000) { 1525 archive_set_error(&a->archive, 1526 ARCHIVE_ERRNO_FILE_FORMAT, 1527 "Unknown encryption flag: %u", zip->flags); 1528 return (ARCHIVE_FAILED); 1529 } 1530 1531 /* 1532 * Read an encrypted random data field. 1533 */ 1534 ts = zip->erd_size; 1535 zip->erd_size = archive_le16dec(p+12); 1536 __archive_read_consume(a, 14); 1537 if ((zip->erd_size & 0xf) != 0 || 1538 (zip->erd_size + 16) > remaining_size || 1539 (zip->erd_size + 16) < zip->erd_size) 1540 goto corrupted; 1541 1542 if (ts < zip->erd_size) { 1543 free(zip->erd); 1544 zip->erd = NULL; 1545 } 1546 p = __archive_read_ahead(a, zip->erd_size, NULL); 1547 if (p == NULL) 1548 goto truncated; 1549 if (zip->erd == NULL) { 1550 zip->erd = malloc(zip->erd_size); 1551 if (zip->erd == NULL) 1552 goto nomem; 1553 } 1554 memcpy(zip->erd, p, zip->erd_size); 1555 __archive_read_consume(a, zip->erd_size); 1556 1557 /* 1558 * Read a reserved data field. 1559 */ 1560 p = __archive_read_ahead(a, 4, NULL); 1561 if (p == NULL) 1562 goto truncated; 1563 /* Reserved data size should be zero. */ 1564 if (archive_le32dec(p) != 0) 1565 goto corrupted; 1566 __archive_read_consume(a, 4); 1567 1568 /* 1569 * Read a password validation data field. 1570 */ 1571 p = __archive_read_ahead(a, 2, NULL); 1572 if (p == NULL) 1573 goto truncated; 1574 ts = zip->v_size; 1575 zip->v_size = archive_le16dec(p); 1576 __archive_read_consume(a, 2); 1577 if ((zip->v_size & 0x0f) != 0 || 1578 (zip->erd_size + zip->v_size + 16) > remaining_size || 1579 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) 1580 goto corrupted; 1581 if (ts < zip->v_size) { 1582 free(zip->v_data); 1583 zip->v_data = NULL; 1584 } 1585 p = __archive_read_ahead(a, zip->v_size, NULL); 1586 if (p == NULL) 1587 goto truncated; 1588 if (zip->v_data == NULL) { 1589 zip->v_data = malloc(zip->v_size); 1590 if (zip->v_data == NULL) 1591 goto nomem; 1592 } 1593 memcpy(zip->v_data, p, zip->v_size); 1594 __archive_read_consume(a, zip->v_size); 1595 1596 p = __archive_read_ahead(a, 4, NULL); 1597 if (p == NULL) 1598 goto truncated; 1599 zip->v_crc32 = archive_le32dec(p); 1600 __archive_read_consume(a, 4); 1601 1602 /*return (ARCHIVE_OK); 1603 * This is not fully implemnted yet.*/ 1604 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1605 "Encrypted file is unsupported"); 1606 return (ARCHIVE_FAILED); 1607 truncated: 1608 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1609 "Truncated ZIP file data"); 1610 return (ARCHIVE_FATAL); 1611 corrupted: 1612 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1613 "Corrupted ZIP file data"); 1614 return (ARCHIVE_FATAL); 1615 nomem: 1616 archive_set_error(&a->archive, ENOMEM, 1617 "No memory for ZIP decryption"); 1618 return (ARCHIVE_FATAL); 1619 } 1620 1621 static int 1622 zip_alloc_decryption_buffer(struct archive_read *a) 1623 { 1624 struct zip *zip = (struct zip *)(a->format->data); 1625 size_t bs = 256 * 1024; 1626 1627 if (zip->decrypted_buffer == NULL) { 1628 zip->decrypted_buffer_size = bs; 1629 zip->decrypted_buffer = malloc(bs); 1630 if (zip->decrypted_buffer == NULL) { 1631 archive_set_error(&a->archive, ENOMEM, 1632 "No memory for ZIP decryption"); 1633 return (ARCHIVE_FATAL); 1634 } 1635 } 1636 zip->decrypted_ptr = zip->decrypted_buffer; 1637 return (ARCHIVE_OK); 1638 } 1639 1640 static int 1641 init_traditional_PKWARE_decryption(struct archive_read *a) 1642 { 1643 struct zip *zip = (struct zip *)(a->format->data); 1644 const void *p; 1645 int retry; 1646 int r; 1647 1648 if (zip->tctx_valid) 1649 return (ARCHIVE_OK); 1650 1651 /* 1652 Read the 12 bytes encryption header stored at 1653 the start of the data area. 1654 */ 1655 #define ENC_HEADER_SIZE 12 1656 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1657 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { 1658 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1659 "Truncated Zip encrypted body: only %jd bytes available", 1660 (intmax_t)zip->entry_bytes_remaining); 1661 return (ARCHIVE_FATAL); 1662 } 1663 1664 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); 1665 if (p == NULL) { 1666 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1667 "Truncated ZIP file data"); 1668 return (ARCHIVE_FATAL); 1669 } 1670 1671 for (retry = 0;; retry++) { 1672 const char *passphrase; 1673 uint8_t crcchk; 1674 1675 passphrase = __archive_read_next_passphrase(a); 1676 if (passphrase == NULL) { 1677 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1678 (retry > 0)? 1679 "Incorrect passphrase": 1680 "Passphrase required for this entry"); 1681 return (ARCHIVE_FAILED); 1682 } 1683 1684 /* 1685 * Initialize ctx for Traditional PKWARE Decyption. 1686 */ 1687 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), 1688 p, ENC_HEADER_SIZE, &crcchk); 1689 if (r == 0 && crcchk == zip->entry->decdat) 1690 break;/* The passphrase is OK. */ 1691 if (retry > 10000) { 1692 /* Avoid infinity loop. */ 1693 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1694 "Too many incorrect passphrases"); 1695 return (ARCHIVE_FAILED); 1696 } 1697 } 1698 1699 __archive_read_consume(a, ENC_HEADER_SIZE); 1700 zip->tctx_valid = 1; 1701 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { 1702 zip->entry_bytes_remaining -= ENC_HEADER_SIZE; 1703 } 1704 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ 1705 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; 1706 zip->decrypted_bytes_remaining = 0; 1707 1708 return (zip_alloc_decryption_buffer(a)); 1709 #undef ENC_HEADER_SIZE 1710 } 1711 1712 static int 1713 init_WinZip_AES_decryption(struct archive_read *a) 1714 { 1715 struct zip *zip = (struct zip *)(a->format->data); 1716 const void *p; 1717 const uint8_t *pv; 1718 size_t key_len, salt_len; 1719 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; 1720 int retry; 1721 int r; 1722 1723 if (zip->cctx_valid || zip->hctx_valid) 1724 return (ARCHIVE_OK); 1725 1726 switch (zip->entry->aes_extra.strength) { 1727 case 1: salt_len = 8; key_len = 16; break; 1728 case 2: salt_len = 12; key_len = 24; break; 1729 case 3: salt_len = 16; key_len = 32; break; 1730 default: goto corrupted; 1731 } 1732 p = __archive_read_ahead(a, salt_len + 2, NULL); 1733 if (p == NULL) 1734 goto truncated; 1735 1736 for (retry = 0;; retry++) { 1737 const char *passphrase; 1738 1739 passphrase = __archive_read_next_passphrase(a); 1740 if (passphrase == NULL) { 1741 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1742 (retry > 0)? 1743 "Incorrect passphrase": 1744 "Passphrase required for this entry"); 1745 return (ARCHIVE_FAILED); 1746 } 1747 memset(derived_key, 0, sizeof(derived_key)); 1748 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), 1749 p, salt_len, 1000, derived_key, key_len * 2 + 2); 1750 if (r != 0) { 1751 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1752 "Decryption is unsupported due to lack of " 1753 "crypto library"); 1754 return (ARCHIVE_FAILED); 1755 } 1756 1757 /* Check password verification value. */ 1758 pv = ((const uint8_t *)p) + salt_len; 1759 if (derived_key[key_len * 2] == pv[0] && 1760 derived_key[key_len * 2 + 1] == pv[1]) 1761 break;/* The passphrase is OK. */ 1762 if (retry > 10000) { 1763 /* Avoid infinity loop. */ 1764 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1765 "Too many incorrect passphrases"); 1766 return (ARCHIVE_FAILED); 1767 } 1768 } 1769 1770 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); 1771 if (r != 0) { 1772 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1773 "Decryption is unsupported due to lack of crypto library"); 1774 return (ARCHIVE_FAILED); 1775 } 1776 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); 1777 if (r != 0) { 1778 archive_decrypto_aes_ctr_release(&zip->cctx); 1779 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1780 "Failed to initialize HMAC-SHA1"); 1781 return (ARCHIVE_FAILED); 1782 } 1783 zip->cctx_valid = zip->hctx_valid = 1; 1784 __archive_read_consume(a, salt_len + 2); 1785 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; 1786 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 1787 && zip->entry_bytes_remaining < 0) 1788 goto corrupted; 1789 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; 1790 zip->decrypted_bytes_remaining = 0; 1791 1792 zip->entry->compression = zip->entry->aes_extra.compression; 1793 return (zip_alloc_decryption_buffer(a)); 1794 1795 truncated: 1796 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1797 "Truncated ZIP file data"); 1798 return (ARCHIVE_FATAL); 1799 corrupted: 1800 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1801 "Corrupted ZIP file data"); 1802 return (ARCHIVE_FATAL); 1803 } 1804 1805 static int 1806 archive_read_format_zip_read_data(struct archive_read *a, 1807 const void **buff, size_t *size, int64_t *offset) 1808 { 1809 int r; 1810 struct zip *zip = (struct zip *)(a->format->data); 1811 1812 if (zip->has_encrypted_entries == 1813 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { 1814 zip->has_encrypted_entries = 0; 1815 } 1816 1817 *offset = zip->entry_uncompressed_bytes_read; 1818 *size = 0; 1819 *buff = NULL; 1820 1821 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ 1822 if (zip->end_of_entry) 1823 return (ARCHIVE_EOF); 1824 1825 /* Return EOF immediately if this is a non-regular file. */ 1826 if (AE_IFREG != (zip->entry->mode & AE_IFMT)) 1827 return (ARCHIVE_EOF); 1828 1829 __archive_read_consume(a, zip->unconsumed); 1830 zip->unconsumed = 0; 1831 1832 if (zip->init_decryption) { 1833 zip->has_encrypted_entries = 1; 1834 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 1835 r = read_decryption_header(a); 1836 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 1837 r = init_WinZip_AES_decryption(a); 1838 else 1839 r = init_traditional_PKWARE_decryption(a); 1840 if (r != ARCHIVE_OK) 1841 return (r); 1842 zip->init_decryption = 0; 1843 } 1844 1845 switch(zip->entry->compression) { 1846 case 0: /* No compression. */ 1847 r = zip_read_data_none(a, buff, size, offset); 1848 break; 1849 #ifdef HAVE_ZLIB_H 1850 case 8: /* Deflate compression. */ 1851 r = zip_read_data_deflate(a, buff, size, offset); 1852 break; 1853 #endif 1854 default: /* Unsupported compression. */ 1855 /* Return a warning. */ 1856 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1857 "Unsupported ZIP compression method (%s)", 1858 compression_name(zip->entry->compression)); 1859 /* We can't decompress this entry, but we will 1860 * be able to skip() it and try the next entry. */ 1861 return (ARCHIVE_FAILED); 1862 break; 1863 } 1864 if (r != ARCHIVE_OK) 1865 return (r); 1866 /* Update checksum */ 1867 if (*size) 1868 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, 1869 (unsigned)*size); 1870 /* If we hit the end, swallow any end-of-data marker. */ 1871 if (zip->end_of_entry) { 1872 /* Check file size, CRC against these values. */ 1873 if (zip->entry->compressed_size != 1874 zip->entry_compressed_bytes_read) { 1875 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1876 "ZIP compressed data is wrong size " 1877 "(read %jd, expected %jd)", 1878 (intmax_t)zip->entry_compressed_bytes_read, 1879 (intmax_t)zip->entry->compressed_size); 1880 return (ARCHIVE_WARN); 1881 } 1882 /* Size field only stores the lower 32 bits of the actual 1883 * size. */ 1884 if ((zip->entry->uncompressed_size & UINT32_MAX) 1885 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { 1886 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1887 "ZIP uncompressed data is wrong size " 1888 "(read %jd, expected %jd)\n", 1889 (intmax_t)zip->entry_uncompressed_bytes_read, 1890 (intmax_t)zip->entry->uncompressed_size); 1891 return (ARCHIVE_WARN); 1892 } 1893 /* Check computed CRC against header */ 1894 if ((!zip->hctx_valid || 1895 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && 1896 zip->entry->crc32 != zip->entry_crc32 1897 && !zip->ignore_crc32) { 1898 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1899 "ZIP bad CRC: 0x%lx should be 0x%lx", 1900 (unsigned long)zip->entry_crc32, 1901 (unsigned long)zip->entry->crc32); 1902 return (ARCHIVE_WARN); 1903 } 1904 } 1905 1906 return (ARCHIVE_OK); 1907 } 1908 1909 static int 1910 archive_read_format_zip_cleanup(struct archive_read *a) 1911 { 1912 struct zip *zip; 1913 struct zip_entry *zip_entry, *next_zip_entry; 1914 1915 zip = (struct zip *)(a->format->data); 1916 #ifdef HAVE_ZLIB_H 1917 if (zip->stream_valid) 1918 inflateEnd(&zip->stream); 1919 free(zip->uncompressed_buffer); 1920 #endif 1921 if (zip->zip_entries) { 1922 zip_entry = zip->zip_entries; 1923 while (zip_entry != NULL) { 1924 next_zip_entry = zip_entry->next; 1925 archive_string_free(&zip_entry->rsrcname); 1926 free(zip_entry); 1927 zip_entry = next_zip_entry; 1928 } 1929 } 1930 free(zip->decrypted_buffer); 1931 if (zip->cctx_valid) 1932 archive_decrypto_aes_ctr_release(&zip->cctx); 1933 if (zip->hctx_valid) 1934 archive_hmac_sha1_cleanup(&zip->hctx); 1935 free(zip->iv); 1936 free(zip->erd); 1937 free(zip->v_data); 1938 archive_string_free(&zip->format_name); 1939 free(zip); 1940 (a->format->data) = NULL; 1941 return (ARCHIVE_OK); 1942 } 1943 1944 static int 1945 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) 1946 { 1947 if (_a && _a->format) { 1948 struct zip * zip = (struct zip *)_a->format->data; 1949 if (zip) { 1950 return zip->has_encrypted_entries; 1951 } 1952 } 1953 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 1954 } 1955 1956 static int 1957 archive_read_format_zip_options(struct archive_read *a, 1958 const char *key, const char *val) 1959 { 1960 struct zip *zip; 1961 int ret = ARCHIVE_FAILED; 1962 1963 zip = (struct zip *)(a->format->data); 1964 if (strcmp(key, "compat-2x") == 0) { 1965 /* Handle filenames as libarchive 2.x */ 1966 zip->init_default_conversion = (val != NULL) ? 1 : 0; 1967 return (ARCHIVE_OK); 1968 } else if (strcmp(key, "hdrcharset") == 0) { 1969 if (val == NULL || val[0] == 0) 1970 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1971 "zip: hdrcharset option needs a character-set name" 1972 ); 1973 else { 1974 zip->sconv = archive_string_conversion_from_charset( 1975 &a->archive, val, 0); 1976 if (zip->sconv != NULL) { 1977 if (strcmp(val, "UTF-8") == 0) 1978 zip->sconv_utf8 = zip->sconv; 1979 ret = ARCHIVE_OK; 1980 } else 1981 ret = ARCHIVE_FATAL; 1982 } 1983 return (ret); 1984 } else if (strcmp(key, "ignorecrc32") == 0) { 1985 /* Mostly useful for testing. */ 1986 if (val == NULL || val[0] == 0) { 1987 zip->crc32func = real_crc32; 1988 zip->ignore_crc32 = 0; 1989 } else { 1990 zip->crc32func = fake_crc32; 1991 zip->ignore_crc32 = 1; 1992 } 1993 return (ARCHIVE_OK); 1994 } else if (strcmp(key, "mac-ext") == 0) { 1995 zip->process_mac_extensions = (val != NULL && val[0] != 0); 1996 return (ARCHIVE_OK); 1997 } 1998 1999 /* Note: The "warn" return is just to inform the options 2000 * supervisor that we didn't handle it. It will generate 2001 * a suitable error if no one used this option. */ 2002 return (ARCHIVE_WARN); 2003 } 2004 2005 int 2006 archive_read_support_format_zip(struct archive *a) 2007 { 2008 int r; 2009 r = archive_read_support_format_zip_streamable(a); 2010 if (r != ARCHIVE_OK) 2011 return r; 2012 return (archive_read_support_format_zip_seekable(a)); 2013 } 2014 2015 /* ------------------------------------------------------------------------ */ 2016 2017 /* 2018 * Streaming-mode support 2019 */ 2020 2021 2022 static int 2023 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) 2024 { 2025 (void)a; /* UNUSED */ 2026 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2027 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2028 } 2029 2030 static int 2031 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) 2032 { 2033 const char *p; 2034 2035 (void)best_bid; /* UNUSED */ 2036 2037 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2038 return (-1); 2039 2040 /* 2041 * Bid of 29 here comes from: 2042 * + 16 bits for "PK", 2043 * + next 16-bit field has 6 options so contributes 2044 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits 2045 * 2046 * So we've effectively verified ~29 total bits of check data. 2047 */ 2048 if (p[0] == 'P' && p[1] == 'K') { 2049 if ((p[2] == '\001' && p[3] == '\002') 2050 || (p[2] == '\003' && p[3] == '\004') 2051 || (p[2] == '\005' && p[3] == '\006') 2052 || (p[2] == '\006' && p[3] == '\006') 2053 || (p[2] == '\007' && p[3] == '\010') 2054 || (p[2] == '0' && p[3] == '0')) 2055 return (29); 2056 } 2057 2058 /* TODO: It's worth looking ahead a little bit for a valid 2059 * PK signature. In particular, that would make it possible 2060 * to read some UUEncoded SFX files or SFX files coming from 2061 * a network socket. */ 2062 2063 return (0); 2064 } 2065 2066 static int 2067 archive_read_format_zip_streamable_read_header(struct archive_read *a, 2068 struct archive_entry *entry) 2069 { 2070 struct zip *zip; 2071 2072 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2073 if (a->archive.archive_format_name == NULL) 2074 a->archive.archive_format_name = "ZIP"; 2075 2076 zip = (struct zip *)(a->format->data); 2077 2078 /* 2079 * It should be sufficient to call archive_read_next_header() for 2080 * a reader to determine if an entry is encrypted or not. If the 2081 * encryption of an entry is only detectable when calling 2082 * archive_read_data(), so be it. We'll do the same check there 2083 * as well. 2084 */ 2085 if (zip->has_encrypted_entries == 2086 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2087 zip->has_encrypted_entries = 0; 2088 2089 /* Make sure we have a zip_entry structure to use. */ 2090 if (zip->zip_entries == NULL) { 2091 zip->zip_entries = malloc(sizeof(struct zip_entry)); 2092 if (zip->zip_entries == NULL) { 2093 archive_set_error(&a->archive, ENOMEM, 2094 "Out of memory"); 2095 return ARCHIVE_FATAL; 2096 } 2097 } 2098 zip->entry = zip->zip_entries; 2099 memset(zip->entry, 0, sizeof(struct zip_entry)); 2100 2101 if (zip->cctx_valid) 2102 archive_decrypto_aes_ctr_release(&zip->cctx); 2103 if (zip->hctx_valid) 2104 archive_hmac_sha1_cleanup(&zip->hctx); 2105 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2106 __archive_read_reset_passphrase(a); 2107 2108 /* Search ahead for the next local file header. */ 2109 __archive_read_consume(a, zip->unconsumed); 2110 zip->unconsumed = 0; 2111 for (;;) { 2112 int64_t skipped = 0; 2113 const char *p, *end; 2114 ssize_t bytes; 2115 2116 p = __archive_read_ahead(a, 4, &bytes); 2117 if (p == NULL) 2118 return (ARCHIVE_FATAL); 2119 end = p + bytes; 2120 2121 while (p + 4 <= end) { 2122 if (p[0] == 'P' && p[1] == 'K') { 2123 if (p[2] == '\003' && p[3] == '\004') { 2124 /* Regular file entry. */ 2125 __archive_read_consume(a, skipped); 2126 return zip_read_local_file_header(a, 2127 entry, zip); 2128 } 2129 2130 /* 2131 * TODO: We cannot restore permissions 2132 * based only on the local file headers. 2133 * Consider scanning the central 2134 * directory and returning additional 2135 * entries for at least directories. 2136 * This would allow us to properly set 2137 * directory permissions. 2138 * 2139 * This won't help us fix symlinks 2140 * and may not help with regular file 2141 * permissions, either. <sigh> 2142 */ 2143 if (p[2] == '\001' && p[3] == '\002') { 2144 return (ARCHIVE_EOF); 2145 } 2146 2147 /* End of central directory? Must be an 2148 * empty archive. */ 2149 if ((p[2] == '\005' && p[3] == '\006') 2150 || (p[2] == '\006' && p[3] == '\006')) 2151 return (ARCHIVE_EOF); 2152 } 2153 ++p; 2154 ++skipped; 2155 } 2156 __archive_read_consume(a, skipped); 2157 } 2158 } 2159 2160 static int 2161 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) 2162 { 2163 struct zip *zip; 2164 int64_t bytes_skipped; 2165 2166 zip = (struct zip *)(a->format->data); 2167 bytes_skipped = __archive_read_consume(a, zip->unconsumed); 2168 zip->unconsumed = 0; 2169 if (bytes_skipped < 0) 2170 return (ARCHIVE_FATAL); 2171 2172 /* If we've already read to end of data, we're done. */ 2173 if (zip->end_of_entry) 2174 return (ARCHIVE_OK); 2175 2176 /* So we know we're streaming... */ 2177 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) 2178 || zip->entry->compressed_size > 0) { 2179 /* We know the compressed length, so we can just skip. */ 2180 bytes_skipped = __archive_read_consume(a, 2181 zip->entry_bytes_remaining); 2182 if (bytes_skipped < 0) 2183 return (ARCHIVE_FATAL); 2184 return (ARCHIVE_OK); 2185 } 2186 2187 if (zip->init_decryption) { 2188 int r; 2189 2190 zip->has_encrypted_entries = 1; 2191 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) 2192 r = read_decryption_header(a); 2193 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) 2194 r = init_WinZip_AES_decryption(a); 2195 else 2196 r = init_traditional_PKWARE_decryption(a); 2197 if (r != ARCHIVE_OK) 2198 return (r); 2199 zip->init_decryption = 0; 2200 } 2201 2202 /* We're streaming and we don't know the length. */ 2203 /* If the body is compressed and we know the format, we can 2204 * find an exact end-of-entry by decompressing it. */ 2205 switch (zip->entry->compression) { 2206 #ifdef HAVE_ZLIB_H 2207 case 8: /* Deflate compression. */ 2208 while (!zip->end_of_entry) { 2209 int64_t offset = 0; 2210 const void *buff = NULL; 2211 size_t size = 0; 2212 int r; 2213 r = zip_read_data_deflate(a, &buff, &size, &offset); 2214 if (r != ARCHIVE_OK) 2215 return (r); 2216 } 2217 return ARCHIVE_OK; 2218 #endif 2219 default: /* Uncompressed or unknown. */ 2220 /* Scan for a PK\007\010 signature. */ 2221 for (;;) { 2222 const char *p, *buff; 2223 ssize_t bytes_avail; 2224 buff = __archive_read_ahead(a, 16, &bytes_avail); 2225 if (bytes_avail < 16) { 2226 archive_set_error(&a->archive, 2227 ARCHIVE_ERRNO_FILE_FORMAT, 2228 "Truncated ZIP file data"); 2229 return (ARCHIVE_FATAL); 2230 } 2231 p = buff; 2232 while (p <= buff + bytes_avail - 16) { 2233 if (p[3] == 'P') { p += 3; } 2234 else if (p[3] == 'K') { p += 2; } 2235 else if (p[3] == '\007') { p += 1; } 2236 else if (p[3] == '\010' && p[2] == '\007' 2237 && p[1] == 'K' && p[0] == 'P') { 2238 if (zip->entry->flags & LA_USED_ZIP64) 2239 __archive_read_consume(a, 2240 p - buff + 24); 2241 else 2242 __archive_read_consume(a, 2243 p - buff + 16); 2244 return ARCHIVE_OK; 2245 } else { p += 4; } 2246 } 2247 __archive_read_consume(a, p - buff); 2248 } 2249 } 2250 } 2251 2252 int 2253 archive_read_support_format_zip_streamable(struct archive *_a) 2254 { 2255 struct archive_read *a = (struct archive_read *)_a; 2256 struct zip *zip; 2257 int r; 2258 2259 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 2260 ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); 2261 2262 zip = (struct zip *)calloc(1, sizeof(*zip)); 2263 if (zip == NULL) { 2264 archive_set_error(&a->archive, ENOMEM, 2265 "Can't allocate zip data"); 2266 return (ARCHIVE_FATAL); 2267 } 2268 2269 /* Streamable reader doesn't support mac extensions. */ 2270 zip->process_mac_extensions = 0; 2271 2272 /* 2273 * Until enough data has been read, we cannot tell about 2274 * any encrypted entries yet. 2275 */ 2276 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 2277 zip->crc32func = real_crc32; 2278 2279 r = __archive_read_register_format(a, 2280 zip, 2281 "zip", 2282 archive_read_format_zip_streamable_bid, 2283 archive_read_format_zip_options, 2284 archive_read_format_zip_streamable_read_header, 2285 archive_read_format_zip_read_data, 2286 archive_read_format_zip_read_data_skip_streamable, 2287 NULL, 2288 archive_read_format_zip_cleanup, 2289 archive_read_support_format_zip_capabilities_streamable, 2290 archive_read_format_zip_has_encrypted_entries); 2291 2292 if (r != ARCHIVE_OK) 2293 free(zip); 2294 return (ARCHIVE_OK); 2295 } 2296 2297 /* ------------------------------------------------------------------------ */ 2298 2299 /* 2300 * Seeking-mode support 2301 */ 2302 2303 static int 2304 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) 2305 { 2306 (void)a; /* UNUSED */ 2307 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | 2308 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); 2309 } 2310 2311 /* 2312 * TODO: This is a performance sink because it forces the read core to 2313 * drop buffered data from the start of file, which will then have to 2314 * be re-read again if this bidder loses. 2315 * 2316 * We workaround this a little by passing in the best bid so far so 2317 * that later bidders can do nothing if they know they'll never 2318 * outbid. But we can certainly do better... 2319 */ 2320 static int 2321 read_eocd(struct zip *zip, const char *p, int64_t current_offset) 2322 { 2323 /* Sanity-check the EOCD we've found. */ 2324 2325 /* This must be the first volume. */ 2326 if (archive_le16dec(p + 4) != 0) 2327 return 0; 2328 /* Central directory must be on this volume. */ 2329 if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) 2330 return 0; 2331 /* All central directory entries must be on this volume. */ 2332 if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) 2333 return 0; 2334 /* Central directory can't extend beyond start of EOCD record. */ 2335 if (archive_le32dec(p + 16) + archive_le32dec(p + 12) 2336 > current_offset) 2337 return 0; 2338 2339 /* Save the central directory location for later use. */ 2340 zip->central_directory_offset = archive_le32dec(p + 16); 2341 2342 /* This is just a tiny bit higher than the maximum 2343 returned by the streaming Zip bidder. This ensures 2344 that the more accurate seeking Zip parser wins 2345 whenever seek is available. */ 2346 return 32; 2347 } 2348 2349 /* 2350 * Examine Zip64 EOCD locator: If it's valid, store the information 2351 * from it. 2352 */ 2353 static void 2354 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) 2355 { 2356 int64_t eocd64_offset; 2357 int64_t eocd64_size; 2358 2359 /* Sanity-check the locator record. */ 2360 2361 /* Central dir must be on first volume. */ 2362 if (archive_le32dec(p + 4) != 0) 2363 return; 2364 /* Must be only a single volume. */ 2365 if (archive_le32dec(p + 16) != 1) 2366 return; 2367 2368 /* Find the Zip64 EOCD record. */ 2369 eocd64_offset = archive_le64dec(p + 8); 2370 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) 2371 return; 2372 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) 2373 return; 2374 /* Make sure we can read all of it. */ 2375 eocd64_size = archive_le64dec(p + 4) + 12; 2376 if (eocd64_size < 56 || eocd64_size > 16384) 2377 return; 2378 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) 2379 return; 2380 2381 /* Sanity-check the EOCD64 */ 2382 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ 2383 return; 2384 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ 2385 return; 2386 /* CD can't be split. */ 2387 if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) 2388 return; 2389 2390 /* Save the central directory offset for later use. */ 2391 zip->central_directory_offset = archive_le64dec(p + 48); 2392 } 2393 2394 static int 2395 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) 2396 { 2397 struct zip *zip = (struct zip *)a->format->data; 2398 int64_t file_size, current_offset; 2399 const char *p; 2400 int i, tail; 2401 2402 /* If someone has already bid more than 32, then avoid 2403 trashing the look-ahead buffers with a seek. */ 2404 if (best_bid > 32) 2405 return (-1); 2406 2407 file_size = __archive_read_seek(a, 0, SEEK_END); 2408 if (file_size <= 0) 2409 return 0; 2410 2411 /* Search last 16k of file for end-of-central-directory 2412 * record (which starts with PK\005\006) */ 2413 tail = (int)zipmin(1024 * 16, file_size); 2414 current_offset = __archive_read_seek(a, -tail, SEEK_END); 2415 if (current_offset < 0) 2416 return 0; 2417 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) 2418 return 0; 2419 /* Boyer-Moore search backwards from the end, since we want 2420 * to match the last EOCD in the file (there can be more than 2421 * one if there is an uncompressed Zip archive as a member 2422 * within this Zip archive). */ 2423 for (i = tail - 22; i > 0;) { 2424 switch (p[i]) { 2425 case 'P': 2426 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2427 int ret = read_eocd(zip, p + i, 2428 current_offset + i); 2429 if (ret > 0) { 2430 /* Zip64 EOCD locator precedes 2431 * regular EOCD if present. */ 2432 if (i >= 20 2433 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { 2434 read_zip64_eocd(a, zip, p + i - 20); 2435 } 2436 return (ret); 2437 } 2438 } 2439 i -= 4; 2440 break; 2441 case 'K': i -= 1; break; 2442 case 005: i -= 2; break; 2443 case 006: i -= 3; break; 2444 default: i -= 4; break; 2445 } 2446 } 2447 return 0; 2448 } 2449 2450 /* The red-black trees are only used in seeking mode to manage 2451 * the in-memory copy of the central directory. */ 2452 2453 static int 2454 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) 2455 { 2456 const struct zip_entry *e1 = (const struct zip_entry *)n1; 2457 const struct zip_entry *e2 = (const struct zip_entry *)n2; 2458 2459 if (e1->local_header_offset > e2->local_header_offset) 2460 return -1; 2461 if (e1->local_header_offset < e2->local_header_offset) 2462 return 1; 2463 return 0; 2464 } 2465 2466 static int 2467 cmp_key(const struct archive_rb_node *n, const void *key) 2468 { 2469 /* This function won't be called */ 2470 (void)n; /* UNUSED */ 2471 (void)key; /* UNUSED */ 2472 return 1; 2473 } 2474 2475 static const struct archive_rb_tree_ops rb_ops = { 2476 &cmp_node, &cmp_key 2477 }; 2478 2479 static int 2480 rsrc_cmp_node(const struct archive_rb_node *n1, 2481 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 return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); 2487 } 2488 2489 static int 2490 rsrc_cmp_key(const struct archive_rb_node *n, const void *key) 2491 { 2492 const struct zip_entry *e = (const struct zip_entry *)n; 2493 return (strcmp((const char *)key, e->rsrcname.s)); 2494 } 2495 2496 static const struct archive_rb_tree_ops rb_rsrc_ops = { 2497 &rsrc_cmp_node, &rsrc_cmp_key 2498 }; 2499 2500 static const char * 2501 rsrc_basename(const char *name, size_t name_length) 2502 { 2503 const char *s, *r; 2504 2505 r = s = name; 2506 for (;;) { 2507 s = memchr(s, '/', name_length - (s - name)); 2508 if (s == NULL) 2509 break; 2510 r = ++s; 2511 } 2512 return (r); 2513 } 2514 2515 static void 2516 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) 2517 { 2518 struct archive_string str; 2519 struct zip_entry *dir; 2520 char *s; 2521 2522 archive_string_init(&str); 2523 archive_strncpy(&str, name, name_length); 2524 for (;;) { 2525 s = strrchr(str.s, '/'); 2526 if (s == NULL) 2527 break; 2528 *s = '\0'; 2529 /* Transfer the parent directory from zip->tree_rsrc RB 2530 * tree to zip->tree RB tree to expose. */ 2531 dir = (struct zip_entry *) 2532 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); 2533 if (dir == NULL) 2534 break; 2535 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); 2536 archive_string_free(&dir->rsrcname); 2537 __archive_rb_tree_insert_node(&zip->tree, &dir->node); 2538 } 2539 archive_string_free(&str); 2540 } 2541 2542 static int 2543 slurp_central_directory(struct archive_read *a, struct zip *zip) 2544 { 2545 ssize_t i; 2546 unsigned found; 2547 int64_t correction; 2548 ssize_t bytes_avail; 2549 const char *p; 2550 2551 /* 2552 * Find the start of the central directory. The end-of-CD 2553 * record has our starting point, but there are lots of 2554 * Zip archives which have had other data prepended to the 2555 * file, which makes the recorded offsets all too small. 2556 * So we search forward from the specified offset until we 2557 * find the real start of the central directory. Then we 2558 * know the correction we need to apply to account for leading 2559 * padding. 2560 */ 2561 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) 2562 return ARCHIVE_FATAL; 2563 2564 found = 0; 2565 while (!found) { 2566 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) 2567 return ARCHIVE_FATAL; 2568 for (found = 0, i = 0; !found && i < bytes_avail - 4;) { 2569 switch (p[i + 3]) { 2570 case 'P': i += 3; break; 2571 case 'K': i += 2; break; 2572 case 001: i += 1; break; 2573 case 002: 2574 if (memcmp(p + i, "PK\001\002", 4) == 0) { 2575 p += i; 2576 found = 1; 2577 } else 2578 i += 4; 2579 break; 2580 case 005: i += 1; break; 2581 case 006: 2582 if (memcmp(p + i, "PK\005\006", 4) == 0) { 2583 p += i; 2584 found = 1; 2585 } else if (memcmp(p + i, "PK\006\006", 4) == 0) { 2586 p += i; 2587 found = 1; 2588 } else 2589 i += 1; 2590 break; 2591 default: i += 4; break; 2592 } 2593 } 2594 __archive_read_consume(a, i); 2595 } 2596 correction = archive_filter_bytes(&a->archive, 0) 2597 - zip->central_directory_offset; 2598 2599 __archive_rb_tree_init(&zip->tree, &rb_ops); 2600 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); 2601 2602 zip->central_directory_entries_total = 0; 2603 while (1) { 2604 struct zip_entry *zip_entry; 2605 size_t filename_length, extra_length, comment_length; 2606 uint32_t external_attributes; 2607 const char *name, *r; 2608 2609 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) 2610 return ARCHIVE_FATAL; 2611 if (memcmp(p, "PK\006\006", 4) == 0 2612 || memcmp(p, "PK\005\006", 4) == 0) { 2613 break; 2614 } else if (memcmp(p, "PK\001\002", 4) != 0) { 2615 archive_set_error(&a->archive, 2616 -1, "Invalid central directory signature"); 2617 return ARCHIVE_FATAL; 2618 } 2619 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) 2620 return ARCHIVE_FATAL; 2621 2622 zip_entry = calloc(1, sizeof(struct zip_entry)); 2623 zip_entry->next = zip->zip_entries; 2624 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; 2625 zip->zip_entries = zip_entry; 2626 zip->central_directory_entries_total++; 2627 2628 /* version = p[4]; */ 2629 zip_entry->system = p[5]; 2630 /* version_required = archive_le16dec(p + 6); */ 2631 zip_entry->zip_flags = archive_le16dec(p + 8); 2632 if (zip_entry->zip_flags 2633 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ 2634 zip->has_encrypted_entries = 1; 2635 } 2636 zip_entry->compression = (char)archive_le16dec(p + 10); 2637 zip_entry->mtime = zip_time(p + 12); 2638 zip_entry->crc32 = archive_le32dec(p + 16); 2639 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) 2640 zip_entry->decdat = p[13]; 2641 else 2642 zip_entry->decdat = p[19]; 2643 zip_entry->compressed_size = archive_le32dec(p + 20); 2644 zip_entry->uncompressed_size = archive_le32dec(p + 24); 2645 filename_length = archive_le16dec(p + 28); 2646 extra_length = archive_le16dec(p + 30); 2647 comment_length = archive_le16dec(p + 32); 2648 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ 2649 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ 2650 external_attributes = archive_le32dec(p + 38); 2651 zip_entry->local_header_offset = 2652 archive_le32dec(p + 42) + correction; 2653 2654 /* If we can't guess the mode, leave it zero here; 2655 when we read the local file header we might get 2656 more information. */ 2657 if (zip_entry->system == 3) { 2658 zip_entry->mode = external_attributes >> 16; 2659 } else if (zip_entry->system == 0) { 2660 // Interpret MSDOS directory bit 2661 if (0x10 == (external_attributes & 0x10)) { 2662 zip_entry->mode = AE_IFDIR | 0775; 2663 } else { 2664 zip_entry->mode = AE_IFREG | 0664; 2665 } 2666 if (0x01 == (external_attributes & 0x01)) { 2667 // Read-only bit; strip write permissions 2668 zip_entry->mode &= 0555; 2669 } 2670 } else { 2671 zip_entry->mode = 0; 2672 } 2673 2674 /* We're done with the regular data; get the filename and 2675 * extra data. */ 2676 __archive_read_consume(a, 46); 2677 p = __archive_read_ahead(a, filename_length + extra_length, 2678 NULL); 2679 if (p == NULL) { 2680 archive_set_error(&a->archive, 2681 ARCHIVE_ERRNO_FILE_FORMAT, 2682 "Truncated ZIP file header"); 2683 return ARCHIVE_FATAL; 2684 } 2685 process_extra(p + filename_length, extra_length, zip_entry); 2686 2687 /* 2688 * Mac resource fork files are stored under the 2689 * "__MACOSX/" directory, so we should check if 2690 * it is. 2691 */ 2692 if (!zip->process_mac_extensions) { 2693 /* Treat every entry as a regular entry. */ 2694 __archive_rb_tree_insert_node(&zip->tree, 2695 &zip_entry->node); 2696 } else { 2697 name = p; 2698 r = rsrc_basename(name, filename_length); 2699 if (filename_length >= 9 && 2700 strncmp("__MACOSX/", name, 9) == 0) { 2701 /* If this file is not a resource fork nor 2702 * a directory. We should treat it as a non 2703 * resource fork file to expose it. */ 2704 if (name[filename_length-1] != '/' && 2705 (r - name < 3 || r[0] != '.' || r[1] != '_')) { 2706 __archive_rb_tree_insert_node( 2707 &zip->tree, &zip_entry->node); 2708 /* Expose its parent directories. */ 2709 expose_parent_dirs(zip, name, 2710 filename_length); 2711 } else { 2712 /* This file is a resource fork file or 2713 * a directory. */ 2714 archive_strncpy(&(zip_entry->rsrcname), 2715 name, filename_length); 2716 __archive_rb_tree_insert_node( 2717 &zip->tree_rsrc, &zip_entry->node); 2718 } 2719 } else { 2720 /* Generate resource fork name to find its 2721 * resource file at zip->tree_rsrc. */ 2722 archive_strcpy(&(zip_entry->rsrcname), 2723 "__MACOSX/"); 2724 archive_strncat(&(zip_entry->rsrcname), 2725 name, r - name); 2726 archive_strcat(&(zip_entry->rsrcname), "._"); 2727 archive_strncat(&(zip_entry->rsrcname), 2728 name + (r - name), 2729 filename_length - (r - name)); 2730 /* Register an entry to RB tree to sort it by 2731 * file offset. */ 2732 __archive_rb_tree_insert_node(&zip->tree, 2733 &zip_entry->node); 2734 } 2735 } 2736 2737 /* Skip the comment too ... */ 2738 __archive_read_consume(a, 2739 filename_length + extra_length + comment_length); 2740 } 2741 2742 return ARCHIVE_OK; 2743 } 2744 2745 static ssize_t 2746 zip_get_local_file_header_size(struct archive_read *a, size_t extra) 2747 { 2748 const char *p; 2749 ssize_t filename_length, extra_length; 2750 2751 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { 2752 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2753 "Truncated ZIP file header"); 2754 return (ARCHIVE_WARN); 2755 } 2756 p += extra; 2757 2758 if (memcmp(p, "PK\003\004", 4) != 0) { 2759 archive_set_error(&a->archive, -1, "Damaged Zip archive"); 2760 return ARCHIVE_WARN; 2761 } 2762 filename_length = archive_le16dec(p + 26); 2763 extra_length = archive_le16dec(p + 28); 2764 2765 return (30 + filename_length + extra_length); 2766 } 2767 2768 static int 2769 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, 2770 struct zip_entry *rsrc) 2771 { 2772 struct zip *zip = (struct zip *)a->format->data; 2773 unsigned char *metadata, *mp; 2774 int64_t offset = archive_filter_bytes(&a->archive, 0); 2775 size_t remaining_bytes, metadata_bytes; 2776 ssize_t hsize; 2777 int ret = ARCHIVE_OK, eof; 2778 2779 switch(rsrc->compression) { 2780 case 0: /* No compression. */ 2781 if (rsrc->uncompressed_size != rsrc->compressed_size) { 2782 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2783 "Malformed OS X metadata entry: inconsistent size"); 2784 return (ARCHIVE_FATAL); 2785 } 2786 #ifdef HAVE_ZLIB_H 2787 case 8: /* Deflate compression. */ 2788 #endif 2789 break; 2790 default: /* Unsupported compression. */ 2791 /* Return a warning. */ 2792 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2793 "Unsupported ZIP compression method (%s)", 2794 compression_name(rsrc->compression)); 2795 /* We can't decompress this entry, but we will 2796 * be able to skip() it and try the next entry. */ 2797 return (ARCHIVE_WARN); 2798 } 2799 2800 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { 2801 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2802 "Mac metadata is too large: %jd > 4M bytes", 2803 (intmax_t)rsrc->uncompressed_size); 2804 return (ARCHIVE_WARN); 2805 } 2806 if (rsrc->compressed_size > (4 * 1024 * 1024)) { 2807 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 2808 "Mac metadata is too large: %jd > 4M bytes", 2809 (intmax_t)rsrc->compressed_size); 2810 return (ARCHIVE_WARN); 2811 } 2812 2813 metadata = malloc((size_t)rsrc->uncompressed_size); 2814 if (metadata == NULL) { 2815 archive_set_error(&a->archive, ENOMEM, 2816 "Can't allocate memory for Mac metadata"); 2817 return (ARCHIVE_FATAL); 2818 } 2819 2820 if (offset < rsrc->local_header_offset) 2821 __archive_read_consume(a, rsrc->local_header_offset - offset); 2822 else if (offset != rsrc->local_header_offset) { 2823 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); 2824 } 2825 2826 hsize = zip_get_local_file_header_size(a, 0); 2827 __archive_read_consume(a, hsize); 2828 2829 remaining_bytes = (size_t)rsrc->compressed_size; 2830 metadata_bytes = (size_t)rsrc->uncompressed_size; 2831 mp = metadata; 2832 eof = 0; 2833 while (!eof && remaining_bytes) { 2834 const unsigned char *p; 2835 ssize_t bytes_avail; 2836 size_t bytes_used; 2837 2838 p = __archive_read_ahead(a, 1, &bytes_avail); 2839 if (p == NULL) { 2840 archive_set_error(&a->archive, 2841 ARCHIVE_ERRNO_FILE_FORMAT, 2842 "Truncated ZIP file header"); 2843 ret = ARCHIVE_WARN; 2844 goto exit_mac_metadata; 2845 } 2846 if ((size_t)bytes_avail > remaining_bytes) 2847 bytes_avail = remaining_bytes; 2848 switch(rsrc->compression) { 2849 case 0: /* No compression. */ 2850 if ((size_t)bytes_avail > metadata_bytes) 2851 bytes_avail = metadata_bytes; 2852 memcpy(mp, p, bytes_avail); 2853 bytes_used = (size_t)bytes_avail; 2854 metadata_bytes -= bytes_used; 2855 mp += bytes_used; 2856 if (metadata_bytes == 0) 2857 eof = 1; 2858 break; 2859 #ifdef HAVE_ZLIB_H 2860 case 8: /* Deflate compression. */ 2861 { 2862 int r; 2863 2864 ret = zip_deflate_init(a, zip); 2865 if (ret != ARCHIVE_OK) 2866 goto exit_mac_metadata; 2867 zip->stream.next_in = 2868 (Bytef *)(uintptr_t)(const void *)p; 2869 zip->stream.avail_in = (uInt)bytes_avail; 2870 zip->stream.total_in = 0; 2871 zip->stream.next_out = mp; 2872 zip->stream.avail_out = (uInt)metadata_bytes; 2873 zip->stream.total_out = 0; 2874 2875 r = inflate(&zip->stream, 0); 2876 switch (r) { 2877 case Z_OK: 2878 break; 2879 case Z_STREAM_END: 2880 eof = 1; 2881 break; 2882 case Z_MEM_ERROR: 2883 archive_set_error(&a->archive, ENOMEM, 2884 "Out of memory for ZIP decompression"); 2885 ret = ARCHIVE_FATAL; 2886 goto exit_mac_metadata; 2887 default: 2888 archive_set_error(&a->archive, 2889 ARCHIVE_ERRNO_MISC, 2890 "ZIP decompression failed (%d)", r); 2891 ret = ARCHIVE_FATAL; 2892 goto exit_mac_metadata; 2893 } 2894 bytes_used = zip->stream.total_in; 2895 metadata_bytes -= zip->stream.total_out; 2896 mp += zip->stream.total_out; 2897 break; 2898 } 2899 #endif 2900 default: 2901 bytes_used = 0; 2902 break; 2903 } 2904 __archive_read_consume(a, bytes_used); 2905 remaining_bytes -= bytes_used; 2906 } 2907 archive_entry_copy_mac_metadata(entry, metadata, 2908 (size_t)rsrc->uncompressed_size - metadata_bytes); 2909 2910 exit_mac_metadata: 2911 __archive_read_seek(a, offset, SEEK_SET); 2912 zip->decompress_init = 0; 2913 free(metadata); 2914 return (ret); 2915 } 2916 2917 static int 2918 archive_read_format_zip_seekable_read_header(struct archive_read *a, 2919 struct archive_entry *entry) 2920 { 2921 struct zip *zip = (struct zip *)a->format->data; 2922 struct zip_entry *rsrc; 2923 int64_t offset; 2924 int r, ret = ARCHIVE_OK; 2925 2926 /* 2927 * It should be sufficient to call archive_read_next_header() for 2928 * a reader to determine if an entry is encrypted or not. If the 2929 * encryption of an entry is only detectable when calling 2930 * archive_read_data(), so be it. We'll do the same check there 2931 * as well. 2932 */ 2933 if (zip->has_encrypted_entries == 2934 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) 2935 zip->has_encrypted_entries = 0; 2936 2937 a->archive.archive_format = ARCHIVE_FORMAT_ZIP; 2938 if (a->archive.archive_format_name == NULL) 2939 a->archive.archive_format_name = "ZIP"; 2940 2941 if (zip->zip_entries == NULL) { 2942 r = slurp_central_directory(a, zip); 2943 if (r != ARCHIVE_OK) 2944 return r; 2945 /* Get first entry whose local header offset is lower than 2946 * other entries in the archive file. */ 2947 zip->entry = 2948 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); 2949 } else if (zip->entry != NULL) { 2950 /* Get next entry in local header offset order. */ 2951 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( 2952 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); 2953 } 2954 2955 if (zip->entry == NULL) 2956 return ARCHIVE_EOF; 2957 2958 if (zip->entry->rsrcname.s) 2959 rsrc = (struct zip_entry *)__archive_rb_tree_find_node( 2960 &zip->tree_rsrc, zip->entry->rsrcname.s); 2961 else 2962 rsrc = NULL; 2963 2964 if (zip->cctx_valid) 2965 archive_decrypto_aes_ctr_release(&zip->cctx); 2966 if (zip->hctx_valid) 2967 archive_hmac_sha1_cleanup(&zip->hctx); 2968 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; 2969 __archive_read_reset_passphrase(a); 2970 2971 /* File entries are sorted by the header offset, we should mostly 2972 * use __archive_read_consume to advance a read point to avoid redundant 2973 * data reading. */ 2974 offset = archive_filter_bytes(&a->archive, 0); 2975 if (offset < zip->entry->local_header_offset) 2976 __archive_read_consume(a, 2977 zip->entry->local_header_offset - offset); 2978 else if (offset != zip->entry->local_header_offset) { 2979 __archive_read_seek(a, zip->entry->local_header_offset, 2980 SEEK_SET); 2981 } 2982 zip->unconsumed = 0; 2983 r = zip_read_local_file_header(a, entry, zip); 2984 if (r != ARCHIVE_OK) 2985 return r; 2986 if (rsrc) { 2987 int ret2 = zip_read_mac_metadata(a, entry, rsrc); 2988 if (ret2 < ret) 2989 ret = ret2; 2990 } 2991 return (ret); 2992 } 2993 2994 /* 2995 * We're going to seek for the next header anyway, so we don't 2996 * need to bother doing anything here. 2997 */ 2998 static int 2999 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) 3000 { 3001 struct zip *zip; 3002 zip = (struct zip *)(a->format->data); 3003 3004 zip->unconsumed = 0; 3005 return (ARCHIVE_OK); 3006 } 3007 3008 int 3009 archive_read_support_format_zip_seekable(struct archive *_a) 3010 { 3011 struct archive_read *a = (struct archive_read *)_a; 3012 struct zip *zip; 3013 int r; 3014 3015 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 3016 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); 3017 3018 zip = (struct zip *)calloc(1, sizeof(*zip)); 3019 if (zip == NULL) { 3020 archive_set_error(&a->archive, ENOMEM, 3021 "Can't allocate zip data"); 3022 return (ARCHIVE_FATAL); 3023 } 3024 3025 #ifdef HAVE_COPYFILE_H 3026 /* Set this by default on Mac OS. */ 3027 zip->process_mac_extensions = 1; 3028 #endif 3029 3030 /* 3031 * Until enough data has been read, we cannot tell about 3032 * any encrypted entries yet. 3033 */ 3034 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; 3035 zip->crc32func = real_crc32; 3036 3037 r = __archive_read_register_format(a, 3038 zip, 3039 "zip", 3040 archive_read_format_zip_seekable_bid, 3041 archive_read_format_zip_options, 3042 archive_read_format_zip_seekable_read_header, 3043 archive_read_format_zip_read_data, 3044 archive_read_format_zip_read_data_skip_seekable, 3045 NULL, 3046 archive_read_format_zip_cleanup, 3047 archive_read_support_format_zip_capabilities_seekable, 3048 archive_read_format_zip_has_encrypted_entries); 3049 3050 if (r != ARCHIVE_OK) 3051 free(zip); 3052 return (ARCHIVE_OK); 3053 } 3054