contrib/libarchive/libarchive/archive_read_support_format_zip.c

   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("$FreeBSD$");
  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 #ifdef HAVE_BZLIB_H
  56 #include <bzlib.h>
  57 #endif
  58 #ifdef HAVE_LZMA_H
  59 #include <lzma.h>
  60 #endif
  61
  62 #include "archive.h"
  63 #include "archive_digest_private.h"
  64 #include "archive_cryptor_private.h"
  65 #include "archive_endian.h"
  66 #include "archive_entry.h"
  67 #include "archive_entry_locale.h"
  68 #include "archive_hmac_private.h"
  69 #include "archive_private.h"
  70 #include "archive_rb.h"
  71 #include "archive_read_private.h"
  72 #include "archive_ppmd8_private.h"
  73
  74 #ifndef HAVE_ZLIB_H
  75 #include "archive_crc32.h"
  76 #endif
  77
  78 struct zip_entry {
  79         struct archive_rb_node  node;
  80         struct zip_entry        *next;
  81         int64_t                 local_header_offset;
  82         int64_t                 compressed_size;
  83         int64_t                 uncompressed_size;
  84         int64_t                 gid;
  85         int64_t                 uid;
  86         struct archive_string   rsrcname;
  87         time_t                  mtime;
  88         time_t                  atime;
  89         time_t                  ctime;
  90         uint32_t                crc32;
  91         uint16_t                mode;
  92         uint16_t                zip_flags; /* From GP Flags Field */
  93         unsigned char           compression;
  94         unsigned char           system; /* From "version written by" */
  95         unsigned char           flags; /* Our extra markers. */
  96         unsigned char           decdat;/* Used for Decryption check */
  97
  98         /* WinZip AES encryption extra field should be available
  99          * when compression is 99. */
 100         struct {
 101                 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */
 102                 unsigned        vendor;
 103 #define AES_VENDOR_AE_1 0x0001
 104 #define AES_VENDOR_AE_2 0x0002
 105                 /* AES encryption strength:
 106                  * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */
 107                 unsigned        strength;
 108                 /* Actual compression method. */
 109                 unsigned char   compression;
 110         }                       aes_extra;
 111 };
 112
 113 struct trad_enc_ctx {
 114         uint32_t        keys[3];
 115 };
 116
 117 /* Bits used in zip_flags. */
 118 #define ZIP_ENCRYPTED   (1 << 0)
 119 #define ZIP_LENGTH_AT_END       (1 << 3)
 120 #define ZIP_STRONG_ENCRYPTED    (1 << 6)
 121 #define ZIP_UTF8_NAME   (1 << 11)
 122 /* See "7.2 Single Password Symmetric Encryption Method"
 123    in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */
 124 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13)
 125
 126 /* Bits used in flags. */
 127 #define LA_USED_ZIP64   (1 << 0)
 128 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1)
 129
 130 /*
 131  * See "WinZip - AES Encryption Information"
 132  *     http://www.winzip.com/aes_info.htm
 133  */
 134 /* Value used in compression method. */
 135 #define WINZIP_AES_ENCRYPTION   99
 136 /* Authentication code size. */
 137 #define AUTH_CODE_SIZE  10
 138 /**/
 139 #define MAX_DERIVED_KEY_BUF_SIZE        (AES_MAX_KEY_SIZE * 2 + 2)
 140
 141 struct zip {
 142         /* Structural information about the archive. */
 143         struct archive_string   format_name;
 144         int64_t                 central_directory_offset;
 145         size_t                  central_directory_entries_total;
 146         size_t                  central_directory_entries_on_this_disk;
 147         int                     has_encrypted_entries;
 148
 149         /* List of entries (seekable Zip only) */
 150         struct zip_entry        *zip_entries;
 151         struct archive_rb_tree  tree;
 152         struct archive_rb_tree  tree_rsrc;
 153
 154         /* Bytes read but not yet consumed via __archive_read_consume() */
 155         size_t                  unconsumed;
 156
 157         /* Information about entry we're currently reading. */
 158         struct zip_entry        *entry;
 159         int64_t                 entry_bytes_remaining;
 160
 161         /* These count the number of bytes actually read for the entry. */
 162         int64_t                 entry_compressed_bytes_read;
 163         int64_t                 entry_uncompressed_bytes_read;
 164
 165         /* Running CRC32 of the decompressed data */
 166         unsigned long           entry_crc32;
 167         unsigned long           (*crc32func)(unsigned long, const void *,
 168                                     size_t);
 169         char                    ignore_crc32;
 170
 171         /* Flags to mark progress of decompression. */
 172         char                    decompress_init;
 173         char                    end_of_entry;
 174
 175         unsigned char           *uncompressed_buffer;
 176         size_t                  uncompressed_buffer_size;
 177
 178 #ifdef HAVE_ZLIB_H
 179         z_stream                stream;
 180         char                    stream_valid;
 181 #endif
 182
 183 #if HAVE_LZMA_H && HAVE_LIBLZMA
 184         lzma_stream             zipx_lzma_stream;
 185         char            zipx_lzma_valid;
 186 #endif
 187
 188 #ifdef HAVE_BZLIB_H
 189         bz_stream               bzstream;
 190         char            bzstream_valid;
 191 #endif
 192
 193         IByteIn                 zipx_ppmd_stream;
 194         ssize_t                 zipx_ppmd_read_compressed;
 195         CPpmd8                  ppmd8;
 196         char                    ppmd8_valid;
 197         char                    ppmd8_stream_failed;
 198
 199         struct archive_string_conv *sconv;
 200         struct archive_string_conv *sconv_default;
 201         struct archive_string_conv *sconv_utf8;
 202         int                     init_default_conversion;
 203         int                     process_mac_extensions;
 204
 205         char                    init_decryption;
 206
 207         /* Decryption buffer. */
 208         /*
 209          * The decrypted data starts at decrypted_ptr and
 210          * extends for decrypted_bytes_remaining.  Decryption
 211          * adds new data to the end of this block, data is returned
 212          * to clients from the beginning.  When the block hits the
 213          * end of decrypted_buffer, it has to be shuffled back to
 214          * the beginning of the buffer.
 215          */
 216         unsigned char           *decrypted_buffer;
 217         unsigned char           *decrypted_ptr;
 218         size_t                  decrypted_buffer_size;
 219         size_t                  decrypted_bytes_remaining;
 220         size_t                  decrypted_unconsumed_bytes;
 221
 222         /* Traditional PKWARE decryption. */
 223         struct trad_enc_ctx     tctx;
 224         char                    tctx_valid;
 225
 226         /* WinZip AES decryption. */
 227         /* Contexts used for AES decryption. */
 228         archive_crypto_ctx      cctx;
 229         char                    cctx_valid;
 230         archive_hmac_sha1_ctx   hctx;
 231         char                    hctx_valid;
 232
 233         /* Strong encryption's decryption header information. */
 234         unsigned                iv_size;
 235         unsigned                alg_id;
 236         unsigned                bit_len;
 237         unsigned                flags;
 238         unsigned                erd_size;
 239         unsigned                v_size;
 240         unsigned                v_crc32;
 241         uint8_t                 *iv;
 242         uint8_t                 *erd;
 243         uint8_t                 *v_data;
 244 };
 245
 246 /* Many systems define min or MIN, but not all. */
 247 #define zipmin(a,b) ((a) < (b) ? (a) : (b))
 248
 249 /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8
 250  * streams inside ZIP files. It has 2 purposes: one is to fetch the next
 251  * compressed byte from the stream, second one is to increase the counter how
 252  * many compressed bytes were read. */
 253 static Byte
 254 ppmd_read(void* p) {
 255         /* Get the handle to current decompression context. */
 256         struct archive_read *a = ((IByteIn*)p)->a;
 257         struct zip *zip = (struct zip*) a->format->data;
 258         ssize_t bytes_avail = 0;
 259
 260         /* Fetch next byte. */
 261         const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail);
 262         if(bytes_avail < 1) {
 263                 zip->ppmd8_stream_failed = 1;
 264                 return 0;
 265         }
 266
 267         __archive_read_consume(a, 1);
 268
 269         /* Increment the counter. */
 270         ++zip->zipx_ppmd_read_compressed;
 271
 272         /* Return the next compressed byte. */
 273         return data[0];
 274 }
 275
 276 /* ------------------------------------------------------------------------ */
 277
 278 /*
 279   Traditional PKWARE Decryption functions.
 280  */
 281
 282 static void
 283 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c)
 284 {
 285         uint8_t t;
 286 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL)
 287
 288         ctx->keys[0] = CRC32(ctx->keys[0], c);
 289         ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1;
 290         t = (ctx->keys[1] >> 24) & 0xff;
 291         ctx->keys[2] = CRC32(ctx->keys[2], t);
 292 #undef CRC32
 293 }
 294
 295 static uint8_t
 296 trad_enc_decrypt_byte(struct trad_enc_ctx *ctx)
 297 {
 298         unsigned temp = ctx->keys[2] | 2;
 299         return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff;
 300 }
 301
 302 static void
 303 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in,
 304     size_t in_len, uint8_t *out, size_t out_len)
 305 {
 306         unsigned i, max;
 307
 308         max = (unsigned)((in_len < out_len)? in_len: out_len);
 309
 310         for (i = 0; i < max; i++) {
 311                 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx);
 312                 out[i] = t;
 313                 trad_enc_update_keys(ctx, t);
 314         }
 315 }
 316
 317 static int
 318 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len,
 319     const uint8_t *key, size_t key_len, uint8_t *crcchk)
 320 {
 321         uint8_t header[12];
 322
 323         if (key_len < 12) {
 324                 *crcchk = 0xff;
 325                 return -1;
 326         }
 327
 328         ctx->keys[0] = 305419896L;
 329         ctx->keys[1] = 591751049L;
 330         ctx->keys[2] = 878082192L;
 331
 332         for (;pw_len; --pw_len)
 333                 trad_enc_update_keys(ctx, *pw++);
 334
 335         trad_enc_decrypt_update(ctx, key, 12, header, 12);
 336         /* Return the last byte for CRC check. */
 337         *crcchk = header[11];
 338         return 0;
 339 }
 340
 341 #if 0
 342 static void
 343 crypt_derive_key_sha1(const void *p, int size, unsigned char *key,
 344     int key_size)
 345 {
 346 #define MD_SIZE 20
 347         archive_sha1_ctx ctx;
 348         unsigned char md1[MD_SIZE];
 349         unsigned char md2[MD_SIZE * 2];
 350         unsigned char mkb[64];
 351         int i;
 352
 353         archive_sha1_init(&ctx);
 354         archive_sha1_update(&ctx, p, size);
 355         archive_sha1_final(&ctx, md1);
 356
 357         memset(mkb, 0x36, sizeof(mkb));
 358         for (i = 0; i < MD_SIZE; i++)
 359                 mkb[i] ^= md1[i];
 360         archive_sha1_init(&ctx);
 361         archive_sha1_update(&ctx, mkb, sizeof(mkb));
 362         archive_sha1_final(&ctx, md2);
 363
 364         memset(mkb, 0x5C, sizeof(mkb));
 365         for (i = 0; i < MD_SIZE; i++)
 366                 mkb[i] ^= md1[i];
 367         archive_sha1_init(&ctx);
 368         archive_sha1_update(&ctx, mkb, sizeof(mkb));
 369         archive_sha1_final(&ctx, md2 + MD_SIZE);
 370
 371         if (key_size > 32)
 372                 key_size = 32;
 373         memcpy(key, md2, key_size);
 374 #undef MD_SIZE
 375 }
 376 #endif
 377
 378 /*
 379  * Common code for streaming or seeking modes.
 380  *
 381  * Includes code to read local file headers, decompress data
 382  * from entry bodies, and common API.
 383  */
 384
 385 static unsigned long
 386 real_crc32(unsigned long crc, const void *buff, size_t len)
 387 {
 388         return crc32(crc, buff, (unsigned int)len);
 389 }
 390
 391 /* Used by "ignorecrc32" option to speed up tests. */
 392 static unsigned long
 393 fake_crc32(unsigned long crc, const void *buff, size_t len)
 394 {
 395         (void)crc; /* UNUSED */
 396         (void)buff; /* UNUSED */
 397         (void)len; /* UNUSED */
 398         return 0;
 399 }
 400
 401 static const struct {
 402         int id;
 403         const char * name;
 404 } compression_methods[] = {
 405         {0, "uncompressed"}, /* The file is stored (no compression) */
 406         {1, "shrinking"}, /* The file is Shrunk */
 407         {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */
 408         {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */
 409         {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */
 410         {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */
 411         {6, "imploded"},  /* The file is Imploded */
 412         {7, "reserved"},  /* Reserved for Tokenizing compression algorithm */
 413         {8, "deflation"}, /* The file is Deflated */
 414         {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */
 415         {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding
 416                            * (old IBM TERSE) */
 417         {11, "reserved"}, /* Reserved by PKWARE */
 418         {12, "bzip"},     /* File is compressed using BZIP2 algorithm */
 419         {13, "reserved"}, /* Reserved by PKWARE */
 420         {14, "lzma"},     /* LZMA (EFS) */
 421         {15, "reserved"}, /* Reserved by PKWARE */
 422         {16, "reserved"}, /* Reserved by PKWARE */
 423         {17, "reserved"}, /* Reserved by PKWARE */
 424         {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */
 425         {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */
 426         {95, "xz"},       /* XZ compressed data */
 427         {96, "jpeg"},     /* JPEG compressed data */
 428         {97, "wav-pack"}, /* WavPack compressed data */
 429         {98, "ppmd-1"},   /* PPMd version I, Rev 1 */
 430         {99, "aes"}       /* WinZip AES encryption  */
 431 };
 432
 433 static const char *
 434 compression_name(const int compression)
 435 {
 436         static const int num_compression_methods =
 437                 sizeof(compression_methods)/sizeof(compression_methods[0]);
 438         int i=0;
 439
 440         while(compression >= 0 && i < num_compression_methods) {
 441                 if (compression_methods[i].id == compression)
 442                         return compression_methods[i].name;
 443                 i++;
 444         }
 445         return "??";
 446 }
 447
 448 /* Convert an MSDOS-style date/time into Unix-style time. */
 449 static time_t
 450 zip_time(const char *p)
 451 {
 452         int msTime, msDate;
 453         struct tm ts;
 454
 455         msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]);
 456         msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]);
 457
 458         memset(&ts, 0, sizeof(ts));
 459         ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */
 460         ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */
 461         ts.tm_mday = msDate & 0x1f; /* Day of month. */
 462         ts.tm_hour = (msTime >> 11) & 0x1f;
 463         ts.tm_min = (msTime >> 5) & 0x3f;
 464         ts.tm_sec = (msTime << 1) & 0x3e;
 465         ts.tm_isdst = -1;
 466         return mktime(&ts);
 467 }
 468
 469 /*
 470  * The extra data is stored as a list of
 471  *      id1+size1+data1 + id2+size2+data2 ...
 472  *  triplets.  id and size are 2 bytes each.
 473  */
 474 static int
 475 process_extra(struct archive_read *a, struct archive_entry *entry,
 476      const char *p, size_t extra_length, struct zip_entry* zip_entry)
 477 {
 478         unsigned offset = 0;
 479         struct zip *zip = (struct zip *)(a->format->data);
 480
 481         if (extra_length == 0) {
 482                 return ARCHIVE_OK;
 483         }
 484
 485         if (extra_length < 4) {
 486                 size_t i = 0;
 487                 /* Some ZIP files may have trailing 0 bytes. Let's check they
 488                  * are all 0 and ignore them instead of returning an error.
 489                  *
 490                  * This is not technically correct, but some ZIP files look
 491                  * like this and other tools support those files - so let's
 492                  * also  support them.
 493                  */
 494                 for (; i < extra_length; i++) {
 495                         if (p[i] != 0) {
 496                                 archive_set_error(&a->archive,
 497                                     ARCHIVE_ERRNO_FILE_FORMAT,
 498                                     "Too-small extra data: "
 499                                     "Need at least 4 bytes, "
 500                                     "but only found %d bytes",
 501                                     (int)extra_length);
 502                                 return ARCHIVE_FAILED;
 503                         }
 504                 }
 505
 506                 return ARCHIVE_OK;
 507         }
 508
 509         while (offset <= extra_length - 4) {
 510                 unsigned short headerid = archive_le16dec(p + offset);
 511                 unsigned short datasize = archive_le16dec(p + offset + 2);
 512
 513                 offset += 4;
 514                 if (offset + datasize > extra_length) {
 515                         archive_set_error(&a->archive,
 516                             ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: "
 517                             "Need %d bytes but only found %d bytes",
 518                             (int)datasize, (int)(extra_length - offset));
 519                         return ARCHIVE_FAILED;
 520                 }
 521 #ifdef DEBUG
 522                 fprintf(stderr, "Header id 0x%04x, length %d\n",
 523                     headerid, datasize);
 524 #endif
 525                 switch (headerid) {
 526                 case 0x0001:
 527                         /* Zip64 extended information extra field. */
 528                         zip_entry->flags |= LA_USED_ZIP64;
 529                         if (zip_entry->uncompressed_size == 0xffffffff) {
 530                                 uint64_t t = 0;
 531                                 if (datasize < 8
 532                                     || (t = archive_le64dec(p + offset)) >
 533                                     INT64_MAX) {
 534                                         archive_set_error(&a->archive,
 535                                             ARCHIVE_ERRNO_FILE_FORMAT,
 536                                             "Malformed 64-bit "
 537                                             "uncompressed size");
 538                                         return ARCHIVE_FAILED;
 539                                 }
 540                                 zip_entry->uncompressed_size = t;
 541                                 offset += 8;
 542                                 datasize -= 8;
 543                         }
 544                         if (zip_entry->compressed_size == 0xffffffff) {
 545                                 uint64_t t = 0;
 546                                 if (datasize < 8
 547                                     || (t = archive_le64dec(p + offset)) >
 548                                     INT64_MAX) {
 549                                         archive_set_error(&a->archive,
 550                                             ARCHIVE_ERRNO_FILE_FORMAT,
 551                                             "Malformed 64-bit "
 552                                             "compressed size");
 553                                         return ARCHIVE_FAILED;
 554                                 }
 555                                 zip_entry->compressed_size = t;
 556                                 offset += 8;
 557                                 datasize -= 8;
 558                         }
 559                         if (zip_entry->local_header_offset == 0xffffffff) {
 560                                 uint64_t t = 0;
 561                                 if (datasize < 8
 562                                     || (t = archive_le64dec(p + offset)) >
 563                                     INT64_MAX) {
 564                                         archive_set_error(&a->archive,
 565                                             ARCHIVE_ERRNO_FILE_FORMAT,
 566                                             "Malformed 64-bit "
 567                                             "local header offset");
 568                                         return ARCHIVE_FAILED;
 569                                 }
 570                                 zip_entry->local_header_offset = t;
 571                                 offset += 8;
 572                                 datasize -= 8;
 573                         }
 574                         /* archive_le32dec(p + offset) gives disk
 575                          * on which file starts, but we don't handle
 576                          * multi-volume Zip files. */
 577                         break;
 578 #ifdef DEBUG
 579                 case 0x0017:
 580                 {
 581                         /* Strong encryption field. */
 582                         if (archive_le16dec(p + offset) == 2) {
 583                                 unsigned algId =
 584                                         archive_le16dec(p + offset + 2);
 585                                 unsigned bitLen =
 586                                         archive_le16dec(p + offset + 4);
 587                                 int      flags =
 588                                         archive_le16dec(p + offset + 6);
 589                                 fprintf(stderr, "algId=0x%04x, bitLen=%u, "
 590                                     "flgas=%d\n", algId, bitLen,flags);
 591                         }
 592                         break;
 593                 }
 594 #endif
 595                 case 0x5455:
 596                 {
 597                         /* Extended time field "UT". */
 598                         int flags;
 599                         if (datasize == 0) {
 600                                 archive_set_error(&a->archive,
 601                                     ARCHIVE_ERRNO_FILE_FORMAT,
 602                                     "Incomplete extended time field");
 603                                 return ARCHIVE_FAILED;
 604                         }
 605                         flags = p[offset];
 606                         offset++;
 607                         datasize--;
 608                         /* Flag bits indicate which dates are present. */
 609                         if (flags & 0x01)
 610                         {
 611 #ifdef DEBUG
 612                                 fprintf(stderr, "mtime: %lld -> %d\n",
 613                                     (long long)zip_entry->mtime,
 614                                     archive_le32dec(p + offset));
 615 #endif
 616                                 if (datasize < 4)
 617                                         break;
 618                                 zip_entry->mtime = archive_le32dec(p + offset);
 619                                 offset += 4;
 620                                 datasize -= 4;
 621                         }
 622                         if (flags & 0x02)
 623                         {
 624                                 if (datasize < 4)
 625                                         break;
 626                                 zip_entry->atime = archive_le32dec(p + offset);
 627                                 offset += 4;
 628                                 datasize -= 4;
 629                         }
 630                         if (flags & 0x04)
 631                         {
 632                                 if (datasize < 4)
 633                                         break;
 634                                 zip_entry->ctime = archive_le32dec(p + offset);
 635                                 offset += 4;
 636                                 datasize -= 4;
 637                         }
 638                         break;
 639                 }
 640                 case 0x5855:
 641                 {
 642                         /* Info-ZIP Unix Extra Field (old version) "UX". */
 643                         if (datasize >= 8) {
 644                                 zip_entry->atime = archive_le32dec(p + offset);
 645                                 zip_entry->mtime =
 646                                     archive_le32dec(p + offset + 4);
 647                         }
 648                         if (datasize >= 12) {
 649                                 zip_entry->uid =
 650                                     archive_le16dec(p + offset + 8);
 651                                 zip_entry->gid =
 652                                     archive_le16dec(p + offset + 10);
 653                         }
 654                         break;
 655                 }
 656                 case 0x6c78:
 657                 {
 658                         /* Experimental 'xl' field */
 659                         /*
 660                          * Introduced Dec 2013 to provide a way to
 661                          * include external file attributes (and other
 662                          * fields that ordinarily appear only in
 663                          * central directory) in local file header.
 664                          * This provides file type and permission
 665                          * information necessary to support full
 666                          * streaming extraction.  Currently being
 667                          * discussed with other Zip developers
 668                          * ... subject to change.
 669                          *
 670                          * Format:
 671                          *  The field starts with a bitmap that specifies
 672                          *  which additional fields are included.  The
 673                          *  bitmap is variable length and can be extended in
 674                          *  the future.
 675                          *
 676                          *  n bytes - feature bitmap: first byte has low-order
 677                          *    7 bits.  If high-order bit is set, a subsequent
 678                          *    byte holds the next 7 bits, etc.
 679                          *
 680                          *  if bitmap & 1, 2 byte "version made by"
 681                          *  if bitmap & 2, 2 byte "internal file attributes"
 682                          *  if bitmap & 4, 4 byte "external file attributes"
 683                          *  if bitmap & 8, 2 byte comment length + n byte
 684                          *  comment
 685                          */
 686                         int bitmap, bitmap_last;
 687
 688                         if (datasize < 1)
 689                                 break;
 690                         bitmap_last = bitmap = 0xff & p[offset];
 691                         offset += 1;
 692                         datasize -= 1;
 693
 694                         /* We only support first 7 bits of bitmap; skip rest. */
 695                         while ((bitmap_last & 0x80) != 0
 696                             && datasize >= 1) {
 697                                 bitmap_last = p[offset];
 698                                 offset += 1;
 699                                 datasize -= 1;
 700                         }
 701
 702                         if (bitmap & 1) {
 703                                 /* 2 byte "version made by" */
 704                                 if (datasize < 2)
 705                                         break;
 706                                 zip_entry->system
 707                                     = archive_le16dec(p + offset) >> 8;
 708                                 offset += 2;
 709                                 datasize -= 2;
 710                         }
 711                         if (bitmap & 2) {
 712                                 /* 2 byte "internal file attributes" */
 713                                 uint32_t internal_attributes;
 714                                 if (datasize < 2)
 715                                         break;
 716                                 internal_attributes
 717                                     = archive_le16dec(p + offset);
 718                                 /* Not used by libarchive at present. */
 719                                 (void)internal_attributes; /* UNUSED */
 720                                 offset += 2;
 721                                 datasize -= 2;
 722                         }
 723                         if (bitmap & 4) {
 724                                 /* 4 byte "external file attributes" */
 725                                 uint32_t external_attributes;
 726                                 if (datasize < 4)
 727                                         break;
 728                                 external_attributes
 729                                     = archive_le32dec(p + offset);
 730                                 if (zip_entry->system == 3) {
 731                                         zip_entry->mode
 732                                             = external_attributes >> 16;
 733                                 } else if (zip_entry->system == 0) {
 734                                         // Interpret MSDOS directory bit
 735                                         if (0x10 == (external_attributes &
 736                                             0x10)) {
 737                                                 zip_entry->mode =
 738                                                     AE_IFDIR | 0775;
 739                                         } else {
 740                                                 zip_entry->mode =
 741                                                     AE_IFREG | 0664;
 742                                         }
 743                                         if (0x01 == (external_attributes &
 744                                             0x01)) {
 745                                                 /* Read-only bit;
 746                                                  * strip write permissions */
 747                                                 zip_entry->mode &= 0555;
 748                                         }
 749                                 } else {
 750                                         zip_entry->mode = 0;
 751                                 }
 752                                 offset += 4;
 753                                 datasize -= 4;
 754                         }
 755                         if (bitmap & 8) {
 756                                 /* 2 byte comment length + comment */
 757                                 uint32_t comment_length;
 758                                 if (datasize < 2)
 759                                         break;
 760                                 comment_length
 761                                     = archive_le16dec(p + offset);
 762                                 offset += 2;
 763                                 datasize -= 2;
 764
 765                                 if (datasize < comment_length)
 766                                         break;
 767                                 /* Comment is not supported by libarchive */
 768                                 offset += comment_length;
 769                                 datasize -= comment_length;
 770                         }
 771                         break;
 772                 }
 773                 case 0x7075:
 774                 {
 775                         /* Info-ZIP Unicode Path Extra Field. */
 776                         if (datasize < 5 || entry == NULL)
 777                                 break;
 778                         offset += 5;
 779                         datasize -= 5;
 780
 781                         /* The path name in this field is always encoded
 782                          * in UTF-8. */
 783                         if (zip->sconv_utf8 == NULL) {
 784                                 zip->sconv_utf8 =
 785                                         archive_string_conversion_from_charset(
 786                                         &a->archive, "UTF-8", 1);
 787                                 /* If the converter from UTF-8 is not
 788                                  * available, then the path name from the main
 789                                  * field will more likely be correct. */
 790                                 if (zip->sconv_utf8 == NULL)
 791                                         break;
 792                         }
 793
 794                         /* Make sure the CRC32 of the filename matches. */
 795                         if (!zip->ignore_crc32) {
 796                                 const char *cp = archive_entry_pathname(entry);
 797                                 if (cp) {
 798                                         unsigned long file_crc =
 799                                             zip->crc32func(0, cp, strlen(cp));
 800                                         unsigned long utf_crc =
 801                                             archive_le32dec(p + offset - 4);
 802                                         if (file_crc != utf_crc) {
 803 #ifdef DEBUG
 804                                                 fprintf(stderr,
 805                                                     "CRC filename mismatch; "
 806                                                     "CDE is %lx, but UTF8 "
 807                                                     "is outdated with %lx\n",
 808                                                     file_crc, utf_crc);
 809 #endif
 810                                                 break;
 811                                         }
 812                                 }
 813                         }
 814
 815                         if (archive_entry_copy_pathname_l(entry,
 816                             p + offset, datasize, zip->sconv_utf8) != 0) {
 817                                 /* Ignore the error, and fallback to the path
 818                                  * name from the main field. */
 819 #ifdef DEBUG
 820                                 fprintf(stderr, "Failed to read the ZIP "
 821                                     "0x7075 extra field path.\n");
 822 #endif
 823                         }
 824                         break;
 825                 }
 826                 case 0x7855:
 827                         /* Info-ZIP Unix Extra Field (type 2) "Ux". */
 828 #ifdef DEBUG
 829                         fprintf(stderr, "uid %d gid %d\n",
 830                             archive_le16dec(p + offset),
 831                             archive_le16dec(p + offset + 2));
 832 #endif
 833                         if (datasize >= 2)
 834                                 zip_entry->uid = archive_le16dec(p + offset);
 835                         if (datasize >= 4)
 836                                 zip_entry->gid =
 837                                     archive_le16dec(p + offset + 2);
 838                         break;
 839                 case 0x7875:
 840                 {
 841                         /* Info-Zip Unix Extra Field (type 3) "ux". */
 842                         int uidsize = 0, gidsize = 0;
 843
 844                         /* TODO: support arbitrary uidsize/gidsize. */
 845                         if (datasize >= 1 && p[offset] == 1) {/* version=1 */
 846                                 if (datasize >= 4) {
 847                                         /* get a uid size. */
 848                                         uidsize = 0xff & (int)p[offset+1];
 849                                         if (uidsize == 2)
 850                                                 zip_entry->uid =
 851                                                     archive_le16dec(
 852                                                         p + offset + 2);
 853                                         else if (uidsize == 4 && datasize >= 6)
 854                                                 zip_entry->uid =
 855                                                     archive_le32dec(
 856                                                         p + offset + 2);
 857                                 }
 858                                 if (datasize >= (2 + uidsize + 3)) {
 859                                         /* get a gid size. */
 860                                         gidsize = 0xff &
 861                                             (int)p[offset+2+uidsize];
 862                                         if (gidsize == 2)
 863                                                 zip_entry->gid =
 864                                                     archive_le16dec(
 865                                                         p+offset+2+uidsize+1);
 866                                         else if (gidsize == 4 &&
 867                                             datasize >= (2 + uidsize + 5))
 868                                                 zip_entry->gid =
 869                                                     archive_le32dec(
 870                                                         p+offset+2+uidsize+1);
 871                                 }
 872                         }
 873                         break;
 874                 }
 875                 case 0x9901:
 876                         /* WinZip AES extra data field. */
 877                         if (datasize < 6) {
 878                                 archive_set_error(&a->archive,
 879                                     ARCHIVE_ERRNO_FILE_FORMAT,
 880                                     "Incomplete AES field");
 881                                 return ARCHIVE_FAILED;
 882                         }
 883                         if (p[offset + 2] == 'A' && p[offset + 3] == 'E') {
 884                                 /* Vendor version. */
 885                                 zip_entry->aes_extra.vendor =
 886                                     archive_le16dec(p + offset);
 887                                 /* AES encryption strength. */
 888                                 zip_entry->aes_extra.strength = p[offset + 4];
 889                                 /* Actual compression method. */
 890                                 zip_entry->aes_extra.compression =
 891                                     p[offset + 5];
 892                         }
 893                         break;
 894                 default:
 895                         break;
 896                 }
 897                 offset += datasize;
 898         }
 899         return ARCHIVE_OK;
 900 }
 901
 902 #if HAVE_LZMA_H && HAVE_LIBLZMA
 903 /*
 904  * Auxiliary function to uncompress data chunk from zipx archive
 905  * (zip with lzma compression).
 906  */
 907 static int
 908 zipx_lzma_uncompress_buffer(const char *compressed_buffer,
 909         size_t compressed_buffer_size,
 910         char *uncompressed_buffer,
 911         size_t uncompressed_buffer_size)
 912 {
 913         int status = ARCHIVE_FATAL;
 914         // length of 'lzma properties data' in lzma compressed
 915         // data segment (stream) inside zip archive
 916         const size_t lzma_params_length = 5;
 917         // offset of 'lzma properties data' from the beginning of lzma stream
 918         const size_t lzma_params_offset = 4;
 919         // end position of 'lzma properties data' in lzma stream
 920         const size_t lzma_params_end = lzma_params_offset + lzma_params_length;
 921         if (compressed_buffer == NULL ||
 922                         compressed_buffer_size < lzma_params_end ||
 923                         uncompressed_buffer == NULL)
 924                 return status;
 925
 926         // prepare header for lzma_alone_decoder to replace zipx header
 927         // (see comments in 'zipx_lzma_alone_init' for justification)
 928 #pragma pack(push)
 929 #pragma pack(1)
 930         struct _alone_header
 931         {
 932                 uint8_t bytes[5]; // lzma_params_length
 933                 uint64_t uncompressed_size;
 934         } alone_header;
 935 #pragma pack(pop)
 936         // copy 'lzma properties data' blob
 937         memcpy(&alone_header.bytes[0], compressed_buffer + lzma_params_offset,
 938                 lzma_params_length);
 939         alone_header.uncompressed_size = UINT64_MAX;
 940
 941         // prepare new compressed buffer, see 'zipx_lzma_alone_init' for details
 942         const size_t lzma_alone_buffer_size =
 943                 compressed_buffer_size - lzma_params_end + sizeof(alone_header);
 944         unsigned char *lzma_alone_compressed_buffer =
 945                 (unsigned char*) malloc(lzma_alone_buffer_size);
 946         if (lzma_alone_compressed_buffer == NULL)
 947                 return status;
 948         // copy lzma_alone header into new buffer
 949         memcpy(lzma_alone_compressed_buffer, (void*) &alone_header,
 950                 sizeof(alone_header));
 951         // copy compressed data into new buffer
 952         memcpy(lzma_alone_compressed_buffer + sizeof(alone_header),
 953                 compressed_buffer + lzma_params_end,
 954                 compressed_buffer_size - lzma_params_end);
 955
 956         // create and fill in lzma_alone_decoder stream
 957         lzma_stream stream = LZMA_STREAM_INIT;
 958         lzma_ret ret = lzma_alone_decoder(&stream, UINT64_MAX);
 959         if (ret == LZMA_OK)
 960         {
 961                 stream.next_in = lzma_alone_compressed_buffer;
 962                 stream.avail_in = lzma_alone_buffer_size;
 963                 stream.total_in = 0;
 964                 stream.next_out = (unsigned char*)uncompressed_buffer;
 965                 stream.avail_out = uncompressed_buffer_size;
 966                 stream.total_out = 0;
 967                 ret = lzma_code(&stream, LZMA_RUN);
 968                 if (ret == LZMA_OK || ret == LZMA_STREAM_END)
 969                         status = ARCHIVE_OK;
 970         }
 971         lzma_end(&stream);
 972         free(lzma_alone_compressed_buffer);
 973         return status;
 974 }
 975 #endif
 976
 977 /*
 978  * Assumes file pointer is at beginning of local file header.
 979  */
 980 static int
 981 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry,
 982     struct zip *zip)
 983 {
 984         const char *p;
 985         const void *h;
 986         const wchar_t *wp;
 987         const char *cp;
 988         size_t len, filename_length, extra_length;
 989         struct archive_string_conv *sconv;
 990         struct zip_entry *zip_entry = zip->entry;
 991         struct zip_entry zip_entry_central_dir;
 992         int ret = ARCHIVE_OK;
 993         char version;
 994
 995         /* Save a copy of the original for consistency checks. */
 996         zip_entry_central_dir = *zip_entry;
 997
 998         zip->decompress_init = 0;
 999         zip->end_of_entry = 0;
1000         zip->entry_uncompressed_bytes_read = 0;
1001         zip->entry_compressed_bytes_read = 0;
1002         zip->entry_crc32 = zip->crc32func(0, NULL, 0);
1003
1004         /* Setup default conversion. */
1005         if (zip->sconv == NULL && !zip->init_default_conversion) {
1006                 zip->sconv_default =
1007                     archive_string_default_conversion_for_read(&(a->archive));
1008                 zip->init_default_conversion = 1;
1009         }
1010
1011         if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) {
1012                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1013                     "Truncated ZIP file header");
1014                 return (ARCHIVE_FATAL);
1015         }
1016
1017         if (memcmp(p, "PK\003\004", 4) != 0) {
1018                 archive_set_error(&a->archive, -1, "Damaged Zip archive");
1019                 return ARCHIVE_FATAL;
1020         }
1021         version = p[4];
1022         zip_entry->system = p[5];
1023         zip_entry->zip_flags = archive_le16dec(p + 6);
1024         if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) {
1025                 zip->has_encrypted_entries = 1;
1026                 archive_entry_set_is_data_encrypted(entry, 1);
1027                 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED &&
1028                         zip_entry->zip_flags & ZIP_ENCRYPTED &&
1029                         zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) {
1030                         archive_entry_set_is_metadata_encrypted(entry, 1);
1031                         return ARCHIVE_FATAL;
1032                 }
1033         }
1034         zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED);
1035         zip_entry->compression = (char)archive_le16dec(p + 8);
1036         zip_entry->mtime = zip_time(p + 10);
1037         zip_entry->crc32 = archive_le32dec(p + 14);
1038         if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1039                 zip_entry->decdat = p[11];
1040         else
1041                 zip_entry->decdat = p[17];
1042         zip_entry->compressed_size = archive_le32dec(p + 18);
1043         zip_entry->uncompressed_size = archive_le32dec(p + 22);
1044         filename_length = archive_le16dec(p + 26);
1045         extra_length = archive_le16dec(p + 28);
1046
1047         __archive_read_consume(a, 30);
1048
1049         /* Read the filename. */
1050         if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) {
1051                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1052                     "Truncated ZIP file header");
1053                 return (ARCHIVE_FATAL);
1054         }
1055         if (zip_entry->zip_flags & ZIP_UTF8_NAME) {
1056                 /* The filename is stored to be UTF-8. */
1057                 if (zip->sconv_utf8 == NULL) {
1058                         zip->sconv_utf8 =
1059                             archive_string_conversion_from_charset(
1060                                 &a->archive, "UTF-8", 1);
1061                         if (zip->sconv_utf8 == NULL)
1062                                 return (ARCHIVE_FATAL);
1063                 }
1064                 sconv = zip->sconv_utf8;
1065         } else if (zip->sconv != NULL)
1066                 sconv = zip->sconv;
1067         else
1068                 sconv = zip->sconv_default;
1069
1070         if (archive_entry_copy_pathname_l(entry,
1071             h, filename_length, sconv) != 0) {
1072                 if (errno == ENOMEM) {
1073                         archive_set_error(&a->archive, ENOMEM,
1074                             "Can't allocate memory for Pathname");
1075                         return (ARCHIVE_FATAL);
1076                 }
1077                 archive_set_error(&a->archive,
1078                     ARCHIVE_ERRNO_FILE_FORMAT,
1079                     "Pathname cannot be converted "
1080                     "from %s to current locale.",
1081                     archive_string_conversion_charset_name(sconv));
1082                 ret = ARCHIVE_WARN;
1083         }
1084         __archive_read_consume(a, filename_length);
1085
1086         /* Read the extra data. */
1087         if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) {
1088                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1089                     "Truncated ZIP file header");
1090                 return (ARCHIVE_FATAL);
1091         }
1092
1093         if (ARCHIVE_OK != process_extra(a, entry, h, extra_length,
1094             zip_entry)) {
1095                 return ARCHIVE_FATAL;
1096         }
1097         __archive_read_consume(a, extra_length);
1098
1099         /* Work around a bug in Info-Zip: When reading from a pipe, it
1100          * stats the pipe instead of synthesizing a file entry. */
1101         if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) {
1102                 zip_entry->mode &= ~ AE_IFMT;
1103                 zip_entry->mode |= AE_IFREG;
1104         }
1105
1106         /* If the mode is totally empty, set some sane default. */
1107         if (zip_entry->mode == 0) {
1108                 zip_entry->mode |= 0664;
1109         }
1110
1111         /* Windows archivers sometimes use backslash as the directory
1112          * separator. Normalize to slash. */
1113         if (zip_entry->system == 0 &&
1114             (wp = archive_entry_pathname_w(entry)) != NULL) {
1115                 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) {
1116                         size_t i;
1117                         struct archive_wstring s;
1118                         archive_string_init(&s);
1119                         archive_wstrcpy(&s, wp);
1120                         for (i = 0; i < archive_strlen(&s); i++) {
1121                                 if (s.s[i] == '\\')
1122                                         s.s[i] = '/';
1123                         }
1124                         archive_entry_copy_pathname_w(entry, s.s);
1125                         archive_wstring_free(&s);
1126                 }
1127         }
1128
1129         /* Make sure that entries with a trailing '/' are marked as directories
1130          * even if the External File Attributes contains bogus values.  If this
1131          * is not a directory and there is no type, assume a regular file. */
1132         if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) {
1133                 int has_slash;
1134
1135                 wp = archive_entry_pathname_w(entry);
1136                 if (wp != NULL) {
1137                         len = wcslen(wp);
1138                         has_slash = len > 0 && wp[len - 1] == L'/';
1139                 } else {
1140                         cp = archive_entry_pathname(entry);
1141                         len = (cp != NULL)?strlen(cp):0;
1142                         has_slash = len > 0 && cp[len - 1] == '/';
1143                 }
1144                 /* Correct file type as needed. */
1145                 if (has_slash) {
1146                         zip_entry->mode &= ~AE_IFMT;
1147                         zip_entry->mode |= AE_IFDIR;
1148                         zip_entry->mode |= 0111;
1149                 } else if ((zip_entry->mode & AE_IFMT) == 0) {
1150                         zip_entry->mode |= AE_IFREG;
1151                 }
1152         }
1153
1154         /* Make sure directories end in '/' */
1155         if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) {
1156                 wp = archive_entry_pathname_w(entry);
1157                 if (wp != NULL) {
1158                         len = wcslen(wp);
1159                         if (len > 0 && wp[len - 1] != L'/') {
1160                                 struct archive_wstring s;
1161                                 archive_string_init(&s);
1162                                 archive_wstrcat(&s, wp);
1163                                 archive_wstrappend_wchar(&s, L'/');
1164                                 archive_entry_copy_pathname_w(entry, s.s);
1165                                 archive_wstring_free(&s);
1166                         }
1167                 } else {
1168                         cp = archive_entry_pathname(entry);
1169                         len = (cp != NULL)?strlen(cp):0;
1170                         if (len > 0 && cp[len - 1] != '/') {
1171                                 struct archive_string s;
1172                                 archive_string_init(&s);
1173                                 archive_strcat(&s, cp);
1174                                 archive_strappend_char(&s, '/');
1175                                 archive_entry_set_pathname(entry, s.s);
1176                                 archive_string_free(&s);
1177                         }
1178                 }
1179         }
1180
1181         if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) {
1182                 /* If this came from the central dir, its size info
1183                  * is definitive, so ignore the length-at-end flag. */
1184                 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END;
1185                 /* If local header is missing a value, use the one from
1186                    the central directory.  If both have it, warn about
1187                    mismatches. */
1188                 if (zip_entry->crc32 == 0) {
1189                         zip_entry->crc32 = zip_entry_central_dir.crc32;
1190                 } else if (!zip->ignore_crc32
1191                     && zip_entry->crc32 != zip_entry_central_dir.crc32) {
1192                         archive_set_error(&a->archive,
1193                             ARCHIVE_ERRNO_FILE_FORMAT,
1194                             "Inconsistent CRC32 values");
1195                         ret = ARCHIVE_WARN;
1196                 }
1197                 if (zip_entry->compressed_size == 0) {
1198                         zip_entry->compressed_size
1199                             = zip_entry_central_dir.compressed_size;
1200                 } else if (zip_entry->compressed_size
1201                     != zip_entry_central_dir.compressed_size) {
1202                         archive_set_error(&a->archive,
1203                             ARCHIVE_ERRNO_FILE_FORMAT,
1204                             "Inconsistent compressed size: "
1205                             "%jd in central directory, %jd in local header",
1206                             (intmax_t)zip_entry_central_dir.compressed_size,
1207                             (intmax_t)zip_entry->compressed_size);
1208                         ret = ARCHIVE_WARN;
1209                 }
1210                 if (zip_entry->uncompressed_size == 0) {
1211                         zip_entry->uncompressed_size
1212                             = zip_entry_central_dir.uncompressed_size;
1213                 } else if (zip_entry->uncompressed_size
1214                     != zip_entry_central_dir.uncompressed_size) {
1215                         archive_set_error(&a->archive,
1216                             ARCHIVE_ERRNO_FILE_FORMAT,
1217                             "Inconsistent uncompressed size: "
1218                             "%jd in central directory, %jd in local header",
1219                             (intmax_t)zip_entry_central_dir.uncompressed_size,
1220                             (intmax_t)zip_entry->uncompressed_size);
1221                         ret = ARCHIVE_WARN;
1222                 }
1223         }
1224
1225         /* Populate some additional entry fields: */
1226         archive_entry_set_mode(entry, zip_entry->mode);
1227         archive_entry_set_uid(entry, zip_entry->uid);
1228         archive_entry_set_gid(entry, zip_entry->gid);
1229         archive_entry_set_mtime(entry, zip_entry->mtime, 0);
1230         archive_entry_set_ctime(entry, zip_entry->ctime, 0);
1231         archive_entry_set_atime(entry, zip_entry->atime, 0);
1232
1233         if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) {
1234                 size_t linkname_length;
1235
1236                 if (zip_entry->compressed_size > 64 * 1024) {
1237                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1238                             "Zip file with oversized link entry");
1239                         return ARCHIVE_FATAL;
1240                 }
1241
1242                 linkname_length = (size_t)zip_entry->compressed_size;
1243
1244                 archive_entry_set_size(entry, 0);
1245                 p = __archive_read_ahead(a, linkname_length, NULL);
1246                 if (p == NULL) {
1247                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1248                             "Truncated Zip file");
1249                         return ARCHIVE_FATAL;
1250                 }
1251                 // take into account link compression if any
1252                 size_t linkname_full_length = linkname_length;
1253                 if (zip->entry->compression != 0)
1254                 {
1255                         // symlink target string appeared to be compressed
1256                         int status = ARCHIVE_FATAL;
1257                         char *uncompressed_buffer =
1258                                 (char*) malloc(zip_entry->uncompressed_size);
1259                         if (uncompressed_buffer == NULL)
1260                         {
1261                                 archive_set_error(&a->archive, ENOMEM,
1262                                         "No memory for lzma decompression");
1263                                 return status;
1264                         }
1265
1266                         switch (zip->entry->compression)
1267                         {
1268 #if HAVE_LZMA_H && HAVE_LIBLZMA
1269                                 case 14: /* ZIPx LZMA compression. */
1270                                         /*(see zip file format specification, section 4.4.5)*/
1271                                         status = zipx_lzma_uncompress_buffer(p,
1272                                                 linkname_length,
1273                                                 uncompressed_buffer,
1274                                                 (size_t)zip_entry->uncompressed_size);
1275                                         break;
1276 #endif
1277                                 default: /* Unsupported compression. */
1278                                         break;
1279                         }
1280                         if (status == ARCHIVE_OK)
1281                         {
1282                                 p = uncompressed_buffer;
1283                                 linkname_full_length =
1284                                         (size_t)zip_entry->uncompressed_size;
1285                         }
1286                         else
1287                         {
1288                                 archive_set_error(&a->archive,
1289                                         ARCHIVE_ERRNO_FILE_FORMAT,
1290                                         "Unsupported ZIP compression method "
1291                                         "during decompression of link entry (%d: %s)",
1292                                         zip->entry->compression,
1293                                         compression_name(zip->entry->compression));
1294                                 return ARCHIVE_FAILED;
1295                         }
1296                 }
1297
1298                 sconv = zip->sconv;
1299                 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME))
1300                         sconv = zip->sconv_utf8;
1301                 if (sconv == NULL)
1302                         sconv = zip->sconv_default;
1303                 if (archive_entry_copy_symlink_l(entry, p, linkname_full_length,
1304                     sconv) != 0) {
1305                         if (errno != ENOMEM && sconv == zip->sconv_utf8 &&
1306                             (zip->entry->zip_flags & ZIP_UTF8_NAME))
1307                             archive_entry_copy_symlink_l(entry, p,
1308                                 linkname_full_length, NULL);
1309                         if (errno == ENOMEM) {
1310                                 archive_set_error(&a->archive, ENOMEM,
1311                                     "Can't allocate memory for Symlink");
1312                                 return (ARCHIVE_FATAL);
1313                         }
1314                         /*
1315                          * Since there is no character-set regulation for
1316                          * symlink name, do not report the conversion error
1317                          * in an automatic conversion.
1318                          */
1319                         if (sconv != zip->sconv_utf8 ||
1320                             (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) {
1321                                 archive_set_error(&a->archive,
1322                                     ARCHIVE_ERRNO_FILE_FORMAT,
1323                                     "Symlink cannot be converted "
1324                                     "from %s to current locale.",
1325                                     archive_string_conversion_charset_name(
1326                                         sconv));
1327                                 ret = ARCHIVE_WARN;
1328                         }
1329                 }
1330                 zip_entry->uncompressed_size = zip_entry->compressed_size = 0;
1331
1332                 if (__archive_read_consume(a, linkname_length) < 0) {
1333                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1334                             "Read error skipping symlink target name");
1335                         return ARCHIVE_FATAL;
1336                 }
1337         } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1338             || zip_entry->uncompressed_size > 0) {
1339                 /* Set the size only if it's meaningful. */
1340                 archive_entry_set_size(entry, zip_entry->uncompressed_size);
1341         }
1342         zip->entry_bytes_remaining = zip_entry->compressed_size;
1343
1344         /* If there's no body, force read_data() to return EOF immediately. */
1345         if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1346             && zip->entry_bytes_remaining < 1)
1347                 zip->end_of_entry = 1;
1348
1349         /* Set up a more descriptive format name. */
1350         archive_string_empty(&zip->format_name);
1351         archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)",
1352             version / 10, version % 10,
1353             compression_name(zip->entry->compression));
1354         a->archive.archive_format_name = zip->format_name.s;
1355
1356         return (ret);
1357 }
1358
1359 static int
1360 check_authentication_code(struct archive_read *a, const void *_p)
1361 {
1362         struct zip *zip = (struct zip *)(a->format->data);
1363
1364         /* Check authentication code. */
1365         if (zip->hctx_valid) {
1366                 const void *p;
1367                 uint8_t hmac[20];
1368                 size_t hmac_len = 20;
1369                 int cmp;
1370
1371                 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len);
1372                 if (_p == NULL) {
1373                         /* Read authentication code. */
1374                         p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL);
1375                         if (p == NULL) {
1376                                 archive_set_error(&a->archive,
1377                                     ARCHIVE_ERRNO_FILE_FORMAT,
1378                                     "Truncated ZIP file data");
1379                                 return (ARCHIVE_FATAL);
1380                         }
1381                 } else {
1382                         p = _p;
1383                 }
1384                 cmp = memcmp(hmac, p, AUTH_CODE_SIZE);
1385                 __archive_read_consume(a, AUTH_CODE_SIZE);
1386                 if (cmp != 0) {
1387                         archive_set_error(&a->archive,
1388                             ARCHIVE_ERRNO_MISC,
1389                             "ZIP bad Authentication code");
1390                         return (ARCHIVE_WARN);
1391                 }
1392         }
1393         return (ARCHIVE_OK);
1394 }
1395
1396 /*
1397  * Read "uncompressed" data.  There are three cases:
1398  *  1) We know the size of the data.  This is always true for the
1399  * seeking reader (we've examined the Central Directory already).
1400  *  2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred.
1401  * Info-ZIP seems to do this; we know the size but have to grab
1402  * the CRC from the data descriptor afterwards.
1403  *  3) We're streaming and ZIP_LENGTH_AT_END was specified and
1404  * we have no size information.  In this case, we can do pretty
1405  * well by watching for the data descriptor record.  The data
1406  * descriptor is 16 bytes and includes a computed CRC that should
1407  * provide a strong check.
1408  *
1409  * TODO: Technically, the PK\007\010 signature is optional.
1410  * In the original spec, the data descriptor contained CRC
1411  * and size fields but had no leading signature.  In practice,
1412  * newer writers seem to provide the signature pretty consistently.
1413  *
1414  * For uncompressed data, the PK\007\010 marker seems essential
1415  * to be sure we've actually seen the end of the entry.
1416  *
1417  * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets
1418  * zip->end_of_entry if it consumes all of the data.
1419  */
1420 static int
1421 zip_read_data_none(struct archive_read *a, const void **_buff,
1422     size_t *size, int64_t *offset)
1423 {
1424         struct zip *zip;
1425         const char *buff;
1426         ssize_t bytes_avail;
1427         int r;
1428
1429         (void)offset; /* UNUSED */
1430
1431         zip = (struct zip *)(a->format->data);
1432
1433         if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) {
1434                 const char *p;
1435                 ssize_t grabbing_bytes = 24;
1436
1437                 if (zip->hctx_valid)
1438                         grabbing_bytes += AUTH_CODE_SIZE;
1439                 /* Grab at least 24 bytes. */
1440                 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail);
1441                 if (bytes_avail < grabbing_bytes) {
1442                         /* Zip archives have end-of-archive markers
1443                            that are longer than this, so a failure to get at
1444                            least 24 bytes really does indicate a truncated
1445                            file. */
1446                         archive_set_error(&a->archive,
1447                             ARCHIVE_ERRNO_FILE_FORMAT,
1448                             "Truncated ZIP file data");
1449                         return (ARCHIVE_FATAL);
1450                 }
1451                 /* Check for a complete PK\007\010 signature, followed
1452                  * by the correct 4-byte CRC. */
1453                 p = buff;
1454                 if (zip->hctx_valid)
1455                         p += AUTH_CODE_SIZE;
1456                 if (p[0] == 'P' && p[1] == 'K'
1457                     && p[2] == '\007' && p[3] == '\010'
1458                     && (archive_le32dec(p + 4) == zip->entry_crc32
1459                         || zip->ignore_crc32
1460                         || (zip->hctx_valid
1461                          && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) {
1462                         if (zip->entry->flags & LA_USED_ZIP64) {
1463                                 uint64_t compressed, uncompressed;
1464                                 zip->entry->crc32 = archive_le32dec(p + 4);
1465                                 compressed = archive_le64dec(p + 8);
1466                                 uncompressed = archive_le64dec(p + 16);
1467                                 if (compressed > INT64_MAX || uncompressed >
1468                                     INT64_MAX) {
1469                                         archive_set_error(&a->archive,
1470                                             ARCHIVE_ERRNO_FILE_FORMAT,
1471                                             "Overflow of 64-bit file sizes");
1472                                         return ARCHIVE_FAILED;
1473                                 }
1474                                 zip->entry->compressed_size = compressed;
1475                                 zip->entry->uncompressed_size = uncompressed;
1476                                 zip->unconsumed = 24;
1477                         } else {
1478                                 zip->entry->crc32 = archive_le32dec(p + 4);
1479                                 zip->entry->compressed_size =
1480                                         archive_le32dec(p + 8);
1481                                 zip->entry->uncompressed_size =
1482                                         archive_le32dec(p + 12);
1483                                 zip->unconsumed = 16;
1484                         }
1485                         if (zip->hctx_valid) {
1486                                 r = check_authentication_code(a, buff);
1487                                 if (r != ARCHIVE_OK)
1488                                         return (r);
1489                         }
1490                         zip->end_of_entry = 1;
1491                         return (ARCHIVE_OK);
1492                 }
1493                 /* If not at EOF, ensure we consume at least one byte. */
1494                 ++p;
1495
1496                 /* Scan forward until we see where a PK\007\010 signature
1497                  * might be. */
1498                 /* Return bytes up until that point.  On the next call,
1499                  * the code above will verify the data descriptor. */
1500                 while (p < buff + bytes_avail - 4) {
1501                         if (p[3] == 'P') { p += 3; }
1502                         else if (p[3] == 'K') { p += 2; }
1503                         else if (p[3] == '\007') { p += 1; }
1504                         else if (p[3] == '\010' && p[2] == '\007'
1505                             && p[1] == 'K' && p[0] == 'P') {
1506                                 if (zip->hctx_valid)
1507                                         p -= AUTH_CODE_SIZE;
1508                                 break;
1509                         } else { p += 4; }
1510                 }
1511                 bytes_avail = p - buff;
1512         } else {
1513                 if (zip->entry_bytes_remaining == 0) {
1514                         zip->end_of_entry = 1;
1515                         if (zip->hctx_valid) {
1516                                 r = check_authentication_code(a, NULL);
1517                                 if (r != ARCHIVE_OK)
1518                                         return (r);
1519                         }
1520                         return (ARCHIVE_OK);
1521                 }
1522                 /* Grab a bunch of bytes. */
1523                 buff = __archive_read_ahead(a, 1, &bytes_avail);
1524                 if (bytes_avail <= 0) {
1525                         archive_set_error(&a->archive,
1526                             ARCHIVE_ERRNO_FILE_FORMAT,
1527                             "Truncated ZIP file data");
1528                         return (ARCHIVE_FATAL);
1529                 }
1530                 if (bytes_avail > zip->entry_bytes_remaining)
1531                         bytes_avail = (ssize_t)zip->entry_bytes_remaining;
1532         }
1533         if (zip->tctx_valid || zip->cctx_valid) {
1534                 size_t dec_size = bytes_avail;
1535
1536                 if (dec_size > zip->decrypted_buffer_size)
1537                         dec_size = zip->decrypted_buffer_size;
1538                 if (zip->tctx_valid) {
1539                         trad_enc_decrypt_update(&zip->tctx,
1540                             (const uint8_t *)buff, dec_size,
1541                             zip->decrypted_buffer, dec_size);
1542                 } else {
1543                         size_t dsize = dec_size;
1544                         archive_hmac_sha1_update(&zip->hctx,
1545                             (const uint8_t *)buff, dec_size);
1546                         archive_decrypto_aes_ctr_update(&zip->cctx,
1547                             (const uint8_t *)buff, dec_size,
1548                             zip->decrypted_buffer, &dsize);
1549                 }
1550                 bytes_avail = dec_size;
1551                 buff = (const char *)zip->decrypted_buffer;
1552         }
1553         *size = bytes_avail;
1554         zip->entry_bytes_remaining -= bytes_avail;
1555         zip->entry_uncompressed_bytes_read += bytes_avail;
1556         zip->entry_compressed_bytes_read += bytes_avail;
1557         zip->unconsumed += bytes_avail;
1558         *_buff = buff;
1559         return (ARCHIVE_OK);
1560 }
1561
1562 static int
1563 consume_optional_marker(struct archive_read *a, struct zip *zip)
1564 {
1565         if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
1566                 const char *p;
1567
1568                 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) {
1569                         archive_set_error(&a->archive,
1570                             ARCHIVE_ERRNO_FILE_FORMAT,
1571                             "Truncated ZIP end-of-file record");
1572                         return (ARCHIVE_FATAL);
1573                 }
1574                 /* Consume the optional PK\007\010 marker. */
1575                 if (p[0] == 'P' && p[1] == 'K' &&
1576                     p[2] == '\007' && p[3] == '\010') {
1577                         p += 4;
1578                         zip->unconsumed = 4;
1579                 }
1580                 if (zip->entry->flags & LA_USED_ZIP64) {
1581                         uint64_t compressed, uncompressed;
1582                         zip->entry->crc32 = archive_le32dec(p);
1583                         compressed = archive_le64dec(p + 4);
1584                         uncompressed = archive_le64dec(p + 12);
1585                         if (compressed > INT64_MAX ||
1586                             uncompressed > INT64_MAX) {
1587                                 archive_set_error(&a->archive,
1588                                     ARCHIVE_ERRNO_FILE_FORMAT,
1589                                     "Overflow of 64-bit file sizes");
1590                                 return ARCHIVE_FAILED;
1591                         }
1592                         zip->entry->compressed_size = compressed;
1593                         zip->entry->uncompressed_size = uncompressed;
1594                         zip->unconsumed += 20;
1595                 } else {
1596                         zip->entry->crc32 = archive_le32dec(p);
1597                         zip->entry->compressed_size = archive_le32dec(p + 4);
1598                         zip->entry->uncompressed_size = archive_le32dec(p + 8);
1599                         zip->unconsumed += 12;
1600                 }
1601         }
1602
1603     return (ARCHIVE_OK);
1604 }
1605
1606 #if HAVE_LZMA_H && HAVE_LIBLZMA
1607 static int
1608 zipx_xz_init(struct archive_read *a, struct zip *zip)
1609 {
1610         lzma_ret r;
1611
1612         if(zip->zipx_lzma_valid) {
1613                 lzma_end(&zip->zipx_lzma_stream);
1614                 zip->zipx_lzma_valid = 0;
1615         }
1616
1617         memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream));
1618         r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0);
1619         if (r != LZMA_OK) {
1620                 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
1621                     "xz initialization failed(%d)",
1622                     r);
1623
1624                 return (ARCHIVE_FAILED);
1625         }
1626
1627         zip->zipx_lzma_valid = 1;
1628
1629         free(zip->uncompressed_buffer);
1630
1631         zip->uncompressed_buffer_size = 256 * 1024;
1632         zip->uncompressed_buffer =
1633             (uint8_t*) malloc(zip->uncompressed_buffer_size);
1634         if (zip->uncompressed_buffer == NULL) {
1635                 archive_set_error(&a->archive, ENOMEM,
1636                     "No memory for xz decompression");
1637                     return (ARCHIVE_FATAL);
1638         }
1639
1640         zip->decompress_init = 1;
1641         return (ARCHIVE_OK);
1642 }
1643
1644 static int
1645 zipx_lzma_alone_init(struct archive_read *a, struct zip *zip)
1646 {
1647         lzma_ret r;
1648         const uint8_t* p;
1649
1650 #pragma pack(push)
1651 #pragma pack(1)
1652         struct _alone_header {
1653             uint8_t bytes[5];
1654             uint64_t uncompressed_size;
1655         } alone_header;
1656 #pragma pack(pop)
1657
1658         if(zip->zipx_lzma_valid) {
1659                 lzma_end(&zip->zipx_lzma_stream);
1660                 zip->zipx_lzma_valid = 0;
1661         }
1662
1663         /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma
1664          * that is a part of XZ Utils. The stream format stored inside ZIPX
1665          * file is a modified "lzma alone" file format, that was used by the
1666          * `lzma` utility which was later deprecated in favour of `xz` utility.          * Since those formats are nearly the same, we can use a standard
1667          * "lzma alone" decoder from XZ Utils. */
1668
1669         memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream));
1670         r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX);
1671         if (r != LZMA_OK) {
1672                 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
1673                     "lzma initialization failed(%d)", r);
1674
1675                 return (ARCHIVE_FAILED);
1676         }
1677
1678         /* Flag the cleanup function that we want our lzma-related structures
1679          * to be freed later. */
1680         zip->zipx_lzma_valid = 1;
1681
1682         /* The "lzma alone" file format and the stream format inside ZIPx are
1683          * almost the same. Here's an example of a structure of "lzma alone"
1684          * format:
1685          *
1686          * $ cat /bin/ls | lzma | xxd | head -n 1
1687          * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814
1688          *
1689          *    5 bytes        8 bytes        n bytes
1690          * <lzma_params><uncompressed_size><data...>
1691          *
1692          * lzma_params is a 5-byte blob that has to be decoded to extract
1693          * parameters of this LZMA stream. The uncompressed_size field is an
1694          * uint64_t value that contains information about the size of the
1695          * uncompressed file, or UINT64_MAX if this value is unknown.
1696          * The <data...> part is the actual lzma-compressed data stream.
1697          *
1698          * Now here's the structure of the stream inside the ZIPX file:
1699          *
1700          * $ cat stream_inside_zipx | xxd | head -n 1
1701          * 00000000: 0914 0500 5d00 8000 0000 2814 .... ....
1702          *
1703          *  2byte   2byte    5 bytes     n bytes
1704          * <magic1><magic2><lzma_params><data...>
1705          *
1706          * This means that the ZIPX file contains an additional magic1 and
1707          * magic2 headers, the lzma_params field contains the same parameter
1708          * set as in the "lzma alone" format, and the <data...> field is the
1709          * same as in the "lzma alone" format as well. Note that also the zipx
1710          * format is missing the uncompressed_size field.
1711          *
1712          * So, in order to use the "lzma alone" decoder for the zipx lzma
1713          * stream, we simply need to shuffle around some fields, prepare a new
1714          * lzma alone header, feed it into lzma alone decoder so it will
1715          * initialize itself properly, and then we can start feeding normal
1716          * zipx lzma stream into the decoder.
1717          */
1718
1719         /* Read magic1,magic2,lzma_params from the ZIPX stream. */
1720         if((p = __archive_read_ahead(a, 9, NULL)) == NULL) {
1721                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1722                     "Truncated lzma data");
1723                 return (ARCHIVE_FATAL);
1724         }
1725
1726         if(p[2] != 0x05 || p[3] != 0x00) {
1727                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1728                     "Invalid lzma data");
1729                 return (ARCHIVE_FATAL);
1730         }
1731
1732         /* Prepare an lzma alone header: copy the lzma_params blob into
1733          * a proper place into the lzma alone header. */
1734         memcpy(&alone_header.bytes[0], p + 4, 5);
1735
1736         /* Initialize the 'uncompressed size' field to unknown; we'll manually
1737          * monitor how many bytes there are still to be uncompressed. */
1738         alone_header.uncompressed_size = UINT64_MAX;
1739
1740         if(!zip->uncompressed_buffer) {
1741                 zip->uncompressed_buffer_size = 256 * 1024;
1742                 zip->uncompressed_buffer =
1743                         (uint8_t*) malloc(zip->uncompressed_buffer_size);
1744
1745                 if (zip->uncompressed_buffer == NULL) {
1746                         archive_set_error(&a->archive, ENOMEM,
1747                             "No memory for lzma decompression");
1748                         return (ARCHIVE_FATAL);
1749                 }
1750         }
1751
1752         zip->zipx_lzma_stream.next_in = (void*) &alone_header;
1753         zip->zipx_lzma_stream.avail_in = sizeof(alone_header);
1754         zip->zipx_lzma_stream.total_in = 0;
1755         zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1756         zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size;
1757         zip->zipx_lzma_stream.total_out = 0;
1758
1759         /* Feed only the header into the lzma alone decoder. This will
1760          * effectively initialize the decoder, and will not produce any
1761          * output bytes yet. */
1762         r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1763         if (r != LZMA_OK) {
1764                 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1765                     "lzma stream initialization error");
1766                 return ARCHIVE_FATAL;
1767         }
1768
1769         /* We've already consumed some bytes, so take this into account. */
1770         __archive_read_consume(a, 9);
1771         zip->entry_bytes_remaining -= 9;
1772         zip->entry_compressed_bytes_read += 9;
1773
1774         zip->decompress_init = 1;
1775         return (ARCHIVE_OK);
1776 }
1777
1778 static int
1779 zip_read_data_zipx_xz(struct archive_read *a, const void **buff,
1780         size_t *size, int64_t *offset)
1781 {
1782         struct zip* zip = (struct zip *)(a->format->data);
1783         int ret;
1784         lzma_ret lz_ret;
1785         const void* compressed_buf;
1786         ssize_t bytes_avail, in_bytes, to_consume = 0;
1787
1788         (void) offset; /* UNUSED */
1789
1790         /* Initialize decompressor if not yet initialized. */
1791         if (!zip->decompress_init) {
1792                 ret = zipx_xz_init(a, zip);
1793                 if (ret != ARCHIVE_OK)
1794                         return (ret);
1795         }
1796
1797         compressed_buf = __archive_read_ahead(a, 1, &bytes_avail);
1798         if (bytes_avail < 0) {
1799                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1800                     "Truncated xz file body");
1801                 return (ARCHIVE_FATAL);
1802         }
1803
1804         in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail);
1805         zip->zipx_lzma_stream.next_in = compressed_buf;
1806         zip->zipx_lzma_stream.avail_in = in_bytes;
1807         zip->zipx_lzma_stream.total_in = 0;
1808         zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1809         zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size;
1810         zip->zipx_lzma_stream.total_out = 0;
1811
1812         /* Perform the decompression. */
1813         lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1814         switch(lz_ret) {
1815                 case LZMA_DATA_ERROR:
1816                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1817                             "xz data error (error %d)", (int) lz_ret);
1818                         return (ARCHIVE_FATAL);
1819
1820                 case LZMA_NO_CHECK:
1821                 case LZMA_OK:
1822                         break;
1823
1824                 default:
1825                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1826                             "xz unknown error %d", (int) lz_ret);
1827                         return (ARCHIVE_FATAL);
1828
1829                 case LZMA_STREAM_END:
1830                         lzma_end(&zip->zipx_lzma_stream);
1831                         zip->zipx_lzma_valid = 0;
1832
1833                         if((int64_t) zip->zipx_lzma_stream.total_in !=
1834                             zip->entry_bytes_remaining)
1835                         {
1836                                 archive_set_error(&a->archive,
1837                                     ARCHIVE_ERRNO_MISC,
1838                                     "xz premature end of stream");
1839                                 return (ARCHIVE_FATAL);
1840                         }
1841
1842                         zip->end_of_entry = 1;
1843                         break;
1844         }
1845
1846         to_consume = zip->zipx_lzma_stream.total_in;
1847
1848         __archive_read_consume(a, to_consume);
1849         zip->entry_bytes_remaining -= to_consume;
1850         zip->entry_compressed_bytes_read += to_consume;
1851         zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out;
1852
1853         *size = zip->zipx_lzma_stream.total_out;
1854         *buff = zip->uncompressed_buffer;
1855
1856         ret = consume_optional_marker(a, zip);
1857         if (ret != ARCHIVE_OK)
1858                 return (ret);
1859
1860         return (ARCHIVE_OK);
1861 }
1862
1863 static int
1864 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff,
1865     size_t *size, int64_t *offset)
1866 {
1867         struct zip* zip = (struct zip *)(a->format->data);
1868         int ret;
1869         lzma_ret lz_ret;
1870         const void* compressed_buf;
1871         ssize_t bytes_avail, in_bytes, to_consume;
1872
1873         (void) offset; /* UNUSED */
1874
1875         /* Initialize decompressor if not yet initialized. */
1876         if (!zip->decompress_init) {
1877                 ret = zipx_lzma_alone_init(a, zip);
1878                 if (ret != ARCHIVE_OK)
1879                         return (ret);
1880         }
1881
1882         /* Fetch more compressed data. The same note as in deflate handler
1883          * applies here as well:
1884          *
1885          * Note: '1' here is a performance optimization. Recall that the
1886          * decompression layer returns a count of available bytes; asking for
1887          * more than that forces the decompressor to combine reads by copying
1888          * data.
1889          */
1890         compressed_buf = __archive_read_ahead(a, 1, &bytes_avail);
1891         if (bytes_avail < 0) {
1892                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1893                     "Truncated lzma file body");
1894                 return (ARCHIVE_FATAL);
1895         }
1896
1897         /* Set decompressor parameters. */
1898         in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail);
1899
1900         zip->zipx_lzma_stream.next_in = compressed_buf;
1901         zip->zipx_lzma_stream.avail_in = in_bytes;
1902         zip->zipx_lzma_stream.total_in = 0;
1903         zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1904         zip->zipx_lzma_stream.avail_out =
1905                 /* These lzma_alone streams lack end of stream marker, so let's
1906                  * make sure the unpacker won't try to unpack more than it's
1907                  * supposed to. */
1908                 zipmin((int64_t) zip->uncompressed_buffer_size,
1909                     zip->entry->uncompressed_size -
1910                     zip->entry_uncompressed_bytes_read);
1911         zip->zipx_lzma_stream.total_out = 0;
1912
1913         /* Perform the decompression. */
1914         lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1915         switch(lz_ret) {
1916                 case LZMA_DATA_ERROR:
1917                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1918                             "lzma data error (error %d)", (int) lz_ret);
1919                         return (ARCHIVE_FATAL);
1920
1921                 /* This case is optional in lzma alone format. It can happen,
1922                  * but most of the files don't have it. (GitHub #1257) */
1923                 case LZMA_STREAM_END:
1924                         lzma_end(&zip->zipx_lzma_stream);
1925                         zip->zipx_lzma_valid = 0;
1926                         if((int64_t) zip->zipx_lzma_stream.total_in !=
1927                             zip->entry_bytes_remaining)
1928                         {
1929                                 archive_set_error(&a->archive,
1930                                     ARCHIVE_ERRNO_MISC,
1931                                     "lzma alone premature end of stream");
1932                                 return (ARCHIVE_FATAL);
1933                         }
1934
1935                         zip->end_of_entry = 1;
1936                         break;
1937
1938                 case LZMA_OK:
1939                         break;
1940
1941                 default:
1942                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1943                             "lzma unknown error %d", (int) lz_ret);
1944                         return (ARCHIVE_FATAL);
1945         }
1946
1947         to_consume = zip->zipx_lzma_stream.total_in;
1948
1949         /* Update pointers. */
1950         __archive_read_consume(a, to_consume);
1951         zip->entry_bytes_remaining -= to_consume;
1952         zip->entry_compressed_bytes_read += to_consume;
1953         zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out;
1954
1955         if(zip->entry_bytes_remaining == 0) {
1956                 zip->end_of_entry = 1;
1957         }
1958
1959         /* Return values. */
1960         *size = zip->zipx_lzma_stream.total_out;
1961         *buff = zip->uncompressed_buffer;
1962
1963         /* Behave the same way as during deflate decompression. */
1964         ret = consume_optional_marker(a, zip);
1965         if (ret != ARCHIVE_OK)
1966                 return (ret);
1967
1968         /* Free lzma decoder handle because we'll no longer need it. */
1969         if(zip->end_of_entry) {
1970                 lzma_end(&zip->zipx_lzma_stream);
1971                 zip->zipx_lzma_valid = 0;
1972         }
1973
1974         /* If we're here, then we're good! */
1975         return (ARCHIVE_OK);
1976 }
1977 #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */
1978
1979 static int
1980 zipx_ppmd8_init(struct archive_read *a, struct zip *zip)
1981 {
1982         const void* p;
1983         uint32_t val;
1984         uint32_t order;
1985         uint32_t mem;
1986         uint32_t restore_method;
1987
1988         /* Remove previous decompression context if it exists. */
1989         if(zip->ppmd8_valid) {
1990                 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
1991                 zip->ppmd8_valid = 0;
1992         }
1993
1994         /* Create a new decompression context. */
1995         __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8);
1996         zip->ppmd8_stream_failed = 0;
1997
1998         /* Setup function pointers required by Ppmd8 decompressor. The
1999          * 'ppmd_read' function will feed new bytes to the decompressor,
2000          * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */
2001         zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream;
2002         zip->zipx_ppmd_stream.a = a;
2003         zip->zipx_ppmd_stream.Read = &ppmd_read;
2004
2005         /* Reset number of read bytes to 0. */
2006         zip->zipx_ppmd_read_compressed = 0;
2007
2008         /* Read Ppmd8 header (2 bytes). */
2009         p = __archive_read_ahead(a, 2, NULL);
2010         if(!p) {
2011                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2012                     "Truncated file data in PPMd8 stream");
2013                 return (ARCHIVE_FATAL);
2014         }
2015         __archive_read_consume(a, 2);
2016
2017         /* Decode the stream's compression parameters. */
2018         val = archive_le16dec(p);
2019         order = (val & 15) + 1;
2020         mem = ((val >> 4) & 0xff) + 1;
2021         restore_method = (val >> 12);
2022
2023         if(order < 2 || restore_method > 2) {
2024                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2025                     "Invalid parameter set in PPMd8 stream (order=%" PRId32 ", "
2026                     "restore=%" PRId32 ")", order, restore_method);
2027                 return (ARCHIVE_FAILED);
2028         }
2029
2030         /* Allocate the memory needed to properly decompress the file. */
2031         if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) {
2032                 archive_set_error(&a->archive, ENOMEM,
2033                     "Unable to allocate memory for PPMd8 stream: %" PRId32 " bytes",
2034                     mem << 20);
2035                 return (ARCHIVE_FATAL);
2036         }
2037
2038         /* Signal the cleanup function to release Ppmd8 context in the
2039          * cleanup phase. */
2040         zip->ppmd8_valid = 1;
2041
2042         /* Perform further Ppmd8 initialization. */
2043         if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) {
2044                 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
2045                     "PPMd8 stream range decoder initialization error");
2046                 return (ARCHIVE_FATAL);
2047         }
2048
2049         __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order,
2050             restore_method);
2051
2052         /* Allocate the buffer that will hold uncompressed data. */
2053         free(zip->uncompressed_buffer);
2054
2055         zip->uncompressed_buffer_size = 256 * 1024;
2056         zip->uncompressed_buffer =
2057             (uint8_t*) malloc(zip->uncompressed_buffer_size);
2058
2059         if(zip->uncompressed_buffer == NULL) {
2060                 archive_set_error(&a->archive, ENOMEM,
2061                     "No memory for PPMd8 decompression");
2062                 return ARCHIVE_FATAL;
2063         }
2064
2065         /* Ppmd8 initialization is done. */
2066         zip->decompress_init = 1;
2067
2068         /* We've already read 2 bytes in the output stream. Additionally,
2069          * Ppmd8 initialization code could read some data as well. So we
2070          * are advancing the stream by 2 bytes plus whatever number of
2071          * bytes Ppmd8 init function used. */
2072         zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed;
2073
2074         return ARCHIVE_OK;
2075 }
2076
2077 static int
2078 zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff,
2079     size_t *size, int64_t *offset)
2080 {
2081         struct zip* zip = (struct zip *)(a->format->data);
2082         int ret;
2083         size_t consumed_bytes = 0;
2084         ssize_t bytes_avail = 0;
2085
2086         (void) offset; /* UNUSED */
2087
2088         /* If we're here for the first time, initialize Ppmd8 decompression
2089          * context first. */
2090         if(!zip->decompress_init) {
2091                 ret = zipx_ppmd8_init(a, zip);
2092                 if(ret != ARCHIVE_OK)
2093                         return ret;
2094         }
2095
2096         /* Fetch for more data. We're reading 1 byte here, but libarchive
2097          * should prefetch more bytes. */
2098         (void) __archive_read_ahead(a, 1, &bytes_avail);
2099         if(bytes_avail < 0) {
2100                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2101                     "Truncated PPMd8 file body");
2102                 return (ARCHIVE_FATAL);
2103         }
2104
2105         /* This counter will be updated inside ppmd_read(), which at one
2106          * point will be called by Ppmd8_DecodeSymbol. */
2107         zip->zipx_ppmd_read_compressed = 0;
2108
2109         /* Decompression loop. */
2110         do {
2111                 int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol(
2112                     &zip->ppmd8);
2113                 if(sym < 0) {
2114                         zip->end_of_entry = 1;
2115                         break;
2116                 }
2117
2118                 /* This field is set by ppmd_read() when there was no more data
2119                  * to be read. */
2120                 if(zip->ppmd8_stream_failed) {
2121                         archive_set_error(&a->archive,
2122                             ARCHIVE_ERRNO_FILE_FORMAT,
2123                             "Truncated PPMd8 file body");
2124                         return (ARCHIVE_FATAL);
2125                 }
2126
2127                 zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym;
2128                 ++consumed_bytes;
2129         } while(consumed_bytes < zip->uncompressed_buffer_size);
2130
2131         /* Update pointers for libarchive. */
2132         *buff = zip->uncompressed_buffer;
2133         *size = consumed_bytes;
2134
2135         /* Update pointers so we can continue decompression in another call. */
2136         zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed;
2137         zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed;
2138         zip->entry_uncompressed_bytes_read += consumed_bytes;
2139
2140         /* If we're at the end of stream, deinitialize Ppmd8 context. */
2141         if(zip->end_of_entry) {
2142                 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
2143                 zip->ppmd8_valid = 0;
2144         }
2145
2146         /* Seek for optional marker, same way as in each zip entry. */
2147         ret = consume_optional_marker(a, zip);
2148         if (ret != ARCHIVE_OK)
2149                 return ret;
2150
2151         return ARCHIVE_OK;
2152 }
2153
2154 #ifdef HAVE_BZLIB_H
2155 static int
2156 zipx_bzip2_init(struct archive_read *a, struct zip *zip)
2157 {
2158         int r;
2159
2160         /* Deallocate already existing BZ2 decompression context if it
2161          * exists. */
2162         if(zip->bzstream_valid) {
2163                 BZ2_bzDecompressEnd(&zip->bzstream);
2164                 zip->bzstream_valid = 0;
2165         }
2166
2167         /* Allocate a new BZ2 decompression context. */
2168         memset(&zip->bzstream, 0, sizeof(bz_stream));
2169         r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1);
2170         if(r != BZ_OK) {
2171                 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
2172                     "bzip2 initialization failed(%d)",
2173                     r);
2174
2175                 return ARCHIVE_FAILED;
2176         }
2177
2178         /* Mark the bzstream field to be released in cleanup phase. */
2179         zip->bzstream_valid = 1;
2180
2181         /* (Re)allocate the buffer that will contain decompressed bytes. */
2182         free(zip->uncompressed_buffer);
2183
2184         zip->uncompressed_buffer_size = 256 * 1024;
2185         zip->uncompressed_buffer =
2186             (uint8_t*) malloc(zip->uncompressed_buffer_size);
2187         if (zip->uncompressed_buffer == NULL) {
2188                 archive_set_error(&a->archive, ENOMEM,
2189                     "No memory for bzip2 decompression");
2190                     return ARCHIVE_FATAL;
2191         }
2192
2193         /* Initialization done. */
2194         zip->decompress_init = 1;
2195         return ARCHIVE_OK;
2196 }
2197
2198 static int
2199 zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff,
2200     size_t *size, int64_t *offset)
2201 {
2202         struct zip *zip = (struct zip *)(a->format->data);
2203         ssize_t bytes_avail = 0, in_bytes, to_consume;
2204         const void *compressed_buff;
2205         int r;
2206         uint64_t total_out;
2207
2208         (void) offset; /* UNUSED */
2209
2210         /* Initialize decompression context if we're here for the first time. */
2211         if(!zip->decompress_init) {
2212                 r = zipx_bzip2_init(a, zip);
2213                 if(r != ARCHIVE_OK)
2214                         return r;
2215         }
2216
2217         /* Fetch more compressed bytes. */
2218         compressed_buff = __archive_read_ahead(a, 1, &bytes_avail);
2219         if(bytes_avail < 0) {
2220                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2221                     "Truncated bzip2 file body");
2222                 return (ARCHIVE_FATAL);
2223         }
2224
2225         in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail);
2226         if(in_bytes < 1) {
2227                 /* libbz2 doesn't complain when caller feeds avail_in == 0.
2228                  * It will actually return success in this case, which is
2229                  * undesirable. This is why we need to make this check
2230                  * manually. */
2231
2232                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2233                     "Truncated bzip2 file body");
2234                 return (ARCHIVE_FATAL);
2235         }
2236
2237         /* Setup buffer boundaries. */
2238         zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff;
2239         zip->bzstream.avail_in = in_bytes;
2240         zip->bzstream.total_in_hi32 = 0;
2241         zip->bzstream.total_in_lo32 = 0;
2242         zip->bzstream.next_out = (char*) zip->uncompressed_buffer;
2243         zip->bzstream.avail_out = zip->uncompressed_buffer_size;
2244         zip->bzstream.total_out_hi32 = 0;
2245         zip->bzstream.total_out_lo32 = 0;
2246
2247         /* Perform the decompression. */
2248         r = BZ2_bzDecompress(&zip->bzstream);
2249         switch(r) {
2250                 case BZ_STREAM_END:
2251                         /* If we're at the end of the stream, deinitialize the
2252                          * decompression context now. */
2253                         switch(BZ2_bzDecompressEnd(&zip->bzstream)) {
2254                                 case BZ_OK:
2255                                         break;
2256                                 default:
2257                                         archive_set_error(&a->archive,
2258                                             ARCHIVE_ERRNO_MISC,
2259                                             "Failed to clean up bzip2 "
2260                                             "decompressor");
2261                                         return ARCHIVE_FATAL;
2262                         }
2263
2264                         zip->end_of_entry = 1;
2265                         break;
2266                 case BZ_OK:
2267                         /* The decompressor has successfully decoded this
2268                          * chunk of data, but more data is still in queue. */
2269                         break;
2270                 default:
2271                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2272                             "bzip2 decompression failed");
2273                         return ARCHIVE_FATAL;
2274         }
2275
2276         /* Update the pointers so decompressor can continue decoding. */
2277         to_consume = zip->bzstream.total_in_lo32;
2278         __archive_read_consume(a, to_consume);
2279
2280         total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) +
2281             zip->bzstream.total_out_lo32;
2282
2283         zip->entry_bytes_remaining -= to_consume;
2284         zip->entry_compressed_bytes_read += to_consume;
2285         zip->entry_uncompressed_bytes_read += total_out;
2286
2287         /* Give libarchive its due. */
2288         *size = total_out;
2289         *buff = zip->uncompressed_buffer;
2290
2291         /* Seek for optional marker, like in other entries. */
2292         r = consume_optional_marker(a, zip);
2293         if(r != ARCHIVE_OK)
2294                 return r;
2295
2296         return ARCHIVE_OK;
2297 }
2298
2299 #endif
2300
2301 #ifdef HAVE_ZLIB_H
2302 static int
2303 zip_deflate_init(struct archive_read *a, struct zip *zip)
2304 {
2305         int r;
2306
2307         /* If we haven't yet read any data, initialize the decompressor. */
2308         if (!zip->decompress_init) {
2309                 if (zip->stream_valid)
2310                         r = inflateReset(&zip->stream);
2311                 else
2312                         r = inflateInit2(&zip->stream,
2313                             -15 /* Don't check for zlib header */);
2314                 if (r != Z_OK) {
2315                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2316                             "Can't initialize ZIP decompression.");
2317                         return (ARCHIVE_FATAL);
2318                 }
2319                 /* Stream structure has been set up. */
2320                 zip->stream_valid = 1;
2321                 /* We've initialized decompression for this stream. */
2322                 zip->decompress_init = 1;
2323         }
2324         return (ARCHIVE_OK);
2325 }
2326
2327 static int
2328 zip_read_data_deflate(struct archive_read *a, const void **buff,
2329     size_t *size, int64_t *offset)
2330 {
2331         struct zip *zip;
2332         ssize_t bytes_avail;
2333         const void *compressed_buff, *sp;
2334         int r;
2335
2336         (void)offset; /* UNUSED */
2337
2338         zip = (struct zip *)(a->format->data);
2339
2340         /* If the buffer hasn't been allocated, allocate it now. */
2341         if (zip->uncompressed_buffer == NULL) {
2342                 zip->uncompressed_buffer_size = 256 * 1024;
2343                 zip->uncompressed_buffer
2344                     = (unsigned char *)malloc(zip->uncompressed_buffer_size);
2345                 if (zip->uncompressed_buffer == NULL) {
2346                         archive_set_error(&a->archive, ENOMEM,
2347                             "No memory for ZIP decompression");
2348                         return (ARCHIVE_FATAL);
2349                 }
2350         }
2351
2352         r = zip_deflate_init(a, zip);
2353         if (r != ARCHIVE_OK)
2354                 return (r);
2355
2356         /*
2357          * Note: '1' here is a performance optimization.
2358          * Recall that the decompression layer returns a count of
2359          * available bytes; asking for more than that forces the
2360          * decompressor to combine reads by copying data.
2361          */
2362         compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail);
2363         if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2364             && bytes_avail > zip->entry_bytes_remaining) {
2365                 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
2366         }
2367         if (bytes_avail < 0) {
2368                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2369                     "Truncated ZIP file body");
2370                 return (ARCHIVE_FATAL);
2371         }
2372
2373         if (zip->tctx_valid || zip->cctx_valid) {
2374                 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) {
2375                         size_t buff_remaining =
2376                             (zip->decrypted_buffer +
2377                             zip->decrypted_buffer_size)
2378                             - (zip->decrypted_ptr +
2379                             zip->decrypted_bytes_remaining);
2380
2381                         if (buff_remaining > (size_t)bytes_avail)
2382                                 buff_remaining = (size_t)bytes_avail;
2383
2384                         if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) &&
2385                               zip->entry_bytes_remaining > 0) {
2386                                 if ((int64_t)(zip->decrypted_bytes_remaining
2387                                     + buff_remaining)
2388                                       > zip->entry_bytes_remaining) {
2389                                         if (zip->entry_bytes_remaining <
2390                                             (int64_t)zip->decrypted_bytes_remaining)
2391                                                 buff_remaining = 0;
2392                                         else
2393                                                 buff_remaining =
2394                                                     (size_t)zip->entry_bytes_remaining
2395                                                     - zip->decrypted_bytes_remaining;
2396                                 }
2397                         }
2398                         if (buff_remaining > 0) {
2399                                 if (zip->tctx_valid) {
2400                                         trad_enc_decrypt_update(&zip->tctx,
2401                                             compressed_buff, buff_remaining,
2402                                             zip->decrypted_ptr
2403                                               + zip->decrypted_bytes_remaining,
2404                                             buff_remaining);
2405                                 } else {
2406                                         size_t dsize = buff_remaining;
2407                                         archive_decrypto_aes_ctr_update(
2408                                             &zip->cctx,
2409                                             compressed_buff, buff_remaining,
2410                                             zip->decrypted_ptr
2411                                               + zip->decrypted_bytes_remaining,
2412                                             &dsize);
2413                                 }
2414                                 zip->decrypted_bytes_remaining +=
2415                                     buff_remaining;
2416                         }
2417                 }
2418                 bytes_avail = zip->decrypted_bytes_remaining;
2419                 compressed_buff = (const char *)zip->decrypted_ptr;
2420         }
2421
2422         /*
2423          * A bug in zlib.h: stream.next_in should be marked 'const'
2424          * but isn't (the library never alters data through the
2425          * next_in pointer, only reads it).  The result: this ugly
2426          * cast to remove 'const'.
2427          */
2428         zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff;
2429         zip->stream.avail_in = (uInt)bytes_avail;
2430         zip->stream.total_in = 0;
2431         zip->stream.next_out = zip->uncompressed_buffer;
2432         zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size;
2433         zip->stream.total_out = 0;
2434
2435         r = inflate(&zip->stream, 0);
2436         switch (r) {
2437         case Z_OK:
2438                 break;
2439         case Z_STREAM_END:
2440                 zip->end_of_entry = 1;
2441                 break;
2442         case Z_MEM_ERROR:
2443                 archive_set_error(&a->archive, ENOMEM,
2444                     "Out of memory for ZIP decompression");
2445                 return (ARCHIVE_FATAL);
2446         default:
2447                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2448                     "ZIP decompression failed (%d)", r);
2449                 return (ARCHIVE_FATAL);
2450         }
2451
2452         /* Consume as much as the compressor actually used. */
2453         bytes_avail = zip->stream.total_in;
2454         if (zip->tctx_valid || zip->cctx_valid) {
2455                 zip->decrypted_bytes_remaining -= bytes_avail;
2456                 if (zip->decrypted_bytes_remaining == 0)
2457                         zip->decrypted_ptr = zip->decrypted_buffer;
2458                 else
2459                         zip->decrypted_ptr += bytes_avail;
2460         }
2461         /* Calculate compressed data as much as we used.*/
2462         if (zip->hctx_valid)
2463                 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail);
2464         __archive_read_consume(a, bytes_avail);
2465         zip->entry_bytes_remaining -= bytes_avail;
2466         zip->entry_compressed_bytes_read += bytes_avail;
2467
2468         *size = zip->stream.total_out;
2469         zip->entry_uncompressed_bytes_read += zip->stream.total_out;
2470         *buff = zip->uncompressed_buffer;
2471
2472         if (zip->end_of_entry && zip->hctx_valid) {
2473                 r = check_authentication_code(a, NULL);
2474                 if (r != ARCHIVE_OK)
2475                         return (r);
2476         }
2477
2478         r = consume_optional_marker(a, zip);
2479         if (r != ARCHIVE_OK)
2480                 return (r);
2481
2482         return (ARCHIVE_OK);
2483 }
2484 #endif
2485
2486 static int
2487 read_decryption_header(struct archive_read *a)
2488 {
2489         struct zip *zip = (struct zip *)(a->format->data);
2490         const char *p;
2491         unsigned int remaining_size;
2492         unsigned int ts;
2493
2494         /*
2495          * Read an initialization vector data field.
2496          */
2497         p = __archive_read_ahead(a, 2, NULL);
2498         if (p == NULL)
2499                 goto truncated;
2500         ts = zip->iv_size;
2501         zip->iv_size = archive_le16dec(p);
2502         __archive_read_consume(a, 2);
2503         if (ts < zip->iv_size) {
2504                 free(zip->iv);
2505                 zip->iv = NULL;
2506         }
2507         p = __archive_read_ahead(a, zip->iv_size, NULL);
2508         if (p == NULL)
2509                 goto truncated;
2510         if (zip->iv == NULL) {
2511                 zip->iv = malloc(zip->iv_size);
2512                 if (zip->iv == NULL)
2513                         goto nomem;
2514         }
2515         memcpy(zip->iv, p, zip->iv_size);
2516         __archive_read_consume(a, zip->iv_size);
2517
2518         /*
2519          * Read a size of remaining decryption header field.
2520          */
2521         p = __archive_read_ahead(a, 14, NULL);
2522         if (p == NULL)
2523                 goto truncated;
2524         remaining_size = archive_le32dec(p);
2525         if (remaining_size < 16 || remaining_size > (1 << 18))
2526                 goto corrupted;
2527
2528         /* Check if format version is supported. */
2529         if (archive_le16dec(p+4) != 3) {
2530                 archive_set_error(&a->archive,
2531                     ARCHIVE_ERRNO_FILE_FORMAT,
2532                     "Unsupported encryption format version: %u",
2533                     archive_le16dec(p+4));
2534                 return (ARCHIVE_FAILED);
2535         }
2536
2537         /*
2538          * Read an encryption algorithm field.
2539          */
2540         zip->alg_id = archive_le16dec(p+6);
2541         switch (zip->alg_id) {
2542         case 0x6601:/* DES */
2543         case 0x6602:/* RC2 */
2544         case 0x6603:/* 3DES 168 */
2545         case 0x6609:/* 3DES 112 */
2546         case 0x660E:/* AES 128 */
2547         case 0x660F:/* AES 192 */
2548         case 0x6610:/* AES 256 */
2549         case 0x6702:/* RC2 (version >= 5.2) */
2550         case 0x6720:/* Blowfish */
2551         case 0x6721:/* Twofish */
2552         case 0x6801:/* RC4 */
2553                 /* Supported encryption algorithm. */
2554                 break;
2555         default:
2556                 archive_set_error(&a->archive,
2557                     ARCHIVE_ERRNO_FILE_FORMAT,
2558                     "Unknown encryption algorithm: %u", zip->alg_id);
2559                 return (ARCHIVE_FAILED);
2560         }
2561
2562         /*
2563          * Read a bit length field.
2564          */
2565         zip->bit_len = archive_le16dec(p+8);
2566
2567         /*
2568          * Read a flags field.
2569          */
2570         zip->flags = archive_le16dec(p+10);
2571         switch (zip->flags & 0xf000) {
2572         case 0x0001: /* Password is required to decrypt. */
2573         case 0x0002: /* Certificates only. */
2574         case 0x0003: /* Password or certificate required to decrypt. */
2575                 break;
2576         default:
2577                 archive_set_error(&a->archive,
2578                     ARCHIVE_ERRNO_FILE_FORMAT,
2579                     "Unknown encryption flag: %u", zip->flags);
2580                 return (ARCHIVE_FAILED);
2581         }
2582         if ((zip->flags & 0xf000) == 0 ||
2583             (zip->flags & 0xf000) == 0x4000) {
2584                 archive_set_error(&a->archive,
2585                     ARCHIVE_ERRNO_FILE_FORMAT,
2586                     "Unknown encryption flag: %u", zip->flags);
2587                 return (ARCHIVE_FAILED);
2588         }
2589
2590         /*
2591          * Read an encrypted random data field.
2592          */
2593         ts = zip->erd_size;
2594         zip->erd_size = archive_le16dec(p+12);
2595         __archive_read_consume(a, 14);
2596         if ((zip->erd_size & 0xf) != 0 ||
2597             (zip->erd_size + 16) > remaining_size ||
2598             (zip->erd_size + 16) < zip->erd_size)
2599                 goto corrupted;
2600
2601         if (ts < zip->erd_size) {
2602                 free(zip->erd);
2603                 zip->erd = NULL;
2604         }
2605         p = __archive_read_ahead(a, zip->erd_size, NULL);
2606         if (p == NULL)
2607                 goto truncated;
2608         if (zip->erd == NULL) {
2609                 zip->erd = malloc(zip->erd_size);
2610                 if (zip->erd == NULL)
2611                         goto nomem;
2612         }
2613         memcpy(zip->erd, p, zip->erd_size);
2614         __archive_read_consume(a, zip->erd_size);
2615
2616         /*
2617          * Read a reserved data field.
2618          */
2619         p = __archive_read_ahead(a, 4, NULL);
2620         if (p == NULL)
2621                 goto truncated;
2622         /* Reserved data size should be zero. */
2623         if (archive_le32dec(p) != 0)
2624                 goto corrupted;
2625         __archive_read_consume(a, 4);
2626
2627         /*
2628          * Read a password validation data field.
2629          */
2630         p = __archive_read_ahead(a, 2, NULL);
2631         if (p == NULL)
2632                 goto truncated;
2633         ts = zip->v_size;
2634         zip->v_size = archive_le16dec(p);
2635         __archive_read_consume(a, 2);
2636         if ((zip->v_size & 0x0f) != 0 ||
2637             (zip->erd_size + zip->v_size + 16) > remaining_size ||
2638             (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size))
2639                 goto corrupted;
2640         if (ts < zip->v_size) {
2641                 free(zip->v_data);
2642                 zip->v_data = NULL;
2643         }
2644         p = __archive_read_ahead(a, zip->v_size, NULL);
2645         if (p == NULL)
2646                 goto truncated;
2647         if (zip->v_data == NULL) {
2648                 zip->v_data = malloc(zip->v_size);
2649                 if (zip->v_data == NULL)
2650                         goto nomem;
2651         }
2652         memcpy(zip->v_data, p, zip->v_size);
2653         __archive_read_consume(a, zip->v_size);
2654
2655         p = __archive_read_ahead(a, 4, NULL);
2656         if (p == NULL)
2657                 goto truncated;
2658         zip->v_crc32 = archive_le32dec(p);
2659         __archive_read_consume(a, 4);
2660
2661         /*return (ARCHIVE_OK);
2662          * This is not fully implemented yet.*/
2663         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2664             "Encrypted file is unsupported");
2665         return (ARCHIVE_FAILED);
2666 truncated:
2667         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2668             "Truncated ZIP file data");
2669         return (ARCHIVE_FATAL);
2670 corrupted:
2671         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2672             "Corrupted ZIP file data");
2673         return (ARCHIVE_FATAL);
2674 nomem:
2675         archive_set_error(&a->archive, ENOMEM,
2676             "No memory for ZIP decryption");
2677         return (ARCHIVE_FATAL);
2678 }
2679
2680 static int
2681 zip_alloc_decryption_buffer(struct archive_read *a)
2682 {
2683         struct zip *zip = (struct zip *)(a->format->data);
2684         size_t bs = 256 * 1024;
2685
2686         if (zip->decrypted_buffer == NULL) {
2687                 zip->decrypted_buffer_size = bs;
2688                 zip->decrypted_buffer = malloc(bs);
2689                 if (zip->decrypted_buffer == NULL) {
2690                         archive_set_error(&a->archive, ENOMEM,
2691                             "No memory for ZIP decryption");
2692                         return (ARCHIVE_FATAL);
2693                 }
2694         }
2695         zip->decrypted_ptr = zip->decrypted_buffer;
2696         return (ARCHIVE_OK);
2697 }
2698
2699 static int
2700 init_traditional_PKWARE_decryption(struct archive_read *a)
2701 {
2702         struct zip *zip = (struct zip *)(a->format->data);
2703         const void *p;
2704         int retry;
2705         int r;
2706
2707         if (zip->tctx_valid)
2708                 return (ARCHIVE_OK);
2709
2710         /*
2711            Read the 12 bytes encryption header stored at
2712            the start of the data area.
2713          */
2714 #define ENC_HEADER_SIZE 12
2715         if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2716             && zip->entry_bytes_remaining < ENC_HEADER_SIZE) {
2717                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2718                     "Truncated Zip encrypted body: only %jd bytes available",
2719                     (intmax_t)zip->entry_bytes_remaining);
2720                 return (ARCHIVE_FATAL);
2721         }
2722
2723         p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL);
2724         if (p == NULL) {
2725                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2726                     "Truncated ZIP file data");
2727                 return (ARCHIVE_FATAL);
2728         }
2729
2730         for (retry = 0;; retry++) {
2731                 const char *passphrase;
2732                 uint8_t crcchk;
2733
2734                 passphrase = __archive_read_next_passphrase(a);
2735                 if (passphrase == NULL) {
2736                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2737                             (retry > 0)?
2738                                 "Incorrect passphrase":
2739                                 "Passphrase required for this entry");
2740                         return (ARCHIVE_FAILED);
2741                 }
2742
2743                 /*
2744                  * Initialize ctx for Traditional PKWARE Decryption.
2745                  */
2746                 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase),
2747                         p, ENC_HEADER_SIZE, &crcchk);
2748                 if (r == 0 && crcchk == zip->entry->decdat)
2749                         break;/* The passphrase is OK. */
2750                 if (retry > 10000) {
2751                         /* Avoid infinity loop. */
2752                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2753                             "Too many incorrect passphrases");
2754                         return (ARCHIVE_FAILED);
2755                 }
2756         }
2757
2758         __archive_read_consume(a, ENC_HEADER_SIZE);
2759         zip->tctx_valid = 1;
2760         if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
2761             zip->entry_bytes_remaining -= ENC_HEADER_SIZE;
2762         }
2763         /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/
2764         zip->entry_compressed_bytes_read += ENC_HEADER_SIZE;
2765         zip->decrypted_bytes_remaining = 0;
2766
2767         return (zip_alloc_decryption_buffer(a));
2768 #undef ENC_HEADER_SIZE
2769 }
2770
2771 static int
2772 init_WinZip_AES_decryption(struct archive_read *a)
2773 {
2774         struct zip *zip = (struct zip *)(a->format->data);
2775         const void *p;
2776         const uint8_t *pv;
2777         size_t key_len, salt_len;
2778         uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE];
2779         int retry;
2780         int r;
2781
2782         if (zip->cctx_valid || zip->hctx_valid)
2783                 return (ARCHIVE_OK);
2784
2785         switch (zip->entry->aes_extra.strength) {
2786         case 1: salt_len = 8;  key_len = 16; break;
2787         case 2: salt_len = 12; key_len = 24; break;
2788         case 3: salt_len = 16; key_len = 32; break;
2789         default: goto corrupted;
2790         }
2791         p = __archive_read_ahead(a, salt_len + 2, NULL);
2792         if (p == NULL)
2793                 goto truncated;
2794
2795         for (retry = 0;; retry++) {
2796                 const char *passphrase;
2797
2798                 passphrase = __archive_read_next_passphrase(a);
2799                 if (passphrase == NULL) {
2800                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2801                             (retry > 0)?
2802                                 "Incorrect passphrase":
2803                                 "Passphrase required for this entry");
2804                         return (ARCHIVE_FAILED);
2805                 }
2806                 memset(derived_key, 0, sizeof(derived_key));
2807                 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase),
2808                     p, salt_len, 1000, derived_key, key_len * 2 + 2);
2809                 if (r != 0) {
2810                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2811                             "Decryption is unsupported due to lack of "
2812                             "crypto library");
2813                         return (ARCHIVE_FAILED);
2814                 }
2815
2816                 /* Check password verification value. */
2817                 pv = ((const uint8_t *)p) + salt_len;
2818                 if (derived_key[key_len * 2] == pv[0] &&
2819                     derived_key[key_len * 2 + 1] == pv[1])
2820                         break;/* The passphrase is OK. */
2821                 if (retry > 10000) {
2822                         /* Avoid infinity loop. */
2823                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2824                             "Too many incorrect passphrases");
2825                         return (ARCHIVE_FAILED);
2826                 }
2827         }
2828
2829         r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len);
2830         if (r != 0) {
2831                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2832                     "Decryption is unsupported due to lack of crypto library");
2833                 return (ARCHIVE_FAILED);
2834         }
2835         r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len);
2836         if (r != 0) {
2837                 archive_decrypto_aes_ctr_release(&zip->cctx);
2838                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2839                     "Failed to initialize HMAC-SHA1");
2840                 return (ARCHIVE_FAILED);
2841         }
2842         zip->cctx_valid = zip->hctx_valid = 1;
2843         __archive_read_consume(a, salt_len + 2);
2844         zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE;
2845         if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2846             && zip->entry_bytes_remaining < 0)
2847                 goto corrupted;
2848         zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE;
2849         zip->decrypted_bytes_remaining = 0;
2850
2851         zip->entry->compression = zip->entry->aes_extra.compression;
2852         return (zip_alloc_decryption_buffer(a));
2853
2854 truncated:
2855         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2856             "Truncated ZIP file data");
2857         return (ARCHIVE_FATAL);
2858 corrupted:
2859         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2860             "Corrupted ZIP file data");
2861         return (ARCHIVE_FATAL);
2862 }
2863
2864 static int
2865 archive_read_format_zip_read_data(struct archive_read *a,
2866     const void **buff, size_t *size, int64_t *offset)
2867 {
2868         int r;
2869         struct zip *zip = (struct zip *)(a->format->data);
2870
2871         if (zip->has_encrypted_entries ==
2872                         ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) {
2873                 zip->has_encrypted_entries = 0;
2874         }
2875
2876         *offset = zip->entry_uncompressed_bytes_read;
2877         *size = 0;
2878         *buff = NULL;
2879
2880         /* If we hit end-of-entry last time, return ARCHIVE_EOF. */
2881         if (zip->end_of_entry)
2882                 return (ARCHIVE_EOF);
2883
2884         /* Return EOF immediately if this is a non-regular file. */
2885         if (AE_IFREG != (zip->entry->mode & AE_IFMT))
2886                 return (ARCHIVE_EOF);
2887
2888         __archive_read_consume(a, zip->unconsumed);
2889         zip->unconsumed = 0;
2890
2891         if (zip->init_decryption) {
2892                 zip->has_encrypted_entries = 1;
2893                 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
2894                         r = read_decryption_header(a);
2895                 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
2896                         r = init_WinZip_AES_decryption(a);
2897                 else
2898                         r = init_traditional_PKWARE_decryption(a);
2899                 if (r != ARCHIVE_OK)
2900                         return (r);
2901                 zip->init_decryption = 0;
2902         }
2903
2904         switch(zip->entry->compression) {
2905         case 0:  /* No compression. */
2906                 r =  zip_read_data_none(a, buff, size, offset);
2907                 break;
2908 #ifdef HAVE_BZLIB_H
2909         case 12: /* ZIPx bzip2 compression. */
2910                 r = zip_read_data_zipx_bzip2(a, buff, size, offset);
2911                 break;
2912 #endif
2913 #if HAVE_LZMA_H && HAVE_LIBLZMA
2914         case 14: /* ZIPx LZMA compression. */
2915                 r = zip_read_data_zipx_lzma_alone(a, buff, size, offset);
2916                 break;
2917         case 95: /* ZIPx XZ compression. */
2918                 r = zip_read_data_zipx_xz(a, buff, size, offset);
2919                 break;
2920 #endif
2921         /* PPMd support is built-in, so we don't need any #if guards. */
2922         case 98: /* ZIPx PPMd compression. */
2923                 r = zip_read_data_zipx_ppmd(a, buff, size, offset);
2924                 break;
2925
2926 #ifdef HAVE_ZLIB_H
2927         case 8: /* Deflate compression. */
2928                 r =  zip_read_data_deflate(a, buff, size, offset);
2929                 break;
2930 #endif
2931         default: /* Unsupported compression. */
2932                 /* Return a warning. */
2933                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2934                     "Unsupported ZIP compression method (%d: %s)",
2935                     zip->entry->compression, compression_name(zip->entry->compression));
2936                 /* We can't decompress this entry, but we will
2937                  * be able to skip() it and try the next entry. */
2938                 return (ARCHIVE_FAILED);
2939                 break;
2940         }
2941         if (r != ARCHIVE_OK)
2942                 return (r);
2943         /* Update checksum */
2944         if (*size)
2945                 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff,
2946                     (unsigned)*size);
2947         /* If we hit the end, swallow any end-of-data marker. */
2948         if (zip->end_of_entry) {
2949                 /* Check file size, CRC against these values. */
2950                 if (zip->entry->compressed_size !=
2951                     zip->entry_compressed_bytes_read) {
2952                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2953                             "ZIP compressed data is wrong size "
2954                             "(read %jd, expected %jd)",
2955                             (intmax_t)zip->entry_compressed_bytes_read,
2956                             (intmax_t)zip->entry->compressed_size);
2957                         return (ARCHIVE_WARN);
2958                 }
2959                 /* Size field only stores the lower 32 bits of the actual
2960                  * size. */
2961                 if ((zip->entry->uncompressed_size & UINT32_MAX)
2962                     != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) {
2963                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2964                             "ZIP uncompressed data is wrong size "
2965                             "(read %jd, expected %jd)\n",
2966                             (intmax_t)zip->entry_uncompressed_bytes_read,
2967                             (intmax_t)zip->entry->uncompressed_size);
2968                         return (ARCHIVE_WARN);
2969                 }
2970                 /* Check computed CRC against header */
2971                 if ((!zip->hctx_valid ||
2972                       zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) &&
2973                    zip->entry->crc32 != zip->entry_crc32
2974                     && !zip->ignore_crc32) {
2975                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2976                             "ZIP bad CRC: 0x%lx should be 0x%lx",
2977                             (unsigned long)zip->entry_crc32,
2978                             (unsigned long)zip->entry->crc32);
2979                         return (ARCHIVE_WARN);
2980                 }
2981         }
2982
2983         return (ARCHIVE_OK);
2984 }
2985
2986 static int
2987 archive_read_format_zip_cleanup(struct archive_read *a)
2988 {
2989         struct zip *zip;
2990         struct zip_entry *zip_entry, *next_zip_entry;
2991
2992         zip = (struct zip *)(a->format->data);
2993
2994 #ifdef HAVE_ZLIB_H
2995         if (zip->stream_valid)
2996                 inflateEnd(&zip->stream);
2997 #endif
2998
2999 #if HAVE_LZMA_H && HAVE_LIBLZMA
3000     if (zip->zipx_lzma_valid) {
3001                 lzma_end(&zip->zipx_lzma_stream);
3002         }
3003 #endif
3004
3005 #ifdef HAVE_BZLIB_H
3006         if (zip->bzstream_valid) {
3007                 BZ2_bzDecompressEnd(&zip->bzstream);
3008         }
3009 #endif
3010
3011         free(zip->uncompressed_buffer);
3012
3013         if (zip->ppmd8_valid)
3014                 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
3015
3016         if (zip->zip_entries) {
3017                 zip_entry = zip->zip_entries;
3018                 while (zip_entry != NULL) {
3019                         next_zip_entry = zip_entry->next;
3020                         archive_string_free(&zip_entry->rsrcname);
3021                         free(zip_entry);
3022                         zip_entry = next_zip_entry;
3023                 }
3024         }
3025         free(zip->decrypted_buffer);
3026         if (zip->cctx_valid)
3027                 archive_decrypto_aes_ctr_release(&zip->cctx);
3028         if (zip->hctx_valid)
3029                 archive_hmac_sha1_cleanup(&zip->hctx);
3030         free(zip->iv);
3031         free(zip->erd);
3032         free(zip->v_data);
3033         archive_string_free(&zip->format_name);
3034         free(zip);
3035         (a->format->data) = NULL;
3036         return (ARCHIVE_OK);
3037 }
3038
3039 static int
3040 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a)
3041 {
3042         if (_a && _a->format) {
3043                 struct zip * zip = (struct zip *)_a->format->data;
3044                 if (zip) {
3045                         return zip->has_encrypted_entries;
3046                 }
3047         }
3048         return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
3049 }
3050
3051 static int
3052 archive_read_format_zip_options(struct archive_read *a,
3053     const char *key, const char *val)
3054 {
3055         struct zip *zip;
3056         int ret = ARCHIVE_FAILED;
3057
3058         zip = (struct zip *)(a->format->data);
3059         if (strcmp(key, "compat-2x")  == 0) {
3060                 /* Handle filenames as libarchive 2.x */
3061                 zip->init_default_conversion = (val != NULL) ? 1 : 0;
3062                 return (ARCHIVE_OK);
3063         } else if (strcmp(key, "hdrcharset")  == 0) {
3064                 if (val == NULL || val[0] == 0)
3065                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3066                             "zip: hdrcharset option needs a character-set name"
3067                         );
3068                 else {
3069                         zip->sconv = archive_string_conversion_from_charset(
3070                             &a->archive, val, 0);
3071                         if (zip->sconv != NULL) {
3072                                 if (strcmp(val, "UTF-8") == 0)
3073                                         zip->sconv_utf8 = zip->sconv;
3074                                 ret = ARCHIVE_OK;
3075                         } else
3076                                 ret = ARCHIVE_FATAL;
3077                 }
3078                 return (ret);
3079         } else if (strcmp(key, "ignorecrc32") == 0) {
3080                 /* Mostly useful for testing. */
3081                 if (val == NULL || val[0] == 0) {
3082                         zip->crc32func = real_crc32;
3083                         zip->ignore_crc32 = 0;
3084                 } else {
3085                         zip->crc32func = fake_crc32;
3086                         zip->ignore_crc32 = 1;
3087                 }
3088                 return (ARCHIVE_OK);
3089         } else if (strcmp(key, "mac-ext") == 0) {
3090                 zip->process_mac_extensions = (val != NULL && val[0] != 0);
3091                 return (ARCHIVE_OK);
3092         }
3093
3094         /* Note: The "warn" return is just to inform the options
3095          * supervisor that we didn't handle it.  It will generate
3096          * a suitable error if no one used this option. */
3097         return (ARCHIVE_WARN);
3098 }
3099
3100 int
3101 archive_read_support_format_zip(struct archive *a)
3102 {
3103         int r;
3104         r = archive_read_support_format_zip_streamable(a);
3105         if (r != ARCHIVE_OK)
3106                 return r;
3107         return (archive_read_support_format_zip_seekable(a));
3108 }
3109
3110 /* ------------------------------------------------------------------------ */
3111
3112 /*
3113  * Streaming-mode support
3114  */
3115
3116
3117 static int
3118 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a)
3119 {
3120         (void)a; /* UNUSED */
3121         return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
3122                 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
3123 }
3124
3125 static int
3126 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid)
3127 {
3128         const char *p;
3129
3130         (void)best_bid; /* UNUSED */
3131
3132         if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
3133                 return (-1);
3134
3135         /*
3136          * Bid of 29 here comes from:
3137          *  + 16 bits for "PK",
3138          *  + next 16-bit field has 6 options so contributes
3139          *    about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits
3140          *
3141          * So we've effectively verified ~29 total bits of check data.
3142          */
3143         if (p[0] == 'P' && p[1] == 'K') {
3144                 if ((p[2] == '\001' && p[3] == '\002')
3145                     || (p[2] == '\003' && p[3] == '\004')
3146                     || (p[2] == '\005' && p[3] == '\006')
3147                     || (p[2] == '\006' && p[3] == '\006')
3148                     || (p[2] == '\007' && p[3] == '\010')
3149                     || (p[2] == '0' && p[3] == '0'))
3150                         return (29);
3151         }
3152
3153         /* TODO: It's worth looking ahead a little bit for a valid
3154          * PK signature.  In particular, that would make it possible
3155          * to read some UUEncoded SFX files or SFX files coming from
3156          * a network socket. */
3157
3158         return (0);
3159 }
3160
3161 static int
3162 archive_read_format_zip_streamable_read_header(struct archive_read *a,
3163     struct archive_entry *entry)
3164 {
3165         struct zip *zip;
3166
3167         a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
3168         if (a->archive.archive_format_name == NULL)
3169                 a->archive.archive_format_name = "ZIP";
3170
3171         zip = (struct zip *)(a->format->data);
3172
3173         /*
3174          * It should be sufficient to call archive_read_next_header() for
3175          * a reader to determine if an entry is encrypted or not. If the
3176          * encryption of an entry is only detectable when calling
3177          * archive_read_data(), so be it. We'll do the same check there
3178          * as well.
3179          */
3180         if (zip->has_encrypted_entries ==
3181                         ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
3182                 zip->has_encrypted_entries = 0;
3183
3184         /* Make sure we have a zip_entry structure to use. */
3185         if (zip->zip_entries == NULL) {
3186                 zip->zip_entries = malloc(sizeof(struct zip_entry));
3187                 if (zip->zip_entries == NULL) {
3188                         archive_set_error(&a->archive, ENOMEM,
3189                             "Out  of memory");
3190                         return ARCHIVE_FATAL;
3191                 }
3192         }
3193         zip->entry = zip->zip_entries;
3194         memset(zip->entry, 0, sizeof(struct zip_entry));
3195
3196         if (zip->cctx_valid)
3197                 archive_decrypto_aes_ctr_release(&zip->cctx);
3198         if (zip->hctx_valid)
3199                 archive_hmac_sha1_cleanup(&zip->hctx);
3200         zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
3201         __archive_read_reset_passphrase(a);
3202
3203         /* Search ahead for the next local file header. */
3204         __archive_read_consume(a, zip->unconsumed);
3205         zip->unconsumed = 0;
3206         for (;;) {
3207                 int64_t skipped = 0;
3208                 const char *p, *end;
3209                 ssize_t bytes;
3210
3211                 p = __archive_read_ahead(a, 4, &bytes);
3212                 if (p == NULL)
3213                         return (ARCHIVE_FATAL);
3214                 end = p + bytes;
3215
3216                 while (p + 4 <= end) {
3217                         if (p[0] == 'P' && p[1] == 'K') {
3218                                 if (p[2] == '\003' && p[3] == '\004') {
3219                                         /* Regular file entry. */
3220                                         __archive_read_consume(a, skipped);
3221                                         return zip_read_local_file_header(a,
3222                                             entry, zip);
3223                                 }
3224
3225                               /*
3226                                * TODO: We cannot restore permissions
3227                                * based only on the local file headers.
3228                                * Consider scanning the central
3229                                * directory and returning additional
3230                                * entries for at least directories.
3231                                * This would allow us to properly set
3232                                * directory permissions.
3233                                *
3234                                * This won't help us fix symlinks
3235                                * and may not help with regular file
3236                                * permissions, either.  <sigh>
3237                                */
3238                               if (p[2] == '\001' && p[3] == '\002') {
3239                                       return (ARCHIVE_EOF);
3240                               }
3241
3242                               /* End of central directory?  Must be an
3243                                * empty archive. */
3244                               if ((p[2] == '\005' && p[3] == '\006')
3245                                   || (p[2] == '\006' && p[3] == '\006'))
3246                                       return (ARCHIVE_EOF);
3247                         }
3248                         ++p;
3249                         ++skipped;
3250                 }
3251                 __archive_read_consume(a, skipped);
3252         }
3253 }
3254
3255 static int
3256 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a)
3257 {
3258         struct zip *zip;
3259         int64_t bytes_skipped;
3260
3261         zip = (struct zip *)(a->format->data);
3262         bytes_skipped = __archive_read_consume(a, zip->unconsumed);
3263         zip->unconsumed = 0;
3264         if (bytes_skipped < 0)
3265                 return (ARCHIVE_FATAL);
3266
3267         /* If we've already read to end of data, we're done. */
3268         if (zip->end_of_entry)
3269                 return (ARCHIVE_OK);
3270
3271         /* So we know we're streaming... */
3272         if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
3273             || zip->entry->compressed_size > 0) {
3274                 /* We know the compressed length, so we can just skip. */
3275                 bytes_skipped = __archive_read_consume(a,
3276                                         zip->entry_bytes_remaining);
3277                 if (bytes_skipped < 0)
3278                         return (ARCHIVE_FATAL);
3279                 return (ARCHIVE_OK);
3280         }
3281
3282         if (zip->init_decryption) {
3283                 int r;
3284
3285                 zip->has_encrypted_entries = 1;
3286                 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
3287                         r = read_decryption_header(a);
3288                 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
3289                         r = init_WinZip_AES_decryption(a);
3290                 else
3291                         r = init_traditional_PKWARE_decryption(a);
3292                 if (r != ARCHIVE_OK)
3293                         return (r);
3294                 zip->init_decryption = 0;
3295         }
3296
3297         /* We're streaming and we don't know the length. */
3298         /* If the body is compressed and we know the format, we can
3299          * find an exact end-of-entry by decompressing it. */
3300         switch (zip->entry->compression) {
3301 #ifdef HAVE_ZLIB_H
3302         case 8: /* Deflate compression. */
3303                 while (!zip->end_of_entry) {
3304                         int64_t offset = 0;
3305                         const void *buff = NULL;
3306                         size_t size = 0;
3307                         int r;
3308                         r =  zip_read_data_deflate(a, &buff, &size, &offset);
3309                         if (r != ARCHIVE_OK)
3310                                 return (r);
3311                 }
3312                 return ARCHIVE_OK;
3313 #endif
3314         default: /* Uncompressed or unknown. */
3315                 /* Scan for a PK\007\010 signature. */
3316                 for (;;) {
3317                         const char *p, *buff;
3318                         ssize_t bytes_avail;
3319                         buff = __archive_read_ahead(a, 16, &bytes_avail);
3320                         if (bytes_avail < 16) {
3321                                 archive_set_error(&a->archive,
3322                                     ARCHIVE_ERRNO_FILE_FORMAT,
3323                                     "Truncated ZIP file data");
3324                                 return (ARCHIVE_FATAL);
3325                         }
3326                         p = buff;
3327                         while (p <= buff + bytes_avail - 16) {
3328                                 if (p[3] == 'P') { p += 3; }
3329                                 else if (p[3] == 'K') { p += 2; }
3330                                 else if (p[3] == '\007') { p += 1; }
3331                                 else if (p[3] == '\010' && p[2] == '\007'
3332                                     && p[1] == 'K' && p[0] == 'P') {
3333                                         if (zip->entry->flags & LA_USED_ZIP64)
3334                                                 __archive_read_consume(a,
3335                                                     p - buff + 24);
3336                                         else
3337                                                 __archive_read_consume(a,
3338                                                     p - buff + 16);
3339                                         return ARCHIVE_OK;
3340                                 } else { p += 4; }
3341                         }
3342                         __archive_read_consume(a, p - buff);
3343                 }
3344         }
3345 }
3346
3347 int
3348 archive_read_support_format_zip_streamable(struct archive *_a)
3349 {
3350         struct archive_read *a = (struct archive_read *)_a;
3351         struct zip *zip;
3352         int r;
3353
3354         archive_check_magic(_a, ARCHIVE_READ_MAGIC,
3355             ARCHIVE_STATE_NEW, "archive_read_support_format_zip");
3356
3357         zip = (struct zip *)calloc(1, sizeof(*zip));
3358         if (zip == NULL) {
3359                 archive_set_error(&a->archive, ENOMEM,
3360                     "Can't allocate zip data");
3361                 return (ARCHIVE_FATAL);
3362         }
3363
3364         /* Streamable reader doesn't support mac extensions. */
3365         zip->process_mac_extensions = 0;
3366
3367         /*
3368          * Until enough data has been read, we cannot tell about
3369          * any encrypted entries yet.
3370          */
3371         zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
3372         zip->crc32func = real_crc32;
3373
3374         r = __archive_read_register_format(a,
3375             zip,
3376             "zip",
3377             archive_read_format_zip_streamable_bid,
3378             archive_read_format_zip_options,
3379             archive_read_format_zip_streamable_read_header,
3380             archive_read_format_zip_read_data,
3381             archive_read_format_zip_read_data_skip_streamable,
3382             NULL,
3383             archive_read_format_zip_cleanup,
3384             archive_read_support_format_zip_capabilities_streamable,
3385             archive_read_format_zip_has_encrypted_entries);
3386
3387         if (r != ARCHIVE_OK)
3388                 free(zip);
3389         return (ARCHIVE_OK);
3390 }
3391
3392 /* ------------------------------------------------------------------------ */
3393
3394 /*
3395  * Seeking-mode support
3396  */
3397
3398 static int
3399 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a)
3400 {
3401         (void)a; /* UNUSED */
3402         return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
3403                 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
3404 }
3405
3406 /*
3407  * TODO: This is a performance sink because it forces the read core to
3408  * drop buffered data from the start of file, which will then have to
3409  * be re-read again if this bidder loses.
3410  *
3411  * We workaround this a little by passing in the best bid so far so
3412  * that later bidders can do nothing if they know they'll never
3413  * outbid.  But we can certainly do better...
3414  */
3415 static int
3416 read_eocd(struct zip *zip, const char *p, int64_t current_offset)
3417 {
3418         /* Sanity-check the EOCD we've found. */
3419
3420         /* This must be the first volume. */
3421         if (archive_le16dec(p + 4) != 0)
3422                 return 0;
3423         /* Central directory must be on this volume. */
3424         if (archive_le16dec(p + 4) != archive_le16dec(p + 6))
3425                 return 0;
3426         /* All central directory entries must be on this volume. */
3427         if (archive_le16dec(p + 10) != archive_le16dec(p + 8))
3428                 return 0;
3429         /* Central directory can't extend beyond start of EOCD record. */
3430         if (archive_le32dec(p + 16) + archive_le32dec(p + 12)
3431             > current_offset)
3432                 return 0;
3433
3434         /* Save the central directory location for later use. */
3435         zip->central_directory_offset = archive_le32dec(p + 16);
3436
3437         /* This is just a tiny bit higher than the maximum
3438            returned by the streaming Zip bidder.  This ensures
3439            that the more accurate seeking Zip parser wins
3440            whenever seek is available. */
3441         return 32;
3442 }
3443
3444 /*
3445  * Examine Zip64 EOCD locator:  If it's valid, store the information
3446  * from it.
3447  */
3448 static int
3449 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p)
3450 {
3451         int64_t eocd64_offset;
3452         int64_t eocd64_size;
3453
3454         /* Sanity-check the locator record. */
3455
3456         /* Central dir must be on first volume. */
3457         if (archive_le32dec(p + 4) != 0)
3458                 return 0;
3459         /* Must be only a single volume. */
3460         if (archive_le32dec(p + 16) != 1)
3461                 return 0;
3462
3463         /* Find the Zip64 EOCD record. */
3464         eocd64_offset = archive_le64dec(p + 8);
3465         if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0)
3466                 return 0;
3467         if ((p = __archive_read_ahead(a, 56, NULL)) == NULL)
3468                 return 0;
3469         /* Make sure we can read all of it. */
3470         eocd64_size = archive_le64dec(p + 4) + 12;
3471         if (eocd64_size < 56 || eocd64_size > 16384)
3472                 return 0;
3473         if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL)
3474                 return 0;
3475
3476         /* Sanity-check the EOCD64 */
3477         if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */
3478                 return 0;
3479         if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */
3480                 return 0;
3481         /* CD can't be split. */
3482         if (archive_le64dec(p + 24) != archive_le64dec(p + 32))
3483                 return 0;
3484
3485         /* Save the central directory offset for later use. */
3486         zip->central_directory_offset = archive_le64dec(p + 48);
3487
3488         return 32;
3489 }
3490
3491 static int
3492 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid)
3493 {
3494         struct zip *zip = (struct zip *)a->format->data;
3495         int64_t file_size, current_offset;
3496         const char *p;
3497         int i, tail;
3498
3499         /* If someone has already bid more than 32, then avoid
3500            trashing the look-ahead buffers with a seek. */
3501         if (best_bid > 32)
3502                 return (-1);
3503
3504         file_size = __archive_read_seek(a, 0, SEEK_END);
3505         if (file_size <= 0)
3506                 return 0;
3507
3508         /* Search last 16k of file for end-of-central-directory
3509          * record (which starts with PK\005\006) */
3510         tail = (int)zipmin(1024 * 16, file_size);
3511         current_offset = __archive_read_seek(a, -tail, SEEK_END);
3512         if (current_offset < 0)
3513                 return 0;
3514         if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL)
3515                 return 0;
3516         /* Boyer-Moore search backwards from the end, since we want
3517          * to match the last EOCD in the file (there can be more than
3518          * one if there is an uncompressed Zip archive as a member
3519          * within this Zip archive). */
3520         for (i = tail - 22; i > 0;) {
3521                 switch (p[i]) {
3522                 case 'P':
3523                         if (memcmp(p + i, "PK\005\006", 4) == 0) {
3524                                 int ret = read_eocd(zip, p + i,
3525                                     current_offset + i);
3526                                 /* Zip64 EOCD locator precedes
3527                                  * regular EOCD if present. */
3528                                 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) {
3529                                         int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20);
3530                                         if (ret_zip64 > ret)
3531                                                 ret = ret_zip64;
3532                                 }
3533                                 return (ret);
3534                         }
3535                         i -= 4;
3536                         break;
3537                 case 'K': i -= 1; break;
3538                 case 005: i -= 2; break;
3539                 case 006: i -= 3; break;
3540                 default: i -= 4; break;
3541                 }
3542         }
3543         return 0;
3544 }
3545
3546 /* The red-black trees are only used in seeking mode to manage
3547  * the in-memory copy of the central directory. */
3548
3549 static int
3550 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2)
3551 {
3552         const struct zip_entry *e1 = (const struct zip_entry *)n1;
3553         const struct zip_entry *e2 = (const struct zip_entry *)n2;
3554
3555         if (e1->local_header_offset > e2->local_header_offset)
3556                 return -1;
3557         if (e1->local_header_offset < e2->local_header_offset)
3558                 return 1;
3559         return 0;
3560 }
3561
3562 static int
3563 cmp_key(const struct archive_rb_node *n, const void *key)
3564 {
3565         /* This function won't be called */
3566         (void)n; /* UNUSED */
3567         (void)key; /* UNUSED */
3568         return 1;
3569 }
3570
3571 static const struct archive_rb_tree_ops rb_ops = {
3572         &cmp_node, &cmp_key
3573 };
3574
3575 static int
3576 rsrc_cmp_node(const struct archive_rb_node *n1,
3577     const struct archive_rb_node *n2)
3578 {
3579         const struct zip_entry *e1 = (const struct zip_entry *)n1;
3580         const struct zip_entry *e2 = (const struct zip_entry *)n2;
3581
3582         return (strcmp(e2->rsrcname.s, e1->rsrcname.s));
3583 }
3584
3585 static int
3586 rsrc_cmp_key(const struct archive_rb_node *n, const void *key)
3587 {
3588         const struct zip_entry *e = (const struct zip_entry *)n;
3589         return (strcmp((const char *)key, e->rsrcname.s));
3590 }
3591
3592 static const struct archive_rb_tree_ops rb_rsrc_ops = {
3593         &rsrc_cmp_node, &rsrc_cmp_key
3594 };
3595
3596 static const char *
3597 rsrc_basename(const char *name, size_t name_length)
3598 {
3599         const char *s, *r;
3600
3601         r = s = name;
3602         for (;;) {
3603                 s = memchr(s, '/', name_length - (s - name));
3604                 if (s == NULL)
3605                         break;
3606                 r = ++s;
3607         }
3608         return (r);
3609 }
3610
3611 static void
3612 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length)
3613 {
3614         struct archive_string str;
3615         struct zip_entry *dir;
3616         char *s;
3617
3618         archive_string_init(&str);
3619         archive_strncpy(&str, name, name_length);
3620         for (;;) {
3621                 s = strrchr(str.s, '/');
3622                 if (s == NULL)
3623                         break;
3624                 *s = '\0';
3625                 /* Transfer the parent directory from zip->tree_rsrc RB
3626                  * tree to zip->tree RB tree to expose. */
3627                 dir = (struct zip_entry *)
3628                     __archive_rb_tree_find_node(&zip->tree_rsrc, str.s);
3629                 if (dir == NULL)
3630                         break;
3631                 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node);
3632                 archive_string_free(&dir->rsrcname);
3633                 __archive_rb_tree_insert_node(&zip->tree, &dir->node);
3634         }
3635         archive_string_free(&str);
3636 }
3637
3638 static int
3639 slurp_central_directory(struct archive_read *a, struct archive_entry* entry,
3640     struct zip *zip)
3641 {
3642         ssize_t i;
3643         unsigned found;
3644         int64_t correction;
3645         ssize_t bytes_avail;
3646         const char *p;
3647
3648         /*
3649          * Find the start of the central directory.  The end-of-CD
3650          * record has our starting point, but there are lots of
3651          * Zip archives which have had other data prepended to the
3652          * file, which makes the recorded offsets all too small.
3653          * So we search forward from the specified offset until we
3654          * find the real start of the central directory.  Then we
3655          * know the correction we need to apply to account for leading
3656          * padding.
3657          */
3658         if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0)
3659                 return ARCHIVE_FATAL;
3660
3661         found = 0;
3662         while (!found) {
3663                 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL)
3664                         return ARCHIVE_FATAL;
3665                 for (found = 0, i = 0; !found && i < bytes_avail - 4;) {
3666                         switch (p[i + 3]) {
3667                         case 'P': i += 3; break;
3668                         case 'K': i += 2; break;
3669                         case 001: i += 1; break;
3670                         case 002:
3671                                 if (memcmp(p + i, "PK\001\002", 4) == 0) {
3672                                         p += i;
3673                                         found = 1;
3674                                 } else
3675                                         i += 4;
3676                                 break;
3677                         case 005: i += 1; break;
3678                         case 006:
3679                                 if (memcmp(p + i, "PK\005\006", 4) == 0) {
3680                                         p += i;
3681                                         found = 1;
3682                                 } else if (memcmp(p + i, "PK\006\006", 4) == 0) {
3683                                         p += i;
3684                                         found = 1;
3685                                 } else
3686                                         i += 1;
3687                                 break;
3688                         default: i += 4; break;
3689                         }
3690                 }
3691                 __archive_read_consume(a, i);
3692         }
3693         correction = archive_filter_bytes(&a->archive, 0)
3694                         - zip->central_directory_offset;
3695
3696         __archive_rb_tree_init(&zip->tree, &rb_ops);
3697         __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops);
3698
3699         zip->central_directory_entries_total = 0;
3700         while (1) {
3701                 struct zip_entry *zip_entry;
3702                 size_t filename_length, extra_length, comment_length;
3703                 uint32_t external_attributes;
3704                 const char *name, *r;
3705
3706                 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
3707                         return ARCHIVE_FATAL;
3708                 if (memcmp(p, "PK\006\006", 4) == 0
3709                     || memcmp(p, "PK\005\006", 4) == 0) {
3710                         break;
3711                 } else if (memcmp(p, "PK\001\002", 4) != 0) {
3712                         archive_set_error(&a->archive,
3713                             -1, "Invalid central directory signature");
3714                         return ARCHIVE_FATAL;
3715                 }
3716                 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL)
3717                         return ARCHIVE_FATAL;
3718
3719                 zip_entry = calloc(1, sizeof(struct zip_entry));
3720                 if (zip_entry == NULL) {
3721                         archive_set_error(&a->archive, ENOMEM,
3722                                 "Can't allocate zip entry");
3723                         return ARCHIVE_FATAL;
3724                 }
3725                 zip_entry->next = zip->zip_entries;
3726                 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY;
3727                 zip->zip_entries = zip_entry;
3728                 zip->central_directory_entries_total++;
3729
3730                 /* version = p[4]; */
3731                 zip_entry->system = p[5];
3732                 /* version_required = archive_le16dec(p + 6); */
3733                 zip_entry->zip_flags = archive_le16dec(p + 8);
3734                 if (zip_entry->zip_flags
3735                       & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){
3736                         zip->has_encrypted_entries = 1;
3737                 }
3738                 zip_entry->compression = (char)archive_le16dec(p + 10);
3739                 zip_entry->mtime = zip_time(p + 12);
3740                 zip_entry->crc32 = archive_le32dec(p + 16);
3741                 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
3742                         zip_entry->decdat = p[13];
3743                 else
3744                         zip_entry->decdat = p[19];
3745                 zip_entry->compressed_size = archive_le32dec(p + 20);
3746                 zip_entry->uncompressed_size = archive_le32dec(p + 24);
3747                 filename_length = archive_le16dec(p + 28);
3748                 extra_length = archive_le16dec(p + 30);
3749                 comment_length = archive_le16dec(p + 32);
3750                 /* disk_start = archive_le16dec(p + 34);
3751                  *   Better be zero.
3752                  * internal_attributes = archive_le16dec(p + 36);
3753                  *   text bit */
3754                 external_attributes = archive_le32dec(p + 38);
3755                 zip_entry->local_header_offset =
3756                     archive_le32dec(p + 42) + correction;
3757
3758                 /* If we can't guess the mode, leave it zero here;
3759                    when we read the local file header we might get
3760                    more information. */
3761                 if (zip_entry->system == 3) {
3762                         zip_entry->mode = external_attributes >> 16;
3763                 } else if (zip_entry->system == 0) {
3764                         // Interpret MSDOS directory bit
3765                         if (0x10 == (external_attributes & 0x10)) {
3766                                 zip_entry->mode = AE_IFDIR | 0775;
3767                         } else {
3768                                 zip_entry->mode = AE_IFREG | 0664;
3769                         }
3770                         if (0x01 == (external_attributes & 0x01)) {
3771                                 // Read-only bit; strip write permissions
3772                                 zip_entry->mode &= 0555;
3773                         }
3774                 } else {
3775                         zip_entry->mode = 0;
3776                 }
3777
3778                 /* We're done with the regular data; get the filename and
3779                  * extra data. */
3780                 __archive_read_consume(a, 46);
3781                 p = __archive_read_ahead(a, filename_length + extra_length,
3782                         NULL);
3783                 if (p == NULL) {
3784                         archive_set_error(&a->archive,
3785                             ARCHIVE_ERRNO_FILE_FORMAT,
3786                             "Truncated ZIP file header");
3787                         return ARCHIVE_FATAL;
3788                 }
3789                 if (ARCHIVE_OK != process_extra(a, entry, p + filename_length,
3790                     extra_length, zip_entry)) {
3791                         return ARCHIVE_FATAL;
3792                 }
3793
3794                 /*
3795                  * Mac resource fork files are stored under the
3796                  * "__MACOSX/" directory, so we should check if
3797                  * it is.
3798                  */
3799                 if (!zip->process_mac_extensions) {
3800                         /* Treat every entry as a regular entry. */
3801                         __archive_rb_tree_insert_node(&zip->tree,
3802                             &zip_entry->node);
3803                 } else {
3804                         name = p;
3805                         r = rsrc_basename(name, filename_length);
3806                         if (filename_length >= 9 &&
3807                             strncmp("__MACOSX/", name, 9) == 0) {
3808                                 /* If this file is not a resource fork nor
3809                                  * a directory. We should treat it as a non
3810                                  * resource fork file to expose it. */
3811                                 if (name[filename_length-1] != '/' &&
3812                                     (r - name < 3 || r[0] != '.' ||
3813                                      r[1] != '_')) {
3814                                         __archive_rb_tree_insert_node(
3815                                             &zip->tree, &zip_entry->node);
3816                                         /* Expose its parent directories. */
3817                                         expose_parent_dirs(zip, name,
3818                                             filename_length);
3819                                 } else {
3820                                         /* This file is a resource fork file or
3821                                          * a directory. */
3822                                         archive_strncpy(&(zip_entry->rsrcname),
3823                                              name, filename_length);
3824                                         __archive_rb_tree_insert_node(
3825                                             &zip->tree_rsrc, &zip_entry->node);
3826                                 }
3827                         } else {
3828                                 /* Generate resource fork name to find its
3829                                  * resource file at zip->tree_rsrc. */
3830                                 archive_strcpy(&(zip_entry->rsrcname),
3831                                     "__MACOSX/");
3832                                 archive_strncat(&(zip_entry->rsrcname),
3833                                     name, r - name);
3834                                 archive_strcat(&(zip_entry->rsrcname), "._");
3835                                 archive_strncat(&(zip_entry->rsrcname),
3836                                     name + (r - name),
3837                                     filename_length - (r - name));
3838                                 /* Register an entry to RB tree to sort it by
3839                                  * file offset. */
3840                                 __archive_rb_tree_insert_node(&zip->tree,
3841                                     &zip_entry->node);
3842                         }
3843                 }
3844
3845                 /* Skip the comment too ... */
3846                 __archive_read_consume(a,
3847                     filename_length + extra_length + comment_length);
3848         }
3849
3850         return ARCHIVE_OK;
3851 }
3852
3853 static ssize_t
3854 zip_get_local_file_header_size(struct archive_read *a, size_t extra)
3855 {
3856         const char *p;
3857         ssize_t filename_length, extra_length;
3858
3859         if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) {
3860                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3861                     "Truncated ZIP file header");
3862                 return (ARCHIVE_WARN);
3863         }
3864         p += extra;
3865
3866         if (memcmp(p, "PK\003\004", 4) != 0) {
3867                 archive_set_error(&a->archive, -1, "Damaged Zip archive");
3868                 return ARCHIVE_WARN;
3869         }
3870         filename_length = archive_le16dec(p + 26);
3871         extra_length = archive_le16dec(p + 28);
3872
3873         return (30 + filename_length + extra_length);
3874 }
3875
3876 static int
3877 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry,
3878     struct zip_entry *rsrc)
3879 {
3880         struct zip *zip = (struct zip *)a->format->data;
3881         unsigned char *metadata, *mp;
3882         int64_t offset = archive_filter_bytes(&a->archive, 0);
3883         size_t remaining_bytes, metadata_bytes;
3884         ssize_t hsize;
3885         int ret = ARCHIVE_OK, eof;
3886
3887         switch(rsrc->compression) {
3888         case 0:  /* No compression. */
3889                 if (rsrc->uncompressed_size != rsrc->compressed_size) {
3890                         archive_set_error(&a->archive,
3891                             ARCHIVE_ERRNO_FILE_FORMAT,
3892                             "Malformed OS X metadata entry: "
3893                             "inconsistent size");
3894                         return (ARCHIVE_FATAL);
3895                 }
3896 #ifdef HAVE_ZLIB_H
3897         case 8: /* Deflate compression. */
3898 #endif
3899                 break;
3900         default: /* Unsupported compression. */
3901                 /* Return a warning. */
3902                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3903                     "Unsupported ZIP compression method (%s)",
3904                     compression_name(rsrc->compression));
3905                 /* We can't decompress this entry, but we will
3906                  * be able to skip() it and try the next entry. */
3907                 return (ARCHIVE_WARN);
3908         }
3909
3910         if (rsrc->uncompressed_size > (4 * 1024 * 1024)) {
3911                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3912                     "Mac metadata is too large: %jd > 4M bytes",
3913                     (intmax_t)rsrc->uncompressed_size);
3914                 return (ARCHIVE_WARN);
3915         }
3916         if (rsrc->compressed_size > (4 * 1024 * 1024)) {
3917                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3918                     "Mac metadata is too large: %jd > 4M bytes",
3919                     (intmax_t)rsrc->compressed_size);
3920                 return (ARCHIVE_WARN);
3921         }
3922
3923         metadata = malloc((size_t)rsrc->uncompressed_size);
3924         if (metadata == NULL) {
3925                 archive_set_error(&a->archive, ENOMEM,
3926                     "Can't allocate memory for Mac metadata");
3927                 return (ARCHIVE_FATAL);
3928         }
3929
3930         if (offset < rsrc->local_header_offset)
3931                 __archive_read_consume(a, rsrc->local_header_offset - offset);
3932         else if (offset != rsrc->local_header_offset) {
3933                 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET);
3934         }
3935
3936         hsize = zip_get_local_file_header_size(a, 0);
3937         __archive_read_consume(a, hsize);
3938
3939         remaining_bytes = (size_t)rsrc->compressed_size;
3940         metadata_bytes = (size_t)rsrc->uncompressed_size;
3941         mp = metadata;
3942         eof = 0;
3943         while (!eof && remaining_bytes) {
3944                 const unsigned char *p;
3945                 ssize_t bytes_avail;
3946                 size_t bytes_used;
3947
3948                 p = __archive_read_ahead(a, 1, &bytes_avail);
3949                 if (p == NULL) {
3950                         archive_set_error(&a->archive,
3951                             ARCHIVE_ERRNO_FILE_FORMAT,
3952                             "Truncated ZIP file header");
3953                         ret = ARCHIVE_WARN;
3954                         goto exit_mac_metadata;
3955                 }
3956                 if ((size_t)bytes_avail > remaining_bytes)
3957                         bytes_avail = remaining_bytes;
3958                 switch(rsrc->compression) {
3959                 case 0:  /* No compression. */
3960                         if ((size_t)bytes_avail > metadata_bytes)
3961                                 bytes_avail = metadata_bytes;
3962                         memcpy(mp, p, bytes_avail);
3963                         bytes_used = (size_t)bytes_avail;
3964                         metadata_bytes -= bytes_used;
3965                         mp += bytes_used;
3966                         if (metadata_bytes == 0)
3967                                 eof = 1;
3968                         break;
3969 #ifdef HAVE_ZLIB_H
3970                 case 8: /* Deflate compression. */
3971                 {
3972                         int r;
3973
3974                         ret = zip_deflate_init(a, zip);
3975                         if (ret != ARCHIVE_OK)
3976                                 goto exit_mac_metadata;
3977                         zip->stream.next_in =
3978                             (Bytef *)(uintptr_t)(const void *)p;
3979                         zip->stream.avail_in = (uInt)bytes_avail;
3980                         zip->stream.total_in = 0;
3981                         zip->stream.next_out = mp;
3982                         zip->stream.avail_out = (uInt)metadata_bytes;
3983                         zip->stream.total_out = 0;
3984
3985                         r = inflate(&zip->stream, 0);
3986                         switch (r) {
3987                         case Z_OK:
3988                                 break;
3989                         case Z_STREAM_END:
3990                                 eof = 1;
3991                                 break;
3992                         case Z_MEM_ERROR:
3993                                 archive_set_error(&a->archive, ENOMEM,
3994                                     "Out of memory for ZIP decompression");
3995                                 ret = ARCHIVE_FATAL;
3996                                 goto exit_mac_metadata;
3997                         default:
3998                                 archive_set_error(&a->archive,
3999                                     ARCHIVE_ERRNO_MISC,
4000                                     "ZIP decompression failed (%d)", r);
4001                                 ret = ARCHIVE_FATAL;
4002                                 goto exit_mac_metadata;
4003                         }
4004                         bytes_used = zip->stream.total_in;
4005                         metadata_bytes -= zip->stream.total_out;
4006                         mp += zip->stream.total_out;
4007                         break;
4008                 }
4009 #endif
4010                 default:
4011                         bytes_used = 0;
4012                         break;
4013                 }
4014                 __archive_read_consume(a, bytes_used);
4015                 remaining_bytes -= bytes_used;
4016         }
4017         archive_entry_copy_mac_metadata(entry, metadata,
4018             (size_t)rsrc->uncompressed_size - metadata_bytes);
4019
4020 exit_mac_metadata:
4021         __archive_read_seek(a, offset, SEEK_SET);
4022         zip->decompress_init = 0;
4023         free(metadata);
4024         return (ret);
4025 }
4026
4027 static int
4028 archive_read_format_zip_seekable_read_header(struct archive_read *a,
4029         struct archive_entry *entry)
4030 {
4031         struct zip *zip = (struct zip *)a->format->data;
4032         struct zip_entry *rsrc;
4033         int64_t offset;
4034         int r, ret = ARCHIVE_OK;
4035
4036         /*
4037          * It should be sufficient to call archive_read_next_header() for
4038          * a reader to determine if an entry is encrypted or not. If the
4039          * encryption of an entry is only detectable when calling
4040          * archive_read_data(), so be it. We'll do the same check there
4041          * as well.
4042          */
4043         if (zip->has_encrypted_entries ==
4044                         ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
4045                 zip->has_encrypted_entries = 0;
4046
4047         a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
4048         if (a->archive.archive_format_name == NULL)
4049                 a->archive.archive_format_name = "ZIP";
4050
4051         if (zip->zip_entries == NULL) {
4052                 r = slurp_central_directory(a, entry, zip);
4053                 if (r != ARCHIVE_OK)
4054                         return r;
4055                 /* Get first entry whose local header offset is lower than
4056                  * other entries in the archive file. */
4057                 zip->entry =
4058                     (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree);
4059         } else if (zip->entry != NULL) {
4060                 /* Get next entry in local header offset order. */
4061                 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate(
4062                     &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT);
4063         }
4064
4065         if (zip->entry == NULL)
4066                 return ARCHIVE_EOF;
4067
4068         if (zip->entry->rsrcname.s)
4069                 rsrc = (struct zip_entry *)__archive_rb_tree_find_node(
4070                     &zip->tree_rsrc, zip->entry->rsrcname.s);
4071         else
4072                 rsrc = NULL;
4073
4074         if (zip->cctx_valid)
4075                 archive_decrypto_aes_ctr_release(&zip->cctx);
4076         if (zip->hctx_valid)
4077                 archive_hmac_sha1_cleanup(&zip->hctx);
4078         zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
4079         __archive_read_reset_passphrase(a);
4080
4081         /* File entries are sorted by the header offset, we should mostly
4082          * use __archive_read_consume to advance a read point to avoid
4083          * redundant data reading.  */
4084         offset = archive_filter_bytes(&a->archive, 0);
4085         if (offset < zip->entry->local_header_offset)
4086                 __archive_read_consume(a,
4087                     zip->entry->local_header_offset - offset);
4088         else if (offset != zip->entry->local_header_offset) {
4089                 __archive_read_seek(a, zip->entry->local_header_offset,
4090                     SEEK_SET);
4091         }
4092         zip->unconsumed = 0;
4093         r = zip_read_local_file_header(a, entry, zip);
4094         if (r != ARCHIVE_OK)
4095                 return r;
4096         if (rsrc) {
4097                 int ret2 = zip_read_mac_metadata(a, entry, rsrc);
4098                 if (ret2 < ret)
4099                         ret = ret2;
4100         }
4101         return (ret);
4102 }
4103
4104 /*
4105  * We're going to seek for the next header anyway, so we don't
4106  * need to bother doing anything here.
4107  */
4108 static int
4109 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a)
4110 {
4111         struct zip *zip;
4112         zip = (struct zip *)(a->format->data);
4113
4114         zip->unconsumed = 0;
4115         return (ARCHIVE_OK);
4116 }
4117
4118 int
4119 archive_read_support_format_zip_seekable(struct archive *_a)
4120 {
4121         struct archive_read *a = (struct archive_read *)_a;
4122         struct zip *zip;
4123         int r;
4124
4125         archive_check_magic(_a, ARCHIVE_READ_MAGIC,
4126             ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable");
4127
4128         zip = (struct zip *)calloc(1, sizeof(*zip));
4129         if (zip == NULL) {
4130                 archive_set_error(&a->archive, ENOMEM,
4131                     "Can't allocate zip data");
4132                 return (ARCHIVE_FATAL);
4133         }
4134
4135 #ifdef HAVE_COPYFILE_H
4136         /* Set this by default on Mac OS. */
4137         zip->process_mac_extensions = 1;
4138 #endif
4139
4140         /*
4141          * Until enough data has been read, we cannot tell about
4142          * any encrypted entries yet.
4143          */
4144         zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
4145         zip->crc32func = real_crc32;
4146
4147         r = __archive_read_register_format(a,
4148             zip,
4149             "zip",
4150             archive_read_format_zip_seekable_bid,
4151             archive_read_format_zip_options,
4152             archive_read_format_zip_seekable_read_header,
4153             archive_read_format_zip_read_data,
4154             archive_read_format_zip_read_data_skip_seekable,
4155             NULL,
4156             archive_read_format_zip_cleanup,
4157             archive_read_support_format_zip_capabilities_seekable,
4158             archive_read_format_zip_has_encrypted_entries);
4159
4160         if (r != ARCHIVE_OK)
4161                 free(zip);
4162         return (ARCHIVE_OK);
4163 }
4164
4165 /*# vim:set noet:*/