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
30 /*
31 * The definitive documentation of the Zip file format is:
32 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT
33 *
34 * The Info-Zip project has pioneered various extensions to better
35 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855
36 * "Ux", and 0x7875 "ux" extensions for time and ownership
37 * information.
38 *
39 * History of this code: The streaming Zip reader was first added to
40 * libarchive in January 2005. Support for seekable input sources was
41 * added in Nov 2011. Zip64 support (including a significant code
42 * refactoring) was added in 2014.
43 */
44
45 #ifdef HAVE_ERRNO_H
46 #include <errno.h>
47 #endif
48 #ifdef HAVE_STDLIB_H
49 #include <stdlib.h>
50 #endif
51 #ifdef HAVE_ZLIB_H
52 #include <zlib.h>
53 #endif
54 #ifdef HAVE_BZLIB_H
55 #include <bzlib.h>
56 #endif
57 #ifdef HAVE_LZMA_H
58 #include <lzma.h>
59 #endif
60 #ifdef HAVE_ZSTD_H
61 #include <zstd.h>
62 #endif
63
64 #include "archive.h"
65 #include "archive_digest_private.h"
66 #include "archive_cryptor_private.h"
67 #include "archive_endian.h"
68 #include "archive_entry.h"
69 #include "archive_entry_locale.h"
70 #include "archive_hmac_private.h"
71 #include "archive_private.h"
72 #include "archive_rb.h"
73 #include "archive_read_private.h"
74 #include "archive_time_private.h"
75 #include "archive_ppmd8_private.h"
76
77 #ifndef HAVE_ZLIB_H
78 #include "archive_crc32.h"
79 #endif
80
81 /* length of local file header, not including filename and extra */
82 #define ZIP_LOCHDR_LEN 30U
83
84 /* maximum length of Mac metadata in MiB */
85 #define ZIP_MAX_METADATA 10U
86
87 struct zip_entry {
88 struct archive_rb_node node;
89 struct zip_entry *next;
90 int64_t local_header_offset;
91 int64_t compressed_size;
92 int64_t uncompressed_size;
93 int64_t gid;
94 int64_t uid;
95 struct archive_string rsrcname;
96 time_t mtime;
97 time_t atime;
98 time_t ctime;
99 uint32_t crc32;
100 uint16_t mode;
101 uint16_t zip_flags; /* From GP Flags Field */
102 unsigned char compression;
103 unsigned char system; /* From "version written by" */
104 unsigned char flags; /* Our extra markers. */
105 unsigned char decdat;/* Used for Decryption check */
106
107 /* WinZip AES encryption extra field should be available
108 * when compression is 99. */
109 struct {
110 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */
111 unsigned vendor;
112 #define AES_VENDOR_AE_1 0x0001
113 #define AES_VENDOR_AE_2 0x0002
114 /* AES encryption strength:
115 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */
116 unsigned strength;
117 /* Actual compression method. */
118 unsigned char compression;
119 } aes_extra;
120 };
121
122 struct trad_enc_ctx {
123 uint32_t keys[3];
124 };
125
126 /* Bits used in zip_flags. */
127 #define ZIP_ENCRYPTED (1 << 0)
128 #define ZIP_LENGTH_AT_END (1 << 3) /* Also called "Streaming bit" */
129 #define ZIP_STRONG_ENCRYPTED (1 << 6)
130 #define ZIP_UTF8_NAME (1 << 11)
131 /* See "7.2 Single Password Symmetric Encryption Method"
132 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */
133 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13)
134
135 /* Bits used in flags. */
136 #define LA_USED_ZIP64 (1 << 0)
137 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1)
138
139 /*
140 * See "WinZip - AES Encryption Information"
141 * http://www.winzip.com/aes_info.htm
142 */
143 /* Value used in compression method. */
144 #define WINZIP_AES_ENCRYPTION 99
145 /* Authentication code size. */
146 #define AUTH_CODE_SIZE 10
147 /**/
148 #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2)
149
150 struct zip {
151 /* Structural information about the archive. */
152 struct archive_string format_name;
153 int64_t central_directory_offset;
154 int64_t central_directory_offset_adjusted;
155 size_t central_directory_entries_total;
156 size_t central_directory_entries_on_this_disk;
157 int has_encrypted_entries;
158
159 /* List of entries (seekable Zip only) */
160 struct zip_entry *zip_entries;
161 struct archive_rb_tree tree;
162 struct archive_rb_tree tree_rsrc;
163
164 /* Bytes read but not yet consumed via __archive_read_consume() */
165 size_t unconsumed;
166
167 /* Information about entry we're currently reading. */
168 struct zip_entry *entry;
169 int64_t entry_bytes_remaining;
170
171 /* These count the number of bytes actually read for the entry. */
172 int64_t entry_compressed_bytes_read;
173 int64_t entry_uncompressed_bytes_read;
174
175 /* Running CRC32 of the decompressed and decrypted data */
176 unsigned long computed_crc32;
177 unsigned long (*crc32func)(unsigned long, const void *,
178 size_t);
179 char ignore_crc32;
180
181 /* Flags to mark progress of decompression. */
182 char decompress_init;
183 char end_of_entry;
184
185 unsigned char *uncompressed_buffer;
186 size_t uncompressed_buffer_size;
187
188 #ifdef HAVE_ZLIB_H
189 z_stream stream;
190 char stream_valid;
191 #endif
192
193 #if HAVE_LZMA_H && HAVE_LIBLZMA
194 lzma_stream zipx_lzma_stream;
195 char zipx_lzma_valid;
196 #endif
197
198 #ifdef HAVE_BZLIB_H
199 bz_stream bzstream;
200 char bzstream_valid;
201 #endif
202
203 #if HAVE_ZSTD_H && HAVE_LIBZSTD
204 ZSTD_DStream *zstdstream;
205 char zstdstream_valid;
206 #endif
207
208 IByteIn zipx_ppmd_stream;
209 ssize_t zipx_ppmd_read_compressed;
210 CPpmd8 ppmd8;
211 char ppmd8_valid;
212 char ppmd8_stream_failed;
213
214 struct archive_string_conv *sconv;
215 struct archive_string_conv *sconv_default;
216 struct archive_string_conv *sconv_utf8;
217 int init_default_conversion;
218 int process_mac_extensions;
219
220 char init_decryption;
221
222 /* Decryption buffer. */
223 /*
224 * The decrypted data starts at decrypted_ptr and
225 * extends for decrypted_bytes_remaining. Decryption
226 * adds new data to the end of this block, data is returned
227 * to clients from the beginning. When the block hits the
228 * end of decrypted_buffer, it has to be shuffled back to
229 * the beginning of the buffer.
230 */
231 unsigned char *decrypted_buffer;
232 unsigned char *decrypted_ptr;
233 size_t decrypted_buffer_size;
234 size_t decrypted_bytes_remaining;
235 size_t decrypted_unconsumed_bytes;
236
237 /* Traditional PKWARE decryption. */
238 struct trad_enc_ctx tctx;
239 char tctx_valid;
240
241 /* WinZip AES decryption. */
242 /* Contexts used for AES decryption. */
243 archive_crypto_ctx cctx;
244 char cctx_valid;
245 archive_hmac_sha1_ctx hctx;
246 char hctx_valid;
247
248 /* Strong encryption's decryption header information. */
249 unsigned iv_size;
250 unsigned alg_id;
251 unsigned bit_len;
252 unsigned flags;
253 unsigned erd_size;
254 unsigned v_size;
255 unsigned v_crc32;
256 uint8_t *iv;
257 uint8_t *erd;
258 uint8_t *v_data;
259 };
260
261 /* Many systems define min or MIN, but not all. */
262 #define zipmin(a,b) ((a) < (b) ? (a) : (b))
263
264 #ifdef HAVE_ZLIB_H
265 static int
266 zip_read_data_deflate(struct archive_read *a, const void **buff,
267 size_t *size, int64_t *offset);
268 #endif
269 #if HAVE_LZMA_H && HAVE_LIBLZMA
270 static int
271 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff,
272 size_t *size, int64_t *offset);
273 #endif
274
275 /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8
276 * streams inside ZIP files. It has 2 purposes: one is to fetch the next
277 * compressed byte from the stream, second one is to increase the counter how
278 * many compressed bytes were read. */
279 static Byte
ppmd_read(void * p)280 ppmd_read(void* p) {
281 /* Get the handle to current decompression context. */
282 struct archive_read *a = ((IByteIn*)p)->a;
283 struct zip *zip = (struct zip*) a->format->data;
284 ssize_t bytes_avail = 0;
285
286 /* Fetch next byte. */
287 const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail);
288 if(bytes_avail < 1) {
289 zip->ppmd8_stream_failed = 1;
290 return 0;
291 }
292
293 __archive_read_consume(a, 1);
294
295 /* Increment the counter. */
296 ++zip->zipx_ppmd_read_compressed;
297
298 /* Return the next compressed byte. */
299 return data[0];
300 }
301
302 /* ------------------------------------------------------------------------ */
303
304 /*
305 Traditional PKWARE Decryption functions.
306 */
307
308 static void
trad_enc_update_keys(struct trad_enc_ctx * ctx,uint8_t c)309 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c)
310 {
311 uint8_t t;
312 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL)
313
314 ctx->keys[0] = CRC32(ctx->keys[0], c);
315 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1;
316 t = (ctx->keys[1] >> 24) & 0xff;
317 ctx->keys[2] = CRC32(ctx->keys[2], t);
318 #undef CRC32
319 }
320
321 static uint8_t
trad_enc_decrypt_byte(struct trad_enc_ctx * ctx)322 trad_enc_decrypt_byte(struct trad_enc_ctx *ctx)
323 {
324 unsigned temp = ctx->keys[2] | 2;
325 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff;
326 }
327
328 static void
trad_enc_decrypt_update(struct trad_enc_ctx * ctx,const uint8_t * in,size_t in_len,uint8_t * out,size_t out_len)329 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in,
330 size_t in_len, uint8_t *out, size_t out_len)
331 {
332 unsigned i, max;
333
334 max = (unsigned)((in_len < out_len)? in_len: out_len);
335
336 for (i = 0; i < max; i++) {
337 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx);
338 out[i] = t;
339 trad_enc_update_keys(ctx, t);
340 }
341 }
342
343 static int
trad_enc_init(struct trad_enc_ctx * ctx,const char * pw,size_t pw_len,const uint8_t * key,size_t key_len,uint8_t * crcchk)344 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len,
345 const uint8_t *key, size_t key_len, uint8_t *crcchk)
346 {
347 uint8_t header[12];
348
349 if (key_len < 12) {
350 *crcchk = 0xff;
351 return -1;
352 }
353
354 ctx->keys[0] = 305419896L;
355 ctx->keys[1] = 591751049L;
356 ctx->keys[2] = 878082192L;
357
358 for (;pw_len; --pw_len)
359 trad_enc_update_keys(ctx, *pw++);
360
361 trad_enc_decrypt_update(ctx, key, 12, header, 12);
362 /* Return the last byte for CRC check. */
363 *crcchk = header[11];
364 return 0;
365 }
366
367 #if 0
368 static void
369 crypt_derive_key_sha1(const void *p, int size, unsigned char *key,
370 int key_size)
371 {
372 #define MD_SIZE 20
373 archive_sha1_ctx ctx;
374 unsigned char md1[MD_SIZE];
375 unsigned char md2[MD_SIZE * 2];
376 unsigned char mkb[64];
377 int i;
378
379 archive_sha1_init(&ctx);
380 archive_sha1_update(&ctx, p, size);
381 archive_sha1_final(&ctx, md1);
382
383 memset(mkb, 0x36, sizeof(mkb));
384 for (i = 0; i < MD_SIZE; i++)
385 mkb[i] ^= md1[i];
386 archive_sha1_init(&ctx);
387 archive_sha1_update(&ctx, mkb, sizeof(mkb));
388 archive_sha1_final(&ctx, md2);
389
390 memset(mkb, 0x5C, sizeof(mkb));
391 for (i = 0; i < MD_SIZE; i++)
392 mkb[i] ^= md1[i];
393 archive_sha1_init(&ctx);
394 archive_sha1_update(&ctx, mkb, sizeof(mkb));
395 archive_sha1_final(&ctx, md2 + MD_SIZE);
396
397 if (key_size > 32)
398 key_size = 32;
399 memcpy(key, md2, key_size);
400 #undef MD_SIZE
401 }
402 #endif
403
404 /*
405 * Common code for streaming or seeking modes.
406 *
407 * Includes code to read local file headers, decompress data
408 * from entry bodies, and common API.
409 */
410
411 static unsigned long
real_crc32(unsigned long crc,const void * buff,size_t len)412 real_crc32(unsigned long crc, const void *buff, size_t len)
413 {
414 return crc32(crc, buff, (unsigned int)len);
415 }
416
417 /* Used by "ignorecrc32" option to speed up tests. */
418 static unsigned long
fake_crc32(unsigned long crc,const void * buff,size_t len)419 fake_crc32(unsigned long crc, const void *buff, size_t len)
420 {
421 (void)crc; /* UNUSED */
422 (void)buff; /* UNUSED */
423 (void)len; /* UNUSED */
424 return 0;
425 }
426
427 static const struct {
428 int id;
429 const char * name;
430 } compression_methods[] = {
431 {0, "uncompressed"}, /* The file is stored (no compression) */
432 {1, "shrinking"}, /* The file is Shrunk */
433 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */
434 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */
435 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */
436 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */
437 {6, "imploded"}, /* The file is Imploded */
438 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */
439 {8, "deflation"}, /* The file is Deflated */
440 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */
441 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding
442 * (old IBM TERSE) */
443 {11, "reserved"}, /* Reserved by PKWARE */
444 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */
445 {13, "reserved"}, /* Reserved by PKWARE */
446 {14, "lzma"}, /* LZMA (EFS) */
447 {15, "reserved"}, /* Reserved by PKWARE */
448 {16, "reserved"}, /* Reserved by PKWARE */
449 {17, "reserved"}, /* Reserved by PKWARE */
450 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */
451 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */
452 {93, "zstd"}, /* Zstandard (zstd) Compression */
453 {95, "xz"}, /* XZ compressed data */
454 {96, "jpeg"}, /* JPEG compressed data */
455 {97, "wav-pack"}, /* WavPack compressed data */
456 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */
457 {99, "aes"} /* WinZip AES encryption */
458 };
459
460 static const char *
compression_name(const int compression)461 compression_name(const int compression)
462 {
463 static const int num_compression_methods =
464 sizeof(compression_methods)/sizeof(compression_methods[0]);
465 int i=0;
466
467 while(compression >= 0 && i < num_compression_methods) {
468 if (compression_methods[i].id == compression)
469 return compression_methods[i].name;
470 i++;
471 }
472 return "??";
473 }
474
475 /*
476 * The extra data is stored as a list of
477 * id1+size1+data1 + id2+size2+data2 ...
478 * triplets. id and size are 2 bytes each.
479 */
480 static int
process_extra(struct archive_read * a,struct archive_entry * entry,const char * p,size_t extra_length,struct zip_entry * zip_entry)481 process_extra(struct archive_read *a, struct archive_entry *entry,
482 const char *p, size_t extra_length, struct zip_entry* zip_entry)
483 {
484 unsigned offset = 0;
485 struct zip *zip = (struct zip *)(a->format->data);
486
487 if (extra_length == 0) {
488 return ARCHIVE_OK;
489 }
490
491 if (extra_length < 4) {
492 size_t i = 0;
493 /* Some ZIP files may have trailing 0 bytes. Let's check they
494 * are all 0 and ignore them instead of returning an error.
495 *
496 * This is not technically correct, but some ZIP files look
497 * like this and other tools support those files - so let's
498 * also support them.
499 */
500 for (; i < extra_length; i++) {
501 if (p[i] != 0) {
502 archive_set_error(&a->archive,
503 ARCHIVE_ERRNO_FILE_FORMAT,
504 "Too-small extra data: "
505 "Need at least 4 bytes, "
506 "but only found %d bytes",
507 (int)extra_length);
508 return ARCHIVE_FAILED;
509 }
510 }
511
512 return ARCHIVE_OK;
513 }
514
515 while (offset <= extra_length - 4) {
516 unsigned short headerid = archive_le16dec(p + offset);
517 unsigned short datasize = archive_le16dec(p + offset + 2);
518
519 offset += 4;
520 if (offset + datasize > extra_length) {
521 archive_set_error(&a->archive,
522 ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: "
523 "Need %d bytes but only found %d bytes",
524 (int)datasize, (int)(extra_length - offset));
525 return ARCHIVE_FAILED;
526 }
527 #ifdef DEBUG
528 fprintf(stderr, "Header id 0x%04x, length %d\n",
529 headerid, datasize);
530 #endif
531 switch (headerid) {
532 case 0x0001:
533 /* Zip64 extended information extra field. */
534 zip_entry->flags |= LA_USED_ZIP64;
535 if (zip_entry->uncompressed_size == 0xffffffff) {
536 uint64_t t = 0;
537 if (datasize < 8
538 || (t = archive_le64dec(p + offset)) >
539 INT64_MAX) {
540 archive_set_error(&a->archive,
541 ARCHIVE_ERRNO_FILE_FORMAT,
542 "Malformed 64-bit "
543 "uncompressed size");
544 return ARCHIVE_FAILED;
545 }
546 zip_entry->uncompressed_size = t;
547 offset += 8;
548 datasize -= 8;
549 }
550 if (zip_entry->compressed_size == 0xffffffff) {
551 uint64_t t = 0;
552 if (datasize < 8
553 || (t = archive_le64dec(p + offset)) >
554 INT64_MAX) {
555 archive_set_error(&a->archive,
556 ARCHIVE_ERRNO_FILE_FORMAT,
557 "Malformed 64-bit "
558 "compressed size");
559 return ARCHIVE_FAILED;
560 }
561 zip_entry->compressed_size = t;
562 offset += 8;
563 datasize -= 8;
564 }
565 if (zip_entry->local_header_offset == 0xffffffff) {
566 uint64_t t = 0;
567 if (datasize < 8
568 || (t = archive_le64dec(p + offset)) >
569 INT64_MAX) {
570 archive_set_error(&a->archive,
571 ARCHIVE_ERRNO_FILE_FORMAT,
572 "Malformed 64-bit "
573 "local header offset");
574 return ARCHIVE_FAILED;
575 }
576 zip_entry->local_header_offset = t;
577 offset += 8;
578 datasize -= 8;
579 }
580 /* archive_le32dec(p + offset) gives disk
581 * on which file starts, but we don't handle
582 * multi-volume Zip files. */
583 break;
584 #ifdef DEBUG
585 case 0x0017:
586 {
587 /* Strong encryption field. */
588 if (archive_le16dec(p + offset) == 2) {
589 unsigned algId =
590 archive_le16dec(p + offset + 2);
591 unsigned bitLen =
592 archive_le16dec(p + offset + 4);
593 int flags =
594 archive_le16dec(p + offset + 6);
595 fprintf(stderr, "algId=0x%04x, bitLen=%u, "
596 "flgas=%d\n", algId, bitLen,flags);
597 }
598 break;
599 }
600 #endif
601 case 0x5455:
602 {
603 /* Extended time field "UT". */
604 int flags;
605 if (datasize == 0) {
606 archive_set_error(&a->archive,
607 ARCHIVE_ERRNO_FILE_FORMAT,
608 "Incomplete extended time field");
609 return ARCHIVE_FAILED;
610 }
611 flags = p[offset];
612 offset++;
613 datasize--;
614 /* Flag bits indicate which dates are present. */
615 if (flags & 0x01)
616 {
617 #ifdef DEBUG
618 fprintf(stderr, "mtime: %lld -> %d\n",
619 (long long)zip_entry->mtime,
620 archive_le32dec(p + offset));
621 #endif
622 if (datasize < 4)
623 break;
624 zip_entry->mtime = archive_le32dec(p + offset);
625 offset += 4;
626 datasize -= 4;
627 }
628 if (flags & 0x02)
629 {
630 if (datasize < 4)
631 break;
632 zip_entry->atime = archive_le32dec(p + offset);
633 offset += 4;
634 datasize -= 4;
635 }
636 if (flags & 0x04)
637 {
638 if (datasize < 4)
639 break;
640 zip_entry->ctime = archive_le32dec(p + offset);
641 offset += 4;
642 datasize -= 4;
643 }
644 break;
645 }
646 case 0x5855:
647 {
648 /* Info-ZIP Unix Extra Field (old version) "UX". */
649 if (datasize >= 8) {
650 zip_entry->atime = archive_le32dec(p + offset);
651 zip_entry->mtime =
652 archive_le32dec(p + offset + 4);
653 }
654 if (datasize >= 12) {
655 zip_entry->uid =
656 archive_le16dec(p + offset + 8);
657 zip_entry->gid =
658 archive_le16dec(p + offset + 10);
659 }
660 break;
661 }
662 case 0x6c78:
663 {
664 /* Experimental 'xl' field */
665 /*
666 * Introduced Dec 2013 to provide a way to
667 * include external file attributes (and other
668 * fields that ordinarily appear only in
669 * central directory) in local file header.
670 * This provides file type and permission
671 * information necessary to support full
672 * streaming extraction. Currently being
673 * discussed with other Zip developers
674 * ... subject to change.
675 *
676 * Format:
677 * The field starts with a bitmap that specifies
678 * which additional fields are included. The
679 * bitmap is variable length and can be extended in
680 * the future.
681 *
682 * n bytes - feature bitmap: first byte has low-order
683 * 7 bits. If high-order bit is set, a subsequent
684 * byte holds the next 7 bits, etc.
685 *
686 * if bitmap & 1, 2 byte "version made by"
687 * if bitmap & 2, 2 byte "internal file attributes"
688 * if bitmap & 4, 4 byte "external file attributes"
689 * if bitmap & 8, 2 byte comment length + n byte
690 * comment
691 */
692 int bitmap, bitmap_last;
693
694 if (datasize < 1)
695 break;
696 bitmap_last = bitmap = 0xff & p[offset];
697 offset += 1;
698 datasize -= 1;
699
700 /* We only support first 7 bits of bitmap; skip rest. */
701 while ((bitmap_last & 0x80) != 0
702 && datasize >= 1) {
703 bitmap_last = p[offset];
704 offset += 1;
705 datasize -= 1;
706 }
707
708 if (bitmap & 1) {
709 /* 2 byte "version made by" */
710 if (datasize < 2)
711 break;
712 zip_entry->system
713 = archive_le16dec(p + offset) >> 8;
714 offset += 2;
715 datasize -= 2;
716 }
717 if (bitmap & 2) {
718 /* 2 byte "internal file attributes" */
719 uint32_t internal_attributes;
720 if (datasize < 2)
721 break;
722 internal_attributes
723 = archive_le16dec(p + offset);
724 /* Not used by libarchive at present. */
725 (void)internal_attributes; /* UNUSED */
726 offset += 2;
727 datasize -= 2;
728 }
729 if (bitmap & 4) {
730 /* 4 byte "external file attributes" */
731 uint32_t external_attributes;
732 if (datasize < 4)
733 break;
734 external_attributes
735 = archive_le32dec(p + offset);
736 if (zip_entry->system == 3) {
737 zip_entry->mode
738 = external_attributes >> 16;
739 } else if (zip_entry->system == 0) {
740 // Interpret MSDOS directory bit
741 if (0x10 == (external_attributes &
742 0x10)) {
743 zip_entry->mode =
744 AE_IFDIR | 0775;
745 } else {
746 zip_entry->mode =
747 AE_IFREG | 0664;
748 }
749 if (0x01 == (external_attributes &
750 0x01)) {
751 /* Read-only bit;
752 * strip write permissions */
753 zip_entry->mode &= 0555;
754 }
755 } else {
756 zip_entry->mode = 0;
757 }
758 offset += 4;
759 datasize -= 4;
760 }
761 if (bitmap & 8) {
762 /* 2 byte comment length + comment */
763 uint32_t comment_length;
764 if (datasize < 2)
765 break;
766 comment_length
767 = archive_le16dec(p + offset);
768 offset += 2;
769 datasize -= 2;
770
771 if (datasize < comment_length)
772 break;
773 /* Comment is not supported by libarchive */
774 offset += comment_length;
775 datasize -= comment_length;
776 }
777 break;
778 }
779 case 0x7075:
780 {
781 /* Info-ZIP Unicode Path Extra Field. */
782 if (datasize < 5 || entry == NULL)
783 break;
784 offset += 5;
785 datasize -= 5;
786
787 /* The path name in this field is always encoded
788 * in UTF-8. */
789 if (zip->sconv_utf8 == NULL) {
790 zip->sconv_utf8 =
791 archive_string_conversion_from_charset(
792 &a->archive, "UTF-8", 1);
793 /* If the converter from UTF-8 is not
794 * available, then the path name from the main
795 * field will more likely be correct. */
796 if (zip->sconv_utf8 == NULL)
797 break;
798 }
799
800 /* Make sure the CRC32 of the filename matches. */
801 if (!zip->ignore_crc32) {
802 const char *cp = archive_entry_pathname(entry);
803 if (cp) {
804 unsigned long file_crc =
805 zip->crc32func(0, cp, strlen(cp));
806 unsigned long utf_crc =
807 archive_le32dec(p + offset - 4);
808 if (file_crc != utf_crc) {
809 #ifdef DEBUG
810 fprintf(stderr,
811 "CRC filename mismatch; "
812 "CDE is %lx, but UTF8 "
813 "is outdated with %lx\n",
814 file_crc, utf_crc);
815 #endif
816 break;
817 }
818 }
819 }
820
821 if (archive_entry_copy_pathname_l(entry,
822 p + offset, datasize, zip->sconv_utf8) != 0) {
823 /* Ignore the error, and fallback to the path
824 * name from the main field. */
825 #ifdef DEBUG
826 fprintf(stderr, "Failed to read the ZIP "
827 "0x7075 extra field path.\n");
828 #endif
829 }
830 break;
831 }
832 case 0x7855:
833 /* Info-ZIP Unix Extra Field (type 2) "Ux". */
834 #ifdef DEBUG
835 fprintf(stderr, "uid %d gid %d\n",
836 archive_le16dec(p + offset),
837 archive_le16dec(p + offset + 2));
838 #endif
839 if (datasize >= 2)
840 zip_entry->uid = archive_le16dec(p + offset);
841 if (datasize >= 4)
842 zip_entry->gid =
843 archive_le16dec(p + offset + 2);
844 break;
845 case 0x7875:
846 {
847 /* Info-Zip Unix Extra Field (type 3) "ux". */
848 int uidsize = 0, gidsize = 0;
849
850 /* TODO: support arbitrary uidsize/gidsize. */
851 if (datasize >= 1 && p[offset] == 1) {/* version=1 */
852 if (datasize >= 4) {
853 /* get a uid size. */
854 uidsize = 0xff & (int)p[offset+1];
855 if (uidsize == 2)
856 zip_entry->uid =
857 archive_le16dec(
858 p + offset + 2);
859 else if (uidsize == 4 && datasize >= 6)
860 zip_entry->uid =
861 archive_le32dec(
862 p + offset + 2);
863 }
864 if (datasize >= (2 + uidsize + 3)) {
865 /* get a gid size. */
866 gidsize = 0xff &
867 (int)p[offset+2+uidsize];
868 if (gidsize == 2)
869 zip_entry->gid =
870 archive_le16dec(
871 p+offset+2+uidsize+1);
872 else if (gidsize == 4 &&
873 datasize >= (2 + uidsize + 5))
874 zip_entry->gid =
875 archive_le32dec(
876 p+offset+2+uidsize+1);
877 }
878 }
879 break;
880 }
881 case 0x9901:
882 /* WinZip AES extra data field. */
883 if (datasize < 6) {
884 archive_set_error(&a->archive,
885 ARCHIVE_ERRNO_FILE_FORMAT,
886 "Incomplete AES field");
887 return ARCHIVE_FAILED;
888 }
889 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') {
890 /* Vendor version. */
891 zip_entry->aes_extra.vendor =
892 archive_le16dec(p + offset);
893 /* AES encryption strength. */
894 zip_entry->aes_extra.strength = p[offset + 4];
895 /* Actual compression method. */
896 zip_entry->aes_extra.compression =
897 p[offset + 5];
898 }
899 break;
900 default:
901 break;
902 }
903 offset += datasize;
904 }
905 return ARCHIVE_OK;
906 }
907
908 /*
909 * Assumes file pointer is at beginning of local file header.
910 */
911 static int
zip_read_local_file_header(struct archive_read * a,struct archive_entry * entry,struct zip * zip)912 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry,
913 struct zip *zip)
914 {
915 const char *p;
916 const void *h;
917 const wchar_t *wp;
918 const char *cp;
919 size_t len, filename_length, extra_length;
920 struct archive_string_conv *sconv;
921 struct zip_entry *zip_entry = zip->entry;
922 struct zip_entry zip_entry_central_dir;
923 int ret = ARCHIVE_OK;
924 char version;
925
926 /* Save a copy of the original for consistency checks. */
927 zip_entry_central_dir = *zip_entry;
928
929 zip->decompress_init = 0;
930 zip->end_of_entry = 0;
931 zip->entry_uncompressed_bytes_read = 0;
932 zip->entry_compressed_bytes_read = 0;
933 zip->computed_crc32 = zip->crc32func(0, NULL, 0);
934
935 /* Setup default conversion. */
936 if (zip->sconv == NULL && !zip->init_default_conversion) {
937 zip->sconv_default =
938 archive_string_default_conversion_for_read(&(a->archive));
939 zip->init_default_conversion = 1;
940 }
941
942 if ((p = __archive_read_ahead(a, ZIP_LOCHDR_LEN, NULL)) == NULL) {
943 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
944 "Truncated ZIP file header");
945 return (ARCHIVE_FATAL);
946 }
947
948 if (memcmp(p, "PK\003\004", 4) != 0) {
949 archive_set_error(&a->archive, -1, "Damaged Zip archive");
950 return ARCHIVE_FATAL;
951 }
952 version = p[4];
953 zip_entry->system = p[5];
954 zip_entry->zip_flags = archive_le16dec(p + 6);
955 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) {
956 zip->has_encrypted_entries = 1;
957 archive_entry_set_is_data_encrypted(entry, 1);
958 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED &&
959 zip_entry->zip_flags & ZIP_ENCRYPTED &&
960 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) {
961 archive_entry_set_is_metadata_encrypted(entry, 1);
962 return ARCHIVE_FATAL;
963 }
964 }
965 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED);
966 zip_entry->compression = (char)archive_le16dec(p + 8);
967 zip_entry->mtime = dos_to_unix(archive_le32dec(p + 10));
968 zip_entry->crc32 = archive_le32dec(p + 14);
969 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
970 zip_entry->decdat = p[11];
971 else
972 zip_entry->decdat = p[17];
973 zip_entry->compressed_size = archive_le32dec(p + 18);
974 zip_entry->uncompressed_size = archive_le32dec(p + 22);
975 filename_length = archive_le16dec(p + 26);
976 extra_length = archive_le16dec(p + 28);
977
978 __archive_read_consume(a, ZIP_LOCHDR_LEN);
979
980 /* Read the filename. */
981 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) {
982 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
983 "Truncated ZIP file header");
984 return (ARCHIVE_FATAL);
985 }
986 if (zip_entry->zip_flags & ZIP_UTF8_NAME) {
987 /* The filename is stored to be UTF-8. */
988 if (zip->sconv_utf8 == NULL) {
989 zip->sconv_utf8 =
990 archive_string_conversion_from_charset(
991 &a->archive, "UTF-8", 1);
992 if (zip->sconv_utf8 == NULL)
993 return (ARCHIVE_FATAL);
994 }
995 sconv = zip->sconv_utf8;
996 } else if (zip->sconv != NULL)
997 sconv = zip->sconv;
998 else
999 sconv = zip->sconv_default;
1000
1001 if (archive_entry_copy_pathname_l(entry,
1002 h, filename_length, sconv) != 0) {
1003 if (errno == ENOMEM) {
1004 archive_set_error(&a->archive, ENOMEM,
1005 "Can't allocate memory for Pathname");
1006 return (ARCHIVE_FATAL);
1007 }
1008 archive_set_error(&a->archive,
1009 ARCHIVE_ERRNO_FILE_FORMAT,
1010 "Pathname cannot be converted "
1011 "from %s to current locale.",
1012 archive_string_conversion_charset_name(sconv));
1013 ret = ARCHIVE_WARN;
1014 }
1015 __archive_read_consume(a, filename_length);
1016
1017 /* Read the extra data. */
1018 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) {
1019 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1020 "Truncated ZIP file header");
1021 return (ARCHIVE_FATAL);
1022 }
1023
1024 if (ARCHIVE_OK != process_extra(a, entry, h, extra_length,
1025 zip_entry)) {
1026 return ARCHIVE_FATAL;
1027 }
1028 __archive_read_consume(a, extra_length);
1029
1030 /* Work around a bug in Info-Zip: When reading from a pipe, it
1031 * stats the pipe instead of synthesizing a file entry. */
1032 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) {
1033 zip_entry->mode &= ~ AE_IFMT;
1034 zip_entry->mode |= AE_IFREG;
1035 }
1036
1037 /* If the mode is totally empty, set some sane default. */
1038 if (zip_entry->mode == 0) {
1039 zip_entry->mode |= 0664;
1040 }
1041
1042 /* Windows archivers sometimes use backslash as the directory
1043 * separator. Normalize to slash. */
1044 if (zip_entry->system == 0 &&
1045 (wp = archive_entry_pathname_w(entry)) != NULL) {
1046 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) {
1047 size_t i;
1048 struct archive_wstring s;
1049 archive_string_init(&s);
1050 archive_wstrcpy(&s, wp);
1051 for (i = 0; i < archive_strlen(&s); i++) {
1052 if (s.s[i] == '\\')
1053 s.s[i] = '/';
1054 }
1055 archive_entry_copy_pathname_w(entry, s.s);
1056 archive_wstring_free(&s);
1057 }
1058 }
1059
1060 /* Make sure that entries with a trailing '/' are marked as directories
1061 * even if the External File Attributes contains bogus values. If this
1062 * is not a directory and there is no type, assume a regular file. */
1063 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) {
1064 int has_slash;
1065
1066 wp = archive_entry_pathname_w(entry);
1067 if (wp != NULL) {
1068 len = wcslen(wp);
1069 has_slash = len > 0 && wp[len - 1] == L'/';
1070 } else {
1071 cp = archive_entry_pathname(entry);
1072 len = (cp != NULL)?strlen(cp):0;
1073 has_slash = len > 0 && cp[len - 1] == '/';
1074 }
1075 /* Correct file type as needed. */
1076 if (has_slash) {
1077 zip_entry->mode &= ~AE_IFMT;
1078 zip_entry->mode |= AE_IFDIR;
1079 zip_entry->mode |= 0111;
1080 } else if ((zip_entry->mode & AE_IFMT) == 0) {
1081 zip_entry->mode |= AE_IFREG;
1082 }
1083 }
1084
1085 /* Make sure directories end in '/' */
1086 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) {
1087 wp = archive_entry_pathname_w(entry);
1088 if (wp != NULL) {
1089 len = wcslen(wp);
1090 if (len > 0 && wp[len - 1] != L'/') {
1091 struct archive_wstring s;
1092 archive_string_init(&s);
1093 archive_wstrcat(&s, wp);
1094 archive_wstrappend_wchar(&s, L'/');
1095 archive_entry_copy_pathname_w(entry, s.s);
1096 archive_wstring_free(&s);
1097 }
1098 } else {
1099 cp = archive_entry_pathname(entry);
1100 len = (cp != NULL)?strlen(cp):0;
1101 if (len > 0 && cp[len - 1] != '/') {
1102 struct archive_string s;
1103 archive_string_init(&s);
1104 archive_strcat(&s, cp);
1105 archive_strappend_char(&s, '/');
1106 archive_entry_set_pathname(entry, s.s);
1107 archive_string_free(&s);
1108 }
1109 }
1110 }
1111
1112 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) {
1113 /* If this came from the central dir, its size info
1114 * is definitive, so ignore the length-at-end flag. */
1115 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END;
1116 /* If local header is missing a value, use the one from
1117 the central directory. If both have it, warn about
1118 mismatches. */
1119 if (zip_entry->crc32 == 0) {
1120 zip_entry->crc32 = zip_entry_central_dir.crc32;
1121 } else if (!zip->ignore_crc32
1122 && zip_entry->crc32 != zip_entry_central_dir.crc32) {
1123 archive_set_error(&a->archive,
1124 ARCHIVE_ERRNO_FILE_FORMAT,
1125 "Inconsistent CRC32 values");
1126 ret = ARCHIVE_WARN;
1127 }
1128 if (zip_entry->compressed_size == 0
1129 || zip_entry->compressed_size == 0xffffffff) {
1130 zip_entry->compressed_size
1131 = zip_entry_central_dir.compressed_size;
1132 } else if (zip_entry->compressed_size
1133 != zip_entry_central_dir.compressed_size) {
1134 archive_set_error(&a->archive,
1135 ARCHIVE_ERRNO_FILE_FORMAT,
1136 "Inconsistent compressed size: "
1137 "%jd in central directory, %jd in local header",
1138 (intmax_t)zip_entry_central_dir.compressed_size,
1139 (intmax_t)zip_entry->compressed_size);
1140 ret = ARCHIVE_WARN;
1141 }
1142 if (zip_entry->uncompressed_size == 0 ||
1143 zip_entry->uncompressed_size == 0xffffffff) {
1144 zip_entry->uncompressed_size
1145 = zip_entry_central_dir.uncompressed_size;
1146 } else if (zip_entry->uncompressed_size
1147 != zip_entry_central_dir.uncompressed_size) {
1148 archive_set_error(&a->archive,
1149 ARCHIVE_ERRNO_FILE_FORMAT,
1150 "Inconsistent uncompressed size: "
1151 "%jd in central directory, %jd in local header",
1152 (intmax_t)zip_entry_central_dir.uncompressed_size,
1153 (intmax_t)zip_entry->uncompressed_size);
1154 ret = ARCHIVE_WARN;
1155 }
1156 }
1157
1158 /* Populate some additional entry fields: */
1159 archive_entry_set_mode(entry, zip_entry->mode);
1160 archive_entry_set_uid(entry, zip_entry->uid);
1161 archive_entry_set_gid(entry, zip_entry->gid);
1162 archive_entry_set_mtime(entry, zip_entry->mtime, 0);
1163 archive_entry_set_ctime(entry, zip_entry->ctime, 0);
1164 archive_entry_set_atime(entry, zip_entry->atime, 0);
1165
1166 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) {
1167 size_t linkname_length;
1168
1169 if (zip_entry->compressed_size > 64 * 1024) {
1170 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1171 "Zip file with oversized link entry");
1172 return ARCHIVE_FATAL;
1173 }
1174
1175 linkname_length = (size_t)zip_entry->compressed_size;
1176
1177 archive_entry_set_size(entry, 0);
1178
1179 // take into account link compression if any
1180 size_t linkname_full_length = linkname_length;
1181 if (zip->entry->compression != 0)
1182 {
1183 // symlink target string appeared to be compressed
1184 int status = ARCHIVE_FATAL;
1185 const void *uncompressed_buffer = NULL;
1186
1187 switch (zip->entry->compression)
1188 {
1189 #if HAVE_ZLIB_H
1190 case 8: /* Deflate compression. */
1191 zip->entry_bytes_remaining = zip_entry->compressed_size;
1192 status = zip_read_data_deflate(a, &uncompressed_buffer,
1193 &linkname_full_length, NULL);
1194 break;
1195 #endif
1196 #if HAVE_LZMA_H && HAVE_LIBLZMA
1197 case 14: /* ZIPx LZMA compression. */
1198 /*(see zip file format specification, section 4.4.5)*/
1199 zip->entry_bytes_remaining = zip_entry->compressed_size;
1200 status = zip_read_data_zipx_lzma_alone(a, &uncompressed_buffer,
1201 &linkname_full_length, NULL);
1202 break;
1203 #endif
1204 default: /* Unsupported compression. */
1205 break;
1206 }
1207 if (status == ARCHIVE_OK)
1208 {
1209 p = uncompressed_buffer;
1210 }
1211 else
1212 {
1213 archive_set_error(&a->archive,
1214 ARCHIVE_ERRNO_FILE_FORMAT,
1215 "Unsupported ZIP compression method "
1216 "during decompression of link entry (%d: %s)",
1217 zip->entry->compression,
1218 compression_name(zip->entry->compression));
1219 return ARCHIVE_FAILED;
1220 }
1221 }
1222 else
1223 {
1224 p = __archive_read_ahead(a, linkname_length, NULL);
1225 }
1226
1227 if (p == NULL) {
1228 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1229 "Truncated Zip file");
1230 return ARCHIVE_FATAL;
1231 }
1232
1233 sconv = zip->sconv;
1234 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME))
1235 sconv = zip->sconv_utf8;
1236 if (sconv == NULL)
1237 sconv = zip->sconv_default;
1238 if (archive_entry_copy_symlink_l(entry, p, linkname_full_length,
1239 sconv) != 0) {
1240 if (errno != ENOMEM && sconv == zip->sconv_utf8 &&
1241 (zip->entry->zip_flags & ZIP_UTF8_NAME))
1242 archive_entry_copy_symlink_l(entry, p,
1243 linkname_full_length, NULL);
1244 if (errno == ENOMEM) {
1245 archive_set_error(&a->archive, ENOMEM,
1246 "Can't allocate memory for Symlink");
1247 return (ARCHIVE_FATAL);
1248 }
1249 /*
1250 * Since there is no character-set regulation for
1251 * symlink name, do not report the conversion error
1252 * in an automatic conversion.
1253 */
1254 if (sconv != zip->sconv_utf8 ||
1255 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) {
1256 archive_set_error(&a->archive,
1257 ARCHIVE_ERRNO_FILE_FORMAT,
1258 "Symlink cannot be converted "
1259 "from %s to current locale.",
1260 archive_string_conversion_charset_name(
1261 sconv));
1262 ret = ARCHIVE_WARN;
1263 }
1264 }
1265 zip_entry->uncompressed_size = zip_entry->compressed_size = 0;
1266
1267 if (__archive_read_consume(a, linkname_length) < 0) {
1268 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1269 "Read error skipping symlink target name");
1270 return ARCHIVE_FATAL;
1271 }
1272 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1273 || (zip_entry->uncompressed_size > 0
1274 && zip_entry->uncompressed_size != 0xffffffff)) {
1275 /* Set the size only if it's meaningful. */
1276 archive_entry_set_size(entry, zip_entry->uncompressed_size);
1277 }
1278 zip->entry_bytes_remaining = zip_entry->compressed_size;
1279
1280 /* If there's no body, force read_data() to return EOF immediately. */
1281 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1282 && zip->entry_bytes_remaining < 1)
1283 zip->end_of_entry = 1;
1284
1285 /* Set up a more descriptive format name. */
1286 archive_string_empty(&zip->format_name);
1287 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)",
1288 version / 10, version % 10,
1289 compression_name(zip->entry->compression));
1290 a->archive.archive_format_name = zip->format_name.s;
1291
1292 return (ret);
1293 }
1294
1295 static int
check_authentication_code(struct archive_read * a,const void * _p)1296 check_authentication_code(struct archive_read *a, const void *_p)
1297 {
1298 struct zip *zip = (struct zip *)(a->format->data);
1299
1300 /* Check authentication code. */
1301 if (zip->hctx_valid) {
1302 const void *p;
1303 uint8_t hmac[20];
1304 size_t hmac_len = 20;
1305 int cmp;
1306
1307 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len);
1308 if (_p == NULL) {
1309 /* Read authentication code. */
1310 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL);
1311 if (p == NULL) {
1312 archive_set_error(&a->archive,
1313 ARCHIVE_ERRNO_FILE_FORMAT,
1314 "Truncated ZIP file data");
1315 return (ARCHIVE_FATAL);
1316 }
1317 } else {
1318 p = _p;
1319 }
1320 cmp = memcmp(hmac, p, AUTH_CODE_SIZE);
1321 __archive_read_consume(a, AUTH_CODE_SIZE);
1322 if (cmp != 0) {
1323 archive_set_error(&a->archive,
1324 ARCHIVE_ERRNO_MISC,
1325 "ZIP bad Authentication code");
1326 return (ARCHIVE_WARN);
1327 }
1328 }
1329 return (ARCHIVE_OK);
1330 }
1331
1332 /*
1333 * The Zip end-of-file marker is inherently ambiguous. The specification
1334 * in APPNOTE.TXT allows any of four possible formats, and there is no
1335 * guaranteed-correct way for a reader to know a priori which one the writer
1336 * will have used. The four formats are:
1337 * 1. 32-bit format with an initial PK78 marker
1338 * 2. 32-bit format without that marker
1339 * 3. 64-bit format with the marker
1340 * 4. 64-bit format without the marker
1341 *
1342 * Mark Adler's `sunzip` streaming unzip program solved this ambiguity
1343 * by just looking at every possible combination and accepting the
1344 * longest one that matches the expected values. His approach always
1345 * consumes the longest possible matching EOF marker, based on an
1346 * analysis of all the possible failures and how the values could
1347 * overlap.
1348 *
1349 * For example, suppose both of the first two formats listed
1350 * above match. In that case, we know the next four
1351 * 32-bit words match this pattern:
1352 * ```
1353 * [PK\07\08] [CRC32] [compressed size] [uncompressed size]
1354 * ```
1355 * but we know they must also match this pattern:
1356 * ```
1357 * [CRC32] [compressed size] [uncompressed size] [other PK marker]
1358 * ```
1359 *
1360 * Since the first word here matches both the PK78 signature in the
1361 * first form and the CRC32 in the second, we know those two values
1362 * are equal, the CRC32 must be exactly 0x08074b50. Similarly, the
1363 * compressed and uncompressed size must also be exactly this value.
1364 * So we know these four words are all 0x08074b50. If we were to
1365 * accept the shorter pattern, it would be immediately followed by
1366 * another PK78 marker, which is not possible in a well-formed ZIP
1367 * archive unless there is garbage between entries. This implies we
1368 * should not accept the shorter form in such a case; we should accept
1369 * the longer form.
1370 *
1371 * If the second and third possibilities above both match, we
1372 * have a slightly different situation. The following words
1373 * must match both the 32-bit format
1374 * ```
1375 * [CRC32] [compressed size] [uncompressed size] [other PK marker]
1376 * ```
1377 * and the 64-bit format
1378 * ```
1379 * [CRC32] [compressed low] [compressed high] [uncompressed low] [uncompressed high] [other PK marker]
1380 * ```
1381 * Since the 32-bit and 64-bit compressed sizes both match, the
1382 * actual size must fit in 32 bits, which implies the high-order
1383 * word of the compressed size is zero. So we know the uncompressed
1384 * low word is zero, which again implies that if we accept the shorter
1385 * format, there will not be a valid PK marker following it.
1386 *
1387 * Similar considerations rule out the shorter form in every other
1388 * possibly-ambiguous pair. So if two of the four possible formats
1389 * match, we should accept the longer option.
1390 *
1391 * If none of the four formats matches, we know the archive must be
1392 * corrupted in some fashion. In particular, it's possible that the
1393 * length-at-end bit was incorrect and we should not really be looking
1394 * for an EOF marker at all. To allow for this possibility, we
1395 * evaluate the following words to collect data for a later error
1396 * report but do not consume any bytes. We instead rely on the later
1397 * search for a new PK marker to re-sync to the next well-formed
1398 * entry.
1399 */
1400 static void
consume_end_of_file_marker(struct archive_read * a,struct zip * zip)1401 consume_end_of_file_marker(struct archive_read *a, struct zip *zip)
1402 {
1403 const char *marker;
1404 const char *p;
1405 uint64_t compressed32, uncompressed32;
1406 uint64_t compressed64, uncompressed64;
1407 uint64_t compressed_actual, uncompressed_actual;
1408 uint32_t crc32_actual;
1409 const uint32_t PK78 = 0x08074B50ULL;
1410 uint8_t crc32_ignored, crc32_may_be_zero;
1411
1412 /* If there shouldn't be a marker, don't consume it. */
1413 if ((zip->entry->zip_flags & ZIP_LENGTH_AT_END) == 0) {
1414 return;
1415 }
1416
1417 /* The longest Zip end-of-file record is 24 bytes. Since an
1418 * end-of-file record can never appear at the end of the
1419 * archive, we know 24 bytes will be available unless
1420 * the archive is severely truncated. */
1421 if (NULL == (marker = __archive_read_ahead(a, 24, NULL))) {
1422 return;
1423 }
1424 p = marker;
1425
1426 /* The end-of-file record comprises:
1427 * = Optional PK\007\010 marker
1428 * = 4-byte CRC32
1429 * = Compressed size
1430 * = Uncompressed size
1431 *
1432 * The last two fields are either both 32 bits or both 64
1433 * bits. We check all possible layouts and accept any one
1434 * that gives us a complete match, else we make a best-effort
1435 * attempt to parse out the pieces.
1436 */
1437
1438 /* CRC32 checking can be tricky:
1439 * * Test suites sometimes ignore the CRC32
1440 * * AES AE-2 always writes zero for the CRC32
1441 * * AES AE-1 sometimes writes zero for the CRC32
1442 */
1443 crc32_ignored = zip->ignore_crc32;
1444 crc32_may_be_zero = 0;
1445 crc32_actual = zip->computed_crc32;
1446 if (zip->hctx_valid) {
1447 switch (zip->entry->aes_extra.vendor) {
1448 case AES_VENDOR_AE_2:
1449 crc32_actual = 0;
1450 break;
1451 case AES_VENDOR_AE_1:
1452 default:
1453 crc32_may_be_zero = 1;
1454 break;
1455 }
1456 }
1457
1458 /* Values computed from the actual data in the archive. */
1459 compressed_actual = (uint64_t)zip->entry_compressed_bytes_read;
1460 uncompressed_actual = (uint64_t)zip->entry_uncompressed_bytes_read;
1461
1462
1463 /* Longest: PK78 marker, all 64-bit fields (24 bytes total) */
1464 if (archive_le32dec(p) == PK78
1465 && ((archive_le32dec(p + 4) == crc32_actual)
1466 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0))
1467 || crc32_ignored)
1468 && (archive_le64dec(p + 8) == compressed_actual)
1469 && (archive_le64dec(p + 16) == uncompressed_actual)) {
1470 if (!crc32_ignored) {
1471 zip->entry->crc32 = crc32_actual;
1472 }
1473 zip->entry->compressed_size = compressed_actual;
1474 zip->entry->uncompressed_size = uncompressed_actual;
1475 zip->unconsumed += 24;
1476 return;
1477 }
1478
1479 /* No PK78 marker, 64-bit fields (20 bytes total) */
1480 if (((archive_le32dec(p) == crc32_actual)
1481 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0))
1482 || crc32_ignored)
1483 && (archive_le64dec(p + 4) == compressed_actual)
1484 && (archive_le64dec(p + 12) == uncompressed_actual)) {
1485 if (!crc32_ignored) {
1486 zip->entry->crc32 = crc32_actual;
1487 }
1488 zip->entry->compressed_size = compressed_actual;
1489 zip->entry->uncompressed_size = uncompressed_actual;
1490 zip->unconsumed += 20;
1491 return;
1492 }
1493
1494 /* PK78 marker and 32-bit fields (16 bytes total) */
1495 if (archive_le32dec(p) == PK78
1496 && ((archive_le32dec(p + 4) == crc32_actual)
1497 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0))
1498 || crc32_ignored)
1499 && (archive_le32dec(p + 8) == compressed_actual)
1500 && (archive_le32dec(p + 12) == uncompressed_actual)) {
1501 if (!crc32_ignored) {
1502 zip->entry->crc32 = crc32_actual;
1503 }
1504 zip->entry->compressed_size = compressed_actual;
1505 zip->entry->uncompressed_size = uncompressed_actual;
1506 zip->unconsumed += 16;
1507 return;
1508 }
1509
1510 /* Shortest: No PK78 marker, all 32-bit fields (12 bytes total) */
1511 if (((archive_le32dec(p) == crc32_actual)
1512 || (crc32_may_be_zero && (archive_le32dec(p + 4) == 0))
1513 || crc32_ignored)
1514 && (archive_le32dec(p + 4) == compressed_actual)
1515 && (archive_le32dec(p + 8) == uncompressed_actual)) {
1516 if (!crc32_ignored) {
1517 zip->entry->crc32 = crc32_actual;
1518 }
1519 zip->entry->compressed_size = compressed_actual;
1520 zip->entry->uncompressed_size = uncompressed_actual;
1521 zip->unconsumed += 12;
1522 return;
1523 }
1524
1525 /* If none of the above patterns gives us a full exact match,
1526 * then there's something definitely amiss. The fallback code
1527 * below will parse out some plausible values for error
1528 * reporting purposes. Note that this won't actually
1529 * consume anything:
1530 *
1531 * = If there really is a marker here, the logic to resync to
1532 * the next entry will suffice to skip it.
1533 *
1534 * = There might not really be a marker: Corruption or bugs
1535 * may have set the length-at-end bit without a marker ever
1536 * having actually been written. In this case, we
1537 * explicitly should not consume any bytes, since that would
1538 * prevent us from correctly reading the next entry.
1539 */
1540 if (archive_le32dec(p) == PK78) {
1541 p += 4; /* Ignore PK78 if it appears to be present */
1542 }
1543 zip->entry->crc32 = archive_le32dec(p); /* Parse CRC32 */
1544 p += 4;
1545
1546 /* Consider both 32- and 64-bit interpretations */
1547 compressed32 = archive_le32dec(p);
1548 uncompressed32 = archive_le32dec(p + 4);
1549 compressed64 = archive_le64dec(p);
1550 uncompressed64 = archive_le64dec(p + 8);
1551
1552 /* The earlier patterns may have failed because of CRC32
1553 * mismatch, so it's still possible that both sizes match.
1554 * Try to match as many as we can...
1555 */
1556 if (compressed32 == compressed_actual
1557 && uncompressed32 == uncompressed_actual) {
1558 /* Both 32-bit fields match */
1559 zip->entry->compressed_size = compressed32;
1560 zip->entry->uncompressed_size = uncompressed32;
1561 } else if (compressed64 == compressed_actual
1562 || uncompressed64 == uncompressed_actual) {
1563 /* One or both 64-bit fields match */
1564 zip->entry->compressed_size = compressed64;
1565 zip->entry->uncompressed_size = uncompressed64;
1566 } else {
1567 /* Zero or one 32-bit fields match */
1568 zip->entry->compressed_size = compressed32;
1569 zip->entry->uncompressed_size = uncompressed32;
1570 }
1571 }
1572
1573 /*
1574 * Read "uncompressed" data.
1575 *
1576 * This is straightforward if we know the size of the data. This is
1577 * always true for the seeking reader (we've examined the Central
1578 * Directory already), and will often be true for the streaming reader
1579 * (the writer was writing uncompressed so probably knows the size).
1580 *
1581 * If we don't know the size, then life is more interesting. Note
1582 * that a careful reading of the Zip specification says that a writer
1583 * must use ZIP_LENGTH_AT_END if it cannot write the CRC into the
1584 * local header. And if it uses ZIP_LENGTH_AT_END, then it is
1585 * prohibited from storing the sizes in the local header. This
1586 * prevents fully-compliant streaming writers from providing any size
1587 * clues to a streaming reader. In this case, we have to scan the
1588 * data as we read to try to locate the end-of-file marker.
1589 *
1590 * We assume here that the end-of-file marker always has the
1591 * PK\007\010 signature. Although it's technically optional, newer
1592 * writers seem to provide it pretty consistently, and it's not clear
1593 * how to efficiently recognize an end-of-file marker that lacks it.
1594 *
1595 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets
1596 * zip->end_of_entry if it consumes all of the data.
1597 */
1598 static int
zip_read_data_none(struct archive_read * a,const void ** _buff,size_t * size,int64_t * offset)1599 zip_read_data_none(struct archive_read *a, const void **_buff,
1600 size_t *size, int64_t *offset)
1601 {
1602 struct zip *zip;
1603 const char *buff;
1604 ssize_t bytes_avail;
1605 ssize_t trailing_extra;
1606 int r;
1607
1608 (void)offset; /* UNUSED */
1609
1610 zip = (struct zip *)(a->format->data);
1611 trailing_extra = zip->hctx_valid ? AUTH_CODE_SIZE : 0;
1612
1613 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) {
1614 const char *p;
1615 ssize_t grabbing_bytes = 24 + trailing_extra;
1616
1617 /* Grab at least 24 bytes. */
1618 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail);
1619 if (bytes_avail < grabbing_bytes) {
1620 /* Zip archives have end-of-archive markers
1621 that are longer than this, so a failure to get at
1622 least 24 bytes really does indicate a truncated
1623 file. */
1624 archive_set_error(&a->archive,
1625 ARCHIVE_ERRNO_FILE_FORMAT,
1626 "Truncated ZIP file data");
1627 return (ARCHIVE_FATAL);
1628 }
1629 /* Check for a complete PK\007\010 signature, followed
1630 * by the correct 4-byte CRC. */
1631 p = buff + trailing_extra;
1632 if (p[0] == 'P' && p[1] == 'K'
1633 && p[2] == '\007' && p[3] == '\010'
1634 && (archive_le32dec(p + 4) == zip->computed_crc32
1635 || zip->ignore_crc32
1636 || (zip->hctx_valid
1637 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) {
1638 zip->end_of_entry = 1;
1639 if (zip->hctx_valid) {
1640 r = check_authentication_code(a, buff);
1641 if (r != ARCHIVE_OK)
1642 return (r);
1643 }
1644 return (ARCHIVE_OK);
1645 }
1646 /* If not at EOF, ensure we consume at least one byte. */
1647 ++p;
1648
1649 /* Scan forward until we see where a PK\007\010 signature
1650 * might be. */
1651 /* Return bytes up until that point. On the next call,
1652 * the code above will verify the data descriptor. */
1653 while (p < buff + bytes_avail - 4) {
1654 if (p[3] == 'P') { p += 3; }
1655 else if (p[3] == 'K') { p += 2; }
1656 else if (p[3] == '\007') { p += 1; }
1657 else if (p[3] == '\010' && p[2] == '\007'
1658 && p[1] == 'K' && p[0] == 'P') {
1659 break;
1660 } else { p += 4; }
1661 }
1662 p -= trailing_extra;
1663 bytes_avail = p - buff;
1664 } else {
1665 if (zip->entry_bytes_remaining == 0) {
1666 zip->end_of_entry = 1;
1667 if (zip->hctx_valid) {
1668 r = check_authentication_code(a, NULL);
1669 if (r != ARCHIVE_OK)
1670 return (r);
1671 }
1672 return (ARCHIVE_OK);
1673 }
1674 /* Grab a bunch of bytes. */
1675 buff = __archive_read_ahead(a, 1, &bytes_avail);
1676 if (bytes_avail <= 0) {
1677 archive_set_error(&a->archive,
1678 ARCHIVE_ERRNO_FILE_FORMAT,
1679 "Truncated ZIP file data");
1680 return (ARCHIVE_FATAL);
1681 }
1682 if (bytes_avail > zip->entry_bytes_remaining)
1683 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
1684 }
1685 if (zip->tctx_valid || zip->cctx_valid) {
1686 size_t dec_size = bytes_avail;
1687
1688 if (dec_size > zip->decrypted_buffer_size)
1689 dec_size = zip->decrypted_buffer_size;
1690 if (zip->tctx_valid) {
1691 trad_enc_decrypt_update(&zip->tctx,
1692 (const uint8_t *)buff, dec_size,
1693 zip->decrypted_buffer, dec_size);
1694 } else {
1695 size_t dsize = dec_size;
1696 archive_hmac_sha1_update(&zip->hctx,
1697 (const uint8_t *)buff, dec_size);
1698 archive_decrypto_aes_ctr_update(&zip->cctx,
1699 (const uint8_t *)buff, dec_size,
1700 zip->decrypted_buffer, &dsize);
1701 }
1702 bytes_avail = dec_size;
1703 buff = (const char *)zip->decrypted_buffer;
1704 }
1705 zip->entry_bytes_remaining -= bytes_avail;
1706 zip->entry_uncompressed_bytes_read += bytes_avail;
1707 zip->entry_compressed_bytes_read += bytes_avail;
1708 zip->unconsumed += bytes_avail;
1709 *size = bytes_avail;
1710 *_buff = buff;
1711 return (ARCHIVE_OK);
1712 }
1713
1714 #if HAVE_LZMA_H && HAVE_LIBLZMA
1715 static int
zipx_xz_init(struct archive_read * a,struct zip * zip)1716 zipx_xz_init(struct archive_read *a, struct zip *zip)
1717 {
1718 lzma_ret r;
1719
1720 if(zip->zipx_lzma_valid) {
1721 lzma_end(&zip->zipx_lzma_stream);
1722 zip->zipx_lzma_valid = 0;
1723 }
1724
1725 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream));
1726 r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0);
1727 if (r != LZMA_OK) {
1728 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
1729 "xz initialization failed(%d)",
1730 r);
1731
1732 return (ARCHIVE_FAILED);
1733 }
1734
1735 zip->zipx_lzma_valid = 1;
1736
1737 free(zip->uncompressed_buffer);
1738
1739 zip->uncompressed_buffer_size = 256 * 1024;
1740 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size);
1741 if (zip->uncompressed_buffer == NULL) {
1742 archive_set_error(&a->archive, ENOMEM,
1743 "No memory for xz decompression");
1744 return (ARCHIVE_FATAL);
1745 }
1746
1747 zip->decompress_init = 1;
1748 return (ARCHIVE_OK);
1749 }
1750
1751 static int
zipx_lzma_alone_init(struct archive_read * a,struct zip * zip)1752 zipx_lzma_alone_init(struct archive_read *a, struct zip *zip)
1753 {
1754 lzma_ret r;
1755 const uint8_t* p;
1756
1757 #pragma pack(push)
1758 #pragma pack(1)
1759 struct _alone_header {
1760 uint8_t bytes[5];
1761 uint64_t uncompressed_size;
1762 } alone_header;
1763 #pragma pack(pop)
1764
1765 if(zip->zipx_lzma_valid) {
1766 lzma_end(&zip->zipx_lzma_stream);
1767 zip->zipx_lzma_valid = 0;
1768 }
1769
1770 /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma
1771 * that is a part of XZ Utils. The stream format stored inside ZIPX
1772 * file is a modified "lzma alone" file format, that was used by the
1773 * `lzma` utility which was later deprecated in favour of `xz` utility.
1774 * Since those formats are nearly the same, we can use a standard
1775 * "lzma alone" decoder from XZ Utils. */
1776
1777 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream));
1778 r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX);
1779 if (r != LZMA_OK) {
1780 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
1781 "lzma initialization failed(%d)", r);
1782
1783 return (ARCHIVE_FAILED);
1784 }
1785
1786 /* Flag the cleanup function that we want our lzma-related structures
1787 * to be freed later. */
1788 zip->zipx_lzma_valid = 1;
1789
1790 /* The "lzma alone" file format and the stream format inside ZIPx are
1791 * almost the same. Here's an example of a structure of "lzma alone"
1792 * format:
1793 *
1794 * $ cat /bin/ls | lzma | xxd | head -n 1
1795 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814
1796 *
1797 * 5 bytes 8 bytes n bytes
1798 * <lzma_params><uncompressed_size><data...>
1799 *
1800 * lzma_params is a 5-byte blob that has to be decoded to extract
1801 * parameters of this LZMA stream. The uncompressed_size field is an
1802 * uint64_t value that contains information about the size of the
1803 * uncompressed file, or UINT64_MAX if this value is unknown.
1804 * The <data...> part is the actual lzma-compressed data stream.
1805 *
1806 * Now here's the structure of the stream inside the ZIPX file:
1807 *
1808 * $ cat stream_inside_zipx | xxd | head -n 1
1809 * 00000000: 0914 0500 5d00 8000 0000 2814 .... ....
1810 *
1811 * 2byte 2byte 5 bytes n bytes
1812 * <magic1><magic2><lzma_params><data...>
1813 *
1814 * This means that the ZIPX file contains an additional magic1 and
1815 * magic2 headers, the lzma_params field contains the same parameter
1816 * set as in the "lzma alone" format, and the <data...> field is the
1817 * same as in the "lzma alone" format as well. Note that also the zipx
1818 * format is missing the uncompressed_size field.
1819 *
1820 * So, in order to use the "lzma alone" decoder for the zipx lzma
1821 * stream, we simply need to shuffle around some fields, prepare a new
1822 * lzma alone header, feed it into lzma alone decoder so it will
1823 * initialize itself properly, and then we can start feeding normal
1824 * zipx lzma stream into the decoder.
1825 */
1826
1827 /* Read magic1,magic2,lzma_params from the ZIPX stream. */
1828 if(zip->entry_bytes_remaining < 9 || (p = __archive_read_ahead(a, 9, NULL)) == NULL) {
1829 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1830 "Truncated lzma data");
1831 return (ARCHIVE_FATAL);
1832 }
1833
1834 if(p[2] != 0x05 || p[3] != 0x00) {
1835 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1836 "Invalid lzma data");
1837 return (ARCHIVE_FATAL);
1838 }
1839
1840 /* Prepare an lzma alone header: copy the lzma_params blob into
1841 * a proper place into the lzma alone header. */
1842 memcpy(&alone_header.bytes[0], p + 4, 5);
1843
1844 /* Initialize the 'uncompressed size' field to unknown; we'll manually
1845 * monitor how many bytes there are still to be uncompressed. */
1846 alone_header.uncompressed_size = UINT64_MAX;
1847
1848 if(!zip->uncompressed_buffer) {
1849 zip->uncompressed_buffer_size = 256 * 1024;
1850 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size);
1851
1852 if (zip->uncompressed_buffer == NULL) {
1853 archive_set_error(&a->archive, ENOMEM,
1854 "No memory for lzma decompression");
1855 return (ARCHIVE_FATAL);
1856 }
1857 }
1858
1859 zip->zipx_lzma_stream.next_in = (void*) &alone_header;
1860 zip->zipx_lzma_stream.avail_in = sizeof(alone_header);
1861 zip->zipx_lzma_stream.total_in = 0;
1862 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1863 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size;
1864 zip->zipx_lzma_stream.total_out = 0;
1865
1866 /* Feed only the header into the lzma alone decoder. This will
1867 * effectively initialize the decoder, and will not produce any
1868 * output bytes yet. */
1869 r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1870 if (r != LZMA_OK) {
1871 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1872 "lzma stream initialization error");
1873 return ARCHIVE_FATAL;
1874 }
1875
1876 /* We've already consumed some bytes, so take this into account. */
1877 __archive_read_consume(a, 9);
1878 zip->entry_bytes_remaining -= 9;
1879 zip->entry_compressed_bytes_read += 9;
1880
1881 zip->decompress_init = 1;
1882 return (ARCHIVE_OK);
1883 }
1884
1885 static int
zip_read_data_zipx_xz(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1886 zip_read_data_zipx_xz(struct archive_read *a, const void **buff,
1887 size_t *size, int64_t *offset)
1888 {
1889 struct zip* zip = (struct zip *)(a->format->data);
1890 int ret;
1891 lzma_ret lz_ret;
1892 const void* compressed_buf;
1893 ssize_t bytes_avail, in_bytes, to_consume = 0;
1894
1895 (void) offset; /* UNUSED */
1896
1897 /* Initialize decompressor if not yet initialized. */
1898 if (!zip->decompress_init) {
1899 ret = zipx_xz_init(a, zip);
1900 if (ret != ARCHIVE_OK)
1901 return (ret);
1902 }
1903
1904 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail);
1905 if (bytes_avail < 0) {
1906 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1907 "Truncated xz file body");
1908 return (ARCHIVE_FATAL);
1909 }
1910
1911 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail);
1912 zip->zipx_lzma_stream.next_in = compressed_buf;
1913 zip->zipx_lzma_stream.avail_in = in_bytes;
1914 zip->zipx_lzma_stream.total_in = 0;
1915 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1916 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size;
1917 zip->zipx_lzma_stream.total_out = 0;
1918
1919 /* Perform the decompression. */
1920 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1921 switch(lz_ret) {
1922 case LZMA_DATA_ERROR:
1923 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1924 "xz data error (error %d)", (int) lz_ret);
1925 return (ARCHIVE_FATAL);
1926
1927 case LZMA_NO_CHECK:
1928 case LZMA_OK:
1929 break;
1930
1931 default:
1932 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1933 "xz unknown error %d", (int) lz_ret);
1934 return (ARCHIVE_FATAL);
1935
1936 case LZMA_STREAM_END:
1937 lzma_end(&zip->zipx_lzma_stream);
1938 zip->zipx_lzma_valid = 0;
1939
1940 if((int64_t) zip->zipx_lzma_stream.total_in !=
1941 zip->entry_bytes_remaining)
1942 {
1943 archive_set_error(&a->archive,
1944 ARCHIVE_ERRNO_MISC,
1945 "xz premature end of stream");
1946 return (ARCHIVE_FATAL);
1947 }
1948
1949 zip->end_of_entry = 1;
1950 break;
1951 }
1952
1953 to_consume = (ssize_t)zip->zipx_lzma_stream.total_in;
1954
1955 __archive_read_consume(a, to_consume);
1956 zip->entry_bytes_remaining -= to_consume;
1957 zip->entry_compressed_bytes_read += to_consume;
1958 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out;
1959
1960 *size = (size_t)zip->zipx_lzma_stream.total_out;
1961 *buff = zip->uncompressed_buffer;
1962
1963 return (ARCHIVE_OK);
1964 }
1965
1966 static int
zip_read_data_zipx_lzma_alone(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1967 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff,
1968 size_t *size, int64_t *offset)
1969 {
1970 struct zip* zip = (struct zip *)(a->format->data);
1971 int ret;
1972 lzma_ret lz_ret;
1973 const void* compressed_buf;
1974 ssize_t bytes_avail, in_bytes, to_consume;
1975
1976 (void) offset; /* UNUSED */
1977
1978 /* Initialize decompressor if not yet initialized. */
1979 if (!zip->decompress_init) {
1980 ret = zipx_lzma_alone_init(a, zip);
1981 if (ret != ARCHIVE_OK)
1982 return (ret);
1983 }
1984
1985 /* Fetch more compressed data. The same note as in deflate handler
1986 * applies here as well:
1987 *
1988 * Note: '1' here is a performance optimization. Recall that the
1989 * decompression layer returns a count of available bytes; asking for
1990 * more than that forces the decompressor to combine reads by copying
1991 * data.
1992 */
1993 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail);
1994 if (bytes_avail < 0) {
1995 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1996 "Truncated lzma file body");
1997 return (ARCHIVE_FATAL);
1998 }
1999
2000 /* Set decompressor parameters. */
2001 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail);
2002
2003 zip->zipx_lzma_stream.next_in = compressed_buf;
2004 zip->zipx_lzma_stream.avail_in = in_bytes;
2005 zip->zipx_lzma_stream.total_in = 0;
2006 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
2007 zip->zipx_lzma_stream.avail_out =
2008 /* These lzma_alone streams lack end of stream marker, so let's
2009 * make sure the unpacker won't try to unpack more than it's
2010 * supposed to. */
2011 (size_t)zipmin((int64_t) zip->uncompressed_buffer_size,
2012 zip->entry->uncompressed_size -
2013 zip->entry_uncompressed_bytes_read);
2014 zip->zipx_lzma_stream.total_out = 0;
2015
2016 /* Perform the decompression. */
2017 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
2018 switch(lz_ret) {
2019 case LZMA_DATA_ERROR:
2020 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2021 "lzma data error (error %d)", (int) lz_ret);
2022 return (ARCHIVE_FATAL);
2023
2024 /* This case is optional in lzma alone format. It can happen,
2025 * but most of the files don't have it. (GitHub #1257) */
2026 case LZMA_STREAM_END:
2027 if((int64_t) zip->zipx_lzma_stream.total_in !=
2028 zip->entry_bytes_remaining)
2029 {
2030 archive_set_error(&a->archive,
2031 ARCHIVE_ERRNO_MISC,
2032 "lzma alone premature end of stream");
2033 return (ARCHIVE_FATAL);
2034 }
2035
2036 zip->end_of_entry = 1;
2037 break;
2038
2039 case LZMA_OK:
2040 break;
2041
2042 default:
2043 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2044 "lzma unknown error %d", (int) lz_ret);
2045 return (ARCHIVE_FATAL);
2046 }
2047
2048 to_consume = (ssize_t)zip->zipx_lzma_stream.total_in;
2049
2050 /* Update pointers. */
2051 __archive_read_consume(a, to_consume);
2052 zip->entry_bytes_remaining -= to_consume;
2053 zip->entry_compressed_bytes_read += to_consume;
2054 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out;
2055
2056 if(zip->entry_bytes_remaining == 0) {
2057 zip->end_of_entry = 1;
2058 }
2059
2060 /* Free lzma decoder handle because we'll no longer need it. */
2061 /* This cannot be folded into LZMA_STREAM_END handling above
2062 * because the stream end marker is not required in this format. */
2063 if(zip->end_of_entry) {
2064 lzma_end(&zip->zipx_lzma_stream);
2065 zip->zipx_lzma_valid = 0;
2066 }
2067
2068 /* Return values. */
2069 *size = (size_t)zip->zipx_lzma_stream.total_out;
2070 *buff = zip->uncompressed_buffer;
2071
2072 /* If we're here, then we're good! */
2073 return (ARCHIVE_OK);
2074 }
2075 #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */
2076
2077 static int
zipx_ppmd8_init(struct archive_read * a,struct zip * zip)2078 zipx_ppmd8_init(struct archive_read *a, struct zip *zip)
2079 {
2080 const void* p;
2081 uint32_t val;
2082 uint32_t order;
2083 uint32_t mem;
2084 uint32_t restore_method;
2085
2086 /* Remove previous decompression context if it exists. */
2087 if(zip->ppmd8_valid) {
2088 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
2089 zip->ppmd8_valid = 0;
2090 }
2091
2092 /* Create a new decompression context. */
2093 __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8);
2094 zip->ppmd8_stream_failed = 0;
2095
2096 /* Setup function pointers required by Ppmd8 decompressor. The
2097 * 'ppmd_read' function will feed new bytes to the decompressor,
2098 * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */
2099 zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream;
2100 zip->zipx_ppmd_stream.a = a;
2101 zip->zipx_ppmd_stream.Read = &ppmd_read;
2102
2103 /* Reset number of read bytes to 0. */
2104 zip->zipx_ppmd_read_compressed = 0;
2105
2106 /* Read Ppmd8 header (2 bytes). */
2107 p = __archive_read_ahead(a, 2, NULL);
2108 if(!p) {
2109 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2110 "Truncated file data in PPMd8 stream");
2111 return (ARCHIVE_FATAL);
2112 }
2113 __archive_read_consume(a, 2);
2114
2115 /* Decode the stream's compression parameters. */
2116 val = archive_le16dec(p);
2117 order = (val & 15) + 1;
2118 mem = ((val >> 4) & 0xff) + 1;
2119 restore_method = (val >> 12);
2120
2121 if(order < 2 || restore_method > 2) {
2122 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2123 "Invalid parameter set in PPMd8 stream (order=%" PRIu32 ", "
2124 "restore=%" PRIu32 ")", order, restore_method);
2125 return (ARCHIVE_FAILED);
2126 }
2127
2128 /* Allocate the memory needed to properly decompress the file. */
2129 if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) {
2130 archive_set_error(&a->archive, ENOMEM,
2131 "Unable to allocate memory for PPMd8 stream: %" PRIu32 " bytes",
2132 mem << 20);
2133 return (ARCHIVE_FATAL);
2134 }
2135
2136 /* Signal the cleanup function to release Ppmd8 context in the
2137 * cleanup phase. */
2138 zip->ppmd8_valid = 1;
2139
2140 /* Perform further Ppmd8 initialization. */
2141 if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) {
2142 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
2143 "PPMd8 stream range decoder initialization error");
2144 return (ARCHIVE_FATAL);
2145 }
2146
2147 __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order,
2148 restore_method);
2149
2150 /* Allocate the buffer that will hold uncompressed data. */
2151 free(zip->uncompressed_buffer);
2152
2153 zip->uncompressed_buffer_size = 256 * 1024;
2154 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size);
2155
2156 if(zip->uncompressed_buffer == NULL) {
2157 archive_set_error(&a->archive, ENOMEM,
2158 "No memory for PPMd8 decompression");
2159 return ARCHIVE_FATAL;
2160 }
2161
2162 /* Ppmd8 initialization is done. */
2163 zip->decompress_init = 1;
2164
2165 /* We've already read 2 bytes in the output stream. Additionally,
2166 * Ppmd8 initialization code could read some data as well. So we
2167 * are advancing the stream by 2 bytes plus whatever number of
2168 * bytes Ppmd8 init function used. */
2169 zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed;
2170
2171 return ARCHIVE_OK;
2172 }
2173
2174 static int
zip_read_data_zipx_ppmd(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2175 zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff,
2176 size_t *size, int64_t *offset)
2177 {
2178 struct zip* zip = (struct zip *)(a->format->data);
2179 int ret;
2180 size_t consumed_bytes = 0;
2181 ssize_t bytes_avail = 0;
2182
2183 (void) offset; /* UNUSED */
2184
2185 /* If we're here for the first time, initialize Ppmd8 decompression
2186 * context first. */
2187 if(!zip->decompress_init) {
2188 ret = zipx_ppmd8_init(a, zip);
2189 if(ret != ARCHIVE_OK)
2190 return ret;
2191 }
2192
2193 /* Fetch for more data. We're reading 1 byte here, but libarchive
2194 * should prefetch more bytes. */
2195 (void) __archive_read_ahead(a, 1, &bytes_avail);
2196 if(bytes_avail < 0) {
2197 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2198 "Truncated PPMd8 file body");
2199 return (ARCHIVE_FATAL);
2200 }
2201
2202 /* This counter will be updated inside ppmd_read(), which at one
2203 * point will be called by Ppmd8_DecodeSymbol. */
2204 zip->zipx_ppmd_read_compressed = 0;
2205
2206 /* Decompression loop. */
2207 do {
2208 int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol(
2209 &zip->ppmd8);
2210 if(sym < 0) {
2211 zip->end_of_entry = 1;
2212 break;
2213 }
2214
2215 /* This field is set by ppmd_read() when there was no more data
2216 * to be read. */
2217 if(zip->ppmd8_stream_failed) {
2218 archive_set_error(&a->archive,
2219 ARCHIVE_ERRNO_FILE_FORMAT,
2220 "Truncated PPMd8 file body");
2221 return (ARCHIVE_FATAL);
2222 }
2223
2224 zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym;
2225 ++consumed_bytes;
2226 } while(consumed_bytes < zip->uncompressed_buffer_size);
2227
2228 /* Update pointers so we can continue decompression in another call. */
2229 zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed;
2230 zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed;
2231 zip->entry_uncompressed_bytes_read += consumed_bytes;
2232
2233 /* If we're at the end of stream, deinitialize Ppmd8 context. */
2234 if(zip->end_of_entry) {
2235 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
2236 zip->ppmd8_valid = 0;
2237 }
2238
2239 /* Update pointers for libarchive. */
2240 *buff = zip->uncompressed_buffer;
2241 *size = consumed_bytes;
2242
2243 return ARCHIVE_OK;
2244 }
2245
2246 #ifdef HAVE_BZLIB_H
2247 static int
zipx_bzip2_init(struct archive_read * a,struct zip * zip)2248 zipx_bzip2_init(struct archive_read *a, struct zip *zip)
2249 {
2250 int r;
2251
2252 /* Deallocate already existing BZ2 decompression context if it
2253 * exists. */
2254 if(zip->bzstream_valid) {
2255 BZ2_bzDecompressEnd(&zip->bzstream);
2256 zip->bzstream_valid = 0;
2257 }
2258
2259 /* Allocate a new BZ2 decompression context. */
2260 memset(&zip->bzstream, 0, sizeof(bz_stream));
2261 r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1);
2262 if(r != BZ_OK) {
2263 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
2264 "bzip2 initialization failed(%d)",
2265 r);
2266
2267 return ARCHIVE_FAILED;
2268 }
2269
2270 /* Mark the bzstream field to be released in cleanup phase. */
2271 zip->bzstream_valid = 1;
2272
2273 /* (Re)allocate the buffer that will contain decompressed bytes. */
2274 free(zip->uncompressed_buffer);
2275
2276 zip->uncompressed_buffer_size = 256 * 1024;
2277 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size);
2278 if (zip->uncompressed_buffer == NULL) {
2279 archive_set_error(&a->archive, ENOMEM,
2280 "No memory for bzip2 decompression");
2281 return ARCHIVE_FATAL;
2282 }
2283
2284 /* Initialization done. */
2285 zip->decompress_init = 1;
2286 return ARCHIVE_OK;
2287 }
2288
2289 static int
zip_read_data_zipx_bzip2(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2290 zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff,
2291 size_t *size, int64_t *offset)
2292 {
2293 struct zip *zip = (struct zip *)(a->format->data);
2294 ssize_t bytes_avail = 0, in_bytes, to_consume;
2295 const void *compressed_buff;
2296 int r;
2297 uint64_t total_out;
2298
2299 (void) offset; /* UNUSED */
2300
2301 /* Initialize decompression context if we're here for the first time. */
2302 if(!zip->decompress_init) {
2303 r = zipx_bzip2_init(a, zip);
2304 if(r != ARCHIVE_OK)
2305 return r;
2306 }
2307
2308 /* Fetch more compressed bytes. */
2309 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail);
2310 if(bytes_avail < 0) {
2311 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2312 "Truncated bzip2 file body");
2313 return (ARCHIVE_FATAL);
2314 }
2315
2316 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail);
2317 if(in_bytes < 1) {
2318 /* libbz2 doesn't complain when caller feeds avail_in == 0.
2319 * It will actually return success in this case, which is
2320 * undesirable. This is why we need to make this check
2321 * manually. */
2322
2323 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2324 "Truncated bzip2 file body");
2325 return (ARCHIVE_FATAL);
2326 }
2327
2328 /* Setup buffer boundaries. */
2329 zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff;
2330 zip->bzstream.avail_in = (uint32_t)in_bytes;
2331 zip->bzstream.total_in_hi32 = 0;
2332 zip->bzstream.total_in_lo32 = 0;
2333 zip->bzstream.next_out = (char*) zip->uncompressed_buffer;
2334 zip->bzstream.avail_out = (uint32_t)zip->uncompressed_buffer_size;
2335 zip->bzstream.total_out_hi32 = 0;
2336 zip->bzstream.total_out_lo32 = 0;
2337
2338 /* Perform the decompression. */
2339 r = BZ2_bzDecompress(&zip->bzstream);
2340 switch(r) {
2341 case BZ_STREAM_END:
2342 /* If we're at the end of the stream, deinitialize the
2343 * decompression context now. */
2344 switch(BZ2_bzDecompressEnd(&zip->bzstream)) {
2345 case BZ_OK:
2346 break;
2347 default:
2348 archive_set_error(&a->archive,
2349 ARCHIVE_ERRNO_MISC,
2350 "Failed to clean up bzip2 "
2351 "decompressor");
2352 return ARCHIVE_FATAL;
2353 }
2354
2355 zip->end_of_entry = 1;
2356 break;
2357 case BZ_OK:
2358 /* The decompressor has successfully decoded this
2359 * chunk of data, but more data is still in queue. */
2360 break;
2361 default:
2362 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2363 "bzip2 decompression failed");
2364 return ARCHIVE_FATAL;
2365 }
2366
2367 /* Update the pointers so decompressor can continue decoding. */
2368 to_consume = zip->bzstream.total_in_lo32;
2369 __archive_read_consume(a, to_consume);
2370
2371 total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) |
2372 zip->bzstream.total_out_lo32;
2373
2374 zip->entry_bytes_remaining -= to_consume;
2375 zip->entry_compressed_bytes_read += to_consume;
2376 zip->entry_uncompressed_bytes_read += total_out;
2377
2378 /* Give libarchive its due. */
2379 *size = (size_t)total_out;
2380 *buff = zip->uncompressed_buffer;
2381
2382 return ARCHIVE_OK;
2383 }
2384
2385 #endif
2386
2387 #if HAVE_ZSTD_H && HAVE_LIBZSTD
2388 static int
zipx_zstd_init(struct archive_read * a,struct zip * zip)2389 zipx_zstd_init(struct archive_read *a, struct zip *zip)
2390 {
2391 size_t r;
2392
2393 /* Deallocate already existing Zstd decompression context if it
2394 * exists. */
2395 if(zip->zstdstream_valid) {
2396 ZSTD_freeDStream(zip->zstdstream);
2397 zip->zstdstream_valid = 0;
2398 }
2399
2400 /* Allocate a new Zstd decompression context. */
2401 zip->zstdstream = ZSTD_createDStream();
2402
2403 r = ZSTD_initDStream(zip->zstdstream);
2404 if (ZSTD_isError(r)) {
2405 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2406 "Error initializing zstd decompressor: %s",
2407 ZSTD_getErrorName(r));
2408
2409 return ARCHIVE_FAILED;
2410 }
2411
2412 /* Mark the zstdstream field to be released in cleanup phase. */
2413 zip->zstdstream_valid = 1;
2414
2415 /* (Re)allocate the buffer that will contain decompressed bytes. */
2416 free(zip->uncompressed_buffer);
2417
2418 zip->uncompressed_buffer_size = ZSTD_DStreamOutSize();
2419 zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size);
2420 if (zip->uncompressed_buffer == NULL) {
2421 archive_set_error(&a->archive, ENOMEM,
2422 "No memory for Zstd decompression");
2423
2424 return ARCHIVE_FATAL;
2425 }
2426
2427 /* Initialization done. */
2428 zip->decompress_init = 1;
2429 return ARCHIVE_OK;
2430 }
2431
2432 static int
zip_read_data_zipx_zstd(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2433 zip_read_data_zipx_zstd(struct archive_read *a, const void **buff,
2434 size_t *size, int64_t *offset)
2435 {
2436 struct zip *zip = (struct zip *)(a->format->data);
2437 ssize_t bytes_avail = 0, in_bytes, to_consume;
2438 const void *compressed_buff;
2439 int r;
2440 size_t ret;
2441 uint64_t total_out;
2442 ZSTD_outBuffer out;
2443 ZSTD_inBuffer in;
2444
2445 (void) offset; /* UNUSED */
2446
2447 /* Initialize decompression context if we're here for the first time. */
2448 if(!zip->decompress_init) {
2449 r = zipx_zstd_init(a, zip);
2450 if(r != ARCHIVE_OK)
2451 return r;
2452 }
2453
2454 /* Fetch more compressed bytes */
2455 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail);
2456 if(bytes_avail < 0) {
2457 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2458 "Truncated zstd file body");
2459 return (ARCHIVE_FATAL);
2460 }
2461
2462 in_bytes = (ssize_t)zipmin(zip->entry_bytes_remaining, bytes_avail);
2463 if(in_bytes < 1) {
2464 /* zstd doesn't complain when caller feeds avail_in == 0.
2465 * It will actually return success in this case, which is
2466 * undesirable. This is why we need to make this check
2467 * manually. */
2468 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2469 "Truncated zstd file body");
2470 return (ARCHIVE_FATAL);
2471 }
2472
2473 /* Setup buffer boundaries */
2474 in.src = compressed_buff;
2475 in.size = in_bytes;
2476 in.pos = 0;
2477 out = (ZSTD_outBuffer) { zip->uncompressed_buffer, zip->uncompressed_buffer_size, 0 };
2478
2479 /* Perform the decompression. */
2480 ret = ZSTD_decompressStream(zip->zstdstream, &out, &in);
2481 if (ZSTD_isError(ret)) {
2482 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2483 "Error during zstd decompression: %s",
2484 ZSTD_getErrorName(ret));
2485 return (ARCHIVE_FATAL);
2486 }
2487
2488 /* Check end of the stream. */
2489 if (ret == 0) {
2490 if ((in.pos == in.size) && (out.pos < out.size)) {
2491 zip->end_of_entry = 1;
2492 ZSTD_freeDStream(zip->zstdstream);
2493 zip->zstdstream_valid = 0;
2494 }
2495 }
2496
2497 /* Update the pointers so decompressor can continue decoding. */
2498 to_consume = in.pos;
2499 __archive_read_consume(a, to_consume);
2500
2501 total_out = out.pos;
2502
2503 zip->entry_bytes_remaining -= to_consume;
2504 zip->entry_compressed_bytes_read += to_consume;
2505 zip->entry_uncompressed_bytes_read += total_out;
2506
2507 /* Give libarchive its due. */
2508 *size = (size_t)total_out;
2509 *buff = zip->uncompressed_buffer;
2510
2511 return ARCHIVE_OK;
2512 }
2513 #endif
2514
2515 #ifdef HAVE_ZLIB_H
2516 static int
zip_deflate_init(struct archive_read * a,struct zip * zip)2517 zip_deflate_init(struct archive_read *a, struct zip *zip)
2518 {
2519 int r;
2520
2521 /* If we haven't yet read any data, initialize the decompressor. */
2522 if (!zip->decompress_init) {
2523 if (zip->stream_valid)
2524 r = inflateReset(&zip->stream);
2525 else
2526 r = inflateInit2(&zip->stream,
2527 -15 /* Don't check for zlib header */);
2528 if (r != Z_OK) {
2529 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2530 "Can't initialize ZIP decompression.");
2531 return (ARCHIVE_FATAL);
2532 }
2533 /* Stream structure has been set up. */
2534 zip->stream_valid = 1;
2535 /* We've initialized decompression for this stream. */
2536 zip->decompress_init = 1;
2537 }
2538 return (ARCHIVE_OK);
2539 }
2540
2541 static int
zip_read_data_deflate(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2542 zip_read_data_deflate(struct archive_read *a, const void **buff,
2543 size_t *size, int64_t *offset)
2544 {
2545 struct zip *zip;
2546 ssize_t bytes_avail, to_consume = 0;
2547 const void *compressed_buff, *sp;
2548 int r;
2549
2550 (void)offset; /* UNUSED */
2551
2552 zip = (struct zip *)(a->format->data);
2553
2554 /* If the buffer hasn't been allocated, allocate it now. */
2555 if (zip->uncompressed_buffer == NULL) {
2556 zip->uncompressed_buffer_size = 256 * 1024;
2557 zip->uncompressed_buffer
2558 = malloc(zip->uncompressed_buffer_size);
2559 if (zip->uncompressed_buffer == NULL) {
2560 archive_set_error(&a->archive, ENOMEM,
2561 "No memory for ZIP decompression");
2562 return (ARCHIVE_FATAL);
2563 }
2564 }
2565
2566 r = zip_deflate_init(a, zip);
2567 if (r != ARCHIVE_OK)
2568 return (r);
2569
2570 /*
2571 * Note: '1' here is a performance optimization.
2572 * Recall that the decompression layer returns a count of
2573 * available bytes; asking for more than that forces the
2574 * decompressor to combine reads by copying data.
2575 */
2576 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail);
2577 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2578 && bytes_avail > zip->entry_bytes_remaining) {
2579 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
2580 }
2581 if (bytes_avail < 0) {
2582 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2583 "Truncated ZIP file body");
2584 return (ARCHIVE_FATAL);
2585 }
2586
2587 if (zip->tctx_valid || zip->cctx_valid) {
2588 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) {
2589 size_t buff_remaining =
2590 (zip->decrypted_buffer +
2591 zip->decrypted_buffer_size)
2592 - (zip->decrypted_ptr +
2593 zip->decrypted_bytes_remaining);
2594
2595 if (buff_remaining > (size_t)bytes_avail)
2596 buff_remaining = (size_t)bytes_avail;
2597
2598 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) &&
2599 zip->entry_bytes_remaining > 0) {
2600 if ((int64_t)(zip->decrypted_bytes_remaining
2601 + buff_remaining)
2602 > zip->entry_bytes_remaining) {
2603 if (zip->entry_bytes_remaining <
2604 (int64_t)zip->decrypted_bytes_remaining)
2605 buff_remaining = 0;
2606 else
2607 buff_remaining =
2608 (size_t)zip->entry_bytes_remaining
2609 - zip->decrypted_bytes_remaining;
2610 }
2611 }
2612 if (buff_remaining > 0) {
2613 if (zip->tctx_valid) {
2614 trad_enc_decrypt_update(&zip->tctx,
2615 compressed_buff, buff_remaining,
2616 zip->decrypted_ptr
2617 + zip->decrypted_bytes_remaining,
2618 buff_remaining);
2619 } else {
2620 size_t dsize = buff_remaining;
2621 archive_decrypto_aes_ctr_update(
2622 &zip->cctx,
2623 compressed_buff, buff_remaining,
2624 zip->decrypted_ptr
2625 + zip->decrypted_bytes_remaining,
2626 &dsize);
2627 }
2628 zip->decrypted_bytes_remaining +=
2629 buff_remaining;
2630 }
2631 }
2632 bytes_avail = zip->decrypted_bytes_remaining;
2633 compressed_buff = (const char *)zip->decrypted_ptr;
2634 }
2635
2636 /*
2637 * A bug in zlib.h: stream.next_in should be marked 'const'
2638 * but isn't (the library never alters data through the
2639 * next_in pointer, only reads it). The result: this ugly
2640 * cast to remove 'const'.
2641 */
2642 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff;
2643 zip->stream.avail_in = (uInt)bytes_avail;
2644 zip->stream.total_in = 0;
2645 zip->stream.next_out = zip->uncompressed_buffer;
2646 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size;
2647 zip->stream.total_out = 0;
2648
2649 r = inflate(&zip->stream, 0);
2650 switch (r) {
2651 case Z_OK:
2652 break;
2653 case Z_STREAM_END:
2654 zip->end_of_entry = 1;
2655 break;
2656 case Z_MEM_ERROR:
2657 archive_set_error(&a->archive, ENOMEM,
2658 "Out of memory for ZIP decompression");
2659 return (ARCHIVE_FATAL);
2660 default:
2661 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2662 "ZIP decompression failed (%d)", r);
2663 return (ARCHIVE_FATAL);
2664 }
2665
2666 /* Consume as much as the compressor actually used. */
2667 to_consume = zip->stream.total_in;
2668 __archive_read_consume(a, to_consume);
2669 zip->entry_bytes_remaining -= to_consume;
2670 zip->entry_compressed_bytes_read += to_consume;
2671 zip->entry_uncompressed_bytes_read += zip->stream.total_out;
2672
2673 if (zip->tctx_valid || zip->cctx_valid) {
2674 zip->decrypted_bytes_remaining -= to_consume;
2675 if (zip->decrypted_bytes_remaining == 0)
2676 zip->decrypted_ptr = zip->decrypted_buffer;
2677 else
2678 zip->decrypted_ptr += to_consume;
2679 }
2680 if (zip->hctx_valid)
2681 archive_hmac_sha1_update(&zip->hctx, sp, to_consume);
2682
2683 if (zip->end_of_entry) {
2684 if (zip->hctx_valid) {
2685 r = check_authentication_code(a, NULL);
2686 if (r != ARCHIVE_OK) {
2687 return (r);
2688 }
2689 }
2690 }
2691
2692 *size = zip->stream.total_out;
2693 *buff = zip->uncompressed_buffer;
2694
2695 return (ARCHIVE_OK);
2696 }
2697 #endif
2698
2699 static int
read_decryption_header(struct archive_read * a)2700 read_decryption_header(struct archive_read *a)
2701 {
2702 struct zip *zip = (struct zip *)(a->format->data);
2703 const char *p;
2704 unsigned int remaining_size;
2705 unsigned int ts;
2706
2707 /*
2708 * Read an initialization vector data field.
2709 */
2710 p = __archive_read_ahead(a, 2, NULL);
2711 if (p == NULL)
2712 goto truncated;
2713 ts = zip->iv_size;
2714 zip->iv_size = archive_le16dec(p);
2715 __archive_read_consume(a, 2);
2716 if (ts < zip->iv_size) {
2717 free(zip->iv);
2718 zip->iv = NULL;
2719 }
2720 p = __archive_read_ahead(a, zip->iv_size, NULL);
2721 if (p == NULL)
2722 goto truncated;
2723 if (zip->iv == NULL) {
2724 zip->iv = malloc(zip->iv_size);
2725 if (zip->iv == NULL)
2726 goto nomem;
2727 }
2728 memcpy(zip->iv, p, zip->iv_size);
2729 __archive_read_consume(a, zip->iv_size);
2730
2731 /*
2732 * Read a size of remaining decryption header field.
2733 */
2734 p = __archive_read_ahead(a, 14, NULL);
2735 if (p == NULL)
2736 goto truncated;
2737 remaining_size = archive_le32dec(p);
2738 if (remaining_size < 16 || remaining_size > (1 << 18))
2739 goto corrupted;
2740
2741 /* Check if format version is supported. */
2742 if (archive_le16dec(p+4) != 3) {
2743 archive_set_error(&a->archive,
2744 ARCHIVE_ERRNO_FILE_FORMAT,
2745 "Unsupported encryption format version: %u",
2746 archive_le16dec(p+4));
2747 return (ARCHIVE_FAILED);
2748 }
2749
2750 /*
2751 * Read an encryption algorithm field.
2752 */
2753 zip->alg_id = archive_le16dec(p+6);
2754 switch (zip->alg_id) {
2755 case 0x6601:/* DES */
2756 case 0x6602:/* RC2 */
2757 case 0x6603:/* 3DES 168 */
2758 case 0x6609:/* 3DES 112 */
2759 case 0x660E:/* AES 128 */
2760 case 0x660F:/* AES 192 */
2761 case 0x6610:/* AES 256 */
2762 case 0x6702:/* RC2 (version >= 5.2) */
2763 case 0x6720:/* Blowfish */
2764 case 0x6721:/* Twofish */
2765 case 0x6801:/* RC4 */
2766 /* Supported encryption algorithm. */
2767 break;
2768 default:
2769 archive_set_error(&a->archive,
2770 ARCHIVE_ERRNO_FILE_FORMAT,
2771 "Unknown encryption algorithm: %u", zip->alg_id);
2772 return (ARCHIVE_FAILED);
2773 }
2774
2775 /*
2776 * Read a bit length field.
2777 */
2778 zip->bit_len = archive_le16dec(p+8);
2779
2780 /*
2781 * Read a flags field.
2782 */
2783 zip->flags = archive_le16dec(p+10);
2784 switch (zip->flags & 0xf000) {
2785 case 0x0001: /* Password is required to decrypt. */
2786 case 0x0002: /* Certificates only. */
2787 case 0x0003: /* Password or certificate required to decrypt. */
2788 break;
2789 default:
2790 archive_set_error(&a->archive,
2791 ARCHIVE_ERRNO_FILE_FORMAT,
2792 "Unknown encryption flag: %u", zip->flags);
2793 return (ARCHIVE_FAILED);
2794 }
2795 if ((zip->flags & 0xf000) == 0 ||
2796 (zip->flags & 0xf000) == 0x4000) {
2797 archive_set_error(&a->archive,
2798 ARCHIVE_ERRNO_FILE_FORMAT,
2799 "Unknown encryption flag: %u", zip->flags);
2800 return (ARCHIVE_FAILED);
2801 }
2802
2803 /*
2804 * Read an encrypted random data field.
2805 */
2806 ts = zip->erd_size;
2807 zip->erd_size = archive_le16dec(p+12);
2808 __archive_read_consume(a, 14);
2809 if ((zip->erd_size & 0xf) != 0 ||
2810 (zip->erd_size + 16) > remaining_size ||
2811 (zip->erd_size + 16) < zip->erd_size)
2812 goto corrupted;
2813
2814 if (ts < zip->erd_size) {
2815 free(zip->erd);
2816 zip->erd = NULL;
2817 }
2818 p = __archive_read_ahead(a, zip->erd_size, NULL);
2819 if (p == NULL)
2820 goto truncated;
2821 if (zip->erd == NULL) {
2822 zip->erd = malloc(zip->erd_size);
2823 if (zip->erd == NULL)
2824 goto nomem;
2825 }
2826 memcpy(zip->erd, p, zip->erd_size);
2827 __archive_read_consume(a, zip->erd_size);
2828
2829 /*
2830 * Read a reserved data field.
2831 */
2832 p = __archive_read_ahead(a, 4, NULL);
2833 if (p == NULL)
2834 goto truncated;
2835 /* Reserved data size should be zero. */
2836 if (archive_le32dec(p) != 0)
2837 goto corrupted;
2838 __archive_read_consume(a, 4);
2839
2840 /*
2841 * Read a password validation data field.
2842 */
2843 p = __archive_read_ahead(a, 2, NULL);
2844 if (p == NULL)
2845 goto truncated;
2846 ts = zip->v_size;
2847 zip->v_size = archive_le16dec(p);
2848 __archive_read_consume(a, 2);
2849 if ((zip->v_size & 0x0f) != 0 ||
2850 (zip->erd_size + zip->v_size + 16) > remaining_size ||
2851 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size))
2852 goto corrupted;
2853 if (ts < zip->v_size) {
2854 free(zip->v_data);
2855 zip->v_data = NULL;
2856 }
2857 p = __archive_read_ahead(a, zip->v_size, NULL);
2858 if (p == NULL)
2859 goto truncated;
2860 if (zip->v_data == NULL) {
2861 zip->v_data = malloc(zip->v_size);
2862 if (zip->v_data == NULL)
2863 goto nomem;
2864 }
2865 memcpy(zip->v_data, p, zip->v_size);
2866 __archive_read_consume(a, zip->v_size);
2867
2868 p = __archive_read_ahead(a, 4, NULL);
2869 if (p == NULL)
2870 goto truncated;
2871 zip->v_crc32 = archive_le32dec(p);
2872 __archive_read_consume(a, 4);
2873
2874 /*return (ARCHIVE_OK);
2875 * This is not fully implemented yet.*/
2876 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2877 "Encrypted file is unsupported");
2878 return (ARCHIVE_FAILED);
2879 truncated:
2880 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2881 "Truncated ZIP file data");
2882 return (ARCHIVE_FATAL);
2883 corrupted:
2884 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2885 "Corrupted ZIP file data");
2886 return (ARCHIVE_FATAL);
2887 nomem:
2888 archive_set_error(&a->archive, ENOMEM,
2889 "No memory for ZIP decryption");
2890 return (ARCHIVE_FATAL);
2891 }
2892
2893 static int
zip_alloc_decryption_buffer(struct archive_read * a)2894 zip_alloc_decryption_buffer(struct archive_read *a)
2895 {
2896 struct zip *zip = (struct zip *)(a->format->data);
2897 size_t bs = 256 * 1024;
2898
2899 if (zip->decrypted_buffer == NULL) {
2900 zip->decrypted_buffer_size = bs;
2901 zip->decrypted_buffer = malloc(bs);
2902 if (zip->decrypted_buffer == NULL) {
2903 archive_set_error(&a->archive, ENOMEM,
2904 "No memory for ZIP decryption");
2905 return (ARCHIVE_FATAL);
2906 }
2907 }
2908 zip->decrypted_ptr = zip->decrypted_buffer;
2909 return (ARCHIVE_OK);
2910 }
2911
2912 static int
init_traditional_PKWARE_decryption(struct archive_read * a)2913 init_traditional_PKWARE_decryption(struct archive_read *a)
2914 {
2915 struct zip *zip = (struct zip *)(a->format->data);
2916 const void *p;
2917 int retry;
2918 int r;
2919
2920 if (zip->tctx_valid)
2921 return (ARCHIVE_OK);
2922
2923 /*
2924 Read the 12 bytes encryption header stored at
2925 the start of the data area.
2926 */
2927 #define ENC_HEADER_SIZE 12
2928 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2929 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) {
2930 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2931 "Truncated Zip encrypted body: only %jd bytes available",
2932 (intmax_t)zip->entry_bytes_remaining);
2933 return (ARCHIVE_FATAL);
2934 }
2935
2936 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL);
2937 if (p == NULL) {
2938 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2939 "Truncated ZIP file data");
2940 return (ARCHIVE_FATAL);
2941 }
2942
2943 for (retry = 0;; retry++) {
2944 const char *passphrase;
2945 uint8_t crcchk;
2946
2947 passphrase = __archive_read_next_passphrase(a);
2948 if (passphrase == NULL) {
2949 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2950 (retry > 0)?
2951 "Incorrect passphrase":
2952 "Passphrase required for this entry");
2953 return (ARCHIVE_FAILED);
2954 }
2955
2956 /*
2957 * Initialize ctx for Traditional PKWARE Decryption.
2958 */
2959 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase),
2960 p, ENC_HEADER_SIZE, &crcchk);
2961 if (r == 0 && crcchk == zip->entry->decdat)
2962 break;/* The passphrase is OK. */
2963 if (retry > 10000) {
2964 /* Avoid infinity loop. */
2965 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2966 "Too many incorrect passphrases");
2967 return (ARCHIVE_FAILED);
2968 }
2969 }
2970
2971 __archive_read_consume(a, ENC_HEADER_SIZE);
2972 zip->tctx_valid = 1;
2973 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
2974 zip->entry_bytes_remaining -= ENC_HEADER_SIZE;
2975 }
2976 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/
2977 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE;
2978 zip->decrypted_bytes_remaining = 0;
2979
2980 return (zip_alloc_decryption_buffer(a));
2981 #undef ENC_HEADER_SIZE
2982 }
2983
2984 static int
init_WinZip_AES_decryption(struct archive_read * a)2985 init_WinZip_AES_decryption(struct archive_read *a)
2986 {
2987 struct zip *zip = (struct zip *)(a->format->data);
2988 const void *p;
2989 const uint8_t *pv;
2990 size_t key_len, salt_len;
2991 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE];
2992 int retry;
2993 int r;
2994
2995 if (zip->cctx_valid || zip->hctx_valid)
2996 return (ARCHIVE_OK);
2997
2998 switch (zip->entry->aes_extra.strength) {
2999 case 1: salt_len = 8; key_len = 16; break;
3000 case 2: salt_len = 12; key_len = 24; break;
3001 case 3: salt_len = 16; key_len = 32; break;
3002 default: goto corrupted;
3003 }
3004 p = __archive_read_ahead(a, salt_len + 2, NULL);
3005 if (p == NULL)
3006 goto truncated;
3007
3008 for (retry = 0;; retry++) {
3009 const char *passphrase;
3010
3011 passphrase = __archive_read_next_passphrase(a);
3012 if (passphrase == NULL) {
3013 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3014 (retry > 0)?
3015 "Incorrect passphrase":
3016 "Passphrase required for this entry");
3017 return (ARCHIVE_FAILED);
3018 }
3019 memset(derived_key, 0, sizeof(derived_key));
3020 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase),
3021 p, salt_len, 1000, derived_key, key_len * 2 + 2);
3022 if (r != 0) {
3023 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3024 r == CRYPTOR_STUB_FUNCTION ? "Decryption is unsupported due "
3025 "to lack of crypto library" : "Failed to process passphrase");
3026 return (ARCHIVE_FAILED);
3027 }
3028
3029 /* Check password verification value. */
3030 pv = ((const uint8_t *)p) + salt_len;
3031 if (derived_key[key_len * 2] == pv[0] &&
3032 derived_key[key_len * 2 + 1] == pv[1])
3033 break;/* The passphrase is OK. */
3034 if (retry > 10000) {
3035 /* Avoid infinity loop. */
3036 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3037 "Too many incorrect passphrases");
3038 return (ARCHIVE_FAILED);
3039 }
3040 }
3041
3042 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len);
3043 if (r != 0) {
3044 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3045 "Decryption is unsupported due to lack of crypto library");
3046 return (ARCHIVE_FAILED);
3047 }
3048 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len);
3049 if (r != 0) {
3050 archive_decrypto_aes_ctr_release(&zip->cctx);
3051 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3052 "Failed to initialize HMAC-SHA1");
3053 return (ARCHIVE_FAILED);
3054 }
3055 zip->cctx_valid = zip->hctx_valid = 1;
3056 __archive_read_consume(a, salt_len + 2);
3057 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE;
3058 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
3059 && zip->entry_bytes_remaining < 0)
3060 goto corrupted;
3061 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE;
3062 zip->decrypted_bytes_remaining = 0;
3063
3064 zip->entry->compression = zip->entry->aes_extra.compression;
3065 return (zip_alloc_decryption_buffer(a));
3066
3067 truncated:
3068 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3069 "Truncated ZIP file data");
3070 return (ARCHIVE_FATAL);
3071 corrupted:
3072 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3073 "Corrupted ZIP file data");
3074 return (ARCHIVE_FATAL);
3075 }
3076
3077 static int
archive_read_format_zip_read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)3078 archive_read_format_zip_read_data(struct archive_read *a,
3079 const void **buff, size_t *size, int64_t *offset)
3080 {
3081 int r;
3082 struct zip *zip = (struct zip *)(a->format->data);
3083
3084 if (zip->has_encrypted_entries ==
3085 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) {
3086 zip->has_encrypted_entries = 0;
3087 }
3088
3089 *offset = zip->entry_uncompressed_bytes_read;
3090 *size = 0;
3091 *buff = NULL;
3092
3093 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */
3094 if (zip->end_of_entry)
3095 return (ARCHIVE_EOF);
3096
3097 /* Return EOF immediately if this is a non-regular file. */
3098 if (AE_IFREG != (zip->entry->mode & AE_IFMT))
3099 return (ARCHIVE_EOF);
3100
3101 __archive_read_consume(a, zip->unconsumed);
3102 zip->unconsumed = 0;
3103
3104 if (zip->init_decryption) {
3105 zip->has_encrypted_entries = 1;
3106 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
3107 r = read_decryption_header(a);
3108 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
3109 r = init_WinZip_AES_decryption(a);
3110 else
3111 r = init_traditional_PKWARE_decryption(a);
3112 if (r != ARCHIVE_OK)
3113 return (r);
3114 zip->init_decryption = 0;
3115 }
3116
3117 switch(zip->entry->compression) {
3118 case 0: /* No compression. */
3119 r = zip_read_data_none(a, buff, size, offset);
3120 break;
3121 #ifdef HAVE_BZLIB_H
3122 case 12: /* ZIPx bzip2 compression. */
3123 r = zip_read_data_zipx_bzip2(a, buff, size, offset);
3124 break;
3125 #endif
3126 #if HAVE_LZMA_H && HAVE_LIBLZMA
3127 case 14: /* ZIPx LZMA compression. */
3128 r = zip_read_data_zipx_lzma_alone(a, buff, size, offset);
3129 break;
3130 case 95: /* ZIPx XZ compression. */
3131 r = zip_read_data_zipx_xz(a, buff, size, offset);
3132 break;
3133 #endif
3134 #if HAVE_ZSTD_H && HAVE_LIBZSTD
3135 case 93: /* ZIPx Zstd compression. */
3136 r = zip_read_data_zipx_zstd(a, buff, size, offset);
3137 break;
3138 #endif
3139 /* PPMd support is built-in, so we don't need any #if guards. */
3140 case 98: /* ZIPx PPMd compression. */
3141 r = zip_read_data_zipx_ppmd(a, buff, size, offset);
3142 break;
3143
3144 #ifdef HAVE_ZLIB_H
3145 case 8: /* Deflate compression. */
3146 r = zip_read_data_deflate(a, buff, size, offset);
3147 break;
3148 #endif
3149 default: /* Unsupported compression. */
3150 /* Return a warning. */
3151 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3152 "Unsupported ZIP compression method (%d: %s)",
3153 zip->entry->compression, compression_name(zip->entry->compression));
3154 /* We can't decompress this entry, but we will
3155 * be able to skip() it and try the next entry. */
3156 return (ARCHIVE_FAILED);
3157 }
3158 if (r != ARCHIVE_OK)
3159 return (r);
3160 if (*size > 0) {
3161 zip->computed_crc32 = zip->crc32func(zip->computed_crc32, *buff,
3162 (unsigned)*size);
3163 }
3164 /* If we hit the end, swallow any end-of-data marker and
3165 * verify the final check values. */
3166 if (zip->end_of_entry) {
3167 consume_end_of_file_marker(a, zip);
3168
3169 /* Check computed CRC against header */
3170 if ((!zip->hctx_valid ||
3171 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) &&
3172 zip->entry->crc32 != zip->computed_crc32
3173 && !zip->ignore_crc32) {
3174 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3175 "ZIP bad CRC: 0x%lx should be 0x%lx",
3176 (unsigned long)zip->computed_crc32,
3177 (unsigned long)zip->entry->crc32);
3178 return (ARCHIVE_FAILED);
3179 }
3180 /* Check file size against header. */
3181 if (zip->entry->compressed_size !=
3182 zip->entry_compressed_bytes_read) {
3183 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3184 "ZIP compressed data is wrong size "
3185 "(read %jd, expected %jd)",
3186 (intmax_t)zip->entry_compressed_bytes_read,
3187 (intmax_t)zip->entry->compressed_size);
3188 return (ARCHIVE_FAILED);
3189 }
3190 /* Size field only stores the lower 32 bits of the actual
3191 * size. */
3192 if ((zip->entry->uncompressed_size & UINT32_MAX)
3193 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) {
3194 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3195 "ZIP uncompressed data is wrong size "
3196 "(read %jd, expected %jd)\n",
3197 (intmax_t)zip->entry_uncompressed_bytes_read,
3198 (intmax_t)zip->entry->uncompressed_size);
3199 return (ARCHIVE_FAILED);
3200 }
3201 }
3202
3203 return (ARCHIVE_OK);
3204 }
3205
3206 static int
archive_read_format_zip_cleanup(struct archive_read * a)3207 archive_read_format_zip_cleanup(struct archive_read *a)
3208 {
3209 struct zip *zip;
3210 struct zip_entry *zip_entry, *next_zip_entry;
3211
3212 zip = (struct zip *)(a->format->data);
3213
3214 #ifdef HAVE_ZLIB_H
3215 if (zip->stream_valid)
3216 inflateEnd(&zip->stream);
3217 #endif
3218
3219 #if HAVE_LZMA_H && HAVE_LIBLZMA
3220 if (zip->zipx_lzma_valid) {
3221 lzma_end(&zip->zipx_lzma_stream);
3222 }
3223 #endif
3224
3225 #ifdef HAVE_BZLIB_H
3226 if (zip->bzstream_valid) {
3227 BZ2_bzDecompressEnd(&zip->bzstream);
3228 }
3229 #endif
3230
3231 #if HAVE_ZSTD_H && HAVE_LIBZSTD
3232 if (zip->zstdstream_valid) {
3233 ZSTD_freeDStream(zip->zstdstream);
3234 }
3235 #endif
3236
3237 free(zip->uncompressed_buffer);
3238
3239 if (zip->ppmd8_valid)
3240 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
3241
3242 if (zip->zip_entries) {
3243 zip_entry = zip->zip_entries;
3244 while (zip_entry != NULL) {
3245 next_zip_entry = zip_entry->next;
3246 archive_string_free(&zip_entry->rsrcname);
3247 free(zip_entry);
3248 zip_entry = next_zip_entry;
3249 }
3250 }
3251 free(zip->decrypted_buffer);
3252 if (zip->cctx_valid)
3253 archive_decrypto_aes_ctr_release(&zip->cctx);
3254 if (zip->hctx_valid)
3255 archive_hmac_sha1_cleanup(&zip->hctx);
3256 free(zip->iv);
3257 free(zip->erd);
3258 free(zip->v_data);
3259 archive_string_free(&zip->format_name);
3260 free(zip);
3261 (a->format->data) = NULL;
3262 return (ARCHIVE_OK);
3263 }
3264
3265 static int
archive_read_format_zip_has_encrypted_entries(struct archive_read * _a)3266 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a)
3267 {
3268 if (_a && _a->format) {
3269 struct zip * zip = (struct zip *)_a->format->data;
3270 if (zip) {
3271 return zip->has_encrypted_entries;
3272 }
3273 }
3274 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
3275 }
3276
3277 static int
archive_read_format_zip_options(struct archive_read * a,const char * key,const char * val)3278 archive_read_format_zip_options(struct archive_read *a,
3279 const char *key, const char *val)
3280 {
3281 struct zip *zip;
3282 int ret = ARCHIVE_FAILED;
3283
3284 zip = (struct zip *)(a->format->data);
3285 if (strcmp(key, "compat-2x") == 0) {
3286 /* Handle filenames as libarchive 2.x */
3287 zip->init_default_conversion = (val != NULL) ? 1 : 0;
3288 return (ARCHIVE_OK);
3289 } else if (strcmp(key, "hdrcharset") == 0) {
3290 if (val == NULL || val[0] == 0)
3291 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3292 "zip: hdrcharset option needs a character-set name"
3293 );
3294 else {
3295 zip->sconv = archive_string_conversion_from_charset(
3296 &a->archive, val, 0);
3297 if (zip->sconv != NULL) {
3298 if (strcmp(val, "UTF-8") == 0)
3299 zip->sconv_utf8 = zip->sconv;
3300 ret = ARCHIVE_OK;
3301 } else
3302 ret = ARCHIVE_FATAL;
3303 }
3304 return (ret);
3305 } else if (strcmp(key, "ignorecrc32") == 0) {
3306 /* Mostly useful for testing. */
3307 if (val == NULL || val[0] == 0) {
3308 zip->crc32func = real_crc32;
3309 zip->ignore_crc32 = 0;
3310 } else {
3311 zip->crc32func = fake_crc32;
3312 zip->ignore_crc32 = 1;
3313 }
3314 return (ARCHIVE_OK);
3315 } else if (strcmp(key, "mac-ext") == 0) {
3316 zip->process_mac_extensions = (val != NULL && val[0] != 0);
3317 return (ARCHIVE_OK);
3318 }
3319
3320 /* Note: The "warn" return is just to inform the options
3321 * supervisor that we didn't handle it. It will generate
3322 * a suitable error if no one used this option. */
3323 return (ARCHIVE_WARN);
3324 }
3325
3326 int
archive_read_support_format_zip(struct archive * a)3327 archive_read_support_format_zip(struct archive *a)
3328 {
3329 int r;
3330 r = archive_read_support_format_zip_streamable(a);
3331 if (r != ARCHIVE_OK)
3332 return r;
3333 return (archive_read_support_format_zip_seekable(a));
3334 }
3335
3336 /* ------------------------------------------------------------------------ */
3337
3338 /*
3339 * Streaming-mode support
3340 */
3341
3342
3343 static int
archive_read_support_format_zip_capabilities_streamable(struct archive_read * a)3344 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a)
3345 {
3346 (void)a; /* UNUSED */
3347 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
3348 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
3349 }
3350
3351 static int
archive_read_format_zip_streamable_bid(struct archive_read * a,int best_bid)3352 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid)
3353 {
3354 const char *p;
3355
3356 (void)best_bid; /* UNUSED */
3357
3358 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
3359 return (-1);
3360
3361 /*
3362 * Bid of 29 here comes from:
3363 * + 16 bits for "PK",
3364 * + next 16-bit field has 6 options so contributes
3365 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits
3366 *
3367 * So we've effectively verified ~29 total bits of check data.
3368 */
3369 if (p[0] == 'P' && p[1] == 'K') {
3370 if ((p[2] == '\001' && p[3] == '\002')
3371 || (p[2] == '\003' && p[3] == '\004')
3372 || (p[2] == '\005' && p[3] == '\006')
3373 || (p[2] == '\006' && p[3] == '\006')
3374 || (p[2] == '\007' && p[3] == '\010')
3375 || (p[2] == '0' && p[3] == '0'))
3376 return (29);
3377 }
3378
3379 /* TODO: It's worth looking ahead a little bit for a valid
3380 * PK signature. In particular, that would make it possible
3381 * to read some UUEncoded SFX files or SFX files coming from
3382 * a network socket. */
3383
3384 return (0);
3385 }
3386
3387 static int
archive_read_format_zip_streamable_read_header(struct archive_read * a,struct archive_entry * entry)3388 archive_read_format_zip_streamable_read_header(struct archive_read *a,
3389 struct archive_entry *entry)
3390 {
3391 struct zip *zip;
3392
3393 a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
3394 if (a->archive.archive_format_name == NULL)
3395 a->archive.archive_format_name = "ZIP";
3396
3397 zip = (struct zip *)(a->format->data);
3398
3399 /*
3400 * It should be sufficient to call archive_read_next_header() for
3401 * a reader to determine if an entry is encrypted or not. If the
3402 * encryption of an entry is only detectable when calling
3403 * archive_read_data(), so be it. We'll do the same check there
3404 * as well.
3405 */
3406 if (zip->has_encrypted_entries ==
3407 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
3408 zip->has_encrypted_entries = 0;
3409
3410 /* Make sure we have a zip_entry structure to use. */
3411 if (zip->zip_entries == NULL) {
3412 zip->zip_entries = malloc(sizeof(struct zip_entry));
3413 if (zip->zip_entries == NULL) {
3414 archive_set_error(&a->archive, ENOMEM,
3415 "Out of memory");
3416 return ARCHIVE_FATAL;
3417 }
3418 }
3419 zip->entry = zip->zip_entries;
3420 memset(zip->entry, 0, sizeof(struct zip_entry));
3421
3422 if (zip->cctx_valid)
3423 archive_decrypto_aes_ctr_release(&zip->cctx);
3424 if (zip->hctx_valid)
3425 archive_hmac_sha1_cleanup(&zip->hctx);
3426 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
3427 __archive_read_reset_passphrase(a);
3428
3429 /* Search ahead for the next local file header. */
3430 __archive_read_consume(a, zip->unconsumed);
3431 zip->unconsumed = 0;
3432 for (;;) {
3433 int64_t skipped = 0;
3434 const char *p, *end;
3435 ssize_t bytes;
3436
3437 p = __archive_read_ahead(a, 4, &bytes);
3438 if (p == NULL)
3439 return (ARCHIVE_FATAL);
3440 end = p + bytes;
3441
3442 while (p + 4 <= end) {
3443 if (p[0] == 'P' && p[1] == 'K') {
3444 if (p[2] == '\003' && p[3] == '\004') {
3445 /* Regular file entry. */
3446 __archive_read_consume(a, skipped);
3447 return zip_read_local_file_header(a,
3448 entry, zip);
3449 }
3450
3451 /*
3452 * TODO: We cannot restore permissions
3453 * based only on the local file headers.
3454 * Consider scanning the central
3455 * directory and returning additional
3456 * entries for at least directories.
3457 * This would allow us to properly set
3458 * directory permissions.
3459 *
3460 * This won't help us fix symlinks
3461 * and may not help with regular file
3462 * permissions, either. <sigh>
3463 */
3464 if (p[2] == '\001' && p[3] == '\002') {
3465 return (ARCHIVE_EOF);
3466 }
3467
3468 /* End of central directory? Must be an
3469 * empty archive. */
3470 if ((p[2] == '\005' && p[3] == '\006')
3471 || (p[2] == '\006' && p[3] == '\006'))
3472 return (ARCHIVE_EOF);
3473 }
3474 ++p;
3475 ++skipped;
3476 }
3477 __archive_read_consume(a, skipped);
3478 }
3479 }
3480
3481 static int
archive_read_format_zip_read_data_skip_streamable(struct archive_read * a)3482 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a)
3483 {
3484 struct zip *zip;
3485 int64_t bytes_skipped;
3486
3487 zip = (struct zip *)(a->format->data);
3488 bytes_skipped = __archive_read_consume(a, zip->unconsumed);
3489 zip->unconsumed = 0;
3490 if (bytes_skipped < 0)
3491 return (ARCHIVE_FATAL);
3492
3493 /* If we've already read to end of data, we're done. */
3494 if (zip->end_of_entry)
3495 return (ARCHIVE_OK);
3496
3497 /* So we know we're streaming... */
3498 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
3499 || zip->entry->compressed_size > 0) {
3500 /* We know the compressed length, so we can just skip. */
3501 bytes_skipped = __archive_read_consume(a,
3502 zip->entry_bytes_remaining);
3503 if (bytes_skipped < 0)
3504 return (ARCHIVE_FATAL);
3505 return (ARCHIVE_OK);
3506 }
3507
3508 if (zip->init_decryption) {
3509 int r;
3510
3511 zip->has_encrypted_entries = 1;
3512 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
3513 r = read_decryption_header(a);
3514 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
3515 r = init_WinZip_AES_decryption(a);
3516 else
3517 r = init_traditional_PKWARE_decryption(a);
3518 if (r != ARCHIVE_OK)
3519 return (r);
3520 zip->init_decryption = 0;
3521 }
3522
3523 /* We're streaming and we don't know the length. */
3524 /* If the body is compressed and we know the format, we can
3525 * find an exact end-of-entry by decompressing it. */
3526 switch (zip->entry->compression) {
3527 #ifdef HAVE_ZLIB_H
3528 case 8: /* Deflate compression. */
3529 while (!zip->end_of_entry) {
3530 int64_t offset = 0;
3531 const void *buff = NULL;
3532 size_t size = 0;
3533 int r;
3534 r = zip_read_data_deflate(a, &buff, &size, &offset);
3535 if (r != ARCHIVE_OK)
3536 return (r);
3537 }
3538 return ARCHIVE_OK;
3539 #endif
3540 default: /* Uncompressed or unknown. */
3541 /* Scan for a PK\007\010 signature. */
3542 for (;;) {
3543 const char *p, *buff;
3544 ssize_t bytes_avail;
3545 buff = __archive_read_ahead(a, 16, &bytes_avail);
3546 if (bytes_avail < 16) {
3547 archive_set_error(&a->archive,
3548 ARCHIVE_ERRNO_FILE_FORMAT,
3549 "Truncated ZIP file data");
3550 return (ARCHIVE_FATAL);
3551 }
3552 p = buff;
3553 while (p <= buff + bytes_avail - 16) {
3554 if (p[3] == 'P') { p += 3; }
3555 else if (p[3] == 'K') { p += 2; }
3556 else if (p[3] == '\007') { p += 1; }
3557 else if (p[3] == '\010' && p[2] == '\007'
3558 && p[1] == 'K' && p[0] == 'P') {
3559 if (zip->entry->flags & LA_USED_ZIP64)
3560 __archive_read_consume(a,
3561 p - buff + 24);
3562 else
3563 __archive_read_consume(a,
3564 p - buff + 16);
3565 return ARCHIVE_OK;
3566 } else { p += 4; }
3567 }
3568 __archive_read_consume(a, p - buff);
3569 }
3570 }
3571 }
3572
3573 int
archive_read_support_format_zip_streamable(struct archive * _a)3574 archive_read_support_format_zip_streamable(struct archive *_a)
3575 {
3576 struct archive_read *a = (struct archive_read *)_a;
3577 struct zip *zip;
3578 int r;
3579
3580 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
3581 ARCHIVE_STATE_NEW, "archive_read_support_format_zip");
3582
3583 zip = calloc(1, sizeof(*zip));
3584 if (zip == NULL) {
3585 archive_set_error(&a->archive, ENOMEM,
3586 "Can't allocate zip data");
3587 return (ARCHIVE_FATAL);
3588 }
3589
3590 /* Streamable reader doesn't support mac extensions. */
3591 zip->process_mac_extensions = 0;
3592
3593 /*
3594 * Until enough data has been read, we cannot tell about
3595 * any encrypted entries yet.
3596 */
3597 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
3598 zip->crc32func = real_crc32;
3599
3600 r = __archive_read_register_format(a,
3601 zip,
3602 "zip",
3603 archive_read_format_zip_streamable_bid,
3604 archive_read_format_zip_options,
3605 archive_read_format_zip_streamable_read_header,
3606 archive_read_format_zip_read_data,
3607 archive_read_format_zip_read_data_skip_streamable,
3608 NULL,
3609 archive_read_format_zip_cleanup,
3610 archive_read_support_format_zip_capabilities_streamable,
3611 archive_read_format_zip_has_encrypted_entries);
3612
3613 if (r != ARCHIVE_OK)
3614 free(zip);
3615 return (ARCHIVE_OK);
3616 }
3617
3618 /* ------------------------------------------------------------------------ */
3619
3620 /*
3621 * Seeking-mode support
3622 */
3623
3624 static int
archive_read_support_format_zip_capabilities_seekable(struct archive_read * a)3625 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a)
3626 {
3627 (void)a; /* UNUSED */
3628 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
3629 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
3630 }
3631
3632 /*
3633 * TODO: This is a performance sink because it forces the read core to
3634 * drop buffered data from the start of file, which will then have to
3635 * be re-read again if this bidder loses.
3636 *
3637 * We workaround this a little by passing in the best bid so far so
3638 * that later bidders can do nothing if they know they'll never
3639 * outbid. But we can certainly do better...
3640 */
3641 static int
read_eocd(struct zip * zip,const char * p,int64_t current_offset)3642 read_eocd(struct zip *zip, const char *p, int64_t current_offset)
3643 {
3644 uint16_t disk_num;
3645 uint32_t cd_size, cd_offset;
3646
3647 disk_num = archive_le16dec(p + 4);
3648 cd_size = archive_le32dec(p + 12);
3649 cd_offset = archive_le32dec(p + 16);
3650
3651 /* Sanity-check the EOCD we've found. */
3652
3653 /* This must be the first volume. */
3654 if (disk_num != 0)
3655 return 0;
3656 /* Central directory must be on this volume. */
3657 if (disk_num != archive_le16dec(p + 6))
3658 return 0;
3659 /* All central directory entries must be on this volume. */
3660 if (archive_le16dec(p + 10) != archive_le16dec(p + 8))
3661 return 0;
3662 /* Central directory can't extend beyond start of EOCD record. */
3663 if ((int64_t)cd_offset + cd_size > current_offset)
3664 return 0;
3665
3666 /* Save the central directory location for later use. */
3667 zip->central_directory_offset = cd_offset;
3668 zip->central_directory_offset_adjusted = current_offset - cd_size;
3669
3670 /* This is just a tiny bit higher than the maximum
3671 returned by the streaming Zip bidder. This ensures
3672 that the more accurate seeking Zip parser wins
3673 whenever seek is available. */
3674 return 32;
3675 }
3676
3677 /*
3678 * Examine Zip64 EOCD locator: If it's valid, store the information
3679 * from it.
3680 */
3681 static int
read_zip64_eocd(struct archive_read * a,struct zip * zip,const char * p)3682 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p)
3683 {
3684 int64_t eocd64_offset;
3685 int64_t eocd64_size;
3686
3687 /* Sanity-check the locator record. */
3688
3689 /* Central dir must be on first volume. */
3690 if (archive_le32dec(p + 4) != 0)
3691 return 0;
3692 /* Must be only a single volume. */
3693 if (archive_le32dec(p + 16) != 1)
3694 return 0;
3695
3696 /* Find the Zip64 EOCD record. */
3697 eocd64_offset = archive_le64dec(p + 8);
3698 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0)
3699 return 0;
3700 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL)
3701 return 0;
3702 /* Make sure we can read all of it. */
3703 eocd64_size = archive_le64dec(p + 4) + 12;
3704 if (eocd64_size < 56 || eocd64_size > 16384)
3705 return 0;
3706 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL)
3707 return 0;
3708
3709 /* Sanity-check the EOCD64 */
3710 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */
3711 return 0;
3712 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */
3713 return 0;
3714 /* CD can't be split. */
3715 if (archive_le64dec(p + 24) != archive_le64dec(p + 32))
3716 return 0;
3717
3718 /* Save the central directory offset for later use. */
3719 zip->central_directory_offset = archive_le64dec(p + 48);
3720 /* TODO: Needs scanning backwards to find the eocd64 instead of assuming */
3721 zip->central_directory_offset_adjusted = zip->central_directory_offset;
3722
3723 return 32;
3724 }
3725
3726 static int
archive_read_format_zip_seekable_bid(struct archive_read * a,int best_bid)3727 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid)
3728 {
3729 struct zip *zip = (struct zip *)a->format->data;
3730 int64_t file_size, current_offset;
3731 const char *p;
3732 int i, tail;
3733
3734 /* If someone has already bid more than 32, then avoid
3735 trashing the look-ahead buffers with a seek. */
3736 if (best_bid > 32)
3737 return (-1);
3738
3739 file_size = __archive_read_seek(a, 0, SEEK_END);
3740 if (file_size <= 0)
3741 return 0;
3742
3743 /* Search last 16k of file for end-of-central-directory
3744 * record (which starts with PK\005\006) */
3745 tail = (int)zipmin(1024 * 16, file_size);
3746 current_offset = __archive_read_seek(a, -tail, SEEK_END);
3747 if (current_offset < 0)
3748 return 0;
3749 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL)
3750 return 0;
3751 /* Boyer-Moore search backwards from the end, since we want
3752 * to match the last EOCD in the file (there can be more than
3753 * one if there is an uncompressed Zip archive as a member
3754 * within this Zip archive). */
3755 for (i = tail - 22; i > 0;) {
3756 switch (p[i]) {
3757 case 'P':
3758 if (memcmp(p + i, "PK\005\006", 4) == 0) {
3759 int ret = read_eocd(zip, p + i,
3760 current_offset + i);
3761 /* Zip64 EOCD locator precedes
3762 * regular EOCD if present. */
3763 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) {
3764 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20);
3765 if (ret_zip64 > ret)
3766 ret = ret_zip64;
3767 }
3768 return (ret);
3769 }
3770 i -= 4;
3771 break;
3772 case 'K': i -= 1; break;
3773 case 005: i -= 2; break;
3774 case 006: i -= 3; break;
3775 default: i -= 4; break;
3776 }
3777 }
3778 return 0;
3779 }
3780
3781 /* The red-black trees are only used in seeking mode to manage
3782 * the in-memory copy of the central directory. */
3783
3784 static int
cmp_node(const struct archive_rb_node * n1,const struct archive_rb_node * n2)3785 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2)
3786 {
3787 const struct zip_entry *e1 = (const struct zip_entry *)n1;
3788 const struct zip_entry *e2 = (const struct zip_entry *)n2;
3789
3790 if (e1->local_header_offset > e2->local_header_offset)
3791 return -1;
3792 if (e1->local_header_offset < e2->local_header_offset)
3793 return 1;
3794 return 0;
3795 }
3796
3797 static int
cmp_key(const struct archive_rb_node * n,const void * key)3798 cmp_key(const struct archive_rb_node *n, const void *key)
3799 {
3800 /* This function won't be called */
3801 (void)n; /* UNUSED */
3802 (void)key; /* UNUSED */
3803 return 1;
3804 }
3805
3806 static const struct archive_rb_tree_ops rb_ops = {
3807 &cmp_node, &cmp_key
3808 };
3809
3810 static int
rsrc_cmp_node(const struct archive_rb_node * n1,const struct archive_rb_node * n2)3811 rsrc_cmp_node(const struct archive_rb_node *n1,
3812 const struct archive_rb_node *n2)
3813 {
3814 const struct zip_entry *e1 = (const struct zip_entry *)n1;
3815 const struct zip_entry *e2 = (const struct zip_entry *)n2;
3816
3817 return (strcmp(e2->rsrcname.s, e1->rsrcname.s));
3818 }
3819
3820 static int
rsrc_cmp_key(const struct archive_rb_node * n,const void * key)3821 rsrc_cmp_key(const struct archive_rb_node *n, const void *key)
3822 {
3823 const struct zip_entry *e = (const struct zip_entry *)n;
3824 return (strcmp((const char *)key, e->rsrcname.s));
3825 }
3826
3827 static const struct archive_rb_tree_ops rb_rsrc_ops = {
3828 &rsrc_cmp_node, &rsrc_cmp_key
3829 };
3830
3831 static const char *
rsrc_basename(const char * name,size_t name_length)3832 rsrc_basename(const char *name, size_t name_length)
3833 {
3834 const char *s, *r;
3835
3836 r = s = name;
3837 for (;;) {
3838 s = memchr(s, '/', name_length - (s - name));
3839 if (s == NULL)
3840 break;
3841 r = ++s;
3842 }
3843 return (r);
3844 }
3845
3846 static void
expose_parent_dirs(struct zip * zip,const char * name,size_t name_length)3847 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length)
3848 {
3849 struct archive_string str;
3850 struct zip_entry *dir;
3851 char *s;
3852
3853 archive_string_init(&str);
3854 archive_strncpy(&str, name, name_length);
3855 for (;;) {
3856 s = strrchr(str.s, '/');
3857 if (s == NULL)
3858 break;
3859 *s = '\0';
3860 /* Transfer the parent directory from zip->tree_rsrc RB
3861 * tree to zip->tree RB tree to expose. */
3862 dir = (struct zip_entry *)
3863 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s);
3864 if (dir == NULL)
3865 break;
3866 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node);
3867 archive_string_free(&dir->rsrcname);
3868 __archive_rb_tree_insert_node(&zip->tree, &dir->node);
3869 }
3870 archive_string_free(&str);
3871 }
3872
3873 static int
slurp_central_directory(struct archive_read * a,struct archive_entry * entry,struct zip * zip)3874 slurp_central_directory(struct archive_read *a, struct archive_entry* entry,
3875 struct zip *zip)
3876 {
3877 ssize_t i;
3878 unsigned found;
3879 int64_t correction;
3880 ssize_t bytes_avail;
3881 const char *p;
3882
3883 /*
3884 * Find the start of the central directory. The end-of-CD
3885 * record has our starting point, but there are lots of
3886 * Zip archives which have had other data prepended to the
3887 * file, which makes the recorded offsets all too small.
3888 * So we search forward from the specified offset until we
3889 * find the real start of the central directory. Then we
3890 * know the correction we need to apply to account for leading
3891 * padding.
3892 */
3893 if (__archive_read_seek(a, zip->central_directory_offset_adjusted, SEEK_SET)
3894 < 0)
3895 return ARCHIVE_FATAL;
3896
3897 found = 0;
3898 while (!found) {
3899 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL)
3900 return ARCHIVE_FATAL;
3901 for (found = 0, i = 0; !found && i < bytes_avail - 4;) {
3902 switch (p[i + 3]) {
3903 case 'P': i += 3; break;
3904 case 'K': i += 2; break;
3905 case 001: i += 1; break;
3906 case 002:
3907 if (memcmp(p + i, "PK\001\002", 4) == 0) {
3908 p += i;
3909 found = 1;
3910 } else
3911 i += 4;
3912 break;
3913 case 005: i += 1; break;
3914 case 006:
3915 if (memcmp(p + i, "PK\005\006", 4) == 0) {
3916 p += i;
3917 found = 1;
3918 } else if (memcmp(p + i, "PK\006\006", 4) == 0) {
3919 p += i;
3920 found = 1;
3921 } else
3922 i += 1;
3923 break;
3924 default: i += 4; break;
3925 }
3926 }
3927 __archive_read_consume(a, i);
3928 }
3929 correction = archive_filter_bytes(&a->archive, 0)
3930 - zip->central_directory_offset;
3931
3932 __archive_rb_tree_init(&zip->tree, &rb_ops);
3933 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops);
3934
3935 zip->central_directory_entries_total = 0;
3936 while (1) {
3937 struct zip_entry *zip_entry;
3938 size_t filename_length, extra_length, comment_length;
3939 uint32_t external_attributes;
3940 const char *name, *r;
3941
3942 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
3943 return ARCHIVE_FATAL;
3944 if (memcmp(p, "PK\006\006", 4) == 0
3945 || memcmp(p, "PK\005\006", 4) == 0) {
3946 break;
3947 } else if (memcmp(p, "PK\001\002", 4) != 0) {
3948 archive_set_error(&a->archive,
3949 -1, "Invalid central directory signature");
3950 return ARCHIVE_FATAL;
3951 }
3952 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL)
3953 return ARCHIVE_FATAL;
3954
3955 zip_entry = calloc(1, sizeof(struct zip_entry));
3956 if (zip_entry == NULL) {
3957 archive_set_error(&a->archive, ENOMEM,
3958 "Can't allocate zip entry");
3959 return ARCHIVE_FATAL;
3960 }
3961 zip_entry->next = zip->zip_entries;
3962 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY;
3963 zip->zip_entries = zip_entry;
3964 zip->central_directory_entries_total++;
3965
3966 /* version = p[4]; */
3967 zip_entry->system = p[5];
3968 /* version_required = archive_le16dec(p + 6); */
3969 zip_entry->zip_flags = archive_le16dec(p + 8);
3970 if (zip_entry->zip_flags
3971 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){
3972 zip->has_encrypted_entries = 1;
3973 }
3974 zip_entry->compression = (char)archive_le16dec(p + 10);
3975 zip_entry->mtime = dos_to_unix(archive_le32dec(p + 12));
3976 zip_entry->crc32 = archive_le32dec(p + 16);
3977 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
3978 zip_entry->decdat = p[13];
3979 else
3980 zip_entry->decdat = p[19];
3981 zip_entry->compressed_size = archive_le32dec(p + 20);
3982 zip_entry->uncompressed_size = archive_le32dec(p + 24);
3983 filename_length = archive_le16dec(p + 28);
3984 extra_length = archive_le16dec(p + 30);
3985 comment_length = archive_le16dec(p + 32);
3986 /* disk_start = archive_le16dec(p + 34);
3987 * Better be zero.
3988 * internal_attributes = archive_le16dec(p + 36);
3989 * text bit */
3990 external_attributes = archive_le32dec(p + 38);
3991 zip_entry->local_header_offset =
3992 archive_le32dec(p + 42) + correction;
3993
3994 /* If we can't guess the mode, leave it zero here;
3995 when we read the local file header we might get
3996 more information. */
3997 if (zip_entry->system == 3) {
3998 zip_entry->mode = external_attributes >> 16;
3999 } else if (zip_entry->system == 0) {
4000 // Interpret MSDOS directory bit
4001 if (0x10 == (external_attributes & 0x10)) {
4002 zip_entry->mode = AE_IFDIR | 0775;
4003 } else {
4004 zip_entry->mode = AE_IFREG | 0664;
4005 }
4006 if (0x01 == (external_attributes & 0x01)) {
4007 // Read-only bit; strip write permissions
4008 zip_entry->mode &= 0555;
4009 }
4010 } else {
4011 zip_entry->mode = 0;
4012 }
4013
4014 /* We're done with the regular data; get the filename and
4015 * extra data. */
4016 __archive_read_consume(a, 46);
4017 p = __archive_read_ahead(a, filename_length + extra_length,
4018 NULL);
4019 if (p == NULL) {
4020 archive_set_error(&a->archive,
4021 ARCHIVE_ERRNO_FILE_FORMAT,
4022 "Truncated ZIP file header");
4023 return ARCHIVE_FATAL;
4024 }
4025 if (ARCHIVE_OK != process_extra(a, entry, p + filename_length,
4026 extra_length, zip_entry)) {
4027 return ARCHIVE_FATAL;
4028 }
4029
4030 /*
4031 * Mac resource fork files are stored under the
4032 * "__MACOSX/" directory, so we should check if
4033 * it is.
4034 */
4035 if (!zip->process_mac_extensions) {
4036 /* Treat every entry as a regular entry. */
4037 __archive_rb_tree_insert_node(&zip->tree,
4038 &zip_entry->node);
4039 } else {
4040 name = p;
4041 r = rsrc_basename(name, filename_length);
4042 if (filename_length >= 9 &&
4043 strncmp("__MACOSX/", name, 9) == 0) {
4044 /* If this file is not a resource fork nor
4045 * a directory. We should treat it as a non
4046 * resource fork file to expose it. */
4047 if (name[filename_length-1] != '/' &&
4048 (r - name < 3 || r[0] != '.' ||
4049 r[1] != '_')) {
4050 __archive_rb_tree_insert_node(
4051 &zip->tree, &zip_entry->node);
4052 /* Expose its parent directories. */
4053 expose_parent_dirs(zip, name,
4054 filename_length);
4055 } else {
4056 /* This file is a resource fork file or
4057 * a directory. */
4058 archive_strncpy(&(zip_entry->rsrcname),
4059 name, filename_length);
4060 __archive_rb_tree_insert_node(
4061 &zip->tree_rsrc, &zip_entry->node);
4062 }
4063 } else {
4064 /* Generate resource fork name to find its
4065 * resource file at zip->tree_rsrc. */
4066
4067 /* If this is an entry ending with slash,
4068 * make the resource for name slash-less
4069 * as the actual resource fork doesn't end with '/'.
4070 */
4071 size_t tmp_length = filename_length;
4072 if (tmp_length > 0 && name[tmp_length - 1] == '/') {
4073 tmp_length--;
4074 r = rsrc_basename(name, tmp_length);
4075 }
4076
4077 archive_strcpy(&(zip_entry->rsrcname),
4078 "__MACOSX/");
4079 archive_strncat(&(zip_entry->rsrcname),
4080 name, r - name);
4081 archive_strcat(&(zip_entry->rsrcname), "._");
4082 archive_strncat(&(zip_entry->rsrcname),
4083 name + (r - name),
4084 tmp_length - (r - name));
4085 /* Register an entry to RB tree to sort it by
4086 * file offset. */
4087 __archive_rb_tree_insert_node(&zip->tree,
4088 &zip_entry->node);
4089 }
4090 }
4091
4092 /* Skip the comment too ... */
4093 __archive_read_consume(a,
4094 filename_length + extra_length + comment_length);
4095 }
4096
4097 return ARCHIVE_OK;
4098 }
4099
4100 static ssize_t
zip_get_local_file_header_size(struct archive_read * a,size_t extra)4101 zip_get_local_file_header_size(struct archive_read *a, size_t extra)
4102 {
4103 const char *p;
4104 ssize_t filename_length, extra_length;
4105
4106 if ((p = __archive_read_ahead(a, extra + ZIP_LOCHDR_LEN, NULL)) == NULL) {
4107 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
4108 "Truncated ZIP file header");
4109 return (ARCHIVE_WARN);
4110 }
4111 p += extra;
4112
4113 if (memcmp(p, "PK\003\004", 4) != 0) {
4114 archive_set_error(&a->archive, -1, "Damaged Zip archive");
4115 return ARCHIVE_WARN;
4116 }
4117 filename_length = archive_le16dec(p + 26);
4118 extra_length = archive_le16dec(p + 28);
4119
4120 return (ZIP_LOCHDR_LEN + filename_length + extra_length);
4121 }
4122
4123 static int
zip_read_mac_metadata(struct archive_read * a,struct archive_entry * entry,struct zip_entry * rsrc)4124 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry,
4125 struct zip_entry *rsrc)
4126 {
4127 struct zip *zip = (struct zip *)a->format->data;
4128 unsigned char *metadata, *mp;
4129 int64_t offset = archive_filter_bytes(&a->archive, 0);
4130 size_t remaining_bytes, metadata_bytes;
4131 ssize_t hsize;
4132 int ret = ARCHIVE_OK, eof;
4133
4134 switch(rsrc->compression) {
4135 case 0: /* No compression. */
4136 if (rsrc->uncompressed_size != rsrc->compressed_size) {
4137 archive_set_error(&a->archive,
4138 ARCHIVE_ERRNO_FILE_FORMAT,
4139 "Malformed OS X metadata entry: "
4140 "inconsistent size");
4141 return (ARCHIVE_FATAL);
4142 }
4143 #ifdef HAVE_ZLIB_H
4144 case 8: /* Deflate compression. */
4145 #endif
4146 break;
4147 default: /* Unsupported compression. */
4148 /* Return a warning. */
4149 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
4150 "Unsupported ZIP compression method (%s)",
4151 compression_name(rsrc->compression));
4152 /* We can't decompress this entry, but we will
4153 * be able to skip() it and try the next entry. */
4154 return (ARCHIVE_WARN);
4155 }
4156
4157 if (rsrc->uncompressed_size > ZIP_MAX_METADATA * 1048576U) {
4158 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
4159 "Mac metadata is too large: %jd > %u MiB",
4160 (intmax_t)rsrc->uncompressed_size, ZIP_MAX_METADATA);
4161 return (ARCHIVE_WARN);
4162 }
4163 if (rsrc->compressed_size > ZIP_MAX_METADATA * 1048576U) {
4164 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
4165 "Mac metadata is too large: %jd > %u MiB",
4166 (intmax_t)rsrc->compressed_size, ZIP_MAX_METADATA);
4167 return (ARCHIVE_WARN);
4168 }
4169
4170 metadata = malloc((size_t)rsrc->uncompressed_size);
4171 if (metadata == NULL) {
4172 archive_set_error(&a->archive, ENOMEM,
4173 "Can't allocate memory for Mac metadata");
4174 return (ARCHIVE_FATAL);
4175 }
4176
4177 if (offset < rsrc->local_header_offset)
4178 __archive_read_consume(a, rsrc->local_header_offset - offset);
4179 else if (offset != rsrc->local_header_offset) {
4180 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET);
4181 }
4182
4183 hsize = zip_get_local_file_header_size(a, 0);
4184 __archive_read_consume(a, hsize);
4185
4186 remaining_bytes = (size_t)rsrc->compressed_size;
4187 metadata_bytes = (size_t)rsrc->uncompressed_size;
4188 mp = metadata;
4189 eof = 0;
4190 while (!eof && remaining_bytes) {
4191 const unsigned char *p;
4192 ssize_t bytes_avail;
4193 size_t bytes_used;
4194
4195 p = __archive_read_ahead(a, 1, &bytes_avail);
4196 if (p == NULL) {
4197 archive_set_error(&a->archive,
4198 ARCHIVE_ERRNO_FILE_FORMAT,
4199 "Truncated ZIP file header");
4200 ret = ARCHIVE_WARN;
4201 goto exit_mac_metadata;
4202 }
4203 if ((size_t)bytes_avail > remaining_bytes)
4204 bytes_avail = remaining_bytes;
4205 switch(rsrc->compression) {
4206 case 0: /* No compression. */
4207 if ((size_t)bytes_avail > metadata_bytes)
4208 bytes_avail = metadata_bytes;
4209 memcpy(mp, p, bytes_avail);
4210 bytes_used = (size_t)bytes_avail;
4211 metadata_bytes -= bytes_used;
4212 mp += bytes_used;
4213 if (metadata_bytes == 0)
4214 eof = 1;
4215 break;
4216 #ifdef HAVE_ZLIB_H
4217 case 8: /* Deflate compression. */
4218 {
4219 int r;
4220
4221 ret = zip_deflate_init(a, zip);
4222 if (ret != ARCHIVE_OK)
4223 goto exit_mac_metadata;
4224 zip->stream.next_in =
4225 (Bytef *)(uintptr_t)(const void *)p;
4226 zip->stream.avail_in = (uInt)bytes_avail;
4227 zip->stream.total_in = 0;
4228 zip->stream.next_out = mp;
4229 zip->stream.avail_out = (uInt)metadata_bytes;
4230 zip->stream.total_out = 0;
4231
4232 r = inflate(&zip->stream, 0);
4233 switch (r) {
4234 case Z_OK:
4235 break;
4236 case Z_STREAM_END:
4237 eof = 1;
4238 break;
4239 case Z_MEM_ERROR:
4240 archive_set_error(&a->archive, ENOMEM,
4241 "Out of memory for ZIP decompression");
4242 ret = ARCHIVE_FATAL;
4243 goto exit_mac_metadata;
4244 default:
4245 archive_set_error(&a->archive,
4246 ARCHIVE_ERRNO_MISC,
4247 "ZIP decompression failed (%d)", r);
4248 ret = ARCHIVE_FATAL;
4249 goto exit_mac_metadata;
4250 }
4251 bytes_used = zip->stream.total_in;
4252 metadata_bytes -= zip->stream.total_out;
4253 mp += zip->stream.total_out;
4254 break;
4255 }
4256 #endif
4257 default:
4258 bytes_used = 0;
4259 break;
4260 }
4261 __archive_read_consume(a, bytes_used);
4262 remaining_bytes -= bytes_used;
4263 }
4264 archive_entry_copy_mac_metadata(entry, metadata,
4265 (size_t)rsrc->uncompressed_size - metadata_bytes);
4266
4267 exit_mac_metadata:
4268 __archive_read_seek(a, offset, SEEK_SET);
4269 zip->decompress_init = 0;
4270 free(metadata);
4271 return (ret);
4272 }
4273
4274 static int
archive_read_format_zip_seekable_read_header(struct archive_read * a,struct archive_entry * entry)4275 archive_read_format_zip_seekable_read_header(struct archive_read *a,
4276 struct archive_entry *entry)
4277 {
4278 struct zip *zip = (struct zip *)a->format->data;
4279 struct zip_entry *rsrc;
4280 int64_t offset;
4281 int r, ret = ARCHIVE_OK;
4282
4283 /*
4284 * It should be sufficient to call archive_read_next_header() for
4285 * a reader to determine if an entry is encrypted or not. If the
4286 * encryption of an entry is only detectable when calling
4287 * archive_read_data(), so be it. We'll do the same check there
4288 * as well.
4289 */
4290 if (zip->has_encrypted_entries ==
4291 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
4292 zip->has_encrypted_entries = 0;
4293
4294 a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
4295 if (a->archive.archive_format_name == NULL)
4296 a->archive.archive_format_name = "ZIP";
4297
4298 if (zip->zip_entries == NULL) {
4299 r = slurp_central_directory(a, entry, zip);
4300 if (r != ARCHIVE_OK)
4301 return r;
4302 /* Get first entry whose local header offset is lower than
4303 * other entries in the archive file. */
4304 zip->entry =
4305 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree);
4306 } else if (zip->entry != NULL) {
4307 /* Get next entry in local header offset order. */
4308 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate(
4309 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT);
4310 }
4311
4312 if (zip->entry == NULL)
4313 return ARCHIVE_EOF;
4314
4315 if (zip->entry->rsrcname.s)
4316 rsrc = (struct zip_entry *)__archive_rb_tree_find_node(
4317 &zip->tree_rsrc, zip->entry->rsrcname.s);
4318 else
4319 rsrc = NULL;
4320
4321 if (zip->cctx_valid)
4322 archive_decrypto_aes_ctr_release(&zip->cctx);
4323 if (zip->hctx_valid)
4324 archive_hmac_sha1_cleanup(&zip->hctx);
4325 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
4326 __archive_read_reset_passphrase(a);
4327
4328 /* File entries are sorted by the header offset, we should mostly
4329 * use __archive_read_consume to advance a read point to avoid
4330 * redundant data reading. */
4331 offset = archive_filter_bytes(&a->archive, 0);
4332 if (offset < zip->entry->local_header_offset)
4333 __archive_read_consume(a,
4334 zip->entry->local_header_offset - offset);
4335 else if (offset != zip->entry->local_header_offset) {
4336 __archive_read_seek(a, zip->entry->local_header_offset,
4337 SEEK_SET);
4338 }
4339 zip->unconsumed = 0;
4340 r = zip_read_local_file_header(a, entry, zip);
4341 if (r != ARCHIVE_OK)
4342 return r;
4343 if (rsrc) {
4344 int ret2 = zip_read_mac_metadata(a, entry, rsrc);
4345 if (ret2 < ret)
4346 ret = ret2;
4347 }
4348 return (ret);
4349 }
4350
4351 /*
4352 * We're going to seek for the next header anyway, so we don't
4353 * need to bother doing anything here.
4354 */
4355 static int
archive_read_format_zip_read_data_skip_seekable(struct archive_read * a)4356 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a)
4357 {
4358 struct zip *zip;
4359 zip = (struct zip *)(a->format->data);
4360
4361 zip->unconsumed = 0;
4362 return (ARCHIVE_OK);
4363 }
4364
4365 int
archive_read_support_format_zip_seekable(struct archive * _a)4366 archive_read_support_format_zip_seekable(struct archive *_a)
4367 {
4368 struct archive_read *a = (struct archive_read *)_a;
4369 struct zip *zip;
4370 int r;
4371
4372 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
4373 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable");
4374
4375 zip = calloc(1, sizeof(*zip));
4376 if (zip == NULL) {
4377 archive_set_error(&a->archive, ENOMEM,
4378 "Can't allocate zip data");
4379 return (ARCHIVE_FATAL);
4380 }
4381
4382 #ifdef HAVE_COPYFILE_H
4383 /* Set this by default on Mac OS. */
4384 zip->process_mac_extensions = 1;
4385 #endif
4386
4387 /*
4388 * Until enough data has been read, we cannot tell about
4389 * any encrypted entries yet.
4390 */
4391 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
4392 zip->crc32func = real_crc32;
4393
4394 r = __archive_read_register_format(a,
4395 zip,
4396 "zip",
4397 archive_read_format_zip_seekable_bid,
4398 archive_read_format_zip_options,
4399 archive_read_format_zip_seekable_read_header,
4400 archive_read_format_zip_read_data,
4401 archive_read_format_zip_read_data_skip_seekable,
4402 NULL,
4403 archive_read_format_zip_cleanup,
4404 archive_read_support_format_zip_capabilities_seekable,
4405 archive_read_format_zip_has_encrypted_entries);
4406
4407 if (r != ARCHIVE_OK)
4408 free(zip);
4409 return (ARCHIVE_OK);
4410 }
4411
4412 /*# vim:set noet:*/
4413