1 /*-
2 * Copyright (c) 2004-2013 Tim Kientzle
3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA
4 * Copyright (c) 2013 Konrad Kleine
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27
28 #include "archive_platform.h"
29 __FBSDID("$FreeBSD$");
30
31 /*
32 * The definitive documentation of the Zip file format is:
33 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT
34 *
35 * The Info-Zip project has pioneered various extensions to better
36 * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855
37 * "Ux", and 0x7875 "ux" extensions for time and ownership
38 * information.
39 *
40 * History of this code: The streaming Zip reader was first added to
41 * libarchive in January 2005. Support for seekable input sources was
42 * added in Nov 2011. Zip64 support (including a significant code
43 * refactoring) was added in 2014.
44 */
45
46 #ifdef HAVE_ERRNO_H
47 #include <errno.h>
48 #endif
49 #ifdef HAVE_STDLIB_H
50 #include <stdlib.h>
51 #endif
52 #ifdef HAVE_ZLIB_H
53 #include <zlib.h>
54 #endif
55 #ifdef HAVE_BZLIB_H
56 #include <bzlib.h>
57 #endif
58 #ifdef HAVE_LZMA_H
59 #include <lzma.h>
60 #endif
61
62 #include "archive.h"
63 #include "archive_digest_private.h"
64 #include "archive_cryptor_private.h"
65 #include "archive_endian.h"
66 #include "archive_entry.h"
67 #include "archive_entry_locale.h"
68 #include "archive_hmac_private.h"
69 #include "archive_private.h"
70 #include "archive_rb.h"
71 #include "archive_read_private.h"
72 #include "archive_ppmd8_private.h"
73
74 #ifndef HAVE_ZLIB_H
75 #include "archive_crc32.h"
76 #endif
77
78 struct zip_entry {
79 struct archive_rb_node node;
80 struct zip_entry *next;
81 int64_t local_header_offset;
82 int64_t compressed_size;
83 int64_t uncompressed_size;
84 int64_t gid;
85 int64_t uid;
86 struct archive_string rsrcname;
87 time_t mtime;
88 time_t atime;
89 time_t ctime;
90 uint32_t crc32;
91 uint16_t mode;
92 uint16_t zip_flags; /* From GP Flags Field */
93 unsigned char compression;
94 unsigned char system; /* From "version written by" */
95 unsigned char flags; /* Our extra markers. */
96 unsigned char decdat;/* Used for Decryption check */
97
98 /* WinZip AES encryption extra field should be available
99 * when compression is 99. */
100 struct {
101 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */
102 unsigned vendor;
103 #define AES_VENDOR_AE_1 0x0001
104 #define AES_VENDOR_AE_2 0x0002
105 /* AES encryption strength:
106 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */
107 unsigned strength;
108 /* Actual compression method. */
109 unsigned char compression;
110 } aes_extra;
111 };
112
113 struct trad_enc_ctx {
114 uint32_t keys[3];
115 };
116
117 /* Bits used in zip_flags. */
118 #define ZIP_ENCRYPTED (1 << 0)
119 #define ZIP_LENGTH_AT_END (1 << 3)
120 #define ZIP_STRONG_ENCRYPTED (1 << 6)
121 #define ZIP_UTF8_NAME (1 << 11)
122 /* See "7.2 Single Password Symmetric Encryption Method"
123 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */
124 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13)
125
126 /* Bits used in flags. */
127 #define LA_USED_ZIP64 (1 << 0)
128 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1)
129
130 /*
131 * See "WinZip - AES Encryption Information"
132 * http://www.winzip.com/aes_info.htm
133 */
134 /* Value used in compression method. */
135 #define WINZIP_AES_ENCRYPTION 99
136 /* Authentication code size. */
137 #define AUTH_CODE_SIZE 10
138 /**/
139 #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2)
140
141 struct zip {
142 /* Structural information about the archive. */
143 struct archive_string format_name;
144 int64_t central_directory_offset;
145 size_t central_directory_entries_total;
146 size_t central_directory_entries_on_this_disk;
147 int has_encrypted_entries;
148
149 /* List of entries (seekable Zip only) */
150 struct zip_entry *zip_entries;
151 struct archive_rb_tree tree;
152 struct archive_rb_tree tree_rsrc;
153
154 /* Bytes read but not yet consumed via __archive_read_consume() */
155 size_t unconsumed;
156
157 /* Information about entry we're currently reading. */
158 struct zip_entry *entry;
159 int64_t entry_bytes_remaining;
160
161 /* These count the number of bytes actually read for the entry. */
162 int64_t entry_compressed_bytes_read;
163 int64_t entry_uncompressed_bytes_read;
164
165 /* Running CRC32 of the decompressed data */
166 unsigned long entry_crc32;
167 unsigned long (*crc32func)(unsigned long, const void *,
168 size_t);
169 char ignore_crc32;
170
171 /* Flags to mark progress of decompression. */
172 char decompress_init;
173 char end_of_entry;
174
175 unsigned char *uncompressed_buffer;
176 size_t uncompressed_buffer_size;
177
178 #ifdef HAVE_ZLIB_H
179 z_stream stream;
180 char stream_valid;
181 #endif
182
183 #if HAVE_LZMA_H && HAVE_LIBLZMA
184 lzma_stream zipx_lzma_stream;
185 char zipx_lzma_valid;
186 #endif
187
188 #ifdef HAVE_BZLIB_H
189 bz_stream bzstream;
190 char bzstream_valid;
191 #endif
192
193 IByteIn zipx_ppmd_stream;
194 ssize_t zipx_ppmd_read_compressed;
195 CPpmd8 ppmd8;
196 char ppmd8_valid;
197 char ppmd8_stream_failed;
198
199 struct archive_string_conv *sconv;
200 struct archive_string_conv *sconv_default;
201 struct archive_string_conv *sconv_utf8;
202 int init_default_conversion;
203 int process_mac_extensions;
204
205 char init_decryption;
206
207 /* Decryption buffer. */
208 /*
209 * The decrypted data starts at decrypted_ptr and
210 * extends for decrypted_bytes_remaining. Decryption
211 * adds new data to the end of this block, data is returned
212 * to clients from the beginning. When the block hits the
213 * end of decrypted_buffer, it has to be shuffled back to
214 * the beginning of the buffer.
215 */
216 unsigned char *decrypted_buffer;
217 unsigned char *decrypted_ptr;
218 size_t decrypted_buffer_size;
219 size_t decrypted_bytes_remaining;
220 size_t decrypted_unconsumed_bytes;
221
222 /* Traditional PKWARE decryption. */
223 struct trad_enc_ctx tctx;
224 char tctx_valid;
225
226 /* WinZip AES decryption. */
227 /* Contexts used for AES decryption. */
228 archive_crypto_ctx cctx;
229 char cctx_valid;
230 archive_hmac_sha1_ctx hctx;
231 char hctx_valid;
232
233 /* Strong encryption's decryption header information. */
234 unsigned iv_size;
235 unsigned alg_id;
236 unsigned bit_len;
237 unsigned flags;
238 unsigned erd_size;
239 unsigned v_size;
240 unsigned v_crc32;
241 uint8_t *iv;
242 uint8_t *erd;
243 uint8_t *v_data;
244 };
245
246 /* Many systems define min or MIN, but not all. */
247 #define zipmin(a,b) ((a) < (b) ? (a) : (b))
248
249 /* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8
250 * streams inside ZIP files. It has 2 purposes: one is to fetch the next
251 * compressed byte from the stream, second one is to increase the counter how
252 * many compressed bytes were read. */
253 static Byte
ppmd_read(void * p)254 ppmd_read(void* p) {
255 /* Get the handle to current decompression context. */
256 struct archive_read *a = ((IByteIn*)p)->a;
257 struct zip *zip = (struct zip*) a->format->data;
258 ssize_t bytes_avail = 0;
259
260 /* Fetch next byte. */
261 const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail);
262 if(bytes_avail < 1) {
263 zip->ppmd8_stream_failed = 1;
264 return 0;
265 }
266
267 __archive_read_consume(a, 1);
268
269 /* Increment the counter. */
270 ++zip->zipx_ppmd_read_compressed;
271
272 /* Return the next compressed byte. */
273 return data[0];
274 }
275
276 /* ------------------------------------------------------------------------ */
277
278 /*
279 Traditional PKWARE Decryption functions.
280 */
281
282 static void
trad_enc_update_keys(struct trad_enc_ctx * ctx,uint8_t c)283 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c)
284 {
285 uint8_t t;
286 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL)
287
288 ctx->keys[0] = CRC32(ctx->keys[0], c);
289 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1;
290 t = (ctx->keys[1] >> 24) & 0xff;
291 ctx->keys[2] = CRC32(ctx->keys[2], t);
292 #undef CRC32
293 }
294
295 static uint8_t
trad_enc_decrypt_byte(struct trad_enc_ctx * ctx)296 trad_enc_decrypt_byte(struct trad_enc_ctx *ctx)
297 {
298 unsigned temp = ctx->keys[2] | 2;
299 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff;
300 }
301
302 static void
trad_enc_decrypt_update(struct trad_enc_ctx * ctx,const uint8_t * in,size_t in_len,uint8_t * out,size_t out_len)303 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in,
304 size_t in_len, uint8_t *out, size_t out_len)
305 {
306 unsigned i, max;
307
308 max = (unsigned)((in_len < out_len)? in_len: out_len);
309
310 for (i = 0; i < max; i++) {
311 uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx);
312 out[i] = t;
313 trad_enc_update_keys(ctx, t);
314 }
315 }
316
317 static int
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)318 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len,
319 const uint8_t *key, size_t key_len, uint8_t *crcchk)
320 {
321 uint8_t header[12];
322
323 if (key_len < 12) {
324 *crcchk = 0xff;
325 return -1;
326 }
327
328 ctx->keys[0] = 305419896L;
329 ctx->keys[1] = 591751049L;
330 ctx->keys[2] = 878082192L;
331
332 for (;pw_len; --pw_len)
333 trad_enc_update_keys(ctx, *pw++);
334
335 trad_enc_decrypt_update(ctx, key, 12, header, 12);
336 /* Return the last byte for CRC check. */
337 *crcchk = header[11];
338 return 0;
339 }
340
341 #if 0
342 static void
343 crypt_derive_key_sha1(const void *p, int size, unsigned char *key,
344 int key_size)
345 {
346 #define MD_SIZE 20
347 archive_sha1_ctx ctx;
348 unsigned char md1[MD_SIZE];
349 unsigned char md2[MD_SIZE * 2];
350 unsigned char mkb[64];
351 int i;
352
353 archive_sha1_init(&ctx);
354 archive_sha1_update(&ctx, p, size);
355 archive_sha1_final(&ctx, md1);
356
357 memset(mkb, 0x36, sizeof(mkb));
358 for (i = 0; i < MD_SIZE; i++)
359 mkb[i] ^= md1[i];
360 archive_sha1_init(&ctx);
361 archive_sha1_update(&ctx, mkb, sizeof(mkb));
362 archive_sha1_final(&ctx, md2);
363
364 memset(mkb, 0x5C, sizeof(mkb));
365 for (i = 0; i < MD_SIZE; i++)
366 mkb[i] ^= md1[i];
367 archive_sha1_init(&ctx);
368 archive_sha1_update(&ctx, mkb, sizeof(mkb));
369 archive_sha1_final(&ctx, md2 + MD_SIZE);
370
371 if (key_size > 32)
372 key_size = 32;
373 memcpy(key, md2, key_size);
374 #undef MD_SIZE
375 }
376 #endif
377
378 /*
379 * Common code for streaming or seeking modes.
380 *
381 * Includes code to read local file headers, decompress data
382 * from entry bodies, and common API.
383 */
384
385 static unsigned long
real_crc32(unsigned long crc,const void * buff,size_t len)386 real_crc32(unsigned long crc, const void *buff, size_t len)
387 {
388 return crc32(crc, buff, (unsigned int)len);
389 }
390
391 /* Used by "ignorecrc32" option to speed up tests. */
392 static unsigned long
fake_crc32(unsigned long crc,const void * buff,size_t len)393 fake_crc32(unsigned long crc, const void *buff, size_t len)
394 {
395 (void)crc; /* UNUSED */
396 (void)buff; /* UNUSED */
397 (void)len; /* UNUSED */
398 return 0;
399 }
400
401 static const struct {
402 int id;
403 const char * name;
404 } compression_methods[] = {
405 {0, "uncompressed"}, /* The file is stored (no compression) */
406 {1, "shrinking"}, /* The file is Shrunk */
407 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */
408 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */
409 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */
410 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */
411 {6, "imploded"}, /* The file is Imploded */
412 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */
413 {8, "deflation"}, /* The file is Deflated */
414 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */
415 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding
416 * (old IBM TERSE) */
417 {11, "reserved"}, /* Reserved by PKWARE */
418 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */
419 {13, "reserved"}, /* Reserved by PKWARE */
420 {14, "lzma"}, /* LZMA (EFS) */
421 {15, "reserved"}, /* Reserved by PKWARE */
422 {16, "reserved"}, /* Reserved by PKWARE */
423 {17, "reserved"}, /* Reserved by PKWARE */
424 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */
425 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */
426 {95, "xz"}, /* XZ compressed data */
427 {96, "jpeg"}, /* JPEG compressed data */
428 {97, "wav-pack"}, /* WavPack compressed data */
429 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */
430 {99, "aes"} /* WinZip AES encryption */
431 };
432
433 static const char *
compression_name(const int compression)434 compression_name(const int compression)
435 {
436 static const int num_compression_methods =
437 sizeof(compression_methods)/sizeof(compression_methods[0]);
438 int i=0;
439
440 while(compression >= 0 && i < num_compression_methods) {
441 if (compression_methods[i].id == compression)
442 return compression_methods[i].name;
443 i++;
444 }
445 return "??";
446 }
447
448 /* Convert an MSDOS-style date/time into Unix-style time. */
449 static time_t
zip_time(const char * p)450 zip_time(const char *p)
451 {
452 int msTime, msDate;
453 struct tm ts;
454
455 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]);
456 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]);
457
458 memset(&ts, 0, sizeof(ts));
459 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */
460 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */
461 ts.tm_mday = msDate & 0x1f; /* Day of month. */
462 ts.tm_hour = (msTime >> 11) & 0x1f;
463 ts.tm_min = (msTime >> 5) & 0x3f;
464 ts.tm_sec = (msTime << 1) & 0x3e;
465 ts.tm_isdst = -1;
466 return mktime(&ts);
467 }
468
469 /*
470 * The extra data is stored as a list of
471 * id1+size1+data1 + id2+size2+data2 ...
472 * triplets. id and size are 2 bytes each.
473 */
474 static int
process_extra(struct archive_read * a,struct archive_entry * entry,const char * p,size_t extra_length,struct zip_entry * zip_entry)475 process_extra(struct archive_read *a, struct archive_entry *entry,
476 const char *p, size_t extra_length, struct zip_entry* zip_entry)
477 {
478 unsigned offset = 0;
479 struct zip *zip = (struct zip *)(a->format->data);
480
481 if (extra_length == 0) {
482 return ARCHIVE_OK;
483 }
484
485 if (extra_length < 4) {
486 size_t i = 0;
487 /* Some ZIP files may have trailing 0 bytes. Let's check they
488 * are all 0 and ignore them instead of returning an error.
489 *
490 * This is not techincally correct, but some ZIP files look
491 * like this and other tools support those files - so let's
492 * also support them.
493 */
494 for (; i < extra_length; i++) {
495 if (p[i] != 0) {
496 archive_set_error(&a->archive,
497 ARCHIVE_ERRNO_FILE_FORMAT,
498 "Too-small extra data: "
499 "Need at least 4 bytes, "
500 "but only found %d bytes",
501 (int)extra_length);
502 return ARCHIVE_FAILED;
503 }
504 }
505
506 return ARCHIVE_OK;
507 }
508
509 while (offset <= extra_length - 4) {
510 unsigned short headerid = archive_le16dec(p + offset);
511 unsigned short datasize = archive_le16dec(p + offset + 2);
512
513 offset += 4;
514 if (offset + datasize > extra_length) {
515 archive_set_error(&a->archive,
516 ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: "
517 "Need %d bytes but only found %d bytes",
518 (int)datasize, (int)(extra_length - offset));
519 return ARCHIVE_FAILED;
520 }
521 #ifdef DEBUG
522 fprintf(stderr, "Header id 0x%04x, length %d\n",
523 headerid, datasize);
524 #endif
525 switch (headerid) {
526 case 0x0001:
527 /* Zip64 extended information extra field. */
528 zip_entry->flags |= LA_USED_ZIP64;
529 if (zip_entry->uncompressed_size == 0xffffffff) {
530 uint64_t t = 0;
531 if (datasize < 8
532 || (t = archive_le64dec(p + offset)) >
533 INT64_MAX) {
534 archive_set_error(&a->archive,
535 ARCHIVE_ERRNO_FILE_FORMAT,
536 "Malformed 64-bit "
537 "uncompressed size");
538 return ARCHIVE_FAILED;
539 }
540 zip_entry->uncompressed_size = t;
541 offset += 8;
542 datasize -= 8;
543 }
544 if (zip_entry->compressed_size == 0xffffffff) {
545 uint64_t t = 0;
546 if (datasize < 8
547 || (t = archive_le64dec(p + offset)) >
548 INT64_MAX) {
549 archive_set_error(&a->archive,
550 ARCHIVE_ERRNO_FILE_FORMAT,
551 "Malformed 64-bit "
552 "compressed size");
553 return ARCHIVE_FAILED;
554 }
555 zip_entry->compressed_size = t;
556 offset += 8;
557 datasize -= 8;
558 }
559 if (zip_entry->local_header_offset == 0xffffffff) {
560 uint64_t t = 0;
561 if (datasize < 8
562 || (t = archive_le64dec(p + offset)) >
563 INT64_MAX) {
564 archive_set_error(&a->archive,
565 ARCHIVE_ERRNO_FILE_FORMAT,
566 "Malformed 64-bit "
567 "local header offset");
568 return ARCHIVE_FAILED;
569 }
570 zip_entry->local_header_offset = t;
571 offset += 8;
572 datasize -= 8;
573 }
574 /* archive_le32dec(p + offset) gives disk
575 * on which file starts, but we don't handle
576 * multi-volume Zip files. */
577 break;
578 #ifdef DEBUG
579 case 0x0017:
580 {
581 /* Strong encryption field. */
582 if (archive_le16dec(p + offset) == 2) {
583 unsigned algId =
584 archive_le16dec(p + offset + 2);
585 unsigned bitLen =
586 archive_le16dec(p + offset + 4);
587 int flags =
588 archive_le16dec(p + offset + 6);
589 fprintf(stderr, "algId=0x%04x, bitLen=%u, "
590 "flgas=%d\n", algId, bitLen,flags);
591 }
592 break;
593 }
594 #endif
595 case 0x5455:
596 {
597 /* Extended time field "UT". */
598 int flags;
599 if (datasize == 0) {
600 archive_set_error(&a->archive,
601 ARCHIVE_ERRNO_FILE_FORMAT,
602 "Incomplete extended time field");
603 return ARCHIVE_FAILED;
604 }
605 flags = p[offset];
606 offset++;
607 datasize--;
608 /* Flag bits indicate which dates are present. */
609 if (flags & 0x01)
610 {
611 #ifdef DEBUG
612 fprintf(stderr, "mtime: %lld -> %d\n",
613 (long long)zip_entry->mtime,
614 archive_le32dec(p + offset));
615 #endif
616 if (datasize < 4)
617 break;
618 zip_entry->mtime = archive_le32dec(p + offset);
619 offset += 4;
620 datasize -= 4;
621 }
622 if (flags & 0x02)
623 {
624 if (datasize < 4)
625 break;
626 zip_entry->atime = archive_le32dec(p + offset);
627 offset += 4;
628 datasize -= 4;
629 }
630 if (flags & 0x04)
631 {
632 if (datasize < 4)
633 break;
634 zip_entry->ctime = archive_le32dec(p + offset);
635 offset += 4;
636 datasize -= 4;
637 }
638 break;
639 }
640 case 0x5855:
641 {
642 /* Info-ZIP Unix Extra Field (old version) "UX". */
643 if (datasize >= 8) {
644 zip_entry->atime = archive_le32dec(p + offset);
645 zip_entry->mtime =
646 archive_le32dec(p + offset + 4);
647 }
648 if (datasize >= 12) {
649 zip_entry->uid =
650 archive_le16dec(p + offset + 8);
651 zip_entry->gid =
652 archive_le16dec(p + offset + 10);
653 }
654 break;
655 }
656 case 0x6c78:
657 {
658 /* Experimental 'xl' field */
659 /*
660 * Introduced Dec 2013 to provide a way to
661 * include external file attributes (and other
662 * fields that ordinarily appear only in
663 * central directory) in local file header.
664 * This provides file type and permission
665 * information necessary to support full
666 * streaming extraction. Currently being
667 * discussed with other Zip developers
668 * ... subject to change.
669 *
670 * Format:
671 * The field starts with a bitmap that specifies
672 * which additional fields are included. The
673 * bitmap is variable length and can be extended in
674 * the future.
675 *
676 * n bytes - feature bitmap: first byte has low-order
677 * 7 bits. If high-order bit is set, a subsequent
678 * byte holds the next 7 bits, etc.
679 *
680 * if bitmap & 1, 2 byte "version made by"
681 * if bitmap & 2, 2 byte "internal file attributes"
682 * if bitmap & 4, 4 byte "external file attributes"
683 * if bitmap & 8, 2 byte comment length + n byte
684 * comment
685 */
686 int bitmap, bitmap_last;
687
688 if (datasize < 1)
689 break;
690 bitmap_last = bitmap = 0xff & p[offset];
691 offset += 1;
692 datasize -= 1;
693
694 /* We only support first 7 bits of bitmap; skip rest. */
695 while ((bitmap_last & 0x80) != 0
696 && datasize >= 1) {
697 bitmap_last = p[offset];
698 offset += 1;
699 datasize -= 1;
700 }
701
702 if (bitmap & 1) {
703 /* 2 byte "version made by" */
704 if (datasize < 2)
705 break;
706 zip_entry->system
707 = archive_le16dec(p + offset) >> 8;
708 offset += 2;
709 datasize -= 2;
710 }
711 if (bitmap & 2) {
712 /* 2 byte "internal file attributes" */
713 uint32_t internal_attributes;
714 if (datasize < 2)
715 break;
716 internal_attributes
717 = archive_le16dec(p + offset);
718 /* Not used by libarchive at present. */
719 (void)internal_attributes; /* UNUSED */
720 offset += 2;
721 datasize -= 2;
722 }
723 if (bitmap & 4) {
724 /* 4 byte "external file attributes" */
725 uint32_t external_attributes;
726 if (datasize < 4)
727 break;
728 external_attributes
729 = archive_le32dec(p + offset);
730 if (zip_entry->system == 3) {
731 zip_entry->mode
732 = external_attributes >> 16;
733 } else if (zip_entry->system == 0) {
734 // Interpret MSDOS directory bit
735 if (0x10 == (external_attributes &
736 0x10)) {
737 zip_entry->mode =
738 AE_IFDIR | 0775;
739 } else {
740 zip_entry->mode =
741 AE_IFREG | 0664;
742 }
743 if (0x01 == (external_attributes &
744 0x01)) {
745 /* Read-only bit;
746 * strip write permissions */
747 zip_entry->mode &= 0555;
748 }
749 } else {
750 zip_entry->mode = 0;
751 }
752 offset += 4;
753 datasize -= 4;
754 }
755 if (bitmap & 8) {
756 /* 2 byte comment length + comment */
757 uint32_t comment_length;
758 if (datasize < 2)
759 break;
760 comment_length
761 = archive_le16dec(p + offset);
762 offset += 2;
763 datasize -= 2;
764
765 if (datasize < comment_length)
766 break;
767 /* Comment is not supported by libarchive */
768 offset += comment_length;
769 datasize -= comment_length;
770 }
771 break;
772 }
773 case 0x7075:
774 {
775 /* Info-ZIP Unicode Path Extra Field. */
776 if (datasize < 5 || entry == NULL)
777 break;
778 offset += 5;
779 datasize -= 5;
780
781 /* The path name in this field is always encoded
782 * in UTF-8. */
783 if (zip->sconv_utf8 == NULL) {
784 zip->sconv_utf8 =
785 archive_string_conversion_from_charset(
786 &a->archive, "UTF-8", 1);
787 /* If the converter from UTF-8 is not
788 * available, then the path name from the main
789 * field will more likely be correct. */
790 if (zip->sconv_utf8 == NULL)
791 break;
792 }
793
794 /* Make sure the CRC32 of the filename matches. */
795 if (!zip->ignore_crc32) {
796 const char *cp = archive_entry_pathname(entry);
797 if (cp) {
798 unsigned long file_crc =
799 zip->crc32func(0, cp, strlen(cp));
800 unsigned long utf_crc =
801 archive_le32dec(p + offset - 4);
802 if (file_crc != utf_crc) {
803 #ifdef DEBUG
804 fprintf(stderr,
805 "CRC filename mismatch; "
806 "CDE is %lx, but UTF8 "
807 "is outdated with %lx\n",
808 file_crc, utf_crc);
809 #endif
810 break;
811 }
812 }
813 }
814
815 if (archive_entry_copy_pathname_l(entry,
816 p + offset, datasize, zip->sconv_utf8) != 0) {
817 /* Ignore the error, and fallback to the path
818 * name from the main field. */
819 #ifdef DEBUG
820 fprintf(stderr, "Failed to read the ZIP "
821 "0x7075 extra field path.\n");
822 #endif
823 }
824 break;
825 }
826 case 0x7855:
827 /* Info-ZIP Unix Extra Field (type 2) "Ux". */
828 #ifdef DEBUG
829 fprintf(stderr, "uid %d gid %d\n",
830 archive_le16dec(p + offset),
831 archive_le16dec(p + offset + 2));
832 #endif
833 if (datasize >= 2)
834 zip_entry->uid = archive_le16dec(p + offset);
835 if (datasize >= 4)
836 zip_entry->gid =
837 archive_le16dec(p + offset + 2);
838 break;
839 case 0x7875:
840 {
841 /* Info-Zip Unix Extra Field (type 3) "ux". */
842 int uidsize = 0, gidsize = 0;
843
844 /* TODO: support arbitrary uidsize/gidsize. */
845 if (datasize >= 1 && p[offset] == 1) {/* version=1 */
846 if (datasize >= 4) {
847 /* get a uid size. */
848 uidsize = 0xff & (int)p[offset+1];
849 if (uidsize == 2)
850 zip_entry->uid =
851 archive_le16dec(
852 p + offset + 2);
853 else if (uidsize == 4 && datasize >= 6)
854 zip_entry->uid =
855 archive_le32dec(
856 p + offset + 2);
857 }
858 if (datasize >= (2 + uidsize + 3)) {
859 /* get a gid size. */
860 gidsize = 0xff &
861 (int)p[offset+2+uidsize];
862 if (gidsize == 2)
863 zip_entry->gid =
864 archive_le16dec(
865 p+offset+2+uidsize+1);
866 else if (gidsize == 4 &&
867 datasize >= (2 + uidsize + 5))
868 zip_entry->gid =
869 archive_le32dec(
870 p+offset+2+uidsize+1);
871 }
872 }
873 break;
874 }
875 case 0x9901:
876 /* WinZip AES extra data field. */
877 if (datasize < 6) {
878 archive_set_error(&a->archive,
879 ARCHIVE_ERRNO_FILE_FORMAT,
880 "Incomplete AES field");
881 return ARCHIVE_FAILED;
882 }
883 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') {
884 /* Vendor version. */
885 zip_entry->aes_extra.vendor =
886 archive_le16dec(p + offset);
887 /* AES encryption strength. */
888 zip_entry->aes_extra.strength = p[offset + 4];
889 /* Actual compression method. */
890 zip_entry->aes_extra.compression =
891 p[offset + 5];
892 }
893 break;
894 default:
895 break;
896 }
897 offset += datasize;
898 }
899 return ARCHIVE_OK;
900 }
901
902 /*
903 * Assumes file pointer is at beginning of local file header.
904 */
905 static int
zip_read_local_file_header(struct archive_read * a,struct archive_entry * entry,struct zip * zip)906 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry,
907 struct zip *zip)
908 {
909 const char *p;
910 const void *h;
911 const wchar_t *wp;
912 const char *cp;
913 size_t len, filename_length, extra_length;
914 struct archive_string_conv *sconv;
915 struct zip_entry *zip_entry = zip->entry;
916 struct zip_entry zip_entry_central_dir;
917 int ret = ARCHIVE_OK;
918 char version;
919
920 /* Save a copy of the original for consistency checks. */
921 zip_entry_central_dir = *zip_entry;
922
923 zip->decompress_init = 0;
924 zip->end_of_entry = 0;
925 zip->entry_uncompressed_bytes_read = 0;
926 zip->entry_compressed_bytes_read = 0;
927 zip->entry_crc32 = zip->crc32func(0, NULL, 0);
928
929 /* Setup default conversion. */
930 if (zip->sconv == NULL && !zip->init_default_conversion) {
931 zip->sconv_default =
932 archive_string_default_conversion_for_read(&(a->archive));
933 zip->init_default_conversion = 1;
934 }
935
936 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) {
937 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
938 "Truncated ZIP file header");
939 return (ARCHIVE_FATAL);
940 }
941
942 if (memcmp(p, "PK\003\004", 4) != 0) {
943 archive_set_error(&a->archive, -1, "Damaged Zip archive");
944 return ARCHIVE_FATAL;
945 }
946 version = p[4];
947 zip_entry->system = p[5];
948 zip_entry->zip_flags = archive_le16dec(p + 6);
949 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) {
950 zip->has_encrypted_entries = 1;
951 archive_entry_set_is_data_encrypted(entry, 1);
952 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED &&
953 zip_entry->zip_flags & ZIP_ENCRYPTED &&
954 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) {
955 archive_entry_set_is_metadata_encrypted(entry, 1);
956 return ARCHIVE_FATAL;
957 }
958 }
959 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED);
960 zip_entry->compression = (char)archive_le16dec(p + 8);
961 zip_entry->mtime = zip_time(p + 10);
962 zip_entry->crc32 = archive_le32dec(p + 14);
963 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
964 zip_entry->decdat = p[11];
965 else
966 zip_entry->decdat = p[17];
967 zip_entry->compressed_size = archive_le32dec(p + 18);
968 zip_entry->uncompressed_size = archive_le32dec(p + 22);
969 filename_length = archive_le16dec(p + 26);
970 extra_length = archive_le16dec(p + 28);
971
972 __archive_read_consume(a, 30);
973
974 /* Read the filename. */
975 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) {
976 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
977 "Truncated ZIP file header");
978 return (ARCHIVE_FATAL);
979 }
980 if (zip_entry->zip_flags & ZIP_UTF8_NAME) {
981 /* The filename is stored to be UTF-8. */
982 if (zip->sconv_utf8 == NULL) {
983 zip->sconv_utf8 =
984 archive_string_conversion_from_charset(
985 &a->archive, "UTF-8", 1);
986 if (zip->sconv_utf8 == NULL)
987 return (ARCHIVE_FATAL);
988 }
989 sconv = zip->sconv_utf8;
990 } else if (zip->sconv != NULL)
991 sconv = zip->sconv;
992 else
993 sconv = zip->sconv_default;
994
995 if (archive_entry_copy_pathname_l(entry,
996 h, filename_length, sconv) != 0) {
997 if (errno == ENOMEM) {
998 archive_set_error(&a->archive, ENOMEM,
999 "Can't allocate memory for Pathname");
1000 return (ARCHIVE_FATAL);
1001 }
1002 archive_set_error(&a->archive,
1003 ARCHIVE_ERRNO_FILE_FORMAT,
1004 "Pathname cannot be converted "
1005 "from %s to current locale.",
1006 archive_string_conversion_charset_name(sconv));
1007 ret = ARCHIVE_WARN;
1008 }
1009 __archive_read_consume(a, filename_length);
1010
1011 /* Read the extra data. */
1012 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) {
1013 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1014 "Truncated ZIP file header");
1015 return (ARCHIVE_FATAL);
1016 }
1017
1018 if (ARCHIVE_OK != process_extra(a, entry, h, extra_length,
1019 zip_entry)) {
1020 return ARCHIVE_FATAL;
1021 }
1022 __archive_read_consume(a, extra_length);
1023
1024 /* Work around a bug in Info-Zip: When reading from a pipe, it
1025 * stats the pipe instead of synthesizing a file entry. */
1026 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) {
1027 zip_entry->mode &= ~ AE_IFMT;
1028 zip_entry->mode |= AE_IFREG;
1029 }
1030
1031 /* If the mode is totally empty, set some sane default. */
1032 if (zip_entry->mode == 0) {
1033 zip_entry->mode |= 0664;
1034 }
1035
1036 /* Windows archivers sometimes use backslash as the directory
1037 * separator. Normalize to slash. */
1038 if (zip_entry->system == 0 &&
1039 (wp = archive_entry_pathname_w(entry)) != NULL) {
1040 if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) {
1041 size_t i;
1042 struct archive_wstring s;
1043 archive_string_init(&s);
1044 archive_wstrcpy(&s, wp);
1045 for (i = 0; i < archive_strlen(&s); i++) {
1046 if (s.s[i] == '\\')
1047 s.s[i] = '/';
1048 }
1049 archive_entry_copy_pathname_w(entry, s.s);
1050 archive_wstring_free(&s);
1051 }
1052 }
1053
1054 /* Make sure that entries with a trailing '/' are marked as directories
1055 * even if the External File Attributes contains bogus values. If this
1056 * is not a directory and there is no type, assume regularfile. */
1057 if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) {
1058 int has_slash;
1059
1060 wp = archive_entry_pathname_w(entry);
1061 if (wp != NULL) {
1062 len = wcslen(wp);
1063 has_slash = len > 0 && wp[len - 1] == L'/';
1064 } else {
1065 cp = archive_entry_pathname(entry);
1066 len = (cp != NULL)?strlen(cp):0;
1067 has_slash = len > 0 && cp[len - 1] == '/';
1068 }
1069 /* Correct file type as needed. */
1070 if (has_slash) {
1071 zip_entry->mode &= ~AE_IFMT;
1072 zip_entry->mode |= AE_IFDIR;
1073 zip_entry->mode |= 0111;
1074 } else if ((zip_entry->mode & AE_IFMT) == 0) {
1075 zip_entry->mode |= AE_IFREG;
1076 }
1077 }
1078
1079 /* Make sure directories end in '/' */
1080 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) {
1081 wp = archive_entry_pathname_w(entry);
1082 if (wp != NULL) {
1083 len = wcslen(wp);
1084 if (len > 0 && wp[len - 1] != L'/') {
1085 struct archive_wstring s;
1086 archive_string_init(&s);
1087 archive_wstrcat(&s, wp);
1088 archive_wstrappend_wchar(&s, L'/');
1089 archive_entry_copy_pathname_w(entry, s.s);
1090 archive_wstring_free(&s);
1091 }
1092 } else {
1093 cp = archive_entry_pathname(entry);
1094 len = (cp != NULL)?strlen(cp):0;
1095 if (len > 0 && cp[len - 1] != '/') {
1096 struct archive_string s;
1097 archive_string_init(&s);
1098 archive_strcat(&s, cp);
1099 archive_strappend_char(&s, '/');
1100 archive_entry_set_pathname(entry, s.s);
1101 archive_string_free(&s);
1102 }
1103 }
1104 }
1105
1106 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) {
1107 /* If this came from the central dir, it's size info
1108 * is definitive, so ignore the length-at-end flag. */
1109 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END;
1110 /* If local header is missing a value, use the one from
1111 the central directory. If both have it, warn about
1112 mismatches. */
1113 if (zip_entry->crc32 == 0) {
1114 zip_entry->crc32 = zip_entry_central_dir.crc32;
1115 } else if (!zip->ignore_crc32
1116 && zip_entry->crc32 != zip_entry_central_dir.crc32) {
1117 archive_set_error(&a->archive,
1118 ARCHIVE_ERRNO_FILE_FORMAT,
1119 "Inconsistent CRC32 values");
1120 ret = ARCHIVE_WARN;
1121 }
1122 if (zip_entry->compressed_size == 0) {
1123 zip_entry->compressed_size
1124 = zip_entry_central_dir.compressed_size;
1125 } else if (zip_entry->compressed_size
1126 != zip_entry_central_dir.compressed_size) {
1127 archive_set_error(&a->archive,
1128 ARCHIVE_ERRNO_FILE_FORMAT,
1129 "Inconsistent compressed size: "
1130 "%jd in central directory, %jd in local header",
1131 (intmax_t)zip_entry_central_dir.compressed_size,
1132 (intmax_t)zip_entry->compressed_size);
1133 ret = ARCHIVE_WARN;
1134 }
1135 if (zip_entry->uncompressed_size == 0) {
1136 zip_entry->uncompressed_size
1137 = zip_entry_central_dir.uncompressed_size;
1138 } else if (zip_entry->uncompressed_size
1139 != zip_entry_central_dir.uncompressed_size) {
1140 archive_set_error(&a->archive,
1141 ARCHIVE_ERRNO_FILE_FORMAT,
1142 "Inconsistent uncompressed size: "
1143 "%jd in central directory, %jd in local header",
1144 (intmax_t)zip_entry_central_dir.uncompressed_size,
1145 (intmax_t)zip_entry->uncompressed_size);
1146 ret = ARCHIVE_WARN;
1147 }
1148 }
1149
1150 /* Populate some additional entry fields: */
1151 archive_entry_set_mode(entry, zip_entry->mode);
1152 archive_entry_set_uid(entry, zip_entry->uid);
1153 archive_entry_set_gid(entry, zip_entry->gid);
1154 archive_entry_set_mtime(entry, zip_entry->mtime, 0);
1155 archive_entry_set_ctime(entry, zip_entry->ctime, 0);
1156 archive_entry_set_atime(entry, zip_entry->atime, 0);
1157
1158 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) {
1159 size_t linkname_length;
1160
1161 if (zip_entry->compressed_size > 64 * 1024) {
1162 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1163 "Zip file with oversized link entry");
1164 return ARCHIVE_FATAL;
1165 }
1166
1167 linkname_length = (size_t)zip_entry->compressed_size;
1168
1169 archive_entry_set_size(entry, 0);
1170 p = __archive_read_ahead(a, linkname_length, NULL);
1171 if (p == NULL) {
1172 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1173 "Truncated Zip file");
1174 return ARCHIVE_FATAL;
1175 }
1176
1177 sconv = zip->sconv;
1178 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME))
1179 sconv = zip->sconv_utf8;
1180 if (sconv == NULL)
1181 sconv = zip->sconv_default;
1182 if (archive_entry_copy_symlink_l(entry, p, linkname_length,
1183 sconv) != 0) {
1184 if (errno != ENOMEM && sconv == zip->sconv_utf8 &&
1185 (zip->entry->zip_flags & ZIP_UTF8_NAME))
1186 archive_entry_copy_symlink_l(entry, p,
1187 linkname_length, NULL);
1188 if (errno == ENOMEM) {
1189 archive_set_error(&a->archive, ENOMEM,
1190 "Can't allocate memory for Symlink");
1191 return (ARCHIVE_FATAL);
1192 }
1193 /*
1194 * Since there is no character-set regulation for
1195 * symlink name, do not report the conversion error
1196 * in an automatic conversion.
1197 */
1198 if (sconv != zip->sconv_utf8 ||
1199 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) {
1200 archive_set_error(&a->archive,
1201 ARCHIVE_ERRNO_FILE_FORMAT,
1202 "Symlink cannot be converted "
1203 "from %s to current locale.",
1204 archive_string_conversion_charset_name(
1205 sconv));
1206 ret = ARCHIVE_WARN;
1207 }
1208 }
1209 zip_entry->uncompressed_size = zip_entry->compressed_size = 0;
1210
1211 if (__archive_read_consume(a, linkname_length) < 0) {
1212 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1213 "Read error skipping symlink target name");
1214 return ARCHIVE_FATAL;
1215 }
1216 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1217 || zip_entry->uncompressed_size > 0) {
1218 /* Set the size only if it's meaningful. */
1219 archive_entry_set_size(entry, zip_entry->uncompressed_size);
1220 }
1221 zip->entry_bytes_remaining = zip_entry->compressed_size;
1222
1223 /* If there's no body, force read_data() to return EOF immediately. */
1224 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1225 && zip->entry_bytes_remaining < 1)
1226 zip->end_of_entry = 1;
1227
1228 /* Set up a more descriptive format name. */
1229 archive_string_empty(&zip->format_name);
1230 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)",
1231 version / 10, version % 10,
1232 compression_name(zip->entry->compression));
1233 a->archive.archive_format_name = zip->format_name.s;
1234
1235 return (ret);
1236 }
1237
1238 static int
check_authentication_code(struct archive_read * a,const void * _p)1239 check_authentication_code(struct archive_read *a, const void *_p)
1240 {
1241 struct zip *zip = (struct zip *)(a->format->data);
1242
1243 /* Check authentication code. */
1244 if (zip->hctx_valid) {
1245 const void *p;
1246 uint8_t hmac[20];
1247 size_t hmac_len = 20;
1248 int cmp;
1249
1250 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len);
1251 if (_p == NULL) {
1252 /* Read authentication code. */
1253 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL);
1254 if (p == NULL) {
1255 archive_set_error(&a->archive,
1256 ARCHIVE_ERRNO_FILE_FORMAT,
1257 "Truncated ZIP file data");
1258 return (ARCHIVE_FATAL);
1259 }
1260 } else {
1261 p = _p;
1262 }
1263 cmp = memcmp(hmac, p, AUTH_CODE_SIZE);
1264 __archive_read_consume(a, AUTH_CODE_SIZE);
1265 if (cmp != 0) {
1266 archive_set_error(&a->archive,
1267 ARCHIVE_ERRNO_MISC,
1268 "ZIP bad Authentication code");
1269 return (ARCHIVE_WARN);
1270 }
1271 }
1272 return (ARCHIVE_OK);
1273 }
1274
1275 /*
1276 * Read "uncompressed" data. There are three cases:
1277 * 1) We know the size of the data. This is always true for the
1278 * seeking reader (we've examined the Central Directory already).
1279 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred.
1280 * Info-ZIP seems to do this; we know the size but have to grab
1281 * the CRC from the data descriptor afterwards.
1282 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and
1283 * we have no size information. In this case, we can do pretty
1284 * well by watching for the data descriptor record. The data
1285 * descriptor is 16 bytes and includes a computed CRC that should
1286 * provide a strong check.
1287 *
1288 * TODO: Technically, the PK\007\010 signature is optional.
1289 * In the original spec, the data descriptor contained CRC
1290 * and size fields but had no leading signature. In practice,
1291 * newer writers seem to provide the signature pretty consistently.
1292 *
1293 * For uncompressed data, the PK\007\010 marker seems essential
1294 * to be sure we've actually seen the end of the entry.
1295 *
1296 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets
1297 * zip->end_of_entry if it consumes all of the data.
1298 */
1299 static int
zip_read_data_none(struct archive_read * a,const void ** _buff,size_t * size,int64_t * offset)1300 zip_read_data_none(struct archive_read *a, const void **_buff,
1301 size_t *size, int64_t *offset)
1302 {
1303 struct zip *zip;
1304 const char *buff;
1305 ssize_t bytes_avail;
1306 int r;
1307
1308 (void)offset; /* UNUSED */
1309
1310 zip = (struct zip *)(a->format->data);
1311
1312 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) {
1313 const char *p;
1314 ssize_t grabbing_bytes = 24;
1315
1316 if (zip->hctx_valid)
1317 grabbing_bytes += AUTH_CODE_SIZE;
1318 /* Grab at least 24 bytes. */
1319 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail);
1320 if (bytes_avail < grabbing_bytes) {
1321 /* Zip archives have end-of-archive markers
1322 that are longer than this, so a failure to get at
1323 least 24 bytes really does indicate a truncated
1324 file. */
1325 archive_set_error(&a->archive,
1326 ARCHIVE_ERRNO_FILE_FORMAT,
1327 "Truncated ZIP file data");
1328 return (ARCHIVE_FATAL);
1329 }
1330 /* Check for a complete PK\007\010 signature, followed
1331 * by the correct 4-byte CRC. */
1332 p = buff;
1333 if (zip->hctx_valid)
1334 p += AUTH_CODE_SIZE;
1335 if (p[0] == 'P' && p[1] == 'K'
1336 && p[2] == '\007' && p[3] == '\010'
1337 && (archive_le32dec(p + 4) == zip->entry_crc32
1338 || zip->ignore_crc32
1339 || (zip->hctx_valid
1340 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) {
1341 if (zip->entry->flags & LA_USED_ZIP64) {
1342 uint64_t compressed, uncompressed;
1343 zip->entry->crc32 = archive_le32dec(p + 4);
1344 compressed = archive_le64dec(p + 8);
1345 uncompressed = archive_le64dec(p + 16);
1346 if (compressed > INT64_MAX || uncompressed >
1347 INT64_MAX) {
1348 archive_set_error(&a->archive,
1349 ARCHIVE_ERRNO_FILE_FORMAT,
1350 "Overflow of 64-bit file sizes");
1351 return ARCHIVE_FAILED;
1352 }
1353 zip->entry->compressed_size = compressed;
1354 zip->entry->uncompressed_size = uncompressed;
1355 zip->unconsumed = 24;
1356 } else {
1357 zip->entry->crc32 = archive_le32dec(p + 4);
1358 zip->entry->compressed_size =
1359 archive_le32dec(p + 8);
1360 zip->entry->uncompressed_size =
1361 archive_le32dec(p + 12);
1362 zip->unconsumed = 16;
1363 }
1364 if (zip->hctx_valid) {
1365 r = check_authentication_code(a, buff);
1366 if (r != ARCHIVE_OK)
1367 return (r);
1368 }
1369 zip->end_of_entry = 1;
1370 return (ARCHIVE_OK);
1371 }
1372 /* If not at EOF, ensure we consume at least one byte. */
1373 ++p;
1374
1375 /* Scan forward until we see where a PK\007\010 signature
1376 * might be. */
1377 /* Return bytes up until that point. On the next call,
1378 * the code above will verify the data descriptor. */
1379 while (p < buff + bytes_avail - 4) {
1380 if (p[3] == 'P') { p += 3; }
1381 else if (p[3] == 'K') { p += 2; }
1382 else if (p[3] == '\007') { p += 1; }
1383 else if (p[3] == '\010' && p[2] == '\007'
1384 && p[1] == 'K' && p[0] == 'P') {
1385 if (zip->hctx_valid)
1386 p -= AUTH_CODE_SIZE;
1387 break;
1388 } else { p += 4; }
1389 }
1390 bytes_avail = p - buff;
1391 } else {
1392 if (zip->entry_bytes_remaining == 0) {
1393 zip->end_of_entry = 1;
1394 if (zip->hctx_valid) {
1395 r = check_authentication_code(a, NULL);
1396 if (r != ARCHIVE_OK)
1397 return (r);
1398 }
1399 return (ARCHIVE_OK);
1400 }
1401 /* Grab a bunch of bytes. */
1402 buff = __archive_read_ahead(a, 1, &bytes_avail);
1403 if (bytes_avail <= 0) {
1404 archive_set_error(&a->archive,
1405 ARCHIVE_ERRNO_FILE_FORMAT,
1406 "Truncated ZIP file data");
1407 return (ARCHIVE_FATAL);
1408 }
1409 if (bytes_avail > zip->entry_bytes_remaining)
1410 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
1411 }
1412 if (zip->tctx_valid || zip->cctx_valid) {
1413 size_t dec_size = bytes_avail;
1414
1415 if (dec_size > zip->decrypted_buffer_size)
1416 dec_size = zip->decrypted_buffer_size;
1417 if (zip->tctx_valid) {
1418 trad_enc_decrypt_update(&zip->tctx,
1419 (const uint8_t *)buff, dec_size,
1420 zip->decrypted_buffer, dec_size);
1421 } else {
1422 size_t dsize = dec_size;
1423 archive_hmac_sha1_update(&zip->hctx,
1424 (const uint8_t *)buff, dec_size);
1425 archive_decrypto_aes_ctr_update(&zip->cctx,
1426 (const uint8_t *)buff, dec_size,
1427 zip->decrypted_buffer, &dsize);
1428 }
1429 bytes_avail = dec_size;
1430 buff = (const char *)zip->decrypted_buffer;
1431 }
1432 *size = bytes_avail;
1433 zip->entry_bytes_remaining -= bytes_avail;
1434 zip->entry_uncompressed_bytes_read += bytes_avail;
1435 zip->entry_compressed_bytes_read += bytes_avail;
1436 zip->unconsumed += bytes_avail;
1437 *_buff = buff;
1438 return (ARCHIVE_OK);
1439 }
1440
1441 static int
consume_optional_marker(struct archive_read * a,struct zip * zip)1442 consume_optional_marker(struct archive_read *a, struct zip *zip)
1443 {
1444 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
1445 const char *p;
1446
1447 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) {
1448 archive_set_error(&a->archive,
1449 ARCHIVE_ERRNO_FILE_FORMAT,
1450 "Truncated ZIP end-of-file record");
1451 return (ARCHIVE_FATAL);
1452 }
1453 /* Consume the optional PK\007\010 marker. */
1454 if (p[0] == 'P' && p[1] == 'K' &&
1455 p[2] == '\007' && p[3] == '\010') {
1456 p += 4;
1457 zip->unconsumed = 4;
1458 }
1459 if (zip->entry->flags & LA_USED_ZIP64) {
1460 uint64_t compressed, uncompressed;
1461 zip->entry->crc32 = archive_le32dec(p);
1462 compressed = archive_le64dec(p + 4);
1463 uncompressed = archive_le64dec(p + 12);
1464 if (compressed > INT64_MAX ||
1465 uncompressed > INT64_MAX) {
1466 archive_set_error(&a->archive,
1467 ARCHIVE_ERRNO_FILE_FORMAT,
1468 "Overflow of 64-bit file sizes");
1469 return ARCHIVE_FAILED;
1470 }
1471 zip->entry->compressed_size = compressed;
1472 zip->entry->uncompressed_size = uncompressed;
1473 zip->unconsumed += 20;
1474 } else {
1475 zip->entry->crc32 = archive_le32dec(p);
1476 zip->entry->compressed_size = archive_le32dec(p + 4);
1477 zip->entry->uncompressed_size = archive_le32dec(p + 8);
1478 zip->unconsumed += 12;
1479 }
1480 }
1481
1482 return (ARCHIVE_OK);
1483 }
1484
1485 #if HAVE_LZMA_H && HAVE_LIBLZMA
1486 static int
zipx_xz_init(struct archive_read * a,struct zip * zip)1487 zipx_xz_init(struct archive_read *a, struct zip *zip)
1488 {
1489 lzma_ret r;
1490
1491 if(zip->zipx_lzma_valid) {
1492 lzma_end(&zip->zipx_lzma_stream);
1493 zip->zipx_lzma_valid = 0;
1494 }
1495
1496 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream));
1497 r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0);
1498 if (r != LZMA_OK) {
1499 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
1500 "xz initialization failed(%d)",
1501 r);
1502
1503 return (ARCHIVE_FAILED);
1504 }
1505
1506 zip->zipx_lzma_valid = 1;
1507
1508 free(zip->uncompressed_buffer);
1509
1510 zip->uncompressed_buffer_size = 256 * 1024;
1511 zip->uncompressed_buffer =
1512 (uint8_t*) malloc(zip->uncompressed_buffer_size);
1513 if (zip->uncompressed_buffer == NULL) {
1514 archive_set_error(&a->archive, ENOMEM,
1515 "No memory for xz decompression");
1516 return (ARCHIVE_FATAL);
1517 }
1518
1519 zip->decompress_init = 1;
1520 return (ARCHIVE_OK);
1521 }
1522
1523 static int
zipx_lzma_alone_init(struct archive_read * a,struct zip * zip)1524 zipx_lzma_alone_init(struct archive_read *a, struct zip *zip)
1525 {
1526 lzma_ret r;
1527 const uint8_t* p;
1528
1529 #pragma pack(push)
1530 #pragma pack(1)
1531 struct _alone_header {
1532 uint8_t bytes[5];
1533 uint64_t uncompressed_size;
1534 } alone_header;
1535 #pragma pack(pop)
1536
1537 if(zip->zipx_lzma_valid) {
1538 lzma_end(&zip->zipx_lzma_stream);
1539 zip->zipx_lzma_valid = 0;
1540 }
1541
1542 /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma
1543 * that is a part of XZ Utils. The stream format stored inside ZIPX
1544 * file is a modified "lzma alone" file format, that was used by the
1545 * `lzma` utility which was later deprecated in favour of `xz` utility. * Since those formats are nearly the same, we can use a standard
1546 * "lzma alone" decoder from XZ Utils. */
1547
1548 memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream));
1549 r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX);
1550 if (r != LZMA_OK) {
1551 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
1552 "lzma initialization failed(%d)", r);
1553
1554 return (ARCHIVE_FAILED);
1555 }
1556
1557 /* Flag the cleanup function that we want our lzma-related structures
1558 * to be freed later. */
1559 zip->zipx_lzma_valid = 1;
1560
1561 /* The "lzma alone" file format and the stream format inside ZIPx are
1562 * almost the same. Here's an example of a structure of "lzma alone"
1563 * format:
1564 *
1565 * $ cat /bin/ls | lzma | xxd | head -n 1
1566 * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814
1567 *
1568 * 5 bytes 8 bytes n bytes
1569 * <lzma_params><uncompressed_size><data...>
1570 *
1571 * lzma_params is a 5-byte blob that has to be decoded to extract
1572 * parameters of this LZMA stream. The uncompressed_size field is an
1573 * uint64_t value that contains information about the size of the
1574 * uncompressed file, or UINT64_MAX if this value is unknown.
1575 * The <data...> part is the actual lzma-compressed data stream.
1576 *
1577 * Now here's the structure of the stream inside the ZIPX file:
1578 *
1579 * $ cat stream_inside_zipx | xxd | head -n 1
1580 * 00000000: 0914 0500 5d00 8000 0000 2814 .... ....
1581 *
1582 * 2byte 2byte 5 bytes n bytes
1583 * <magic1><magic2><lzma_params><data...>
1584 *
1585 * This means that the ZIPX file contains an additional magic1 and
1586 * magic2 headers, the lzma_params field contains the same parameter
1587 * set as in the "lzma alone" format, and the <data...> field is the
1588 * same as in the "lzma alone" format as well. Note that also the zipx
1589 * format is missing the uncompressed_size field.
1590 *
1591 * So, in order to use the "lzma alone" decoder for the zipx lzma
1592 * stream, we simply need to shuffle around some fields, prepare a new
1593 * lzma alone header, feed it into lzma alone decoder so it will
1594 * initialize itself properly, and then we can start feeding normal
1595 * zipx lzma stream into the decoder.
1596 */
1597
1598 /* Read magic1,magic2,lzma_params from the ZIPX stream. */
1599 if((p = __archive_read_ahead(a, 9, NULL)) == NULL) {
1600 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1601 "Truncated lzma data");
1602 return (ARCHIVE_FATAL);
1603 }
1604
1605 if(p[2] != 0x05 || p[3] != 0x00) {
1606 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1607 "Invalid lzma data");
1608 return (ARCHIVE_FATAL);
1609 }
1610
1611 /* Prepare an lzma alone header: copy the lzma_params blob into
1612 * a proper place into the lzma alone header. */
1613 memcpy(&alone_header.bytes[0], p + 4, 5);
1614
1615 /* Initialize the 'uncompressed size' field to unknown; we'll manually
1616 * monitor how many bytes there are still to be uncompressed. */
1617 alone_header.uncompressed_size = UINT64_MAX;
1618
1619 if(!zip->uncompressed_buffer) {
1620 zip->uncompressed_buffer_size = 256 * 1024;
1621 zip->uncompressed_buffer =
1622 (uint8_t*) malloc(zip->uncompressed_buffer_size);
1623
1624 if (zip->uncompressed_buffer == NULL) {
1625 archive_set_error(&a->archive, ENOMEM,
1626 "No memory for lzma decompression");
1627 return (ARCHIVE_FATAL);
1628 }
1629 }
1630
1631 zip->zipx_lzma_stream.next_in = (void*) &alone_header;
1632 zip->zipx_lzma_stream.avail_in = sizeof(alone_header);
1633 zip->zipx_lzma_stream.total_in = 0;
1634 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1635 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size;
1636 zip->zipx_lzma_stream.total_out = 0;
1637
1638 /* Feed only the header into the lzma alone decoder. This will
1639 * effectively initialize the decoder, and will not produce any
1640 * output bytes yet. */
1641 r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1642 if (r != LZMA_OK) {
1643 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1644 "lzma stream initialization error");
1645 return ARCHIVE_FATAL;
1646 }
1647
1648 /* We've already consumed some bytes, so take this into account. */
1649 __archive_read_consume(a, 9);
1650 zip->entry_bytes_remaining -= 9;
1651 zip->entry_compressed_bytes_read += 9;
1652
1653 zip->decompress_init = 1;
1654 return (ARCHIVE_OK);
1655 }
1656
1657 static int
zip_read_data_zipx_xz(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1658 zip_read_data_zipx_xz(struct archive_read *a, const void **buff,
1659 size_t *size, int64_t *offset)
1660 {
1661 struct zip* zip = (struct zip *)(a->format->data);
1662 int ret;
1663 lzma_ret lz_ret;
1664 const void* compressed_buf;
1665 ssize_t bytes_avail, in_bytes, to_consume = 0;
1666
1667 (void) offset; /* UNUSED */
1668
1669 /* Initialize decompressor if not yet initialized. */
1670 if (!zip->decompress_init) {
1671 ret = zipx_xz_init(a, zip);
1672 if (ret != ARCHIVE_OK)
1673 return (ret);
1674 }
1675
1676 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail);
1677 if (bytes_avail < 0) {
1678 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1679 "Truncated xz file body");
1680 return (ARCHIVE_FATAL);
1681 }
1682
1683 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail);
1684 zip->zipx_lzma_stream.next_in = compressed_buf;
1685 zip->zipx_lzma_stream.avail_in = in_bytes;
1686 zip->zipx_lzma_stream.total_in = 0;
1687 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1688 zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size;
1689 zip->zipx_lzma_stream.total_out = 0;
1690
1691 /* Perform the decompression. */
1692 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1693 switch(lz_ret) {
1694 case LZMA_DATA_ERROR:
1695 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1696 "xz data error (error %d)", (int) lz_ret);
1697 return (ARCHIVE_FATAL);
1698
1699 case LZMA_NO_CHECK:
1700 case LZMA_OK:
1701 break;
1702
1703 default:
1704 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1705 "xz unknown error %d", (int) lz_ret);
1706 return (ARCHIVE_FATAL);
1707
1708 case LZMA_STREAM_END:
1709 lzma_end(&zip->zipx_lzma_stream);
1710 zip->zipx_lzma_valid = 0;
1711
1712 if((int64_t) zip->zipx_lzma_stream.total_in !=
1713 zip->entry_bytes_remaining)
1714 {
1715 archive_set_error(&a->archive,
1716 ARCHIVE_ERRNO_MISC,
1717 "xz premature end of stream");
1718 return (ARCHIVE_FATAL);
1719 }
1720
1721 zip->end_of_entry = 1;
1722 break;
1723 }
1724
1725 to_consume = zip->zipx_lzma_stream.total_in;
1726
1727 __archive_read_consume(a, to_consume);
1728 zip->entry_bytes_remaining -= to_consume;
1729 zip->entry_compressed_bytes_read += to_consume;
1730 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out;
1731
1732 *size = zip->zipx_lzma_stream.total_out;
1733 *buff = zip->uncompressed_buffer;
1734
1735 ret = consume_optional_marker(a, zip);
1736 if (ret != ARCHIVE_OK)
1737 return (ret);
1738
1739 return (ARCHIVE_OK);
1740 }
1741
1742 static int
zip_read_data_zipx_lzma_alone(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1743 zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff,
1744 size_t *size, int64_t *offset)
1745 {
1746 struct zip* zip = (struct zip *)(a->format->data);
1747 int ret;
1748 lzma_ret lz_ret;
1749 const void* compressed_buf;
1750 ssize_t bytes_avail, in_bytes, to_consume;
1751
1752 (void) offset; /* UNUSED */
1753
1754 /* Initialize decompressor if not yet initialized. */
1755 if (!zip->decompress_init) {
1756 ret = zipx_lzma_alone_init(a, zip);
1757 if (ret != ARCHIVE_OK)
1758 return (ret);
1759 }
1760
1761 /* Fetch more compressed data. The same note as in deflate handler
1762 * applies here as well:
1763 *
1764 * Note: '1' here is a performance optimization. Recall that the
1765 * decompression layer returns a count of available bytes; asking for
1766 * more than that forces the decompressor to combine reads by copying
1767 * data.
1768 */
1769 compressed_buf = __archive_read_ahead(a, 1, &bytes_avail);
1770 if (bytes_avail < 0) {
1771 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1772 "Truncated lzma file body");
1773 return (ARCHIVE_FATAL);
1774 }
1775
1776 /* Set decompressor parameters. */
1777 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail);
1778
1779 zip->zipx_lzma_stream.next_in = compressed_buf;
1780 zip->zipx_lzma_stream.avail_in = in_bytes;
1781 zip->zipx_lzma_stream.total_in = 0;
1782 zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer;
1783 zip->zipx_lzma_stream.avail_out =
1784 /* These lzma_alone streams lack end of stream marker, so let's
1785 * make sure the unpacker won't try to unpack more than it's
1786 * supposed to. */
1787 zipmin((int64_t) zip->uncompressed_buffer_size,
1788 zip->entry->uncompressed_size -
1789 zip->entry_uncompressed_bytes_read);
1790 zip->zipx_lzma_stream.total_out = 0;
1791
1792 /* Perform the decompression. */
1793 lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN);
1794 switch(lz_ret) {
1795 case LZMA_DATA_ERROR:
1796 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1797 "lzma data error (error %d)", (int) lz_ret);
1798 return (ARCHIVE_FATAL);
1799
1800 case LZMA_OK:
1801 break;
1802
1803 default:
1804 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1805 "lzma unknown error %d", (int) lz_ret);
1806 return (ARCHIVE_FATAL);
1807 }
1808
1809 to_consume = zip->zipx_lzma_stream.total_in;
1810
1811 /* Update pointers. */
1812 __archive_read_consume(a, to_consume);
1813 zip->entry_bytes_remaining -= to_consume;
1814 zip->entry_compressed_bytes_read += to_consume;
1815 zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out;
1816
1817 if(zip->entry_bytes_remaining == 0) {
1818 zip->end_of_entry = 1;
1819 }
1820
1821 /* Return values. */
1822 *size = zip->zipx_lzma_stream.total_out;
1823 *buff = zip->uncompressed_buffer;
1824
1825 /* Behave the same way as during deflate decompression. */
1826 ret = consume_optional_marker(a, zip);
1827 if (ret != ARCHIVE_OK)
1828 return (ret);
1829
1830 /* Free lzma decoder handle because we'll no longer need it. */
1831 if(zip->end_of_entry) {
1832 lzma_end(&zip->zipx_lzma_stream);
1833 zip->zipx_lzma_valid = 0;
1834 }
1835
1836 /* If we're here, then we're good! */
1837 return (ARCHIVE_OK);
1838 }
1839 #endif /* HAVE_LZMA_H && HAVE_LIBLZMA */
1840
1841 static int
zipx_ppmd8_init(struct archive_read * a,struct zip * zip)1842 zipx_ppmd8_init(struct archive_read *a, struct zip *zip)
1843 {
1844 const void* p;
1845 uint32_t val;
1846 uint32_t order;
1847 uint32_t mem;
1848 uint32_t restore_method;
1849
1850 /* Remove previous decompression context if it exists. */
1851 if(zip->ppmd8_valid) {
1852 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
1853 zip->ppmd8_valid = 0;
1854 }
1855
1856 /* Create a new decompression context. */
1857 __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8);
1858 zip->ppmd8_stream_failed = 0;
1859
1860 /* Setup function pointers required by Ppmd8 decompressor. The
1861 * 'ppmd_read' function will feed new bytes to the decompressor,
1862 * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */
1863 zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream;
1864 zip->zipx_ppmd_stream.a = a;
1865 zip->zipx_ppmd_stream.Read = &ppmd_read;
1866
1867 /* Reset number of read bytes to 0. */
1868 zip->zipx_ppmd_read_compressed = 0;
1869
1870 /* Read Ppmd8 header (2 bytes). */
1871 p = __archive_read_ahead(a, 2, NULL);
1872 if(!p) {
1873 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1874 "Truncated file data in PPMd8 stream");
1875 return (ARCHIVE_FATAL);
1876 }
1877 __archive_read_consume(a, 2);
1878
1879 /* Decode the stream's compression parameters. */
1880 val = archive_le16dec(p);
1881 order = (val & 15) + 1;
1882 mem = ((val >> 4) & 0xff) + 1;
1883 restore_method = (val >> 12);
1884
1885 if(order < 2 || restore_method > 2) {
1886 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1887 "Invalid parameter set in PPMd8 stream (order=%d, "
1888 "restore=%d)", order, restore_method);
1889 return (ARCHIVE_FAILED);
1890 }
1891
1892 /* Allocate the memory needed to properly decompress the file. */
1893 if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) {
1894 archive_set_error(&a->archive, ENOMEM,
1895 "Unable to allocate memory for PPMd8 stream: %d bytes",
1896 mem << 20);
1897 return (ARCHIVE_FATAL);
1898 }
1899
1900 /* Signal the cleanup function to release Ppmd8 context in the
1901 * cleanup phase. */
1902 zip->ppmd8_valid = 1;
1903
1904 /* Perform further Ppmd8 initialization. */
1905 if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) {
1906 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1907 "PPMd8 stream range decoder initialization error");
1908 return (ARCHIVE_FATAL);
1909 }
1910
1911 __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order,
1912 restore_method);
1913
1914 /* Allocate the buffer that will hold uncompressed data. */
1915 free(zip->uncompressed_buffer);
1916
1917 zip->uncompressed_buffer_size = 256 * 1024;
1918 zip->uncompressed_buffer =
1919 (uint8_t*) malloc(zip->uncompressed_buffer_size);
1920
1921 if(zip->uncompressed_buffer == NULL) {
1922 archive_set_error(&a->archive, ENOMEM,
1923 "No memory for PPMd8 decompression");
1924 return ARCHIVE_FATAL;
1925 }
1926
1927 /* Ppmd8 initialization is done. */
1928 zip->decompress_init = 1;
1929
1930 /* We've already read 2 bytes in the output stream. Additionally,
1931 * Ppmd8 initialization code could read some data as well. So we
1932 * are advancing the stream by 2 bytes plus whatever number of
1933 * bytes Ppmd8 init function used. */
1934 zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed;
1935
1936 return ARCHIVE_OK;
1937 }
1938
1939 static int
zip_read_data_zipx_ppmd(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1940 zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff,
1941 size_t *size, int64_t *offset)
1942 {
1943 struct zip* zip = (struct zip *)(a->format->data);
1944 int ret;
1945 size_t consumed_bytes = 0;
1946 ssize_t bytes_avail = 0;
1947
1948 (void) offset; /* UNUSED */
1949
1950 /* If we're here for the first time, initialize Ppmd8 decompression
1951 * context first. */
1952 if(!zip->decompress_init) {
1953 ret = zipx_ppmd8_init(a, zip);
1954 if(ret != ARCHIVE_OK)
1955 return ret;
1956 }
1957
1958 /* Fetch for more data. We're reading 1 byte here, but libarchive
1959 * should prefetch more bytes. */
1960 (void) __archive_read_ahead(a, 1, &bytes_avail);
1961 if(bytes_avail < 0) {
1962 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1963 "Truncated PPMd8 file body");
1964 return (ARCHIVE_FATAL);
1965 }
1966
1967 /* This counter will be updated inside ppmd_read(), which at one
1968 * point will be called by Ppmd8_DecodeSymbol. */
1969 zip->zipx_ppmd_read_compressed = 0;
1970
1971 /* Decompression loop. */
1972 do {
1973 int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol(
1974 &zip->ppmd8);
1975 if(sym < 0) {
1976 zip->end_of_entry = 1;
1977 break;
1978 }
1979
1980 /* This field is set by ppmd_read() when there was no more data
1981 * to be read. */
1982 if(zip->ppmd8_stream_failed) {
1983 archive_set_error(&a->archive,
1984 ARCHIVE_ERRNO_FILE_FORMAT,
1985 "Truncated PPMd8 file body");
1986 return (ARCHIVE_FATAL);
1987 }
1988
1989 zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym;
1990 ++consumed_bytes;
1991 } while(consumed_bytes < zip->uncompressed_buffer_size);
1992
1993 /* Update pointers for libarchive. */
1994 *buff = zip->uncompressed_buffer;
1995 *size = consumed_bytes;
1996
1997 /* Update pointers so we can continue decompression in another call. */
1998 zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed;
1999 zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed;
2000 zip->entry_uncompressed_bytes_read += consumed_bytes;
2001
2002 /* If we're at the end of stream, deinitialize Ppmd8 context. */
2003 if(zip->end_of_entry) {
2004 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
2005 zip->ppmd8_valid = 0;
2006 }
2007
2008 /* Seek for optional marker, same way as in each zip entry. */
2009 ret = consume_optional_marker(a, zip);
2010 if (ret != ARCHIVE_OK)
2011 return ret;
2012
2013 return ARCHIVE_OK;
2014 }
2015
2016 #ifdef HAVE_BZLIB_H
2017 static int
zipx_bzip2_init(struct archive_read * a,struct zip * zip)2018 zipx_bzip2_init(struct archive_read *a, struct zip *zip)
2019 {
2020 int r;
2021
2022 /* Deallocate already existing BZ2 decompression context if it
2023 * exists. */
2024 if(zip->bzstream_valid) {
2025 BZ2_bzDecompressEnd(&zip->bzstream);
2026 zip->bzstream_valid = 0;
2027 }
2028
2029 /* Allocate a new BZ2 decompression context. */
2030 memset(&zip->bzstream, 0, sizeof(bz_stream));
2031 r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1);
2032 if(r != BZ_OK) {
2033 archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC,
2034 "bzip2 initialization failed(%d)",
2035 r);
2036
2037 return ARCHIVE_FAILED;
2038 }
2039
2040 /* Mark the bzstream field to be released in cleanup phase. */
2041 zip->bzstream_valid = 1;
2042
2043 /* (Re)allocate the buffer that will contain decompressed bytes. */
2044 free(zip->uncompressed_buffer);
2045
2046 zip->uncompressed_buffer_size = 256 * 1024;
2047 zip->uncompressed_buffer =
2048 (uint8_t*) malloc(zip->uncompressed_buffer_size);
2049 if (zip->uncompressed_buffer == NULL) {
2050 archive_set_error(&a->archive, ENOMEM,
2051 "No memory for bzip2 decompression");
2052 return ARCHIVE_FATAL;
2053 }
2054
2055 /* Initialization done. */
2056 zip->decompress_init = 1;
2057 return ARCHIVE_OK;
2058 }
2059
2060 static int
zip_read_data_zipx_bzip2(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2061 zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff,
2062 size_t *size, int64_t *offset)
2063 {
2064 struct zip *zip = (struct zip *)(a->format->data);
2065 ssize_t bytes_avail = 0, in_bytes, to_consume;
2066 const void *compressed_buff;
2067 int r;
2068 uint64_t total_out;
2069
2070 (void) offset; /* UNUSED */
2071
2072 /* Initialize decompression context if we're here for the first time. */
2073 if(!zip->decompress_init) {
2074 r = zipx_bzip2_init(a, zip);
2075 if(r != ARCHIVE_OK)
2076 return r;
2077 }
2078
2079 /* Fetch more compressed bytes. */
2080 compressed_buff = __archive_read_ahead(a, 1, &bytes_avail);
2081 if(bytes_avail < 0) {
2082 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2083 "Truncated bzip2 file body");
2084 return (ARCHIVE_FATAL);
2085 }
2086
2087 in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail);
2088 if(in_bytes < 1) {
2089 /* libbz2 doesn't complain when caller feeds avail_in == 0.
2090 * It will actually return success in this case, which is
2091 * undesirable. This is why we need to make this check
2092 * manually. */
2093
2094 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2095 "Truncated bzip2 file body");
2096 return (ARCHIVE_FATAL);
2097 }
2098
2099 /* Setup buffer boundaries. */
2100 zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff;
2101 zip->bzstream.avail_in = in_bytes;
2102 zip->bzstream.total_in_hi32 = 0;
2103 zip->bzstream.total_in_lo32 = 0;
2104 zip->bzstream.next_out = (char*) zip->uncompressed_buffer;
2105 zip->bzstream.avail_out = zip->uncompressed_buffer_size;
2106 zip->bzstream.total_out_hi32 = 0;
2107 zip->bzstream.total_out_lo32 = 0;
2108
2109 /* Perform the decompression. */
2110 r = BZ2_bzDecompress(&zip->bzstream);
2111 switch(r) {
2112 case BZ_STREAM_END:
2113 /* If we're at the end of the stream, deinitialize the
2114 * decompression context now. */
2115 switch(BZ2_bzDecompressEnd(&zip->bzstream)) {
2116 case BZ_OK:
2117 break;
2118 default:
2119 archive_set_error(&a->archive,
2120 ARCHIVE_ERRNO_MISC,
2121 "Failed to clean up bzip2 "
2122 "decompressor");
2123 return ARCHIVE_FATAL;
2124 }
2125
2126 zip->end_of_entry = 1;
2127 break;
2128 case BZ_OK:
2129 /* The decompressor has successfully decoded this
2130 * chunk of data, but more data is still in queue. */
2131 break;
2132 default:
2133 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2134 "bzip2 decompression failed");
2135 return ARCHIVE_FATAL;
2136 }
2137
2138 /* Update the pointers so decompressor can continue decoding. */
2139 to_consume = zip->bzstream.total_in_lo32;
2140 __archive_read_consume(a, to_consume);
2141
2142 total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) +
2143 zip->bzstream.total_out_lo32;
2144
2145 zip->entry_bytes_remaining -= to_consume;
2146 zip->entry_compressed_bytes_read += to_consume;
2147 zip->entry_uncompressed_bytes_read += total_out;
2148
2149 /* Give libarchive its due. */
2150 *size = total_out;
2151 *buff = zip->uncompressed_buffer;
2152
2153 /* Seek for optional marker, like in other entries. */
2154 r = consume_optional_marker(a, zip);
2155 if(r != ARCHIVE_OK)
2156 return r;
2157
2158 return ARCHIVE_OK;
2159 }
2160
2161 #endif
2162
2163 #ifdef HAVE_ZLIB_H
2164 static int
zip_deflate_init(struct archive_read * a,struct zip * zip)2165 zip_deflate_init(struct archive_read *a, struct zip *zip)
2166 {
2167 int r;
2168
2169 /* If we haven't yet read any data, initialize the decompressor. */
2170 if (!zip->decompress_init) {
2171 if (zip->stream_valid)
2172 r = inflateReset(&zip->stream);
2173 else
2174 r = inflateInit2(&zip->stream,
2175 -15 /* Don't check for zlib header */);
2176 if (r != Z_OK) {
2177 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2178 "Can't initialize ZIP decompression.");
2179 return (ARCHIVE_FATAL);
2180 }
2181 /* Stream structure has been set up. */
2182 zip->stream_valid = 1;
2183 /* We've initialized decompression for this stream. */
2184 zip->decompress_init = 1;
2185 }
2186 return (ARCHIVE_OK);
2187 }
2188
2189 static int
zip_read_data_deflate(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2190 zip_read_data_deflate(struct archive_read *a, const void **buff,
2191 size_t *size, int64_t *offset)
2192 {
2193 struct zip *zip;
2194 ssize_t bytes_avail;
2195 const void *compressed_buff, *sp;
2196 int r;
2197
2198 (void)offset; /* UNUSED */
2199
2200 zip = (struct zip *)(a->format->data);
2201
2202 /* If the buffer hasn't been allocated, allocate it now. */
2203 if (zip->uncompressed_buffer == NULL) {
2204 zip->uncompressed_buffer_size = 256 * 1024;
2205 zip->uncompressed_buffer
2206 = (unsigned char *)malloc(zip->uncompressed_buffer_size);
2207 if (zip->uncompressed_buffer == NULL) {
2208 archive_set_error(&a->archive, ENOMEM,
2209 "No memory for ZIP decompression");
2210 return (ARCHIVE_FATAL);
2211 }
2212 }
2213
2214 r = zip_deflate_init(a, zip);
2215 if (r != ARCHIVE_OK)
2216 return (r);
2217
2218 /*
2219 * Note: '1' here is a performance optimization.
2220 * Recall that the decompression layer returns a count of
2221 * available bytes; asking for more than that forces the
2222 * decompressor to combine reads by copying data.
2223 */
2224 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail);
2225 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2226 && bytes_avail > zip->entry_bytes_remaining) {
2227 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
2228 }
2229 if (bytes_avail < 0) {
2230 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2231 "Truncated ZIP file body");
2232 return (ARCHIVE_FATAL);
2233 }
2234
2235 if (zip->tctx_valid || zip->cctx_valid) {
2236 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) {
2237 size_t buff_remaining =
2238 (zip->decrypted_buffer +
2239 zip->decrypted_buffer_size)
2240 - (zip->decrypted_ptr +
2241 zip->decrypted_bytes_remaining);
2242
2243 if (buff_remaining > (size_t)bytes_avail)
2244 buff_remaining = (size_t)bytes_avail;
2245
2246 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) &&
2247 zip->entry_bytes_remaining > 0) {
2248 if ((int64_t)(zip->decrypted_bytes_remaining
2249 + buff_remaining)
2250 > zip->entry_bytes_remaining) {
2251 if (zip->entry_bytes_remaining <
2252 (int64_t)zip->decrypted_bytes_remaining)
2253 buff_remaining = 0;
2254 else
2255 buff_remaining =
2256 (size_t)zip->entry_bytes_remaining
2257 - zip->decrypted_bytes_remaining;
2258 }
2259 }
2260 if (buff_remaining > 0) {
2261 if (zip->tctx_valid) {
2262 trad_enc_decrypt_update(&zip->tctx,
2263 compressed_buff, buff_remaining,
2264 zip->decrypted_ptr
2265 + zip->decrypted_bytes_remaining,
2266 buff_remaining);
2267 } else {
2268 size_t dsize = buff_remaining;
2269 archive_decrypto_aes_ctr_update(
2270 &zip->cctx,
2271 compressed_buff, buff_remaining,
2272 zip->decrypted_ptr
2273 + zip->decrypted_bytes_remaining,
2274 &dsize);
2275 }
2276 zip->decrypted_bytes_remaining +=
2277 buff_remaining;
2278 }
2279 }
2280 bytes_avail = zip->decrypted_bytes_remaining;
2281 compressed_buff = (const char *)zip->decrypted_ptr;
2282 }
2283
2284 /*
2285 * A bug in zlib.h: stream.next_in should be marked 'const'
2286 * but isn't (the library never alters data through the
2287 * next_in pointer, only reads it). The result: this ugly
2288 * cast to remove 'const'.
2289 */
2290 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff;
2291 zip->stream.avail_in = (uInt)bytes_avail;
2292 zip->stream.total_in = 0;
2293 zip->stream.next_out = zip->uncompressed_buffer;
2294 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size;
2295 zip->stream.total_out = 0;
2296
2297 r = inflate(&zip->stream, 0);
2298 switch (r) {
2299 case Z_OK:
2300 break;
2301 case Z_STREAM_END:
2302 zip->end_of_entry = 1;
2303 break;
2304 case Z_MEM_ERROR:
2305 archive_set_error(&a->archive, ENOMEM,
2306 "Out of memory for ZIP decompression");
2307 return (ARCHIVE_FATAL);
2308 default:
2309 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2310 "ZIP decompression failed (%d)", r);
2311 return (ARCHIVE_FATAL);
2312 }
2313
2314 /* Consume as much as the compressor actually used. */
2315 bytes_avail = zip->stream.total_in;
2316 if (zip->tctx_valid || zip->cctx_valid) {
2317 zip->decrypted_bytes_remaining -= bytes_avail;
2318 if (zip->decrypted_bytes_remaining == 0)
2319 zip->decrypted_ptr = zip->decrypted_buffer;
2320 else
2321 zip->decrypted_ptr += bytes_avail;
2322 }
2323 /* Calculate compressed data as much as we used.*/
2324 if (zip->hctx_valid)
2325 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail);
2326 __archive_read_consume(a, bytes_avail);
2327 zip->entry_bytes_remaining -= bytes_avail;
2328 zip->entry_compressed_bytes_read += bytes_avail;
2329
2330 *size = zip->stream.total_out;
2331 zip->entry_uncompressed_bytes_read += zip->stream.total_out;
2332 *buff = zip->uncompressed_buffer;
2333
2334 if (zip->end_of_entry && zip->hctx_valid) {
2335 r = check_authentication_code(a, NULL);
2336 if (r != ARCHIVE_OK)
2337 return (r);
2338 }
2339
2340 r = consume_optional_marker(a, zip);
2341 if (r != ARCHIVE_OK)
2342 return (r);
2343
2344 return (ARCHIVE_OK);
2345 }
2346 #endif
2347
2348 static int
read_decryption_header(struct archive_read * a)2349 read_decryption_header(struct archive_read *a)
2350 {
2351 struct zip *zip = (struct zip *)(a->format->data);
2352 const char *p;
2353 unsigned int remaining_size;
2354 unsigned int ts;
2355
2356 /*
2357 * Read an initialization vector data field.
2358 */
2359 p = __archive_read_ahead(a, 2, NULL);
2360 if (p == NULL)
2361 goto truncated;
2362 ts = zip->iv_size;
2363 zip->iv_size = archive_le16dec(p);
2364 __archive_read_consume(a, 2);
2365 if (ts < zip->iv_size) {
2366 free(zip->iv);
2367 zip->iv = NULL;
2368 }
2369 p = __archive_read_ahead(a, zip->iv_size, NULL);
2370 if (p == NULL)
2371 goto truncated;
2372 if (zip->iv == NULL) {
2373 zip->iv = malloc(zip->iv_size);
2374 if (zip->iv == NULL)
2375 goto nomem;
2376 }
2377 memcpy(zip->iv, p, zip->iv_size);
2378 __archive_read_consume(a, zip->iv_size);
2379
2380 /*
2381 * Read a size of remaining decryption header field.
2382 */
2383 p = __archive_read_ahead(a, 14, NULL);
2384 if (p == NULL)
2385 goto truncated;
2386 remaining_size = archive_le32dec(p);
2387 if (remaining_size < 16 || remaining_size > (1 << 18))
2388 goto corrupted;
2389
2390 /* Check if format version is supported. */
2391 if (archive_le16dec(p+4) != 3) {
2392 archive_set_error(&a->archive,
2393 ARCHIVE_ERRNO_FILE_FORMAT,
2394 "Unsupported encryption format version: %u",
2395 archive_le16dec(p+4));
2396 return (ARCHIVE_FAILED);
2397 }
2398
2399 /*
2400 * Read an encryption algorithm field.
2401 */
2402 zip->alg_id = archive_le16dec(p+6);
2403 switch (zip->alg_id) {
2404 case 0x6601:/* DES */
2405 case 0x6602:/* RC2 */
2406 case 0x6603:/* 3DES 168 */
2407 case 0x6609:/* 3DES 112 */
2408 case 0x660E:/* AES 128 */
2409 case 0x660F:/* AES 192 */
2410 case 0x6610:/* AES 256 */
2411 case 0x6702:/* RC2 (version >= 5.2) */
2412 case 0x6720:/* Blowfish */
2413 case 0x6721:/* Twofish */
2414 case 0x6801:/* RC4 */
2415 /* Supported encryption algorithm. */
2416 break;
2417 default:
2418 archive_set_error(&a->archive,
2419 ARCHIVE_ERRNO_FILE_FORMAT,
2420 "Unknown encryption algorithm: %u", zip->alg_id);
2421 return (ARCHIVE_FAILED);
2422 }
2423
2424 /*
2425 * Read a bit length field.
2426 */
2427 zip->bit_len = archive_le16dec(p+8);
2428
2429 /*
2430 * Read a flags field.
2431 */
2432 zip->flags = archive_le16dec(p+10);
2433 switch (zip->flags & 0xf000) {
2434 case 0x0001: /* Password is required to decrypt. */
2435 case 0x0002: /* Certificates only. */
2436 case 0x0003: /* Password or certificate required to decrypt. */
2437 break;
2438 default:
2439 archive_set_error(&a->archive,
2440 ARCHIVE_ERRNO_FILE_FORMAT,
2441 "Unknown encryption flag: %u", zip->flags);
2442 return (ARCHIVE_FAILED);
2443 }
2444 if ((zip->flags & 0xf000) == 0 ||
2445 (zip->flags & 0xf000) == 0x4000) {
2446 archive_set_error(&a->archive,
2447 ARCHIVE_ERRNO_FILE_FORMAT,
2448 "Unknown encryption flag: %u", zip->flags);
2449 return (ARCHIVE_FAILED);
2450 }
2451
2452 /*
2453 * Read an encrypted random data field.
2454 */
2455 ts = zip->erd_size;
2456 zip->erd_size = archive_le16dec(p+12);
2457 __archive_read_consume(a, 14);
2458 if ((zip->erd_size & 0xf) != 0 ||
2459 (zip->erd_size + 16) > remaining_size ||
2460 (zip->erd_size + 16) < zip->erd_size)
2461 goto corrupted;
2462
2463 if (ts < zip->erd_size) {
2464 free(zip->erd);
2465 zip->erd = NULL;
2466 }
2467 p = __archive_read_ahead(a, zip->erd_size, NULL);
2468 if (p == NULL)
2469 goto truncated;
2470 if (zip->erd == NULL) {
2471 zip->erd = malloc(zip->erd_size);
2472 if (zip->erd == NULL)
2473 goto nomem;
2474 }
2475 memcpy(zip->erd, p, zip->erd_size);
2476 __archive_read_consume(a, zip->erd_size);
2477
2478 /*
2479 * Read a reserved data field.
2480 */
2481 p = __archive_read_ahead(a, 4, NULL);
2482 if (p == NULL)
2483 goto truncated;
2484 /* Reserved data size should be zero. */
2485 if (archive_le32dec(p) != 0)
2486 goto corrupted;
2487 __archive_read_consume(a, 4);
2488
2489 /*
2490 * Read a password validation data field.
2491 */
2492 p = __archive_read_ahead(a, 2, NULL);
2493 if (p == NULL)
2494 goto truncated;
2495 ts = zip->v_size;
2496 zip->v_size = archive_le16dec(p);
2497 __archive_read_consume(a, 2);
2498 if ((zip->v_size & 0x0f) != 0 ||
2499 (zip->erd_size + zip->v_size + 16) > remaining_size ||
2500 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size))
2501 goto corrupted;
2502 if (ts < zip->v_size) {
2503 free(zip->v_data);
2504 zip->v_data = NULL;
2505 }
2506 p = __archive_read_ahead(a, zip->v_size, NULL);
2507 if (p == NULL)
2508 goto truncated;
2509 if (zip->v_data == NULL) {
2510 zip->v_data = malloc(zip->v_size);
2511 if (zip->v_data == NULL)
2512 goto nomem;
2513 }
2514 memcpy(zip->v_data, p, zip->v_size);
2515 __archive_read_consume(a, zip->v_size);
2516
2517 p = __archive_read_ahead(a, 4, NULL);
2518 if (p == NULL)
2519 goto truncated;
2520 zip->v_crc32 = archive_le32dec(p);
2521 __archive_read_consume(a, 4);
2522
2523 /*return (ARCHIVE_OK);
2524 * This is not fully implemented yet.*/
2525 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2526 "Encrypted file is unsupported");
2527 return (ARCHIVE_FAILED);
2528 truncated:
2529 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2530 "Truncated ZIP file data");
2531 return (ARCHIVE_FATAL);
2532 corrupted:
2533 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2534 "Corrupted ZIP file data");
2535 return (ARCHIVE_FATAL);
2536 nomem:
2537 archive_set_error(&a->archive, ENOMEM,
2538 "No memory for ZIP decryption");
2539 return (ARCHIVE_FATAL);
2540 }
2541
2542 static int
zip_alloc_decryption_buffer(struct archive_read * a)2543 zip_alloc_decryption_buffer(struct archive_read *a)
2544 {
2545 struct zip *zip = (struct zip *)(a->format->data);
2546 size_t bs = 256 * 1024;
2547
2548 if (zip->decrypted_buffer == NULL) {
2549 zip->decrypted_buffer_size = bs;
2550 zip->decrypted_buffer = malloc(bs);
2551 if (zip->decrypted_buffer == NULL) {
2552 archive_set_error(&a->archive, ENOMEM,
2553 "No memory for ZIP decryption");
2554 return (ARCHIVE_FATAL);
2555 }
2556 }
2557 zip->decrypted_ptr = zip->decrypted_buffer;
2558 return (ARCHIVE_OK);
2559 }
2560
2561 static int
init_traditional_PKWARE_decryption(struct archive_read * a)2562 init_traditional_PKWARE_decryption(struct archive_read *a)
2563 {
2564 struct zip *zip = (struct zip *)(a->format->data);
2565 const void *p;
2566 int retry;
2567 int r;
2568
2569 if (zip->tctx_valid)
2570 return (ARCHIVE_OK);
2571
2572 /*
2573 Read the 12 bytes encryption header stored at
2574 the start of the data area.
2575 */
2576 #define ENC_HEADER_SIZE 12
2577 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2578 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) {
2579 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2580 "Truncated Zip encrypted body: only %jd bytes available",
2581 (intmax_t)zip->entry_bytes_remaining);
2582 return (ARCHIVE_FATAL);
2583 }
2584
2585 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL);
2586 if (p == NULL) {
2587 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2588 "Truncated ZIP file data");
2589 return (ARCHIVE_FATAL);
2590 }
2591
2592 for (retry = 0;; retry++) {
2593 const char *passphrase;
2594 uint8_t crcchk;
2595
2596 passphrase = __archive_read_next_passphrase(a);
2597 if (passphrase == NULL) {
2598 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2599 (retry > 0)?
2600 "Incorrect passphrase":
2601 "Passphrase required for this entry");
2602 return (ARCHIVE_FAILED);
2603 }
2604
2605 /*
2606 * Initialize ctx for Traditional PKWARE Decryption.
2607 */
2608 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase),
2609 p, ENC_HEADER_SIZE, &crcchk);
2610 if (r == 0 && crcchk == zip->entry->decdat)
2611 break;/* The passphrase is OK. */
2612 if (retry > 10000) {
2613 /* Avoid infinity loop. */
2614 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2615 "Too many incorrect passphrases");
2616 return (ARCHIVE_FAILED);
2617 }
2618 }
2619
2620 __archive_read_consume(a, ENC_HEADER_SIZE);
2621 zip->tctx_valid = 1;
2622 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
2623 zip->entry_bytes_remaining -= ENC_HEADER_SIZE;
2624 }
2625 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/
2626 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE;
2627 zip->decrypted_bytes_remaining = 0;
2628
2629 return (zip_alloc_decryption_buffer(a));
2630 #undef ENC_HEADER_SIZE
2631 }
2632
2633 static int
init_WinZip_AES_decryption(struct archive_read * a)2634 init_WinZip_AES_decryption(struct archive_read *a)
2635 {
2636 struct zip *zip = (struct zip *)(a->format->data);
2637 const void *p;
2638 const uint8_t *pv;
2639 size_t key_len, salt_len;
2640 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE];
2641 int retry;
2642 int r;
2643
2644 if (zip->cctx_valid || zip->hctx_valid)
2645 return (ARCHIVE_OK);
2646
2647 switch (zip->entry->aes_extra.strength) {
2648 case 1: salt_len = 8; key_len = 16; break;
2649 case 2: salt_len = 12; key_len = 24; break;
2650 case 3: salt_len = 16; key_len = 32; break;
2651 default: goto corrupted;
2652 }
2653 p = __archive_read_ahead(a, salt_len + 2, NULL);
2654 if (p == NULL)
2655 goto truncated;
2656
2657 for (retry = 0;; retry++) {
2658 const char *passphrase;
2659
2660 passphrase = __archive_read_next_passphrase(a);
2661 if (passphrase == NULL) {
2662 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2663 (retry > 0)?
2664 "Incorrect passphrase":
2665 "Passphrase required for this entry");
2666 return (ARCHIVE_FAILED);
2667 }
2668 memset(derived_key, 0, sizeof(derived_key));
2669 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase),
2670 p, salt_len, 1000, derived_key, key_len * 2 + 2);
2671 if (r != 0) {
2672 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2673 "Decryption is unsupported due to lack of "
2674 "crypto library");
2675 return (ARCHIVE_FAILED);
2676 }
2677
2678 /* Check password verification value. */
2679 pv = ((const uint8_t *)p) + salt_len;
2680 if (derived_key[key_len * 2] == pv[0] &&
2681 derived_key[key_len * 2 + 1] == pv[1])
2682 break;/* The passphrase is OK. */
2683 if (retry > 10000) {
2684 /* Avoid infinity loop. */
2685 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2686 "Too many incorrect passphrases");
2687 return (ARCHIVE_FAILED);
2688 }
2689 }
2690
2691 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len);
2692 if (r != 0) {
2693 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2694 "Decryption is unsupported due to lack of crypto library");
2695 return (ARCHIVE_FAILED);
2696 }
2697 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len);
2698 if (r != 0) {
2699 archive_decrypto_aes_ctr_release(&zip->cctx);
2700 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2701 "Failed to initialize HMAC-SHA1");
2702 return (ARCHIVE_FAILED);
2703 }
2704 zip->cctx_valid = zip->hctx_valid = 1;
2705 __archive_read_consume(a, salt_len + 2);
2706 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE;
2707 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2708 && zip->entry_bytes_remaining < 0)
2709 goto corrupted;
2710 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE;
2711 zip->decrypted_bytes_remaining = 0;
2712
2713 zip->entry->compression = zip->entry->aes_extra.compression;
2714 return (zip_alloc_decryption_buffer(a));
2715
2716 truncated:
2717 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2718 "Truncated ZIP file data");
2719 return (ARCHIVE_FATAL);
2720 corrupted:
2721 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2722 "Corrupted ZIP file data");
2723 return (ARCHIVE_FATAL);
2724 }
2725
2726 static int
archive_read_format_zip_read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)2727 archive_read_format_zip_read_data(struct archive_read *a,
2728 const void **buff, size_t *size, int64_t *offset)
2729 {
2730 int r;
2731 struct zip *zip = (struct zip *)(a->format->data);
2732
2733 if (zip->has_encrypted_entries ==
2734 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) {
2735 zip->has_encrypted_entries = 0;
2736 }
2737
2738 *offset = zip->entry_uncompressed_bytes_read;
2739 *size = 0;
2740 *buff = NULL;
2741
2742 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */
2743 if (zip->end_of_entry)
2744 return (ARCHIVE_EOF);
2745
2746 /* Return EOF immediately if this is a non-regular file. */
2747 if (AE_IFREG != (zip->entry->mode & AE_IFMT))
2748 return (ARCHIVE_EOF);
2749
2750 __archive_read_consume(a, zip->unconsumed);
2751 zip->unconsumed = 0;
2752
2753 if (zip->init_decryption) {
2754 zip->has_encrypted_entries = 1;
2755 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
2756 r = read_decryption_header(a);
2757 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
2758 r = init_WinZip_AES_decryption(a);
2759 else
2760 r = init_traditional_PKWARE_decryption(a);
2761 if (r != ARCHIVE_OK)
2762 return (r);
2763 zip->init_decryption = 0;
2764 }
2765
2766 switch(zip->entry->compression) {
2767 case 0: /* No compression. */
2768 r = zip_read_data_none(a, buff, size, offset);
2769 break;
2770 #ifdef HAVE_BZLIB_H
2771 case 12: /* ZIPx bzip2 compression. */
2772 r = zip_read_data_zipx_bzip2(a, buff, size, offset);
2773 break;
2774 #endif
2775 #if HAVE_LZMA_H && HAVE_LIBLZMA
2776 case 14: /* ZIPx LZMA compression. */
2777 r = zip_read_data_zipx_lzma_alone(a, buff, size, offset);
2778 break;
2779 case 95: /* ZIPx XZ compression. */
2780 r = zip_read_data_zipx_xz(a, buff, size, offset);
2781 break;
2782 #endif
2783 /* PPMd support is built-in, so we don't need any #if guards. */
2784 case 98: /* ZIPx PPMd compression. */
2785 r = zip_read_data_zipx_ppmd(a, buff, size, offset);
2786 break;
2787
2788 #ifdef HAVE_ZLIB_H
2789 case 8: /* Deflate compression. */
2790 r = zip_read_data_deflate(a, buff, size, offset);
2791 break;
2792 #endif
2793 default: /* Unsupported compression. */
2794 /* Return a warning. */
2795 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2796 "Unsupported ZIP compression method (%d: %s)",
2797 zip->entry->compression, compression_name(zip->entry->compression));
2798 /* We can't decompress this entry, but we will
2799 * be able to skip() it and try the next entry. */
2800 return (ARCHIVE_FAILED);
2801 break;
2802 }
2803 if (r != ARCHIVE_OK)
2804 return (r);
2805 /* Update checksum */
2806 if (*size)
2807 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff,
2808 (unsigned)*size);
2809 /* If we hit the end, swallow any end-of-data marker. */
2810 if (zip->end_of_entry) {
2811 /* Check file size, CRC against these values. */
2812 if (zip->entry->compressed_size !=
2813 zip->entry_compressed_bytes_read) {
2814 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2815 "ZIP compressed data is wrong size "
2816 "(read %jd, expected %jd)",
2817 (intmax_t)zip->entry_compressed_bytes_read,
2818 (intmax_t)zip->entry->compressed_size);
2819 return (ARCHIVE_WARN);
2820 }
2821 /* Size field only stores the lower 32 bits of the actual
2822 * size. */
2823 if ((zip->entry->uncompressed_size & UINT32_MAX)
2824 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) {
2825 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2826 "ZIP uncompressed data is wrong size "
2827 "(read %jd, expected %jd)\n",
2828 (intmax_t)zip->entry_uncompressed_bytes_read,
2829 (intmax_t)zip->entry->uncompressed_size);
2830 return (ARCHIVE_WARN);
2831 }
2832 /* Check computed CRC against header */
2833 if ((!zip->hctx_valid ||
2834 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) &&
2835 zip->entry->crc32 != zip->entry_crc32
2836 && !zip->ignore_crc32) {
2837 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2838 "ZIP bad CRC: 0x%lx should be 0x%lx",
2839 (unsigned long)zip->entry_crc32,
2840 (unsigned long)zip->entry->crc32);
2841 return (ARCHIVE_WARN);
2842 }
2843 }
2844
2845 return (ARCHIVE_OK);
2846 }
2847
2848 static int
archive_read_format_zip_cleanup(struct archive_read * a)2849 archive_read_format_zip_cleanup(struct archive_read *a)
2850 {
2851 struct zip *zip;
2852 struct zip_entry *zip_entry, *next_zip_entry;
2853
2854 zip = (struct zip *)(a->format->data);
2855
2856 #ifdef HAVE_ZLIB_H
2857 if (zip->stream_valid)
2858 inflateEnd(&zip->stream);
2859 #endif
2860
2861 #if HAVE_LZMA_H && HAVE_LIBLZMA
2862 if (zip->zipx_lzma_valid) {
2863 lzma_end(&zip->zipx_lzma_stream);
2864 }
2865 #endif
2866
2867 #ifdef HAVE_BZLIB_H
2868 if (zip->bzstream_valid) {
2869 BZ2_bzDecompressEnd(&zip->bzstream);
2870 }
2871 #endif
2872
2873 free(zip->uncompressed_buffer);
2874
2875 if (zip->ppmd8_valid)
2876 __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8);
2877
2878 if (zip->zip_entries) {
2879 zip_entry = zip->zip_entries;
2880 while (zip_entry != NULL) {
2881 next_zip_entry = zip_entry->next;
2882 archive_string_free(&zip_entry->rsrcname);
2883 free(zip_entry);
2884 zip_entry = next_zip_entry;
2885 }
2886 }
2887 free(zip->decrypted_buffer);
2888 if (zip->cctx_valid)
2889 archive_decrypto_aes_ctr_release(&zip->cctx);
2890 if (zip->hctx_valid)
2891 archive_hmac_sha1_cleanup(&zip->hctx);
2892 free(zip->iv);
2893 free(zip->erd);
2894 free(zip->v_data);
2895 archive_string_free(&zip->format_name);
2896 free(zip);
2897 (a->format->data) = NULL;
2898 return (ARCHIVE_OK);
2899 }
2900
2901 static int
archive_read_format_zip_has_encrypted_entries(struct archive_read * _a)2902 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a)
2903 {
2904 if (_a && _a->format) {
2905 struct zip * zip = (struct zip *)_a->format->data;
2906 if (zip) {
2907 return zip->has_encrypted_entries;
2908 }
2909 }
2910 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
2911 }
2912
2913 static int
archive_read_format_zip_options(struct archive_read * a,const char * key,const char * val)2914 archive_read_format_zip_options(struct archive_read *a,
2915 const char *key, const char *val)
2916 {
2917 struct zip *zip;
2918 int ret = ARCHIVE_FAILED;
2919
2920 zip = (struct zip *)(a->format->data);
2921 if (strcmp(key, "compat-2x") == 0) {
2922 /* Handle filenames as libarchive 2.x */
2923 zip->init_default_conversion = (val != NULL) ? 1 : 0;
2924 return (ARCHIVE_OK);
2925 } else if (strcmp(key, "hdrcharset") == 0) {
2926 if (val == NULL || val[0] == 0)
2927 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2928 "zip: hdrcharset option needs a character-set name"
2929 );
2930 else {
2931 zip->sconv = archive_string_conversion_from_charset(
2932 &a->archive, val, 0);
2933 if (zip->sconv != NULL) {
2934 if (strcmp(val, "UTF-8") == 0)
2935 zip->sconv_utf8 = zip->sconv;
2936 ret = ARCHIVE_OK;
2937 } else
2938 ret = ARCHIVE_FATAL;
2939 }
2940 return (ret);
2941 } else if (strcmp(key, "ignorecrc32") == 0) {
2942 /* Mostly useful for testing. */
2943 if (val == NULL || val[0] == 0) {
2944 zip->crc32func = real_crc32;
2945 zip->ignore_crc32 = 0;
2946 } else {
2947 zip->crc32func = fake_crc32;
2948 zip->ignore_crc32 = 1;
2949 }
2950 return (ARCHIVE_OK);
2951 } else if (strcmp(key, "mac-ext") == 0) {
2952 zip->process_mac_extensions = (val != NULL && val[0] != 0);
2953 return (ARCHIVE_OK);
2954 }
2955
2956 /* Note: The "warn" return is just to inform the options
2957 * supervisor that we didn't handle it. It will generate
2958 * a suitable error if no one used this option. */
2959 return (ARCHIVE_WARN);
2960 }
2961
2962 int
archive_read_support_format_zip(struct archive * a)2963 archive_read_support_format_zip(struct archive *a)
2964 {
2965 int r;
2966 r = archive_read_support_format_zip_streamable(a);
2967 if (r != ARCHIVE_OK)
2968 return r;
2969 return (archive_read_support_format_zip_seekable(a));
2970 }
2971
2972 /* ------------------------------------------------------------------------ */
2973
2974 /*
2975 * Streaming-mode support
2976 */
2977
2978
2979 static int
archive_read_support_format_zip_capabilities_streamable(struct archive_read * a)2980 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a)
2981 {
2982 (void)a; /* UNUSED */
2983 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
2984 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
2985 }
2986
2987 static int
archive_read_format_zip_streamable_bid(struct archive_read * a,int best_bid)2988 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid)
2989 {
2990 const char *p;
2991
2992 (void)best_bid; /* UNUSED */
2993
2994 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
2995 return (-1);
2996
2997 /*
2998 * Bid of 29 here comes from:
2999 * + 16 bits for "PK",
3000 * + next 16-bit field has 6 options so contributes
3001 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits
3002 *
3003 * So we've effectively verified ~29 total bits of check data.
3004 */
3005 if (p[0] == 'P' && p[1] == 'K') {
3006 if ((p[2] == '\001' && p[3] == '\002')
3007 || (p[2] == '\003' && p[3] == '\004')
3008 || (p[2] == '\005' && p[3] == '\006')
3009 || (p[2] == '\006' && p[3] == '\006')
3010 || (p[2] == '\007' && p[3] == '\010')
3011 || (p[2] == '0' && p[3] == '0'))
3012 return (29);
3013 }
3014
3015 /* TODO: It's worth looking ahead a little bit for a valid
3016 * PK signature. In particular, that would make it possible
3017 * to read some UUEncoded SFX files or SFX files coming from
3018 * a network socket. */
3019
3020 return (0);
3021 }
3022
3023 static int
archive_read_format_zip_streamable_read_header(struct archive_read * a,struct archive_entry * entry)3024 archive_read_format_zip_streamable_read_header(struct archive_read *a,
3025 struct archive_entry *entry)
3026 {
3027 struct zip *zip;
3028
3029 a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
3030 if (a->archive.archive_format_name == NULL)
3031 a->archive.archive_format_name = "ZIP";
3032
3033 zip = (struct zip *)(a->format->data);
3034
3035 /*
3036 * It should be sufficient to call archive_read_next_header() for
3037 * a reader to determine if an entry is encrypted or not. If the
3038 * encryption of an entry is only detectable when calling
3039 * archive_read_data(), so be it. We'll do the same check there
3040 * as well.
3041 */
3042 if (zip->has_encrypted_entries ==
3043 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
3044 zip->has_encrypted_entries = 0;
3045
3046 /* Make sure we have a zip_entry structure to use. */
3047 if (zip->zip_entries == NULL) {
3048 zip->zip_entries = malloc(sizeof(struct zip_entry));
3049 if (zip->zip_entries == NULL) {
3050 archive_set_error(&a->archive, ENOMEM,
3051 "Out of memory");
3052 return ARCHIVE_FATAL;
3053 }
3054 }
3055 zip->entry = zip->zip_entries;
3056 memset(zip->entry, 0, sizeof(struct zip_entry));
3057
3058 if (zip->cctx_valid)
3059 archive_decrypto_aes_ctr_release(&zip->cctx);
3060 if (zip->hctx_valid)
3061 archive_hmac_sha1_cleanup(&zip->hctx);
3062 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
3063 __archive_read_reset_passphrase(a);
3064
3065 /* Search ahead for the next local file header. */
3066 __archive_read_consume(a, zip->unconsumed);
3067 zip->unconsumed = 0;
3068 for (;;) {
3069 int64_t skipped = 0;
3070 const char *p, *end;
3071 ssize_t bytes;
3072
3073 p = __archive_read_ahead(a, 4, &bytes);
3074 if (p == NULL)
3075 return (ARCHIVE_FATAL);
3076 end = p + bytes;
3077
3078 while (p + 4 <= end) {
3079 if (p[0] == 'P' && p[1] == 'K') {
3080 if (p[2] == '\003' && p[3] == '\004') {
3081 /* Regular file entry. */
3082 __archive_read_consume(a, skipped);
3083 return zip_read_local_file_header(a,
3084 entry, zip);
3085 }
3086
3087 /*
3088 * TODO: We cannot restore permissions
3089 * based only on the local file headers.
3090 * Consider scanning the central
3091 * directory and returning additional
3092 * entries for at least directories.
3093 * This would allow us to properly set
3094 * directory permissions.
3095 *
3096 * This won't help us fix symlinks
3097 * and may not help with regular file
3098 * permissions, either. <sigh>
3099 */
3100 if (p[2] == '\001' && p[3] == '\002') {
3101 return (ARCHIVE_EOF);
3102 }
3103
3104 /* End of central directory? Must be an
3105 * empty archive. */
3106 if ((p[2] == '\005' && p[3] == '\006')
3107 || (p[2] == '\006' && p[3] == '\006'))
3108 return (ARCHIVE_EOF);
3109 }
3110 ++p;
3111 ++skipped;
3112 }
3113 __archive_read_consume(a, skipped);
3114 }
3115 }
3116
3117 static int
archive_read_format_zip_read_data_skip_streamable(struct archive_read * a)3118 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a)
3119 {
3120 struct zip *zip;
3121 int64_t bytes_skipped;
3122
3123 zip = (struct zip *)(a->format->data);
3124 bytes_skipped = __archive_read_consume(a, zip->unconsumed);
3125 zip->unconsumed = 0;
3126 if (bytes_skipped < 0)
3127 return (ARCHIVE_FATAL);
3128
3129 /* If we've already read to end of data, we're done. */
3130 if (zip->end_of_entry)
3131 return (ARCHIVE_OK);
3132
3133 /* So we know we're streaming... */
3134 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
3135 || zip->entry->compressed_size > 0) {
3136 /* We know the compressed length, so we can just skip. */
3137 bytes_skipped = __archive_read_consume(a,
3138 zip->entry_bytes_remaining);
3139 if (bytes_skipped < 0)
3140 return (ARCHIVE_FATAL);
3141 return (ARCHIVE_OK);
3142 }
3143
3144 if (zip->init_decryption) {
3145 int r;
3146
3147 zip->has_encrypted_entries = 1;
3148 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
3149 r = read_decryption_header(a);
3150 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
3151 r = init_WinZip_AES_decryption(a);
3152 else
3153 r = init_traditional_PKWARE_decryption(a);
3154 if (r != ARCHIVE_OK)
3155 return (r);
3156 zip->init_decryption = 0;
3157 }
3158
3159 /* We're streaming and we don't know the length. */
3160 /* If the body is compressed and we know the format, we can
3161 * find an exact end-of-entry by decompressing it. */
3162 switch (zip->entry->compression) {
3163 #ifdef HAVE_ZLIB_H
3164 case 8: /* Deflate compression. */
3165 while (!zip->end_of_entry) {
3166 int64_t offset = 0;
3167 const void *buff = NULL;
3168 size_t size = 0;
3169 int r;
3170 r = zip_read_data_deflate(a, &buff, &size, &offset);
3171 if (r != ARCHIVE_OK)
3172 return (r);
3173 }
3174 return ARCHIVE_OK;
3175 #endif
3176 default: /* Uncompressed or unknown. */
3177 /* Scan for a PK\007\010 signature. */
3178 for (;;) {
3179 const char *p, *buff;
3180 ssize_t bytes_avail;
3181 buff = __archive_read_ahead(a, 16, &bytes_avail);
3182 if (bytes_avail < 16) {
3183 archive_set_error(&a->archive,
3184 ARCHIVE_ERRNO_FILE_FORMAT,
3185 "Truncated ZIP file data");
3186 return (ARCHIVE_FATAL);
3187 }
3188 p = buff;
3189 while (p <= buff + bytes_avail - 16) {
3190 if (p[3] == 'P') { p += 3; }
3191 else if (p[3] == 'K') { p += 2; }
3192 else if (p[3] == '\007') { p += 1; }
3193 else if (p[3] == '\010' && p[2] == '\007'
3194 && p[1] == 'K' && p[0] == 'P') {
3195 if (zip->entry->flags & LA_USED_ZIP64)
3196 __archive_read_consume(a,
3197 p - buff + 24);
3198 else
3199 __archive_read_consume(a,
3200 p - buff + 16);
3201 return ARCHIVE_OK;
3202 } else { p += 4; }
3203 }
3204 __archive_read_consume(a, p - buff);
3205 }
3206 }
3207 }
3208
3209 int
archive_read_support_format_zip_streamable(struct archive * _a)3210 archive_read_support_format_zip_streamable(struct archive *_a)
3211 {
3212 struct archive_read *a = (struct archive_read *)_a;
3213 struct zip *zip;
3214 int r;
3215
3216 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
3217 ARCHIVE_STATE_NEW, "archive_read_support_format_zip");
3218
3219 zip = (struct zip *)calloc(1, sizeof(*zip));
3220 if (zip == NULL) {
3221 archive_set_error(&a->archive, ENOMEM,
3222 "Can't allocate zip data");
3223 return (ARCHIVE_FATAL);
3224 }
3225
3226 /* Streamable reader doesn't support mac extensions. */
3227 zip->process_mac_extensions = 0;
3228
3229 /*
3230 * Until enough data has been read, we cannot tell about
3231 * any encrypted entries yet.
3232 */
3233 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
3234 zip->crc32func = real_crc32;
3235
3236 r = __archive_read_register_format(a,
3237 zip,
3238 "zip",
3239 archive_read_format_zip_streamable_bid,
3240 archive_read_format_zip_options,
3241 archive_read_format_zip_streamable_read_header,
3242 archive_read_format_zip_read_data,
3243 archive_read_format_zip_read_data_skip_streamable,
3244 NULL,
3245 archive_read_format_zip_cleanup,
3246 archive_read_support_format_zip_capabilities_streamable,
3247 archive_read_format_zip_has_encrypted_entries);
3248
3249 if (r != ARCHIVE_OK)
3250 free(zip);
3251 return (ARCHIVE_OK);
3252 }
3253
3254 /* ------------------------------------------------------------------------ */
3255
3256 /*
3257 * Seeking-mode support
3258 */
3259
3260 static int
archive_read_support_format_zip_capabilities_seekable(struct archive_read * a)3261 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a)
3262 {
3263 (void)a; /* UNUSED */
3264 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
3265 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
3266 }
3267
3268 /*
3269 * TODO: This is a performance sink because it forces the read core to
3270 * drop buffered data from the start of file, which will then have to
3271 * be re-read again if this bidder loses.
3272 *
3273 * We workaround this a little by passing in the best bid so far so
3274 * that later bidders can do nothing if they know they'll never
3275 * outbid. But we can certainly do better...
3276 */
3277 static int
read_eocd(struct zip * zip,const char * p,int64_t current_offset)3278 read_eocd(struct zip *zip, const char *p, int64_t current_offset)
3279 {
3280 /* Sanity-check the EOCD we've found. */
3281
3282 /* This must be the first volume. */
3283 if (archive_le16dec(p + 4) != 0)
3284 return 0;
3285 /* Central directory must be on this volume. */
3286 if (archive_le16dec(p + 4) != archive_le16dec(p + 6))
3287 return 0;
3288 /* All central directory entries must be on this volume. */
3289 if (archive_le16dec(p + 10) != archive_le16dec(p + 8))
3290 return 0;
3291 /* Central directory can't extend beyond start of EOCD record. */
3292 if (archive_le32dec(p + 16) + archive_le32dec(p + 12)
3293 > current_offset)
3294 return 0;
3295
3296 /* Save the central directory location for later use. */
3297 zip->central_directory_offset = archive_le32dec(p + 16);
3298
3299 /* This is just a tiny bit higher than the maximum
3300 returned by the streaming Zip bidder. This ensures
3301 that the more accurate seeking Zip parser wins
3302 whenever seek is available. */
3303 return 32;
3304 }
3305
3306 /*
3307 * Examine Zip64 EOCD locator: If it's valid, store the information
3308 * from it.
3309 */
3310 static int
read_zip64_eocd(struct archive_read * a,struct zip * zip,const char * p)3311 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p)
3312 {
3313 int64_t eocd64_offset;
3314 int64_t eocd64_size;
3315
3316 /* Sanity-check the locator record. */
3317
3318 /* Central dir must be on first volume. */
3319 if (archive_le32dec(p + 4) != 0)
3320 return 0;
3321 /* Must be only a single volume. */
3322 if (archive_le32dec(p + 16) != 1)
3323 return 0;
3324
3325 /* Find the Zip64 EOCD record. */
3326 eocd64_offset = archive_le64dec(p + 8);
3327 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0)
3328 return 0;
3329 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL)
3330 return 0;
3331 /* Make sure we can read all of it. */
3332 eocd64_size = archive_le64dec(p + 4) + 12;
3333 if (eocd64_size < 56 || eocd64_size > 16384)
3334 return 0;
3335 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL)
3336 return 0;
3337
3338 /* Sanity-check the EOCD64 */
3339 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */
3340 return 0;
3341 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */
3342 return 0;
3343 /* CD can't be split. */
3344 if (archive_le64dec(p + 24) != archive_le64dec(p + 32))
3345 return 0;
3346
3347 /* Save the central directory offset for later use. */
3348 zip->central_directory_offset = archive_le64dec(p + 48);
3349
3350 return 32;
3351 }
3352
3353 static int
archive_read_format_zip_seekable_bid(struct archive_read * a,int best_bid)3354 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid)
3355 {
3356 struct zip *zip = (struct zip *)a->format->data;
3357 int64_t file_size, current_offset;
3358 const char *p;
3359 int i, tail;
3360
3361 /* If someone has already bid more than 32, then avoid
3362 trashing the look-ahead buffers with a seek. */
3363 if (best_bid > 32)
3364 return (-1);
3365
3366 file_size = __archive_read_seek(a, 0, SEEK_END);
3367 if (file_size <= 0)
3368 return 0;
3369
3370 /* Search last 16k of file for end-of-central-directory
3371 * record (which starts with PK\005\006) */
3372 tail = (int)zipmin(1024 * 16, file_size);
3373 current_offset = __archive_read_seek(a, -tail, SEEK_END);
3374 if (current_offset < 0)
3375 return 0;
3376 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL)
3377 return 0;
3378 /* Boyer-Moore search backwards from the end, since we want
3379 * to match the last EOCD in the file (there can be more than
3380 * one if there is an uncompressed Zip archive as a member
3381 * within this Zip archive). */
3382 for (i = tail - 22; i > 0;) {
3383 switch (p[i]) {
3384 case 'P':
3385 if (memcmp(p + i, "PK\005\006", 4) == 0) {
3386 int ret = read_eocd(zip, p + i,
3387 current_offset + i);
3388 /* Zip64 EOCD locator precedes
3389 * regular EOCD if present. */
3390 if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) {
3391 int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20);
3392 if (ret_zip64 > ret)
3393 ret = ret_zip64;
3394 }
3395 return (ret);
3396 }
3397 i -= 4;
3398 break;
3399 case 'K': i -= 1; break;
3400 case 005: i -= 2; break;
3401 case 006: i -= 3; break;
3402 default: i -= 4; break;
3403 }
3404 }
3405 return 0;
3406 }
3407
3408 /* The red-black trees are only used in seeking mode to manage
3409 * the in-memory copy of the central directory. */
3410
3411 static int
cmp_node(const struct archive_rb_node * n1,const struct archive_rb_node * n2)3412 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2)
3413 {
3414 const struct zip_entry *e1 = (const struct zip_entry *)n1;
3415 const struct zip_entry *e2 = (const struct zip_entry *)n2;
3416
3417 if (e1->local_header_offset > e2->local_header_offset)
3418 return -1;
3419 if (e1->local_header_offset < e2->local_header_offset)
3420 return 1;
3421 return 0;
3422 }
3423
3424 static int
cmp_key(const struct archive_rb_node * n,const void * key)3425 cmp_key(const struct archive_rb_node *n, const void *key)
3426 {
3427 /* This function won't be called */
3428 (void)n; /* UNUSED */
3429 (void)key; /* UNUSED */
3430 return 1;
3431 }
3432
3433 static const struct archive_rb_tree_ops rb_ops = {
3434 &cmp_node, &cmp_key
3435 };
3436
3437 static int
rsrc_cmp_node(const struct archive_rb_node * n1,const struct archive_rb_node * n2)3438 rsrc_cmp_node(const struct archive_rb_node *n1,
3439 const struct archive_rb_node *n2)
3440 {
3441 const struct zip_entry *e1 = (const struct zip_entry *)n1;
3442 const struct zip_entry *e2 = (const struct zip_entry *)n2;
3443
3444 return (strcmp(e2->rsrcname.s, e1->rsrcname.s));
3445 }
3446
3447 static int
rsrc_cmp_key(const struct archive_rb_node * n,const void * key)3448 rsrc_cmp_key(const struct archive_rb_node *n, const void *key)
3449 {
3450 const struct zip_entry *e = (const struct zip_entry *)n;
3451 return (strcmp((const char *)key, e->rsrcname.s));
3452 }
3453
3454 static const struct archive_rb_tree_ops rb_rsrc_ops = {
3455 &rsrc_cmp_node, &rsrc_cmp_key
3456 };
3457
3458 static const char *
rsrc_basename(const char * name,size_t name_length)3459 rsrc_basename(const char *name, size_t name_length)
3460 {
3461 const char *s, *r;
3462
3463 r = s = name;
3464 for (;;) {
3465 s = memchr(s, '/', name_length - (s - name));
3466 if (s == NULL)
3467 break;
3468 r = ++s;
3469 }
3470 return (r);
3471 }
3472
3473 static void
expose_parent_dirs(struct zip * zip,const char * name,size_t name_length)3474 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length)
3475 {
3476 struct archive_string str;
3477 struct zip_entry *dir;
3478 char *s;
3479
3480 archive_string_init(&str);
3481 archive_strncpy(&str, name, name_length);
3482 for (;;) {
3483 s = strrchr(str.s, '/');
3484 if (s == NULL)
3485 break;
3486 *s = '\0';
3487 /* Transfer the parent directory from zip->tree_rsrc RB
3488 * tree to zip->tree RB tree to expose. */
3489 dir = (struct zip_entry *)
3490 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s);
3491 if (dir == NULL)
3492 break;
3493 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node);
3494 archive_string_free(&dir->rsrcname);
3495 __archive_rb_tree_insert_node(&zip->tree, &dir->node);
3496 }
3497 archive_string_free(&str);
3498 }
3499
3500 static int
slurp_central_directory(struct archive_read * a,struct archive_entry * entry,struct zip * zip)3501 slurp_central_directory(struct archive_read *a, struct archive_entry* entry,
3502 struct zip *zip)
3503 {
3504 ssize_t i;
3505 unsigned found;
3506 int64_t correction;
3507 ssize_t bytes_avail;
3508 const char *p;
3509
3510 /*
3511 * Find the start of the central directory. The end-of-CD
3512 * record has our starting point, but there are lots of
3513 * Zip archives which have had other data prepended to the
3514 * file, which makes the recorded offsets all too small.
3515 * So we search forward from the specified offset until we
3516 * find the real start of the central directory. Then we
3517 * know the correction we need to apply to account for leading
3518 * padding.
3519 */
3520 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0)
3521 return ARCHIVE_FATAL;
3522
3523 found = 0;
3524 while (!found) {
3525 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL)
3526 return ARCHIVE_FATAL;
3527 for (found = 0, i = 0; !found && i < bytes_avail - 4;) {
3528 switch (p[i + 3]) {
3529 case 'P': i += 3; break;
3530 case 'K': i += 2; break;
3531 case 001: i += 1; break;
3532 case 002:
3533 if (memcmp(p + i, "PK\001\002", 4) == 0) {
3534 p += i;
3535 found = 1;
3536 } else
3537 i += 4;
3538 break;
3539 case 005: i += 1; break;
3540 case 006:
3541 if (memcmp(p + i, "PK\005\006", 4) == 0) {
3542 p += i;
3543 found = 1;
3544 } else if (memcmp(p + i, "PK\006\006", 4) == 0) {
3545 p += i;
3546 found = 1;
3547 } else
3548 i += 1;
3549 break;
3550 default: i += 4; break;
3551 }
3552 }
3553 __archive_read_consume(a, i);
3554 }
3555 correction = archive_filter_bytes(&a->archive, 0)
3556 - zip->central_directory_offset;
3557
3558 __archive_rb_tree_init(&zip->tree, &rb_ops);
3559 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops);
3560
3561 zip->central_directory_entries_total = 0;
3562 while (1) {
3563 struct zip_entry *zip_entry;
3564 size_t filename_length, extra_length, comment_length;
3565 uint32_t external_attributes;
3566 const char *name, *r;
3567
3568 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
3569 return ARCHIVE_FATAL;
3570 if (memcmp(p, "PK\006\006", 4) == 0
3571 || memcmp(p, "PK\005\006", 4) == 0) {
3572 break;
3573 } else if (memcmp(p, "PK\001\002", 4) != 0) {
3574 archive_set_error(&a->archive,
3575 -1, "Invalid central directory signature");
3576 return ARCHIVE_FATAL;
3577 }
3578 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL)
3579 return ARCHIVE_FATAL;
3580
3581 zip_entry = calloc(1, sizeof(struct zip_entry));
3582 if (zip_entry == NULL) {
3583 archive_set_error(&a->archive, ENOMEM,
3584 "Can't allocate zip entry");
3585 return ARCHIVE_FATAL;
3586 }
3587 zip_entry->next = zip->zip_entries;
3588 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY;
3589 zip->zip_entries = zip_entry;
3590 zip->central_directory_entries_total++;
3591
3592 /* version = p[4]; */
3593 zip_entry->system = p[5];
3594 /* version_required = archive_le16dec(p + 6); */
3595 zip_entry->zip_flags = archive_le16dec(p + 8);
3596 if (zip_entry->zip_flags
3597 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){
3598 zip->has_encrypted_entries = 1;
3599 }
3600 zip_entry->compression = (char)archive_le16dec(p + 10);
3601 zip_entry->mtime = zip_time(p + 12);
3602 zip_entry->crc32 = archive_le32dec(p + 16);
3603 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
3604 zip_entry->decdat = p[13];
3605 else
3606 zip_entry->decdat = p[19];
3607 zip_entry->compressed_size = archive_le32dec(p + 20);
3608 zip_entry->uncompressed_size = archive_le32dec(p + 24);
3609 filename_length = archive_le16dec(p + 28);
3610 extra_length = archive_le16dec(p + 30);
3611 comment_length = archive_le16dec(p + 32);
3612 /* disk_start = archive_le16dec(p + 34);
3613 * Better be zero.
3614 * internal_attributes = archive_le16dec(p + 36);
3615 * text bit */
3616 external_attributes = archive_le32dec(p + 38);
3617 zip_entry->local_header_offset =
3618 archive_le32dec(p + 42) + correction;
3619
3620 /* If we can't guess the mode, leave it zero here;
3621 when we read the local file header we might get
3622 more information. */
3623 if (zip_entry->system == 3) {
3624 zip_entry->mode = external_attributes >> 16;
3625 } else if (zip_entry->system == 0) {
3626 // Interpret MSDOS directory bit
3627 if (0x10 == (external_attributes & 0x10)) {
3628 zip_entry->mode = AE_IFDIR | 0775;
3629 } else {
3630 zip_entry->mode = AE_IFREG | 0664;
3631 }
3632 if (0x01 == (external_attributes & 0x01)) {
3633 // Read-only bit; strip write permissions
3634 zip_entry->mode &= 0555;
3635 }
3636 } else {
3637 zip_entry->mode = 0;
3638 }
3639
3640 /* We're done with the regular data; get the filename and
3641 * extra data. */
3642 __archive_read_consume(a, 46);
3643 p = __archive_read_ahead(a, filename_length + extra_length,
3644 NULL);
3645 if (p == NULL) {
3646 archive_set_error(&a->archive,
3647 ARCHIVE_ERRNO_FILE_FORMAT,
3648 "Truncated ZIP file header");
3649 return ARCHIVE_FATAL;
3650 }
3651 if (ARCHIVE_OK != process_extra(a, entry, p + filename_length,
3652 extra_length, zip_entry)) {
3653 return ARCHIVE_FATAL;
3654 }
3655
3656 /*
3657 * Mac resource fork files are stored under the
3658 * "__MACOSX/" directory, so we should check if
3659 * it is.
3660 */
3661 if (!zip->process_mac_extensions) {
3662 /* Treat every entry as a regular entry. */
3663 __archive_rb_tree_insert_node(&zip->tree,
3664 &zip_entry->node);
3665 } else {
3666 name = p;
3667 r = rsrc_basename(name, filename_length);
3668 if (filename_length >= 9 &&
3669 strncmp("__MACOSX/", name, 9) == 0) {
3670 /* If this file is not a resource fork nor
3671 * a directory. We should treat it as a non
3672 * resource fork file to expose it. */
3673 if (name[filename_length-1] != '/' &&
3674 (r - name < 3 || r[0] != '.' ||
3675 r[1] != '_')) {
3676 __archive_rb_tree_insert_node(
3677 &zip->tree, &zip_entry->node);
3678 /* Expose its parent directories. */
3679 expose_parent_dirs(zip, name,
3680 filename_length);
3681 } else {
3682 /* This file is a resource fork file or
3683 * a directory. */
3684 archive_strncpy(&(zip_entry->rsrcname),
3685 name, filename_length);
3686 __archive_rb_tree_insert_node(
3687 &zip->tree_rsrc, &zip_entry->node);
3688 }
3689 } else {
3690 /* Generate resource fork name to find its
3691 * resource file at zip->tree_rsrc. */
3692 archive_strcpy(&(zip_entry->rsrcname),
3693 "__MACOSX/");
3694 archive_strncat(&(zip_entry->rsrcname),
3695 name, r - name);
3696 archive_strcat(&(zip_entry->rsrcname), "._");
3697 archive_strncat(&(zip_entry->rsrcname),
3698 name + (r - name),
3699 filename_length - (r - name));
3700 /* Register an entry to RB tree to sort it by
3701 * file offset. */
3702 __archive_rb_tree_insert_node(&zip->tree,
3703 &zip_entry->node);
3704 }
3705 }
3706
3707 /* Skip the comment too ... */
3708 __archive_read_consume(a,
3709 filename_length + extra_length + comment_length);
3710 }
3711
3712 return ARCHIVE_OK;
3713 }
3714
3715 static ssize_t
zip_get_local_file_header_size(struct archive_read * a,size_t extra)3716 zip_get_local_file_header_size(struct archive_read *a, size_t extra)
3717 {
3718 const char *p;
3719 ssize_t filename_length, extra_length;
3720
3721 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) {
3722 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3723 "Truncated ZIP file header");
3724 return (ARCHIVE_WARN);
3725 }
3726 p += extra;
3727
3728 if (memcmp(p, "PK\003\004", 4) != 0) {
3729 archive_set_error(&a->archive, -1, "Damaged Zip archive");
3730 return ARCHIVE_WARN;
3731 }
3732 filename_length = archive_le16dec(p + 26);
3733 extra_length = archive_le16dec(p + 28);
3734
3735 return (30 + filename_length + extra_length);
3736 }
3737
3738 static int
zip_read_mac_metadata(struct archive_read * a,struct archive_entry * entry,struct zip_entry * rsrc)3739 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry,
3740 struct zip_entry *rsrc)
3741 {
3742 struct zip *zip = (struct zip *)a->format->data;
3743 unsigned char *metadata, *mp;
3744 int64_t offset = archive_filter_bytes(&a->archive, 0);
3745 size_t remaining_bytes, metadata_bytes;
3746 ssize_t hsize;
3747 int ret = ARCHIVE_OK, eof;
3748
3749 switch(rsrc->compression) {
3750 case 0: /* No compression. */
3751 if (rsrc->uncompressed_size != rsrc->compressed_size) {
3752 archive_set_error(&a->archive,
3753 ARCHIVE_ERRNO_FILE_FORMAT,
3754 "Malformed OS X metadata entry: "
3755 "inconsistent size");
3756 return (ARCHIVE_FATAL);
3757 }
3758 #ifdef HAVE_ZLIB_H
3759 case 8: /* Deflate compression. */
3760 #endif
3761 break;
3762 default: /* Unsupported compression. */
3763 /* Return a warning. */
3764 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3765 "Unsupported ZIP compression method (%s)",
3766 compression_name(rsrc->compression));
3767 /* We can't decompress this entry, but we will
3768 * be able to skip() it and try the next entry. */
3769 return (ARCHIVE_WARN);
3770 }
3771
3772 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) {
3773 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3774 "Mac metadata is too large: %jd > 4M bytes",
3775 (intmax_t)rsrc->uncompressed_size);
3776 return (ARCHIVE_WARN);
3777 }
3778 if (rsrc->compressed_size > (4 * 1024 * 1024)) {
3779 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
3780 "Mac metadata is too large: %jd > 4M bytes",
3781 (intmax_t)rsrc->compressed_size);
3782 return (ARCHIVE_WARN);
3783 }
3784
3785 metadata = malloc((size_t)rsrc->uncompressed_size);
3786 if (metadata == NULL) {
3787 archive_set_error(&a->archive, ENOMEM,
3788 "Can't allocate memory for Mac metadata");
3789 return (ARCHIVE_FATAL);
3790 }
3791
3792 if (offset < rsrc->local_header_offset)
3793 __archive_read_consume(a, rsrc->local_header_offset - offset);
3794 else if (offset != rsrc->local_header_offset) {
3795 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET);
3796 }
3797
3798 hsize = zip_get_local_file_header_size(a, 0);
3799 __archive_read_consume(a, hsize);
3800
3801 remaining_bytes = (size_t)rsrc->compressed_size;
3802 metadata_bytes = (size_t)rsrc->uncompressed_size;
3803 mp = metadata;
3804 eof = 0;
3805 while (!eof && remaining_bytes) {
3806 const unsigned char *p;
3807 ssize_t bytes_avail;
3808 size_t bytes_used;
3809
3810 p = __archive_read_ahead(a, 1, &bytes_avail);
3811 if (p == NULL) {
3812 archive_set_error(&a->archive,
3813 ARCHIVE_ERRNO_FILE_FORMAT,
3814 "Truncated ZIP file header");
3815 ret = ARCHIVE_WARN;
3816 goto exit_mac_metadata;
3817 }
3818 if ((size_t)bytes_avail > remaining_bytes)
3819 bytes_avail = remaining_bytes;
3820 switch(rsrc->compression) {
3821 case 0: /* No compression. */
3822 if ((size_t)bytes_avail > metadata_bytes)
3823 bytes_avail = metadata_bytes;
3824 memcpy(mp, p, bytes_avail);
3825 bytes_used = (size_t)bytes_avail;
3826 metadata_bytes -= bytes_used;
3827 mp += bytes_used;
3828 if (metadata_bytes == 0)
3829 eof = 1;
3830 break;
3831 #ifdef HAVE_ZLIB_H
3832 case 8: /* Deflate compression. */
3833 {
3834 int r;
3835
3836 ret = zip_deflate_init(a, zip);
3837 if (ret != ARCHIVE_OK)
3838 goto exit_mac_metadata;
3839 zip->stream.next_in =
3840 (Bytef *)(uintptr_t)(const void *)p;
3841 zip->stream.avail_in = (uInt)bytes_avail;
3842 zip->stream.total_in = 0;
3843 zip->stream.next_out = mp;
3844 zip->stream.avail_out = (uInt)metadata_bytes;
3845 zip->stream.total_out = 0;
3846
3847 r = inflate(&zip->stream, 0);
3848 switch (r) {
3849 case Z_OK:
3850 break;
3851 case Z_STREAM_END:
3852 eof = 1;
3853 break;
3854 case Z_MEM_ERROR:
3855 archive_set_error(&a->archive, ENOMEM,
3856 "Out of memory for ZIP decompression");
3857 ret = ARCHIVE_FATAL;
3858 goto exit_mac_metadata;
3859 default:
3860 archive_set_error(&a->archive,
3861 ARCHIVE_ERRNO_MISC,
3862 "ZIP decompression failed (%d)", r);
3863 ret = ARCHIVE_FATAL;
3864 goto exit_mac_metadata;
3865 }
3866 bytes_used = zip->stream.total_in;
3867 metadata_bytes -= zip->stream.total_out;
3868 mp += zip->stream.total_out;
3869 break;
3870 }
3871 #endif
3872 default:
3873 bytes_used = 0;
3874 break;
3875 }
3876 __archive_read_consume(a, bytes_used);
3877 remaining_bytes -= bytes_used;
3878 }
3879 archive_entry_copy_mac_metadata(entry, metadata,
3880 (size_t)rsrc->uncompressed_size - metadata_bytes);
3881
3882 exit_mac_metadata:
3883 __archive_read_seek(a, offset, SEEK_SET);
3884 zip->decompress_init = 0;
3885 free(metadata);
3886 return (ret);
3887 }
3888
3889 static int
archive_read_format_zip_seekable_read_header(struct archive_read * a,struct archive_entry * entry)3890 archive_read_format_zip_seekable_read_header(struct archive_read *a,
3891 struct archive_entry *entry)
3892 {
3893 struct zip *zip = (struct zip *)a->format->data;
3894 struct zip_entry *rsrc;
3895 int64_t offset;
3896 int r, ret = ARCHIVE_OK;
3897
3898 /*
3899 * It should be sufficient to call archive_read_next_header() for
3900 * a reader to determine if an entry is encrypted or not. If the
3901 * encryption of an entry is only detectable when calling
3902 * archive_read_data(), so be it. We'll do the same check there
3903 * as well.
3904 */
3905 if (zip->has_encrypted_entries ==
3906 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
3907 zip->has_encrypted_entries = 0;
3908
3909 a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
3910 if (a->archive.archive_format_name == NULL)
3911 a->archive.archive_format_name = "ZIP";
3912
3913 if (zip->zip_entries == NULL) {
3914 r = slurp_central_directory(a, entry, zip);
3915 if (r != ARCHIVE_OK)
3916 return r;
3917 /* Get first entry whose local header offset is lower than
3918 * other entries in the archive file. */
3919 zip->entry =
3920 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree);
3921 } else if (zip->entry != NULL) {
3922 /* Get next entry in local header offset order. */
3923 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate(
3924 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT);
3925 }
3926
3927 if (zip->entry == NULL)
3928 return ARCHIVE_EOF;
3929
3930 if (zip->entry->rsrcname.s)
3931 rsrc = (struct zip_entry *)__archive_rb_tree_find_node(
3932 &zip->tree_rsrc, zip->entry->rsrcname.s);
3933 else
3934 rsrc = NULL;
3935
3936 if (zip->cctx_valid)
3937 archive_decrypto_aes_ctr_release(&zip->cctx);
3938 if (zip->hctx_valid)
3939 archive_hmac_sha1_cleanup(&zip->hctx);
3940 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
3941 __archive_read_reset_passphrase(a);
3942
3943 /* File entries are sorted by the header offset, we should mostly
3944 * use __archive_read_consume to advance a read point to avoid
3945 * redundant data reading. */
3946 offset = archive_filter_bytes(&a->archive, 0);
3947 if (offset < zip->entry->local_header_offset)
3948 __archive_read_consume(a,
3949 zip->entry->local_header_offset - offset);
3950 else if (offset != zip->entry->local_header_offset) {
3951 __archive_read_seek(a, zip->entry->local_header_offset,
3952 SEEK_SET);
3953 }
3954 zip->unconsumed = 0;
3955 r = zip_read_local_file_header(a, entry, zip);
3956 if (r != ARCHIVE_OK)
3957 return r;
3958 if (rsrc) {
3959 int ret2 = zip_read_mac_metadata(a, entry, rsrc);
3960 if (ret2 < ret)
3961 ret = ret2;
3962 }
3963 return (ret);
3964 }
3965
3966 /*
3967 * We're going to seek for the next header anyway, so we don't
3968 * need to bother doing anything here.
3969 */
3970 static int
archive_read_format_zip_read_data_skip_seekable(struct archive_read * a)3971 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a)
3972 {
3973 struct zip *zip;
3974 zip = (struct zip *)(a->format->data);
3975
3976 zip->unconsumed = 0;
3977 return (ARCHIVE_OK);
3978 }
3979
3980 int
archive_read_support_format_zip_seekable(struct archive * _a)3981 archive_read_support_format_zip_seekable(struct archive *_a)
3982 {
3983 struct archive_read *a = (struct archive_read *)_a;
3984 struct zip *zip;
3985 int r;
3986
3987 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
3988 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable");
3989
3990 zip = (struct zip *)calloc(1, sizeof(*zip));
3991 if (zip == NULL) {
3992 archive_set_error(&a->archive, ENOMEM,
3993 "Can't allocate zip data");
3994 return (ARCHIVE_FATAL);
3995 }
3996
3997 #ifdef HAVE_COPYFILE_H
3998 /* Set this by default on Mac OS. */
3999 zip->process_mac_extensions = 1;
4000 #endif
4001
4002 /*
4003 * Until enough data has been read, we cannot tell about
4004 * any encrypted entries yet.
4005 */
4006 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
4007 zip->crc32func = real_crc32;
4008
4009 r = __archive_read_register_format(a,
4010 zip,
4011 "zip",
4012 archive_read_format_zip_seekable_bid,
4013 archive_read_format_zip_options,
4014 archive_read_format_zip_seekable_read_header,
4015 archive_read_format_zip_read_data,
4016 archive_read_format_zip_read_data_skip_seekable,
4017 NULL,
4018 archive_read_format_zip_cleanup,
4019 archive_read_support_format_zip_capabilities_seekable,
4020 archive_read_format_zip_has_encrypted_entries);
4021
4022 if (r != ARCHIVE_OK)
4023 free(zip);
4024 return (ARCHIVE_OK);
4025 }
4026
4027 /*# vim:set noet:*/
4028