2 * Copyright (c) 2004-2013 Tim Kientzle
3 * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA
4 * Copyright (c) 2013 Konrad Kleine
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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.
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.
28 #include "archive_platform.h"
29 __FBSDID("$FreeBSD$");
32 * The definitive documentation of the Zip file format is:
33 * http://www.pkware.com/documents/casestudies/APPNOTE.TXT
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
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.
57 #include "archive_digest_private.h"
58 #include "archive_cryptor_private.h"
59 #include "archive_endian.h"
60 #include "archive_entry.h"
61 #include "archive_entry_locale.h"
62 #include "archive_hmac_private.h"
63 #include "archive_private.h"
64 #include "archive_rb.h"
65 #include "archive_read_private.h"
68 #include "archive_crc32.h"
72 struct archive_rb_node node;
73 struct zip_entry *next;
74 int64_t local_header_offset;
75 int64_t compressed_size;
76 int64_t uncompressed_size;
79 struct archive_string rsrcname;
85 uint16_t zip_flags; /* From GP Flags Field */
86 unsigned char compression;
87 unsigned char system; /* From "version written by" */
88 unsigned char flags; /* Our extra markers. */
89 unsigned char decdat;/* Used for Decryption check */
91 /* WinZip AES encryption extra field should be available
92 * when compression is 99. */
94 /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */
96 #define AES_VENDOR_AE_1 0x0001
97 #define AES_VENDOR_AE_2 0x0002
98 /* AES encryption strength:
99 * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */
101 /* Actual compression method. */
102 unsigned char compression;
106 struct trad_enc_ctx {
110 /* Bits used in zip_flags. */
111 #define ZIP_ENCRYPTED (1 << 0)
112 #define ZIP_LENGTH_AT_END (1 << 3)
113 #define ZIP_STRONG_ENCRYPTED (1 << 6)
114 #define ZIP_UTF8_NAME (1 << 11)
115 /* See "7.2 Single Password Symmetric Encryption Method"
116 in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */
117 #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13)
119 /* Bits used in flags. */
120 #define LA_USED_ZIP64 (1 << 0)
121 #define LA_FROM_CENTRAL_DIRECTORY (1 << 1)
124 * See "WinZip - AES Encryption Information"
125 * http://www.winzip.com/aes_info.htm
127 /* Value used in compression method. */
128 #define WINZIP_AES_ENCRYPTION 99
129 /* Authentication code size. */
130 #define AUTH_CODE_SIZE 10
132 #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2)
135 /* Structural information about the archive. */
136 struct archive_string format_name;
137 int64_t central_directory_offset;
138 size_t central_directory_entries_total;
139 size_t central_directory_entries_on_this_disk;
140 int has_encrypted_entries;
142 /* List of entries (seekable Zip only) */
143 struct zip_entry *zip_entries;
144 struct archive_rb_tree tree;
145 struct archive_rb_tree tree_rsrc;
147 /* Bytes read but not yet consumed via __archive_read_consume() */
150 /* Information about entry we're currently reading. */
151 struct zip_entry *entry;
152 int64_t entry_bytes_remaining;
154 /* These count the number of bytes actually read for the entry. */
155 int64_t entry_compressed_bytes_read;
156 int64_t entry_uncompressed_bytes_read;
158 /* Running CRC32 of the decompressed data */
159 unsigned long entry_crc32;
160 unsigned long (*crc32func)(unsigned long, const void *,
164 /* Flags to mark progress of decompression. */
165 char decompress_init;
169 unsigned char *uncompressed_buffer;
170 size_t uncompressed_buffer_size;
175 struct archive_string_conv *sconv;
176 struct archive_string_conv *sconv_default;
177 struct archive_string_conv *sconv_utf8;
178 int init_default_conversion;
179 int process_mac_extensions;
181 char init_decryption;
183 /* Decryption buffer. */
185 * The decrypted data starts at decrypted_ptr and
186 * extends for decrypted_bytes_remaining. Decryption
187 * adds new data to the end of this block, data is returned
188 * to clients from the beginning. When the block hits the
189 * end of decrypted_buffer, it has to be shuffled back to
190 * the beginning of the buffer.
192 unsigned char *decrypted_buffer;
193 unsigned char *decrypted_ptr;
194 size_t decrypted_buffer_size;
195 size_t decrypted_bytes_remaining;
196 size_t decrypted_unconsumed_bytes;
198 /* Traditional PKWARE decryption. */
199 struct trad_enc_ctx tctx;
202 /* WinZip AES decyption. */
203 /* Contexts used for AES decryption. */
204 archive_crypto_ctx cctx;
206 archive_hmac_sha1_ctx hctx;
209 /* Strong encryption's decryption header information. */
222 /* Many systems define min or MIN, but not all. */
223 #define zipmin(a,b) ((a) < (b) ? (a) : (b))
225 /* ------------------------------------------------------------------------ */
228 Traditional PKWARE Decryption functions.
232 trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c)
235 #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL)
237 ctx->keys[0] = CRC32(ctx->keys[0], c);
238 ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1;
239 t = (ctx->keys[1] >> 24) & 0xff;
240 ctx->keys[2] = CRC32(ctx->keys[2], t);
245 trad_enc_decypt_byte(struct trad_enc_ctx *ctx)
247 unsigned temp = ctx->keys[2] | 2;
248 return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff;
252 trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in,
253 size_t in_len, uint8_t *out, size_t out_len)
257 max = (unsigned)((in_len < out_len)? in_len: out_len);
259 for (i = 0; i < max; i++) {
260 uint8_t t = in[i] ^ trad_enc_decypt_byte(ctx);
262 trad_enc_update_keys(ctx, t);
267 trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len,
268 const uint8_t *key, size_t key_len, uint8_t *crcchk)
277 ctx->keys[0] = 305419896L;
278 ctx->keys[1] = 591751049L;
279 ctx->keys[2] = 878082192L;
281 for (;pw_len; --pw_len)
282 trad_enc_update_keys(ctx, *pw++);
284 trad_enc_decrypt_update(ctx, key, 12, header, 12);
285 /* Return the last byte for CRC check. */
286 *crcchk = header[11];
292 crypt_derive_key_sha1(const void *p, int size, unsigned char *key,
296 archive_sha1_ctx ctx;
297 unsigned char md1[MD_SIZE];
298 unsigned char md2[MD_SIZE * 2];
299 unsigned char mkb[64];
302 archive_sha1_init(&ctx);
303 archive_sha1_update(&ctx, p, size);
304 archive_sha1_final(&ctx, md1);
306 memset(mkb, 0x36, sizeof(mkb));
307 for (i = 0; i < MD_SIZE; i++)
309 archive_sha1_init(&ctx);
310 archive_sha1_update(&ctx, mkb, sizeof(mkb));
311 archive_sha1_final(&ctx, md2);
313 memset(mkb, 0x5C, sizeof(mkb));
314 for (i = 0; i < MD_SIZE; i++)
316 archive_sha1_init(&ctx);
317 archive_sha1_update(&ctx, mkb, sizeof(mkb));
318 archive_sha1_final(&ctx, md2 + MD_SIZE);
322 memcpy(key, md2, key_size);
328 * Common code for streaming or seeking modes.
330 * Includes code to read local file headers, decompress data
331 * from entry bodies, and common API.
335 real_crc32(unsigned long crc, const void *buff, size_t len)
337 return crc32(crc, buff, (unsigned int)len);
340 /* Used by "ignorecrc32" option to speed up tests. */
342 fake_crc32(unsigned long crc, const void *buff, size_t len)
344 (void)crc; /* UNUSED */
345 (void)buff; /* UNUSED */
346 (void)len; /* UNUSED */
353 } compression_methods[] = {
354 {0, "uncompressed"}, /* The file is stored (no compression) */
355 {1, "shrinking"}, /* The file is Shrunk */
356 {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */
357 {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */
358 {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */
359 {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */
360 {6, "imploded"}, /* The file is Imploded */
361 {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */
362 {8, "deflation"}, /* The file is Deflated */
363 {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */
364 {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding
366 {11, "reserved"}, /* Reserved by PKWARE */
367 {12, "bzip"}, /* File is compressed using BZIP2 algorithm */
368 {13, "reserved"}, /* Reserved by PKWARE */
369 {14, "lzma"}, /* LZMA (EFS) */
370 {15, "reserved"}, /* Reserved by PKWARE */
371 {16, "reserved"}, /* Reserved by PKWARE */
372 {17, "reserved"}, /* Reserved by PKWARE */
373 {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */
374 {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */
375 {97, "wav-pack"}, /* WavPack compressed data */
376 {98, "ppmd-1"}, /* PPMd version I, Rev 1 */
377 {99, "aes"} /* WinZip AES encryption */
381 compression_name(const int compression)
383 static const int num_compression_methods =
384 sizeof(compression_methods)/sizeof(compression_methods[0]);
387 while(compression >= 0 && i < num_compression_methods) {
388 if (compression_methods[i].id == compression)
389 return compression_methods[i].name;
395 /* Convert an MSDOS-style date/time into Unix-style time. */
397 zip_time(const char *p)
402 msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]);
403 msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]);
405 memset(&ts, 0, sizeof(ts));
406 ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */
407 ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */
408 ts.tm_mday = msDate & 0x1f; /* Day of month. */
409 ts.tm_hour = (msTime >> 11) & 0x1f;
410 ts.tm_min = (msTime >> 5) & 0x3f;
411 ts.tm_sec = (msTime << 1) & 0x3e;
417 * The extra data is stored as a list of
418 * id1+size1+data1 + id2+size2+data2 ...
419 * triplets. id and size are 2 bytes each.
422 process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry)
426 if (extra_length == 0) {
430 if (extra_length < 4) {
431 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
432 "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length);
433 return ARCHIVE_FAILED;
435 while (offset <= extra_length - 4) {
436 unsigned short headerid = archive_le16dec(p + offset);
437 unsigned short datasize = archive_le16dec(p + offset + 2);
440 if (offset + datasize > extra_length) {
441 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
442 "Extra data overflow: Need %d bytes but only found %d bytes",
443 (int)datasize, (int)(extra_length - offset));
444 return ARCHIVE_FAILED;
447 fprintf(stderr, "Header id 0x%04x, length %d\n",
452 /* Zip64 extended information extra field. */
453 zip_entry->flags |= LA_USED_ZIP64;
454 if (zip_entry->uncompressed_size == 0xffffffff) {
457 zip_entry->uncompressed_size =
458 archive_le64dec(p + offset);
462 if (zip_entry->compressed_size == 0xffffffff) {
465 zip_entry->compressed_size =
466 archive_le64dec(p + offset);
470 if (zip_entry->local_header_offset == 0xffffffff) {
473 zip_entry->local_header_offset =
474 archive_le64dec(p + offset);
478 /* archive_le32dec(p + offset) gives disk
479 * on which file starts, but we don't handle
480 * multi-volume Zip files. */
485 /* Strong encryption field. */
486 if (archive_le16dec(p + offset) == 2) {
488 archive_le16dec(p + offset + 2);
490 archive_le16dec(p + offset + 4);
492 archive_le16dec(p + offset + 6);
493 fprintf(stderr, "algId=0x%04x, bitLen=%u, "
494 "flgas=%d\n", algId, bitLen,flags);
501 /* Extended time field "UT". */
502 int flags = p[offset];
505 /* Flag bits indicate which dates are present. */
509 fprintf(stderr, "mtime: %lld -> %d\n",
510 (long long)zip_entry->mtime,
511 archive_le32dec(p + offset));
515 zip_entry->mtime = archive_le32dec(p + offset);
523 zip_entry->atime = archive_le32dec(p + offset);
531 zip_entry->ctime = archive_le32dec(p + offset);
539 /* Info-ZIP Unix Extra Field (old version) "UX". */
541 zip_entry->atime = archive_le32dec(p + offset);
543 archive_le32dec(p + offset + 4);
545 if (datasize >= 12) {
547 archive_le16dec(p + offset + 8);
549 archive_le16dec(p + offset + 10);
555 /* Experimental 'xl' field */
557 * Introduced Dec 2013 to provide a way to
558 * include external file attributes (and other
559 * fields that ordinarily appear only in
560 * central directory) in local file header.
561 * This provides file type and permission
562 * information necessary to support full
563 * streaming extraction. Currently being
564 * discussed with other Zip developers
565 * ... subject to change.
568 * The field starts with a bitmap that specifies
569 * which additional fields are included. The
570 * bitmap is variable length and can be extended in
573 * n bytes - feature bitmap: first byte has low-order
574 * 7 bits. If high-order bit is set, a subsequent
575 * byte holds the next 7 bits, etc.
577 * if bitmap & 1, 2 byte "version made by"
578 * if bitmap & 2, 2 byte "internal file attributes"
579 * if bitmap & 4, 4 byte "external file attributes"
580 * if bitmap & 8, 2 byte comment length + n byte comment
582 int bitmap, bitmap_last;
586 bitmap_last = bitmap = 0xff & p[offset];
590 /* We only support first 7 bits of bitmap; skip rest. */
591 while ((bitmap_last & 0x80) != 0
593 bitmap_last = p[offset];
599 /* 2 byte "version made by" */
603 = archive_le16dec(p + offset) >> 8;
608 /* 2 byte "internal file attributes" */
609 uint32_t internal_attributes;
613 = archive_le16dec(p + offset);
614 /* Not used by libarchive at present. */
615 (void)internal_attributes; /* UNUSED */
620 /* 4 byte "external file attributes" */
621 uint32_t external_attributes;
625 = archive_le32dec(p + offset);
626 if (zip_entry->system == 3) {
628 = external_attributes >> 16;
629 } else if (zip_entry->system == 0) {
630 // Interpret MSDOS directory bit
631 if (0x10 == (external_attributes & 0x10)) {
632 zip_entry->mode = AE_IFDIR | 0775;
634 zip_entry->mode = AE_IFREG | 0664;
636 if (0x01 == (external_attributes & 0x01)) {
637 // Read-only bit; strip write permissions
638 zip_entry->mode &= 0555;
647 /* 2 byte comment length + comment */
648 uint32_t comment_length;
652 = archive_le16dec(p + offset);
656 if (datasize < comment_length)
658 /* Comment is not supported by libarchive */
659 offset += comment_length;
660 datasize -= comment_length;
665 /* Info-ZIP Unix Extra Field (type 2) "Ux". */
667 fprintf(stderr, "uid %d gid %d\n",
668 archive_le16dec(p + offset),
669 archive_le16dec(p + offset + 2));
672 zip_entry->uid = archive_le16dec(p + offset);
675 archive_le16dec(p + offset + 2);
679 /* Info-Zip Unix Extra Field (type 3) "ux". */
680 int uidsize = 0, gidsize = 0;
682 /* TODO: support arbitrary uidsize/gidsize. */
683 if (datasize >= 1 && p[offset] == 1) {/* version=1 */
685 /* get a uid size. */
686 uidsize = 0xff & (int)p[offset+1];
691 else if (uidsize == 4 && datasize >= 6)
696 if (datasize >= (2 + uidsize + 3)) {
697 /* get a gid size. */
698 gidsize = 0xff & (int)p[offset+2+uidsize];
702 p+offset+2+uidsize+1);
703 else if (gidsize == 4 &&
704 datasize >= (2 + uidsize + 5))
707 p+offset+2+uidsize+1);
713 /* WinZIp AES extra data field. */
714 if (p[offset + 2] == 'A' && p[offset + 3] == 'E') {
715 /* Vendor version. */
716 zip_entry->aes_extra.vendor =
717 archive_le16dec(p + offset);
718 /* AES encryption strength. */
719 zip_entry->aes_extra.strength = p[offset + 4];
720 /* Actual compression method. */
721 zip_entry->aes_extra.compression =
730 if (offset != extra_length) {
731 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
732 "Malformed extra data: Consumed %d bytes of %d bytes",
733 (int)offset, (int)extra_length);
734 return ARCHIVE_FAILED;
740 * Assumes file pointer is at beginning of local file header.
743 zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry,
750 size_t len, filename_length, extra_length;
751 struct archive_string_conv *sconv;
752 struct zip_entry *zip_entry = zip->entry;
753 struct zip_entry zip_entry_central_dir;
754 int ret = ARCHIVE_OK;
757 /* Save a copy of the original for consistency checks. */
758 zip_entry_central_dir = *zip_entry;
760 zip->decompress_init = 0;
761 zip->end_of_entry = 0;
762 zip->entry_uncompressed_bytes_read = 0;
763 zip->entry_compressed_bytes_read = 0;
764 zip->entry_crc32 = zip->crc32func(0, NULL, 0);
766 /* Setup default conversion. */
767 if (zip->sconv == NULL && !zip->init_default_conversion) {
769 archive_string_default_conversion_for_read(&(a->archive));
770 zip->init_default_conversion = 1;
773 if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) {
774 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
775 "Truncated ZIP file header");
776 return (ARCHIVE_FATAL);
779 if (memcmp(p, "PK\003\004", 4) != 0) {
780 archive_set_error(&a->archive, -1, "Damaged Zip archive");
781 return ARCHIVE_FATAL;
784 zip_entry->system = p[5];
785 zip_entry->zip_flags = archive_le16dec(p + 6);
786 if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) {
787 zip->has_encrypted_entries = 1;
788 archive_entry_set_is_data_encrypted(entry, 1);
789 if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED &&
790 zip_entry->zip_flags & ZIP_ENCRYPTED &&
791 zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) {
792 archive_entry_set_is_metadata_encrypted(entry, 1);
793 return ARCHIVE_FATAL;
796 zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED);
797 zip_entry->compression = (char)archive_le16dec(p + 8);
798 zip_entry->mtime = zip_time(p + 10);
799 zip_entry->crc32 = archive_le32dec(p + 14);
800 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
801 zip_entry->decdat = p[11];
803 zip_entry->decdat = p[17];
804 zip_entry->compressed_size = archive_le32dec(p + 18);
805 zip_entry->uncompressed_size = archive_le32dec(p + 22);
806 filename_length = archive_le16dec(p + 26);
807 extra_length = archive_le16dec(p + 28);
809 __archive_read_consume(a, 30);
811 /* Read the filename. */
812 if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) {
813 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
814 "Truncated ZIP file header");
815 return (ARCHIVE_FATAL);
817 if (zip_entry->zip_flags & ZIP_UTF8_NAME) {
818 /* The filename is stored to be UTF-8. */
819 if (zip->sconv_utf8 == NULL) {
821 archive_string_conversion_from_charset(
822 &a->archive, "UTF-8", 1);
823 if (zip->sconv_utf8 == NULL)
824 return (ARCHIVE_FATAL);
826 sconv = zip->sconv_utf8;
827 } else if (zip->sconv != NULL)
830 sconv = zip->sconv_default;
832 if (archive_entry_copy_pathname_l(entry,
833 h, filename_length, sconv) != 0) {
834 if (errno == ENOMEM) {
835 archive_set_error(&a->archive, ENOMEM,
836 "Can't allocate memory for Pathname");
837 return (ARCHIVE_FATAL);
839 archive_set_error(&a->archive,
840 ARCHIVE_ERRNO_FILE_FORMAT,
841 "Pathname cannot be converted "
842 "from %s to current locale.",
843 archive_string_conversion_charset_name(sconv));
846 __archive_read_consume(a, filename_length);
848 /* Read the extra data. */
849 if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) {
850 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
851 "Truncated ZIP file header");
852 return (ARCHIVE_FATAL);
855 if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) {
856 return ARCHIVE_FATAL;
858 __archive_read_consume(a, extra_length);
860 /* Work around a bug in Info-Zip: When reading from a pipe, it
861 * stats the pipe instead of synthesizing a file entry. */
862 if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) {
863 zip_entry->mode &= ~ AE_IFMT;
864 zip_entry->mode |= AE_IFREG;
867 if ((zip_entry->mode & AE_IFMT) == 0) {
868 /* Especially in streaming mode, we can end up
869 here without having seen proper mode information.
870 Guess from the filename. */
871 wp = archive_entry_pathname_w(entry);
874 if (len > 0 && wp[len - 1] == L'/')
875 zip_entry->mode |= AE_IFDIR;
877 zip_entry->mode |= AE_IFREG;
879 cp = archive_entry_pathname(entry);
880 len = (cp != NULL)?strlen(cp):0;
881 if (len > 0 && cp[len - 1] == '/')
882 zip_entry->mode |= AE_IFDIR;
884 zip_entry->mode |= AE_IFREG;
886 if (zip_entry->mode == AE_IFDIR) {
887 zip_entry->mode |= 0775;
888 } else if (zip_entry->mode == AE_IFREG) {
889 zip_entry->mode |= 0664;
893 /* Make sure directories end in '/' */
894 if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) {
895 wp = archive_entry_pathname_w(entry);
898 if (len > 0 && wp[len - 1] != L'/') {
899 struct archive_wstring s;
900 archive_string_init(&s);
901 archive_wstrcat(&s, wp);
902 archive_wstrappend_wchar(&s, L'/');
903 archive_entry_copy_pathname_w(entry, s.s);
906 cp = archive_entry_pathname(entry);
907 len = (cp != NULL)?strlen(cp):0;
908 if (len > 0 && cp[len - 1] != '/') {
909 struct archive_string s;
910 archive_string_init(&s);
911 archive_strcat(&s, cp);
912 archive_strappend_char(&s, '/');
913 archive_entry_set_pathname(entry, s.s);
918 if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) {
919 /* If this came from the central dir, it's size info
920 * is definitive, so ignore the length-at-end flag. */
921 zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END;
922 /* If local header is missing a value, use the one from
923 the central directory. If both have it, warn about
925 if (zip_entry->crc32 == 0) {
926 zip_entry->crc32 = zip_entry_central_dir.crc32;
927 } else if (!zip->ignore_crc32
928 && zip_entry->crc32 != zip_entry_central_dir.crc32) {
929 archive_set_error(&a->archive,
930 ARCHIVE_ERRNO_FILE_FORMAT,
931 "Inconsistent CRC32 values");
934 if (zip_entry->compressed_size == 0) {
935 zip_entry->compressed_size
936 = zip_entry_central_dir.compressed_size;
937 } else if (zip_entry->compressed_size
938 != zip_entry_central_dir.compressed_size) {
939 archive_set_error(&a->archive,
940 ARCHIVE_ERRNO_FILE_FORMAT,
941 "Inconsistent compressed size: "
942 "%jd in central directory, %jd in local header",
943 (intmax_t)zip_entry_central_dir.compressed_size,
944 (intmax_t)zip_entry->compressed_size);
947 if (zip_entry->uncompressed_size == 0) {
948 zip_entry->uncompressed_size
949 = zip_entry_central_dir.uncompressed_size;
950 } else if (zip_entry->uncompressed_size
951 != zip_entry_central_dir.uncompressed_size) {
952 archive_set_error(&a->archive,
953 ARCHIVE_ERRNO_FILE_FORMAT,
954 "Inconsistent uncompressed size: "
955 "%jd in central directory, %jd in local header",
956 (intmax_t)zip_entry_central_dir.uncompressed_size,
957 (intmax_t)zip_entry->uncompressed_size);
962 /* Populate some additional entry fields: */
963 archive_entry_set_mode(entry, zip_entry->mode);
964 archive_entry_set_uid(entry, zip_entry->uid);
965 archive_entry_set_gid(entry, zip_entry->gid);
966 archive_entry_set_mtime(entry, zip_entry->mtime, 0);
967 archive_entry_set_ctime(entry, zip_entry->ctime, 0);
968 archive_entry_set_atime(entry, zip_entry->atime, 0);
970 if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) {
971 size_t linkname_length;
973 if (zip_entry->compressed_size > 64 * 1024) {
974 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
975 "Zip file with oversized link entry");
976 return ARCHIVE_FATAL;
979 linkname_length = (size_t)zip_entry->compressed_size;
981 archive_entry_set_size(entry, 0);
982 p = __archive_read_ahead(a, linkname_length, NULL);
984 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
985 "Truncated Zip file");
986 return ARCHIVE_FATAL;
990 if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME))
991 sconv = zip->sconv_utf8;
993 sconv = zip->sconv_default;
994 if (archive_entry_copy_symlink_l(entry, p, linkname_length,
996 if (errno != ENOMEM && sconv == zip->sconv_utf8 &&
997 (zip->entry->zip_flags & ZIP_UTF8_NAME))
998 archive_entry_copy_symlink_l(entry, p,
999 linkname_length, NULL);
1000 if (errno == ENOMEM) {
1001 archive_set_error(&a->archive, ENOMEM,
1002 "Can't allocate memory for Symlink");
1003 return (ARCHIVE_FATAL);
1006 * Since there is no character-set regulation for
1007 * symlink name, do not report the conversion error
1008 * in an automatic conversion.
1010 if (sconv != zip->sconv_utf8 ||
1011 (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) {
1012 archive_set_error(&a->archive,
1013 ARCHIVE_ERRNO_FILE_FORMAT,
1014 "Symlink cannot be converted "
1015 "from %s to current locale.",
1016 archive_string_conversion_charset_name(
1021 zip_entry->uncompressed_size = zip_entry->compressed_size = 0;
1023 if (__archive_read_consume(a, linkname_length) < 0) {
1024 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1025 "Read error skipping symlink target name");
1026 return ARCHIVE_FATAL;
1028 } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1029 || zip_entry->uncompressed_size > 0) {
1030 /* Set the size only if it's meaningful. */
1031 archive_entry_set_size(entry, zip_entry->uncompressed_size);
1033 zip->entry_bytes_remaining = zip_entry->compressed_size;
1035 /* If there's no body, force read_data() to return EOF immediately. */
1036 if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
1037 && zip->entry_bytes_remaining < 1)
1038 zip->end_of_entry = 1;
1040 /* Set up a more descriptive format name. */
1041 archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)",
1042 version / 10, version % 10,
1043 compression_name(zip->entry->compression));
1044 a->archive.archive_format_name = zip->format_name.s;
1050 check_authentication_code(struct archive_read *a, const void *_p)
1052 struct zip *zip = (struct zip *)(a->format->data);
1054 /* Check authentication code. */
1055 if (zip->hctx_valid) {
1058 size_t hmac_len = 20;
1061 archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len);
1063 /* Read authentication code. */
1064 p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL);
1066 archive_set_error(&a->archive,
1067 ARCHIVE_ERRNO_FILE_FORMAT,
1068 "Truncated ZIP file data");
1069 return (ARCHIVE_FATAL);
1074 cmp = memcmp(hmac, p, AUTH_CODE_SIZE);
1075 __archive_read_consume(a, AUTH_CODE_SIZE);
1077 archive_set_error(&a->archive,
1079 "ZIP bad Authentication code");
1080 return (ARCHIVE_WARN);
1083 return (ARCHIVE_OK);
1087 * Read "uncompressed" data. There are three cases:
1088 * 1) We know the size of the data. This is always true for the
1089 * seeking reader (we've examined the Central Directory already).
1090 * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred.
1091 * Info-ZIP seems to do this; we know the size but have to grab
1092 * the CRC from the data descriptor afterwards.
1093 * 3) We're streaming and ZIP_LENGTH_AT_END was specified and
1094 * we have no size information. In this case, we can do pretty
1095 * well by watching for the data descriptor record. The data
1096 * descriptor is 16 bytes and includes a computed CRC that should
1097 * provide a strong check.
1099 * TODO: Technically, the PK\007\010 signature is optional.
1100 * In the original spec, the data descriptor contained CRC
1101 * and size fields but had no leading signature. In practice,
1102 * newer writers seem to provide the signature pretty consistently.
1104 * For uncompressed data, the PK\007\010 marker seems essential
1105 * to be sure we've actually seen the end of the entry.
1107 * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets
1108 * zip->end_of_entry if it consumes all of the data.
1111 zip_read_data_none(struct archive_read *a, const void **_buff,
1112 size_t *size, int64_t *offset)
1116 ssize_t bytes_avail;
1119 (void)offset; /* UNUSED */
1121 zip = (struct zip *)(a->format->data);
1123 if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) {
1125 ssize_t grabbing_bytes = 24;
1127 if (zip->hctx_valid)
1128 grabbing_bytes += AUTH_CODE_SIZE;
1129 /* Grab at least 24 bytes. */
1130 buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail);
1131 if (bytes_avail < grabbing_bytes) {
1132 /* Zip archives have end-of-archive markers
1133 that are longer than this, so a failure to get at
1134 least 24 bytes really does indicate a truncated
1136 archive_set_error(&a->archive,
1137 ARCHIVE_ERRNO_FILE_FORMAT,
1138 "Truncated ZIP file data");
1139 return (ARCHIVE_FATAL);
1141 /* Check for a complete PK\007\010 signature, followed
1142 * by the correct 4-byte CRC. */
1144 if (zip->hctx_valid)
1145 p += AUTH_CODE_SIZE;
1146 if (p[0] == 'P' && p[1] == 'K'
1147 && p[2] == '\007' && p[3] == '\010'
1148 && (archive_le32dec(p + 4) == zip->entry_crc32
1149 || zip->ignore_crc32
1151 && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) {
1152 if (zip->entry->flags & LA_USED_ZIP64) {
1153 zip->entry->crc32 = archive_le32dec(p + 4);
1154 zip->entry->compressed_size =
1155 archive_le64dec(p + 8);
1156 zip->entry->uncompressed_size =
1157 archive_le64dec(p + 16);
1158 zip->unconsumed = 24;
1160 zip->entry->crc32 = archive_le32dec(p + 4);
1161 zip->entry->compressed_size =
1162 archive_le32dec(p + 8);
1163 zip->entry->uncompressed_size =
1164 archive_le32dec(p + 12);
1165 zip->unconsumed = 16;
1167 if (zip->hctx_valid) {
1168 r = check_authentication_code(a, buff);
1169 if (r != ARCHIVE_OK)
1172 zip->end_of_entry = 1;
1173 return (ARCHIVE_OK);
1175 /* If not at EOF, ensure we consume at least one byte. */
1178 /* Scan forward until we see where a PK\007\010 signature
1180 /* Return bytes up until that point. On the next call,
1181 * the code above will verify the data descriptor. */
1182 while (p < buff + bytes_avail - 4) {
1183 if (p[3] == 'P') { p += 3; }
1184 else if (p[3] == 'K') { p += 2; }
1185 else if (p[3] == '\007') { p += 1; }
1186 else if (p[3] == '\010' && p[2] == '\007'
1187 && p[1] == 'K' && p[0] == 'P') {
1188 if (zip->hctx_valid)
1189 p -= AUTH_CODE_SIZE;
1193 bytes_avail = p - buff;
1195 if (zip->entry_bytes_remaining == 0) {
1196 zip->end_of_entry = 1;
1197 if (zip->hctx_valid) {
1198 r = check_authentication_code(a, NULL);
1199 if (r != ARCHIVE_OK)
1202 return (ARCHIVE_OK);
1204 /* Grab a bunch of bytes. */
1205 buff = __archive_read_ahead(a, 1, &bytes_avail);
1206 if (bytes_avail <= 0) {
1207 archive_set_error(&a->archive,
1208 ARCHIVE_ERRNO_FILE_FORMAT,
1209 "Truncated ZIP file data");
1210 return (ARCHIVE_FATAL);
1212 if (bytes_avail > zip->entry_bytes_remaining)
1213 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
1215 if (zip->tctx_valid || zip->cctx_valid) {
1216 size_t dec_size = bytes_avail;
1218 if (dec_size > zip->decrypted_buffer_size)
1219 dec_size = zip->decrypted_buffer_size;
1220 if (zip->tctx_valid) {
1221 trad_enc_decrypt_update(&zip->tctx,
1222 (const uint8_t *)buff, dec_size,
1223 zip->decrypted_buffer, dec_size);
1225 size_t dsize = dec_size;
1226 archive_hmac_sha1_update(&zip->hctx,
1227 (const uint8_t *)buff, dec_size);
1228 archive_decrypto_aes_ctr_update(&zip->cctx,
1229 (const uint8_t *)buff, dec_size,
1230 zip->decrypted_buffer, &dsize);
1232 bytes_avail = dec_size;
1233 buff = (const char *)zip->decrypted_buffer;
1235 *size = bytes_avail;
1236 zip->entry_bytes_remaining -= bytes_avail;
1237 zip->entry_uncompressed_bytes_read += bytes_avail;
1238 zip->entry_compressed_bytes_read += bytes_avail;
1239 zip->unconsumed += bytes_avail;
1241 return (ARCHIVE_OK);
1246 zip_deflate_init(struct archive_read *a, struct zip *zip)
1250 /* If we haven't yet read any data, initialize the decompressor. */
1251 if (!zip->decompress_init) {
1252 if (zip->stream_valid)
1253 r = inflateReset(&zip->stream);
1255 r = inflateInit2(&zip->stream,
1256 -15 /* Don't check for zlib header */);
1258 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1259 "Can't initialize ZIP decompression.");
1260 return (ARCHIVE_FATAL);
1262 /* Stream structure has been set up. */
1263 zip->stream_valid = 1;
1264 /* We've initialized decompression for this stream. */
1265 zip->decompress_init = 1;
1267 return (ARCHIVE_OK);
1271 zip_read_data_deflate(struct archive_read *a, const void **buff,
1272 size_t *size, int64_t *offset)
1275 ssize_t bytes_avail;
1276 const void *compressed_buff, *sp;
1279 (void)offset; /* UNUSED */
1281 zip = (struct zip *)(a->format->data);
1283 /* If the buffer hasn't been allocated, allocate it now. */
1284 if (zip->uncompressed_buffer == NULL) {
1285 zip->uncompressed_buffer_size = 256 * 1024;
1286 zip->uncompressed_buffer
1287 = (unsigned char *)malloc(zip->uncompressed_buffer_size);
1288 if (zip->uncompressed_buffer == NULL) {
1289 archive_set_error(&a->archive, ENOMEM,
1290 "No memory for ZIP decompression");
1291 return (ARCHIVE_FATAL);
1295 r = zip_deflate_init(a, zip);
1296 if (r != ARCHIVE_OK)
1300 * Note: '1' here is a performance optimization.
1301 * Recall that the decompression layer returns a count of
1302 * available bytes; asking for more than that forces the
1303 * decompressor to combine reads by copying data.
1305 compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail);
1306 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
1307 && bytes_avail > zip->entry_bytes_remaining) {
1308 bytes_avail = (ssize_t)zip->entry_bytes_remaining;
1310 if (bytes_avail < 0) {
1311 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1312 "Truncated ZIP file body");
1313 return (ARCHIVE_FATAL);
1316 if (zip->tctx_valid || zip->cctx_valid) {
1317 if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) {
1318 size_t buff_remaining =
1319 (zip->decrypted_buffer + zip->decrypted_buffer_size)
1320 - (zip->decrypted_ptr + zip->decrypted_bytes_remaining);
1322 if (buff_remaining > (size_t)bytes_avail)
1323 buff_remaining = (size_t)bytes_avail;
1325 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) &&
1326 zip->entry_bytes_remaining > 0) {
1327 if ((int64_t)(zip->decrypted_bytes_remaining
1329 > zip->entry_bytes_remaining) {
1330 if (zip->entry_bytes_remaining <
1331 (int64_t)zip->decrypted_bytes_remaining)
1335 (size_t)zip->entry_bytes_remaining
1336 - zip->decrypted_bytes_remaining;
1339 if (buff_remaining > 0) {
1340 if (zip->tctx_valid) {
1341 trad_enc_decrypt_update(&zip->tctx,
1342 compressed_buff, buff_remaining,
1344 + zip->decrypted_bytes_remaining,
1347 size_t dsize = buff_remaining;
1348 archive_decrypto_aes_ctr_update(
1350 compressed_buff, buff_remaining,
1352 + zip->decrypted_bytes_remaining,
1355 zip->decrypted_bytes_remaining += buff_remaining;
1358 bytes_avail = zip->decrypted_bytes_remaining;
1359 compressed_buff = (const char *)zip->decrypted_ptr;
1363 * A bug in zlib.h: stream.next_in should be marked 'const'
1364 * but isn't (the library never alters data through the
1365 * next_in pointer, only reads it). The result: this ugly
1366 * cast to remove 'const'.
1368 zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff;
1369 zip->stream.avail_in = (uInt)bytes_avail;
1370 zip->stream.total_in = 0;
1371 zip->stream.next_out = zip->uncompressed_buffer;
1372 zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size;
1373 zip->stream.total_out = 0;
1375 r = inflate(&zip->stream, 0);
1380 zip->end_of_entry = 1;
1383 archive_set_error(&a->archive, ENOMEM,
1384 "Out of memory for ZIP decompression");
1385 return (ARCHIVE_FATAL);
1387 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1388 "ZIP decompression failed (%d)", r);
1389 return (ARCHIVE_FATAL);
1392 /* Consume as much as the compressor actually used. */
1393 bytes_avail = zip->stream.total_in;
1394 if (zip->tctx_valid || zip->cctx_valid) {
1395 zip->decrypted_bytes_remaining -= bytes_avail;
1396 if (zip->decrypted_bytes_remaining == 0)
1397 zip->decrypted_ptr = zip->decrypted_buffer;
1399 zip->decrypted_ptr += bytes_avail;
1401 /* Calculate compressed data as much as we used.*/
1402 if (zip->hctx_valid)
1403 archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail);
1404 __archive_read_consume(a, bytes_avail);
1405 zip->entry_bytes_remaining -= bytes_avail;
1406 zip->entry_compressed_bytes_read += bytes_avail;
1408 *size = zip->stream.total_out;
1409 zip->entry_uncompressed_bytes_read += zip->stream.total_out;
1410 *buff = zip->uncompressed_buffer;
1412 if (zip->end_of_entry && zip->hctx_valid) {
1413 r = check_authentication_code(a, NULL);
1414 if (r != ARCHIVE_OK)
1418 if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
1421 if (NULL == (p = __archive_read_ahead(a, 24, NULL))) {
1422 archive_set_error(&a->archive,
1423 ARCHIVE_ERRNO_FILE_FORMAT,
1424 "Truncated ZIP end-of-file record");
1425 return (ARCHIVE_FATAL);
1427 /* Consume the optional PK\007\010 marker. */
1428 if (p[0] == 'P' && p[1] == 'K' &&
1429 p[2] == '\007' && p[3] == '\010') {
1431 zip->unconsumed = 4;
1433 if (zip->entry->flags & LA_USED_ZIP64) {
1434 zip->entry->crc32 = archive_le32dec(p);
1435 zip->entry->compressed_size = archive_le64dec(p + 4);
1436 zip->entry->uncompressed_size = archive_le64dec(p + 12);
1437 zip->unconsumed += 20;
1439 zip->entry->crc32 = archive_le32dec(p);
1440 zip->entry->compressed_size = archive_le32dec(p + 4);
1441 zip->entry->uncompressed_size = archive_le32dec(p + 8);
1442 zip->unconsumed += 12;
1446 return (ARCHIVE_OK);
1451 read_decryption_header(struct archive_read *a)
1453 struct zip *zip = (struct zip *)(a->format->data);
1455 unsigned int remaining_size;
1459 * Read an initialization vector data field.
1461 p = __archive_read_ahead(a, 2, NULL);
1465 zip->iv_size = archive_le16dec(p);
1466 __archive_read_consume(a, 2);
1467 if (ts < zip->iv_size) {
1471 p = __archive_read_ahead(a, zip->iv_size, NULL);
1474 if (zip->iv == NULL) {
1475 zip->iv = malloc(zip->iv_size);
1476 if (zip->iv == NULL)
1479 memcpy(zip->iv, p, zip->iv_size);
1480 __archive_read_consume(a, zip->iv_size);
1483 * Read a size of remaining decryption header field.
1485 p = __archive_read_ahead(a, 14, NULL);
1488 remaining_size = archive_le32dec(p);
1489 if (remaining_size < 16 || remaining_size > (1 << 18))
1492 /* Check if format version is supported. */
1493 if (archive_le16dec(p+4) != 3) {
1494 archive_set_error(&a->archive,
1495 ARCHIVE_ERRNO_FILE_FORMAT,
1496 "Unsupported encryption format version: %u",
1497 archive_le16dec(p+4));
1498 return (ARCHIVE_FAILED);
1502 * Read an encryption algorithm field.
1504 zip->alg_id = archive_le16dec(p+6);
1505 switch (zip->alg_id) {
1506 case 0x6601:/* DES */
1507 case 0x6602:/* RC2 */
1508 case 0x6603:/* 3DES 168 */
1509 case 0x6609:/* 3DES 112 */
1510 case 0x660E:/* AES 128 */
1511 case 0x660F:/* AES 192 */
1512 case 0x6610:/* AES 256 */
1513 case 0x6702:/* RC2 (version >= 5.2) */
1514 case 0x6720:/* Blowfish */
1515 case 0x6721:/* Twofish */
1516 case 0x6801:/* RC4 */
1517 /* Suuported encryption algorithm. */
1520 archive_set_error(&a->archive,
1521 ARCHIVE_ERRNO_FILE_FORMAT,
1522 "Unknown encryption algorithm: %u", zip->alg_id);
1523 return (ARCHIVE_FAILED);
1527 * Read a bit length field.
1529 zip->bit_len = archive_le16dec(p+8);
1532 * Read a flags field.
1534 zip->flags = archive_le16dec(p+10);
1535 switch (zip->flags & 0xf000) {
1536 case 0x0001: /* Password is required to decrypt. */
1537 case 0x0002: /* Certificates only. */
1538 case 0x0003: /* Password or certificate required to decrypt. */
1541 archive_set_error(&a->archive,
1542 ARCHIVE_ERRNO_FILE_FORMAT,
1543 "Unknown encryption flag: %u", zip->flags);
1544 return (ARCHIVE_FAILED);
1546 if ((zip->flags & 0xf000) == 0 ||
1547 (zip->flags & 0xf000) == 0x4000) {
1548 archive_set_error(&a->archive,
1549 ARCHIVE_ERRNO_FILE_FORMAT,
1550 "Unknown encryption flag: %u", zip->flags);
1551 return (ARCHIVE_FAILED);
1555 * Read an encrypted random data field.
1558 zip->erd_size = archive_le16dec(p+12);
1559 __archive_read_consume(a, 14);
1560 if ((zip->erd_size & 0xf) != 0 ||
1561 (zip->erd_size + 16) > remaining_size ||
1562 (zip->erd_size + 16) < zip->erd_size)
1565 if (ts < zip->erd_size) {
1569 p = __archive_read_ahead(a, zip->erd_size, NULL);
1572 if (zip->erd == NULL) {
1573 zip->erd = malloc(zip->erd_size);
1574 if (zip->erd == NULL)
1577 memcpy(zip->erd, p, zip->erd_size);
1578 __archive_read_consume(a, zip->erd_size);
1581 * Read a reserved data field.
1583 p = __archive_read_ahead(a, 4, NULL);
1586 /* Reserved data size should be zero. */
1587 if (archive_le32dec(p) != 0)
1589 __archive_read_consume(a, 4);
1592 * Read a password validation data field.
1594 p = __archive_read_ahead(a, 2, NULL);
1598 zip->v_size = archive_le16dec(p);
1599 __archive_read_consume(a, 2);
1600 if ((zip->v_size & 0x0f) != 0 ||
1601 (zip->erd_size + zip->v_size + 16) > remaining_size ||
1602 (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size))
1604 if (ts < zip->v_size) {
1608 p = __archive_read_ahead(a, zip->v_size, NULL);
1611 if (zip->v_data == NULL) {
1612 zip->v_data = malloc(zip->v_size);
1613 if (zip->v_data == NULL)
1616 memcpy(zip->v_data, p, zip->v_size);
1617 __archive_read_consume(a, zip->v_size);
1619 p = __archive_read_ahead(a, 4, NULL);
1622 zip->v_crc32 = archive_le32dec(p);
1623 __archive_read_consume(a, 4);
1625 /*return (ARCHIVE_OK);
1626 * This is not fully implemnted yet.*/
1627 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1628 "Encrypted file is unsupported");
1629 return (ARCHIVE_FAILED);
1631 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1632 "Truncated ZIP file data");
1633 return (ARCHIVE_FATAL);
1635 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1636 "Corrupted ZIP file data");
1637 return (ARCHIVE_FATAL);
1639 archive_set_error(&a->archive, ENOMEM,
1640 "No memory for ZIP decryption");
1641 return (ARCHIVE_FATAL);
1645 zip_alloc_decryption_buffer(struct archive_read *a)
1647 struct zip *zip = (struct zip *)(a->format->data);
1648 size_t bs = 256 * 1024;
1650 if (zip->decrypted_buffer == NULL) {
1651 zip->decrypted_buffer_size = bs;
1652 zip->decrypted_buffer = malloc(bs);
1653 if (zip->decrypted_buffer == NULL) {
1654 archive_set_error(&a->archive, ENOMEM,
1655 "No memory for ZIP decryption");
1656 return (ARCHIVE_FATAL);
1659 zip->decrypted_ptr = zip->decrypted_buffer;
1660 return (ARCHIVE_OK);
1664 init_traditional_PKWARE_decryption(struct archive_read *a)
1666 struct zip *zip = (struct zip *)(a->format->data);
1671 if (zip->tctx_valid)
1672 return (ARCHIVE_OK);
1675 Read the 12 bytes encryption header stored at
1676 the start of the data area.
1678 #define ENC_HEADER_SIZE 12
1679 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
1680 && zip->entry_bytes_remaining < ENC_HEADER_SIZE) {
1681 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1682 "Truncated Zip encrypted body: only %jd bytes available",
1683 (intmax_t)zip->entry_bytes_remaining);
1684 return (ARCHIVE_FATAL);
1687 p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL);
1689 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1690 "Truncated ZIP file data");
1691 return (ARCHIVE_FATAL);
1694 for (retry = 0;; retry++) {
1695 const char *passphrase;
1698 passphrase = __archive_read_next_passphrase(a);
1699 if (passphrase == NULL) {
1700 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1702 "Incorrect passphrase":
1703 "Passphrase required for this entry");
1704 return (ARCHIVE_FAILED);
1708 * Initialize ctx for Traditional PKWARE Decyption.
1710 r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase),
1711 p, ENC_HEADER_SIZE, &crcchk);
1712 if (r == 0 && crcchk == zip->entry->decdat)
1713 break;/* The passphrase is OK. */
1714 if (retry > 10000) {
1715 /* Avoid infinity loop. */
1716 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1717 "Too many incorrect passphrases");
1718 return (ARCHIVE_FAILED);
1722 __archive_read_consume(a, ENC_HEADER_SIZE);
1723 zip->tctx_valid = 1;
1724 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) {
1725 zip->entry_bytes_remaining -= ENC_HEADER_SIZE;
1727 /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/
1728 zip->entry_compressed_bytes_read += ENC_HEADER_SIZE;
1729 zip->decrypted_bytes_remaining = 0;
1731 return (zip_alloc_decryption_buffer(a));
1732 #undef ENC_HEADER_SIZE
1736 init_WinZip_AES_decryption(struct archive_read *a)
1738 struct zip *zip = (struct zip *)(a->format->data);
1741 size_t key_len, salt_len;
1742 uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE];
1746 if (zip->cctx_valid || zip->hctx_valid)
1747 return (ARCHIVE_OK);
1749 switch (zip->entry->aes_extra.strength) {
1750 case 1: salt_len = 8; key_len = 16; break;
1751 case 2: salt_len = 12; key_len = 24; break;
1752 case 3: salt_len = 16; key_len = 32; break;
1753 default: goto corrupted;
1755 p = __archive_read_ahead(a, salt_len + 2, NULL);
1759 for (retry = 0;; retry++) {
1760 const char *passphrase;
1762 passphrase = __archive_read_next_passphrase(a);
1763 if (passphrase == NULL) {
1764 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1766 "Incorrect passphrase":
1767 "Passphrase required for this entry");
1768 return (ARCHIVE_FAILED);
1770 memset(derived_key, 0, sizeof(derived_key));
1771 r = archive_pbkdf2_sha1(passphrase, strlen(passphrase),
1772 p, salt_len, 1000, derived_key, key_len * 2 + 2);
1774 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1775 "Decryption is unsupported due to lack of "
1777 return (ARCHIVE_FAILED);
1780 /* Check password verification value. */
1781 pv = ((const uint8_t *)p) + salt_len;
1782 if (derived_key[key_len * 2] == pv[0] &&
1783 derived_key[key_len * 2 + 1] == pv[1])
1784 break;/* The passphrase is OK. */
1785 if (retry > 10000) {
1786 /* Avoid infinity loop. */
1787 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1788 "Too many incorrect passphrases");
1789 return (ARCHIVE_FAILED);
1793 r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len);
1795 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1796 "Decryption is unsupported due to lack of crypto library");
1797 return (ARCHIVE_FAILED);
1799 r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len);
1801 archive_decrypto_aes_ctr_release(&zip->cctx);
1802 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1803 "Failed to initialize HMAC-SHA1");
1804 return (ARCHIVE_FAILED);
1806 zip->cctx_valid = zip->hctx_valid = 1;
1807 __archive_read_consume(a, salt_len + 2);
1808 zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE;
1809 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
1810 && zip->entry_bytes_remaining < 0)
1812 zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE;
1813 zip->decrypted_bytes_remaining = 0;
1815 zip->entry->compression = zip->entry->aes_extra.compression;
1816 return (zip_alloc_decryption_buffer(a));
1819 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1820 "Truncated ZIP file data");
1821 return (ARCHIVE_FATAL);
1823 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1824 "Corrupted ZIP file data");
1825 return (ARCHIVE_FATAL);
1829 archive_read_format_zip_read_data(struct archive_read *a,
1830 const void **buff, size_t *size, int64_t *offset)
1833 struct zip *zip = (struct zip *)(a->format->data);
1835 if (zip->has_encrypted_entries ==
1836 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) {
1837 zip->has_encrypted_entries = 0;
1840 *offset = zip->entry_uncompressed_bytes_read;
1844 /* If we hit end-of-entry last time, return ARCHIVE_EOF. */
1845 if (zip->end_of_entry)
1846 return (ARCHIVE_EOF);
1848 /* Return EOF immediately if this is a non-regular file. */
1849 if (AE_IFREG != (zip->entry->mode & AE_IFMT))
1850 return (ARCHIVE_EOF);
1852 __archive_read_consume(a, zip->unconsumed);
1853 zip->unconsumed = 0;
1855 if (zip->init_decryption) {
1856 zip->has_encrypted_entries = 1;
1857 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
1858 r = read_decryption_header(a);
1859 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
1860 r = init_WinZip_AES_decryption(a);
1862 r = init_traditional_PKWARE_decryption(a);
1863 if (r != ARCHIVE_OK)
1865 zip->init_decryption = 0;
1868 switch(zip->entry->compression) {
1869 case 0: /* No compression. */
1870 r = zip_read_data_none(a, buff, size, offset);
1873 case 8: /* Deflate compression. */
1874 r = zip_read_data_deflate(a, buff, size, offset);
1877 default: /* Unsupported compression. */
1878 /* Return a warning. */
1879 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1880 "Unsupported ZIP compression method (%s)",
1881 compression_name(zip->entry->compression));
1882 /* We can't decompress this entry, but we will
1883 * be able to skip() it and try the next entry. */
1884 return (ARCHIVE_FAILED);
1887 if (r != ARCHIVE_OK)
1889 /* Update checksum */
1891 zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff,
1893 /* If we hit the end, swallow any end-of-data marker. */
1894 if (zip->end_of_entry) {
1895 /* Check file size, CRC against these values. */
1896 if (zip->entry->compressed_size !=
1897 zip->entry_compressed_bytes_read) {
1898 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1899 "ZIP compressed data is wrong size "
1900 "(read %jd, expected %jd)",
1901 (intmax_t)zip->entry_compressed_bytes_read,
1902 (intmax_t)zip->entry->compressed_size);
1903 return (ARCHIVE_WARN);
1905 /* Size field only stores the lower 32 bits of the actual
1907 if ((zip->entry->uncompressed_size & UINT32_MAX)
1908 != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) {
1909 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1910 "ZIP uncompressed data is wrong size "
1911 "(read %jd, expected %jd)\n",
1912 (intmax_t)zip->entry_uncompressed_bytes_read,
1913 (intmax_t)zip->entry->uncompressed_size);
1914 return (ARCHIVE_WARN);
1916 /* Check computed CRC against header */
1917 if ((!zip->hctx_valid ||
1918 zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) &&
1919 zip->entry->crc32 != zip->entry_crc32
1920 && !zip->ignore_crc32) {
1921 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1922 "ZIP bad CRC: 0x%lx should be 0x%lx",
1923 (unsigned long)zip->entry_crc32,
1924 (unsigned long)zip->entry->crc32);
1925 return (ARCHIVE_WARN);
1929 return (ARCHIVE_OK);
1933 archive_read_format_zip_cleanup(struct archive_read *a)
1936 struct zip_entry *zip_entry, *next_zip_entry;
1938 zip = (struct zip *)(a->format->data);
1940 if (zip->stream_valid)
1941 inflateEnd(&zip->stream);
1942 free(zip->uncompressed_buffer);
1944 if (zip->zip_entries) {
1945 zip_entry = zip->zip_entries;
1946 while (zip_entry != NULL) {
1947 next_zip_entry = zip_entry->next;
1948 archive_string_free(&zip_entry->rsrcname);
1950 zip_entry = next_zip_entry;
1953 free(zip->decrypted_buffer);
1954 if (zip->cctx_valid)
1955 archive_decrypto_aes_ctr_release(&zip->cctx);
1956 if (zip->hctx_valid)
1957 archive_hmac_sha1_cleanup(&zip->hctx);
1961 archive_string_free(&zip->format_name);
1963 (a->format->data) = NULL;
1964 return (ARCHIVE_OK);
1968 archive_read_format_zip_has_encrypted_entries(struct archive_read *_a)
1970 if (_a && _a->format) {
1971 struct zip * zip = (struct zip *)_a->format->data;
1973 return zip->has_encrypted_entries;
1976 return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
1980 archive_read_format_zip_options(struct archive_read *a,
1981 const char *key, const char *val)
1984 int ret = ARCHIVE_FAILED;
1986 zip = (struct zip *)(a->format->data);
1987 if (strcmp(key, "compat-2x") == 0) {
1988 /* Handle filenames as libarchive 2.x */
1989 zip->init_default_conversion = (val != NULL) ? 1 : 0;
1990 return (ARCHIVE_OK);
1991 } else if (strcmp(key, "hdrcharset") == 0) {
1992 if (val == NULL || val[0] == 0)
1993 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1994 "zip: hdrcharset option needs a character-set name"
1997 zip->sconv = archive_string_conversion_from_charset(
1998 &a->archive, val, 0);
1999 if (zip->sconv != NULL) {
2000 if (strcmp(val, "UTF-8") == 0)
2001 zip->sconv_utf8 = zip->sconv;
2004 ret = ARCHIVE_FATAL;
2007 } else if (strcmp(key, "ignorecrc32") == 0) {
2008 /* Mostly useful for testing. */
2009 if (val == NULL || val[0] == 0) {
2010 zip->crc32func = real_crc32;
2011 zip->ignore_crc32 = 0;
2013 zip->crc32func = fake_crc32;
2014 zip->ignore_crc32 = 1;
2016 return (ARCHIVE_OK);
2017 } else if (strcmp(key, "mac-ext") == 0) {
2018 zip->process_mac_extensions = (val != NULL && val[0] != 0);
2019 return (ARCHIVE_OK);
2022 /* Note: The "warn" return is just to inform the options
2023 * supervisor that we didn't handle it. It will generate
2024 * a suitable error if no one used this option. */
2025 return (ARCHIVE_WARN);
2029 archive_read_support_format_zip(struct archive *a)
2032 r = archive_read_support_format_zip_streamable(a);
2033 if (r != ARCHIVE_OK)
2035 return (archive_read_support_format_zip_seekable(a));
2038 /* ------------------------------------------------------------------------ */
2041 * Streaming-mode support
2046 archive_read_support_format_zip_capabilities_streamable(struct archive_read * a)
2048 (void)a; /* UNUSED */
2049 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
2050 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
2054 archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid)
2058 (void)best_bid; /* UNUSED */
2060 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
2064 * Bid of 29 here comes from:
2065 * + 16 bits for "PK",
2066 * + next 16-bit field has 6 options so contributes
2067 * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits
2069 * So we've effectively verified ~29 total bits of check data.
2071 if (p[0] == 'P' && p[1] == 'K') {
2072 if ((p[2] == '\001' && p[3] == '\002')
2073 || (p[2] == '\003' && p[3] == '\004')
2074 || (p[2] == '\005' && p[3] == '\006')
2075 || (p[2] == '\006' && p[3] == '\006')
2076 || (p[2] == '\007' && p[3] == '\010')
2077 || (p[2] == '0' && p[3] == '0'))
2081 /* TODO: It's worth looking ahead a little bit for a valid
2082 * PK signature. In particular, that would make it possible
2083 * to read some UUEncoded SFX files or SFX files coming from
2084 * a network socket. */
2090 archive_read_format_zip_streamable_read_header(struct archive_read *a,
2091 struct archive_entry *entry)
2095 a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
2096 if (a->archive.archive_format_name == NULL)
2097 a->archive.archive_format_name = "ZIP";
2099 zip = (struct zip *)(a->format->data);
2102 * It should be sufficient to call archive_read_next_header() for
2103 * a reader to determine if an entry is encrypted or not. If the
2104 * encryption of an entry is only detectable when calling
2105 * archive_read_data(), so be it. We'll do the same check there
2108 if (zip->has_encrypted_entries ==
2109 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
2110 zip->has_encrypted_entries = 0;
2112 /* Make sure we have a zip_entry structure to use. */
2113 if (zip->zip_entries == NULL) {
2114 zip->zip_entries = malloc(sizeof(struct zip_entry));
2115 if (zip->zip_entries == NULL) {
2116 archive_set_error(&a->archive, ENOMEM,
2118 return ARCHIVE_FATAL;
2121 zip->entry = zip->zip_entries;
2122 memset(zip->entry, 0, sizeof(struct zip_entry));
2124 if (zip->cctx_valid)
2125 archive_decrypto_aes_ctr_release(&zip->cctx);
2126 if (zip->hctx_valid)
2127 archive_hmac_sha1_cleanup(&zip->hctx);
2128 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
2129 __archive_read_reset_passphrase(a);
2131 /* Search ahead for the next local file header. */
2132 __archive_read_consume(a, zip->unconsumed);
2133 zip->unconsumed = 0;
2135 int64_t skipped = 0;
2136 const char *p, *end;
2139 p = __archive_read_ahead(a, 4, &bytes);
2141 return (ARCHIVE_FATAL);
2144 while (p + 4 <= end) {
2145 if (p[0] == 'P' && p[1] == 'K') {
2146 if (p[2] == '\003' && p[3] == '\004') {
2147 /* Regular file entry. */
2148 __archive_read_consume(a, skipped);
2149 return zip_read_local_file_header(a,
2154 * TODO: We cannot restore permissions
2155 * based only on the local file headers.
2156 * Consider scanning the central
2157 * directory and returning additional
2158 * entries for at least directories.
2159 * This would allow us to properly set
2160 * directory permissions.
2162 * This won't help us fix symlinks
2163 * and may not help with regular file
2164 * permissions, either. <sigh>
2166 if (p[2] == '\001' && p[3] == '\002') {
2167 return (ARCHIVE_EOF);
2170 /* End of central directory? Must be an
2172 if ((p[2] == '\005' && p[3] == '\006')
2173 || (p[2] == '\006' && p[3] == '\006'))
2174 return (ARCHIVE_EOF);
2179 __archive_read_consume(a, skipped);
2184 archive_read_format_zip_read_data_skip_streamable(struct archive_read *a)
2187 int64_t bytes_skipped;
2189 zip = (struct zip *)(a->format->data);
2190 bytes_skipped = __archive_read_consume(a, zip->unconsumed);
2191 zip->unconsumed = 0;
2192 if (bytes_skipped < 0)
2193 return (ARCHIVE_FATAL);
2195 /* If we've already read to end of data, we're done. */
2196 if (zip->end_of_entry)
2197 return (ARCHIVE_OK);
2199 /* So we know we're streaming... */
2200 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)
2201 || zip->entry->compressed_size > 0) {
2202 /* We know the compressed length, so we can just skip. */
2203 bytes_skipped = __archive_read_consume(a,
2204 zip->entry_bytes_remaining);
2205 if (bytes_skipped < 0)
2206 return (ARCHIVE_FATAL);
2207 return (ARCHIVE_OK);
2210 if (zip->init_decryption) {
2213 zip->has_encrypted_entries = 1;
2214 if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED)
2215 r = read_decryption_header(a);
2216 else if (zip->entry->compression == WINZIP_AES_ENCRYPTION)
2217 r = init_WinZip_AES_decryption(a);
2219 r = init_traditional_PKWARE_decryption(a);
2220 if (r != ARCHIVE_OK)
2222 zip->init_decryption = 0;
2225 /* We're streaming and we don't know the length. */
2226 /* If the body is compressed and we know the format, we can
2227 * find an exact end-of-entry by decompressing it. */
2228 switch (zip->entry->compression) {
2230 case 8: /* Deflate compression. */
2231 while (!zip->end_of_entry) {
2233 const void *buff = NULL;
2236 r = zip_read_data_deflate(a, &buff, &size, &offset);
2237 if (r != ARCHIVE_OK)
2242 default: /* Uncompressed or unknown. */
2243 /* Scan for a PK\007\010 signature. */
2245 const char *p, *buff;
2246 ssize_t bytes_avail;
2247 buff = __archive_read_ahead(a, 16, &bytes_avail);
2248 if (bytes_avail < 16) {
2249 archive_set_error(&a->archive,
2250 ARCHIVE_ERRNO_FILE_FORMAT,
2251 "Truncated ZIP file data");
2252 return (ARCHIVE_FATAL);
2255 while (p <= buff + bytes_avail - 16) {
2256 if (p[3] == 'P') { p += 3; }
2257 else if (p[3] == 'K') { p += 2; }
2258 else if (p[3] == '\007') { p += 1; }
2259 else if (p[3] == '\010' && p[2] == '\007'
2260 && p[1] == 'K' && p[0] == 'P') {
2261 if (zip->entry->flags & LA_USED_ZIP64)
2262 __archive_read_consume(a,
2265 __archive_read_consume(a,
2270 __archive_read_consume(a, p - buff);
2276 archive_read_support_format_zip_streamable(struct archive *_a)
2278 struct archive_read *a = (struct archive_read *)_a;
2282 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
2283 ARCHIVE_STATE_NEW, "archive_read_support_format_zip");
2285 zip = (struct zip *)calloc(1, sizeof(*zip));
2287 archive_set_error(&a->archive, ENOMEM,
2288 "Can't allocate zip data");
2289 return (ARCHIVE_FATAL);
2292 /* Streamable reader doesn't support mac extensions. */
2293 zip->process_mac_extensions = 0;
2296 * Until enough data has been read, we cannot tell about
2297 * any encrypted entries yet.
2299 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
2300 zip->crc32func = real_crc32;
2302 r = __archive_read_register_format(a,
2305 archive_read_format_zip_streamable_bid,
2306 archive_read_format_zip_options,
2307 archive_read_format_zip_streamable_read_header,
2308 archive_read_format_zip_read_data,
2309 archive_read_format_zip_read_data_skip_streamable,
2311 archive_read_format_zip_cleanup,
2312 archive_read_support_format_zip_capabilities_streamable,
2313 archive_read_format_zip_has_encrypted_entries);
2315 if (r != ARCHIVE_OK)
2317 return (ARCHIVE_OK);
2320 /* ------------------------------------------------------------------------ */
2323 * Seeking-mode support
2327 archive_read_support_format_zip_capabilities_seekable(struct archive_read * a)
2329 (void)a; /* UNUSED */
2330 return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA |
2331 ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA);
2335 * TODO: This is a performance sink because it forces the read core to
2336 * drop buffered data from the start of file, which will then have to
2337 * be re-read again if this bidder loses.
2339 * We workaround this a little by passing in the best bid so far so
2340 * that later bidders can do nothing if they know they'll never
2341 * outbid. But we can certainly do better...
2344 read_eocd(struct zip *zip, const char *p, int64_t current_offset)
2346 /* Sanity-check the EOCD we've found. */
2348 /* This must be the first volume. */
2349 if (archive_le16dec(p + 4) != 0)
2351 /* Central directory must be on this volume. */
2352 if (archive_le16dec(p + 4) != archive_le16dec(p + 6))
2354 /* All central directory entries must be on this volume. */
2355 if (archive_le16dec(p + 10) != archive_le16dec(p + 8))
2357 /* Central directory can't extend beyond start of EOCD record. */
2358 if (archive_le32dec(p + 16) + archive_le32dec(p + 12)
2362 /* Save the central directory location for later use. */
2363 zip->central_directory_offset = archive_le32dec(p + 16);
2365 /* This is just a tiny bit higher than the maximum
2366 returned by the streaming Zip bidder. This ensures
2367 that the more accurate seeking Zip parser wins
2368 whenever seek is available. */
2373 * Examine Zip64 EOCD locator: If it's valid, store the information
2377 read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p)
2379 int64_t eocd64_offset;
2380 int64_t eocd64_size;
2382 /* Sanity-check the locator record. */
2384 /* Central dir must be on first volume. */
2385 if (archive_le32dec(p + 4) != 0)
2387 /* Must be only a single volume. */
2388 if (archive_le32dec(p + 16) != 1)
2391 /* Find the Zip64 EOCD record. */
2392 eocd64_offset = archive_le64dec(p + 8);
2393 if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0)
2395 if ((p = __archive_read_ahead(a, 56, NULL)) == NULL)
2397 /* Make sure we can read all of it. */
2398 eocd64_size = archive_le64dec(p + 4) + 12;
2399 if (eocd64_size < 56 || eocd64_size > 16384)
2401 if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL)
2404 /* Sanity-check the EOCD64 */
2405 if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */
2407 if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */
2409 /* CD can't be split. */
2410 if (archive_le64dec(p + 24) != archive_le64dec(p + 32))
2413 /* Save the central directory offset for later use. */
2414 zip->central_directory_offset = archive_le64dec(p + 48);
2418 archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid)
2420 struct zip *zip = (struct zip *)a->format->data;
2421 int64_t file_size, current_offset;
2425 /* If someone has already bid more than 32, then avoid
2426 trashing the look-ahead buffers with a seek. */
2430 file_size = __archive_read_seek(a, 0, SEEK_END);
2434 /* Search last 16k of file for end-of-central-directory
2435 * record (which starts with PK\005\006) */
2436 tail = (int)zipmin(1024 * 16, file_size);
2437 current_offset = __archive_read_seek(a, -tail, SEEK_END);
2438 if (current_offset < 0)
2440 if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL)
2442 /* Boyer-Moore search backwards from the end, since we want
2443 * to match the last EOCD in the file (there can be more than
2444 * one if there is an uncompressed Zip archive as a member
2445 * within this Zip archive). */
2446 for (i = tail - 22; i > 0;) {
2449 if (memcmp(p + i, "PK\005\006", 4) == 0) {
2450 int ret = read_eocd(zip, p + i,
2451 current_offset + i);
2453 /* Zip64 EOCD locator precedes
2454 * regular EOCD if present. */
2456 && memcmp(p + i - 20, "PK\006\007", 4) == 0) {
2457 read_zip64_eocd(a, zip, p + i - 20);
2464 case 'K': i -= 1; break;
2465 case 005: i -= 2; break;
2466 case 006: i -= 3; break;
2467 default: i -= 4; break;
2473 /* The red-black trees are only used in seeking mode to manage
2474 * the in-memory copy of the central directory. */
2477 cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2)
2479 const struct zip_entry *e1 = (const struct zip_entry *)n1;
2480 const struct zip_entry *e2 = (const struct zip_entry *)n2;
2482 if (e1->local_header_offset > e2->local_header_offset)
2484 if (e1->local_header_offset < e2->local_header_offset)
2490 cmp_key(const struct archive_rb_node *n, const void *key)
2492 /* This function won't be called */
2493 (void)n; /* UNUSED */
2494 (void)key; /* UNUSED */
2498 static const struct archive_rb_tree_ops rb_ops = {
2503 rsrc_cmp_node(const struct archive_rb_node *n1,
2504 const struct archive_rb_node *n2)
2506 const struct zip_entry *e1 = (const struct zip_entry *)n1;
2507 const struct zip_entry *e2 = (const struct zip_entry *)n2;
2509 return (strcmp(e2->rsrcname.s, e1->rsrcname.s));
2513 rsrc_cmp_key(const struct archive_rb_node *n, const void *key)
2515 const struct zip_entry *e = (const struct zip_entry *)n;
2516 return (strcmp((const char *)key, e->rsrcname.s));
2519 static const struct archive_rb_tree_ops rb_rsrc_ops = {
2520 &rsrc_cmp_node, &rsrc_cmp_key
2524 rsrc_basename(const char *name, size_t name_length)
2530 s = memchr(s, '/', name_length - (s - name));
2539 expose_parent_dirs(struct zip *zip, const char *name, size_t name_length)
2541 struct archive_string str;
2542 struct zip_entry *dir;
2545 archive_string_init(&str);
2546 archive_strncpy(&str, name, name_length);
2548 s = strrchr(str.s, '/');
2552 /* Transfer the parent directory from zip->tree_rsrc RB
2553 * tree to zip->tree RB tree to expose. */
2554 dir = (struct zip_entry *)
2555 __archive_rb_tree_find_node(&zip->tree_rsrc, str.s);
2558 __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node);
2559 archive_string_free(&dir->rsrcname);
2560 __archive_rb_tree_insert_node(&zip->tree, &dir->node);
2562 archive_string_free(&str);
2566 slurp_central_directory(struct archive_read *a, struct zip *zip)
2571 ssize_t bytes_avail;
2575 * Find the start of the central directory. The end-of-CD
2576 * record has our starting point, but there are lots of
2577 * Zip archives which have had other data prepended to the
2578 * file, which makes the recorded offsets all too small.
2579 * So we search forward from the specified offset until we
2580 * find the real start of the central directory. Then we
2581 * know the correction we need to apply to account for leading
2584 if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0)
2585 return ARCHIVE_FATAL;
2589 if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL)
2590 return ARCHIVE_FATAL;
2591 for (found = 0, i = 0; !found && i < bytes_avail - 4;) {
2593 case 'P': i += 3; break;
2594 case 'K': i += 2; break;
2595 case 001: i += 1; break;
2597 if (memcmp(p + i, "PK\001\002", 4) == 0) {
2603 case 005: i += 1; break;
2605 if (memcmp(p + i, "PK\005\006", 4) == 0) {
2608 } else if (memcmp(p + i, "PK\006\006", 4) == 0) {
2614 default: i += 4; break;
2617 __archive_read_consume(a, i);
2619 correction = archive_filter_bytes(&a->archive, 0)
2620 - zip->central_directory_offset;
2622 __archive_rb_tree_init(&zip->tree, &rb_ops);
2623 __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops);
2625 zip->central_directory_entries_total = 0;
2627 struct zip_entry *zip_entry;
2628 size_t filename_length, extra_length, comment_length;
2629 uint32_t external_attributes;
2630 const char *name, *r;
2632 if ((p = __archive_read_ahead(a, 4, NULL)) == NULL)
2633 return ARCHIVE_FATAL;
2634 if (memcmp(p, "PK\006\006", 4) == 0
2635 || memcmp(p, "PK\005\006", 4) == 0) {
2637 } else if (memcmp(p, "PK\001\002", 4) != 0) {
2638 archive_set_error(&a->archive,
2639 -1, "Invalid central directory signature");
2640 return ARCHIVE_FATAL;
2642 if ((p = __archive_read_ahead(a, 46, NULL)) == NULL)
2643 return ARCHIVE_FATAL;
2645 zip_entry = calloc(1, sizeof(struct zip_entry));
2646 zip_entry->next = zip->zip_entries;
2647 zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY;
2648 zip->zip_entries = zip_entry;
2649 zip->central_directory_entries_total++;
2651 /* version = p[4]; */
2652 zip_entry->system = p[5];
2653 /* version_required = archive_le16dec(p + 6); */
2654 zip_entry->zip_flags = archive_le16dec(p + 8);
2655 if (zip_entry->zip_flags
2656 & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){
2657 zip->has_encrypted_entries = 1;
2659 zip_entry->compression = (char)archive_le16dec(p + 10);
2660 zip_entry->mtime = zip_time(p + 12);
2661 zip_entry->crc32 = archive_le32dec(p + 16);
2662 if (zip_entry->zip_flags & ZIP_LENGTH_AT_END)
2663 zip_entry->decdat = p[13];
2665 zip_entry->decdat = p[19];
2666 zip_entry->compressed_size = archive_le32dec(p + 20);
2667 zip_entry->uncompressed_size = archive_le32dec(p + 24);
2668 filename_length = archive_le16dec(p + 28);
2669 extra_length = archive_le16dec(p + 30);
2670 comment_length = archive_le16dec(p + 32);
2671 /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */
2672 /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */
2673 external_attributes = archive_le32dec(p + 38);
2674 zip_entry->local_header_offset =
2675 archive_le32dec(p + 42) + correction;
2677 /* If we can't guess the mode, leave it zero here;
2678 when we read the local file header we might get
2679 more information. */
2680 if (zip_entry->system == 3) {
2681 zip_entry->mode = external_attributes >> 16;
2682 } else if (zip_entry->system == 0) {
2683 // Interpret MSDOS directory bit
2684 if (0x10 == (external_attributes & 0x10)) {
2685 zip_entry->mode = AE_IFDIR | 0775;
2687 zip_entry->mode = AE_IFREG | 0664;
2689 if (0x01 == (external_attributes & 0x01)) {
2690 // Read-only bit; strip write permissions
2691 zip_entry->mode &= 0555;
2694 zip_entry->mode = 0;
2697 /* We're done with the regular data; get the filename and
2699 __archive_read_consume(a, 46);
2700 p = __archive_read_ahead(a, filename_length + extra_length,
2703 archive_set_error(&a->archive,
2704 ARCHIVE_ERRNO_FILE_FORMAT,
2705 "Truncated ZIP file header");
2706 return ARCHIVE_FATAL;
2708 if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) {
2709 return ARCHIVE_FATAL;
2713 * Mac resource fork files are stored under the
2714 * "__MACOSX/" directory, so we should check if
2717 if (!zip->process_mac_extensions) {
2718 /* Treat every entry as a regular entry. */
2719 __archive_rb_tree_insert_node(&zip->tree,
2723 r = rsrc_basename(name, filename_length);
2724 if (filename_length >= 9 &&
2725 strncmp("__MACOSX/", name, 9) == 0) {
2726 /* If this file is not a resource fork nor
2727 * a directory. We should treat it as a non
2728 * resource fork file to expose it. */
2729 if (name[filename_length-1] != '/' &&
2730 (r - name < 3 || r[0] != '.' || r[1] != '_')) {
2731 __archive_rb_tree_insert_node(
2732 &zip->tree, &zip_entry->node);
2733 /* Expose its parent directories. */
2734 expose_parent_dirs(zip, name,
2737 /* This file is a resource fork file or
2739 archive_strncpy(&(zip_entry->rsrcname),
2740 name, filename_length);
2741 __archive_rb_tree_insert_node(
2742 &zip->tree_rsrc, &zip_entry->node);
2745 /* Generate resource fork name to find its
2746 * resource file at zip->tree_rsrc. */
2747 archive_strcpy(&(zip_entry->rsrcname),
2749 archive_strncat(&(zip_entry->rsrcname),
2751 archive_strcat(&(zip_entry->rsrcname), "._");
2752 archive_strncat(&(zip_entry->rsrcname),
2754 filename_length - (r - name));
2755 /* Register an entry to RB tree to sort it by
2757 __archive_rb_tree_insert_node(&zip->tree,
2762 /* Skip the comment too ... */
2763 __archive_read_consume(a,
2764 filename_length + extra_length + comment_length);
2771 zip_get_local_file_header_size(struct archive_read *a, size_t extra)
2774 ssize_t filename_length, extra_length;
2776 if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) {
2777 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2778 "Truncated ZIP file header");
2779 return (ARCHIVE_WARN);
2783 if (memcmp(p, "PK\003\004", 4) != 0) {
2784 archive_set_error(&a->archive, -1, "Damaged Zip archive");
2785 return ARCHIVE_WARN;
2787 filename_length = archive_le16dec(p + 26);
2788 extra_length = archive_le16dec(p + 28);
2790 return (30 + filename_length + extra_length);
2794 zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry,
2795 struct zip_entry *rsrc)
2797 struct zip *zip = (struct zip *)a->format->data;
2798 unsigned char *metadata, *mp;
2799 int64_t offset = archive_filter_bytes(&a->archive, 0);
2800 size_t remaining_bytes, metadata_bytes;
2802 int ret = ARCHIVE_OK, eof;
2804 switch(rsrc->compression) {
2805 case 0: /* No compression. */
2806 if (rsrc->uncompressed_size != rsrc->compressed_size) {
2807 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2808 "Malformed OS X metadata entry: inconsistent size");
2809 return (ARCHIVE_FATAL);
2812 case 8: /* Deflate compression. */
2815 default: /* Unsupported compression. */
2816 /* Return a warning. */
2817 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2818 "Unsupported ZIP compression method (%s)",
2819 compression_name(rsrc->compression));
2820 /* We can't decompress this entry, but we will
2821 * be able to skip() it and try the next entry. */
2822 return (ARCHIVE_WARN);
2825 if (rsrc->uncompressed_size > (4 * 1024 * 1024)) {
2826 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2827 "Mac metadata is too large: %jd > 4M bytes",
2828 (intmax_t)rsrc->uncompressed_size);
2829 return (ARCHIVE_WARN);
2831 if (rsrc->compressed_size > (4 * 1024 * 1024)) {
2832 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2833 "Mac metadata is too large: %jd > 4M bytes",
2834 (intmax_t)rsrc->compressed_size);
2835 return (ARCHIVE_WARN);
2838 metadata = malloc((size_t)rsrc->uncompressed_size);
2839 if (metadata == NULL) {
2840 archive_set_error(&a->archive, ENOMEM,
2841 "Can't allocate memory for Mac metadata");
2842 return (ARCHIVE_FATAL);
2845 if (offset < rsrc->local_header_offset)
2846 __archive_read_consume(a, rsrc->local_header_offset - offset);
2847 else if (offset != rsrc->local_header_offset) {
2848 __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET);
2851 hsize = zip_get_local_file_header_size(a, 0);
2852 __archive_read_consume(a, hsize);
2854 remaining_bytes = (size_t)rsrc->compressed_size;
2855 metadata_bytes = (size_t)rsrc->uncompressed_size;
2858 while (!eof && remaining_bytes) {
2859 const unsigned char *p;
2860 ssize_t bytes_avail;
2863 p = __archive_read_ahead(a, 1, &bytes_avail);
2865 archive_set_error(&a->archive,
2866 ARCHIVE_ERRNO_FILE_FORMAT,
2867 "Truncated ZIP file header");
2869 goto exit_mac_metadata;
2871 if ((size_t)bytes_avail > remaining_bytes)
2872 bytes_avail = remaining_bytes;
2873 switch(rsrc->compression) {
2874 case 0: /* No compression. */
2875 if ((size_t)bytes_avail > metadata_bytes)
2876 bytes_avail = metadata_bytes;
2877 memcpy(mp, p, bytes_avail);
2878 bytes_used = (size_t)bytes_avail;
2879 metadata_bytes -= bytes_used;
2881 if (metadata_bytes == 0)
2885 case 8: /* Deflate compression. */
2889 ret = zip_deflate_init(a, zip);
2890 if (ret != ARCHIVE_OK)
2891 goto exit_mac_metadata;
2892 zip->stream.next_in =
2893 (Bytef *)(uintptr_t)(const void *)p;
2894 zip->stream.avail_in = (uInt)bytes_avail;
2895 zip->stream.total_in = 0;
2896 zip->stream.next_out = mp;
2897 zip->stream.avail_out = (uInt)metadata_bytes;
2898 zip->stream.total_out = 0;
2900 r = inflate(&zip->stream, 0);
2908 archive_set_error(&a->archive, ENOMEM,
2909 "Out of memory for ZIP decompression");
2910 ret = ARCHIVE_FATAL;
2911 goto exit_mac_metadata;
2913 archive_set_error(&a->archive,
2915 "ZIP decompression failed (%d)", r);
2916 ret = ARCHIVE_FATAL;
2917 goto exit_mac_metadata;
2919 bytes_used = zip->stream.total_in;
2920 metadata_bytes -= zip->stream.total_out;
2921 mp += zip->stream.total_out;
2929 __archive_read_consume(a, bytes_used);
2930 remaining_bytes -= bytes_used;
2932 archive_entry_copy_mac_metadata(entry, metadata,
2933 (size_t)rsrc->uncompressed_size - metadata_bytes);
2936 __archive_read_seek(a, offset, SEEK_SET);
2937 zip->decompress_init = 0;
2943 archive_read_format_zip_seekable_read_header(struct archive_read *a,
2944 struct archive_entry *entry)
2946 struct zip *zip = (struct zip *)a->format->data;
2947 struct zip_entry *rsrc;
2949 int r, ret = ARCHIVE_OK;
2952 * It should be sufficient to call archive_read_next_header() for
2953 * a reader to determine if an entry is encrypted or not. If the
2954 * encryption of an entry is only detectable when calling
2955 * archive_read_data(), so be it. We'll do the same check there
2958 if (zip->has_encrypted_entries ==
2959 ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW)
2960 zip->has_encrypted_entries = 0;
2962 a->archive.archive_format = ARCHIVE_FORMAT_ZIP;
2963 if (a->archive.archive_format_name == NULL)
2964 a->archive.archive_format_name = "ZIP";
2966 if (zip->zip_entries == NULL) {
2967 r = slurp_central_directory(a, zip);
2968 if (r != ARCHIVE_OK)
2970 /* Get first entry whose local header offset is lower than
2971 * other entries in the archive file. */
2973 (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree);
2974 } else if (zip->entry != NULL) {
2975 /* Get next entry in local header offset order. */
2976 zip->entry = (struct zip_entry *)__archive_rb_tree_iterate(
2977 &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT);
2980 if (zip->entry == NULL)
2983 if (zip->entry->rsrcname.s)
2984 rsrc = (struct zip_entry *)__archive_rb_tree_find_node(
2985 &zip->tree_rsrc, zip->entry->rsrcname.s);
2989 if (zip->cctx_valid)
2990 archive_decrypto_aes_ctr_release(&zip->cctx);
2991 if (zip->hctx_valid)
2992 archive_hmac_sha1_cleanup(&zip->hctx);
2993 zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0;
2994 __archive_read_reset_passphrase(a);
2996 /* File entries are sorted by the header offset, we should mostly
2997 * use __archive_read_consume to advance a read point to avoid redundant
2999 offset = archive_filter_bytes(&a->archive, 0);
3000 if (offset < zip->entry->local_header_offset)
3001 __archive_read_consume(a,
3002 zip->entry->local_header_offset - offset);
3003 else if (offset != zip->entry->local_header_offset) {
3004 __archive_read_seek(a, zip->entry->local_header_offset,
3007 zip->unconsumed = 0;
3008 r = zip_read_local_file_header(a, entry, zip);
3009 if (r != ARCHIVE_OK)
3012 int ret2 = zip_read_mac_metadata(a, entry, rsrc);
3020 * We're going to seek for the next header anyway, so we don't
3021 * need to bother doing anything here.
3024 archive_read_format_zip_read_data_skip_seekable(struct archive_read *a)
3027 zip = (struct zip *)(a->format->data);
3029 zip->unconsumed = 0;
3030 return (ARCHIVE_OK);
3034 archive_read_support_format_zip_seekable(struct archive *_a)
3036 struct archive_read *a = (struct archive_read *)_a;
3040 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
3041 ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable");
3043 zip = (struct zip *)calloc(1, sizeof(*zip));
3045 archive_set_error(&a->archive, ENOMEM,
3046 "Can't allocate zip data");
3047 return (ARCHIVE_FATAL);
3050 #ifdef HAVE_COPYFILE_H
3051 /* Set this by default on Mac OS. */
3052 zip->process_mac_extensions = 1;
3056 * Until enough data has been read, we cannot tell about
3057 * any encrypted entries yet.
3059 zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW;
3060 zip->crc32func = real_crc32;
3062 r = __archive_read_register_format(a,
3065 archive_read_format_zip_seekable_bid,
3066 archive_read_format_zip_options,
3067 archive_read_format_zip_seekable_read_header,
3068 archive_read_format_zip_read_data,
3069 archive_read_format_zip_read_data_skip_seekable,
3071 archive_read_format_zip_cleanup,
3072 archive_read_support_format_zip_capabilities_seekable,
3073 archive_read_format_zip_has_encrypted_entries);
3075 if (r != ARCHIVE_OK)
3077 return (ARCHIVE_OK);