blob: 0d00d8dc921ff129051b688bea02b1094a0d39f6 [file] [log] [blame]
Glenn L McGrath87ac7022002-01-02 13:52:26 +00001/* vi: set sw=4 ts=4: */
2/*
3 * Mini unzip implementation for busybox
4 *
Paul Fox0840b762005-07-20 20:26:49 +00005 * Copyright (C) 2004 by Ed Clark
6 *
7 * Loosely based on original busybox unzip applet by Laurence Anderson.
8 * All options and features should work in this version.
Glenn L McGrath87ac7022002-01-02 13:52:26 +00009 *
Denys Vlasenko0ef64bd2010-08-16 20:14:46 +020010 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
Glenn L McGrath87ac7022002-01-02 13:52:26 +000011 */
Glenn L McGrathf34b0e92004-06-06 10:22:43 +000012/* For reference see
Paul Fox0840b762005-07-20 20:26:49 +000013 * http://www.pkware.com/company/standards/appnote/
Glenn L McGrathf34b0e92004-06-06 10:22:43 +000014 * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
Denys Vlasenko66620fa2013-11-14 09:53:52 +010015 *
16 * TODO
Paul Fox0840b762005-07-20 20:26:49 +000017 * Zip64 + other methods
Paul Fox0840b762005-07-20 20:26:49 +000018 */
Denys Vlasenkof6beef62013-11-14 11:39:00 +010019//config:config UNZIP
Denys Vlasenko4eed2c62017-07-18 22:01:24 +020020//config: bool "unzip (24 kb)"
Denys Vlasenkof6beef62013-11-14 11:39:00 +010021//config: default y
22//config: help
Denys Vlasenko72089cf2017-07-21 09:50:55 +020023//config: unzip will list or extract files from a ZIP archive,
24//config: commonly found on DOS/WIN systems. The default behavior
25//config: (with no options) is to extract the archive into the
26//config: current directory.
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +010027//config:
28//config:config FEATURE_UNZIP_CDF
29//config: bool "Read and use Central Directory data"
30//config: default y
31//config: depends on UNZIP
32//config: help
Denys Vlasenko72089cf2017-07-21 09:50:55 +020033//config: If you know that you only need to deal with simple
34//config: ZIP files without deleted/updated files, SFX archives etc,
35//config: you can reduce code size by unselecting this option.
36//config: To support less trivial ZIPs, say Y.
Denys Vlasenko2a0867a2017-01-09 10:58:37 +010037//config:
38//config:config FEATURE_UNZIP_BZIP2
39//config: bool "Support compression method 12 (bzip2)"
40//config: default y
41//config: depends on FEATURE_UNZIP_CDF && DESKTOP
42// FEATURE_UNZIP_CDF is needed, otherwise we can't find start of next file
43// DESKTOP is needed to get back uncompressed length
44//config:
45//config:config FEATURE_UNZIP_LZMA
46//config: bool "Support compression method 14 (lzma)"
47//config: default y
48//config: depends on FEATURE_UNZIP_CDF && DESKTOP
Denys Vlasenko6b4f4b52017-01-09 11:12:01 +010049//config:
50//config:config FEATURE_UNZIP_XZ
51//config: bool "Support compression method 95 (xz)"
52//config: default y
53//config: depends on FEATURE_UNZIP_CDF && DESKTOP
Denys Vlasenkof6beef62013-11-14 11:39:00 +010054
Denys Vlasenkoac216872013-11-14 11:38:18 +010055//applet:IF_UNZIP(APPLET(unzip, BB_DIR_USR_BIN, BB_SUID_DROP))
Denys Vlasenko66620fa2013-11-14 09:53:52 +010056//kbuild:lib-$(CONFIG_UNZIP) += unzip.o
57
Pere Orga1f4447b2011-03-27 22:40:30 +020058//usage:#define unzip_trivial_usage
Eugene Rudoy978810d2017-11-07 08:03:37 +010059//usage: "[-lnojpq] FILE[.zip] [FILE]... [-x FILE...] [-d DIR]"
Pere Orga1f4447b2011-03-27 22:40:30 +020060//usage:#define unzip_full_usage "\n\n"
Denys Vlasenkoe3e0d2b2012-06-19 12:46:59 +020061//usage: "Extract FILEs from ZIP archive\n"
62//usage: "\n -l List contents (with -q for short form)"
Denys Vlasenkoc5b01012012-06-15 16:43:26 +020063//usage: "\n -n Never overwrite files (default: ask)"
Pere Orga1f4447b2011-03-27 22:40:30 +020064//usage: "\n -o Overwrite"
Denys Vlasenko997ad2c2017-07-20 20:04:49 +020065//usage: "\n -j Do not restore paths"
Denys Vlasenkoe3e0d2b2012-06-19 12:46:59 +020066//usage: "\n -p Print to stdout"
Pere Orga1f4447b2011-03-27 22:40:30 +020067//usage: "\n -q Quiet"
Denys Vlasenkoe3e0d2b2012-06-19 12:46:59 +020068//usage: "\n -x FILE Exclude FILEs"
69//usage: "\n -d DIR Extract into DIR"
Pere Orga1f4447b2011-03-27 22:40:30 +020070
Denis Vlasenkob6adbf12007-05-26 19:00:18 +000071#include "libbb.h"
Denys Vlasenkod184a722011-09-22 12:45:14 +020072#include "bb_archive.h"
Glenn L McGrath87ac7022002-01-02 13:52:26 +000073
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +020074#if 0
75# define dbg(...) bb_error_msg(__VA_ARGS__)
76#else
77# define dbg(...) ((void)0)
78#endif
79
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +000080enum {
81#if BB_BIG_ENDIAN
82 ZIP_FILEHEADER_MAGIC = 0x504b0304,
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +010083 ZIP_CDF_MAGIC = 0x504b0102, /* CDF item */
84 ZIP_CDE_MAGIC = 0x504b0506, /* End of CDF */
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +000085 ZIP_DD_MAGIC = 0x504b0708,
86#else
87 ZIP_FILEHEADER_MAGIC = 0x04034b50,
Denys Vlasenko4e8ff732010-05-24 04:33:02 +020088 ZIP_CDF_MAGIC = 0x02014b50,
Denis Vlasenko006e8622008-09-21 01:01:46 +000089 ZIP_CDE_MAGIC = 0x06054b50,
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +000090 ZIP_DD_MAGIC = 0x08074b50,
91#endif
92};
Glenn L McGrath7ca04f32002-09-25 02:47:48 +000093
Paul Foxcb981632007-11-05 23:09:03 +000094#define ZIP_HEADER_LEN 26
95
Paul Fox0840b762005-07-20 20:26:49 +000096typedef union {
Paul Foxcb981632007-11-05 23:09:03 +000097 uint8_t raw[ZIP_HEADER_LEN];
Paul Fox0840b762005-07-20 20:26:49 +000098 struct {
Denis Vlasenko006e8622008-09-21 01:01:46 +000099 uint16_t version; /* 0-1 */
Denys Vlasenkoe98884b2010-05-24 04:46:18 +0200100 uint16_t zip_flags; /* 2-3 */
Denis Vlasenko006e8622008-09-21 01:01:46 +0000101 uint16_t method; /* 4-5 */
102 uint16_t modtime; /* 6-7 */
103 uint16_t moddate; /* 8-9 */
104 uint32_t crc32 PACKED; /* 10-13 */
105 uint32_t cmpsize PACKED; /* 14-17 */
106 uint32_t ucmpsize PACKED; /* 18-21 */
107 uint16_t filename_len; /* 22-23 */
108 uint16_t extra_len; /* 24-25 */
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100109 /* filename follows (not NUL terminated) */
110 /* extra field follows */
111 /* data follows */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100112 } fmt PACKED;
Denis Vlasenkoa60f84e2008-07-05 09:18:54 +0000113} zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
Paul Fox0840b762005-07-20 20:26:49 +0000114
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100115#define FIX_ENDIANNESS_ZIP(zip) \
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100116do { if (BB_BIG_ENDIAN) { \
Natanael Copa46f3f162017-07-25 20:44:50 +0200117 (zip).fmt.method = SWAP_LE16((zip).fmt.method ); \
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100118 (zip).fmt.crc32 = SWAP_LE32((zip).fmt.crc32 ); \
119 (zip).fmt.cmpsize = SWAP_LE32((zip).fmt.cmpsize ); \
120 (zip).fmt.ucmpsize = SWAP_LE32((zip).fmt.ucmpsize ); \
121 (zip).fmt.filename_len = SWAP_LE16((zip).fmt.filename_len); \
122 (zip).fmt.extra_len = SWAP_LE16((zip).fmt.extra_len ); \
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100123}} while (0)
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000124
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200125#define CDF_HEADER_LEN 42
Denis Vlasenko006e8622008-09-21 01:01:46 +0000126
127typedef union {
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200128 uint8_t raw[CDF_HEADER_LEN];
Denis Vlasenko006e8622008-09-21 01:01:46 +0000129 struct {
130 /* uint32_t signature; 50 4b 01 02 */
131 uint16_t version_made_by; /* 0-1 */
132 uint16_t version_needed; /* 2-3 */
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200133 uint16_t cdf_flags; /* 4-5 */
Denis Vlasenko006e8622008-09-21 01:01:46 +0000134 uint16_t method; /* 6-7 */
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100135 uint16_t modtime; /* 8-9 */
136 uint16_t moddate; /* 10-11 */
Denis Vlasenko006e8622008-09-21 01:01:46 +0000137 uint32_t crc32; /* 12-15 */
138 uint32_t cmpsize; /* 16-19 */
139 uint32_t ucmpsize; /* 20-23 */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100140 uint16_t filename_len; /* 24-25 */
141 uint16_t extra_len; /* 26-27 */
Denis Vlasenko006e8622008-09-21 01:01:46 +0000142 uint16_t file_comment_length; /* 28-29 */
143 uint16_t disk_number_start; /* 30-31 */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100144 uint16_t internal_attributes; /* 32-33 */
145 uint32_t external_attributes PACKED; /* 34-37 */
Denis Vlasenko006e8622008-09-21 01:01:46 +0000146 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100147 /* filename follows (not NUL terminated) */
148 /* extra field follows */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100149 /* file comment follows */
150 } fmt PACKED;
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200151} cdf_header_t;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000152
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100153#define FIX_ENDIANNESS_CDF(cdf) \
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100154do { if (BB_BIG_ENDIAN) { \
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100155 (cdf).fmt.version_made_by = SWAP_LE16((cdf).fmt.version_made_by); \
156 (cdf).fmt.version_needed = SWAP_LE16((cdf).fmt.version_needed); \
157 (cdf).fmt.method = SWAP_LE16((cdf).fmt.method ); \
158 (cdf).fmt.modtime = SWAP_LE16((cdf).fmt.modtime ); \
159 (cdf).fmt.moddate = SWAP_LE16((cdf).fmt.moddate ); \
160 (cdf).fmt.crc32 = SWAP_LE32((cdf).fmt.crc32 ); \
161 (cdf).fmt.cmpsize = SWAP_LE32((cdf).fmt.cmpsize ); \
162 (cdf).fmt.ucmpsize = SWAP_LE32((cdf).fmt.ucmpsize ); \
163 (cdf).fmt.filename_len = SWAP_LE16((cdf).fmt.filename_len); \
164 (cdf).fmt.extra_len = SWAP_LE16((cdf).fmt.extra_len ); \
165 (cdf).fmt.file_comment_length = SWAP_LE16((cdf).fmt.file_comment_length); \
166 (cdf).fmt.external_attributes = SWAP_LE32((cdf).fmt.external_attributes); \
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100167}} while (0)
Denis Vlasenko006e8622008-09-21 01:01:46 +0000168
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100169#define CDE_LEN 16
Denis Vlasenko006e8622008-09-21 01:01:46 +0000170
171typedef union {
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100172 uint8_t raw[CDE_LEN];
Denis Vlasenko006e8622008-09-21 01:01:46 +0000173 struct {
174 /* uint32_t signature; 50 4b 05 06 */
175 uint16_t this_disk_no;
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200176 uint16_t disk_with_cdf_no;
177 uint16_t cdf_entries_on_this_disk;
178 uint16_t cdf_entries_total;
179 uint32_t cdf_size;
180 uint32_t cdf_offset;
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100181 /* uint16_t archive_comment_length; */
182 /* archive comment follows */
183 } fmt PACKED;
184} cde_t;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000185
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100186#define FIX_ENDIANNESS_CDE(cde) \
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100187do { if (BB_BIG_ENDIAN) { \
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100188 (cde).fmt.cdf_offset = SWAP_LE32((cde).fmt.cdf_offset); \
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100189}} while (0)
190
191struct BUG {
192 /* Check the offset of the last element, not the length. This leniency
193 * allows for poor packing, whereby the overall struct may be too long,
194 * even though the elements are all in the right place.
195 */
196 char BUG_zip_header_must_be_26_bytes[
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100197 offsetof(zip_header_t, fmt.extra_len) + 2
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100198 == ZIP_HEADER_LEN ? 1 : -1];
199 char BUG_cdf_header_must_be_42_bytes[
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100200 offsetof(cdf_header_t, fmt.relative_offset_of_local_header) + 4
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100201 == CDF_HEADER_LEN ? 1 : -1];
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100202 char BUG_cde_must_be_16_bytes[
203 sizeof(cde_t) == CDE_LEN ? 1 : -1];
Denis Vlasenko006e8622008-09-21 01:01:46 +0000204};
205
Denis Vlasenko006e8622008-09-21 01:01:46 +0000206
207enum { zip_fd = 3 };
208
209
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100210/* This value means that we failed to find CDF */
211#define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
Dan Fandrichb76f18d2010-06-17 21:39:44 -0700212
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100213#if !ENABLE_FEATURE_UNZIP_CDF
214
215# define find_cdf_offset() BAD_CDF_OFFSET
216
217#else
Denys Vlasenko5e87e8a2013-07-20 15:20:46 +0200218/* Seen in the wild:
219 * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
220 * where CDE was nearly 48 kbytes before EOF.
221 * (Surprisingly, it also apparently has *another* CDE structure
222 * closer to the end, with bogus cdf_offset).
223 * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
224 */
225#define PEEK_FROM_END (64*1024)
Denis Vlasenko006e8622008-09-21 01:01:46 +0000226/* NB: does not preserve file position! */
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200227static uint32_t find_cdf_offset(void)
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000228{
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100229 cde_t cde;
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100230 unsigned char *buf;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000231 unsigned char *p;
232 off_t end;
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200233 uint32_t found;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000234
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100235 end = lseek(zip_fd, 0, SEEK_END);
236 if (end == (off_t) -1)
237 return BAD_CDF_OFFSET;
238
Dan Fandrichb76f18d2010-06-17 21:39:44 -0700239 end -= PEEK_FROM_END;
Denys Vlasenkofc2bb8f2010-05-24 13:07:55 +0200240 if (end < 0)
241 end = 0;
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100242
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200243 dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
Denys Vlasenko876c1212017-03-24 15:00:12 +0100244 xlseek(zip_fd, end, SEEK_SET);
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100245 buf = xzalloc(PEEK_FROM_END);
Dan Fandrichb76f18d2010-06-17 21:39:44 -0700246 full_read(zip_fd, buf, PEEK_FROM_END);
Denis Vlasenko006e8622008-09-21 01:01:46 +0000247
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200248 found = BAD_CDF_OFFSET;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000249 p = buf;
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100250 while (p <= buf + PEEK_FROM_END - CDE_LEN - 4) {
Denis Vlasenko006e8622008-09-21 01:01:46 +0000251 if (*p != 'P') {
252 p++;
253 continue;
254 }
255 if (*++p != 'K')
256 continue;
257 if (*++p != 5)
258 continue;
259 if (*++p != 6)
260 continue;
261 /* we found CDE! */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100262 memcpy(cde.raw, p + 1, CDE_LEN);
263 FIX_ENDIANNESS_CDE(cde);
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200264 /*
265 * I've seen .ZIP files with seemingly valid CDEs
266 * where cdf_offset points past EOF - ??
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200267 * This check ignores such CDEs:
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200268 */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100269 if (cde.fmt.cdf_offset < end + (p - buf)) {
270 found = cde.fmt.cdf_offset;
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200271 dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
272 (unsigned)found, end + (p-3 - buf));
273 dbg(" cdf_offset+cdf_size:0x%x",
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100274 (unsigned)(found + SWAP_LE32(cde.fmt.cdf_size)));
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200275 /*
276 * We do not "break" here because only the last CDE is valid.
277 * I've seen a .zip archive which contained a .zip file,
278 * uncompressed, and taking the first CDE was using
279 * the CDE inside that file!
280 */
281 }
Denis Vlasenko006e8622008-09-21 01:01:46 +0000282 }
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200283 free(buf);
Denys Vlasenko0ccf52a2016-04-17 21:05:34 +0200284 dbg("Found cdf_offset:0x%x", (unsigned)found);
285 return found;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000286};
287
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100288static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf)
Denis Vlasenko006e8622008-09-21 01:01:46 +0000289{
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100290 uint32_t magic;
291
292 if (cdf_offset == BAD_CDF_OFFSET)
293 return cdf_offset;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000294
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100295 dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
296 xlseek(zip_fd, cdf_offset, SEEK_SET);
297 xread(zip_fd, &magic, 4);
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100298 /* Central Directory End? Assume CDF has ended.
299 * (more correct method is to use cde.cdf_entries_total counter)
300 */
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100301 if (magic == ZIP_CDE_MAGIC) {
302 dbg("got ZIP_CDE_MAGIC");
303 return 0; /* EOF */
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200304 }
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100305 xread(zip_fd, cdf->raw, CDF_HEADER_LEN);
Denis Vlasenko006e8622008-09-21 01:01:46 +0000306
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100307 FIX_ENDIANNESS_CDF(*cdf);
308 dbg(" filename_len:%u extra_len:%u file_comment_length:%u",
309 (unsigned)cdf->fmt.filename_len,
310 (unsigned)cdf->fmt.extra_len,
311 (unsigned)cdf->fmt.file_comment_length
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100312 );
313 cdf_offset += 4 + CDF_HEADER_LEN
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100314 + cdf->fmt.filename_len
315 + cdf->fmt.extra_len
316 + cdf->fmt.file_comment_length;
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100317
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200318 return cdf_offset;
Denis Vlasenko006e8622008-09-21 01:01:46 +0000319};
320#endif
321
Denys Vlasenkoad37abf2017-07-20 20:21:50 +0200322static void die_if_bad_fnamesize(unsigned sz)
323{
324 if (sz > 0xfff) /* more than 4k?! no funny business please */
325 bb_error_msg_and_die("bad archive");
326}
327
Denis Vlasenko006e8622008-09-21 01:01:46 +0000328static void unzip_skip(off_t skip)
329{
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200330 if (skip != 0)
331 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
332 bb_copyfd_exact_size(zip_fd, -1, skip);
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000333}
334
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000335static void unzip_create_leading_dirs(const char *fn)
Paul Fox0840b762005-07-20 20:26:49 +0000336{
337 /* Create all leading directories */
Rob Landleyd921b2e2006-08-03 15:41:12 +0000338 char *name = xstrdup(fn);
Denys Vlasenko5cdd1202018-02-06 17:59:32 +0100339
340 /* mode of -1: set mode according to umask */
341 if (bb_make_directory(dirname(name), -1, FILEUTILS_RECUR)) {
Denys Vlasenko831756b2011-09-09 17:30:55 +0200342 xfunc_die(); /* bb_make_directory is noisy */
Paul Fox0840b762005-07-20 20:26:49 +0000343 }
344 free(name);
345}
346
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200347#if ENABLE_FEATURE_UNZIP_CDF
Denys Vlasenkoa84db182018-02-20 15:57:45 +0100348static void unzip_extract_symlink(llist_t **symlink_placeholders,
349 zip_header_t *zip,
350 const char *dst_fn)
Denys Vlasenko13ae85e2017-07-20 18:34:51 +0200351{
352 char *target;
353
Denys Vlasenkoad37abf2017-07-20 20:21:50 +0200354 die_if_bad_fnamesize(zip->fmt.ucmpsize);
Denys Vlasenko13ae85e2017-07-20 18:34:51 +0200355
356 if (zip->fmt.method == 0) {
357 /* Method 0 - stored (not compressed) */
358 target = xzalloc(zip->fmt.ucmpsize + 1);
359 xread(zip_fd, target, zip->fmt.ucmpsize);
360 } else {
361#if 1
362 bb_error_msg_and_die("compressed symlink is not supported");
363#else
364 transformer_state_t xstate;
365 init_transformer_state(&xstate);
366 xstate.mem_output_size_max = zip->fmt.ucmpsize;
367 /* ...unpack... */
368 if (!xstate.mem_output_buf)
369 WTF();
370 target = xstate.mem_output_buf;
371 target = xrealloc(target, xstate.mem_output_size + 1);
372 target[xstate.mem_output_size] = '\0';
373#endif
374 }
Denys Vlasenkoa84db182018-02-20 15:57:45 +0100375 create_or_remember_symlink(symlink_placeholders,
376 target,
377 dst_fn);
Denys Vlasenko13ae85e2017-07-20 18:34:51 +0200378 free(target);
379}
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200380#endif
Denys Vlasenko13ae85e2017-07-20 18:34:51 +0200381
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100382static void unzip_extract(zip_header_t *zip, int dst_fd)
Paul Fox0840b762005-07-20 20:26:49 +0000383{
Denys Vlasenko2a0867a2017-01-09 10:58:37 +0100384 transformer_state_t xstate;
385
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100386 if (zip->fmt.method == 0) {
Paul Fox0840b762005-07-20 20:26:49 +0000387 /* Method 0 - stored (not compressed) */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100388 off_t size = zip->fmt.ucmpsize;
Denis Vlasenko714701c2006-12-22 00:21:07 +0000389 if (size)
Denis Vlasenko006e8622008-09-21 01:01:46 +0000390 bb_copyfd_exact_size(zip_fd, dst_fd, size);
Denys Vlasenko2a0867a2017-01-09 10:58:37 +0100391 return;
392 }
393
394 init_transformer_state(&xstate);
395 xstate.bytes_in = zip->fmt.cmpsize;
396 xstate.src_fd = zip_fd;
397 xstate.dst_fd = dst_fd;
398 if (zip->fmt.method == 8) {
Paul Fox0840b762005-07-20 20:26:49 +0000399 /* Method 8 - inflate */
Denys Vlasenkob4c11c12014-12-07 00:44:00 +0100400 if (inflate_unzip(&xstate) < 0)
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000401 bb_error_msg_and_die("inflate error");
Paul Fox0840b762005-07-20 20:26:49 +0000402 /* Validate decompression - crc */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100403 if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000404 bb_error_msg_and_die("crc error");
Paul Fox0840b762005-07-20 20:26:49 +0000405 }
Paul Fox0840b762005-07-20 20:26:49 +0000406 }
Denys Vlasenko2a0867a2017-01-09 10:58:37 +0100407#if ENABLE_FEATURE_UNZIP_BZIP2
408 else if (zip->fmt.method == 12) {
409 /* Tested. Unpacker reads too much, but we use CDF
410 * and will seek to the correct beginning of next file.
411 */
412 xstate.bytes_out = unpack_bz2_stream(&xstate);
413 if (xstate.bytes_out < 0)
414 bb_error_msg_and_die("inflate error");
415 }
416#endif
417#if ENABLE_FEATURE_UNZIP_LZMA
418 else if (zip->fmt.method == 14) {
419 /* Not tested yet */
420 xstate.bytes_out = unpack_lzma_stream(&xstate);
421 if (xstate.bytes_out < 0)
422 bb_error_msg_and_die("inflate error");
423 }
424#endif
Denys Vlasenko6b4f4b52017-01-09 11:12:01 +0100425#if ENABLE_FEATURE_UNZIP_XZ
426 else if (zip->fmt.method == 95) {
427 /* Not tested yet */
428 xstate.bytes_out = unpack_xz_stream(&xstate);
429 if (xstate.bytes_out < 0)
430 bb_error_msg_and_die("inflate error");
431 }
432#endif
Denys Vlasenko2a0867a2017-01-09 10:58:37 +0100433 else {
434 bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
435 }
436
437 /* Validate decompression - size */
438 if (zip->fmt.ucmpsize != xstate.bytes_out) {
439 /* Don't die. Who knows, maybe len calculation
440 * was botched somewhere. After all, crc matched! */
441 bb_error_msg("bad length");
442 }
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000443}
444
Denys Vlasenkobf998072013-01-22 11:16:08 +0100445static void my_fgets80(char *buf80)
446{
447 fflush_all();
448 if (!fgets(buf80, 80, stdin)) {
449 bb_perror_msg_and_die("can't read standard input");
450 }
451}
452
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200453static int get_lstat_mode(const char *dst_fn)
454{
455 struct stat stat_buf;
456 if (lstat(dst_fn, &stat_buf) == -1) {
457 if (errno != ENOENT) {
458 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
459 }
460 /* File does not exist */
461 return -1;
462 }
463 return stat_buf.st_mode;
464}
465
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +0000466int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Rob Landleydfba7412006-03-06 20:47:33 +0000467int unzip_main(int argc, char **argv)
Glenn L McGrath87ac7022002-01-02 13:52:26 +0000468{
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200469 enum {
470 OPT_l = (1 << 0),
471 OPT_x = (1 << 1),
472 OPT_j = (1 << 2),
473 };
474 unsigned opts;
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200475 smallint quiet = 0;
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100476 IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200477 enum { O_PROMPT, O_NEVER, O_ALWAYS };
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000478 smallint overwrite = O_PROMPT;
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200479 uint32_t cdf_offset;
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200480 unsigned long total_usize;
481 unsigned long total_size;
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000482 unsigned total_entries;
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000483 int dst_fd = -1;
484 char *src_fn = NULL;
485 char *dst_fn = NULL;
Mike Frysinger69024552005-07-30 07:30:26 +0000486 llist_t *zaccept = NULL;
487 llist_t *zreject = NULL;
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000488 char *base_dir = NULL;
Denys Vlasenkoa84db182018-02-20 15:57:45 +0100489#if ENABLE_FEATURE_UNZIP_CDF
490 llist_t *symlink_placeholders = NULL;
491#endif
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200492 int i;
Denys Vlasenkobf998072013-01-22 11:16:08 +0100493 char key_buf[80]; /* must match size used by my_fgets80 */
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000494
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200495/* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
496 *
497 * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
498 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
499 * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
500 * Length Method Size Ratio Date Time CRC-32 Name
501 * -------- ------ ------- ----- ---- ---- ------ ----
502 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
503 * -------- ------- --- -------
504 * 204372 35278 83% 1 file
505 * # /usr/bin/unzip -v decompress_unlzma.i.zip
506 * Archive: decompress_unlzma.i.zip
507 * Length Method Size Ratio Date Time CRC-32 Name
508 * -------- ------ ------- ----- ---- ---- ------ ----
509 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
510 * -------- ------- --- -------
511 * 204372 35278 83% 1 file
512 * # unzip -v decompress_unlzma.i.zip
513 * Archive: decompress_unlzma.i.zip
514 * Length Date Time Name
515 * -------- ---- ---- ----
516 * 204372 09-06-09 14:23 decompress_unlzma.i
517 * -------- -------
518 * 204372 1 files
519 * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
520 * 204372 09-06-09 14:23 decompress_unlzma.i
521 * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
522 * Length Date Time Name
523 * -------- ---- ---- ----
524 * 204372 09-06-09 14:23 decompress_unlzma.i
525 * -------- -------
526 * 204372 1 file
527 * # /usr/bin/unzip -l decompress_unlzma.i.zip
528 * Archive: decompress_unlzma.i.zip
529 * Length Date Time Name
530 * -------- ---- ---- ----
531 * 204372 09-06-09 14:23 decompress_unlzma.i
532 * -------- -------
533 * 204372 1 file
534 */
535
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200536 opts = 0;
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000537 /* '-' makes getopt return 1 for non-options */
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200538 while ((i = getopt(argc, argv, "-d:lnopqxjv")) != -1) {
539 switch (i) {
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200540 case 'd': /* Extract to base directory */
541 base_dir = optarg;
542 break;
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000543
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200544 case 'l': /* List */
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200545 opts |= OPT_l;
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200546 break;
Paul Fox0840b762005-07-20 20:26:49 +0000547
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200548 case 'n': /* Never overwrite existing files */
549 overwrite = O_NEVER;
550 break;
Paul Fox0840b762005-07-20 20:26:49 +0000551
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200552 case 'o': /* Always overwrite existing files */
553 overwrite = O_ALWAYS;
554 break;
Paul Fox0840b762005-07-20 20:26:49 +0000555
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200556 case 'p': /* Extract files to stdout and fall through to set verbosity */
557 dst_fd = STDOUT_FILENO;
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200558
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200559 case 'q': /* Be quiet */
560 quiet++;
561 break;
Paul Fox0840b762005-07-20 20:26:49 +0000562
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200563 case 'v': /* Verbose list */
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100564 IF_FEATURE_UNZIP_CDF(verbose++;)
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200565 opts |= OPT_l;
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200566 break;
567
568 case 'x':
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200569 opts |= OPT_x;
570 break;
571
572 case 'j':
573 opts |= OPT_j;
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200574 break;
575
576 case 1:
577 if (!src_fn) {
578 /* The zip file */
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000579 /* +5: space for ".zip" and NUL */
580 src_fn = xmalloc(strlen(optarg) + 5);
Denis Vlasenko666c40c2007-03-31 10:17:24 +0000581 strcpy(src_fn, optarg);
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200582 } else if (!(opts & OPT_x)) {
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200583 /* Include files */
584 llist_add_to(&zaccept, optarg);
585 } else {
586 /* Exclude files */
587 llist_add_to(&zreject, optarg);
Paul Fox0840b762005-07-20 20:26:49 +0000588 }
589 break;
590
Paul Fox0840b762005-07-20 20:26:49 +0000591 default:
592 bb_show_usage();
Glenn L McGrath87ac7022002-01-02 13:52:26 +0000593 }
594 }
595
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200596#ifndef __GLIBC__
597 /*
598 * This code is needed for non-GNU getopt
599 * which doesn't understand "-" in option string.
600 * The -x option won't work properly in this case:
601 * "unzip a.zip q -x w e" will be interpreted as
602 * "unzip a.zip q w e -x" = "unzip a.zip q w e"
603 */
604 argv += optind;
605 if (argv[0]) {
606 /* +5: space for ".zip" and NUL */
607 src_fn = xmalloc(strlen(argv[0]) + 5);
608 strcpy(src_fn, argv[0]);
609 while (*++argv)
610 llist_add_to(&zaccept, *argv);
611 }
612#endif
613
614 if (!src_fn) {
Manuel Novoa III cad53642003-03-19 09:13:01 +0000615 bb_show_usage();
Glenn L McGrath87ac7022002-01-02 13:52:26 +0000616 }
617
Paul Fox0840b762005-07-20 20:26:49 +0000618 /* Open input file */
Denis Vlasenko9f739442006-12-16 23:49:13 +0000619 if (LONE_DASH(src_fn)) {
Denis Vlasenko006e8622008-09-21 01:01:46 +0000620 xdup2(STDIN_FILENO, zip_fd);
Paul Fox0840b762005-07-20 20:26:49 +0000621 /* Cannot use prompt mode since zip data is arriving on STDIN */
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000622 if (overwrite == O_PROMPT)
623 overwrite = O_NEVER;
Glenn L McGrath237ae422002-11-03 14:05:15 +0000624 } else {
Denys Vlasenko3e134eb2016-04-22 18:09:21 +0200625 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200626 char *ext = src_fn + strlen(src_fn);
627 int src_fd;
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000628
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200629 i = 0;
630 for (;;) {
Paul Fox0840b762005-07-20 20:26:49 +0000631 src_fd = open(src_fn, O_RDONLY);
Denys Vlasenkoc5b01012012-06-15 16:43:26 +0200632 if (src_fd >= 0)
633 break;
634 if (++i > 2) {
635 *ext = '\0';
636 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
637 }
638 strcpy(ext, extn[i - 1]);
Paul Fox0840b762005-07-20 20:26:49 +0000639 }
Denis Vlasenko006e8622008-09-21 01:01:46 +0000640 xmove_fd(src_fd, zip_fd);
Glenn L McGrath237ae422002-11-03 14:05:15 +0000641 }
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000642
Paul Fox0840b762005-07-20 20:26:49 +0000643 /* Change dir if necessary */
Bernhard Reutner-Fischerd9cf7ac2006-04-12 18:39:58 +0000644 if (base_dir)
Rob Landleyd921b2e2006-08-03 15:41:12 +0000645 xchdir(base_dir);
Glenn L McGrath87ac7022002-01-02 13:52:26 +0000646
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200647 if (quiet <= 1) { /* not -qq */
648 if (quiet == 0)
649 printf("Archive: %s\n", src_fn);
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200650 if (opts & OPT_l) {
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200651 puts(verbose ?
Denys Vlasenko07bd9792016-04-18 01:43:24 +0200652 " Length Method Size Cmpr Date Time CRC-32 Name\n"
653 "-------- ------ ------- ---- ---------- ----- -------- ----"
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200654 :
Denys Vlasenko07bd9792016-04-18 01:43:24 +0200655 " Length Date Time Name\n"
656 "--------- ---------- ----- ----"
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200657 );
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000658 }
659 }
Glenn L McGrath87ac7022002-01-02 13:52:26 +0000660
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200661/* Example of an archive with one 0-byte long file named 'z'
662 * created by Zip 2.31 on Unix:
663 * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
664 * sig........ vneed flags compr mtime mdate crc32>
665 * 0010 00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
666 * >..... csize...... usize...... fnlen exlen fn ex>
667 * 0020 54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
668 * >tra_field......................................
669 * 0030 00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
670 * ........... sig........ vmade vneed flags compr
671 * 0040 42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
672 * mtime mdate crc32...... csize...... usize......
673 * 0050 01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
674 * fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
675 * 0060 00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
676 * >..... fn extra_field...........................
677 * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
678 * 0080 34 00 00 00 00 00 |4.....|
679 */
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200680 total_usize = 0;
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000681 total_size = 0;
682 total_entries = 0;
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100683 cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000684 while (1) {
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100685 zip_header_t zip;
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200686 mode_t dir_mode = 0777;
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100687#if ENABLE_FEATURE_UNZIP_CDF
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200688 mode_t file_mode = 0666;
689#endif
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000690
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100691 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
692 /* Normally happens when input is unseekable.
693 *
694 * Valid ZIP file has Central Directory at the end
695 * with central directory file headers (CDFs).
696 * After it, there is a Central Directory End structure.
697 * CDFs identify what files are in the ZIP and where
698 * they are located. This allows ZIP readers to load
699 * the list of files without reading the entire ZIP archive.
700 * ZIP files may be appended to, only files specified in
701 * the CD are valid. Scanning for local file headers is
702 * not a correct algorithm.
703 *
704 * We try to do the above, and resort to "linear" reading
705 * of ZIP file only if seek failed or CDE wasn't found.
706 */
707 uint32_t magic;
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000708
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100709 /* Check magic number */
710 xread(zip_fd, &magic, 4);
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100711 /* CDF item? Assume there are no more files, exit */
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100712 if (magic == ZIP_CDF_MAGIC) {
713 dbg("got ZIP_CDF_MAGIC");
714 break;
715 }
716 /* Data descriptor? It was a streaming file, go on */
717 if (magic == ZIP_DD_MAGIC) {
718 dbg("got ZIP_DD_MAGIC");
719 /* skip over duplicate crc32, cmpsize and ucmpsize */
720 unzip_skip(3 * 4);
721 continue;
722 }
723 if (magic != ZIP_FILEHEADER_MAGIC)
724 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
725 dbg("got ZIP_FILEHEADER_MAGIC");
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200726
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100727 xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
728 FIX_ENDIANNESS_ZIP(zip);
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100729 if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
730 bb_error_msg_and_die("zip flag %s is not supported",
731 "8 (streaming)");
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100732 }
733 }
734#if ENABLE_FEATURE_UNZIP_CDF
735 else {
736 /* cdf_offset is valid (and we know the file is seekable) */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100737 cdf_header_t cdf;
738 cdf_offset = read_next_cdf(cdf_offset, &cdf);
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100739 if (cdf_offset == 0) /* EOF? */
740 break;
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100741# if 1
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100742 xlseek(zip_fd,
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100743 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100744 SEEK_SET);
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100745 xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
746 FIX_ENDIANNESS_ZIP(zip);
747 if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200748 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100749 * only from Central Directory.
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200750 */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100751 zip.fmt.crc32 = cdf.fmt.crc32;
752 zip.fmt.cmpsize = cdf.fmt.cmpsize;
753 zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200754 }
Denys Vlasenko38d96692017-07-11 22:10:52 +0200755// Seen in some zipfiles: central directory 9 byte extra field contains
756// a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
757// Local header version:
758// u16 0x5455 ("UT")
759// u16 size (1 + 4 * n)
760// u8 flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
761// u32 mtime
762// u32 atime
763// u32 ctime
764// Central header version:
765// u16 0x5455 ("UT")
766// u16 size (5 (or 1?))
767// u8 flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
768// u32 mtime (CDF does not store atime/ctime)
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100769# else
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100770 /* CDF has the same data as local header, no need to read the latter...
771 * ...not really. An archive was seen with cdf.extra_len == 6 but
772 * zip.extra_len == 0.
773 */
774 memcpy(&zip.fmt.version,
775 &cdf.fmt.version_needed, ZIP_HEADER_LEN);
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100776 xlseek(zip_fd,
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100777 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100778 SEEK_SET);
779# endif
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100780 if ((cdf.fmt.version_made_by >> 8) == 3) {
Denys Vlasenko26cd90c2013-07-21 02:31:08 +0200781 /* This archive is created on Unix */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100782 dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200783 }
Denis Vlasenko48a99712008-07-26 17:32:41 +0000784 }
Denis Vlasenko006e8622008-09-21 01:01:46 +0000785#endif
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100786
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100787 if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100788 /* 0x0001 - encrypted */
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100789 bb_error_msg_and_die("zip flag %s is not supported",
790 "1 (encryption)");
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100791 }
Denys Vlasenkobca4dee2016-04-18 01:14:05 +0200792 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100793 (unsigned)zip.fmt.cmpsize,
794 (unsigned)zip.fmt.extra_len,
795 (unsigned)zip.fmt.ucmpsize
Denys Vlasenkobca4dee2016-04-18 01:14:05 +0200796 );
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000797
798 /* Read filename */
Paul Fox0840b762005-07-20 20:26:49 +0000799 free(dst_fn);
Denys Vlasenkoad37abf2017-07-20 20:21:50 +0200800 die_if_bad_fnamesize(zip.fmt.filename_len);
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100801 dst_fn = xzalloc(zip.fmt.filename_len + 1);
802 xread(zip_fd, dst_fn, zip.fmt.filename_len);
Paul Fox0840b762005-07-20 20:26:49 +0000803 /* Skip extra header bytes */
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100804 unzip_skip(zip.fmt.extra_len);
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000805
Denys Vlasenko8c06bc62015-02-10 01:30:43 +0100806 /* Guard against "/abspath", "/../" and similar attacks */
807 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
808
Paul Fox0840b762005-07-20 20:26:49 +0000809 /* Filter zip entries */
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000810 if (find_list_entry(zreject, dst_fn)
811 || (zaccept && !find_list_entry(zaccept, dst_fn))
812 ) { /* Skip entry */
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100813 goto skip_cmpsize;
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000814 }
815
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200816 if (opts & OPT_l) {
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100817 /* List entry */
818 char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
819 sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
820 (zip.fmt.moddate >> 5) & 0xf, // mm: 0x01e0
821 (zip.fmt.moddate) & 0x1f, // dd: 0x001f
822 (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
823 (zip.fmt.modtime >> 11), // hh: 0xf800
824 (zip.fmt.modtime >> 5) & 0x3f // mm: 0x07e0
825 // seconds/2 not shown, encoded in -- 0x001f
826 );
827 if (!verbose) {
828 // " Length Date Time Name\n"
829 // "--------- ---------- ----- ----"
830 printf( "%9u " "%s " "%s\n",
831 (unsigned)zip.fmt.ucmpsize,
832 dtbuf,
833 dst_fn);
834 } else {
835 char method6[7];
836 unsigned long percents;
837
838 sprintf(method6, "%6u", zip.fmt.method);
839 if (zip.fmt.method == 0) {
840 strcpy(method6, "Stored");
841 }
842 if (zip.fmt.method == 8) {
843 strcpy(method6, "Defl:N");
844 /* normal, maximum, fast, superfast */
845 IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
846 }
847 percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
848 if ((int32_t)percents < 0)
849 percents = 0; /* happens if ucmpsize < cmpsize */
850 percents = percents * 100;
851 if (zip.fmt.ucmpsize)
852 percents /= zip.fmt.ucmpsize;
853 // " Length Method Size Cmpr Date Time CRC-32 Name\n"
854 // "-------- ------ ------- ---- ---------- ----- -------- ----"
855 printf( "%8u %s" "%9u%4u%% " "%s " "%08x " "%s\n",
856 (unsigned)zip.fmt.ucmpsize,
857 method6,
858 (unsigned)zip.fmt.cmpsize,
859 (unsigned)percents,
860 dtbuf,
861 zip.fmt.crc32,
862 dst_fn);
863 total_size += zip.fmt.cmpsize;
864 }
865 total_usize += zip.fmt.ucmpsize;
866 goto skip_cmpsize;
867 }
868
869 if (dst_fd == STDOUT_FILENO) {
870 /* Extracting to STDOUT */
871 goto do_extract;
872 }
Eugene Rudoyc6f213a2017-11-07 08:03:36 +0100873
874 /* Strip paths (after -l: unzip -lj a.zip lists full names) */
875 if (opts & OPT_j)
876 overlapping_strcpy(dst_fn, bb_basename(dst_fn));
877 /* Did this strip everything ("DIR/" case)? Then skip */
878 if (!dst_fn[0])
879 goto skip_cmpsize;
880
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100881 if (last_char_is(dst_fn, '/')) {
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200882 int mode;
883
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100884 /* Extract directory */
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200885 mode = get_lstat_mode(dst_fn);
886 if (mode == -1) { /* ENOENT */
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100887 if (!quiet) {
888 printf(" creating: %s\n", dst_fn);
889 }
890 unzip_create_leading_dirs(dst_fn);
891 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
892 xfunc_die();
893 }
894 } else {
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200895 if (!S_ISDIR(mode)) {
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100896 bb_error_msg_and_die("'%s' exists but is not a %s",
897 dst_fn, "directory");
898 }
899 }
900 goto skip_cmpsize;
901 }
902 check_file:
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200903 /* Does target file already exist? */
904 {
905 int mode = get_lstat_mode(dst_fn);
906 if (mode == -1) {
907 /* ENOENT: does not exist */
908 goto do_open_and_extract;
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100909 }
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200910 if (overwrite == O_NEVER) {
911 goto skip_cmpsize;
912 }
913 if (!S_ISREG(mode)) {
914 fishy:
915 bb_error_msg_and_die("'%s' exists but is not a %s",
916 dst_fn, "regular file");
917 }
918 if (overwrite == O_ALWAYS) {
919 goto do_open_and_extract;
920 }
921 printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
922 my_fgets80(key_buf);
923 /* User input could take a long time. Is it still a regular file? */
924 mode = get_lstat_mode(dst_fn);
925 if (!S_ISREG(mode))
926 goto fishy;
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100927 }
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100928
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200929 /* Extract (or skip) it */
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100930 switch (key_buf[0]) {
Paul Fox0840b762005-07-20 20:26:49 +0000931 case 'A':
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000932 overwrite = O_ALWAYS;
Paul Fox0840b762005-07-20 20:26:49 +0000933 case 'y': /* Open file and fall into unzip */
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100934 do_open_and_extract:
Paul Fox0840b762005-07-20 20:26:49 +0000935 unzip_create_leading_dirs(dst_fn);
Denys Vlasenkoe3c4db82017-01-05 11:43:53 +0100936#if ENABLE_FEATURE_UNZIP_CDF
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200937 dst_fd = -1;
938 if (!S_ISLNK(file_mode)) {
939 dst_fd = xopen3(dst_fn,
940 O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW,
941 file_mode);
942 }
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200943#else
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200944 /* O_NOFOLLOW defends against symlink attacks */
945 dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW);
Denys Vlasenko4e8ff732010-05-24 04:33:02 +0200946#endif
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100947 do_extract:
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200948 if (!quiet) {
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100949 printf(/* zip.fmt.method == 0
950 ? " extracting: %s\n"
951 : */ " inflating: %s\n", dst_fn);
Paul Fox0840b762005-07-20 20:26:49 +0000952 }
Denys Vlasenko13ae85e2017-07-20 18:34:51 +0200953#if ENABLE_FEATURE_UNZIP_CDF
954 if (S_ISLNK(file_mode)) {
Denys Vlasenkobff9bbc2017-07-20 18:56:05 +0200955 if (dst_fd != STDOUT_FILENO) /* not -p? */
Denys Vlasenkoa84db182018-02-20 15:57:45 +0100956 unzip_extract_symlink(&symlink_placeholders, &zip, dst_fn);
Denys Vlasenko13ae85e2017-07-20 18:34:51 +0200957 } else
958#endif
959 {
960 unzip_extract(&zip, dst_fd);
961 if (dst_fd != STDOUT_FILENO) {
962 /* closing STDOUT is potentially bad for future business */
963 close(dst_fd);
964 }
Paul Fox0840b762005-07-20 20:26:49 +0000965 }
966 break;
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000967
Paul Fox0840b762005-07-20 20:26:49 +0000968 case 'N':
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +0000969 overwrite = O_NEVER;
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100970 case 'n': /* Skip entry data */
971 skip_cmpsize:
Denys Vlasenko0ffac1c2017-01-08 14:14:19 +0100972 unzip_skip(zip.fmt.cmpsize);
Paul Fox0840b762005-07-20 20:26:49 +0000973 break;
974
975 case 'r':
976 /* Prompt for new name */
977 printf("new name: ");
Denys Vlasenkobf998072013-01-22 11:16:08 +0100978 my_fgets80(key_buf);
Paul Fox0840b762005-07-20 20:26:49 +0000979 free(dst_fn);
Rob Landleyd921b2e2006-08-03 15:41:12 +0000980 dst_fn = xstrdup(key_buf);
Paul Fox0840b762005-07-20 20:26:49 +0000981 chomp(dst_fn);
Denis Vlasenko48a99712008-07-26 17:32:41 +0000982 goto check_file;
Paul Fox0840b762005-07-20 20:26:49 +0000983
984 default:
Denys Vlasenko8c1d8572017-01-09 13:10:10 +0100985 printf("error: invalid response [%c]\n", (char)key_buf[0]);
Denis Vlasenko48a99712008-07-26 17:32:41 +0000986 goto check_file;
Paul Fox0840b762005-07-20 20:26:49 +0000987 }
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000988
Denis Vlasenko006e8622008-09-21 01:01:46 +0000989 total_entries++;
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000990 }
Glenn L McGrath7ca04f32002-09-25 02:47:48 +0000991
Denys Vlasenkoa84db182018-02-20 15:57:45 +0100992#if ENABLE_FEATURE_UNZIP_CDF
993 create_symlinks_from_list(symlink_placeholders);
994#endif
995
Denys Vlasenko997ad2c2017-07-20 20:04:49 +0200996 if ((opts & OPT_l) && quiet <= 1) {
Denys Vlasenko386bc9f2009-09-06 16:52:50 +0200997 if (!verbose) {
Denys Vlasenko07bd9792016-04-18 01:43:24 +0200998 // " Length Date Time Name\n"
999 // "--------- ---------- ----- ----"
1000 printf( " --------%21s" "-------\n"
1001 "%9lu%21s" "%u files\n",
1002 "",
1003 total_usize, "", total_entries);
Denys Vlasenko386bc9f2009-09-06 16:52:50 +02001004 } else {
1005 unsigned long percents = total_usize - total_size;
Denys Vlasenko07bd9792016-04-18 01:43:24 +02001006 if ((long)percents < 0)
1007 percents = 0; /* happens if usize < size */
Denys Vlasenko386bc9f2009-09-06 16:52:50 +02001008 percents = percents * 100;
1009 if (total_usize)
1010 percents /= total_usize;
Denys Vlasenko07bd9792016-04-18 01:43:24 +02001011 // " Length Method Size Cmpr Date Time CRC-32 Name\n"
1012 // "-------- ------ ------- ---- ---------- ----- -------- ----"
1013 printf( "-------- ------- ----%28s" "----\n"
1014 "%8lu" "%17lu%4u%%%28s" "%u files\n",
1015 "",
1016 total_usize, total_size, (unsigned)percents, "",
Denys Vlasenko386bc9f2009-09-06 16:52:50 +02001017 total_entries);
1018 }
Glenn L McGrath7ca04f32002-09-25 02:47:48 +00001019 }
1020
Denis Vlasenkobc7c5d02007-10-18 23:27:46 +00001021 return 0;
Glenn L McGrath87ac7022002-01-02 13:52:26 +00001022}