1 /* vi: set sw=4 ts=4: */
3 * Mini unzip implementation for busybox
5 * Copyright (C) 2004 by Ed Clark
7 * Loosely based on original busybox unzip applet by Laurence Anderson.
8 * All options and features should work in this version.
10 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
14 * http://www.pkware.com/company/standards/appnote/
15 * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
19 * Zip64 + other methods
22 //usage:#define unzip_trivial_usage
23 //usage: "[-opts[modifiers]] FILE[.zip] [LIST] [-x XLIST] [-d DIR]"
24 //usage:#define unzip_full_usage "\n\n"
25 //usage: "Extract files from ZIP archives\n"
26 //usage: "\n -l List archive contents (with -q for short form)"
27 //usage: "\n -n Never overwrite files (default)"
28 //usage: "\n -o Overwrite"
29 //usage: "\n -p Send output to stdout"
30 //usage: "\n -q Quiet"
31 //usage: "\n -x XLST Exclude these files"
32 //usage: "\n -d DIR Extract files into DIR"
35 #include "bb_archive.h"
39 ZIP_FILEHEADER_MAGIC
= 0x504b0304,
40 ZIP_CDF_MAGIC
= 0x504b0102, /* central directory's file header */
41 ZIP_CDE_MAGIC
= 0x504b0506, /* "end of central directory" record */
42 ZIP_DD_MAGIC
= 0x504b0708,
44 ZIP_FILEHEADER_MAGIC
= 0x04034b50,
45 ZIP_CDF_MAGIC
= 0x02014b50,
46 ZIP_CDE_MAGIC
= 0x06054b50,
47 ZIP_DD_MAGIC
= 0x08074b50,
51 #define ZIP_HEADER_LEN 26
54 uint8_t raw
[ZIP_HEADER_LEN
];
56 uint16_t version
; /* 0-1 */
57 uint16_t zip_flags
; /* 2-3 */
58 uint16_t method
; /* 4-5 */
59 uint16_t modtime
; /* 6-7 */
60 uint16_t moddate
; /* 8-9 */
61 uint32_t crc32 PACKED
; /* 10-13 */
62 uint32_t cmpsize PACKED
; /* 14-17 */
63 uint32_t ucmpsize PACKED
; /* 18-21 */
64 uint16_t filename_len
; /* 22-23 */
65 uint16_t extra_len
; /* 24-25 */
67 } zip_header_t
; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
69 /* Check the offset of the last element, not the length. This leniency
70 * allows for poor packing, whereby the overall struct may be too long,
71 * even though the elements are all in the right place.
73 struct BUG_zip_header_must_be_26_bytes
{
74 char BUG_zip_header_must_be_26_bytes
[
75 offsetof(zip_header_t
, formatted
.extra_len
) + 2
76 == ZIP_HEADER_LEN
? 1 : -1];
79 #define FIX_ENDIANNESS_ZIP(zip_header) do { \
80 (zip_header).formatted.version = SWAP_LE16((zip_header).formatted.version ); \
81 (zip_header).formatted.method = SWAP_LE16((zip_header).formatted.method ); \
82 (zip_header).formatted.modtime = SWAP_LE16((zip_header).formatted.modtime ); \
83 (zip_header).formatted.moddate = SWAP_LE16((zip_header).formatted.moddate ); \
84 (zip_header).formatted.crc32 = SWAP_LE32((zip_header).formatted.crc32 ); \
85 (zip_header).formatted.cmpsize = SWAP_LE32((zip_header).formatted.cmpsize ); \
86 (zip_header).formatted.ucmpsize = SWAP_LE32((zip_header).formatted.ucmpsize ); \
87 (zip_header).formatted.filename_len = SWAP_LE16((zip_header).formatted.filename_len); \
88 (zip_header).formatted.extra_len = SWAP_LE16((zip_header).formatted.extra_len ); \
91 #define CDF_HEADER_LEN 42
94 uint8_t raw
[CDF_HEADER_LEN
];
96 /* uint32_t signature; 50 4b 01 02 */
97 uint16_t version_made_by
; /* 0-1 */
98 uint16_t version_needed
; /* 2-3 */
99 uint16_t cdf_flags
; /* 4-5 */
100 uint16_t method
; /* 6-7 */
101 uint16_t mtime
; /* 8-9 */
102 uint16_t mdate
; /* 10-11 */
103 uint32_t crc32
; /* 12-15 */
104 uint32_t cmpsize
; /* 16-19 */
105 uint32_t ucmpsize
; /* 20-23 */
106 uint16_t file_name_length
; /* 24-25 */
107 uint16_t extra_field_length
; /* 26-27 */
108 uint16_t file_comment_length
; /* 28-29 */
109 uint16_t disk_number_start
; /* 30-31 */
110 uint16_t internal_file_attributes
; /* 32-33 */
111 uint32_t external_file_attributes PACKED
; /* 34-37 */
112 uint32_t relative_offset_of_local_header PACKED
; /* 38-41 */
116 struct BUG_cdf_header_must_be_42_bytes
{
117 char BUG_cdf_header_must_be_42_bytes
[
118 offsetof(cdf_header_t
, formatted
.relative_offset_of_local_header
) + 4
119 == CDF_HEADER_LEN
? 1 : -1];
122 #define FIX_ENDIANNESS_CDF(cdf_header) do { \
123 (cdf_header).formatted.crc32 = SWAP_LE32((cdf_header).formatted.crc32 ); \
124 (cdf_header).formatted.cmpsize = SWAP_LE32((cdf_header).formatted.cmpsize ); \
125 (cdf_header).formatted.ucmpsize = SWAP_LE32((cdf_header).formatted.ucmpsize ); \
126 (cdf_header).formatted.file_name_length = SWAP_LE16((cdf_header).formatted.file_name_length); \
127 (cdf_header).formatted.extra_field_length = SWAP_LE16((cdf_header).formatted.extra_field_length); \
128 (cdf_header).formatted.file_comment_length = SWAP_LE16((cdf_header).formatted.file_comment_length); \
130 (cdf_header).formatted.version_made_by = SWAP_LE16((cdf_header).formatted.version_made_by); \
131 (cdf_header).formatted.external_file_attributes = SWAP_LE32((cdf_header).formatted.external_file_attributes); \
135 #define CDE_HEADER_LEN 16
138 uint8_t raw
[CDE_HEADER_LEN
];
140 /* uint32_t signature; 50 4b 05 06 */
141 uint16_t this_disk_no
;
142 uint16_t disk_with_cdf_no
;
143 uint16_t cdf_entries_on_this_disk
;
144 uint16_t cdf_entries_total
;
147 /* uint16_t file_comment_length; */
148 /* .ZIP file comment (variable size) */
152 struct BUG_cde_header_must_be_16_bytes
{
153 char BUG_cde_header_must_be_16_bytes
[
154 sizeof(cde_header_t
) == CDE_HEADER_LEN
? 1 : -1];
157 #define FIX_ENDIANNESS_CDE(cde_header) do { \
158 (cde_header).formatted.cdf_offset = SWAP_LE32((cde_header).formatted.cdf_offset); \
166 #define PEEK_FROM_END 16384
168 /* NB: does not preserve file position! */
169 static uint32_t find_cdf_offset(void)
171 cde_header_t cde_header
;
174 unsigned char *buf
= xzalloc(PEEK_FROM_END
);
176 end
= xlseek(zip_fd
, 0, SEEK_END
);
177 end
-= PEEK_FROM_END
;
180 xlseek(zip_fd
, end
, SEEK_SET
);
181 full_read(zip_fd
, buf
, PEEK_FROM_END
);
184 while (p
<= buf
+ PEEK_FROM_END
- CDE_HEADER_LEN
- 4) {
196 memcpy(cde_header
.raw
, p
+ 1, CDE_HEADER_LEN
);
197 FIX_ENDIANNESS_CDE(cde_header
);
199 return cde_header
.formatted
.cdf_offset
;
202 bb_error_msg_and_die("can't find file table");
205 static uint32_t read_next_cdf(uint32_t cdf_offset
, cdf_header_t
*cdf_ptr
)
209 org
= xlseek(zip_fd
, 0, SEEK_CUR
);
212 cdf_offset
= find_cdf_offset();
214 xlseek(zip_fd
, cdf_offset
+ 4, SEEK_SET
);
215 xread(zip_fd
, cdf_ptr
->raw
, CDF_HEADER_LEN
);
216 FIX_ENDIANNESS_CDF(*cdf_ptr
);
217 cdf_offset
+= 4 + CDF_HEADER_LEN
218 + cdf_ptr
->formatted
.file_name_length
219 + cdf_ptr
->formatted
.extra_field_length
220 + cdf_ptr
->formatted
.file_comment_length
;
222 xlseek(zip_fd
, org
, SEEK_SET
);
227 static void unzip_skip(off_t skip
)
229 if (lseek(zip_fd
, skip
, SEEK_CUR
) == (off_t
)-1)
230 bb_copyfd_exact_size(zip_fd
, -1, skip
);
233 static void unzip_create_leading_dirs(const char *fn
)
235 /* Create all leading directories */
236 char *name
= xstrdup(fn
);
237 if (bb_make_directory(dirname(name
), 0777, FILEUTILS_RECUR
)) {
238 xfunc_die(); /* bb_make_directory is noisy */
243 static void unzip_extract(zip_header_t
*zip_header
, int dst_fd
)
245 if (zip_header
->formatted
.method
== 0) {
246 /* Method 0 - stored (not compressed) */
247 off_t size
= zip_header
->formatted
.ucmpsize
;
249 bb_copyfd_exact_size(zip_fd
, dst_fd
, size
);
251 /* Method 8 - inflate */
252 transformer_aux_data_t aux
;
253 init_transformer_aux_data(&aux
);
254 aux
.bytes_in
= zip_header
->formatted
.cmpsize
;
255 if (inflate_unzip(&aux
, zip_fd
, dst_fd
) < 0)
256 bb_error_msg_and_die("inflate error");
257 /* Validate decompression - crc */
258 if (zip_header
->formatted
.crc32
!= (aux
.crc32
^ 0xffffffffL
)) {
259 bb_error_msg_and_die("crc error");
261 /* Validate decompression - size */
262 if (zip_header
->formatted
.ucmpsize
!= aux
.bytes_out
) {
263 /* Don't die. Who knows, maybe len calculation
264 * was botched somewhere. After all, crc matched! */
265 bb_error_msg("bad length");
270 int unzip_main(int argc
, char **argv
) MAIN_EXTERNALLY_VISIBLE
;
271 int unzip_main(int argc
, char **argv
)
273 enum { O_PROMPT
, O_NEVER
, O_ALWAYS
};
275 zip_header_t zip_header
;
277 IF_NOT_DESKTOP(const) smallint verbose
= 0;
278 smallint listing
= 0;
279 smallint overwrite
= O_PROMPT
;
283 unsigned long total_usize
;
284 unsigned long total_size
;
285 unsigned total_entries
;
289 llist_t
*zaccept
= NULL
;
290 llist_t
*zreject
= NULL
;
291 char *base_dir
= NULL
;
295 struct stat stat_buf
;
297 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
299 * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
300 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
301 * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
302 * Length Method Size Ratio Date Time CRC-32 Name
303 * -------- ------ ------- ----- ---- ---- ------ ----
304 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
305 * -------- ------- --- -------
306 * 204372 35278 83% 1 file
307 * # /usr/bin/unzip -v decompress_unlzma.i.zip
308 * Archive: decompress_unlzma.i.zip
309 * Length Method Size Ratio Date Time CRC-32 Name
310 * -------- ------ ------- ----- ---- ---- ------ ----
311 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
312 * -------- ------- --- -------
313 * 204372 35278 83% 1 file
314 * # unzip -v decompress_unlzma.i.zip
315 * Archive: decompress_unlzma.i.zip
316 * Length Date Time Name
317 * -------- ---- ---- ----
318 * 204372 09-06-09 14:23 decompress_unlzma.i
321 * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
322 * 204372 09-06-09 14:23 decompress_unlzma.i
323 * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
324 * Length Date Time Name
325 * -------- ---- ---- ----
326 * 204372 09-06-09 14:23 decompress_unlzma.i
329 * # /usr/bin/unzip -l decompress_unlzma.i.zip
330 * Archive: decompress_unlzma.i.zip
331 * Length Date Time Name
332 * -------- ---- ---- ----
333 * 204372 09-06-09 14:23 decompress_unlzma.i
338 /* '-' makes getopt return 1 for non-options */
339 while ((opt
= getopt(argc
, argv
, "-d:lnopqxv")) != -1) {
341 case 0: /* Options */
347 case 'n': /* Never overwrite existing files */
351 case 'o': /* Always overwrite existing files */
352 overwrite
= O_ALWAYS
;
355 case 'p': /* Extract files to stdout and fall through to set verbosity */
356 dst_fd
= STDOUT_FILENO
;
358 case 'q': /* Be quiet */
362 case 'v': /* Verbose list */
363 IF_DESKTOP(verbose
++;)
367 case 1: /* The zip file */
368 /* +5: space for ".zip" and NUL */
369 src_fn
= xmalloc(strlen(optarg
) + 5);
370 strcpy(src_fn
, optarg
);
379 case 1: /* Include files */
381 llist_add_to(&zaccept
, optarg
);
395 case 2 : /* Exclude files */
397 llist_add_to(&zreject
, optarg
);
400 if (opt
== 'd') { /* Extract to base directory */
412 if (src_fn
== NULL
) {
416 /* Open input file */
417 if (LONE_DASH(src_fn
)) {
418 xdup2(STDIN_FILENO
, zip_fd
);
419 /* Cannot use prompt mode since zip data is arriving on STDIN */
420 if (overwrite
== O_PROMPT
)
423 static const char extn
[][5] = {"", ".zip", ".ZIP"};
424 int orig_src_fn_len
= strlen(src_fn
);
427 for (i
= 0; (i
< 3) && (src_fd
== -1); i
++) {
428 strcpy(src_fn
+ orig_src_fn_len
, extn
[i
]);
429 src_fd
= open(src_fn
, O_RDONLY
);
432 src_fn
[orig_src_fn_len
] = '\0';
433 bb_error_msg_and_die("can't open %s, %s.zip, %s.ZIP", src_fn
, src_fn
, src_fn
);
435 xmove_fd(src_fd
, zip_fd
);
438 /* Change dir if necessary */
442 if (quiet
<= 1) { /* not -qq */
444 printf("Archive: %s\n", src_fn
);
447 " Length Method Size Ratio Date Time CRC-32 Name\n"
448 "-------- ------ ------- ----- ---- ---- ------ ----"
450 " Length Date Time Name\n"
451 " -------- ---- ---- ----"
456 /* Example of an archive with one 0-byte long file named 'z'
457 * created by Zip 2.31 on Unix:
458 * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
459 * sig........ vneed flags compr mtime mdate crc32>
460 * 0010 00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
461 * >..... csize...... usize...... fnlen exlen fn ex>
462 * 0020 54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
463 * >tra_field......................................
464 * 0030 00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
465 * ........... sig........ vmade vneed flags compr
466 * 0040 42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
467 * mtime mdate crc32...... csize...... usize......
468 * 0050 01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
469 * fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
470 * 0060 00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
471 * >..... fn extra_field...........................
472 * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
473 * 0080 34 00 00 00 00 00 |4.....|
483 mode_t dir_mode
= 0777;
485 mode_t file_mode
= 0666;
488 /* Check magic number */
489 xread(zip_fd
, &magic
, 4);
490 /* Central directory? It's at the end, so exit */
491 if (magic
== ZIP_CDF_MAGIC
)
494 /* Data descriptor? It was a streaming file, go on */
495 if (magic
== ZIP_DD_MAGIC
) {
496 /* skip over duplicate crc32, cmpsize and ucmpsize */
501 if (magic
!= ZIP_FILEHEADER_MAGIC
)
502 bb_error_msg_and_die("invalid zip magic %08X", (int)magic
);
504 /* Read the file header */
505 xread(zip_fd
, zip_header
.raw
, ZIP_HEADER_LEN
);
506 FIX_ENDIANNESS_ZIP(zip_header
);
507 if ((zip_header
.formatted
.method
!= 0) && (zip_header
.formatted
.method
!= 8)) {
508 bb_error_msg_and_die("unsupported method %d", zip_header
.formatted
.method
);
511 if (zip_header
.formatted
.zip_flags
& SWAP_LE16(0x0009)) {
512 bb_error_msg_and_die("zip flags 1 and 8 are not supported");
515 if (zip_header
.formatted
.zip_flags
& SWAP_LE16(0x0001)) {
516 /* 0x0001 - encrypted */
517 bb_error_msg_and_die("zip flag 1 (encryption) is not supported");
521 cdf_header_t cdf_header
;
522 cdf_offset
= read_next_cdf(cdf_offset
, &cdf_header
);
523 if (zip_header
.formatted
.zip_flags
& SWAP_LE16(0x0008)) {
524 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
525 * only from Central Directory. See unzip_doc.txt */
526 zip_header
.formatted
.crc32
= cdf_header
.formatted
.crc32
;
527 zip_header
.formatted
.cmpsize
= cdf_header
.formatted
.cmpsize
;
528 zip_header
.formatted
.ucmpsize
= cdf_header
.formatted
.ucmpsize
;
530 if ((cdf_header
.formatted
.version_made_by
>> 8) == 3) {
531 /* this archive is created on Unix */
532 dir_mode
= file_mode
= (cdf_header
.formatted
.external_file_attributes
>> 16);
539 dst_fn
= xzalloc(zip_header
.formatted
.filename_len
+ 1);
540 xread(zip_fd
, dst_fn
, zip_header
.formatted
.filename_len
);
542 /* Skip extra header bytes */
543 unzip_skip(zip_header
.formatted
.extra_len
);
545 /* Filter zip entries */
546 if (find_list_entry(zreject
, dst_fn
)
547 || (zaccept
&& !find_list_entry(zaccept
, dst_fn
))
551 } else { /* Extract entry */
552 if (listing
) { /* List entry */
553 unsigned dostime
= zip_header
.formatted
.modtime
| (zip_header
.formatted
.moddate
<< 16);
555 // " Length Date Time Name\n"
556 // " -------- ---- ---- ----"
557 printf( "%9u %02u-%02u-%02u %02u:%02u %s\n",
558 (unsigned)zip_header
.formatted
.ucmpsize
,
559 (dostime
& 0x01e00000) >> 21,
560 (dostime
& 0x001f0000) >> 16,
561 (((dostime
& 0xfe000000) >> 25) + 1980) % 100,
562 (dostime
& 0x0000f800) >> 11,
563 (dostime
& 0x000007e0) >> 5,
565 total_usize
+= zip_header
.formatted
.ucmpsize
;
567 unsigned long percents
= zip_header
.formatted
.ucmpsize
- zip_header
.formatted
.cmpsize
;
568 percents
= percents
* 100;
569 if (zip_header
.formatted
.ucmpsize
)
570 percents
/= zip_header
.formatted
.ucmpsize
;
571 // " Length Method Size Ratio Date Time CRC-32 Name\n"
572 // "-------- ------ ------- ----- ---- ---- ------ ----"
573 printf( "%8u Defl:N" "%9u%4u%% %02u-%02u-%02u %02u:%02u %08x %s\n",
574 (unsigned)zip_header
.formatted
.ucmpsize
,
575 (unsigned)zip_header
.formatted
.cmpsize
,
577 (dostime
& 0x01e00000) >> 21,
578 (dostime
& 0x001f0000) >> 16,
579 (((dostime
& 0xfe000000) >> 25) + 1980) % 100,
580 (dostime
& 0x0000f800) >> 11,
581 (dostime
& 0x000007e0) >> 5,
582 zip_header
.formatted
.crc32
,
584 total_usize
+= zip_header
.formatted
.ucmpsize
;
585 total_size
+= zip_header
.formatted
.cmpsize
;
588 } else if (dst_fd
== STDOUT_FILENO
) { /* Extracting to STDOUT */
590 } else if (last_char_is(dst_fn
, '/')) { /* Extract directory */
591 if (stat(dst_fn
, &stat_buf
) == -1) {
592 if (errno
!= ENOENT
) {
593 bb_perror_msg_and_die("can't stat '%s'", dst_fn
);
596 printf(" creating: %s\n", dst_fn
);
598 unzip_create_leading_dirs(dst_fn
);
599 if (bb_make_directory(dst_fn
, dir_mode
, 0)) {
603 if (!S_ISDIR(stat_buf
.st_mode
)) {
604 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn
);
609 } else { /* Extract file */
611 if (stat(dst_fn
, &stat_buf
) == -1) { /* File does not exist */
612 if (errno
!= ENOENT
) {
613 bb_perror_msg_and_die("can't stat '%s'", dst_fn
);
616 } else { /* File already exists */
617 if (overwrite
== O_NEVER
) {
619 } else if (S_ISREG(stat_buf
.st_mode
)) { /* File is regular file */
620 if (overwrite
== O_ALWAYS
) {
623 printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn
);
625 if (!fgets(key_buf
, sizeof(key_buf
), stdin
)) {
626 bb_perror_msg_and_die("can't read input");
630 } else { /* File is not regular file */
631 bb_error_msg_and_die("'%s' exists but is not regular file", dst_fn
);
639 overwrite
= O_ALWAYS
;
640 case 'y': /* Open file and fall into unzip */
641 unzip_create_leading_dirs(dst_fn
);
643 dst_fd
= xopen3(dst_fn
, O_WRONLY
| O_CREAT
| O_TRUNC
, file_mode
);
645 dst_fd
= xopen(dst_fn
, O_WRONLY
| O_CREAT
| O_TRUNC
);
649 printf(" inflating: %s\n", dst_fn
);
651 unzip_extract(&zip_header
, dst_fd
);
652 if (dst_fd
!= STDOUT_FILENO
) {
653 /* closing STDOUT is potentially bad for future business */
661 /* Skip entry data */
662 unzip_skip(zip_header
.formatted
.cmpsize
);
666 /* Prompt for new name */
667 printf("new name: ");
668 if (!fgets(key_buf
, sizeof(key_buf
), stdin
)) {
669 bb_perror_msg_and_die("can't read input");
672 dst_fn
= xstrdup(key_buf
);
677 printf("error: invalid response [%c]\n", (char)i
);
684 if (listing
&& quiet
<= 1) {
686 // " Length Date Time Name\n"
687 // " -------- ---- ---- ----"
688 printf( " -------- -------\n"
689 "%9lu" " %u files\n",
690 total_usize
, total_entries
);
692 unsigned long percents
= total_usize
- total_size
;
693 percents
= percents
* 100;
695 percents
/= total_usize
;
696 // " Length Method Size Ratio Date Time CRC-32 Name\n"
697 // "-------- ------ ------- ----- ---- ---- ------ ----"
698 printf( "-------- ------- --- -------\n"
699 "%8lu" "%17lu%4u%% %u files\n",
700 total_usize
, total_size
, (unsigned)percents
,