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: "[-lnopq] FILE[.zip] [FILE]... [-x FILE...] [-d DIR]"
24 //usage:#define unzip_full_usage "\n\n"
25 //usage: "Extract FILEs from ZIP archive\n"
26 //usage: "\n -l List contents (with -q for short form)"
27 //usage: "\n -n Never overwrite files (default: ask)"
28 //usage: "\n -o Overwrite"
29 //usage: "\n -p Print to stdout"
30 //usage: "\n -q Quiet"
31 //usage: "\n -x FILE Exclude FILEs"
32 //usage: "\n -d DIR Extract 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 static void my_fgets80(char *buf80)
273 if (!fgets(buf80, 80, stdin)) {
274 bb_perror_msg_and_die("can't read standard input");
278 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
279 int unzip_main(int argc, char **argv)
281 enum { O_PROMPT, O_NEVER, O_ALWAYS };
283 zip_header_t zip_header;
285 IF_NOT_DESKTOP(const) smallint verbose = 0;
286 smallint listing = 0;
287 smallint overwrite = O_PROMPT;
292 unsigned long total_usize;
293 unsigned long total_size;
294 unsigned total_entries;
298 llist_t *zaccept = NULL;
299 llist_t *zreject = NULL;
300 char *base_dir = NULL;
302 char key_buf[80]; /* must match size used by my_fgets80 */
303 struct stat stat_buf;
305 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
307 * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
308 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
309 * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
310 * Length Method Size Ratio Date Time CRC-32 Name
311 * -------- ------ ------- ----- ---- ---- ------ ----
312 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
313 * -------- ------- --- -------
314 * 204372 35278 83% 1 file
315 * # /usr/bin/unzip -v decompress_unlzma.i.zip
316 * Archive: decompress_unlzma.i.zip
317 * Length Method Size Ratio Date Time CRC-32 Name
318 * -------- ------ ------- ----- ---- ---- ------ ----
319 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
320 * -------- ------- --- -------
321 * 204372 35278 83% 1 file
322 * # unzip -v decompress_unlzma.i.zip
323 * Archive: 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 -qq decompress_unlzma.i.zip
330 * 204372 09-06-09 14:23 decompress_unlzma.i
331 * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
332 * Length Date Time Name
333 * -------- ---- ---- ----
334 * 204372 09-06-09 14:23 decompress_unlzma.i
337 * # /usr/bin/unzip -l decompress_unlzma.i.zip
338 * Archive: decompress_unlzma.i.zip
339 * Length Date Time Name
340 * -------- ---- ---- ----
341 * 204372 09-06-09 14:23 decompress_unlzma.i
347 /* '-' makes getopt return 1 for non-options */
348 while ((opt = getopt(argc, argv, "-d:lnopqxv")) != -1) {
350 case 'd': /* Extract to base directory */
358 case 'n': /* Never overwrite existing files */
362 case 'o': /* Always overwrite existing files */
363 overwrite = O_ALWAYS;
366 case 'p': /* Extract files to stdout and fall through to set verbosity */
367 dst_fd = STDOUT_FILENO;
369 case 'q': /* Be quiet */
373 case 'v': /* Verbose list */
374 IF_DESKTOP(verbose++;)
385 /* +5: space for ".zip" and NUL */
386 src_fn = xmalloc(strlen(optarg) + 5);
387 strcpy(src_fn, optarg);
388 } else if (!x_opt_seen) {
390 llist_add_to(&zaccept, optarg);
393 llist_add_to(&zreject, optarg);
404 * This code is needed for non-GNU getopt
405 * which doesn't understand "-" in option string.
406 * The -x option won't work properly in this case:
407 * "unzip a.zip q -x w e" will be interpreted as
408 * "unzip a.zip q w e -x" = "unzip a.zip q w e"
412 /* +5: space for ".zip" and NUL */
413 src_fn = xmalloc(strlen(argv[0]) + 5);
414 strcpy(src_fn, argv[0]);
416 llist_add_to(&zaccept, *argv);
424 /* Open input file */
425 if (LONE_DASH(src_fn)) {
426 xdup2(STDIN_FILENO, zip_fd);
427 /* Cannot use prompt mode since zip data is arriving on STDIN */
428 if (overwrite == O_PROMPT)
431 static const char extn[][5] = { ".zip", ".ZIP" };
432 char *ext = src_fn + strlen(src_fn);
437 src_fd = open(src_fn, O_RDONLY);
442 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
444 strcpy(ext, extn[i - 1]);
446 xmove_fd(src_fd, zip_fd);
449 /* Change dir if necessary */
453 if (quiet <= 1) { /* not -qq */
455 printf("Archive: %s\n", src_fn);
458 " Length Method Size Ratio Date Time CRC-32 Name\n"
459 "-------- ------ ------- ----- ---- ---- ------ ----"
461 " Length Date Time Name\n"
462 " -------- ---- ---- ----"
467 /* Example of an archive with one 0-byte long file named 'z'
468 * created by Zip 2.31 on Unix:
469 * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
470 * sig........ vneed flags compr mtime mdate crc32>
471 * 0010 00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
472 * >..... csize...... usize...... fnlen exlen fn ex>
473 * 0020 54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
474 * >tra_field......................................
475 * 0030 00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
476 * ........... sig........ vmade vneed flags compr
477 * 0040 42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
478 * mtime mdate crc32...... csize...... usize......
479 * 0050 01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
480 * fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
481 * 0060 00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
482 * >..... fn extra_field...........................
483 * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
484 * 0080 34 00 00 00 00 00 |4.....|
494 mode_t dir_mode = 0777;
496 mode_t file_mode = 0666;
499 /* Check magic number */
500 xread(zip_fd, &magic, 4);
501 /* Central directory? It's at the end, so exit */
502 if (magic == ZIP_CDF_MAGIC)
505 /* Data descriptor? It was a streaming file, go on */
506 if (magic == ZIP_DD_MAGIC) {
507 /* skip over duplicate crc32, cmpsize and ucmpsize */
512 if (magic != ZIP_FILEHEADER_MAGIC)
513 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
515 /* Read the file header */
516 xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);
517 FIX_ENDIANNESS_ZIP(zip_header);
518 if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
519 bb_error_msg_and_die("unsupported method %d", zip_header.formatted.method);
522 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0009)) {
523 bb_error_msg_and_die("zip flags 1 and 8 are not supported");
526 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0001)) {
527 /* 0x0001 - encrypted */
528 bb_error_msg_and_die("zip flag 1 (encryption) is not supported");
532 cdf_header_t cdf_header;
533 cdf_offset = read_next_cdf(cdf_offset, &cdf_header);
534 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0008)) {
535 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
536 * only from Central Directory. See unzip_doc.txt */
537 zip_header.formatted.crc32 = cdf_header.formatted.crc32;
538 zip_header.formatted.cmpsize = cdf_header.formatted.cmpsize;
539 zip_header.formatted.ucmpsize = cdf_header.formatted.ucmpsize;
541 if ((cdf_header.formatted.version_made_by >> 8) == 3) {
542 /* this archive is created on Unix */
543 dir_mode = file_mode = (cdf_header.formatted.external_file_attributes >> 16);
550 dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
551 xread(zip_fd, dst_fn, zip_header.formatted.filename_len);
553 /* Skip extra header bytes */
554 unzip_skip(zip_header.formatted.extra_len);
556 /* Filter zip entries */
557 if (find_list_entry(zreject, dst_fn)
558 || (zaccept && !find_list_entry(zaccept, dst_fn))
562 } else { /* Extract entry */
563 if (listing) { /* List entry */
564 unsigned dostime = zip_header.formatted.modtime | (zip_header.formatted.moddate << 16);
566 // " Length Date Time Name\n"
567 // " -------- ---- ---- ----"
568 printf( "%9u %02u-%02u-%02u %02u:%02u %s\n",
569 (unsigned)zip_header.formatted.ucmpsize,
570 (dostime & 0x01e00000) >> 21,
571 (dostime & 0x001f0000) >> 16,
572 (((dostime & 0xfe000000) >> 25) + 1980) % 100,
573 (dostime & 0x0000f800) >> 11,
574 (dostime & 0x000007e0) >> 5,
576 total_usize += zip_header.formatted.ucmpsize;
578 unsigned long percents = zip_header.formatted.ucmpsize - zip_header.formatted.cmpsize;
579 percents = percents * 100;
580 if (zip_header.formatted.ucmpsize)
581 percents /= zip_header.formatted.ucmpsize;
582 // " Length Method Size Ratio Date Time CRC-32 Name\n"
583 // "-------- ------ ------- ----- ---- ---- ------ ----"
584 printf( "%8u Defl:N" "%9u%4u%% %02u-%02u-%02u %02u:%02u %08x %s\n",
585 (unsigned)zip_header.formatted.ucmpsize,
586 (unsigned)zip_header.formatted.cmpsize,
588 (dostime & 0x01e00000) >> 21,
589 (dostime & 0x001f0000) >> 16,
590 (((dostime & 0xfe000000) >> 25) + 1980) % 100,
591 (dostime & 0x0000f800) >> 11,
592 (dostime & 0x000007e0) >> 5,
593 zip_header.formatted.crc32,
595 total_usize += zip_header.formatted.ucmpsize;
596 total_size += zip_header.formatted.cmpsize;
599 } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
601 } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
602 if (stat(dst_fn, &stat_buf) == -1) {
603 if (errno != ENOENT) {
604 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
607 printf(" creating: %s\n", dst_fn);
609 unzip_create_leading_dirs(dst_fn);
610 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
614 if (!S_ISDIR(stat_buf.st_mode)) {
615 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
620 } else { /* Extract file */
622 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
623 if (errno != ENOENT) {
624 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
627 } else { /* File already exists */
628 if (overwrite == O_NEVER) {
630 } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
631 if (overwrite == O_ALWAYS) {
634 printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
638 } else { /* File is not regular file */
639 bb_error_msg_and_die("'%s' exists but is not regular file", dst_fn);
647 overwrite = O_ALWAYS;
648 case 'y': /* Open file and fall into unzip */
649 unzip_create_leading_dirs(dst_fn);
651 dst_fd = xopen3(dst_fn, O_WRONLY | O_CREAT | O_TRUNC, file_mode);
653 dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
657 printf(" inflating: %s\n", dst_fn);
659 unzip_extract(&zip_header, dst_fd);
660 if (dst_fd != STDOUT_FILENO) {
661 /* closing STDOUT is potentially bad for future business */
669 /* Skip entry data */
670 unzip_skip(zip_header.formatted.cmpsize);
674 /* Prompt for new name */
675 printf("new name: ");
678 dst_fn = xstrdup(key_buf);
683 printf("error: invalid response [%c]\n", (char)i);
690 if (listing && quiet <= 1) {
692 // " Length Date Time Name\n"
693 // " -------- ---- ---- ----"
694 printf( " -------- -------\n"
695 "%9lu" " %u files\n",
696 total_usize, total_entries);
698 unsigned long percents = total_usize - total_size;
699 percents = percents * 100;
701 percents /= total_usize;
702 // " Length Method Size Ratio Date Time CRC-32 Name\n"
703 // "-------- ------ ------- ----- ---- ---- ------ ----"
704 printf( "-------- ------- --- -------\n"
705 "%8lu" "%17lu%4u%% %u files\n",
706 total_usize, total_size, (unsigned)percents,