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
27 ZIP_FILEHEADER_MAGIC = 0x504b0304,
28 ZIP_CDF_MAGIC = 0x504b0102, /* central directory's file header */
29 ZIP_CDE_MAGIC = 0x504b0506, /* "end of central directory" record */
30 ZIP_DD_MAGIC = 0x504b0708,
32 ZIP_FILEHEADER_MAGIC = 0x04034b50,
33 ZIP_CDF_MAGIC = 0x02014b50,
34 ZIP_CDE_MAGIC = 0x06054b50,
35 ZIP_DD_MAGIC = 0x08074b50,
39 #define ZIP_HEADER_LEN 26
42 uint8_t raw[ZIP_HEADER_LEN];
44 uint16_t version; /* 0-1 */
45 uint16_t zip_flags; /* 2-3 */
46 uint16_t method; /* 4-5 */
47 uint16_t modtime; /* 6-7 */
48 uint16_t moddate; /* 8-9 */
49 uint32_t crc32 PACKED; /* 10-13 */
50 uint32_t cmpsize PACKED; /* 14-17 */
51 uint32_t ucmpsize PACKED; /* 18-21 */
52 uint16_t filename_len; /* 22-23 */
53 uint16_t extra_len; /* 24-25 */
55 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
57 /* Check the offset of the last element, not the length. This leniency
58 * allows for poor packing, whereby the overall struct may be too long,
59 * even though the elements are all in the right place.
61 struct BUG_zip_header_must_be_26_bytes {
62 char BUG_zip_header_must_be_26_bytes[
63 offsetof(zip_header_t, formatted.extra_len) + 2
64 == ZIP_HEADER_LEN ? 1 : -1];
67 #define FIX_ENDIANNESS_ZIP(zip_header) do { \
68 (zip_header).formatted.version = SWAP_LE16((zip_header).formatted.version ); \
69 (zip_header).formatted.method = SWAP_LE16((zip_header).formatted.method ); \
70 (zip_header).formatted.modtime = SWAP_LE16((zip_header).formatted.modtime ); \
71 (zip_header).formatted.moddate = SWAP_LE16((zip_header).formatted.moddate ); \
72 (zip_header).formatted.crc32 = SWAP_LE32((zip_header).formatted.crc32 ); \
73 (zip_header).formatted.cmpsize = SWAP_LE32((zip_header).formatted.cmpsize ); \
74 (zip_header).formatted.ucmpsize = SWAP_LE32((zip_header).formatted.ucmpsize ); \
75 (zip_header).formatted.filename_len = SWAP_LE16((zip_header).formatted.filename_len); \
76 (zip_header).formatted.extra_len = SWAP_LE16((zip_header).formatted.extra_len ); \
79 #define CDF_HEADER_LEN 42
82 uint8_t raw[CDF_HEADER_LEN];
84 /* uint32_t signature; 50 4b 01 02 */
85 uint16_t version_made_by; /* 0-1 */
86 uint16_t version_needed; /* 2-3 */
87 uint16_t cdf_flags; /* 4-5 */
88 uint16_t method; /* 6-7 */
89 uint16_t mtime; /* 8-9 */
90 uint16_t mdate; /* 10-11 */
91 uint32_t crc32; /* 12-15 */
92 uint32_t cmpsize; /* 16-19 */
93 uint32_t ucmpsize; /* 20-23 */
94 uint16_t file_name_length; /* 24-25 */
95 uint16_t extra_field_length; /* 26-27 */
96 uint16_t file_comment_length; /* 28-29 */
97 uint16_t disk_number_start; /* 30-31 */
98 uint16_t internal_file_attributes; /* 32-33 */
99 uint32_t external_file_attributes PACKED; /* 34-37 */
100 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
104 struct BUG_cdf_header_must_be_42_bytes {
105 char BUG_cdf_header_must_be_42_bytes[
106 offsetof(cdf_header_t, formatted.relative_offset_of_local_header) + 4
107 == CDF_HEADER_LEN ? 1 : -1];
110 #define FIX_ENDIANNESS_CDF(cdf_header) do { \
111 (cdf_header).formatted.crc32 = SWAP_LE32((cdf_header).formatted.crc32 ); \
112 (cdf_header).formatted.cmpsize = SWAP_LE32((cdf_header).formatted.cmpsize ); \
113 (cdf_header).formatted.ucmpsize = SWAP_LE32((cdf_header).formatted.ucmpsize ); \
114 (cdf_header).formatted.file_name_length = SWAP_LE16((cdf_header).formatted.file_name_length); \
115 (cdf_header).formatted.extra_field_length = SWAP_LE16((cdf_header).formatted.extra_field_length); \
116 (cdf_header).formatted.file_comment_length = SWAP_LE16((cdf_header).formatted.file_comment_length); \
118 (cdf_header).formatted.version_made_by = SWAP_LE16((cdf_header).formatted.version_made_by); \
119 (cdf_header).formatted.external_file_attributes = SWAP_LE32((cdf_header).formatted.external_file_attributes); \
123 #define CDE_HEADER_LEN 16
126 uint8_t raw[CDE_HEADER_LEN];
128 /* uint32_t signature; 50 4b 05 06 */
129 uint16_t this_disk_no;
130 uint16_t disk_with_cdf_no;
131 uint16_t cdf_entries_on_this_disk;
132 uint16_t cdf_entries_total;
135 /* uint16_t file_comment_length; */
136 /* .ZIP file comment (variable size) */
140 struct BUG_cde_header_must_be_16_bytes {
141 char BUG_cde_header_must_be_16_bytes[
142 sizeof(cde_header_t) == CDE_HEADER_LEN ? 1 : -1];
145 #define FIX_ENDIANNESS_CDE(cde_header) do { \
146 (cde_header).formatted.cdf_offset = SWAP_LE32((cde_header).formatted.cdf_offset); \
154 #define PEEK_FROM_END 16384
156 /* NB: does not preserve file position! */
157 static uint32_t find_cdf_offset(void)
159 cde_header_t cde_header;
162 unsigned char *buf = xzalloc(PEEK_FROM_END);
164 end = xlseek(zip_fd, 0, SEEK_END);
165 end -= PEEK_FROM_END;
168 xlseek(zip_fd, end, SEEK_SET);
169 full_read(zip_fd, buf, PEEK_FROM_END);
172 while (p <= buf + PEEK_FROM_END - CDE_HEADER_LEN - 4) {
184 memcpy(cde_header.raw, p + 1, CDE_HEADER_LEN);
185 FIX_ENDIANNESS_CDE(cde_header);
187 return cde_header.formatted.cdf_offset;
190 bb_error_msg_and_die("can't find file table");
193 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf_ptr)
197 org = xlseek(zip_fd, 0, SEEK_CUR);
200 cdf_offset = find_cdf_offset();
202 xlseek(zip_fd, cdf_offset + 4, SEEK_SET);
203 xread(zip_fd, cdf_ptr->raw, CDF_HEADER_LEN);
204 FIX_ENDIANNESS_CDF(*cdf_ptr);
205 cdf_offset += 4 + CDF_HEADER_LEN
206 + cdf_ptr->formatted.file_name_length
207 + cdf_ptr->formatted.extra_field_length
208 + cdf_ptr->formatted.file_comment_length;
210 xlseek(zip_fd, org, SEEK_SET);
215 static void unzip_skip(off_t skip)
217 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
218 bb_copyfd_exact_size(zip_fd, -1, skip);
221 static void unzip_create_leading_dirs(const char *fn)
223 /* Create all leading directories */
224 char *name = xstrdup(fn);
225 if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
226 bb_error_msg_and_die("exiting"); /* bb_make_directory is noisy */
231 static void unzip_extract(zip_header_t *zip_header, int dst_fd)
233 if (zip_header->formatted.method == 0) {
234 /* Method 0 - stored (not compressed) */
235 off_t size = zip_header->formatted.ucmpsize;
237 bb_copyfd_exact_size(zip_fd, dst_fd, size);
239 /* Method 8 - inflate */
240 inflate_unzip_result res;
241 if (inflate_unzip(&res, zip_header->formatted.cmpsize, zip_fd, dst_fd) < 0)
242 bb_error_msg_and_die("inflate error");
243 /* Validate decompression - crc */
244 if (zip_header->formatted.crc32 != (res.crc ^ 0xffffffffL)) {
245 bb_error_msg_and_die("crc error");
247 /* Validate decompression - size */
248 if (zip_header->formatted.ucmpsize != res.bytes_out) {
249 /* Don't die. Who knows, maybe len calculation
250 * was botched somewhere. After all, crc matched! */
251 bb_error_msg("bad length");
256 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
257 int unzip_main(int argc, char **argv)
259 enum { O_PROMPT, O_NEVER, O_ALWAYS };
261 zip_header_t zip_header;
263 IF_NOT_DESKTOP(const) smallint verbose = 0;
264 smallint listing = 0;
265 smallint overwrite = O_PROMPT;
269 unsigned long total_usize;
270 unsigned long total_size;
271 unsigned total_entries;
275 llist_t *zaccept = NULL;
276 llist_t *zreject = NULL;
277 char *base_dir = NULL;
281 struct stat stat_buf;
283 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
285 * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
286 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
287 * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
288 * Length Method Size Ratio Date Time CRC-32 Name
289 * -------- ------ ------- ----- ---- ---- ------ ----
290 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
291 * -------- ------- --- -------
292 * 204372 35278 83% 1 file
293 * # /usr/bin/unzip -v decompress_unlzma.i.zip
294 * Archive: decompress_unlzma.i.zip
295 * Length Method Size Ratio Date Time CRC-32 Name
296 * -------- ------ ------- ----- ---- ---- ------ ----
297 * 204372 Defl:N 35278 83% 09-06-09 14:23 0d056252 decompress_unlzma.i
298 * -------- ------- --- -------
299 * 204372 35278 83% 1 file
300 * # unzip -v decompress_unlzma.i.zip
301 * Archive: decompress_unlzma.i.zip
302 * Length Date Time Name
303 * -------- ---- ---- ----
304 * 204372 09-06-09 14:23 decompress_unlzma.i
307 * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
308 * 204372 09-06-09 14:23 decompress_unlzma.i
309 * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
310 * Length Date Time Name
311 * -------- ---- ---- ----
312 * 204372 09-06-09 14:23 decompress_unlzma.i
315 * # /usr/bin/unzip -l decompress_unlzma.i.zip
316 * Archive: decompress_unlzma.i.zip
317 * Length Date Time Name
318 * -------- ---- ---- ----
319 * 204372 09-06-09 14:23 decompress_unlzma.i
324 /* '-' makes getopt return 1 for non-options */
325 while ((opt = getopt(argc, argv, "-d:lnopqxv")) != -1) {
327 case 0: /* Options */
333 case 'n': /* Never overwrite existing files */
337 case 'o': /* Always overwrite existing files */
338 overwrite = O_ALWAYS;
341 case 'p': /* Extract files to stdout and fall through to set verbosity */
342 dst_fd = STDOUT_FILENO;
344 case 'q': /* Be quiet */
348 case 'v': /* Verbose list */
349 IF_DESKTOP(verbose++;)
353 case 1: /* The zip file */
354 /* +5: space for ".zip" and NUL */
355 src_fn = xmalloc(strlen(optarg) + 5);
356 strcpy(src_fn, optarg);
365 case 1: /* Include files */
367 llist_add_to(&zaccept, optarg);
381 case 2 : /* Exclude files */
383 llist_add_to(&zreject, optarg);
386 if (opt == 'd') { /* Extract to base directory */
398 if (src_fn == NULL) {
402 /* Open input file */
403 if (LONE_DASH(src_fn)) {
404 xdup2(STDIN_FILENO, zip_fd);
405 /* Cannot use prompt mode since zip data is arriving on STDIN */
406 if (overwrite == O_PROMPT)
409 static const char extn[][5] = {"", ".zip", ".ZIP"};
410 int orig_src_fn_len = strlen(src_fn);
413 for (i = 0; (i < 3) && (src_fd == -1); i++) {
414 strcpy(src_fn + orig_src_fn_len, extn[i]);
415 src_fd = open(src_fn, O_RDONLY);
418 src_fn[orig_src_fn_len] = '\0';
419 bb_error_msg_and_die("can't open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
421 xmove_fd(src_fd, zip_fd);
424 /* Change dir if necessary */
428 if (quiet <= 1) { /* not -qq */
430 printf("Archive: %s\n", src_fn);
433 " Length Method Size Ratio Date Time CRC-32 Name\n"
434 "-------- ------ ------- ----- ---- ---- ------ ----"
436 " Length Date Time Name\n"
437 " -------- ---- ---- ----"
442 /* Example of an archive with one 0-byte long file named 'z'
443 * created by Zip 2.31 on Unix:
444 * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
445 * sig........ vneed flags compr mtime mdate crc32>
446 * 0010 00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
447 * >..... csize...... usize...... fnlen exlen fn ex>
448 * 0020 54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
449 * >tra_field......................................
450 * 0030 00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
451 * ........... sig........ vmade vneed flags compr
452 * 0040 42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
453 * mtime mdate crc32...... csize...... usize......
454 * 0050 01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
455 * fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
456 * 0060 00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
457 * >..... fn extra_field...........................
458 * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
459 * 0080 34 00 00 00 00 00 |4.....|
469 mode_t dir_mode = 0777;
471 mode_t file_mode = 0666;
474 /* Check magic number */
475 xread(zip_fd, &magic, 4);
476 /* Central directory? It's at the end, so exit */
477 if (magic == ZIP_CDF_MAGIC)
480 /* Data descriptor? It was a streaming file, go on */
481 if (magic == ZIP_DD_MAGIC) {
482 /* skip over duplicate crc32, cmpsize and ucmpsize */
487 if (magic != ZIP_FILEHEADER_MAGIC)
488 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
490 /* Read the file header */
491 xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);
492 FIX_ENDIANNESS_ZIP(zip_header);
493 if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
494 bb_error_msg_and_die("unsupported method %d", zip_header.formatted.method);
497 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0009)) {
498 bb_error_msg_and_die("zip flags 1 and 8 are not supported");
501 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0001)) {
502 /* 0x0001 - encrypted */
503 bb_error_msg_and_die("zip flag 1 (encryption) is not supported");
507 cdf_header_t cdf_header;
508 cdf_offset = read_next_cdf(cdf_offset, &cdf_header);
509 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0008)) {
510 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
511 * only from Central Directory. See unzip_doc.txt */
512 zip_header.formatted.crc32 = cdf_header.formatted.crc32;
513 zip_header.formatted.cmpsize = cdf_header.formatted.cmpsize;
514 zip_header.formatted.ucmpsize = cdf_header.formatted.ucmpsize;
516 if ((cdf_header.formatted.version_made_by >> 8) == 3) {
517 /* this archive is created on Unix */
518 dir_mode = file_mode = (cdf_header.formatted.external_file_attributes >> 16);
525 dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
526 xread(zip_fd, dst_fn, zip_header.formatted.filename_len);
528 /* Skip extra header bytes */
529 unzip_skip(zip_header.formatted.extra_len);
531 /* Filter zip entries */
532 if (find_list_entry(zreject, dst_fn)
533 || (zaccept && !find_list_entry(zaccept, dst_fn))
537 } else { /* Extract entry */
538 if (listing) { /* List entry */
539 unsigned dostime = zip_header.formatted.modtime | (zip_header.formatted.moddate << 16);
541 // " Length Date Time Name\n"
542 // " -------- ---- ---- ----"
543 printf( "%9u %02u-%02u-%02u %02u:%02u %s\n",
544 (unsigned)zip_header.formatted.ucmpsize,
545 (dostime & 0x01e00000) >> 21,
546 (dostime & 0x001f0000) >> 16,
547 (((dostime & 0xfe000000) >> 25) + 1980) % 100,
548 (dostime & 0x0000f800) >> 11,
549 (dostime & 0x000007e0) >> 5,
551 total_usize += zip_header.formatted.ucmpsize;
553 unsigned long percents = zip_header.formatted.ucmpsize - zip_header.formatted.cmpsize;
554 percents = percents * 100;
555 if (zip_header.formatted.ucmpsize)
556 percents /= zip_header.formatted.ucmpsize;
557 // " Length Method Size Ratio Date Time CRC-32 Name\n"
558 // "-------- ------ ------- ----- ---- ---- ------ ----"
559 printf( "%8u Defl:N" "%9u%4u%% %02u-%02u-%02u %02u:%02u %08x %s\n",
560 (unsigned)zip_header.formatted.ucmpsize,
561 (unsigned)zip_header.formatted.cmpsize,
563 (dostime & 0x01e00000) >> 21,
564 (dostime & 0x001f0000) >> 16,
565 (((dostime & 0xfe000000) >> 25) + 1980) % 100,
566 (dostime & 0x0000f800) >> 11,
567 (dostime & 0x000007e0) >> 5,
568 zip_header.formatted.crc32,
570 total_usize += zip_header.formatted.ucmpsize;
571 total_size += zip_header.formatted.cmpsize;
574 } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
576 } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
577 if (stat(dst_fn, &stat_buf) == -1) {
578 if (errno != ENOENT) {
579 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
582 printf(" creating: %s\n", dst_fn);
584 unzip_create_leading_dirs(dst_fn);
585 if (bb_make_directory(dst_fn, dir_mode, 0)) {
586 bb_error_msg_and_die("exiting");
589 if (!S_ISDIR(stat_buf.st_mode)) {
590 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
595 } else { /* Extract file */
597 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
598 if (errno != ENOENT) {
599 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
602 } else { /* File already exists */
603 if (overwrite == O_NEVER) {
605 } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
606 if (overwrite == O_ALWAYS) {
609 printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
610 if (!fgets(key_buf, sizeof(key_buf), stdin)) {
611 bb_perror_msg_and_die("can't read input");
615 } else { /* File is not regular file */
616 bb_error_msg_and_die("'%s' exists but is not regular file", dst_fn);
624 overwrite = O_ALWAYS;
625 case 'y': /* Open file and fall into unzip */
626 unzip_create_leading_dirs(dst_fn);
628 dst_fd = xopen3(dst_fn, O_WRONLY | O_CREAT | O_TRUNC, file_mode);
630 dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
634 printf(" inflating: %s\n", dst_fn);
636 unzip_extract(&zip_header, dst_fd);
637 if (dst_fd != STDOUT_FILENO) {
638 /* closing STDOUT is potentially bad for future business */
646 /* Skip entry data */
647 unzip_skip(zip_header.formatted.cmpsize);
651 /* Prompt for new name */
652 printf("new name: ");
653 if (!fgets(key_buf, sizeof(key_buf), stdin)) {
654 bb_perror_msg_and_die("can't read input");
657 dst_fn = xstrdup(key_buf);
662 printf("error: invalid response [%c]\n", (char)i);
669 if (listing && quiet <= 1) {
671 // " Length Date Time Name\n"
672 // " -------- ---- ---- ----"
673 printf( " -------- -------\n"
674 "%9lu" " %u files\n",
675 total_usize, total_entries);
677 unsigned long percents = total_usize - total_size;
678 percents = percents * 100;
680 percents /= total_usize;
681 // " Length Method Size Ratio Date Time CRC-32 Name\n"
682 // "-------- ------ ------- ----- ---- ---- ------ ----"
683 printf( "-------- ------- --- -------\n"
684 "%8lu" "%17lu%4u%% %u files\n",
685 total_usize, total_size, (unsigned)percents,