unzip: hmm... gcc doesn't like ATTRIBUTE_PACKED?? Document that...
[people/mcb30/busybox.git] / archival / unzip.c
1 /* vi: set sw=4 ts=4: */
2 /*
3  * Mini unzip implementation for busybox
4  *
5  * 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.
9  *
10  * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
11  */
12
13 /* For reference see
14  * http://www.pkware.com/company/standards/appnote/
15  * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
16  */
17
18 /* TODO
19  * Endian issues
20  * Zip64 + other methods
21  * Improve handling of zip format, ie.
22  * - deferred CRC, comp. & uncomp. lengths (zip header flags bit 3)
23  * - unix file permissions, etc.
24  * - central directory
25  */
26
27 #include "libbb.h"
28 #include "unarchive.h"
29
30 enum {
31 #if BB_BIG_ENDIAN
32         ZIP_FILEHEADER_MAGIC = 0x504b0304,
33         ZIP_CDS_MAGIC        = 0x504b0102,
34         ZIP_CDS_END_MAGIC    = 0x504b0506,
35         ZIP_DD_MAGIC         = 0x504b0708,
36 #else
37         ZIP_FILEHEADER_MAGIC = 0x04034b50,
38         ZIP_CDS_MAGIC        = 0x02014b50,
39         ZIP_CDS_END_MAGIC    = 0x06054b50,
40         ZIP_DD_MAGIC         = 0x08074b50,
41 #endif
42 };
43
44 #define ZIP_HEADER_LEN 26
45
46 typedef union {
47         uint8_t raw[ZIP_HEADER_LEN];
48         struct {
49                 uint16_t version;                       /* 0-1 */
50                 uint16_t flags;                         /* 2-3 */
51                 uint16_t method;                        /* 4-5 */
52                 uint16_t modtime;                       /* 6-7 */
53                 uint16_t moddate;                       /* 8-9 */
54                 uint32_t crc32 ATTRIBUTE_PACKED;        /* 10-13 */
55                 uint32_t cmpsize ATTRIBUTE_PACKED;      /* 14-17 */
56                 uint32_t ucmpsize ATTRIBUTE_PACKED;     /* 18-21 */
57                 uint16_t filename_len;                  /* 22-23 */
58                 uint16_t extra_len;                     /* 24-25 */
59         } formatted ATTRIBUTE_PACKED;
60 } zip_header_t; /* ATTRIBUTE_PACKED - gcc 4.2.1 doesn't like it (spews warning) */
61
62 /* Check the offset of the last element, not the length.  This leniency
63  * allows for poor packing, whereby the overall struct may be too long,
64  * even though the elements are all in the right place.
65  */
66 struct BUG_zip_header_must_be_26_bytes {
67         char BUG_zip_header_must_be_26_bytes[
68                 offsetof(zip_header_t, formatted.extra_len) + 2 ==
69                         ZIP_HEADER_LEN ? 1 : -1];
70 };
71
72 #define FIX_ENDIANNESS(zip_header) do { \
73         (zip_header).formatted.version      = SWAP_LE16((zip_header).formatted.version     ); \
74         (zip_header).formatted.flags        = SWAP_LE16((zip_header).formatted.flags       ); \
75         (zip_header).formatted.method       = SWAP_LE16((zip_header).formatted.method      ); \
76         (zip_header).formatted.modtime      = SWAP_LE16((zip_header).formatted.modtime     ); \
77         (zip_header).formatted.moddate      = SWAP_LE16((zip_header).formatted.moddate     ); \
78         (zip_header).formatted.crc32        = SWAP_LE32((zip_header).formatted.crc32       ); \
79         (zip_header).formatted.cmpsize      = SWAP_LE32((zip_header).formatted.cmpsize     ); \
80         (zip_header).formatted.ucmpsize     = SWAP_LE32((zip_header).formatted.ucmpsize    ); \
81         (zip_header).formatted.filename_len = SWAP_LE16((zip_header).formatted.filename_len); \
82         (zip_header).formatted.extra_len    = SWAP_LE16((zip_header).formatted.extra_len   ); \
83 } while (0)
84
85 static void unzip_skip(int fd, off_t skip)
86 {
87         bb_copyfd_exact_size(fd, -1, skip);
88 }
89
90 static void unzip_create_leading_dirs(const char *fn)
91 {
92         /* Create all leading directories */
93         char *name = xstrdup(fn);
94         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
95                 bb_error_msg_and_die("exiting"); /* bb_make_directory is noisy */
96         }
97         free(name);
98 }
99
100 static void unzip_extract(zip_header_t *zip_header, int src_fd, int dst_fd)
101 {
102         if (zip_header->formatted.method == 0) {
103                 /* Method 0 - stored (not compressed) */
104                 off_t size = zip_header->formatted.ucmpsize;
105                 if (size)
106                         bb_copyfd_exact_size(src_fd, dst_fd, size);
107         } else {
108                 /* Method 8 - inflate */
109                 inflate_unzip_result res;
110                 if (inflate_unzip(&res, zip_header->formatted.cmpsize, src_fd, dst_fd) < 0)
111                         bb_error_msg_and_die("inflate error");
112                 /* Validate decompression - crc */
113                 if (zip_header->formatted.crc32 != (res.crc ^ 0xffffffffL)) {
114                         bb_error_msg_and_die("crc error");
115                 }
116                 /* Validate decompression - size */
117                 if (zip_header->formatted.ucmpsize != res.bytes_out) {
118                         /* Don't die. Who knows, maybe len calculation
119                          * was botched somewhere. After all, crc matched! */
120                         bb_error_msg("bad length");
121                 }
122         }
123 }
124
125 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
126 int unzip_main(int argc, char **argv)
127 {
128         enum { O_PROMPT, O_NEVER, O_ALWAYS };
129
130         zip_header_t zip_header;
131         smallint verbose = 1;
132         smallint listing = 0;
133         smallint overwrite = O_PROMPT;
134         unsigned total_size;
135         unsigned total_entries;
136         int src_fd = -1;
137         int dst_fd = -1;
138         char *src_fn = NULL;
139         char *dst_fn = NULL;
140         llist_t *zaccept = NULL;
141         llist_t *zreject = NULL;
142         char *base_dir = NULL;
143         int i, opt;
144         int opt_range = 0;
145         char key_buf[80];
146         struct stat stat_buf;
147
148         /* '-' makes getopt return 1 for non-options */
149         while ((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
150                 switch (opt_range) {
151                 case 0: /* Options */
152                         switch (opt) {
153                         case 'l': /* List */
154                                 listing = 1;
155                                 break;
156
157                         case 'n': /* Never overwrite existing files */
158                                 overwrite = O_NEVER;
159                                 break;
160
161                         case 'o': /* Always overwrite existing files */
162                                 overwrite = O_ALWAYS;
163                                 break;
164
165                         case 'p': /* Extract files to stdout and fall through to set verbosity */
166                                 dst_fd = STDOUT_FILENO;
167
168                         case 'q': /* Be quiet */
169                                 verbose = 0;
170                                 break;
171
172                         case 1: /* The zip file */
173                                 /* +5: space for ".zip" and NUL */
174                                 src_fn = xmalloc(strlen(optarg) + 5);
175                                 strcpy(src_fn, optarg);
176                                 opt_range++;
177                                 break;
178
179                         default:
180                                 bb_show_usage();
181
182                         }
183                         break;
184
185                 case 1: /* Include files */
186                         if (opt == 1) {
187                                 llist_add_to(&zaccept, optarg);
188                                 break;
189                         }
190                         if (opt == 'd') {
191                                 base_dir = optarg;
192                                 opt_range += 2;
193                                 break;
194                         }
195                         if (opt == 'x') {
196                                 opt_range++;
197                                 break;
198                         }
199                         bb_show_usage();
200
201                 case 2 : /* Exclude files */
202                         if (opt == 1) {
203                                 llist_add_to(&zreject, optarg);
204                                 break;
205                         }
206                         if (opt == 'd') { /* Extract to base directory */
207                                 base_dir = optarg;
208                                 opt_range++;
209                                 break;
210                         }
211                         /* fall through */
212
213                 default:
214                         bb_show_usage();
215                 }
216         }
217
218         if (src_fn == NULL) {
219                 bb_show_usage();
220         }
221
222         /* Open input file */
223         if (LONE_DASH(src_fn)) {
224                 src_fd = STDIN_FILENO;
225                 /* Cannot use prompt mode since zip data is arriving on STDIN */
226                 if (overwrite == O_PROMPT)
227                         overwrite = O_NEVER;
228         } else {
229                 static const char extn[][5] = {"", ".zip", ".ZIP"};
230                 int orig_src_fn_len = strlen(src_fn);
231
232                 for (i = 0; (i < 3) && (src_fd == -1); i++) {
233                         strcpy(src_fn + orig_src_fn_len, extn[i]);
234                         src_fd = open(src_fn, O_RDONLY);
235                 }
236                 if (src_fd == -1) {
237                         src_fn[orig_src_fn_len] = '\0';
238                         bb_error_msg_and_die("can't open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
239                 }
240         }
241
242         /* Change dir if necessary */
243         if (base_dir)
244                 xchdir(base_dir);
245
246         if (verbose) {
247                 printf("Archive:  %s\n", src_fn);
248                 if (listing){
249                         puts("  Length     Date   Time    Name\n"
250                              " --------    ----   ----    ----");
251                 }
252         }
253
254         total_size = 0;
255         total_entries = 0;
256         while (1) {
257                 uint32_t magic;
258
259                 /* Check magic number */
260                 xread(src_fd, &magic, 4);
261                 if (magic == ZIP_CDS_MAGIC)
262                         break;
263                 if (magic != ZIP_FILEHEADER_MAGIC)
264                         bb_error_msg_and_die("invalid zip magic %08X", magic);
265
266                 /* Read the file header */
267                 xread(src_fd, zip_header.raw, ZIP_HEADER_LEN);
268                 FIX_ENDIANNESS(zip_header);
269                 if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
270                         bb_error_msg_and_die("unsupported method %d", zip_header.formatted.method);
271                 }
272
273                 /* Read filename */
274                 free(dst_fn);
275                 dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
276                 xread(src_fd, dst_fn, zip_header.formatted.filename_len);
277
278                 /* Skip extra header bytes */
279                 unzip_skip(src_fd, zip_header.formatted.extra_len);
280
281                 /* Filter zip entries */
282                 if (find_list_entry(zreject, dst_fn)
283                  || (zaccept && !find_list_entry(zaccept, dst_fn))
284                 ) { /* Skip entry */
285                         i = 'n';
286
287                 } else { /* Extract entry */
288                         if (listing) { /* List entry */
289                                 if (verbose) {
290                                         unsigned dostime = zip_header.formatted.modtime | (zip_header.formatted.moddate << 16);
291                                         printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
292                                            zip_header.formatted.ucmpsize,
293                                            (dostime & 0x01e00000) >> 21,
294                                            (dostime & 0x001f0000) >> 16,
295                                            (((dostime & 0xfe000000) >> 25) + 1980) % 100,
296                                            (dostime & 0x0000f800) >> 11,
297                                            (dostime & 0x000007e0) >> 5,
298                                            dst_fn);
299                                         total_size += zip_header.formatted.ucmpsize;
300                                         total_entries++;
301                                 } else {
302                                         /* short listing -- filenames only */
303                                         puts(dst_fn);
304                                 }
305                                 i = 'n';
306                         } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
307                                 i = -1;
308                         } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
309                                 if (stat(dst_fn, &stat_buf) == -1) {
310                                         if (errno != ENOENT) {
311                                                 bb_perror_msg_and_die("cannot stat '%s'",dst_fn);
312                                         }
313                                         if (verbose) {
314                                                 printf("   creating: %s\n", dst_fn);
315                                         }
316                                         unzip_create_leading_dirs(dst_fn);
317                                         if (bb_make_directory(dst_fn, 0777, 0)) {
318                                                 bb_error_msg_and_die("exiting");
319                                         }
320                                 } else {
321                                         if (!S_ISDIR(stat_buf.st_mode)) {
322                                                 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
323                                         }
324                                 }
325                                 i = 'n';
326
327                         } else {  /* Extract file */
328  _check_file:
329                                 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
330                                         if (errno != ENOENT) {
331                                                 bb_perror_msg_and_die("cannot stat '%s'",dst_fn);
332                                         }
333                                         i = 'y';
334                                 } else { /* File already exists */
335                                         if (overwrite == O_NEVER) {
336                                                 i = 'n';
337                                         } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
338                                                 if (overwrite == O_ALWAYS) {
339                                                         i = 'y';
340                                                 } else {
341                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
342                                                         if (!fgets(key_buf, sizeof(key_buf), stdin)) {
343                                                                 bb_perror_msg_and_die("cannot read input");
344                                                         }
345                                                         i = key_buf[0];
346                                                 }
347                                         } else { /* File is not regular file */
348                                                 bb_error_msg_and_die("'%s' exists but is not regular file",dst_fn);
349                                         }
350                                 }
351                         }
352                 }
353
354                 switch (i) {
355                 case 'A':
356                         overwrite = O_ALWAYS;
357                 case 'y': /* Open file and fall into unzip */
358                         unzip_create_leading_dirs(dst_fn);
359                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
360                 case -1: /* Unzip */
361                         if (verbose) {
362                                 printf("  inflating: %s\n", dst_fn);
363                         }
364                         unzip_extract(&zip_header, src_fd, dst_fd);
365                         if (dst_fd != STDOUT_FILENO) {
366                                 /* closing STDOUT is potentially bad for future business */
367                                 close(dst_fd);
368                         }
369                         break;
370
371                 case 'N':
372                         overwrite = O_NEVER;
373                 case 'n':
374                         /* Skip entry data */
375                         unzip_skip(src_fd, zip_header.formatted.cmpsize);
376                         break;
377
378                 case 'r':
379                         /* Prompt for new name */
380                         printf("new name: ");
381                         if (!fgets(key_buf, sizeof(key_buf), stdin)) {
382                                 bb_perror_msg_and_die("cannot read input");
383                         }
384                         free(dst_fn);
385                         dst_fn = xstrdup(key_buf);
386                         chomp(dst_fn);
387                         goto _check_file;
388
389                 default:
390                         printf("error: invalid response [%c]\n",(char)i);
391                         goto _check_file;
392                 }
393
394                 /* Data descriptor section */
395                 if (zip_header.formatted.flags & 4) {
396                         /* skip over duplicate crc, compressed size and uncompressed size */
397                         unzip_skip(src_fd, 12);
398                 }
399         }
400
401         if (listing && verbose) {
402                 printf(" --------                   -------\n"
403                        "%9d                   %d files\n",
404                        total_size, total_entries);
405         }
406
407         return 0;
408 }