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#define ZIP_FILEHEADER_MAGIC		SWAP_LE32(0x04034b50)
31#define ZIP_CDS_MAGIC			SWAP_LE32(0x02014b50)
32#define ZIP_CDS_END_MAGIC		SWAP_LE32(0x06054b50)
33#define ZIP_DD_MAGIC			SWAP_LE32(0x08074b50)
34
35typedef union {
36	unsigned char raw[26];
37	struct {
38		unsigned short version;	/* 0-1 */
39		unsigned short flags;	/* 2-3 */
40		unsigned short method;	/* 4-5 */
41		unsigned short modtime;	/* 6-7 */
42		unsigned short moddate;	/* 8-9 */
43		unsigned int crc32 ATTRIBUTE_PACKED;	/* 10-13 */
44		unsigned int cmpsize ATTRIBUTE_PACKED;	/* 14-17 */
45		unsigned int ucmpsize ATTRIBUTE_PACKED;	/* 18-21 */
46		unsigned short filename_len;	/* 22-23 */
47		unsigned short extra_len;		/* 24-25 */
48	} formatted ATTRIBUTE_PACKED;
49} zip_header_t;
50
51static void unzip_skip(int fd, off_t skip)
52{
53	if (lseek(fd, skip, SEEK_CUR) == (off_t)-1) {
54		if (errno != ESPIPE)
55			bb_error_msg_and_die("seek failure");
56		bb_copyfd_exact_size(fd, -1, skip);
57	}
58}
59
60static void unzip_create_leading_dirs(char *fn)
61{
62	/* Create all leading directories */
63	char *name = xstrdup(fn);
64	if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
65		bb_error_msg_and_die("exiting"); /* bb_make_directory is noisy */
66	}
67	free(name);
68}
69
70static int unzip_extract(zip_header_t *zip_header, int src_fd, int dst_fd)
71{
72	if (zip_header->formatted.method == 0) {
73		/* Method 0 - stored (not compressed) */
74		off_t size = zip_header->formatted.ucmpsize;
75		if (size)
76			bb_copyfd_exact_size(src_fd, dst_fd, size);
77	} else {
78		/* Method 8 - inflate */
79		inflate_unzip_result res;
80		/* err = */ inflate_unzip(&res, zip_header->formatted.cmpsize, src_fd, dst_fd);
81// we should check for -1 error return
82		/* Validate decompression - crc */
83		if (zip_header->formatted.crc32 != (res.crc ^ 0xffffffffL)) {
84			bb_error_msg("invalid compressed data--%s error", "crc");
85			return 1;
86		}
87		/* Validate decompression - size */
88		if (zip_header->formatted.ucmpsize != res.bytes_out) {
89			bb_error_msg("invalid compressed data--%s error", "length");
90			return 1;
91		}
92	}
93	return 0;
94}
95
96int unzip_main(int argc, char **argv);
97int unzip_main(int argc, char **argv)
98{
99	zip_header_t zip_header;
100	enum {v_silent, v_normal, v_list} verbosity = v_normal;
101	enum {o_prompt, o_never, o_always} overwrite = o_prompt;
102	unsigned int total_size = 0;
103	unsigned int total_entries = 0;
104	int src_fd = -1, dst_fd = -1;
105	char *src_fn = NULL, *dst_fn = NULL;
106	llist_t *zaccept = NULL;
107	llist_t *zreject = NULL;
108	char *base_dir = NULL;
109	int failed, i, opt, opt_range = 0, list_header_done = 0;
110	char key_buf[512];
111	struct stat stat_buf;
112
113	while ((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
114		switch (opt_range) {
115		case 0: /* Options */
116			switch (opt) {
117			case 'l': /* List */
118				verbosity = v_list;
119				break;
120
121			case 'n': /* Never overwrite existing files */
122				overwrite = o_never;
123				break;
124
125			case 'o': /* Always overwrite existing files */
126				overwrite = o_always;
127				break;
128
129			case 'p': /* Extract files to stdout and fall through to set verbosity */
130				dst_fd = STDOUT_FILENO;
131
132			case 'q': /* Be quiet */
133				verbosity = (verbosity == v_normal) ? v_silent : verbosity;
134				break;
135
136			case 1 : /* The zip file */
137				src_fn = xmalloc(strlen(optarg)+4);
138				strcpy(src_fn, optarg);
139				opt_range++;
140				break;
141
142			default:
143				bb_show_usage();
144
145			}
146			break;
147
148		case 1: /* Include files */
149			if (opt == 1) {
150				llist_add_to(&zaccept, optarg);
151
152			} else if (opt == 'd') {
153				base_dir = optarg;
154				opt_range += 2;
155
156			} else if (opt == 'x') {
157				opt_range++;
158
159			} else {
160				bb_show_usage();
161			}
162			break;
163
164		case 2 : /* Exclude files */
165			if (opt == 1) {
166				llist_add_to(&zreject, optarg);
167
168			} else if (opt == 'd') { /* Extract to base directory */
169				base_dir = optarg;
170				opt_range++;
171
172			} else {
173				bb_show_usage();
174			}
175			break;
176
177		default:
178			bb_show_usage();
179		}
180	}
181
182	if (src_fn == NULL) {
183		bb_show_usage();
184	}
185
186	/* Open input file */
187	if (LONE_DASH(src_fn)) {
188		src_fd = STDIN_FILENO;
189		/* Cannot use prompt mode since zip data is arriving on STDIN */
190		overwrite = (overwrite == o_prompt) ? o_never : overwrite;
191	} else {
192		static const char *const extn[] = {"", ".zip", ".ZIP"};
193		int orig_src_fn_len = strlen(src_fn);
194		for (i = 0; (i < 3) && (src_fd == -1); i++) {
195			strcpy(src_fn + orig_src_fn_len, extn[i]);
196			src_fd = open(src_fn, O_RDONLY);
197		}
198		if (src_fd == -1) {
199			src_fn[orig_src_fn_len] = '\0';
200			bb_error_msg_and_die("cannot open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
201		}
202	}
203
204	/* Change dir if necessary */
205	if (base_dir)
206		xchdir(base_dir);
207
208	if (verbosity != v_silent)
209		printf("Archive:  %s\n", src_fn);
210
211	failed = 0;
212
213	while (1) {
214		unsigned int magic;
215
216		/* Check magic number */
217		xread(src_fd, &magic, 4);
218		if (magic == ZIP_CDS_MAGIC) {
219			break;
220		} else if (magic != ZIP_FILEHEADER_MAGIC) {
221			bb_error_msg_and_die("invalid zip magic %08X", magic);
222		}
223
224		/* Read the file header */
225		xread(src_fd, zip_header.raw, 26);
226		zip_header.formatted.version = SWAP_LE32(zip_header.formatted.version);
227		zip_header.formatted.flags = SWAP_LE32(zip_header.formatted.flags);
228		zip_header.formatted.method = SWAP_LE32(zip_header.formatted.method);
229		zip_header.formatted.modtime = SWAP_LE32(zip_header.formatted.modtime);
230		zip_header.formatted.moddate = SWAP_LE32(zip_header.formatted.moddate);
231		zip_header.formatted.crc32 = SWAP_LE32(zip_header.formatted.crc32);
232		zip_header.formatted.cmpsize = SWAP_LE32(zip_header.formatted.cmpsize);
233		zip_header.formatted.ucmpsize = SWAP_LE32(zip_header.formatted.ucmpsize);
234		zip_header.formatted.filename_len = SWAP_LE32(zip_header.formatted.filename_len);
235		zip_header.formatted.extra_len = SWAP_LE32(zip_header.formatted.extra_len);
236		if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
237			bb_error_msg_and_die("unsupported compression method %d", zip_header.formatted.method);
238		}
239
240		/* Read filename */
241		free(dst_fn);
242		dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
243		xread(src_fd, dst_fn, zip_header.formatted.filename_len);
244
245		/* Skip extra header bytes */
246		unzip_skip(src_fd, zip_header.formatted.extra_len);
247
248		if ((verbosity == v_list) && !list_header_done){
249			puts("  Length     Date   Time    Name\n"
250			     " --------    ----   ----    ----");
251			list_header_done = 1;
252		}
253
254		/* Filter zip entries */
255		if (find_list_entry(zreject, dst_fn) ||
256			(zaccept && !find_list_entry(zaccept, dst_fn))) { /* Skip entry */
257			i = 'n';
258
259		} else { /* Extract entry */
260			total_size += zip_header.formatted.ucmpsize;
261
262			if (verbosity == v_list) { /* List entry */
263				unsigned int dostime = zip_header.formatted.modtime | (zip_header.formatted.moddate << 16);
264				printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
265					   zip_header.formatted.ucmpsize,
266					   (dostime & 0x01e00000) >> 21,
267					   (dostime & 0x001f0000) >> 16,
268					   (((dostime & 0xfe000000) >> 25) + 1980) % 100,
269					   (dostime & 0x0000f800) >> 11,
270					   (dostime & 0x000007e0) >> 5,
271					   dst_fn);
272				total_entries++;
273				i = 'n';
274			} else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
275				i = -1;
276			} else if (last_char_is(dst_fn, '/')) { /* Extract directory */
277				if (stat(dst_fn, &stat_buf) == -1) {
278					if (errno != ENOENT) {
279						bb_perror_msg_and_die("cannot stat '%s'",dst_fn);
280					}
281					if (verbosity == v_normal) {
282						printf("   creating: %s\n", dst_fn);
283					}
284					unzip_create_leading_dirs(dst_fn);
285					if (bb_make_directory(dst_fn, 0777, 0)) {
286						bb_error_msg_and_die("exiting");
287					}
288				} else {
289					if (!S_ISDIR(stat_buf.st_mode)) {
290						bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
291					}
292				}
293				i = 'n';
294
295			} else {  /* Extract file */
296 _check_file:
297				if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
298					if (errno != ENOENT) {
299						bb_perror_msg_and_die("cannot stat '%s'",dst_fn);
300					}
301					i = 'y';
302				} else { /* File already exists */
303					if (overwrite == o_never) {
304						i = 'n';
305					} else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
306						if (overwrite == o_always) {
307							i = 'y';
308						} else {
309							printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
310							if (!fgets(key_buf, 512, stdin)) {
311								bb_perror_msg_and_die("cannot read input");
312							}
313							i = key_buf[0];
314						}
315					} else { /* File is not regular file */
316						bb_error_msg_and_die("'%s' exists but is not regular file",dst_fn);
317					}
318				}
319			}
320		}
321
322		switch (i) {
323		case 'A':
324			overwrite = o_always;
325		case 'y': /* Open file and fall into unzip */
326			unzip_create_leading_dirs(dst_fn);
327			dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
328		case -1: /* Unzip */
329			if (verbosity == v_normal) {
330				printf("  inflating: %s\n", dst_fn);
331			}
332			if (unzip_extract(&zip_header, src_fd, dst_fd)) {
333				failed = 1;
334			}
335			if (dst_fd != STDOUT_FILENO) {
336				/* closing STDOUT is potentially bad for future business */
337				close(dst_fd);
338			}
339			break;
340
341		case 'N':
342			overwrite = o_never;
343		case 'n':
344			/* Skip entry data */
345			unzip_skip(src_fd, zip_header.formatted.cmpsize);
346			break;
347
348		case 'r':
349			/* Prompt for new name */
350			printf("new name: ");
351			if (!fgets(key_buf, 512, stdin)) {
352				bb_perror_msg_and_die("cannot read input");
353			}
354			free(dst_fn);
355			dst_fn = xstrdup(key_buf);
356			chomp(dst_fn);
357			goto _check_file;
358
359		default:
360			printf("error: invalid response [%c]\n",(char)i);
361			goto _check_file;
362		}
363
364		/* Data descriptor section */
365		if (zip_header.formatted.flags & 4) {
366			/* skip over duplicate crc, compressed size and uncompressed size */
367			unzip_skip(src_fd, 12);
368		}
369	}
370
371	if (verbosity == v_list) {
372		printf(" --------                   -------\n"
373		       "%9d                   %d files\n", total_size, total_entries);
374	}
375
376	return failed;
377}
378