1 /*-
2  * Copyright (c) 2003-2007 Tim Kientzle
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
15  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
16  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
17  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
18  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
19  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
20  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
21  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
23  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24  */
25 
26 /*
27  * This code borrows heavily from "compress" source code, which is
28  * protected by the following copyright.  (Clause 3 dropped by request
29  * of the Regents.)
30  */
31 
32 /*-
33  * Copyright (c) 1985, 1986, 1992, 1993
34  *	The Regents of the University of California.  All rights reserved.
35  *
36  * This code is derived from software contributed to Berkeley by
37  * Diomidis Spinellis and James A. Woods, derived from original
38  * work by Spencer Thomas and Joseph Orost.
39  *
40  * Redistribution and use in source and binary forms, with or without
41  * modification, are permitted provided that the following conditions
42  * are met:
43  * 1. Redistributions of source code must retain the above copyright
44  *    notice, this list of conditions and the following disclaimer.
45  * 2. Redistributions in binary form must reproduce the above copyright
46  *    notice, this list of conditions and the following disclaimer in the
47  *    documentation and/or other materials provided with the distribution.
48  * 4. Neither the name of the University nor the names of its contributors
49  *    may be used to endorse or promote products derived from this software
50  *    without specific prior written permission.
51  *
52  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62  * SUCH DAMAGE.
63  */
64 
65 
66 #include "archive_platform.h"
67 __FBSDID("$FreeBSD$");
68 
69 #ifdef HAVE_ERRNO_H
70 #include <errno.h>
71 #endif
72 #ifdef HAVE_STDLIB_H
73 #include <stdlib.h>
74 #endif
75 #ifdef HAVE_STRING_H
76 #include <string.h>
77 #endif
78 #ifdef HAVE_UNISTD_H
79 #include <unistd.h>
80 #endif
81 
82 #include "archive.h"
83 #include "archive_private.h"
84 #include "archive_read_private.h"
85 
86 /*
87  * Because LZW decompression is pretty simple, I've just implemented
88  * the whole decompressor here (cribbing from "compress" source code,
89  * of course), rather than relying on an external library.  I have
90  * made an effort to clarify and simplify the algorithm, so the
91  * names and structure here don't exactly match those used by compress.
92  */
93 
94 struct private_data {
95 	/* Input variables. */
96 	const unsigned char	*next_in;
97 	size_t			 avail_in;
98 	size_t			 consume_unnotified;
99 	int			 bit_buffer;
100 	int			 bits_avail;
101 	size_t			 bytes_in_section;
102 
103 	/* Output variables. */
104 	size_t			 out_block_size;
105 	void			*out_block;
106 
107 	/* Decompression status variables. */
108 	int			 use_reset_code;
109 	int			 end_of_stream;	/* EOF status. */
110 	int			 maxcode;	/* Largest code. */
111 	int			 maxcode_bits;	/* Length of largest code. */
112 	int			 section_end_code; /* When to increase bits. */
113 	int			 bits;		/* Current code length. */
114 	int			 oldcode;	/* Previous code. */
115 	int			 finbyte;	/* Last byte of prev code. */
116 
117 	/* Dictionary. */
118 	int			 free_ent;       /* Next dictionary entry. */
119 	unsigned char		 suffix[65536];
120 	uint16_t		 prefix[65536];
121 
122 	/*
123 	 * Scratch area for expanding dictionary entries.  Note:
124 	 * "worst" case here comes from compressing /dev/zero: the
125 	 * last code in the dictionary will code a sequence of
126 	 * 65536-256 zero bytes.  Thus, we need stack space to expand
127 	 * a 65280-byte dictionary entry.  (Of course, 32640:1
128 	 * compression could also be considered the "best" case. ;-)
129 	 */
130 	unsigned char		*stackp;
131 	unsigned char		 stack[65300];
132 };
133 
134 static int	compress_bidder_bid(struct archive_read_filter_bidder *, struct archive_read_filter *);
135 static int	compress_bidder_init(struct archive_read_filter *);
136 static int	compress_bidder_free(struct archive_read_filter_bidder *);
137 
138 static ssize_t	compress_filter_read(struct archive_read_filter *, const void **);
139 static int	compress_filter_close(struct archive_read_filter *);
140 
141 static int	getbits(struct archive_read_filter *, int n);
142 static int	next_code(struct archive_read_filter *);
143 
144 #if ARCHIVE_VERSION_NUMBER < 4000000
145 /* Deprecated; remove in libarchive 4.0 */
146 int
147 archive_read_support_compression_compress(struct archive *a)
148 {
149 	return archive_read_support_filter_compress(a);
150 }
151 #endif
152 
153 int
154 archive_read_support_filter_compress(struct archive *_a)
155 {
156 	struct archive_read *a = (struct archive_read *)_a;
157 	struct archive_read_filter_bidder *bidder;
158 
159 	archive_check_magic(_a, ARCHIVE_READ_MAGIC,
160 	    ARCHIVE_STATE_NEW, "archive_read_support_filter_compress");
161 
162 	if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK)
163 		return (ARCHIVE_FATAL);
164 
165 	bidder->data = NULL;
166 	bidder->name = "compress (.Z)";
167 	bidder->bid = compress_bidder_bid;
168 	bidder->init = compress_bidder_init;
169 	bidder->options = NULL;
170 	bidder->free = compress_bidder_free;
171 	return (ARCHIVE_OK);
172 }
173 
174 /*
175  * Test whether we can handle this data.
176  * This logic returns zero if any part of the signature fails.
177  */
178 static int
179 compress_bidder_bid(struct archive_read_filter_bidder *self,
180     struct archive_read_filter *filter)
181 {
182 	const unsigned char *buffer;
183 	ssize_t avail;
184 	int bits_checked;
185 
186 	(void)self; /* UNUSED */
187 
188 	/* Shortest valid compress file is 3 bytes. */
189 	buffer = __archive_read_filter_ahead(filter, 3, &avail);
190 
191 	if (buffer == NULL)
192 		return (0);
193 
194 	bits_checked = 0;
195 	/* First two bytes are the magic value */
196 	if (buffer[0] != 0x1F || buffer[1] != 0x9D)
197 		return (0);
198 	/* Third byte holds compression parameters. */
199 	if (buffer[2] & 0x20) /* Reserved bit, must be zero. */
200 		return (0);
201 	if (buffer[2] & 0x40) /* Reserved bit, must be zero. */
202 		return (0);
203 	bits_checked += 18;
204 
205 	return (bits_checked);
206 }
207 
208 /*
209  * Setup the callbacks.
210  */
211 static int
212 compress_bidder_init(struct archive_read_filter *self)
213 {
214 	struct private_data *state;
215 	static const size_t out_block_size = 64 * 1024;
216 	void *out_block;
217 	int code;
218 
219 	self->code = ARCHIVE_FILTER_COMPRESS;
220 	self->name = "compress (.Z)";
221 
222 	state = (struct private_data *)calloc(sizeof(*state), 1);
223 	out_block = malloc(out_block_size);
224 	if (state == NULL || out_block == NULL) {
225 		free(out_block);
226 		free(state);
227 		archive_set_error(&self->archive->archive, ENOMEM,
228 		    "Can't allocate data for %s decompression",
229 		    self->name);
230 		return (ARCHIVE_FATAL);
231 	}
232 
233 	self->data = state;
234 	state->out_block_size = out_block_size;
235 	state->out_block = out_block;
236 	self->read = compress_filter_read;
237 	self->skip = NULL; /* not supported */
238 	self->close = compress_filter_close;
239 
240 	/* XXX MOVE THE FOLLOWING OUT OF INIT() XXX */
241 
242 	(void)getbits(self, 8); /* Skip first signature byte. */
243 	(void)getbits(self, 8); /* Skip second signature byte. */
244 
245 	/* Get compression parameters. */
246 	code = getbits(self, 8);
247 	if ((code & 0x1f) > 16) {
248 		archive_set_error(&self->archive->archive, -1,
249 		    "Invalid compressed data");
250 		return (ARCHIVE_FATAL);
251 	}
252 	state->maxcode_bits = code & 0x1f;
253 	state->maxcode = (1 << state->maxcode_bits);
254 	state->use_reset_code = code & 0x80;
255 
256 	/* Initialize decompressor. */
257 	state->free_ent = 256;
258 	state->stackp = state->stack;
259 	if (state->use_reset_code)
260 		state->free_ent++;
261 	state->bits = 9;
262 	state->section_end_code = (1<<state->bits) - 1;
263 	state->oldcode = -1;
264 	for (code = 255; code >= 0; code--) {
265 		state->prefix[code] = 0;
266 		state->suffix[code] = code;
267 	}
268 	next_code(self);
269 
270 	return (ARCHIVE_OK);
271 }
272 
273 /*
274  * Return a block of data from the decompression buffer.  Decompress more
275  * as necessary.
276  */
277 static ssize_t
278 compress_filter_read(struct archive_read_filter *self, const void **pblock)
279 {
280 	struct private_data *state;
281 	unsigned char *p, *start, *end;
282 	int ret;
283 
284 	state = (struct private_data *)self->data;
285 	if (state->end_of_stream) {
286 		*pblock = NULL;
287 		return (0);
288 	}
289 	p = start = (unsigned char *)state->out_block;
290 	end = start + state->out_block_size;
291 
292 	while (p < end && !state->end_of_stream) {
293 		if (state->stackp > state->stack) {
294 			*p++ = *--state->stackp;
295 		} else {
296 			ret = next_code(self);
297 			if (ret == -1)
298 				state->end_of_stream = ret;
299 			else if (ret != ARCHIVE_OK)
300 				return (ret);
301 		}
302 	}
303 
304 	*pblock = start;
305 	return (p - start);
306 }
307 
308 /*
309  * Clean up the reader.
310  */
311 static int
312 compress_bidder_free(struct archive_read_filter_bidder *self)
313 {
314 	self->data = NULL;
315 	return (ARCHIVE_OK);
316 }
317 
318 /*
319  * Close and release the filter.
320  */
321 static int
322 compress_filter_close(struct archive_read_filter *self)
323 {
324 	struct private_data *state = (struct private_data *)self->data;
325 
326 	free(state->out_block);
327 	free(state);
328 	return (ARCHIVE_OK);
329 }
330 
331 /*
332  * Process the next code and fill the stack with the expansion
333  * of the code.  Returns ARCHIVE_FATAL if there is a fatal I/O or
334  * format error, ARCHIVE_EOF if we hit end of data, ARCHIVE_OK otherwise.
335  */
336 static int
337 next_code(struct archive_read_filter *self)
338 {
339 	struct private_data *state = (struct private_data *)self->data;
340 	int code, newcode;
341 
342 	static int debug_buff[1024];
343 	static unsigned debug_index;
344 
345 	code = newcode = getbits(self, state->bits);
346 	if (code < 0)
347 		return (code);
348 
349 	debug_buff[debug_index++] = code;
350 	if (debug_index >= sizeof(debug_buff)/sizeof(debug_buff[0]))
351 		debug_index = 0;
352 
353 	/* If it's a reset code, reset the dictionary. */
354 	if ((code == 256) && state->use_reset_code) {
355 		/*
356 		 * The original 'compress' implementation blocked its
357 		 * I/O in a manner that resulted in junk bytes being
358 		 * inserted after every reset.  The next section skips
359 		 * this junk.  (Yes, the number of *bytes* to skip is
360 		 * a function of the current *bit* length.)
361 		 */
362 		int skip_bytes =  state->bits -
363 		    (state->bytes_in_section % state->bits);
364 		skip_bytes %= state->bits;
365 		state->bits_avail = 0; /* Discard rest of this byte. */
366 		while (skip_bytes-- > 0) {
367 			code = getbits(self, 8);
368 			if (code < 0)
369 				return (code);
370 		}
371 		/* Now, actually do the reset. */
372 		state->bytes_in_section = 0;
373 		state->bits = 9;
374 		state->section_end_code = (1 << state->bits) - 1;
375 		state->free_ent = 257;
376 		state->oldcode = -1;
377 		return (next_code(self));
378 	}
379 
380 	if (code > state->free_ent
381 	    || (code == state->free_ent && state->oldcode < 0)) {
382 		/* An invalid code is a fatal error. */
383 		archive_set_error(&(self->archive->archive), -1,
384 		    "Invalid compressed data");
385 		return (ARCHIVE_FATAL);
386 	}
387 
388 	/* Special case for KwKwK string. */
389 	if (code >= state->free_ent) {
390 		*state->stackp++ = state->finbyte;
391 		code = state->oldcode;
392 	}
393 
394 	/* Generate output characters in reverse order. */
395 	while (code >= 256) {
396 		*state->stackp++ = state->suffix[code];
397 		code = state->prefix[code];
398 	}
399 	*state->stackp++ = state->finbyte = code;
400 
401 	/* Generate the new entry. */
402 	code = state->free_ent;
403 	if (code < state->maxcode && state->oldcode >= 0) {
404 		state->prefix[code] = state->oldcode;
405 		state->suffix[code] = state->finbyte;
406 		++state->free_ent;
407 	}
408 	if (state->free_ent > state->section_end_code) {
409 		state->bits++;
410 		state->bytes_in_section = 0;
411 		if (state->bits == state->maxcode_bits)
412 			state->section_end_code = state->maxcode;
413 		else
414 			state->section_end_code = (1 << state->bits) - 1;
415 	}
416 
417 	/* Remember previous code. */
418 	state->oldcode = newcode;
419 	return (ARCHIVE_OK);
420 }
421 
422 /*
423  * Return next 'n' bits from stream.
424  *
425  * -1 indicates end of available data.
426  */
427 static int
428 getbits(struct archive_read_filter *self, int n)
429 {
430 	struct private_data *state = (struct private_data *)self->data;
431 	int code;
432 	ssize_t ret;
433 	static const int mask[] = {
434 		0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff,
435 		0x1ff, 0x3ff, 0x7ff, 0xfff, 0x1fff, 0x3fff, 0x7fff, 0xffff
436 	};
437 
438 	while (state->bits_avail < n) {
439 		if (state->avail_in <= 0) {
440 			if (state->consume_unnotified) {
441 				__archive_read_filter_consume(self->upstream,
442 					state->consume_unnotified);
443 				state->consume_unnotified = 0;
444 			}
445 			state->next_in
446 			    = __archive_read_filter_ahead(self->upstream,
447 				1, &ret);
448 			if (ret == 0)
449 				return (-1);
450 			if (ret < 0 || state->next_in == NULL)
451 				return (ARCHIVE_FATAL);
452 			state->consume_unnotified = state->avail_in = ret;
453 		}
454 		state->bit_buffer |= *state->next_in++ << state->bits_avail;
455 		state->avail_in--;
456 		state->bits_avail += 8;
457 		state->bytes_in_section++;
458 	}
459 
460 	code = state->bit_buffer;
461 	state->bit_buffer >>= n;
462 	state->bits_avail -= n;
463 
464 	return (code & mask[n]);
465 }
466