xref: /freebsd/usr.bin/gzip/zuncompress.c (revision 38f0b757fd84d17d0fc24739a7cda160c4516d81)
1 /*	$NetBSD: zuncompress.c,v 1.11 2011/08/16 13:55:02 joerg Exp $ */
2 
3 /*-
4  * Copyright (c) 1985, 1986, 1992, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Diomidis Spinellis and James A. Woods, derived from original
9  * work by Spencer Thomas and Joseph Orost.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  *
35  * from: NetBSD: zopen.c,v 1.8 2003/08/07 11:13:29 agc Exp
36  * $FreeBSD$
37  */
38 
39 /* This file is #included by gzip.c */
40 
41 static int	zread(void *, char *, int);
42 
43 #define	tab_prefixof(i)	(zs->zs_codetab[i])
44 #define	tab_suffixof(i)	((char_type *)(zs->zs_htab))[i]
45 #define	de_stack	((char_type *)&tab_suffixof(1 << BITS))
46 
47 #define BITS		16		/* Default bits. */
48 #define HSIZE		69001		/* 95% occupancy */ /* XXX may not need HSIZE */
49 #define BIT_MASK	0x1f		/* Defines for third byte of header. */
50 #define BLOCK_MASK	0x80
51 #define CHECK_GAP	10000		/* Ratio check interval. */
52 #define BUFSIZE		(64 * 1024)
53 
54 /*
55  * Masks 0x40 and 0x20 are free.  I think 0x20 should mean that there is
56  * a fourth header byte (for expansion).
57  */
58 #define INIT_BITS	9	/* Initial number of bits/code. */
59 
60 /*
61  * the next two codes should not be changed lightly, as they must not
62  * lie within the contiguous general code space.
63  */
64 #define	FIRST	257		/* First free entry. */
65 #define	CLEAR	256		/* Table clear output code. */
66 
67 
68 #define MAXCODE(n_bits)	((1 << (n_bits)) - 1)
69 
70 typedef long	code_int;
71 typedef long	count_int;
72 typedef u_char	char_type;
73 
74 static char_type magic_header[] =
75 	{'\037', '\235'};	/* 1F 9D */
76 
77 static char_type rmask[9] =
78 	{0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
79 
80 static off_t total_compressed_bytes;
81 static size_t compressed_prelen;
82 static char *compressed_pre;
83 
84 struct s_zstate {
85 	FILE *zs_fp;			/* File stream for I/O */
86 	char zs_mode;			/* r or w */
87 	enum {
88 		S_START, S_MIDDLE, S_EOF
89 	} zs_state;			/* State of computation */
90 	int zs_n_bits;			/* Number of bits/code. */
91 	int zs_maxbits;			/* User settable max # bits/code. */
92 	code_int zs_maxcode;		/* Maximum code, given n_bits. */
93 	code_int zs_maxmaxcode;		/* Should NEVER generate this code. */
94 	count_int zs_htab [HSIZE];
95 	u_short zs_codetab [HSIZE];
96 	code_int zs_hsize;		/* For dynamic table sizing. */
97 	code_int zs_free_ent;		/* First unused entry. */
98 	/*
99 	 * Block compression parameters -- after all codes are used up,
100 	 * and compression rate changes, start over.
101 	 */
102 	int zs_block_compress;
103 	int zs_clear_flg;
104 	long zs_ratio;
105 	count_int zs_checkpoint;
106 	int zs_offset;
107 	long zs_in_count;		/* Length of input. */
108 	long zs_bytes_out;		/* Length of compressed output. */
109 	long zs_out_count;		/* # of codes output (for debugging). */
110 	char_type zs_buf[BITS];
111 	union {
112 		struct {
113 			long zs_fcode;
114 			code_int zs_ent;
115 			code_int zs_hsize_reg;
116 			int zs_hshift;
117 		} w;			/* Write parameters */
118 		struct {
119 			char_type *zs_stackp;
120 			int zs_finchar;
121 			code_int zs_code, zs_oldcode, zs_incode;
122 			int zs_roffset, zs_size;
123 			char_type zs_gbuf[BITS];
124 		} r;			/* Read parameters */
125 	} u;
126 };
127 
128 static code_int	getcode(struct s_zstate *zs);
129 
130 static off_t
131 zuncompress(FILE *in, FILE *out, char *pre, size_t prelen,
132 	    off_t *compressed_bytes)
133 {
134 	off_t bin, bout = 0;
135 	char *buf;
136 
137 	buf = malloc(BUFSIZE);
138 	if (buf == NULL)
139 		return -1;
140 
141 	/* XXX */
142 	compressed_prelen = prelen;
143 	if (prelen != 0)
144 		compressed_pre = pre;
145 	else
146 		compressed_pre = NULL;
147 
148 	while ((bin = fread(buf, 1, sizeof(buf), in)) != 0) {
149 		if (tflag == 0 && (off_t)fwrite(buf, 1, bin, out) != bin) {
150 			free(buf);
151 			return -1;
152 		}
153 		bout += bin;
154 	}
155 
156 	if (compressed_bytes)
157 		*compressed_bytes = total_compressed_bytes;
158 
159 	free(buf);
160 	return bout;
161 }
162 
163 static int
164 zclose(void *zs)
165 {
166 	free(zs);
167 	/* We leave the caller to close the fd passed to zdopen() */
168 	return 0;
169 }
170 
171 FILE *
172 zdopen(int fd)
173 {
174 	struct s_zstate *zs;
175 
176 	if ((zs = calloc(1, sizeof(struct s_zstate))) == NULL)
177 		return (NULL);
178 
179 	zs->zs_state = S_START;
180 
181 	/* XXX we can get rid of some of these */
182 	zs->zs_hsize = HSIZE;			/* For dynamic table sizing. */
183 	zs->zs_free_ent = 0;			/* First unused entry. */
184 	zs->zs_block_compress = BLOCK_MASK;
185 	zs->zs_clear_flg = 0;			/* XXX we calloc()'d this structure why = 0? */
186 	zs->zs_ratio = 0;
187 	zs->zs_checkpoint = CHECK_GAP;
188 	zs->zs_in_count = 1;			/* Length of input. */
189 	zs->zs_out_count = 0;			/* # of codes output (for debugging). */
190 	zs->u.r.zs_roffset = 0;
191 	zs->u.r.zs_size = 0;
192 
193 	/*
194 	 * Layering compress on top of stdio in order to provide buffering,
195 	 * and ensure that reads and write work with the data specified.
196 	 */
197 	if ((zs->zs_fp = fdopen(fd, "r")) == NULL) {
198 		free(zs);
199 		return NULL;
200 	}
201 
202 	return funopen(zs, zread, NULL, NULL, zclose);
203 }
204 
205 /*
206  * Decompress read.  This routine adapts to the codes in the file building
207  * the "string" table on-the-fly; requiring no table to be stored in the
208  * compressed file.  The tables used herein are shared with those of the
209  * compress() routine.  See the definitions above.
210  */
211 static int
212 zread(void *cookie, char *rbp, int num)
213 {
214 	u_int count, i;
215 	struct s_zstate *zs;
216 	u_char *bp, header[3];
217 
218 	if (num == 0)
219 		return (0);
220 
221 	zs = cookie;
222 	count = num;
223 	bp = (u_char *)rbp;
224 	switch (zs->zs_state) {
225 	case S_START:
226 		zs->zs_state = S_MIDDLE;
227 		break;
228 	case S_MIDDLE:
229 		goto middle;
230 	case S_EOF:
231 		goto eof;
232 	}
233 
234 	/* Check the magic number */
235 	for (i = 0; i < 3 && compressed_prelen; i++, compressed_prelen--)
236 		header[i] = *compressed_pre++;
237 
238 	if (fread(header + i, 1, sizeof(header) - i, zs->zs_fp) !=
239 		  sizeof(header) - i ||
240 	    memcmp(header, magic_header, sizeof(magic_header)) != 0) {
241 		errno = EFTYPE;
242 		return (-1);
243 	}
244 	total_compressed_bytes = 0;
245 	zs->zs_maxbits = header[2];	/* Set -b from file. */
246 	zs->zs_block_compress = zs->zs_maxbits & BLOCK_MASK;
247 	zs->zs_maxbits &= BIT_MASK;
248 	zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
249 	if (zs->zs_maxbits > BITS || zs->zs_maxbits < 12) {
250 		errno = EFTYPE;
251 		return (-1);
252 	}
253 	/* As above, initialize the first 256 entries in the table. */
254 	zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
255 	for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0; zs->u.r.zs_code--) {
256 		tab_prefixof(zs->u.r.zs_code) = 0;
257 		tab_suffixof(zs->u.r.zs_code) = (char_type) zs->u.r.zs_code;
258 	}
259 	zs->zs_free_ent = zs->zs_block_compress ? FIRST : 256;
260 
261 	zs->u.r.zs_oldcode = -1;
262 	zs->u.r.zs_stackp = de_stack;
263 
264 	while ((zs->u.r.zs_code = getcode(zs)) > -1) {
265 
266 		if ((zs->u.r.zs_code == CLEAR) && zs->zs_block_compress) {
267 			for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0;
268 			    zs->u.r.zs_code--)
269 				tab_prefixof(zs->u.r.zs_code) = 0;
270 			zs->zs_clear_flg = 1;
271 			zs->zs_free_ent = FIRST;
272 			zs->u.r.zs_oldcode = -1;
273 			continue;
274 		}
275 		zs->u.r.zs_incode = zs->u.r.zs_code;
276 
277 		/* Special case for KwKwK string. */
278 		if (zs->u.r.zs_code >= zs->zs_free_ent) {
279 			if (zs->u.r.zs_code > zs->zs_free_ent ||
280 			    zs->u.r.zs_oldcode == -1) {
281 				/* Bad stream. */
282 				errno = EINVAL;
283 				return (-1);
284 			}
285 			*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar;
286 			zs->u.r.zs_code = zs->u.r.zs_oldcode;
287 		}
288 		/*
289 		 * The above condition ensures that code < free_ent.
290 		 * The construction of tab_prefixof in turn guarantees that
291 		 * each iteration decreases code and therefore stack usage is
292 		 * bound by 1 << BITS - 256.
293 		 */
294 
295 		/* Generate output characters in reverse order. */
296 		while (zs->u.r.zs_code >= 256) {
297 			*zs->u.r.zs_stackp++ = tab_suffixof(zs->u.r.zs_code);
298 			zs->u.r.zs_code = tab_prefixof(zs->u.r.zs_code);
299 		}
300 		*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar = tab_suffixof(zs->u.r.zs_code);
301 
302 		/* And put them out in forward order.  */
303 middle:		do {
304 			if (count-- == 0)
305 				return (num);
306 			*bp++ = *--zs->u.r.zs_stackp;
307 		} while (zs->u.r.zs_stackp > de_stack);
308 
309 		/* Generate the new entry. */
310 		if ((zs->u.r.zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode &&
311 		    zs->u.r.zs_oldcode != -1) {
312 			tab_prefixof(zs->u.r.zs_code) = (u_short) zs->u.r.zs_oldcode;
313 			tab_suffixof(zs->u.r.zs_code) = zs->u.r.zs_finchar;
314 			zs->zs_free_ent = zs->u.r.zs_code + 1;
315 		}
316 
317 		/* Remember previous code. */
318 		zs->u.r.zs_oldcode = zs->u.r.zs_incode;
319 	}
320 	zs->zs_state = S_EOF;
321 eof:	return (num - count);
322 }
323 
324 /*-
325  * Read one code from the standard input.  If EOF, return -1.
326  * Inputs:
327  * 	stdin
328  * Outputs:
329  * 	code or -1 is returned.
330  */
331 static code_int
332 getcode(struct s_zstate *zs)
333 {
334 	code_int gcode;
335 	int r_off, bits, i;
336 	char_type *bp;
337 
338 	bp = zs->u.r.zs_gbuf;
339 	if (zs->zs_clear_flg > 0 || zs->u.r.zs_roffset >= zs->u.r.zs_size ||
340 	    zs->zs_free_ent > zs->zs_maxcode) {
341 		/*
342 		 * If the next entry will be too big for the current gcode
343 		 * size, then we must increase the size.  This implies reading
344 		 * a new buffer full, too.
345 		 */
346 		if (zs->zs_free_ent > zs->zs_maxcode) {
347 			zs->zs_n_bits++;
348 			if (zs->zs_n_bits == zs->zs_maxbits)	/* Won't get any bigger now. */
349 				zs->zs_maxcode = zs->zs_maxmaxcode;
350 			else
351 				zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
352 		}
353 		if (zs->zs_clear_flg > 0) {
354 			zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
355 			zs->zs_clear_flg = 0;
356 		}
357 		/* XXX */
358 		for (i = 0; i < zs->zs_n_bits && compressed_prelen; i++, compressed_prelen--)
359 			zs->u.r.zs_gbuf[i] = *compressed_pre++;
360 		zs->u.r.zs_size = fread(zs->u.r.zs_gbuf + i, 1, zs->zs_n_bits - i, zs->zs_fp);
361 		zs->u.r.zs_size += i;
362 		if (zs->u.r.zs_size <= 0)			/* End of file. */
363 			return (-1);
364 		zs->u.r.zs_roffset = 0;
365 
366 		total_compressed_bytes += zs->u.r.zs_size;
367 
368 		/* Round size down to integral number of codes. */
369 		zs->u.r.zs_size = (zs->u.r.zs_size << 3) - (zs->zs_n_bits - 1);
370 	}
371 	r_off = zs->u.r.zs_roffset;
372 	bits = zs->zs_n_bits;
373 
374 	/* Get to the first byte. */
375 	bp += (r_off >> 3);
376 	r_off &= 7;
377 
378 	/* Get first part (low order bits). */
379 	gcode = (*bp++ >> r_off);
380 	bits -= (8 - r_off);
381 	r_off = 8 - r_off;	/* Now, roffset into gcode word. */
382 
383 	/* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
384 	if (bits >= 8) {
385 		gcode |= *bp++ << r_off;
386 		r_off += 8;
387 		bits -= 8;
388 	}
389 
390 	/* High order bits. */
391 	gcode |= (*bp & rmask[bits]) << r_off;
392 	zs->u.r.zs_roffset += zs->zs_n_bits;
393 
394 	return (gcode);
395 }
396 
397