xref: /freebsd/usr.bin/gzip/zuncompress.c (revision 10ff414c)
1 /*	$NetBSD: zuncompress.c,v 1.11 2011/08/16 13:55:02 joerg Exp $ */
2 
3 /*-
4  * SPDX-License-Identifier: BSD-3-Clause
5  *
6  * Copyright (c) 1985, 1986, 1992, 1993
7  *	The Regents of the University of California.  All rights reserved.
8  *
9  * This code is derived from software contributed to Berkeley by
10  * Diomidis Spinellis and James A. Woods, derived from original
11  * work by Spencer Thomas and Joseph Orost.
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
16  * 1. Redistributions of source code must retain the above copyright
17  *    notice, this list of conditions and the following disclaimer.
18  * 2. Redistributions in binary form must reproduce the above copyright
19  *    notice, this list of conditions and the following disclaimer in the
20  *    documentation and/or other materials provided with the distribution.
21  * 3. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  *
37  * from: NetBSD: zopen.c,v 1.8 2003/08/07 11:13:29 agc Exp
38  * $FreeBSD$
39  */
40 
41 /* This file is #included by gzip.c */
42 
43 static int	zread(void *, char *, int);
44 
45 #define	tab_prefixof(i)	(zs->zs_codetab[i])
46 #define	tab_suffixof(i)	((char_type *)(zs->zs_htab))[i]
47 #define	de_stack	((char_type *)&tab_suffixof(1 << BITS))
48 
49 #define BITS		16		/* Default bits. */
50 #define HSIZE		69001		/* 95% occupancy */ /* XXX may not need HSIZE */
51 #define BIT_MASK	0x1f		/* Defines for third byte of header. */
52 #define BLOCK_MASK	0x80
53 #define CHECK_GAP	10000		/* Ratio check interval. */
54 #define BUFSIZE		(64 * 1024)
55 
56 /*
57  * Masks 0x40 and 0x20 are free.  I think 0x20 should mean that there is
58  * a fourth header byte (for expansion).
59  */
60 #define INIT_BITS	9	/* Initial number of bits/code. */
61 
62 /*
63  * the next two codes should not be changed lightly, as they must not
64  * lie within the contiguous general code space.
65  */
66 #define	FIRST	257		/* First free entry. */
67 #define	CLEAR	256		/* Table clear output code. */
68 
69 
70 #define MAXCODE(n_bits)	((1 << (n_bits)) - 1)
71 
72 typedef long	code_int;
73 typedef long	count_int;
74 typedef u_char	char_type;
75 
76 static char_type magic_header[] =
77 	{'\037', '\235'};	/* 1F 9D */
78 
79 static char_type rmask[9] =
80 	{0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
81 
82 static off_t total_compressed_bytes;
83 static size_t compressed_prelen;
84 static char *compressed_pre;
85 
86 struct s_zstate {
87 	FILE *zs_fp;			/* File stream for I/O */
88 	char zs_mode;			/* r or w */
89 	enum {
90 		S_START, S_MIDDLE, S_EOF
91 	} zs_state;			/* State of computation */
92 	int zs_n_bits;			/* Number of bits/code. */
93 	int zs_maxbits;			/* User settable max # bits/code. */
94 	code_int zs_maxcode;		/* Maximum code, given n_bits. */
95 	code_int zs_maxmaxcode;		/* Should NEVER generate this code. */
96 	count_int zs_htab [HSIZE];
97 	u_short zs_codetab [HSIZE];
98 	code_int zs_hsize;		/* For dynamic table sizing. */
99 	code_int zs_free_ent;		/* First unused entry. */
100 	/*
101 	 * Block compression parameters -- after all codes are used up,
102 	 * and compression rate changes, start over.
103 	 */
104 	int zs_block_compress;
105 	int zs_clear_flg;
106 	long zs_ratio;
107 	count_int zs_checkpoint;
108 	int zs_offset;
109 	long zs_in_count;		/* Length of input. */
110 	long zs_bytes_out;		/* Length of compressed output. */
111 	long zs_out_count;		/* # of codes output (for debugging). */
112 	char_type zs_buf[BITS];
113 	union {
114 		struct {
115 			long zs_fcode;
116 			code_int zs_ent;
117 			code_int zs_hsize_reg;
118 			int zs_hshift;
119 		} w;			/* Write parameters */
120 		struct {
121 			char_type *zs_stackp;
122 			int zs_finchar;
123 			code_int zs_code, zs_oldcode, zs_incode;
124 			int zs_roffset, zs_size;
125 			char_type zs_gbuf[BITS];
126 		} r;			/* Read parameters */
127 	} u;
128 };
129 
130 static code_int	getcode(struct s_zstate *zs);
131 
132 static off_t
133 zuncompress(FILE *in, FILE *out, char *pre, size_t prelen,
134 	    off_t *compressed_bytes)
135 {
136 	off_t bin, bout = 0;
137 	char *buf;
138 
139 	buf = malloc(BUFSIZE);
140 	if (buf == NULL)
141 		return -1;
142 
143 	/* XXX */
144 	compressed_prelen = prelen;
145 	if (prelen != 0)
146 		compressed_pre = pre;
147 	else
148 		compressed_pre = NULL;
149 
150 	while ((bin = fread(buf, 1, BUFSIZE, in)) != 0) {
151 		if (tflag == 0 && (off_t)fwrite(buf, 1, bin, out) != bin) {
152 			free(buf);
153 			return -1;
154 		}
155 		bout += bin;
156 	}
157 
158 	if (compressed_bytes)
159 		*compressed_bytes = total_compressed_bytes;
160 
161 	free(buf);
162 	return bout;
163 }
164 
165 static int
166 zclose(void *zs)
167 {
168 	free(zs);
169 	/* We leave the caller to close the fd passed to zdopen() */
170 	return 0;
171 }
172 
173 FILE *
174 zdopen(int fd)
175 {
176 	struct s_zstate *zs;
177 
178 	if ((zs = calloc(1, sizeof(struct s_zstate))) == NULL)
179 		return (NULL);
180 
181 	zs->zs_state = S_START;
182 
183 	/* XXX we can get rid of some of these */
184 	zs->zs_hsize = HSIZE;			/* For dynamic table sizing. */
185 	zs->zs_free_ent = 0;			/* First unused entry. */
186 	zs->zs_block_compress = BLOCK_MASK;
187 	zs->zs_clear_flg = 0;			/* XXX we calloc()'d this structure why = 0? */
188 	zs->zs_ratio = 0;
189 	zs->zs_checkpoint = CHECK_GAP;
190 	zs->zs_in_count = 1;			/* Length of input. */
191 	zs->zs_out_count = 0;			/* # of codes output (for debugging). */
192 	zs->u.r.zs_roffset = 0;
193 	zs->u.r.zs_size = 0;
194 
195 	/*
196 	 * Layering compress on top of stdio in order to provide buffering,
197 	 * and ensure that reads and write work with the data specified.
198 	 */
199 	if ((zs->zs_fp = fdopen(fd, "r")) == NULL) {
200 		free(zs);
201 		return NULL;
202 	}
203 
204 	return funopen(zs, zread, NULL, NULL, zclose);
205 }
206 
207 /*
208  * Decompress read.  This routine adapts to the codes in the file building
209  * the "string" table on-the-fly; requiring no table to be stored in the
210  * compressed file.  The tables used herein are shared with those of the
211  * compress() routine.  See the definitions above.
212  */
213 static int
214 zread(void *cookie, char *rbp, int num)
215 {
216 	u_int count, i;
217 	struct s_zstate *zs;
218 	u_char *bp, header[3];
219 
220 	if (num == 0)
221 		return (0);
222 
223 	zs = cookie;
224 	count = num;
225 	bp = (u_char *)rbp;
226 	switch (zs->zs_state) {
227 	case S_START:
228 		zs->zs_state = S_MIDDLE;
229 		break;
230 	case S_MIDDLE:
231 		goto middle;
232 	case S_EOF:
233 		goto eof;
234 	}
235 
236 	/* Check the magic number */
237 	for (i = 0; i < 3 && compressed_prelen; i++, compressed_prelen--)
238 		header[i] = *compressed_pre++;
239 
240 	if (fread(header + i, 1, sizeof(header) - i, zs->zs_fp) !=
241 		  sizeof(header) - i ||
242 	    memcmp(header, magic_header, sizeof(magic_header)) != 0) {
243 		errno = EFTYPE;
244 		return (-1);
245 	}
246 	total_compressed_bytes = 0;
247 	zs->zs_maxbits = header[2];	/* Set -b from file. */
248 	zs->zs_block_compress = zs->zs_maxbits & BLOCK_MASK;
249 	zs->zs_maxbits &= BIT_MASK;
250 	zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
251 	if (zs->zs_maxbits > BITS || zs->zs_maxbits < 12) {
252 		errno = EFTYPE;
253 		return (-1);
254 	}
255 	/* As above, initialize the first 256 entries in the table. */
256 	zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
257 	for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0; zs->u.r.zs_code--) {
258 		tab_prefixof(zs->u.r.zs_code) = 0;
259 		tab_suffixof(zs->u.r.zs_code) = (char_type) zs->u.r.zs_code;
260 	}
261 	zs->zs_free_ent = zs->zs_block_compress ? FIRST : 256;
262 
263 	zs->u.r.zs_oldcode = -1;
264 	zs->u.r.zs_stackp = de_stack;
265 
266 	while ((zs->u.r.zs_code = getcode(zs)) > -1) {
267 
268 		if ((zs->u.r.zs_code == CLEAR) && zs->zs_block_compress) {
269 			for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0;
270 			    zs->u.r.zs_code--)
271 				tab_prefixof(zs->u.r.zs_code) = 0;
272 			zs->zs_clear_flg = 1;
273 			zs->zs_free_ent = FIRST;
274 			zs->u.r.zs_oldcode = -1;
275 			continue;
276 		}
277 		zs->u.r.zs_incode = zs->u.r.zs_code;
278 
279 		/* Special case for KwKwK string. */
280 		if (zs->u.r.zs_code >= zs->zs_free_ent) {
281 			if (zs->u.r.zs_code > zs->zs_free_ent ||
282 			    zs->u.r.zs_oldcode == -1) {
283 				/* Bad stream. */
284 				errno = EFTYPE;
285 				return (-1);
286 			}
287 			*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar;
288 			zs->u.r.zs_code = zs->u.r.zs_oldcode;
289 		}
290 		/*
291 		 * The above condition ensures that code < free_ent.
292 		 * The construction of tab_prefixof in turn guarantees that
293 		 * each iteration decreases code and therefore stack usage is
294 		 * bound by 1 << BITS - 256.
295 		 */
296 
297 		/* Generate output characters in reverse order. */
298 		while (zs->u.r.zs_code >= 256) {
299 			*zs->u.r.zs_stackp++ = tab_suffixof(zs->u.r.zs_code);
300 			zs->u.r.zs_code = tab_prefixof(zs->u.r.zs_code);
301 		}
302 		*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar = tab_suffixof(zs->u.r.zs_code);
303 
304 		/* And put them out in forward order.  */
305 middle:		do {
306 			if (count-- == 0)
307 				return (num);
308 			*bp++ = *--zs->u.r.zs_stackp;
309 		} while (zs->u.r.zs_stackp > de_stack);
310 
311 		/* Generate the new entry. */
312 		if ((zs->u.r.zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode &&
313 		    zs->u.r.zs_oldcode != -1) {
314 			tab_prefixof(zs->u.r.zs_code) = (u_short) zs->u.r.zs_oldcode;
315 			tab_suffixof(zs->u.r.zs_code) = zs->u.r.zs_finchar;
316 			zs->zs_free_ent = zs->u.r.zs_code + 1;
317 		}
318 
319 		/* Remember previous code. */
320 		zs->u.r.zs_oldcode = zs->u.r.zs_incode;
321 	}
322 	zs->zs_state = S_EOF;
323 eof:	return (num - count);
324 }
325 
326 /*-
327  * Read one code from the standard input.  If EOF, return -1.
328  * Inputs:
329  * 	stdin
330  * Outputs:
331  * 	code or -1 is returned.
332  */
333 static code_int
334 getcode(struct s_zstate *zs)
335 {
336 	code_int gcode;
337 	int r_off, bits, i;
338 	char_type *bp;
339 
340 	bp = zs->u.r.zs_gbuf;
341 	if (zs->zs_clear_flg > 0 || zs->u.r.zs_roffset >= zs->u.r.zs_size ||
342 	    zs->zs_free_ent > zs->zs_maxcode) {
343 		/*
344 		 * If the next entry will be too big for the current gcode
345 		 * size, then we must increase the size.  This implies reading
346 		 * a new buffer full, too.
347 		 */
348 		if (zs->zs_free_ent > zs->zs_maxcode) {
349 			zs->zs_n_bits++;
350 			if (zs->zs_n_bits == zs->zs_maxbits)	/* Won't get any bigger now. */
351 				zs->zs_maxcode = zs->zs_maxmaxcode;
352 			else
353 				zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
354 		}
355 		if (zs->zs_clear_flg > 0) {
356 			zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
357 			zs->zs_clear_flg = 0;
358 		}
359 		/* XXX */
360 		for (i = 0; i < zs->zs_n_bits && compressed_prelen; i++, compressed_prelen--)
361 			zs->u.r.zs_gbuf[i] = *compressed_pre++;
362 		zs->u.r.zs_size = fread(zs->u.r.zs_gbuf + i, 1, zs->zs_n_bits - i, zs->zs_fp);
363 		zs->u.r.zs_size += i;
364 		if (zs->u.r.zs_size <= 0)			/* End of file. */
365 			return (-1);
366 		zs->u.r.zs_roffset = 0;
367 
368 		total_compressed_bytes += zs->u.r.zs_size;
369 
370 		/* Round size down to integral number of codes. */
371 		zs->u.r.zs_size = (zs->u.r.zs_size << 3) - (zs->zs_n_bits - 1);
372 	}
373 	r_off = zs->u.r.zs_roffset;
374 	bits = zs->zs_n_bits;
375 
376 	/* Get to the first byte. */
377 	bp += (r_off >> 3);
378 	r_off &= 7;
379 
380 	/* Get first part (low order bits). */
381 	gcode = (*bp++ >> r_off);
382 	bits -= (8 - r_off);
383 	r_off = 8 - r_off;	/* Now, roffset into gcode word. */
384 
385 	/* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
386 	if (bits >= 8) {
387 		gcode |= *bp++ << r_off;
388 		r_off += 8;
389 		bits -= 8;
390 	}
391 
392 	/* High order bits. */
393 	gcode |= (*bp & rmask[bits]) << r_off;
394 	zs->u.r.zs_roffset += zs->zs_n_bits;
395 
396 	return (gcode);
397 }
398 
399