xref: /freebsd/usr.bin/gzip/zuncompress.c (revision 4543ef516683042d46f3bd3bb8a4f3f746e00499)
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  */
39 
40 /* This file is #included by gzip.c */
41 
42 static int	zread(void *, char *, int);
43 
44 #define	tab_prefixof(i)	(zs->zs_codetab[i])
45 #define	tab_suffixof(i)	((char_type *)(zs->zs_htab))[i]
46 #define	de_stack	((char_type *)&tab_suffixof(1 << BITS))
47 
48 #define BITS		16		/* Default bits. */
49 #define HSIZE		69001		/* 95% occupancy */ /* XXX may not need HSIZE */
50 #define BIT_MASK	0x1f		/* Defines for third byte of header. */
51 #define BLOCK_MASK	0x80
52 #define CHECK_GAP	10000		/* Ratio check interval. */
53 #define BUFSIZE		(64 * 1024)
54 
55 /*
56  * Masks 0x40 and 0x20 are free.  I think 0x20 should mean that there is
57  * a fourth header byte (for expansion).
58  */
59 #define INIT_BITS	9	/* Initial number of bits/code. */
60 
61 /*
62  * the next two codes should not be changed lightly, as they must not
63  * lie within the contiguous general code space.
64  */
65 #define	FIRST	257		/* First free entry. */
66 #define	CLEAR	256		/* Table clear output code. */
67 
68 
69 #define MAXCODE(n_bits)	((1 << (n_bits)) - 1)
70 
71 typedef long	code_int;
72 typedef long	count_int;
73 typedef u_char	char_type;
74 
75 static char_type magic_header[] =
76 	{'\037', '\235'};	/* 1F 9D */
77 
78 static char_type rmask[9] =
79 	{0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff};
80 
81 static off_t total_compressed_bytes;
82 static size_t compressed_prelen;
83 static char *compressed_pre;
84 
85 struct s_zstate {
86 	FILE *zs_fp;			/* File stream for I/O */
87 	char zs_mode;			/* r or w */
88 	enum {
89 		S_START, S_MIDDLE, S_EOF
90 	} zs_state;			/* State of computation */
91 	int zs_n_bits;			/* Number of bits/code. */
92 	int zs_maxbits;			/* User settable max # bits/code. */
93 	code_int zs_maxcode;		/* Maximum code, given n_bits. */
94 	code_int zs_maxmaxcode;		/* Should NEVER generate this code. */
95 	count_int zs_htab [HSIZE];
96 	u_short zs_codetab [HSIZE];
97 	code_int zs_hsize;		/* For dynamic table sizing. */
98 	code_int zs_free_ent;		/* First unused entry. */
99 	/*
100 	 * Block compression parameters -- after all codes are used up,
101 	 * and compression rate changes, start over.
102 	 */
103 	int zs_block_compress;
104 	int zs_clear_flg;
105 	long zs_ratio;
106 	count_int zs_checkpoint;
107 	int zs_offset;
108 	long zs_in_count;		/* Length of input. */
109 	long zs_bytes_out;		/* Length of compressed output. */
110 	long zs_out_count;		/* # of codes output (for debugging). */
111 	char_type zs_buf[BITS];
112 	union {
113 		struct {
114 			long zs_fcode;
115 			code_int zs_ent;
116 			code_int zs_hsize_reg;
117 			int zs_hshift;
118 		} w;			/* Write parameters */
119 		struct {
120 			char_type *zs_stackp;
121 			int zs_finchar;
122 			code_int zs_code, zs_oldcode, zs_incode;
123 			int zs_roffset, zs_size;
124 			char_type zs_gbuf[BITS];
125 		} r;			/* Read parameters */
126 	} u;
127 };
128 
129 static code_int	getcode(struct s_zstate *zs);
130 
131 static off_t
132 zuncompress(FILE *in, FILE *out, char *pre, size_t prelen,
133 	    off_t *compressed_bytes)
134 {
135 	off_t bin, bout = 0;
136 	char *buf;
137 
138 	buf = malloc(BUFSIZE);
139 	if (buf == NULL)
140 		return -1;
141 
142 	/* XXX */
143 	compressed_prelen = prelen;
144 	if (prelen != 0)
145 		compressed_pre = pre;
146 	else
147 		compressed_pre = NULL;
148 
149 	while ((bin = fread(buf, 1, BUFSIZE, in)) != 0) {
150 		if (tflag == 0 && (off_t)fwrite(buf, 1, bin, out) != bin) {
151 			free(buf);
152 			return -1;
153 		}
154 		bout += bin;
155 	}
156 
157 	if (compressed_bytes)
158 		*compressed_bytes = total_compressed_bytes;
159 
160 	free(buf);
161 	return bout;
162 }
163 
164 static int
165 zclose(void *zs)
166 {
167 	free(zs);
168 	/* We leave the caller to close the fd passed to zdopen() */
169 	return 0;
170 }
171 
172 FILE *
173 zdopen(int fd)
174 {
175 	struct s_zstate *zs;
176 
177 	if ((zs = calloc(1, sizeof(struct s_zstate))) == NULL)
178 		return (NULL);
179 
180 	zs->zs_state = S_START;
181 
182 	/* XXX we can get rid of some of these */
183 	zs->zs_hsize = HSIZE;			/* For dynamic table sizing. */
184 	zs->zs_free_ent = 0;			/* First unused entry. */
185 	zs->zs_block_compress = BLOCK_MASK;
186 	zs->zs_clear_flg = 0;			/* XXX we calloc()'d this structure why = 0? */
187 	zs->zs_ratio = 0;
188 	zs->zs_checkpoint = CHECK_GAP;
189 	zs->zs_in_count = 1;			/* Length of input. */
190 	zs->zs_out_count = 0;			/* # of codes output (for debugging). */
191 	zs->u.r.zs_roffset = 0;
192 	zs->u.r.zs_size = 0;
193 
194 	/*
195 	 * Layering compress on top of stdio in order to provide buffering,
196 	 * and ensure that reads and write work with the data specified.
197 	 */
198 	if ((zs->zs_fp = fdopen(fd, "r")) == NULL) {
199 		free(zs);
200 		return NULL;
201 	}
202 
203 	return funopen(zs, zread, NULL, NULL, zclose);
204 }
205 
206 /*
207  * Decompress read.  This routine adapts to the codes in the file building
208  * the "string" table on-the-fly; requiring no table to be stored in the
209  * compressed file.  The tables used herein are shared with those of the
210  * compress() routine.  See the definitions above.
211  */
212 static int
213 zread(void *cookie, char *rbp, int num)
214 {
215 	u_int count, i;
216 	struct s_zstate *zs;
217 	u_char *bp, header[3];
218 
219 	if (num == 0)
220 		return (0);
221 
222 	zs = cookie;
223 	count = num;
224 	bp = (u_char *)rbp;
225 	switch (zs->zs_state) {
226 	case S_START:
227 		zs->zs_state = S_MIDDLE;
228 		break;
229 	case S_MIDDLE:
230 		goto middle;
231 	case S_EOF:
232 		goto eof;
233 	}
234 
235 	/* Check the magic number */
236 	for (i = 0; i < 3 && compressed_prelen; i++, compressed_prelen--)
237 		header[i] = *compressed_pre++;
238 
239 	if (fread(header + i, 1, sizeof(header) - i, zs->zs_fp) !=
240 		  sizeof(header) - i ||
241 	    memcmp(header, magic_header, sizeof(magic_header)) != 0) {
242 		errno = EFTYPE;
243 		return (-1);
244 	}
245 	total_compressed_bytes = 0;
246 	zs->zs_maxbits = header[2];	/* Set -b from file. */
247 	zs->zs_block_compress = zs->zs_maxbits & BLOCK_MASK;
248 	zs->zs_maxbits &= BIT_MASK;
249 	zs->zs_maxmaxcode = 1L << zs->zs_maxbits;
250 	if (zs->zs_maxbits > BITS || zs->zs_maxbits < 12) {
251 		errno = EFTYPE;
252 		return (-1);
253 	}
254 	/* As above, initialize the first 256 entries in the table. */
255 	zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
256 	for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0; zs->u.r.zs_code--) {
257 		tab_prefixof(zs->u.r.zs_code) = 0;
258 		tab_suffixof(zs->u.r.zs_code) = (char_type) zs->u.r.zs_code;
259 	}
260 	zs->zs_free_ent = zs->zs_block_compress ? FIRST : 256;
261 
262 	zs->u.r.zs_oldcode = -1;
263 	zs->u.r.zs_stackp = de_stack;
264 
265 	while ((zs->u.r.zs_code = getcode(zs)) > -1) {
266 
267 		if ((zs->u.r.zs_code == CLEAR) && zs->zs_block_compress) {
268 			for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0;
269 			    zs->u.r.zs_code--)
270 				tab_prefixof(zs->u.r.zs_code) = 0;
271 			zs->zs_clear_flg = 1;
272 			zs->zs_free_ent = FIRST;
273 			zs->u.r.zs_oldcode = -1;
274 			continue;
275 		}
276 		zs->u.r.zs_incode = zs->u.r.zs_code;
277 
278 		/* Special case for KwKwK string. */
279 		if (zs->u.r.zs_code >= zs->zs_free_ent) {
280 			if (zs->u.r.zs_code > zs->zs_free_ent ||
281 			    zs->u.r.zs_oldcode == -1) {
282 				/* Bad stream. */
283 				errno = EFTYPE;
284 				return (-1);
285 			}
286 			*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar;
287 			zs->u.r.zs_code = zs->u.r.zs_oldcode;
288 		}
289 		/*
290 		 * The above condition ensures that code < free_ent.
291 		 * The construction of tab_prefixof in turn guarantees that
292 		 * each iteration decreases code and therefore stack usage is
293 		 * bound by 1 << BITS - 256.
294 		 */
295 
296 		/* Generate output characters in reverse order. */
297 		while (zs->u.r.zs_code >= 256) {
298 			*zs->u.r.zs_stackp++ = tab_suffixof(zs->u.r.zs_code);
299 			zs->u.r.zs_code = tab_prefixof(zs->u.r.zs_code);
300 		}
301 		*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar = tab_suffixof(zs->u.r.zs_code);
302 
303 		/* And put them out in forward order.  */
304 middle:		do {
305 			if (count-- == 0)
306 				return (num);
307 			*bp++ = *--zs->u.r.zs_stackp;
308 		} while (zs->u.r.zs_stackp > de_stack);
309 
310 		/* Generate the new entry. */
311 		if ((zs->u.r.zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode &&
312 		    zs->u.r.zs_oldcode != -1) {
313 			tab_prefixof(zs->u.r.zs_code) = (u_short) zs->u.r.zs_oldcode;
314 			tab_suffixof(zs->u.r.zs_code) = zs->u.r.zs_finchar;
315 			zs->zs_free_ent = zs->u.r.zs_code + 1;
316 		}
317 
318 		/* Remember previous code. */
319 		zs->u.r.zs_oldcode = zs->u.r.zs_incode;
320 	}
321 	zs->zs_state = S_EOF;
322 eof:	return (num - count);
323 }
324 
325 /*-
326  * Read one code from the standard input.  If EOF, return -1.
327  * Inputs:
328  * 	stdin
329  * Outputs:
330  * 	code or -1 is returned.
331  */
332 static code_int
333 getcode(struct s_zstate *zs)
334 {
335 	code_int gcode;
336 	int r_off, bits, i;
337 	char_type *bp;
338 
339 	bp = zs->u.r.zs_gbuf;
340 	if (zs->zs_clear_flg > 0 || zs->u.r.zs_roffset >= zs->u.r.zs_size ||
341 	    zs->zs_free_ent > zs->zs_maxcode) {
342 		/*
343 		 * If the next entry will be too big for the current gcode
344 		 * size, then we must increase the size.  This implies reading
345 		 * a new buffer full, too.
346 		 */
347 		if (zs->zs_free_ent > zs->zs_maxcode) {
348 			zs->zs_n_bits++;
349 			if (zs->zs_n_bits == zs->zs_maxbits)	/* Won't get any bigger now. */
350 				zs->zs_maxcode = zs->zs_maxmaxcode;
351 			else
352 				zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
353 		}
354 		if (zs->zs_clear_flg > 0) {
355 			zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
356 			zs->zs_clear_flg = 0;
357 		}
358 		/* XXX */
359 		for (i = 0; i < zs->zs_n_bits && compressed_prelen; i++, compressed_prelen--)
360 			zs->u.r.zs_gbuf[i] = *compressed_pre++;
361 		zs->u.r.zs_size = fread(zs->u.r.zs_gbuf + i, 1, zs->zs_n_bits - i, zs->zs_fp);
362 		zs->u.r.zs_size += i;
363 		if (zs->u.r.zs_size <= 0)			/* End of file. */
364 			return (-1);
365 		zs->u.r.zs_roffset = 0;
366 
367 		total_compressed_bytes += zs->u.r.zs_size;
368 
369 		/* Round size down to integral number of codes. */
370 		zs->u.r.zs_size = (zs->u.r.zs_size << 3) - (zs->zs_n_bits - 1);
371 	}
372 	r_off = zs->u.r.zs_roffset;
373 	bits = zs->zs_n_bits;
374 
375 	/* Get to the first byte. */
376 	bp += (r_off >> 3);
377 	r_off &= 7;
378 
379 	/* Get first part (low order bits). */
380 	gcode = (*bp++ >> r_off);
381 	bits -= (8 - r_off);
382 	r_off = 8 - r_off;	/* Now, roffset into gcode word. */
383 
384 	/* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
385 	if (bits >= 8) {
386 		gcode |= *bp++ << r_off;
387 		r_off += 8;
388 		bits -= 8;
389 	}
390 
391 	/* High order bits. */
392 	gcode |= (*bp & rmask[bits]) << r_off;
393 	zs->u.r.zs_roffset += zs->zs_n_bits;
394 
395 	return (gcode);
396 }
397 
398