xref: /freebsd/usr.bin/gzip/zuncompress.c (revision 8fc257994d0ce2396196d7a06d50d20c8015f4b7)
1 /*	$NetBSD: zuncompress.c,v 1.7 2009/04/12 10:31:14 lukem 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 /* XXX zuncompress global */
81 off_t total_compressed_bytes;
82 size_t compressed_prelen;
83 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 paramenters */
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, sizeof(buf), 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) {
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_finchar = zs->u.r.zs_oldcode = getcode(zs);
263 	if (zs->u.r.zs_oldcode == -1)	/* EOF already? */
264 		return (0);	/* Get out of here */
265 
266 	/* First code must be 8 bits = char. */
267 	*bp++ = (u_char)zs->u.r.zs_finchar;
268 	count--;
269 	zs->u.r.zs_stackp = de_stack;
270 
271 	while ((zs->u.r.zs_code = getcode(zs)) > -1) {
272 
273 		if ((zs->u.r.zs_code == CLEAR) && zs->zs_block_compress) {
274 			for (zs->u.r.zs_code = 255; zs->u.r.zs_code >= 0;
275 			    zs->u.r.zs_code--)
276 				tab_prefixof(zs->u.r.zs_code) = 0;
277 			zs->zs_clear_flg = 1;
278 			zs->zs_free_ent = FIRST - 1;
279 			if ((zs->u.r.zs_code = getcode(zs)) == -1)	/* O, untimely death! */
280 				break;
281 		}
282 		zs->u.r.zs_incode = zs->u.r.zs_code;
283 
284 		/* Special case for KwKwK string. */
285 		if (zs->u.r.zs_code >= zs->zs_free_ent) {
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 		/* Generate output characters in reverse order. */
291 		while (zs->u.r.zs_code >= 256) {
292 			*zs->u.r.zs_stackp++ = tab_suffixof(zs->u.r.zs_code);
293 			zs->u.r.zs_code = tab_prefixof(zs->u.r.zs_code);
294 		}
295 		*zs->u.r.zs_stackp++ = zs->u.r.zs_finchar = tab_suffixof(zs->u.r.zs_code);
296 
297 		/* And put them out in forward order.  */
298 middle:		do {
299 			if (count-- == 0)
300 				return (num);
301 			*bp++ = *--zs->u.r.zs_stackp;
302 		} while (zs->u.r.zs_stackp > de_stack);
303 
304 		/* Generate the new entry. */
305 		if ((zs->u.r.zs_code = zs->zs_free_ent) < zs->zs_maxmaxcode) {
306 			tab_prefixof(zs->u.r.zs_code) = (u_short) zs->u.r.zs_oldcode;
307 			tab_suffixof(zs->u.r.zs_code) = zs->u.r.zs_finchar;
308 			zs->zs_free_ent = zs->u.r.zs_code + 1;
309 		}
310 
311 		/* Remember previous code. */
312 		zs->u.r.zs_oldcode = zs->u.r.zs_incode;
313 	}
314 	zs->zs_state = S_EOF;
315 eof:	return (num - count);
316 }
317 
318 /*-
319  * Read one code from the standard input.  If EOF, return -1.
320  * Inputs:
321  * 	stdin
322  * Outputs:
323  * 	code or -1 is returned.
324  */
325 static code_int
326 getcode(struct s_zstate *zs)
327 {
328 	code_int gcode;
329 	int r_off, bits, i;
330 	char_type *bp;
331 
332 	bp = zs->u.r.zs_gbuf;
333 	if (zs->zs_clear_flg > 0 || zs->u.r.zs_roffset >= zs->u.r.zs_size ||
334 	    zs->zs_free_ent > zs->zs_maxcode) {
335 		/*
336 		 * If the next entry will be too big for the current gcode
337 		 * size, then we must increase the size.  This implies reading
338 		 * a new buffer full, too.
339 		 */
340 		if (zs->zs_free_ent > zs->zs_maxcode) {
341 			zs->zs_n_bits++;
342 			if (zs->zs_n_bits == zs->zs_maxbits)	/* Won't get any bigger now. */
343 				zs->zs_maxcode = zs->zs_maxmaxcode;
344 			else
345 				zs->zs_maxcode = MAXCODE(zs->zs_n_bits);
346 		}
347 		if (zs->zs_clear_flg > 0) {
348 			zs->zs_maxcode = MAXCODE(zs->zs_n_bits = INIT_BITS);
349 			zs->zs_clear_flg = 0;
350 		}
351 		/* XXX */
352 		for (i = 0; i < zs->zs_n_bits && compressed_prelen; i++, compressed_prelen--)
353 			zs->u.r.zs_gbuf[i] = *compressed_pre++;
354 		zs->u.r.zs_size = fread(zs->u.r.zs_gbuf + i, 1, zs->zs_n_bits - i, zs->zs_fp);
355 		zs->u.r.zs_size += i;
356 		if (zs->u.r.zs_size <= 0)			/* End of file. */
357 			return (-1);
358 		zs->u.r.zs_roffset = 0;
359 
360 		total_compressed_bytes += zs->u.r.zs_size;
361 
362 		/* Round size down to integral number of codes. */
363 		zs->u.r.zs_size = (zs->u.r.zs_size << 3) - (zs->zs_n_bits - 1);
364 	}
365 	r_off = zs->u.r.zs_roffset;
366 	bits = zs->zs_n_bits;
367 
368 	/* Get to the first byte. */
369 	bp += (r_off >> 3);
370 	r_off &= 7;
371 
372 	/* Get first part (low order bits). */
373 	gcode = (*bp++ >> r_off);
374 	bits -= (8 - r_off);
375 	r_off = 8 - r_off;	/* Now, roffset into gcode word. */
376 
377 	/* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */
378 	if (bits >= 8) {
379 		gcode |= *bp++ << r_off;
380 		r_off += 8;
381 		bits -= 8;
382 	}
383 
384 	/* High order bits. */
385 	gcode |= (*bp & rmask[bits]) << r_off;
386 	zs->u.r.zs_roffset += zs->zs_n_bits;
387 
388 	return (gcode);
389 }
390 
391