xref: /freebsd/stand/libsa/bzipfs.c (revision 4fe1295c964fa712dd763e3852187da8724ef79a)
1 /*
2  * Copyright (c) 1998 Michael Smith.
3  * Copyright (c) 2000 Maxim Sobolev
4  * All rights reserved.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
19  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25  * SUCH DAMAGE.
26  */
27 
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
30 
31 #ifndef REGRESSION
32 #include "stand.h"
33 #else
34 #include <stdlib.h>
35 #include <sys/errno.h>
36 #include <sys/fcntl.h>
37 #include <sys/types.h>
38 #include <sys/unistd.h>
39 
40 struct open_file {
41     int                 f_flags;        /* see F_* below */
42     void                *f_fsdata;      /* file system specific data */
43 };
44 #define F_READ          0x0001  /* file opened for reading */
45 #define EOFFSET (ELAST+8)       /* relative seek not supported */
46 static inline u_int min(u_int a, u_int b) { return(a < b ? a : b); }
47 #define panic(x, y) abort()
48 #endif
49 
50 #include <sys/stat.h>
51 #include <string.h>
52 #include <bzlib.h>
53 
54 #define BZ_BUFSIZE 2048	/* XXX larger? */
55 
56 struct bz_file
57 {
58     int			bzf_rawfd;
59     bz_stream		bzf_bzstream;
60     char		bzf_buf[BZ_BUFSIZE];
61     int			bzf_endseen;
62 };
63 
64 static int	bzf_fill(struct bz_file *z);
65 static int	bzf_open(const char *path, struct open_file *f);
66 static int	bzf_close(struct open_file *f);
67 static int	bzf_read(struct open_file *f, void *buf, size_t size, size_t *resid);
68 static off_t	bzf_seek(struct open_file *f, off_t offset, int where);
69 static int	bzf_stat(struct open_file *f, struct stat *sb);
70 
71 #ifndef REGRESSION
72 struct fs_ops bzipfs_fsops = {
73     "bzip",
74     bzf_open,
75     bzf_close,
76     bzf_read,
77     null_write,
78     bzf_seek,
79     bzf_stat,
80     null_readdir
81 };
82 #endif
83 
84 static int
85 bzf_fill(struct bz_file *bzf)
86 {
87     int		result;
88     int		req;
89 
90     req = BZ_BUFSIZE - bzf->bzf_bzstream.avail_in;
91     result = 0;
92 
93     /* If we need more */
94     if (req > 0) {
95 	/* move old data to bottom of buffer */
96 	if (req < BZ_BUFSIZE)
97 	    bcopy(bzf->bzf_buf + req, bzf->bzf_buf, BZ_BUFSIZE - req);
98 
99 	/* read to fill buffer and update availibility data */
100 	result = read(bzf->bzf_rawfd, bzf->bzf_buf + bzf->bzf_bzstream.avail_in, req);
101 	bzf->bzf_bzstream.next_in = bzf->bzf_buf;
102 	if (result >= 0)
103 	    bzf->bzf_bzstream.avail_in += result;
104     }
105     return(result);
106 }
107 
108 /*
109  * Adapted from get_byte/check_header in libz
110  *
111  * Returns 0 if the header is OK, nonzero if not.
112  */
113 static int
114 get_byte(struct bz_file *bzf)
115 {
116     if ((bzf->bzf_bzstream.avail_in == 0) && (bzf_fill(bzf) == -1))
117 	return(-1);
118     bzf->bzf_bzstream.avail_in--;
119     return(*(bzf->bzf_bzstream.next_in)++);
120 }
121 
122 static int bz_magic[3] = {'B', 'Z', 'h'}; /* bzip2 magic header */
123 
124 static int
125 check_header(struct bz_file *bzf)
126 {
127     unsigned int len;
128     int		 c;
129 
130     /* Check the bzip2 magic header */
131     for (len = 0; len < 3; len++) {
132 	c = get_byte(bzf);
133 	if (c != bz_magic[len]) {
134 	    return(1);
135 	}
136     }
137     /* Check that the block size is valid */
138     c = get_byte(bzf);
139     if (c < '1' || c > '9')
140 	return(1);
141 
142     /* Put back bytes that we've took from the input stream */
143     bzf->bzf_bzstream.next_in -= 4;
144     bzf->bzf_bzstream.avail_in += 4;
145 
146     return(0);
147 }
148 
149 static int
150 bzf_open(const char *fname, struct open_file *f)
151 {
152     static char		*bzfname;
153     int			rawfd;
154     struct bz_file	*bzf;
155     char		*cp;
156     int			error;
157     struct stat		sb;
158 
159     /* Have to be in "just read it" mode */
160     if (f->f_flags != F_READ)
161 	return(EPERM);
162 
163     /* If the name already ends in .gz or .bz2, ignore it */
164     if ((cp = strrchr(fname, '.')) && (!strcmp(cp, ".gz")
165 	    || !strcmp(cp, ".bz2") || !strcmp(cp, ".split")))
166 	return(ENOENT);
167 
168     /* Construct new name */
169     bzfname = malloc(strlen(fname) + 5);
170     if (bzfname == NULL)
171 	return(ENOMEM);
172     sprintf(bzfname, "%s.bz2", fname);
173 
174     /* Try to open the compressed datafile */
175     rawfd = open(bzfname, O_RDONLY);
176     free(bzfname);
177     if (rawfd == -1)
178 	return(ENOENT);
179 
180     if (fstat(rawfd, &sb) < 0) {
181 	printf("bzf_open: stat failed\n");
182 	close(rawfd);
183 	return(ENOENT);
184     }
185     if (!S_ISREG(sb.st_mode)) {
186 	printf("bzf_open: not a file\n");
187 	close(rawfd);
188 	return(EISDIR);			/* best guess */
189     }
190 
191     /* Allocate a bz_file structure, populate it */
192     bzf = malloc(sizeof(struct bz_file));
193     if (bzf == NULL)
194 	return(ENOMEM);
195     bzero(bzf, sizeof(struct bz_file));
196     bzf->bzf_rawfd = rawfd;
197 
198     /* Verify that the file is bzipped */
199     if (check_header(bzf)) {
200 	close(bzf->bzf_rawfd);
201 	free(bzf);
202 	return(EFTYPE);
203     }
204 
205     /* Initialise the inflation engine */
206     if ((error = BZ2_bzDecompressInit(&(bzf->bzf_bzstream), 0, 1)) != BZ_OK) {
207 	printf("bzf_open: BZ2_bzDecompressInit returned %d\n", error);
208 	close(bzf->bzf_rawfd);
209 	free(bzf);
210 	return(EIO);
211     }
212 
213     /* Looks OK, we'll take it */
214     f->f_fsdata = bzf;
215     return(0);
216 }
217 
218 static int
219 bzf_close(struct open_file *f)
220 {
221     struct bz_file	*bzf = (struct bz_file *)f->f_fsdata;
222 
223     BZ2_bzDecompressEnd(&(bzf->bzf_bzstream));
224     close(bzf->bzf_rawfd);
225     free(bzf);
226     return(0);
227 }
228 
229 static int
230 bzf_read(struct open_file *f, void *buf, size_t size, size_t *resid)
231 {
232     struct bz_file	*bzf = (struct bz_file *)f->f_fsdata;
233     int			error;
234 
235     bzf->bzf_bzstream.next_out = buf;			/* where and how much */
236     bzf->bzf_bzstream.avail_out = size;
237 
238     while (bzf->bzf_bzstream.avail_out && bzf->bzf_endseen == 0) {
239 	if ((bzf->bzf_bzstream.avail_in == 0) && (bzf_fill(bzf) == -1)) {
240 	    printf("bzf_read: fill error\n");
241 	    return(EIO);
242 	}
243 	if (bzf->bzf_bzstream.avail_in == 0) {		/* oops, unexpected EOF */
244 	    printf("bzf_read: unexpected EOF\n");
245 	    if (bzf->bzf_bzstream.avail_out == size)
246 		return(EIO);
247 	    break;
248 	}
249 
250 	error = BZ2_bzDecompress(&bzf->bzf_bzstream);	/* decompression pass */
251 	if (error == BZ_STREAM_END) {			/* EOF, all done */
252 	    bzf->bzf_endseen = 1;
253 	    break;
254 	}
255 	if (error != BZ_OK) {				/* argh, decompression error */
256 	    printf("bzf_read: BZ2_bzDecompress returned %d\n", error);
257 	    return(EIO);
258 	}
259     }
260     if (resid != NULL)
261 	*resid = bzf->bzf_bzstream.avail_out;
262     return(0);
263 }
264 
265 static int
266 bzf_rewind(struct open_file *f)
267 {
268     struct bz_file	*bzf = (struct bz_file *)f->f_fsdata;
269     struct bz_file	*bzf_tmp;
270 
271     /*
272      * Since bzip2 does not have an equivalent inflateReset function a crude
273      * one needs to be provided.  The functions all called in such a way that
274      * at any time an error occurs a roll back can be done (effectively making
275      * this rewind 'atomic', either the reset occurs successfully or not at all,
276      * with no 'undefined' state happening).
277      */
278 
279     /* Allocate a bz_file structure, populate it */
280     bzf_tmp = malloc(sizeof(struct bz_file));
281     if (bzf_tmp == NULL)
282 	return(-1);
283     bzero(bzf_tmp, sizeof(struct bz_file));
284     bzf_tmp->bzf_rawfd = bzf->bzf_rawfd;
285 
286     /* Initialise the inflation engine */
287     if (BZ2_bzDecompressInit(&(bzf_tmp->bzf_bzstream), 0, 1) != BZ_OK) {
288 	free(bzf_tmp);
289 	return(-1);
290     }
291 
292     /* Seek back to the beginning of the file */
293     if (lseek(bzf->bzf_rawfd, 0, SEEK_SET) == -1) {
294 	BZ2_bzDecompressEnd(&(bzf_tmp->bzf_bzstream));
295 	free(bzf_tmp);
296 	return(-1);
297     }
298 
299     /* Free old bz_file data */
300     BZ2_bzDecompressEnd(&(bzf->bzf_bzstream));
301     free(bzf);
302 
303     /* Use the new bz_file data */
304     f->f_fsdata = bzf_tmp;
305 
306     return(0);
307 }
308 
309 static off_t
310 bzf_seek(struct open_file *f, off_t offset, int where)
311 {
312     struct bz_file	*bzf = (struct bz_file *)f->f_fsdata;
313     off_t		target;
314     char		discard[16];
315 
316     switch (where) {
317     case SEEK_SET:
318 	target = offset;
319 	break;
320     case SEEK_CUR:
321 	target = offset + bzf->bzf_bzstream.total_out_lo32;
322 	break;
323     default:
324 	errno = EINVAL;
325 	return(-1);
326     }
327 
328     /* Can we get there from here? */
329     if (target < bzf->bzf_bzstream.total_out_lo32 && bzf_rewind(f) != 0) {
330 	errno = EOFFSET;
331 	return -1;
332     }
333 
334     /* if bzf_rewind was called then bzf has changed */
335     bzf = (struct bz_file *)f->f_fsdata;
336 
337     /* skip forwards if required */
338     while (target > bzf->bzf_bzstream.total_out_lo32) {
339 	errno = bzf_read(f, discard, min(sizeof(discard),
340 	    target - bzf->bzf_bzstream.total_out_lo32), NULL);
341 	if (errno)
342 	    return(-1);
343 	/* Break out of loop if end of file has been reached. */
344 	if (bzf->bzf_endseen)
345 	    break;
346     }
347     /* This is where we are (be honest if we overshot) */
348     return(bzf->bzf_bzstream.total_out_lo32);
349 }
350 
351 static int
352 bzf_stat(struct open_file *f, struct stat *sb)
353 {
354     struct bz_file	*bzf = (struct bz_file *)f->f_fsdata;
355     int			result;
356 
357     /* stat as normal, but indicate that size is unknown */
358     if ((result = fstat(bzf->bzf_rawfd, sb)) == 0)
359 	sb->st_size = -1;
360     return(result);
361 }
362 
363 void
364 bz_internal_error(int errorcode)
365 {
366     panic("bzipfs: critical error %d in bzip2 library occurred", errorcode);
367 }
368 
369 #ifdef REGRESSION
370 /* Small test case, open and decompress test.bz2 */
371 int main()
372 {
373     struct open_file f;
374     char buf[1024];
375     size_t resid;
376     int err;
377 
378     memset(&f, '\0', sizeof(f));
379     f.f_flags = F_READ;
380     err = bzf_open("test", &f);
381     if (err != 0)
382 	exit(1);
383     do {
384 	err = bzf_read(&f, buf, sizeof(buf), &resid);
385     } while (err == 0 && resid != sizeof(buf));
386 
387     if (err != 0)
388 	exit(2);
389     exit(0);
390 }
391 #endif
392