1 /* vi: set sw=4 ts=4: */
3 * Gzip implementation for busybox
5 * Based on GNU gzip Copyright (C) 1992-1993 Jean-loup Gailly.
7 * Originally adjusted for busybox by Sven Rudolph <sr1@inf.tu-dresden.de>
8 * based on gzip sources
10 * Adjusted further by Erik Andersen <andersen@lineo.com>, <andersee@debian.org>
11 * to support files as well as stdin/stdout, and to generally behave itself wrt
12 * command line handling.
14 * This program is free software; you can redistribute it and/or modify
15 * it under the terms of the GNU General Public License as published by
16 * the Free Software Foundation; either version 2 of the License, or
17 * (at your option) any later version.
19 * This program is distributed in the hope that it will be useful,
20 * but WITHOUT ANY WARRANTY; without even the implied warranty of
21 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22 * General Public License for more details.
24 * You should have received a copy of the GNU General Public License
25 * along with this program; if not, write to the Free Software
26 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
33 /* These defines are very important for BusyBox. Without these,
34 * huge chunks of ram are pre-allocated making the BusyBox bss
35 * size Freaking Huge(tm), which is a bad thing.*/
39 #define BB_DECLARE_EXTERN
40 #define bb_need_memory_exhausted
41 #define bb_need_name_too_long
45 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
46 * Copyright (C) 1992-1993 Jean-loup Gailly
47 * The unzip code was written and put in the public domain by Mark Adler.
48 * Portions of the lzw code are derived from the public domain 'compress'
49 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
50 * Ken Turkowski, Dave Mack and Peter Jannesen.
52 * See the license_msg below and the file COPYING for the software license.
53 * See the file algorithm.doc for the compression algorithms and file formats.
57 static char *license_msg[] = {
58 " Copyright (C) 1992-1993 Jean-loup Gailly",
59 " This program is free software; you can redistribute it and/or modify",
60 " it under the terms of the GNU General Public License as published by",
61 " the Free Software Foundation; either version 2, or (at your option)",
62 " any later version.",
64 " This program is distributed in the hope that it will be useful,",
65 " but WITHOUT ANY WARRANTY; without even the implied warranty of",
66 " MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the",
67 " GNU General Public License for more details.",
69 " You should have received a copy of the GNU General Public License",
70 " along with this program; if not, write to the Free Software",
71 " Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.",
76 /* Compress files with zip algorithm and 'compress' interface.
77 * See usage() and help() functions below for all options.
79 * file.gz: compressed file with same mode, owner, and utimes
80 * or stdout with -c option or if stdin used as input.
81 * If the output file name had to be truncated, the original name is kept
82 * in the compressed file.
83 * On MSDOS, file.tmp -> file.tmz. On VMS, file.tmp -> file.tmp-gz.
85 * Using gz on MSDOS would create too many file name conflicts. For
86 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
87 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
88 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
89 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
91 * For the meaning of all compilation flags, see comments in Makefile.in.
95 #include <sys/types.h>
99 /* #include "tailor.h" */
101 /* tailor.h -- target dependent definitions
102 * Copyright (C) 1992-1993 Jean-loup Gailly.
103 * This is free software; you can redistribute it and/or modify it under the
104 * terms of the GNU General Public License, see the file COPYING.
107 /* The target dependent definitions should be defined here only.
108 * The target dependent functions should be defined in tailor.c.
113 #define get_char() get_byte()
114 #define put_char(c) put_byte(c)
117 /* I don't like nested includes, but the string and io functions are used
121 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
123 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
126 # define memzero(s, n) memset ((void *)(s), 0, (n))
128 # include <strings.h>
129 # define strchr index
130 # define strrchr rindex
131 # define memcpy(d, s, n) bcopy((s), (d), (n))
132 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
133 # define memzero(s, n) bzero((s), (n))
137 # define RETSIGTYPE void
142 typedef unsigned char uch;
143 typedef unsigned short ush;
144 typedef unsigned long ulg;
146 /* Return codes from gzip */
151 /* Compression methods (see algorithm.doc) */
154 extern int method; /* compression method */
156 /* To save memory for 16 bit systems, some arrays are overlaid between
157 * the various modules:
158 * deflate: prev+head window d_buf l_buf outbuf
159 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
160 * inflate: window inbuf
161 * unpack: window inbuf prefix_len
162 * unlzh: left+right window c_table inbuf c_len
163 * For compression, input is done in window[]. For decompression, output
164 * is done in window except for unlzw.
169 # define INBUFSIZ 0x2000 /* input buffer size */
171 # define INBUFSIZ 0x8000 /* input buffer size */
174 #define INBUF_EXTRA 64 /* required by unlzw() */
178 # define OUTBUFSIZ 8192 /* output buffer size */
180 # define OUTBUFSIZ 16384 /* output buffer size */
183 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
189 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
191 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
195 /*#define DYN_ALLOC*/
198 # define EXTERN(type, array) extern type * array
199 # define DECLARE(type, array, size) type * array
200 # define ALLOC(type, array, size) { \
201 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
202 if (array == NULL) errorMsg(memory_exhausted); \
204 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
206 # define EXTERN(type, array) extern type array[]
207 # define DECLARE(type, array, size) type array[size]
208 # define ALLOC(type, array, size)
212 EXTERN(uch, inbuf); /* input buffer */
213 EXTERN(uch, outbuf); /* output buffer */
214 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
215 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
216 #define tab_suffix window
218 # define tab_prefix prev /* hash link (see deflate.c) */
219 # define head (prev+WSIZE) /* hash head (see deflate.c) */
220 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
222 # define tab_prefix0 prev
223 # define head tab_prefix1
224 EXTERN(ush, tab_prefix0); /* prefix for even codes */
225 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
228 extern unsigned insize; /* valid bytes in inbuf */
229 extern unsigned inptr; /* index of next byte to be processed in inbuf */
230 extern unsigned outcnt; /* bytes in output buffer */
232 extern long bytes_in; /* number of input bytes */
233 extern long bytes_out; /* number of output bytes */
234 extern long header_bytes; /* number of bytes in gzip header */
236 extern long ifile_size; /* input file size, -1 for devices (debug only) */
238 typedef int file_t; /* Do not use stdio */
240 #define NO_FILE (-1) /* in memory compression */
243 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
246 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
247 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
248 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
249 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
250 #define COMMENT 0x10 /* bit 4 set: file comment present */
251 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
252 #define RESERVED 0xC0 /* bit 6,7: reserved */
255 # define WSIZE 0x8000 /* window size--must be a power of two, and */
256 #endif /* at least 32K for zip's deflate method */
259 #define MAX_MATCH 258
260 /* The minimum and maximum match lengths */
262 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
263 /* Minimum amount of lookahead, except at the end of the input file.
264 * See deflate.c for comments about the MIN_MATCH+1.
267 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
268 /* In order to simplify the code, particularly on 16 bit machines, match
269 * distances are limited to MAX_DIST instead of WSIZE.
272 extern int exit_code; /* program exit code */
273 extern int verbose; /* be verbose (-v) */
274 extern int level; /* compression level */
275 extern int test; /* check .z file integrity */
276 extern int save_orig_name; /* set if original name must be saved */
278 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
279 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
281 /* put_byte is used for the compressed output, put_ubyte for the
282 * uncompressed output. However unlzw() uses window for its
283 * suffix table instead of its output buffer, so it does not use put_ubyte
284 * (to be cleaned up).
286 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
288 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
291 /* Output a 16 bit value, lsb first */
292 #define put_short(w) \
293 { if (outcnt < OUTBUFSIZ-2) { \
294 outbuf[outcnt++] = (uch) ((w) & 0xff); \
295 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
297 put_byte((uch)((w) & 0xff)); \
298 put_byte((uch)((ush)(w) >> 8)); \
302 /* Output a 32 bit value to the bit stream, lsb first */
303 #define put_long(n) { \
304 put_short((n) & 0xffff); \
305 put_short(((ulg)(n)) >> 16); \
308 #define seekable() 0 /* force sequential output */
309 #define translate_eol 0 /* no option -a yet */
311 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
313 /* Macros for getting two-byte and four-byte header values */
314 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
315 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
317 /* Diagnostic functions */
319 # define Assert(cond,msg) {if(!(cond)) errorMsg(msg);}
320 # define Trace(x) fprintf x
321 # define Tracev(x) {if (verbose) fprintf x ;}
322 # define Tracevv(x) {if (verbose>1) fprintf x ;}
323 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
324 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
326 # define Assert(cond,msg)
331 # define Tracecv(c,x)
334 #define WARN(msg) {fprintf msg ; \
335 if (exit_code == OK) exit_code = WARNING;}
338 extern int unzip (int in, int out);
341 RETSIGTYPE abort_gzip (void);
344 void lm_init (int pack_level, ush * flags);
348 void ct_init (ush * attr, int *method);
349 int ct_tally (int dist, int lc);
350 ulg flush_block (char *buf, ulg stored_len, int eof);
353 void bi_init (file_t zipfile);
354 void send_bits (int value, int length);
355 unsigned bi_reverse (unsigned value, int length);
356 void bi_windup (void);
357 void copy_block (char *buf, unsigned len, int header);
360 extern ulg updcrc (uch * s, unsigned n);
361 extern void clear_bufs (void);
362 static int fill_inbuf (int eof_ok);
363 extern void flush_outbuf (void);
364 static void flush_window (void);
365 extern void write_buf (int fd, void * buf, unsigned cnt);
367 void read_error_msg (void);
368 void write_error_msg (void);
371 static int inflate (void);
373 /* #include "lzw.h" */
375 /* lzw.h -- define the lzw functions.
376 * Copyright (C) 1992-1993 Jean-loup Gailly.
377 * This is free software; you can redistribute it and/or modify it under the
378 * terms of the GNU General Public License, see the file COPYING.
381 #if !defined(OF) && defined(lint)
388 #define INIT_BITS 9 /* Initial number of bits per code */
390 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
392 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
393 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
394 * It's a pity that old uncompress does not check bit 0x20. That makes
395 * extension of the format actually undesirable because old compress
396 * would just crash on the new format instead of giving a meaningful
397 * error message. It does check the number of bits, but it's more
398 * helpful to say "unsupported format, get a new version" than
399 * "can only handle 16 bits".
402 #define BLOCK_MODE 0x80
403 /* Block compression: if table is full and compression rate is dropping,
404 * clear the dictionary.
407 #define LZW_RESERVED 0x60 /* reserved bits */
409 #define CLEAR 256 /* flush the dictionary */
410 #define FIRST (CLEAR+1) /* first free entry */
412 extern int maxbits; /* max bits per code for LZW */
413 extern int block_mode; /* block compress mode -C compatible with 2.0 */
415 extern int lzw (int in, int out);
416 extern int unlzw (int in, int out);
419 /* #include "revision.h" */
421 /* revision.h -- define the version number
422 * Copyright (C) 1992-1993 Jean-loup Gailly.
423 * This is free software; you can redistribute it and/or modify it under the
424 * terms of the GNU General Public License, see the file COPYING.
427 #define VERSION "1.2.4"
429 #define REVDATE "18 Aug 93"
431 /* This version does not support compression into old compress format: */
442 typedef struct dirent dir_type;
444 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
445 # define DIR_OPT "DIRENT"
447 # define NLENGTH(dirent) ((dirent)->d_namlen)
449 # include <sys/dir.h>
450 typedef struct direct dir_type;
452 # define DIR_OPT "SYSDIR"
455 # include <sys/ndir.h>
456 typedef struct direct dir_type;
458 # define DIR_OPT "SYSNDIR"
462 typedef struct direct dir_type;
464 # define DIR_OPT "NDIR"
467 # define DIR_OPT "NO_DIR"
472 #if !defined(S_ISDIR) && defined(S_IFDIR)
473 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
475 #if !defined(S_ISREG) && defined(S_IFREG)
476 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
478 typedef RETSIGTYPE(*sig_type) (int);
481 # define O_BINARY 0 /* creation mode for open() */
485 /* Pure BSD system? */
486 # include <sys/file.h>
488 # define O_CREAT FCREAT
491 # define O_EXCL FEXCL
496 # define S_IRUSR 0400
499 # define S_IWUSR 0200
501 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
503 #ifndef MAX_PATH_LEN /* max pathname length */
505 # define MAX_PATH_LEN BUFSIZ
507 # define MAX_PATH_LEN 1024
517 off_t lseek (int fd, off_t offset, int whence);
523 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
524 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
525 DECLARE(ush, d_buf, DIST_BUFSIZE);
526 DECLARE(uch, window, 2L * WSIZE);
528 DECLARE(ush, tab_prefix, 1L << BITS);
530 DECLARE(ush, tab_prefix0, 1L << (BITS - 1));
531 DECLARE(ush, tab_prefix1, 1L << (BITS - 1));
534 /* local variables */
536 int test_mode = 0; /* check file integrity option */
537 int foreground; /* set if program run in foreground */
538 int maxbits = BITS; /* max bits per code for LZW */
539 int method = DEFLATED; /* compression method */
540 int exit_code = OK; /* program exit code */
541 int last_member; /* set for .zip and .Z files */
542 int part_nb; /* number of parts in .gz file */
543 long ifile_size; /* input file size, -1 for devices (debug only) */
545 long bytes_in; /* number of input bytes */
546 long bytes_out; /* number of output bytes */
547 long total_in = 0; /* input bytes for all files */
548 long total_out = 0; /* output bytes for all files */
549 struct stat istat; /* status for input file */
550 int ifd; /* input file descriptor */
551 int ofd; /* output file descriptor */
552 unsigned insize; /* valid bytes in inbuf */
553 unsigned inptr; /* index of next byte to be processed in inbuf */
554 unsigned outcnt; /* bytes in output buffer */
556 long header_bytes; /* number of bytes in gzip header */
558 /* local functions */
560 local int get_method (int in);
562 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
564 /* ======================================================================== */
565 int gunzip_main(int argc, char **argv)
567 int file_count; /* number of files to precess */
573 int delInputFile = 0;
577 char ifname[MAX_PATH_LEN + 1]; /* input file name */
578 char ofname[MAX_PATH_LEN + 1]; /* output file name */
580 if (strcmp(applet_name, "zcat") == 0) {
585 /* Parse any options */
586 while (--argc > 0 && **(++argv) == '-') {
587 if (*((*argv) + 1) == '\0') {
590 while (*(++(*argv))) {
612 if (isatty(fileno(stdin)) && fromstdin==1 && force==0)
613 fatalError( "data not read from terminal. Use -f to force it.\n");
614 if (isatty(fileno(stdout)) && tostdout==1 && force==0)
615 fatalError( "data not written to terminal. Use -f to force it.\n");
618 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
620 (void) signal(SIGINT, (sig_type) abort_gzip);
623 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
624 (void) signal(SIGTERM, (sig_type) abort_gzip);
628 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
629 (void) signal(SIGHUP, (sig_type) abort_gzip);
633 file_count = argc - optind;
635 /* Allocate all global buffers (for DYN_ALLOC option) */
636 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
637 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
638 ALLOC(ush, d_buf, DIST_BUFSIZE);
639 ALLOC(uch, window, 2L * WSIZE);
641 ALLOC(ush, tab_prefix, 1L << BITS);
643 ALLOC(ush, tab_prefix0, 1L << (BITS - 1));
644 ALLOC(ush, tab_prefix1, 1L << (BITS - 1));
647 if (fromstdin == 1) {
648 strcpy(ofname, "stdin");
650 inFileNum = fileno(stdin);
651 ifile_size = -1L; /* convention for unknown size */
653 /* Open up the input file */
656 if (strlen(*argv) > MAX_PATH_LEN) {
657 errorMsg(name_too_long);
660 strcpy(ifname, *argv);
662 /* Open input fille */
663 inFileNum = open(ifname, O_RDONLY);
668 /* Get the time stamp on the input file. */
669 result = stat(ifname, &statBuf);
674 ifile_size = statBuf.st_size;
678 /* And get to work */
679 strcpy(ofname, "stdout");
680 outFileNum = fileno(stdout);
682 clear_bufs(); /* clear input and output buffers */
685 /* Actually do the compression/decompression. */
686 unzip(inFileNum, outFileNum);
688 } else if (test_mode) {
689 /* Actually do the compression/decompression. */
694 /* And get to work */
695 if (strlen(ifname) > MAX_PATH_LEN - 4) {
696 errorMsg(name_too_long);
699 strcpy(ofname, ifname);
700 pos = strstr(ofname, ".gz");
705 pos = strstr(ofname, ".tgz");
713 /* Open output fille */
714 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
715 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
717 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL);
719 if (outFileNum < 0) {
723 /* Set permissions on the file */
724 fchmod(outFileNum, statBuf.st_mode);
726 clear_bufs(); /* clear input and output buffers */
729 /* Actually do the compression/decompression. */
730 result = unzip(inFileNum, outFileNum);
734 /* Delete the original file */
736 delFileName = ifname;
738 delFileName = ofname;
740 if (delInputFile == 1 && unlink(delFileName) < 0) {
749 /* ========================================================================
750 * Check the magic number of the input file and update ofname if an
751 * original name was given and tostdout is not set.
752 * Return the compression method, -1 for error, -2 for warning.
753 * Set inptr to the offset of the next byte to be processed.
754 * Updates time_stamp if there is one and --no-time is not used.
755 * This function may be called repeatedly for an input file consisting
756 * of several contiguous gzip'ed members.
757 * IN assertions: there is at least one remaining compressed member.
758 * If the member is a zip file, it must be the only one.
760 local int get_method(in)
761 int in; /* input file descriptor */
763 uch flags; /* compression flags */
764 char magic[2]; /* magic header */
766 magic[0] = (char) get_byte();
767 magic[1] = (char) get_byte();
768 method = -1; /* unknown yet */
769 part_nb++; /* number of parts in gzip file */
771 last_member = RECORD_IO;
772 /* assume multiple members in gzip file except for record oriented I/O */
774 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
776 method = (int) get_byte();
777 if (method != DEFLATED) {
778 errorMsg("unknown method %d -- get newer version of gzip\n",
783 flags = (uch) get_byte();
785 (ulg) get_byte(); /* Ignore time stamp */
790 (void) get_byte(); /* Ignore extra flags for the moment */
791 (void) get_byte(); /* Ignore OS type for the moment */
793 if ((flags & EXTRA_FIELD) != 0) {
794 unsigned len = (unsigned) get_byte();
796 len |= ((unsigned) get_byte()) << 8;
802 /* Discard original name if any */
803 if ((flags & ORIG_NAME) != 0) {
804 while (get_char() != 0) /* null */
808 /* Discard file comment if any */
809 if ((flags & COMMENT) != 0) {
810 while (get_char() != 0) /* null */
814 header_bytes = inptr + 2 * sizeof(long); /* include crc and size */
823 fprintf(stderr, "\nnot in gzip format\n");
827 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
832 /* ========================================================================
833 * Signal and error handler.
835 RETSIGTYPE abort_gzip()
840 /* unzip.c -- decompress files in gzip or pkzip format.
841 * Copyright (C) 1992-1993 Jean-loup Gailly
842 * This is free software; you can redistribute it and/or modify it under the
843 * terms of the GNU General Public License, see the file COPYING.
845 * The code in this file is derived from the file funzip.c written
846 * and put in the public domain by Mark Adler.
850 This version can extract files in gzip or pkzip format.
851 For the latter, only the first entry is extracted, and it has to be
852 either deflated or stored.
855 /* #include "crypt.h" */
857 /* crypt.h (dummy version) -- do not perform encryption
858 * Hardly worth copyrighting :-)
862 # undef CRYPT /* dummy version */
865 #define RAND_HEAD_LEN 12 /* length of encryption random header */
870 /* PKZIP header definitions */
871 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
872 #define LOCFLG 6 /* offset of bit flag */
873 #define CRPFLG 1 /* bit for encrypted entry */
874 #define EXTFLG 8 /* bit for extended local header */
875 #define LOCHOW 8 /* offset of compression method */
876 #define LOCTIM 10 /* file mod time (for decryption) */
877 #define LOCCRC 14 /* offset of crc */
878 #define LOCSIZ 18 /* offset of compressed size */
879 #define LOCLEN 22 /* offset of uncompressed length */
880 #define LOCFIL 26 /* offset of file name field length */
881 #define LOCEXT 28 /* offset of extra field length */
882 #define LOCHDR 30 /* size of local header, including sig */
883 #define EXTHDR 16 /* size of extended local header, inc sig */
888 char *key; /* not used--needed to link crypt.c */
889 int pkzip = 0; /* set for a pkzip file */
890 int ext_header = 0; /* set if extended local header */
892 /* ===========================================================================
893 * Unzip in to out. This routine works on both gzip and pkzip files.
895 * IN assertions: the buffer inbuf contains already the beginning of
896 * the compressed data, from offsets inptr to insize-1 included.
897 * The magic header has already been checked. The output buffer is cleared.
900 int in, out; /* input and output file descriptors */
902 ulg orig_crc = 0; /* original crc */
903 ulg orig_len = 0; /* original uncompressed length */
905 uch buf[EXTHDR]; /* extended local header */
909 method = get_method(ifd);
911 exit(exit_code); /* error message already emitted */
914 updcrc(NULL, 0); /* initialize crc */
916 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
917 orig_crc = LG(inbuf + LOCCRC);
918 orig_len = LG(inbuf + LOCLEN);
922 if (method == DEFLATED) {
927 errorMsg(memory_exhausted);
928 } else if (res != 0) {
929 errorMsg("invalid compressed data--format violated");
933 errorMsg("internal error, invalid method");
936 /* Get the crc and original length */
938 /* crc32 (see algorithm.doc)
939 * uncompressed input size modulo 2^32
941 for (n = 0; n < 8; n++) {
942 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
945 orig_len = LG(buf + 4);
947 } else if (ext_header) { /* If extended header, check it */
948 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
950 * compressed size 4-bytes
951 * uncompressed size 4-bytes
953 for (n = 0; n < EXTHDR; n++) {
954 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
956 orig_crc = LG(buf + 4);
957 orig_len = LG(buf + 12);
960 /* Validate decompression */
961 if (orig_crc != updcrc(outbuf, 0)) {
962 errorMsg("invalid compressed data--crc error");
964 if (orig_len != (ulg) bytes_out) {
965 errorMsg("invalid compressed data--length error");
968 /* Check if there are more entries in a pkzip file */
969 if (pkzip && inptr + 4 < insize && LG(inbuf + inptr) == LOCSIG) {
970 WARN((stderr, "has more than one entry--rest ignored\n"));
972 ext_header = pkzip = 0; /* for next file */
976 /* util.c -- utility functions for gzip support
977 * Copyright (C) 1992-1993 Jean-loup Gailly
978 * This is free software; you can redistribute it and/or modify it under the
979 * terms of the GNU General Public License, see the file COPYING.
984 #include <sys/types.h>
993 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
999 static const ulg crc_32_tab[]; /* crc table, defined below */
1001 /* ===========================================================================
1002 * Run a set of bytes through the crc shift register. If s is a NULL
1003 * pointer, then initialize the crc shift register contents instead.
1004 * Return the current crc in either case.
1007 uch *s; /* pointer to bytes to pump through */
1008 unsigned n; /* number of bytes in s[] */
1010 register ulg c; /* temporary variable */
1012 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
1020 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
1024 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1027 /* ===========================================================================
1028 * Clear input and output buffers
1030 void clear_bufs(void)
1034 bytes_in = bytes_out = 0L;
1037 /* ===========================================================================
1038 * Fill the input buffer. This is called only when the buffer is empty.
1040 int fill_inbuf(eof_ok)
1041 int eof_ok; /* set if EOF acceptable as a result */
1045 /* Read as much as possible */
1049 len = read(ifd, (char *) inbuf + insize, INBUFSIZ - insize);
1050 if (len == 0 || len == EOF)
1053 } while (insize < INBUFSIZ);
1060 bytes_in += (ulg) insize;
1065 /* ===========================================================================
1066 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1067 * (used for the compressed data only)
1075 write_buf(ofd, (char *) outbuf, outcnt);
1076 bytes_out += (ulg) outcnt;
1080 /* ===========================================================================
1081 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1082 * (Used for the decompressed data only.)
1088 updcrc(window, outcnt);
1091 write_buf(ofd, (char *) window, outcnt);
1092 bytes_out += (ulg) outcnt;
1096 /* ===========================================================================
1097 * Does the same as write(), but also handles partial pipe writes and checks
1100 void write_buf(fd, buf, cnt)
1107 while ((n = write(fd, buf, cnt)) != cnt) {
1108 if (n == (unsigned) (-1)) {
1112 buf = (void *) ((char *) buf + n);
1116 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1118 /* Provide missing strspn and strcspn functions. */
1124 int strspn (const char *s, const char *accept);
1125 int strcspn (const char *s, const char *reject);
1127 /* ========================================================================
1128 * Return the length of the maximum initial segment
1129 * of s which contains only characters in accept.
1131 int strspn(s, accept)
1135 register const char *p;
1136 register const char *a;
1137 register int count = 0;
1139 for (p = s; *p != '\0'; ++p) {
1140 for (a = accept; *a != '\0'; ++a) {
1151 /* ========================================================================
1152 * Return the length of the maximum inital segment of s
1153 * which contains no characters from reject.
1155 int strcspn(s, reject)
1159 register int count = 0;
1161 while (*s != '\0') {
1162 if (strchr(reject, *s++) != NULL)
1169 #endif /* NO_STRING_H */
1172 /* ========================================================================
1175 void read_error_msg()
1177 fprintf(stderr, "\n");
1181 fprintf(stderr, "unexpected end of file\n");
1186 void write_error_msg()
1188 fprintf(stderr, "\n");
1194 /* ========================================================================
1195 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1197 static const ulg crc_32_tab[] = {
1198 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1199 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1200 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1201 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1202 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1203 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1204 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1205 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1206 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1207 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1208 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1209 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1210 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1211 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1212 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1213 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1214 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1215 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1216 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1217 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1218 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1219 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1220 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1221 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1222 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1223 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1224 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1225 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1226 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1227 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1228 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1229 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1230 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1231 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1232 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1233 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1234 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1235 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1236 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1237 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1238 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1239 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1240 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1241 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1242 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1243 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1244 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1245 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1246 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1247 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1248 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1252 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1253 version c10p1, 10 January 1993 */
1255 /* You can do whatever you like with this source file, though I would
1256 prefer that if you modify it and redistribute it that you include
1257 comments to that effect with your name and the date. Thank you.
1258 [The history has been moved to the file ChangeLog.]
1262 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1263 method searches for as much of the current string of bytes (up to a
1264 length of 258) in the previous 32K bytes. If it doesn't find any
1265 matches (of at least length 3), it codes the next byte. Otherwise, it
1266 codes the length of the matched string and its distance backwards from
1267 the current position. There is a single Huffman code that codes both
1268 single bytes (called "literals") and match lengths. A second Huffman
1269 code codes the distance information, which follows a length code. Each
1270 length or distance code actually represents a base value and a number
1271 of "extra" (sometimes zero) bits to get to add to the base value. At
1272 the end of each deflated block is a special end-of-block (EOB) literal/
1273 length code. The decoding process is basically: get a literal/length
1274 code; if EOB then done; if a literal, emit the decoded byte; if a
1275 length then get the distance and emit the referred-to bytes from the
1276 sliding window of previously emitted data.
1278 There are (currently) three kinds of inflate blocks: stored, fixed, and
1279 dynamic. The compressor deals with some chunk of data at a time, and
1280 decides which method to use on a chunk-by-chunk basis. A chunk might
1281 typically be 32K or 64K. If the chunk is uncompressible, then the
1282 "stored" method is used. In this case, the bytes are simply stored as
1283 is, eight bits per byte, with none of the above coding. The bytes are
1284 preceded by a count, since there is no longer an EOB code.
1286 If the data is compressible, then either the fixed or dynamic methods
1287 are used. In the dynamic method, the compressed data is preceded by
1288 an encoding of the literal/length and distance Huffman codes that are
1289 to be used to decode this block. The representation is itself Huffman
1290 coded, and so is preceded by a description of that code. These code
1291 descriptions take up a little space, and so for small blocks, there is
1292 a predefined set of codes, called the fixed codes. The fixed method is
1293 used if the block codes up smaller that way (usually for quite small
1294 chunks), otherwise the dynamic method is used. In the latter case, the
1295 codes are customized to the probabilities in the current block, and so
1296 can code it much better than the pre-determined fixed codes.
1298 The Huffman codes themselves are decoded using a mutli-level table
1299 lookup, in order to maximize the speed of decoding plus the speed of
1300 building the decoding tables. See the comments below that precede the
1301 lbits and dbits tuning parameters.
1306 Notes beyond the 1.93a appnote.txt:
1308 1. Distance pointers never point before the beginning of the output
1310 2. Distance pointers can point back across blocks, up to 32k away.
1311 3. There is an implied maximum of 7 bits for the bit length table and
1312 15 bits for the actual data.
1313 4. If only one code exists, then it is encoded using one bit. (Zero
1314 would be more efficient, but perhaps a little confusing.) If two
1315 codes exist, they are coded using one bit each (0 and 1).
1316 5. There is no way of sending zero distance codes--a dummy must be
1317 sent if there are none. (History: a pre 2.0 version of PKZIP would
1318 store blocks with no distance codes, but this was discovered to be
1319 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1320 zero distance codes, which is sent as one code of zero bits in
1322 6. There are up to 286 literal/length codes. Code 256 represents the
1323 end-of-block. Note however that the static length tree defines
1324 288 codes just to fill out the Huffman codes. Codes 286 and 287
1325 cannot be used though, since there is no length base or extra bits
1326 defined for them. Similarly, there are up to 30 distance codes.
1327 However, static trees define 32 codes (all 5 bits) to fill out the
1328 Huffman codes, but the last two had better not show up in the data.
1329 7. Unzip can check dynamic Huffman blocks for complete code sets.
1330 The exception is that a single code would not be complete (see #4).
1331 8. The five bits following the block type is really the number of
1332 literal codes sent minus 257.
1333 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1334 (1+6+6). Therefore, to output three times the length, you output
1335 three codes (1+1+1), whereas to output four times the same length,
1336 you only need two codes (1+3). Hmm.
1337 10. In the tree reconstruction algorithm, Code = Code + Increment
1338 only if BitLength(i) is not zero. (Pretty obvious.)
1339 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1340 12. Note: length code 284 can represent 227-258, but length code 285
1341 really is 258. The last length deserves its own, short code
1342 since it gets used a lot in very redundant files. The length
1343 258 is special since 258 - 3 (the min match length) is 255.
1344 13. The literal/length and distance code bit lengths are read as a
1345 single stream of lengths. It is possible (and advantageous) for
1346 a repeat code (16, 17, or 18) to go across the boundary between
1347 the two sets of lengths.
1350 #include <sys/types.h>
1352 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1353 # include <stdlib.h>
1357 #define slide window
1359 /* Huffman code lookup table entry--this entry is four bytes for machines
1360 that have 16-bit pointers (e.g. PC's in the small or medium model).
1361 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1362 means that v is a literal, 16 < e < 32 means that v is a pointer to
1363 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1364 an unused code. If a code with e == 99 is looked up, this implies an
1365 error in the data. */
1367 uch e; /* number of extra bits or operation */
1368 uch b; /* number of bits in this code or subcode */
1370 ush n; /* literal, length base, or distance base */
1371 struct huft *t; /* pointer to next level of table */
1376 /* Function prototypes */
1377 int huft_build (unsigned *, unsigned, unsigned, ush *, ush *,
1378 struct huft **, int *);
1379 int huft_free (struct huft *);
1380 int inflate_codes (struct huft *, struct huft *, int, int);
1381 int inflate_stored (void);
1382 int inflate_fixed (void);
1383 int inflate_dynamic (void);
1384 int inflate_block (int *);
1388 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1389 stream to find repeated byte strings. This is implemented here as a
1390 circular buffer. The index is updated simply by incrementing and then
1391 and'ing with 0x7fff (32K-1). */
1392 /* It is left to other modules to supply the 32K area. It is assumed
1393 to be usable as if it were declared "uch slide[32768];" or as just
1394 "uch *slide;" and then malloc'ed in the latter case. The definition
1395 must be in unzip.h, included above. */
1396 /* unsigned wp; current position in slide */
1398 #define flush_output(w) (wp=(w),flush_window())
1400 /* Tables for deflate from PKZIP's appnote.txt. */
1401 static unsigned border[] = { /* Order of the bit length code lengths */
1402 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15
1404 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1405 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1406 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0
1409 /* note: see note #13 above about the 258 in this list. */
1410 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1411 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1412 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99
1413 }; /* 99==invalid */
1414 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1415 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1416 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1417 8193, 12289, 16385, 24577
1419 static ush cpdext[] = { /* Extra bits for distance codes */
1420 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1421 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1427 /* Macros for inflate() bit peeking and grabbing.
1431 x = b & mask_bits[j];
1434 where NEEDBITS makes sure that b has at least j bits in it, and
1435 DUMPBITS removes the bits from b. The macros use the variable k
1436 for the number of bits in b. Normally, b and k are register
1437 variables for speed, and are initialized at the beginning of a
1438 routine that uses these macros from a global bit buffer and count.
1440 If we assume that EOB will be the longest code, then we will never
1441 ask for bits with NEEDBITS that are beyond the end of the stream.
1442 So, NEEDBITS should not read any more bytes than are needed to
1443 meet the request. Then no bytes need to be "returned" to the buffer
1444 at the end of the last block.
1446 However, this assumption is not true for fixed blocks--the EOB code
1447 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1448 (The EOB code is shorter than other codes because fixed blocks are
1449 generally short. So, while a block always has an EOB, many other
1450 literal/length codes have a significantly lower probability of
1451 showing up at all.) However, by making the first table have a
1452 lookup of seven bits, the EOB code will be found in that first
1453 lookup, and so will not require that too many bits be pulled from
1457 ulg bb; /* bit buffer */
1458 unsigned bk; /* bits in bit buffer */
1462 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1463 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1469 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1471 # define NEXTBYTE() (uch)get_byte()
1473 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1474 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1478 Huffman code decoding is performed using a multi-level table lookup.
1479 The fastest way to decode is to simply build a lookup table whose
1480 size is determined by the longest code. However, the time it takes
1481 to build this table can also be a factor if the data being decoded
1482 is not very long. The most common codes are necessarily the
1483 shortest codes, so those codes dominate the decoding time, and hence
1484 the speed. The idea is you can have a shorter table that decodes the
1485 shorter, more probable codes, and then point to subsidiary tables for
1486 the longer codes. The time it costs to decode the longer codes is
1487 then traded against the time it takes to make longer tables.
1489 This results of this trade are in the variables lbits and dbits
1490 below. lbits is the number of bits the first level table for literal/
1491 length codes can decode in one step, and dbits is the same thing for
1492 the distance codes. Subsequent tables are also less than or equal to
1493 those sizes. These values may be adjusted either when all of the
1494 codes are shorter than that, in which case the longest code length in
1495 bits is used, or when the shortest code is *longer* than the requested
1496 table size, in which case the length of the shortest code in bits is
1499 There are two different values for the two tables, since they code a
1500 different number of possibilities each. The literal/length table
1501 codes 286 possible values, or in a flat code, a little over eight
1502 bits. The distance table codes 30 possible values, or a little less
1503 than five bits, flat. The optimum values for speed end up being
1504 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1505 The optimum values may differ though from machine to machine, and
1506 possibly even between compilers. Your mileage may vary.
1510 int lbits = 9; /* bits in base literal/length lookup table */
1511 int dbits = 6; /* bits in base distance lookup table */
1514 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1515 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1516 #define N_MAX 288 /* maximum number of codes in any set */
1519 unsigned hufts; /* track memory usage */
1522 int huft_build(b, n, s, d, e, t, m)
1523 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1524 unsigned n; /* number of codes (assumed <= N_MAX) */
1525 unsigned s; /* number of simple-valued codes (0..s-1) */
1526 ush *d; /* list of base values for non-simple codes */
1527 ush *e; /* list of extra bits for non-simple codes */
1528 struct huft **t; /* result: starting table */
1529 int *m; /* maximum lookup bits, returns actual */
1531 /* Given a list of code lengths and a maximum table size, make a set of
1532 tables to decode that set of codes. Return zero on success, one if
1533 the given code set is incomplete (the tables are still built in this
1534 case), two if the input is invalid (all zero length codes or an
1535 oversubscribed set of lengths), and three if not enough memory. */
1537 unsigned a; /* counter for codes of length k */
1538 unsigned c[BMAX + 1]; /* bit length count table */
1539 unsigned f; /* i repeats in table every f entries */
1540 int g; /* maximum code length */
1541 int h; /* table level */
1542 register unsigned i; /* counter, current code */
1543 register unsigned j; /* counter */
1544 register int k; /* number of bits in current code */
1545 int l; /* bits per table (returned in m) */
1546 register unsigned *p; /* pointer into c[], b[], or v[] */
1547 register struct huft *q; /* points to current table */
1548 struct huft r; /* table entry for structure assignment */
1549 struct huft *u[BMAX]; /* table stack */
1550 unsigned v[N_MAX]; /* values in order of bit length */
1551 register int w; /* bits before this table == (l * h) */
1552 unsigned x[BMAX + 1]; /* bit offsets, then code stack */
1553 unsigned *xp; /* pointer into x */
1554 int y; /* number of dummy codes added */
1555 unsigned z; /* number of entries in current table */
1558 /* Generate counts for each bit length */
1559 memzero(c, sizeof(c));
1566 && n - i <= '~' ? "%c %d\n" : "0x%x %d\n"), n - i, *p));
1567 c[*p]++; /* assume all entries <= BMAX */
1568 p++; /* Can't combine with above line (Solaris bug) */
1570 if (c[0] == n) { /* null input--all zero length codes */
1571 *t = (struct huft *) NULL;
1577 /* Find minimum and maximum length, bound *m by those */
1579 for (j = 1; j <= BMAX; j++)
1582 k = j; /* minimum code length */
1583 if ((unsigned) l < j)
1585 for (i = BMAX; i; i--)
1588 g = i; /* maximum code length */
1589 if ((unsigned) l > i)
1594 /* Adjust last length count to fill out codes, if needed */
1595 for (y = 1 << j; j < i; j++, y <<= 1)
1596 if ((y -= c[j]) < 0)
1597 return 2; /* bad input: more codes than bits */
1598 if ((y -= c[i]) < 0)
1603 /* Generate starting offsets into the value table for each length */
1607 while (--i) { /* note that i == g from above */
1608 *xp++ = (j += *p++);
1612 /* Make a table of values in order of bit lengths */
1616 if ((j = *p++) != 0)
1621 /* Generate the Huffman codes and for each, make the table entries */
1622 x[0] = i = 0; /* first Huffman code is zero */
1623 p = v; /* grab values in bit order */
1624 h = -1; /* no tables yet--level -1 */
1625 w = -l; /* bits decoded == (l * h) */
1626 u[0] = (struct huft *) NULL; /* just to keep compilers happy */
1627 q = (struct huft *) NULL; /* ditto */
1630 /* go through the bit lengths (k already is bits in shortest code) */
1631 for (; k <= g; k++) {
1634 /* here i is the Huffman code of length k bits for value *p */
1635 /* make tables up to required level */
1638 w += l; /* previous table always l bits */
1640 /* compute minimum size table less than or equal to l bits */
1641 z = (z = g - w) > (unsigned) l ? l : z; /* upper limit on table size */
1642 if ((f = 1 << (j = k - w)) > a + 1) { /* try a k-w bit table *//* too few codes for k-w bit table */
1643 f -= a + 1; /* deduct codes from patterns left */
1645 while (++j < z) { /* try smaller tables up to z bits */
1646 if ((f <<= 1) <= *++xp)
1647 break; /* enough codes to use up j bits */
1648 f -= *xp; /* else deduct codes from patterns */
1651 z = 1 << j; /* table entries for j-bit table */
1653 /* allocate and link in new table */
1656 (struct huft *) malloc((z + 1) *
1657 sizeof(struct huft))) ==
1658 (struct huft *) NULL) {
1661 return 3; /* not enough memory */
1663 hufts += z + 1; /* track memory usage */
1664 *t = q + 1; /* link to list for huft_free() */
1665 *(t = &(q->v.t)) = (struct huft *) NULL;
1666 u[h] = ++q; /* table starts after link */
1668 /* connect to last table, if there is one */
1670 x[h] = i; /* save pattern for backing up */
1671 r.b = (uch) l; /* bits to dump before this table */
1672 r.e = (uch) (16 + j); /* bits in this table */
1673 r.v.t = q; /* pointer to this table */
1674 j = i >> (w - l); /* (get around Turbo C bug) */
1675 u[h - 1][j] = r; /* connect to last table */
1679 /* set up table entry in r */
1680 r.b = (uch) (k - w);
1682 r.e = 99; /* out of values--invalid code */
1684 r.e = (uch) (*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1685 r.v.n = (ush) (*p); /* simple code is just the value */
1686 p++; /* one compiler does not like *p++ */
1688 r.e = (uch) e[*p - s]; /* non-simple--look up in lists */
1689 r.v.n = d[*p++ - s];
1692 /* fill code-like entries with r */
1694 for (j = i >> w; j < z; j += f)
1697 /* backwards increment the k-bit code i */
1698 for (j = 1 << (k - 1); i & j; j >>= 1)
1702 /* backup over finished tables */
1703 while ((i & ((1 << w) - 1)) != x[h]) {
1704 h--; /* don't need to update q */
1711 /* Return true (1) if we were given an incomplete table */
1712 return y != 0 && g != 1;
1718 struct huft *t; /* table to free */
1720 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1721 list of the tables it made, with the links in a dummy first entry of
1724 register struct huft *p, *q;
1727 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1729 while (p != (struct huft *) NULL) {
1738 int inflate_codes(tl, td, bl, bd)
1739 struct huft *tl, *td; /* literal/length and distance decoder tables */
1740 int bl, bd; /* number of bits decoded by tl[] and td[] */
1742 /* inflate (decompress) the codes in a deflated (compressed) block.
1743 Return an error code or zero if it all goes ok. */
1745 register unsigned e; /* table entry flag/number of extra bits */
1746 unsigned n, d; /* length and index for copy */
1747 unsigned w; /* current window position */
1748 struct huft *t; /* pointer to table entry */
1749 unsigned ml, md; /* masks for bl and bd bits */
1750 register ulg b; /* bit buffer */
1751 register unsigned k; /* number of bits in bit buffer */
1754 /* make local copies of globals */
1755 b = bb; /* initialize bit buffer */
1757 w = wp; /* initialize window position */
1759 /* inflate the coded data */
1760 ml = mask_bits[bl]; /* precompute masks for speed */
1762 for (;;) { /* do until end of block */
1763 NEEDBITS((unsigned) bl)
1764 if ((e = (t = tl + ((unsigned) b & ml))->e) > 16)
1771 } while ((e = (t = t->v.t + ((unsigned) b & mask_bits[e]))->e)
1774 if (e == 16) { /* then it's a literal */
1775 slide[w++] = (uch) t->v.n;
1776 Tracevv((stderr, "%c", slide[w - 1]));
1781 } else { /* it's an EOB or a length */
1783 /* exit if end of block */
1787 /* get length of block to copy */
1789 n = t->v.n + ((unsigned) b & mask_bits[e]);
1792 /* decode distance of block to copy */
1793 NEEDBITS((unsigned) bd)
1794 if ((e = (t = td + ((unsigned) b & md))->e) > 16)
1805 t->v.t + ((unsigned) b & mask_bits[e]))->e) >
1809 d = w - t->v.n - ((unsigned) b & mask_bits[e]);
1811 Tracevv((stderr, "\\[%d,%d]", w - d, n));
1817 WSIZE - ((d &= WSIZE - 1) > w ? d : w)) >
1819 #if !defined(NOMEMCPY) && !defined(DEBUG)
1820 if (w - d >= e) { /* (this test assumes unsigned comparison) */
1821 memcpy(slide + w, slide + d, e);
1824 } else /* do it slow to avoid memcpy() overlap */
1825 #endif /* !NOMEMCPY */
1827 slide[w++] = slide[d++];
1828 Tracevv((stderr, "%c", slide[w - 1]));
1839 /* restore the globals from the locals */
1840 wp = w; /* restore global window pointer */
1841 bb = b; /* restore global bit buffer */
1850 int inflate_stored()
1851 /* "decompress" an inflated type 0 (stored) block. */
1853 unsigned n; /* number of bytes in block */
1854 unsigned w; /* current window position */
1855 register ulg b; /* bit buffer */
1856 register unsigned k; /* number of bits in bit buffer */
1859 /* make local copies of globals */
1860 b = bb; /* initialize bit buffer */
1862 w = wp; /* initialize window position */
1865 /* go to byte boundary */
1870 /* get the length and its complement */
1872 n = ((unsigned) b & 0xffff);
1875 if (n != (unsigned) ((~b) & 0xffff))
1876 return 1; /* error in compressed data */
1880 /* read and output the compressed data */
1883 slide[w++] = (uch) b;
1892 /* restore the globals from the locals */
1893 wp = w; /* restore global window pointer */
1894 bb = b; /* restore global bit buffer */
1902 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
1903 either replace this with a custom decoder, or at least precompute the
1906 int i; /* temporary variable */
1907 struct huft *tl; /* literal/length code table */
1908 struct huft *td; /* distance code table */
1909 int bl; /* lookup bits for tl */
1910 int bd; /* lookup bits for td */
1911 unsigned l[288]; /* length list for huft_build */
1914 /* set up literal table */
1915 for (i = 0; i < 144; i++)
1917 for (; i < 256; i++)
1919 for (; i < 280; i++)
1921 for (; i < 288; i++) /* make a complete, but wrong code set */
1924 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
1928 /* set up distance table */
1929 for (i = 0; i < 30; i++) /* make an incomplete code set */
1932 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1) {
1938 /* decompress until an end-of-block code */
1939 if (inflate_codes(tl, td, bl, bd))
1943 /* free the decoding tables, return */
1951 int inflate_dynamic()
1952 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
1954 int i; /* temporary variables */
1956 unsigned l; /* last length */
1957 unsigned m; /* mask for bit lengths table */
1958 unsigned n; /* number of lengths to get */
1959 struct huft *tl; /* literal/length code table */
1960 struct huft *td; /* distance code table */
1961 int bl; /* lookup bits for tl */
1962 int bd; /* lookup bits for td */
1963 unsigned nb; /* number of bit length codes */
1964 unsigned nl; /* number of literal/length codes */
1965 unsigned nd; /* number of distance codes */
1967 #ifdef PKZIP_BUG_WORKAROUND
1968 unsigned ll[288 + 32]; /* literal/length and distance code lengths */
1970 unsigned ll[286 + 30]; /* literal/length and distance code lengths */
1972 register ulg b; /* bit buffer */
1973 register unsigned k; /* number of bits in bit buffer */
1976 /* make local bit buffer */
1981 /* read in table lengths */
1983 nl = 257 + ((unsigned) b & 0x1f); /* number of literal/length codes */
1986 nd = 1 + ((unsigned) b & 0x1f); /* number of distance codes */
1989 nb = 4 + ((unsigned) b & 0xf); /* number of bit length codes */
1991 #ifdef PKZIP_BUG_WORKAROUND
1992 if (nl > 288 || nd > 32)
1994 if (nl > 286 || nd > 30)
1996 return 1; /* bad lengths */
1999 /* read in bit-length-code lengths */
2000 for (j = 0; j < nb; j++) {
2002 ll[border[j]] = (unsigned) b & 7;
2009 /* build decoding table for trees--single level, 7 bit lookup */
2011 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0) {
2014 return i; /* incomplete code set */
2018 /* read in literal and distance code lengths */
2022 while ((unsigned) i < n) {
2023 NEEDBITS((unsigned) bl)
2024 j = (td = tl + ((unsigned) b & m))->b;
2027 if (j < 16) /* length of code in bits (0..15) */
2028 ll[i++] = l = j; /* save last length in l */
2029 else if (j == 16) { /* repeat last length 3 to 6 times */
2031 j = 3 + ((unsigned) b & 3);
2033 if ((unsigned) i + j > n)
2037 } else if (j == 17) { /* 3 to 10 zero length codes */
2039 j = 3 + ((unsigned) b & 7);
2041 if ((unsigned) i + j > n)
2046 } else { /* j == 18: 11 to 138 zero length codes */
2049 j = 11 + ((unsigned) b & 0x7f);
2051 if ((unsigned) i + j > n)
2060 /* free decoding table for trees */
2064 /* restore the global bit buffer */
2069 /* build the decoding tables for literal/length and distance codes */
2071 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0) {
2073 fprintf(stderr, " incomplete literal tree\n");
2076 return i; /* incomplete code set */
2079 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0) {
2081 fprintf(stderr, " incomplete distance tree\n");
2082 #ifdef PKZIP_BUG_WORKAROUND
2089 return i; /* incomplete code set */
2094 /* decompress until an end-of-block code */
2095 if (inflate_codes(tl, td, bl, bd))
2099 /* free the decoding tables, return */
2107 int inflate_block(e)
2108 int *e; /* last block flag */
2110 /* decompress an inflated block */
2112 unsigned t; /* block type */
2113 register ulg b; /* bit buffer */
2114 register unsigned k; /* number of bits in bit buffer */
2117 /* make local bit buffer */
2122 /* read in last block bit */
2128 /* read in block type */
2130 t = (unsigned) b & 3;
2134 /* restore the global bit buffer */
2139 /* inflate that block type */
2141 return inflate_dynamic();
2143 return inflate_stored();
2145 return inflate_fixed();
2148 /* bad block type */
2155 /* decompress an inflated entry */
2157 int e; /* last block flag */
2158 int r; /* result code */
2159 unsigned h; /* maximum struct huft's malloc'ed */
2162 /* initialize window, bit buffer */
2168 /* decompress until the last block */
2172 if ((r = inflate_block(&e)) != 0)
2178 /* Undo too much lookahead. The next read will be byte aligned so we
2179 * can discard unused bits in the last meaningful byte.
2186 /* flush out slide */
2190 /* return success */
2192 fprintf(stderr, "<%u> ", h);