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
32 static const char gunzip_usage[] =
33 "gunzip [OPTION]... FILE\n"
34 #ifndef BB_FEATURE_TRIVIAL_HELP
35 "\nUncompress FILE (or standard input if FILE is '-').\n\n"
38 "\t-c\tWrite output to standard output\n"
39 "\t-t\tTest compressed file integrity\n"
44 /* These defines are very important for BusyBox. Without these,
45 * huge chunks of ram are pre-allocated making the BusyBox bss
46 * size Freaking Huge(tm), which is a bad thing.*/
50 #define BB_DECLARE_EXTERN
51 #define bb_need_memory_exhausted
52 #define bb_need_name_too_long
56 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
57 * Copyright (C) 1992-1993 Jean-loup Gailly
58 * The unzip code was written and put in the public domain by Mark Adler.
59 * Portions of the lzw code are derived from the public domain 'compress'
60 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
61 * Ken Turkowski, Dave Mack and Peter Jannesen.
63 * See the license_msg below and the file COPYING for the software license.
64 * See the file algorithm.doc for the compression algorithms and file formats.
68 static char *license_msg[] = {
69 " Copyright (C) 1992-1993 Jean-loup Gailly",
70 " This program is free software; you can redistribute it and/or modify",
71 " it under the terms of the GNU General Public License as published by",
72 " the Free Software Foundation; either version 2, or (at your option)",
73 " any later version.",
75 " This program is distributed in the hope that it will be useful,",
76 " but WITHOUT ANY WARRANTY; without even the implied warranty of",
77 " MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the",
78 " GNU General Public License for more details.",
80 " You should have received a copy of the GNU General Public License",
81 " along with this program; if not, write to the Free Software",
82 " Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.",
87 /* Compress files with zip algorithm and 'compress' interface.
88 * See usage() and help() functions below for all options.
90 * file.gz: compressed file with same mode, owner, and utimes
91 * or stdout with -c option or if stdin used as input.
92 * If the output file name had to be truncated, the original name is kept
93 * in the compressed file.
94 * On MSDOS, file.tmp -> file.tmz. On VMS, file.tmp -> file.tmp-gz.
96 * Using gz on MSDOS would create too many file name conflicts. For
97 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
98 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
99 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
100 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
102 * For the meaning of all compilation flags, see comments in Makefile.in.
106 #include <sys/types.h>
108 #include <sys/stat.h>
111 /* #include "tailor.h" */
113 /* tailor.h -- target dependent definitions
114 * Copyright (C) 1992-1993 Jean-loup Gailly.
115 * This is free software; you can redistribute it and/or modify it under the
116 * terms of the GNU General Public License, see the file COPYING.
119 /* The target dependent definitions should be defined here only.
120 * The target dependent functions should be defined in tailor.c.
125 #define get_char() get_byte()
126 #define put_char(c) put_byte(c)
129 /* I don't like nested includes, but the string and io functions are used
133 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
135 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
138 # define memzero(s, n) memset ((void *)(s), 0, (n))
140 # include <strings.h>
141 # define strchr index
142 # define strrchr rindex
143 # define memcpy(d, s, n) bcopy((s), (d), (n))
144 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
145 # define memzero(s, n) bzero((s), (n))
149 # define RETSIGTYPE void
154 typedef unsigned char uch;
155 typedef unsigned short ush;
156 typedef unsigned long ulg;
158 /* Return codes from gzip */
163 /* Compression methods (see algorithm.doc) */
166 extern int method; /* compression method */
168 /* To save memory for 16 bit systems, some arrays are overlaid between
169 * the various modules:
170 * deflate: prev+head window d_buf l_buf outbuf
171 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
172 * inflate: window inbuf
173 * unpack: window inbuf prefix_len
174 * unlzh: left+right window c_table inbuf c_len
175 * For compression, input is done in window[]. For decompression, output
176 * is done in window except for unlzw.
181 # define INBUFSIZ 0x2000 /* input buffer size */
183 # define INBUFSIZ 0x8000 /* input buffer size */
186 #define INBUF_EXTRA 64 /* required by unlzw() */
190 # define OUTBUFSIZ 8192 /* output buffer size */
192 # define OUTBUFSIZ 16384 /* output buffer size */
195 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
201 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
203 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
207 /*#define DYN_ALLOC*/
210 # define EXTERN(type, array) extern type * array
211 # define DECLARE(type, array, size) type * array
212 # define ALLOC(type, array, size) { \
213 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
214 if (array == NULL) errorMsg(memory_exhausted, "gunzip"); \
216 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
218 # define EXTERN(type, array) extern type array[]
219 # define DECLARE(type, array, size) type array[size]
220 # define ALLOC(type, array, size)
224 EXTERN(uch, inbuf); /* input buffer */
225 EXTERN(uch, outbuf); /* output buffer */
226 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
227 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
228 #define tab_suffix window
230 # define tab_prefix prev /* hash link (see deflate.c) */
231 # define head (prev+WSIZE) /* hash head (see deflate.c) */
232 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
234 # define tab_prefix0 prev
235 # define head tab_prefix1
236 EXTERN(ush, tab_prefix0); /* prefix for even codes */
237 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
240 extern unsigned insize; /* valid bytes in inbuf */
241 extern unsigned inptr; /* index of next byte to be processed in inbuf */
242 extern unsigned outcnt; /* bytes in output buffer */
244 extern long bytes_in; /* number of input bytes */
245 extern long bytes_out; /* number of output bytes */
246 extern long header_bytes; /* number of bytes in gzip header */
248 extern long ifile_size; /* input file size, -1 for devices (debug only) */
250 typedef int file_t; /* Do not use stdio */
252 #define NO_FILE (-1) /* in memory compression */
255 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
258 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
259 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
260 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
261 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
262 #define COMMENT 0x10 /* bit 4 set: file comment present */
263 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
264 #define RESERVED 0xC0 /* bit 6,7: reserved */
267 # define WSIZE 0x8000 /* window size--must be a power of two, and */
268 #endif /* at least 32K for zip's deflate method */
271 #define MAX_MATCH 258
272 /* The minimum and maximum match lengths */
274 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
275 /* Minimum amount of lookahead, except at the end of the input file.
276 * See deflate.c for comments about the MIN_MATCH+1.
279 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
280 /* In order to simplify the code, particularly on 16 bit machines, match
281 * distances are limited to MAX_DIST instead of WSIZE.
284 extern int exit_code; /* program exit code */
285 extern int verbose; /* be verbose (-v) */
286 extern int level; /* compression level */
287 extern int test; /* check .z file integrity */
288 extern int save_orig_name; /* set if original name must be saved */
290 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
291 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
293 /* put_byte is used for the compressed output, put_ubyte for the
294 * uncompressed output. However unlzw() uses window for its
295 * suffix table instead of its output buffer, so it does not use put_ubyte
296 * (to be cleaned up).
298 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
300 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
303 /* Output a 16 bit value, lsb first */
304 #define put_short(w) \
305 { if (outcnt < OUTBUFSIZ-2) { \
306 outbuf[outcnt++] = (uch) ((w) & 0xff); \
307 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
309 put_byte((uch)((w) & 0xff)); \
310 put_byte((uch)((ush)(w) >> 8)); \
314 /* Output a 32 bit value to the bit stream, lsb first */
315 #define put_long(n) { \
316 put_short((n) & 0xffff); \
317 put_short(((ulg)(n)) >> 16); \
320 #define seekable() 0 /* force sequential output */
321 #define translate_eol 0 /* no option -a yet */
323 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
325 /* Macros for getting two-byte and four-byte header values */
326 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
327 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
329 /* Diagnostic functions */
331 # define Assert(cond,msg) {if(!(cond)) errorMsg(msg);}
332 # define Trace(x) fprintf x
333 # define Tracev(x) {if (verbose) fprintf x ;}
334 # define Tracevv(x) {if (verbose>1) fprintf x ;}
335 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
336 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
338 # define Assert(cond,msg)
343 # define Tracecv(c,x)
346 #define WARN(msg) {fprintf msg ; \
347 if (exit_code == OK) exit_code = WARNING;}
350 extern int unzip (int in, int out);
353 RETSIGTYPE abort_gzip (void);
356 void lm_init (int pack_level, ush * flags);
360 void ct_init (ush * attr, int *method);
361 int ct_tally (int dist, int lc);
362 ulg flush_block (char *buf, ulg stored_len, int eof);
365 void bi_init (file_t zipfile);
366 void send_bits (int value, int length);
367 unsigned bi_reverse (unsigned value, int length);
368 void bi_windup (void);
369 void copy_block (char *buf, unsigned len, int header);
372 extern ulg updcrc (uch * s, unsigned n);
373 extern void clear_bufs (void);
374 static int fill_inbuf (int eof_ok);
375 extern void flush_outbuf (void);
376 static void flush_window (void);
377 extern void write_buf (int fd, void * buf, unsigned cnt);
380 static char *basename (char *fname);
381 #endif /* not __linux__ */
382 void read_error_msg (void);
383 void write_error_msg (void);
386 static int inflate (void);
388 /* #include "lzw.h" */
390 /* lzw.h -- define the lzw functions.
391 * Copyright (C) 1992-1993 Jean-loup Gailly.
392 * This is free software; you can redistribute it and/or modify it under the
393 * terms of the GNU General Public License, see the file COPYING.
396 #if !defined(OF) && defined(lint)
403 #define INIT_BITS 9 /* Initial number of bits per code */
405 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
407 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
408 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
409 * It's a pity that old uncompress does not check bit 0x20. That makes
410 * extension of the format actually undesirable because old compress
411 * would just crash on the new format instead of giving a meaningful
412 * error message. It does check the number of bits, but it's more
413 * helpful to say "unsupported format, get a new version" than
414 * "can only handle 16 bits".
417 #define BLOCK_MODE 0x80
418 /* Block compression: if table is full and compression rate is dropping,
419 * clear the dictionary.
422 #define LZW_RESERVED 0x60 /* reserved bits */
424 #define CLEAR 256 /* flush the dictionary */
425 #define FIRST (CLEAR+1) /* first free entry */
427 extern int maxbits; /* max bits per code for LZW */
428 extern int block_mode; /* block compress mode -C compatible with 2.0 */
430 extern int lzw (int in, int out);
431 extern int unlzw (int in, int out);
434 /* #include "revision.h" */
436 /* revision.h -- define the version number
437 * Copyright (C) 1992-1993 Jean-loup Gailly.
438 * This is free software; you can redistribute it and/or modify it under the
439 * terms of the GNU General Public License, see the file COPYING.
442 #define VERSION "1.2.4"
444 #define REVDATE "18 Aug 93"
446 /* This version does not support compression into old compress format: */
457 typedef struct dirent dir_type;
459 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
460 # define DIR_OPT "DIRENT"
462 # define NLENGTH(dirent) ((dirent)->d_namlen)
464 # include <sys/dir.h>
465 typedef struct direct dir_type;
467 # define DIR_OPT "SYSDIR"
470 # include <sys/ndir.h>
471 typedef struct direct dir_type;
473 # define DIR_OPT "SYSNDIR"
477 typedef struct direct dir_type;
479 # define DIR_OPT "NDIR"
482 # define DIR_OPT "NO_DIR"
487 #if !defined(S_ISDIR) && defined(S_IFDIR)
488 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
490 #if !defined(S_ISREG) && defined(S_IFREG)
491 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
493 typedef RETSIGTYPE(*sig_type) (int);
496 # define O_BINARY 0 /* creation mode for open() */
500 /* Pure BSD system? */
501 # include <sys/file.h>
503 # define O_CREAT FCREAT
506 # define O_EXCL FEXCL
511 # define S_IRUSR 0400
514 # define S_IWUSR 0200
516 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
518 #ifndef MAX_PATH_LEN /* max pathname length */
520 # define MAX_PATH_LEN BUFSIZ
522 # define MAX_PATH_LEN 1024
532 off_t lseek (int fd, off_t offset, int whence);
538 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
539 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
540 DECLARE(ush, d_buf, DIST_BUFSIZE);
541 DECLARE(uch, window, 2L * WSIZE);
543 DECLARE(ush, tab_prefix, 1L << BITS);
545 DECLARE(ush, tab_prefix0, 1L << (BITS - 1));
546 DECLARE(ush, tab_prefix1, 1L << (BITS - 1));
549 /* local variables */
551 int test_mode = 0; /* check file integrity option */
552 int foreground; /* set if program run in foreground */
553 int maxbits = BITS; /* max bits per code for LZW */
554 int method = DEFLATED; /* compression method */
555 int exit_code = OK; /* program exit code */
556 int last_member; /* set for .zip and .Z files */
557 int part_nb; /* number of parts in .gz file */
558 long ifile_size; /* input file size, -1 for devices (debug only) */
560 long bytes_in; /* number of input bytes */
561 long bytes_out; /* number of output bytes */
562 long total_in = 0; /* input bytes for all files */
563 long total_out = 0; /* output bytes for all files */
564 struct stat istat; /* status for input file */
565 int ifd; /* input file descriptor */
566 int ofd; /* output file descriptor */
567 unsigned insize; /* valid bytes in inbuf */
568 unsigned inptr; /* index of next byte to be processed in inbuf */
569 unsigned outcnt; /* bytes in output buffer */
571 long header_bytes; /* number of bytes in gzip header */
573 /* local functions */
575 local int get_method (int in);
577 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
579 /* ======================================================================== */
580 int gunzip_main(int argc, char **argv)
582 int file_count; /* number of files to precess */
588 int delInputFile = 0;
591 char ifname[MAX_PATH_LEN + 1]; /* input file name */
592 char ofname[MAX_PATH_LEN + 1]; /* output file name */
594 if (strcmp(*argv, "zcat") == 0) {
601 /* Parse any options */
602 while (--argc > 0 && **(++argv) == '-') {
603 if (*((*argv) + 1) == '\0') {
607 while (*(++(*argv))) {
622 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
624 (void) signal(SIGINT, (sig_type) abort_gzip);
627 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
628 (void) signal(SIGTERM, (sig_type) abort_gzip);
632 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
633 (void) signal(SIGHUP, (sig_type) abort_gzip);
637 file_count = argc - optind;
639 /* Allocate all global buffers (for DYN_ALLOC option) */
640 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
641 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
642 ALLOC(ush, d_buf, DIST_BUFSIZE);
643 ALLOC(uch, window, 2L * WSIZE);
645 ALLOC(ush, tab_prefix, 1L << BITS);
647 ALLOC(ush, tab_prefix0, 1L << (BITS - 1));
648 ALLOC(ush, tab_prefix1, 1L << (BITS - 1));
651 if (fromstdin == 1) {
652 strcpy(ofname, "stdin");
654 inFileNum = fileno(stdin);
655 ifile_size = -1L; /* convention for unknown size */
657 /* Open up the input file */
660 if (strlen(*argv) > MAX_PATH_LEN) {
661 fprintf(stderr, name_too_long, "gunzip");
664 strcpy(ifname, *argv);
666 /* Open input fille */
667 inFileNum = open(ifname, O_RDONLY);
672 /* Get the time stamp on the input file. */
673 result = stat(ifname, &statBuf);
678 ifile_size = statBuf.st_size;
681 if (to_stdout == 1) {
682 /* And get to work */
683 strcpy(ofname, "stdout");
684 outFileNum = fileno(stdout);
686 clear_bufs(); /* clear input and output buffers */
689 /* Actually do the compression/decompression. */
690 unzip(inFileNum, outFileNum);
692 } else if (test_mode) {
693 /* Actually do the compression/decompression. */
698 /* And get to work */
699 if (strlen(ifname) > MAX_PATH_LEN - 4) {
700 fprintf(stderr, name_too_long, "gunzip");
703 strcpy(ofname, ifname);
704 pos = strstr(ofname, ".gz");
709 pos = strstr(ofname, ".tgz");
717 /* Open output fille */
718 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
719 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
721 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL);
723 if (outFileNum < 0) {
727 /* Set permissions on the file */
728 fchmod(outFileNum, statBuf.st_mode);
730 clear_bufs(); /* clear input and output buffers */
733 /* Actually do the compression/decompression. */
734 result = unzip(inFileNum, outFileNum);
738 /* Delete the original file */
740 delFileName = ifname;
742 delFileName = ofname;
744 if (delInputFile == 1 && unlink(delFileName) < 0) {
753 /* ========================================================================
754 * Check the magic number of the input file and update ofname if an
755 * original name was given and to_stdout is not set.
756 * Return the compression method, -1 for error, -2 for warning.
757 * Set inptr to the offset of the next byte to be processed.
758 * Updates time_stamp if there is one and --no-time is not used.
759 * This function may be called repeatedly for an input file consisting
760 * of several contiguous gzip'ed members.
761 * IN assertions: there is at least one remaining compressed member.
762 * If the member is a zip file, it must be the only one.
764 local int get_method(in)
765 int in; /* input file descriptor */
767 uch flags; /* compression flags */
768 char magic[2]; /* magic header */
770 magic[0] = (char) get_byte();
771 magic[1] = (char) get_byte();
772 method = -1; /* unknown yet */
773 part_nb++; /* number of parts in gzip file */
775 last_member = RECORD_IO;
776 /* assume multiple members in gzip file except for record oriented I/O */
778 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
780 method = (int) get_byte();
781 if (method != DEFLATED) {
783 "unknown method %d -- get newer version of gzip\n",
788 flags = (uch) get_byte();
790 (ulg) get_byte(); /* Ignore time stamp */
795 (void) get_byte(); /* Ignore extra flags for the moment */
796 (void) get_byte(); /* Ignore OS type for the moment */
798 if ((flags & EXTRA_FIELD) != 0) {
799 unsigned len = (unsigned) get_byte();
801 len |= ((unsigned) get_byte()) << 8;
807 /* Discard original name if any */
808 if ((flags & ORIG_NAME) != 0) {
809 while (get_char() != 0) /* null */
813 /* Discard file comment if any */
814 if ((flags & COMMENT) != 0) {
815 while (get_char() != 0) /* null */
819 header_bytes = inptr + 2 * sizeof(long); /* include crc and size */
828 fprintf(stderr, "\nnot in gzip format\n");
832 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
837 /* ========================================================================
838 * Signal and error handler.
840 RETSIGTYPE abort_gzip()
845 /* unzip.c -- decompress files in gzip or pkzip format.
846 * Copyright (C) 1992-1993 Jean-loup Gailly
847 * This is free software; you can redistribute it and/or modify it under the
848 * terms of the GNU General Public License, see the file COPYING.
850 * The code in this file is derived from the file funzip.c written
851 * and put in the public domain by Mark Adler.
855 This version can extract files in gzip or pkzip format.
856 For the latter, only the first entry is extracted, and it has to be
857 either deflated or stored.
860 /* #include "crypt.h" */
862 /* crypt.h (dummy version) -- do not perform encryption
863 * Hardly worth copyrighting :-)
867 # undef CRYPT /* dummy version */
870 #define RAND_HEAD_LEN 12 /* length of encryption random header */
875 /* PKZIP header definitions */
876 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
877 #define LOCFLG 6 /* offset of bit flag */
878 #define CRPFLG 1 /* bit for encrypted entry */
879 #define EXTFLG 8 /* bit for extended local header */
880 #define LOCHOW 8 /* offset of compression method */
881 #define LOCTIM 10 /* file mod time (for decryption) */
882 #define LOCCRC 14 /* offset of crc */
883 #define LOCSIZ 18 /* offset of compressed size */
884 #define LOCLEN 22 /* offset of uncompressed length */
885 #define LOCFIL 26 /* offset of file name field length */
886 #define LOCEXT 28 /* offset of extra field length */
887 #define LOCHDR 30 /* size of local header, including sig */
888 #define EXTHDR 16 /* size of extended local header, inc sig */
893 char *key; /* not used--needed to link crypt.c */
894 int pkzip = 0; /* set for a pkzip file */
895 int ext_header = 0; /* set if extended local header */
897 /* ===========================================================================
898 * Unzip in to out. This routine works on both gzip and pkzip files.
900 * IN assertions: the buffer inbuf contains already the beginning of
901 * the compressed data, from offsets inptr to insize-1 included.
902 * The magic header has already been checked. The output buffer is cleared.
905 int in, out; /* input and output file descriptors */
907 ulg orig_crc = 0; /* original crc */
908 ulg orig_len = 0; /* original uncompressed length */
910 uch buf[EXTHDR]; /* extended local header */
914 method = get_method(ifd);
916 exit(exit_code); /* error message already emitted */
919 updcrc(NULL, 0); /* initialize crc */
921 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
922 orig_crc = LG(inbuf + LOCCRC);
923 orig_len = LG(inbuf + LOCLEN);
927 if (method == DEFLATED) {
932 errorMsg(memory_exhausted, "gunzip");
933 } else if (res != 0) {
934 errorMsg("invalid compressed data--format violated");
938 errorMsg("internal error, invalid method");
941 /* Get the crc and original length */
943 /* crc32 (see algorithm.doc)
944 * uncompressed input size modulo 2^32
946 for (n = 0; n < 8; n++) {
947 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
950 orig_len = LG(buf + 4);
952 } else if (ext_header) { /* If extended header, check it */
953 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
955 * compressed size 4-bytes
956 * uncompressed size 4-bytes
958 for (n = 0; n < EXTHDR; n++) {
959 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
961 orig_crc = LG(buf + 4);
962 orig_len = LG(buf + 12);
965 /* Validate decompression */
966 if (orig_crc != updcrc(outbuf, 0)) {
967 errorMsg("invalid compressed data--crc error");
969 if (orig_len != (ulg) bytes_out) {
970 errorMsg("invalid compressed data--length error");
973 /* Check if there are more entries in a pkzip file */
974 if (pkzip && inptr + 4 < insize && LG(inbuf + inptr) == LOCSIG) {
975 WARN((stderr, "has more than one entry--rest ignored\n"));
977 ext_header = pkzip = 0; /* for next file */
981 /* util.c -- utility functions for gzip support
982 * Copyright (C) 1992-1993 Jean-loup Gailly
983 * This is free software; you can redistribute it and/or modify it under the
984 * terms of the GNU General Public License, see the file COPYING.
989 #include <sys/types.h>
998 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1004 static const ulg crc_32_tab[]; /* crc table, defined below */
1006 /* ===========================================================================
1007 * Run a set of bytes through the crc shift register. If s is a NULL
1008 * pointer, then initialize the crc shift register contents instead.
1009 * Return the current crc in either case.
1012 uch *s; /* pointer to bytes to pump through */
1013 unsigned n; /* number of bytes in s[] */
1015 register ulg c; /* temporary variable */
1017 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
1025 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
1029 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1032 /* ===========================================================================
1033 * Clear input and output buffers
1035 void clear_bufs(void)
1039 bytes_in = bytes_out = 0L;
1042 /* ===========================================================================
1043 * Fill the input buffer. This is called only when the buffer is empty.
1045 int fill_inbuf(eof_ok)
1046 int eof_ok; /* set if EOF acceptable as a result */
1050 /* Read as much as possible */
1054 len = read(ifd, (char *) inbuf + insize, INBUFSIZ - insize);
1055 if (len == 0 || len == EOF)
1058 } while (insize < INBUFSIZ);
1065 bytes_in += (ulg) insize;
1070 /* ===========================================================================
1071 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1072 * (used for the compressed data only)
1080 write_buf(ofd, (char *) outbuf, outcnt);
1081 bytes_out += (ulg) outcnt;
1085 /* ===========================================================================
1086 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1087 * (Used for the decompressed data only.)
1093 updcrc(window, outcnt);
1096 write_buf(ofd, (char *) window, outcnt);
1097 bytes_out += (ulg) outcnt;
1101 /* ===========================================================================
1102 * Does the same as write(), but also handles partial pipe writes and checks
1105 void write_buf(fd, buf, cnt)
1112 while ((n = write(fd, buf, cnt)) != cnt) {
1113 if (n == (unsigned) (-1)) {
1117 buf = (void *) ((char *) buf + n);
1121 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1123 /* Provide missing strspn and strcspn functions. */
1129 int strspn (const char *s, const char *accept);
1130 int strcspn (const char *s, const char *reject);
1132 /* ========================================================================
1133 * Return the length of the maximum initial segment
1134 * of s which contains only characters in accept.
1136 int strspn(s, accept)
1140 register const char *p;
1141 register const char *a;
1142 register int count = 0;
1144 for (p = s; *p != '\0'; ++p) {
1145 for (a = accept; *a != '\0'; ++a) {
1156 /* ========================================================================
1157 * Return the length of the maximum inital segment of s
1158 * which contains no characters from reject.
1160 int strcspn(s, reject)
1164 register int count = 0;
1166 while (*s != '\0') {
1167 if (strchr(reject, *s++) != NULL)
1174 #endif /* NO_STRING_H */
1177 /* ========================================================================
1180 void read_error_msg()
1182 fprintf(stderr, "\n");
1186 fprintf(stderr, "unexpected end of file\n");
1191 void write_error_msg()
1193 fprintf(stderr, "\n");
1199 /* ========================================================================
1200 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1202 static const ulg crc_32_tab[] = {
1203 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1204 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1205 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1206 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1207 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1208 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1209 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1210 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1211 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1212 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1213 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1214 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1215 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1216 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1217 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1218 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1219 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1220 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1221 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1222 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1223 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1224 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1225 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1226 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1227 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1228 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1229 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1230 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1231 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1232 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1233 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1234 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1235 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1236 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1237 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1238 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1239 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1240 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1241 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1242 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1243 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1244 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1245 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1246 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1247 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1248 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1249 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1250 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1251 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1252 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1253 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1257 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1258 version c10p1, 10 January 1993 */
1260 /* You can do whatever you like with this source file, though I would
1261 prefer that if you modify it and redistribute it that you include
1262 comments to that effect with your name and the date. Thank you.
1263 [The history has been moved to the file ChangeLog.]
1267 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1268 method searches for as much of the current string of bytes (up to a
1269 length of 258) in the previous 32K bytes. If it doesn't find any
1270 matches (of at least length 3), it codes the next byte. Otherwise, it
1271 codes the length of the matched string and its distance backwards from
1272 the current position. There is a single Huffman code that codes both
1273 single bytes (called "literals") and match lengths. A second Huffman
1274 code codes the distance information, which follows a length code. Each
1275 length or distance code actually represents a base value and a number
1276 of "extra" (sometimes zero) bits to get to add to the base value. At
1277 the end of each deflated block is a special end-of-block (EOB) literal/
1278 length code. The decoding process is basically: get a literal/length
1279 code; if EOB then done; if a literal, emit the decoded byte; if a
1280 length then get the distance and emit the referred-to bytes from the
1281 sliding window of previously emitted data.
1283 There are (currently) three kinds of inflate blocks: stored, fixed, and
1284 dynamic. The compressor deals with some chunk of data at a time, and
1285 decides which method to use on a chunk-by-chunk basis. A chunk might
1286 typically be 32K or 64K. If the chunk is uncompressible, then the
1287 "stored" method is used. In this case, the bytes are simply stored as
1288 is, eight bits per byte, with none of the above coding. The bytes are
1289 preceded by a count, since there is no longer an EOB code.
1291 If the data is compressible, then either the fixed or dynamic methods
1292 are used. In the dynamic method, the compressed data is preceded by
1293 an encoding of the literal/length and distance Huffman codes that are
1294 to be used to decode this block. The representation is itself Huffman
1295 coded, and so is preceded by a description of that code. These code
1296 descriptions take up a little space, and so for small blocks, there is
1297 a predefined set of codes, called the fixed codes. The fixed method is
1298 used if the block codes up smaller that way (usually for quite small
1299 chunks), otherwise the dynamic method is used. In the latter case, the
1300 codes are customized to the probabilities in the current block, and so
1301 can code it much better than the pre-determined fixed codes.
1303 The Huffman codes themselves are decoded using a mutli-level table
1304 lookup, in order to maximize the speed of decoding plus the speed of
1305 building the decoding tables. See the comments below that precede the
1306 lbits and dbits tuning parameters.
1311 Notes beyond the 1.93a appnote.txt:
1313 1. Distance pointers never point before the beginning of the output
1315 2. Distance pointers can point back across blocks, up to 32k away.
1316 3. There is an implied maximum of 7 bits for the bit length table and
1317 15 bits for the actual data.
1318 4. If only one code exists, then it is encoded using one bit. (Zero
1319 would be more efficient, but perhaps a little confusing.) If two
1320 codes exist, they are coded using one bit each (0 and 1).
1321 5. There is no way of sending zero distance codes--a dummy must be
1322 sent if there are none. (History: a pre 2.0 version of PKZIP would
1323 store blocks with no distance codes, but this was discovered to be
1324 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1325 zero distance codes, which is sent as one code of zero bits in
1327 6. There are up to 286 literal/length codes. Code 256 represents the
1328 end-of-block. Note however that the static length tree defines
1329 288 codes just to fill out the Huffman codes. Codes 286 and 287
1330 cannot be used though, since there is no length base or extra bits
1331 defined for them. Similarly, there are up to 30 distance codes.
1332 However, static trees define 32 codes (all 5 bits) to fill out the
1333 Huffman codes, but the last two had better not show up in the data.
1334 7. Unzip can check dynamic Huffman blocks for complete code sets.
1335 The exception is that a single code would not be complete (see #4).
1336 8. The five bits following the block type is really the number of
1337 literal codes sent minus 257.
1338 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1339 (1+6+6). Therefore, to output three times the length, you output
1340 three codes (1+1+1), whereas to output four times the same length,
1341 you only need two codes (1+3). Hmm.
1342 10. In the tree reconstruction algorithm, Code = Code + Increment
1343 only if BitLength(i) is not zero. (Pretty obvious.)
1344 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1345 12. Note: length code 284 can represent 227-258, but length code 285
1346 really is 258. The last length deserves its own, short code
1347 since it gets used a lot in very redundant files. The length
1348 258 is special since 258 - 3 (the min match length) is 255.
1349 13. The literal/length and distance code bit lengths are read as a
1350 single stream of lengths. It is possible (and advantageous) for
1351 a repeat code (16, 17, or 18) to go across the boundary between
1352 the two sets of lengths.
1355 #include <sys/types.h>
1357 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1358 # include <stdlib.h>
1362 #define slide window
1364 /* Huffman code lookup table entry--this entry is four bytes for machines
1365 that have 16-bit pointers (e.g. PC's in the small or medium model).
1366 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1367 means that v is a literal, 16 < e < 32 means that v is a pointer to
1368 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1369 an unused code. If a code with e == 99 is looked up, this implies an
1370 error in the data. */
1372 uch e; /* number of extra bits or operation */
1373 uch b; /* number of bits in this code or subcode */
1375 ush n; /* literal, length base, or distance base */
1376 struct huft *t; /* pointer to next level of table */
1381 /* Function prototypes */
1382 int huft_build (unsigned *, unsigned, unsigned, ush *, ush *,
1383 struct huft **, int *);
1384 int huft_free (struct huft *);
1385 int inflate_codes (struct huft *, struct huft *, int, int);
1386 int inflate_stored (void);
1387 int inflate_fixed (void);
1388 int inflate_dynamic (void);
1389 int inflate_block (int *);
1393 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1394 stream to find repeated byte strings. This is implemented here as a
1395 circular buffer. The index is updated simply by incrementing and then
1396 and'ing with 0x7fff (32K-1). */
1397 /* It is left to other modules to supply the 32K area. It is assumed
1398 to be usable as if it were declared "uch slide[32768];" or as just
1399 "uch *slide;" and then malloc'ed in the latter case. The definition
1400 must be in unzip.h, included above. */
1401 /* unsigned wp; current position in slide */
1403 #define flush_output(w) (wp=(w),flush_window())
1405 /* Tables for deflate from PKZIP's appnote.txt. */
1406 static unsigned border[] = { /* Order of the bit length code lengths */
1407 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15
1409 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1410 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1411 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0
1414 /* note: see note #13 above about the 258 in this list. */
1415 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1416 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1417 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99
1418 }; /* 99==invalid */
1419 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1420 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1421 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1422 8193, 12289, 16385, 24577
1424 static ush cpdext[] = { /* Extra bits for distance codes */
1425 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1426 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1432 /* Macros for inflate() bit peeking and grabbing.
1436 x = b & mask_bits[j];
1439 where NEEDBITS makes sure that b has at least j bits in it, and
1440 DUMPBITS removes the bits from b. The macros use the variable k
1441 for the number of bits in b. Normally, b and k are register
1442 variables for speed, and are initialized at the beginning of a
1443 routine that uses these macros from a global bit buffer and count.
1445 If we assume that EOB will be the longest code, then we will never
1446 ask for bits with NEEDBITS that are beyond the end of the stream.
1447 So, NEEDBITS should not read any more bytes than are needed to
1448 meet the request. Then no bytes need to be "returned" to the buffer
1449 at the end of the last block.
1451 However, this assumption is not true for fixed blocks--the EOB code
1452 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1453 (The EOB code is shorter than other codes because fixed blocks are
1454 generally short. So, while a block always has an EOB, many other
1455 literal/length codes have a significantly lower probability of
1456 showing up at all.) However, by making the first table have a
1457 lookup of seven bits, the EOB code will be found in that first
1458 lookup, and so will not require that too many bits be pulled from
1462 ulg bb; /* bit buffer */
1463 unsigned bk; /* bits in bit buffer */
1467 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1468 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1474 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1476 # define NEXTBYTE() (uch)get_byte()
1478 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1479 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1483 Huffman code decoding is performed using a multi-level table lookup.
1484 The fastest way to decode is to simply build a lookup table whose
1485 size is determined by the longest code. However, the time it takes
1486 to build this table can also be a factor if the data being decoded
1487 is not very long. The most common codes are necessarily the
1488 shortest codes, so those codes dominate the decoding time, and hence
1489 the speed. The idea is you can have a shorter table that decodes the
1490 shorter, more probable codes, and then point to subsidiary tables for
1491 the longer codes. The time it costs to decode the longer codes is
1492 then traded against the time it takes to make longer tables.
1494 This results of this trade are in the variables lbits and dbits
1495 below. lbits is the number of bits the first level table for literal/
1496 length codes can decode in one step, and dbits is the same thing for
1497 the distance codes. Subsequent tables are also less than or equal to
1498 those sizes. These values may be adjusted either when all of the
1499 codes are shorter than that, in which case the longest code length in
1500 bits is used, or when the shortest code is *longer* than the requested
1501 table size, in which case the length of the shortest code in bits is
1504 There are two different values for the two tables, since they code a
1505 different number of possibilities each. The literal/length table
1506 codes 286 possible values, or in a flat code, a little over eight
1507 bits. The distance table codes 30 possible values, or a little less
1508 than five bits, flat. The optimum values for speed end up being
1509 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1510 The optimum values may differ though from machine to machine, and
1511 possibly even between compilers. Your mileage may vary.
1515 int lbits = 9; /* bits in base literal/length lookup table */
1516 int dbits = 6; /* bits in base distance lookup table */
1519 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1520 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1521 #define N_MAX 288 /* maximum number of codes in any set */
1524 unsigned hufts; /* track memory usage */
1527 int huft_build(b, n, s, d, e, t, m)
1528 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1529 unsigned n; /* number of codes (assumed <= N_MAX) */
1530 unsigned s; /* number of simple-valued codes (0..s-1) */
1531 ush *d; /* list of base values for non-simple codes */
1532 ush *e; /* list of extra bits for non-simple codes */
1533 struct huft **t; /* result: starting table */
1534 int *m; /* maximum lookup bits, returns actual */
1536 /* Given a list of code lengths and a maximum table size, make a set of
1537 tables to decode that set of codes. Return zero on success, one if
1538 the given code set is incomplete (the tables are still built in this
1539 case), two if the input is invalid (all zero length codes or an
1540 oversubscribed set of lengths), and three if not enough memory. */
1542 unsigned a; /* counter for codes of length k */
1543 unsigned c[BMAX + 1]; /* bit length count table */
1544 unsigned f; /* i repeats in table every f entries */
1545 int g; /* maximum code length */
1546 int h; /* table level */
1547 register unsigned i; /* counter, current code */
1548 register unsigned j; /* counter */
1549 register int k; /* number of bits in current code */
1550 int l; /* bits per table (returned in m) */
1551 register unsigned *p; /* pointer into c[], b[], or v[] */
1552 register struct huft *q; /* points to current table */
1553 struct huft r; /* table entry for structure assignment */
1554 struct huft *u[BMAX]; /* table stack */
1555 unsigned v[N_MAX]; /* values in order of bit length */
1556 register int w; /* bits before this table == (l * h) */
1557 unsigned x[BMAX + 1]; /* bit offsets, then code stack */
1558 unsigned *xp; /* pointer into x */
1559 int y; /* number of dummy codes added */
1560 unsigned z; /* number of entries in current table */
1563 /* Generate counts for each bit length */
1564 memzero(c, sizeof(c));
1571 && n - i <= '~' ? "%c %d\n" : "0x%x %d\n"), n - i, *p));
1572 c[*p]++; /* assume all entries <= BMAX */
1573 p++; /* Can't combine with above line (Solaris bug) */
1575 if (c[0] == n) { /* null input--all zero length codes */
1576 *t = (struct huft *) NULL;
1582 /* Find minimum and maximum length, bound *m by those */
1584 for (j = 1; j <= BMAX; j++)
1587 k = j; /* minimum code length */
1588 if ((unsigned) l < j)
1590 for (i = BMAX; i; i--)
1593 g = i; /* maximum code length */
1594 if ((unsigned) l > i)
1599 /* Adjust last length count to fill out codes, if needed */
1600 for (y = 1 << j; j < i; j++, y <<= 1)
1601 if ((y -= c[j]) < 0)
1602 return 2; /* bad input: more codes than bits */
1603 if ((y -= c[i]) < 0)
1608 /* Generate starting offsets into the value table for each length */
1612 while (--i) { /* note that i == g from above */
1613 *xp++ = (j += *p++);
1617 /* Make a table of values in order of bit lengths */
1621 if ((j = *p++) != 0)
1626 /* Generate the Huffman codes and for each, make the table entries */
1627 x[0] = i = 0; /* first Huffman code is zero */
1628 p = v; /* grab values in bit order */
1629 h = -1; /* no tables yet--level -1 */
1630 w = -l; /* bits decoded == (l * h) */
1631 u[0] = (struct huft *) NULL; /* just to keep compilers happy */
1632 q = (struct huft *) NULL; /* ditto */
1635 /* go through the bit lengths (k already is bits in shortest code) */
1636 for (; k <= g; k++) {
1639 /* here i is the Huffman code of length k bits for value *p */
1640 /* make tables up to required level */
1643 w += l; /* previous table always l bits */
1645 /* compute minimum size table less than or equal to l bits */
1646 z = (z = g - w) > (unsigned) l ? l : z; /* upper limit on table size */
1647 if ((f = 1 << (j = k - w)) > a + 1) { /* try a k-w bit table *//* too few codes for k-w bit table */
1648 f -= a + 1; /* deduct codes from patterns left */
1650 while (++j < z) { /* try smaller tables up to z bits */
1651 if ((f <<= 1) <= *++xp)
1652 break; /* enough codes to use up j bits */
1653 f -= *xp; /* else deduct codes from patterns */
1656 z = 1 << j; /* table entries for j-bit table */
1658 /* allocate and link in new table */
1661 (struct huft *) malloc((z + 1) *
1662 sizeof(struct huft))) ==
1663 (struct huft *) NULL) {
1666 return 3; /* not enough memory */
1668 hufts += z + 1; /* track memory usage */
1669 *t = q + 1; /* link to list for huft_free() */
1670 *(t = &(q->v.t)) = (struct huft *) NULL;
1671 u[h] = ++q; /* table starts after link */
1673 /* connect to last table, if there is one */
1675 x[h] = i; /* save pattern for backing up */
1676 r.b = (uch) l; /* bits to dump before this table */
1677 r.e = (uch) (16 + j); /* bits in this table */
1678 r.v.t = q; /* pointer to this table */
1679 j = i >> (w - l); /* (get around Turbo C bug) */
1680 u[h - 1][j] = r; /* connect to last table */
1684 /* set up table entry in r */
1685 r.b = (uch) (k - w);
1687 r.e = 99; /* out of values--invalid code */
1689 r.e = (uch) (*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1690 r.v.n = (ush) (*p); /* simple code is just the value */
1691 p++; /* one compiler does not like *p++ */
1693 r.e = (uch) e[*p - s]; /* non-simple--look up in lists */
1694 r.v.n = d[*p++ - s];
1697 /* fill code-like entries with r */
1699 for (j = i >> w; j < z; j += f)
1702 /* backwards increment the k-bit code i */
1703 for (j = 1 << (k - 1); i & j; j >>= 1)
1707 /* backup over finished tables */
1708 while ((i & ((1 << w) - 1)) != x[h]) {
1709 h--; /* don't need to update q */
1716 /* Return true (1) if we were given an incomplete table */
1717 return y != 0 && g != 1;
1723 struct huft *t; /* table to free */
1725 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1726 list of the tables it made, with the links in a dummy first entry of
1729 register struct huft *p, *q;
1732 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1734 while (p != (struct huft *) NULL) {
1743 int inflate_codes(tl, td, bl, bd)
1744 struct huft *tl, *td; /* literal/length and distance decoder tables */
1745 int bl, bd; /* number of bits decoded by tl[] and td[] */
1747 /* inflate (decompress) the codes in a deflated (compressed) block.
1748 Return an error code or zero if it all goes ok. */
1750 register unsigned e; /* table entry flag/number of extra bits */
1751 unsigned n, d; /* length and index for copy */
1752 unsigned w; /* current window position */
1753 struct huft *t; /* pointer to table entry */
1754 unsigned ml, md; /* masks for bl and bd bits */
1755 register ulg b; /* bit buffer */
1756 register unsigned k; /* number of bits in bit buffer */
1759 /* make local copies of globals */
1760 b = bb; /* initialize bit buffer */
1762 w = wp; /* initialize window position */
1764 /* inflate the coded data */
1765 ml = mask_bits[bl]; /* precompute masks for speed */
1767 for (;;) { /* do until end of block */
1768 NEEDBITS((unsigned) bl)
1769 if ((e = (t = tl + ((unsigned) b & ml))->e) > 16)
1776 } while ((e = (t = t->v.t + ((unsigned) b & mask_bits[e]))->e)
1779 if (e == 16) { /* then it's a literal */
1780 slide[w++] = (uch) t->v.n;
1781 Tracevv((stderr, "%c", slide[w - 1]));
1786 } else { /* it's an EOB or a length */
1788 /* exit if end of block */
1792 /* get length of block to copy */
1794 n = t->v.n + ((unsigned) b & mask_bits[e]);
1797 /* decode distance of block to copy */
1798 NEEDBITS((unsigned) bd)
1799 if ((e = (t = td + ((unsigned) b & md))->e) > 16)
1810 t->v.t + ((unsigned) b & mask_bits[e]))->e) >
1814 d = w - t->v.n - ((unsigned) b & mask_bits[e]);
1816 Tracevv((stderr, "\\[%d,%d]", w - d, n));
1822 WSIZE - ((d &= WSIZE - 1) > w ? d : w)) >
1824 #if !defined(NOMEMCPY) && !defined(DEBUG)
1825 if (w - d >= e) { /* (this test assumes unsigned comparison) */
1826 memcpy(slide + w, slide + d, e);
1829 } else /* do it slow to avoid memcpy() overlap */
1830 #endif /* !NOMEMCPY */
1832 slide[w++] = slide[d++];
1833 Tracevv((stderr, "%c", slide[w - 1]));
1844 /* restore the globals from the locals */
1845 wp = w; /* restore global window pointer */
1846 bb = b; /* restore global bit buffer */
1855 int inflate_stored()
1856 /* "decompress" an inflated type 0 (stored) block. */
1858 unsigned n; /* number of bytes in block */
1859 unsigned w; /* current window position */
1860 register ulg b; /* bit buffer */
1861 register unsigned k; /* number of bits in bit buffer */
1864 /* make local copies of globals */
1865 b = bb; /* initialize bit buffer */
1867 w = wp; /* initialize window position */
1870 /* go to byte boundary */
1875 /* get the length and its complement */
1877 n = ((unsigned) b & 0xffff);
1880 if (n != (unsigned) ((~b) & 0xffff))
1881 return 1; /* error in compressed data */
1885 /* read and output the compressed data */
1888 slide[w++] = (uch) b;
1897 /* restore the globals from the locals */
1898 wp = w; /* restore global window pointer */
1899 bb = b; /* restore global bit buffer */
1907 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
1908 either replace this with a custom decoder, or at least precompute the
1911 int i; /* temporary variable */
1912 struct huft *tl; /* literal/length code table */
1913 struct huft *td; /* distance code table */
1914 int bl; /* lookup bits for tl */
1915 int bd; /* lookup bits for td */
1916 unsigned l[288]; /* length list for huft_build */
1919 /* set up literal table */
1920 for (i = 0; i < 144; i++)
1922 for (; i < 256; i++)
1924 for (; i < 280; i++)
1926 for (; i < 288; i++) /* make a complete, but wrong code set */
1929 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
1933 /* set up distance table */
1934 for (i = 0; i < 30; i++) /* make an incomplete code set */
1937 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1) {
1943 /* decompress until an end-of-block code */
1944 if (inflate_codes(tl, td, bl, bd))
1948 /* free the decoding tables, return */
1956 int inflate_dynamic()
1957 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
1959 int i; /* temporary variables */
1961 unsigned l; /* last length */
1962 unsigned m; /* mask for bit lengths table */
1963 unsigned n; /* number of lengths to get */
1964 struct huft *tl; /* literal/length code table */
1965 struct huft *td; /* distance code table */
1966 int bl; /* lookup bits for tl */
1967 int bd; /* lookup bits for td */
1968 unsigned nb; /* number of bit length codes */
1969 unsigned nl; /* number of literal/length codes */
1970 unsigned nd; /* number of distance codes */
1972 #ifdef PKZIP_BUG_WORKAROUND
1973 unsigned ll[288 + 32]; /* literal/length and distance code lengths */
1975 unsigned ll[286 + 30]; /* literal/length and distance code lengths */
1977 register ulg b; /* bit buffer */
1978 register unsigned k; /* number of bits in bit buffer */
1981 /* make local bit buffer */
1986 /* read in table lengths */
1988 nl = 257 + ((unsigned) b & 0x1f); /* number of literal/length codes */
1991 nd = 1 + ((unsigned) b & 0x1f); /* number of distance codes */
1994 nb = 4 + ((unsigned) b & 0xf); /* number of bit length codes */
1996 #ifdef PKZIP_BUG_WORKAROUND
1997 if (nl > 288 || nd > 32)
1999 if (nl > 286 || nd > 30)
2001 return 1; /* bad lengths */
2004 /* read in bit-length-code lengths */
2005 for (j = 0; j < nb; j++) {
2007 ll[border[j]] = (unsigned) b & 7;
2014 /* build decoding table for trees--single level, 7 bit lookup */
2016 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0) {
2019 return i; /* incomplete code set */
2023 /* read in literal and distance code lengths */
2027 while ((unsigned) i < n) {
2028 NEEDBITS((unsigned) bl)
2029 j = (td = tl + ((unsigned) b & m))->b;
2032 if (j < 16) /* length of code in bits (0..15) */
2033 ll[i++] = l = j; /* save last length in l */
2034 else if (j == 16) { /* repeat last length 3 to 6 times */
2036 j = 3 + ((unsigned) b & 3);
2038 if ((unsigned) i + j > n)
2042 } else if (j == 17) { /* 3 to 10 zero length codes */
2044 j = 3 + ((unsigned) b & 7);
2046 if ((unsigned) i + j > n)
2051 } else { /* j == 18: 11 to 138 zero length codes */
2054 j = 11 + ((unsigned) b & 0x7f);
2056 if ((unsigned) i + j > n)
2065 /* free decoding table for trees */
2069 /* restore the global bit buffer */
2074 /* build the decoding tables for literal/length and distance codes */
2076 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0) {
2078 fprintf(stderr, " incomplete literal tree\n");
2081 return i; /* incomplete code set */
2084 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0) {
2086 fprintf(stderr, " incomplete distance tree\n");
2087 #ifdef PKZIP_BUG_WORKAROUND
2094 return i; /* incomplete code set */
2099 /* decompress until an end-of-block code */
2100 if (inflate_codes(tl, td, bl, bd))
2104 /* free the decoding tables, return */
2112 int inflate_block(e)
2113 int *e; /* last block flag */
2115 /* decompress an inflated block */
2117 unsigned t; /* block type */
2118 register ulg b; /* bit buffer */
2119 register unsigned k; /* number of bits in bit buffer */
2122 /* make local bit buffer */
2127 /* read in last block bit */
2133 /* read in block type */
2135 t = (unsigned) b & 3;
2139 /* restore the global bit buffer */
2144 /* inflate that block type */
2146 return inflate_dynamic();
2148 return inflate_stored();
2150 return inflate_fixed();
2153 /* bad block type */
2160 /* decompress an inflated entry */
2162 int e; /* last block flag */
2163 int r; /* result code */
2164 unsigned h; /* maximum struct huft's malloc'ed */
2167 /* initialize window, bit buffer */
2173 /* decompress until the last block */
2177 if ((r = inflate_block(&e)) != 0)
2183 /* Undo too much lookahead. The next read will be byte aligned so we
2184 * can discard unused bits in the last meaningful byte.
2191 /* flush out slide */
2195 /* return success */
2197 fprintf(stderr, "<%u> ", h);