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
31 static const char gunzip_usage[] =
32 "gunzip [OPTION]... FILE\n\n"
33 "Uncompress FILE (or standard input if FILE is '-').\n\n"
36 "\t-c\tWrite output to standard output\n"
37 "\t-t\tTest compressed file integrity\n";
40 /* These defines are very important for BusyBox. Without these,
41 * huge chunks of ram are pre-allocated making the BusyBox bss
42 * size Freaking Huge(tm), which is a bad thing.*/
46 #define bb_need_name_too_long
47 #define BB_DECLARE_EXTERN
51 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
52 * Copyright (C) 1992-1993 Jean-loup Gailly
53 * The unzip code was written and put in the public domain by Mark Adler.
54 * Portions of the lzw code are derived from the public domain 'compress'
55 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
56 * Ken Turkowski, Dave Mack and Peter Jannesen.
58 * See the license_msg below and the file COPYING for the software license.
59 * See the file algorithm.doc for the compression algorithms and file formats.
63 static char *license_msg[] = {
64 " Copyright (C) 1992-1993 Jean-loup Gailly",
65 " This program is free software; you can redistribute it and/or modify",
66 " it under the terms of the GNU General Public License as published by",
67 " the Free Software Foundation; either version 2, or (at your option)",
68 " any later version.",
70 " This program is distributed in the hope that it will be useful,",
71 " but WITHOUT ANY WARRANTY; without even the implied warranty of",
72 " MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the",
73 " GNU General Public License for more details.",
75 " You should have received a copy of the GNU General Public License",
76 " along with this program; if not, write to the Free Software",
77 " Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.",
82 /* Compress files with zip algorithm and 'compress' interface.
83 * See usage() and help() functions below for all options.
85 * file.gz: compressed file with same mode, owner, and utimes
86 * or stdout with -c option or if stdin used as input.
87 * If the output file name had to be truncated, the original name is kept
88 * in the compressed file.
89 * On MSDOS, file.tmp -> file.tmz. On VMS, file.tmp -> file.tmp-gz.
91 * Using gz on MSDOS would create too many file name conflicts. For
92 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
93 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
94 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
95 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
97 * For the meaning of all compilation flags, see comments in Makefile.in.
101 #include <sys/types.h>
103 #include <sys/stat.h>
105 #include <sys/param.h> /* for PATH_MAX */
107 /* #include "tailor.h" */
109 /* tailor.h -- target dependent definitions
110 * Copyright (C) 1992-1993 Jean-loup Gailly.
111 * This is free software; you can redistribute it and/or modify it under the
112 * terms of the GNU General Public License, see the file COPYING.
115 /* The target dependent definitions should be defined here only.
116 * The target dependent functions should be defined in tailor.c.
121 #define get_char() get_byte()
122 #define put_char(c) put_byte(c)
125 /* I don't like nested includes, but the string and io functions are used
129 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
131 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
134 # define memzero(s, n) memset ((void *)(s), 0, (n))
136 # include <strings.h>
137 # define strchr index
138 # define strrchr rindex
139 # define memcpy(d, s, n) bcopy((s), (d), (n))
140 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
141 # define memzero(s, n) bzero((s), (n))
145 # define RETSIGTYPE void
150 typedef unsigned char uch;
151 typedef unsigned short ush;
152 typedef unsigned long ulg;
154 /* Return codes from gzip */
159 /* Compression methods (see algorithm.doc) */
162 extern int method; /* compression method */
164 /* To save memory for 16 bit systems, some arrays are overlaid between
165 * the various modules:
166 * deflate: prev+head window d_buf l_buf outbuf
167 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
168 * inflate: window inbuf
169 * unpack: window inbuf prefix_len
170 * unlzh: left+right window c_table inbuf c_len
171 * For compression, input is done in window[]. For decompression, output
172 * is done in window except for unlzw.
177 # define INBUFSIZ 0x2000 /* input buffer size */
179 # define INBUFSIZ 0x8000 /* input buffer size */
182 #define INBUF_EXTRA 64 /* required by unlzw() */
186 # define OUTBUFSIZ 8192 /* output buffer size */
188 # define OUTBUFSIZ 16384 /* output buffer size */
191 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
197 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
199 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
203 /*#define DYN_ALLOC*/
206 # define EXTERN(type, array) extern type * array
207 # define DECLARE(type, array, size) type * array
208 # define ALLOC(type, array, size) { \
209 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
210 if (array == NULL) errorMsg("insufficient memory"); \
212 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
214 # define EXTERN(type, array) extern type array[]
215 # define DECLARE(type, array, size) type array[size]
216 # define ALLOC(type, array, size)
220 EXTERN(uch, inbuf); /* input buffer */
221 EXTERN(uch, outbuf); /* output buffer */
222 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
223 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
224 #define tab_suffix window
226 # define tab_prefix prev /* hash link (see deflate.c) */
227 # define head (prev+WSIZE) /* hash head (see deflate.c) */
228 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
230 # define tab_prefix0 prev
231 # define head tab_prefix1
232 EXTERN(ush, tab_prefix0); /* prefix for even codes */
233 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
236 extern unsigned insize; /* valid bytes in inbuf */
237 extern unsigned inptr; /* index of next byte to be processed in inbuf */
238 extern unsigned outcnt; /* bytes in output buffer */
240 extern long bytes_in; /* number of input bytes */
241 extern long bytes_out; /* number of output bytes */
242 extern long header_bytes; /* number of bytes in gzip header */
244 extern long ifile_size; /* input file size, -1 for devices (debug only) */
246 typedef int file_t; /* Do not use stdio */
248 #define NO_FILE (-1) /* in memory compression */
251 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
254 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
255 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
256 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
257 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
258 #define COMMENT 0x10 /* bit 4 set: file comment present */
259 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
260 #define RESERVED 0xC0 /* bit 6,7: reserved */
263 # define WSIZE 0x8000 /* window size--must be a power of two, and */
264 #endif /* at least 32K for zip's deflate method */
267 #define MAX_MATCH 258
268 /* The minimum and maximum match lengths */
270 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
271 /* Minimum amount of lookahead, except at the end of the input file.
272 * See deflate.c for comments about the MIN_MATCH+1.
275 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
276 /* In order to simplify the code, particularly on 16 bit machines, match
277 * distances are limited to MAX_DIST instead of WSIZE.
280 extern int exit_code; /* program exit code */
281 extern int verbose; /* be verbose (-v) */
282 extern int level; /* compression level */
283 extern int test; /* check .z file integrity */
284 extern int save_orig_name; /* set if original name must be saved */
286 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
287 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
289 /* put_byte is used for the compressed output, put_ubyte for the
290 * uncompressed output. However unlzw() uses window for its
291 * suffix table instead of its output buffer, so it does not use put_ubyte
292 * (to be cleaned up).
294 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
296 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
299 /* Output a 16 bit value, lsb first */
300 #define put_short(w) \
301 { if (outcnt < OUTBUFSIZ-2) { \
302 outbuf[outcnt++] = (uch) ((w) & 0xff); \
303 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
305 put_byte((uch)((w) & 0xff)); \
306 put_byte((uch)((ush)(w) >> 8)); \
310 /* Output a 32 bit value to the bit stream, lsb first */
311 #define put_long(n) { \
312 put_short((n) & 0xffff); \
313 put_short(((ulg)(n)) >> 16); \
316 #define seekable() 0 /* force sequential output */
317 #define translate_eol 0 /* no option -a yet */
319 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
321 /* Macros for getting two-byte and four-byte header values */
322 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
323 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
325 /* Diagnostic functions */
327 # define Assert(cond,msg) {if(!(cond)) errorMsg(msg);}
328 # define Trace(x) fprintf x
329 # define Tracev(x) {if (verbose) fprintf x ;}
330 # define Tracevv(x) {if (verbose>1) fprintf x ;}
331 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
332 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
334 # define Assert(cond,msg)
339 # define Tracecv(c,x)
342 #define WARN(msg) {fprintf msg ; \
343 if (exit_code == OK) exit_code = WARNING;}
346 extern int unzip (int in, int out);
349 RETSIGTYPE abort_gzip (void);
352 void lm_init (int pack_level, ush * flags);
356 void ct_init (ush * attr, int *method);
357 int ct_tally (int dist, int lc);
358 ulg flush_block (char *buf, ulg stored_len, int eof);
361 void bi_init (file_t zipfile);
362 void send_bits (int value, int length);
363 unsigned bi_reverse (unsigned value, int length);
364 void bi_windup (void);
365 void copy_block (char *buf, unsigned len, int header);
366 extern int (*read_buf) (char *buf, unsigned size);
369 extern int copy (int in, int out);
370 extern ulg updcrc (uch * s, unsigned n);
371 extern void clear_bufs (void);
372 extern int fill_inbuf (int eof_ok);
373 extern void flush_outbuf (void);
374 extern void flush_window (void);
375 extern void write_buf (int fd, void * buf, unsigned cnt);
378 extern char *basename (char *fname);
379 #endif /* not __linux__ */
380 extern void read_error (void);
381 extern void write_error (void);
384 extern int inflate (void);
386 /* #include "lzw.h" */
388 /* lzw.h -- define the lzw functions.
389 * Copyright (C) 1992-1993 Jean-loup Gailly.
390 * This is free software; you can redistribute it and/or modify it under the
391 * terms of the GNU General Public License, see the file COPYING.
394 #if !defined(OF) && defined(lint)
401 #define INIT_BITS 9 /* Initial number of bits per code */
403 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
405 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
406 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
407 * It's a pity that old uncompress does not check bit 0x20. That makes
408 * extension of the format actually undesirable because old compress
409 * would just crash on the new format instead of giving a meaningful
410 * error message. It does check the number of bits, but it's more
411 * helpful to say "unsupported format, get a new version" than
412 * "can only handle 16 bits".
415 #define BLOCK_MODE 0x80
416 /* Block compression: if table is full and compression rate is dropping,
417 * clear the dictionary.
420 #define LZW_RESERVED 0x60 /* reserved bits */
422 #define CLEAR 256 /* flush the dictionary */
423 #define FIRST (CLEAR+1) /* first free entry */
425 extern int maxbits; /* max bits per code for LZW */
426 extern int block_mode; /* block compress mode -C compatible with 2.0 */
428 extern int lzw (int in, int out);
429 extern int unlzw (int in, int out);
432 /* #include "revision.h" */
434 /* revision.h -- define the version number
435 * Copyright (C) 1992-1993 Jean-loup Gailly.
436 * This is free software; you can redistribute it and/or modify it under the
437 * terms of the GNU General Public License, see the file COPYING.
440 #define VERSION "1.2.4"
442 #define REVDATE "18 Aug 93"
444 /* This version does not support compression into old compress format: */
449 /* #include "getopt.h" */
451 /* Declarations for getopt.
452 Copyright (C) 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc.
454 This program is free software; you can redistribute it and/or modify it
455 under the terms of the GNU General Public License as published by the
456 Free Software Foundation; either version 2, or (at your option) any
459 This program is distributed in the hope that it will be useful,
460 but WITHOUT ANY WARRANTY; without even the implied warranty of
461 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
462 GNU General Public License for more details.
464 You should have received a copy of the GNU General Public License
465 along with this program; if not, write to the Free Software
466 Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
474 /* For communication from `getopt' to the caller.
475 When `getopt' finds an option that takes an argument,
476 the argument value is returned here.
477 Also, when `ordering' is RETURN_IN_ORDER,
478 each non-option ARGV-element is returned here. */
481 /* Index in ARGV of the next element to be scanned.
482 This is used for communication to and from the caller
483 and for communication between successive calls to `getopt'.
485 On entry to `getopt', zero means this is the first call; initialize.
487 When `getopt' returns EOF, this is the index of the first of the
488 non-option elements that the caller should itself scan.
490 Otherwise, `optind' communicates from one call to the next
491 how much of ARGV has been scanned so far. */
495 /* Callers store zero here to inhibit the error message `getopt' prints
496 for unrecognized options. */
500 /* Set to an option character which was unrecognized. */
504 /* Describe the long-named options requested by the application.
505 The LONG_OPTIONS argument to getopt_long or getopt_long_only is a vector
506 of `struct option' terminated by an element containing a name which is
509 The field `has_arg' is:
510 no_argument (or 0) if the option does not take an argument,
511 required_argument (or 1) if the option requires an argument,
512 optional_argument (or 2) if the option takes an optional argument.
514 If the field `flag' is not NULL, it points to a variable that is set
515 to the value given in the field `val' when the option is found, but
516 left unchanged if the option is not found.
518 To have a long-named option do something other than set an `int' to
519 a compiled-in constant, such as set a value from `optarg', set the
520 option's `flag' field to zero and its `val' field to a nonzero
521 value (the equivalent single-letter option character, if there is
522 one). For long options that have a zero `flag' field, `getopt'
523 returns the contents of the `val' field. */
531 /* has_arg can't be an enum because some compilers complain about
532 type mismatches in all the code that assumes it is an int. */
538 /* Names for the values of the `has_arg' field of `struct option'. */
540 #define no_argument 0
541 #define required_argument 1
542 #define optional_argument 2
544 #if __STDC__ || defined(PROTO)
545 #if defined(__GNU_LIBRARY__)
546 /* Many other libraries have conflicting prototypes for getopt, with
547 differences in the consts, in stdlib.h. To avoid compilation
548 errors, only prototype getopt for the GNU C library. */
549 extern int getopt(int argc, char *const *argv, const char *shortopts);
550 #endif /* not __GNU_LIBRARY__ */
551 extern int getopt_long(int argc, char *const *argv,
552 const char *shortopts,
553 const struct option *longopts, int *longind);
554 extern int getopt_long_only(int argc, char *const *argv,
555 const char *shortopts,
556 const struct option *longopts,
559 /* Internal only. Users should not call this directly. */
560 extern int _getopt_internal(int argc, char *const *argv,
561 const char *shortopts,
562 const struct option *longopts,
563 int *longind, int long_only);
564 #else /* not __STDC__ */
566 extern int getopt_long();
567 extern int getopt_long_only();
569 extern int _getopt_internal();
570 #endif /* not __STDC__ */
575 #endif /* _GETOPT_H */
582 typedef struct dirent dir_type;
584 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
585 # define DIR_OPT "DIRENT"
587 # define NLENGTH(dirent) ((dirent)->d_namlen)
589 # include <sys/dir.h>
590 typedef struct direct dir_type;
592 # define DIR_OPT "SYSDIR"
595 # include <sys/ndir.h>
596 typedef struct direct dir_type;
598 # define DIR_OPT "SYSNDIR"
602 typedef struct direct dir_type;
604 # define DIR_OPT "NDIR"
607 # define DIR_OPT "NO_DIR"
612 #if !defined(S_ISDIR) && defined(S_IFDIR)
613 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
615 #if !defined(S_ISREG) && defined(S_IFREG)
616 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
618 typedef RETSIGTYPE(*sig_type) (int);
621 # define O_BINARY 0 /* creation mode for open() */
625 /* Pure BSD system? */
626 # include <sys/file.h>
628 # define O_CREAT FCREAT
631 # define O_EXCL FEXCL
636 # define S_IRUSR 0400
639 # define S_IWUSR 0200
641 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
643 #ifndef MAX_PATH_LEN /* max pathname length */
645 # define MAX_PATH_LEN PATH_MAX
647 # define MAX_PATH_LEN 1024
657 off_t lseek (int fd, off_t offset, int whence);
663 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
664 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
665 DECLARE(ush, d_buf, DIST_BUFSIZE);
666 DECLARE(uch, window, 2L * WSIZE);
668 DECLARE(ush, tab_prefix, 1L << BITS);
670 DECLARE(ush, tab_prefix0, 1L << (BITS - 1));
671 DECLARE(ush, tab_prefix1, 1L << (BITS - 1));
674 /* local variables */
676 int test_mode = 0; /* check file integrity option */
677 int foreground; /* set if program run in foreground */
678 int maxbits = BITS; /* max bits per code for LZW */
679 int method = DEFLATED; /* compression method */
680 int exit_code = OK; /* program exit code */
681 int last_member; /* set for .zip and .Z files */
682 int part_nb; /* number of parts in .gz file */
683 long ifile_size; /* input file size, -1 for devices (debug only) */
685 long bytes_in; /* number of input bytes */
686 long bytes_out; /* number of output bytes */
687 long total_in = 0; /* input bytes for all files */
688 long total_out = 0; /* output bytes for all files */
689 struct stat istat; /* status for input file */
690 int ifd; /* input file descriptor */
691 int ofd; /* output file descriptor */
692 unsigned insize; /* valid bytes in inbuf */
693 unsigned inptr; /* index of next byte to be processed in inbuf */
694 unsigned outcnt; /* bytes in output buffer */
696 long header_bytes; /* number of bytes in gzip header */
698 /* local functions */
700 local int get_method (int in);
702 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
704 /* ======================================================================== */
705 int gunzip_main(int argc, char **argv)
707 int file_count; /* number of files to precess */
713 int delInputFile = 0;
716 char ifname[MAX_PATH_LEN + 1]; /* input file name */
717 char ofname[MAX_PATH_LEN + 1]; /* output file name */
722 if (strcmp(*argv, "zcat") == 0)
725 /* Parse any options */
726 while (--argc > 0 && **(++argv) == '-') {
727 if (*((*argv) + 1) == '\0') {
731 while (*(++(*argv))) {
746 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
748 (void) signal(SIGINT, (sig_type) abort_gzip);
751 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
752 (void) signal(SIGTERM, (sig_type) abort_gzip);
756 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
757 (void) signal(SIGHUP, (sig_type) abort_gzip);
761 file_count = argc - optind;
763 /* Allocate all global buffers (for DYN_ALLOC option) */
764 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
765 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
766 ALLOC(ush, d_buf, DIST_BUFSIZE);
767 ALLOC(uch, window, 2L * WSIZE);
769 ALLOC(ush, tab_prefix, 1L << BITS);
771 ALLOC(ush, tab_prefix0, 1L << (BITS - 1));
772 ALLOC(ush, tab_prefix1, 1L << (BITS - 1));
775 if (fromstdin == 1) {
776 strcpy(ofname, "stdin");
778 inFileNum = fileno(stdin);
779 ifile_size = -1L; /* convention for unknown size */
781 /* Open up the input file */
784 if (strlen(*argv) > MAX_PATH_LEN) {
785 fprintf(stderr, name_too_long, "gunzip");
788 strcpy(ifname, *argv);
790 /* Open input fille */
791 inFileNum = open(ifname, O_RDONLY);
796 /* Get the time stamp on the input file. */
797 result = stat(ifname, &statBuf);
802 ifile_size = statBuf.st_size;
805 if (to_stdout == 1) {
806 /* And get to work */
807 strcpy(ofname, "stdout");
808 outFileNum = fileno(stdout);
810 clear_bufs(); /* clear input and output buffers */
813 /* Actually do the compression/decompression. */
814 unzip(inFileNum, outFileNum);
816 } else if (test_mode) {
817 /* Actually do the compression/decompression. */
822 /* And get to work */
823 if (strlen(ifname) > MAX_PATH_LEN - 4) {
824 fprintf(stderr, name_too_long, "gunzip");
827 strcpy(ofname, ifname);
828 pos = strstr(ofname, ".gz");
833 pos = strstr(ofname, ".tgz");
841 /* Open output fille */
842 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
843 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
845 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL);
847 if (outFileNum < 0) {
851 /* Set permissions on the file */
852 fchmod(outFileNum, statBuf.st_mode);
854 clear_bufs(); /* clear input and output buffers */
857 /* Actually do the compression/decompression. */
858 result = unzip(inFileNum, outFileNum);
862 /* Delete the original file */
864 delFileName = ifname;
866 delFileName = ofname;
868 if (delInputFile == 1 && unlink(delFileName) < 0) {
877 /* ========================================================================
878 * Check the magic number of the input file and update ofname if an
879 * original name was given and to_stdout is not set.
880 * Return the compression method, -1 for error, -2 for warning.
881 * Set inptr to the offset of the next byte to be processed.
882 * Updates time_stamp if there is one and --no-time is not used.
883 * This function may be called repeatedly for an input file consisting
884 * of several contiguous gzip'ed members.
885 * IN assertions: there is at least one remaining compressed member.
886 * If the member is a zip file, it must be the only one.
888 local int get_method(in)
889 int in; /* input file descriptor */
891 uch flags; /* compression flags */
892 char magic[2]; /* magic header */
894 magic[0] = (char) get_byte();
895 magic[1] = (char) get_byte();
896 method = -1; /* unknown yet */
897 part_nb++; /* number of parts in gzip file */
899 last_member = RECORD_IO;
900 /* assume multiple members in gzip file except for record oriented I/O */
902 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
904 method = (int) get_byte();
905 if (method != DEFLATED) {
907 "unknown method %d -- get newer version of gzip\n",
912 flags = (uch) get_byte();
914 (ulg) get_byte(); /* Ignore time stamp */
919 (void) get_byte(); /* Ignore extra flags for the moment */
920 (void) get_byte(); /* Ignore OS type for the moment */
922 if ((flags & EXTRA_FIELD) != 0) {
923 unsigned len = (unsigned) get_byte();
925 len |= ((unsigned) get_byte()) << 8;
931 /* Discard original name if any */
932 if ((flags & ORIG_NAME) != 0) {
933 while (get_char() != 0) /* null */
937 /* Discard file comment if any */
938 if ((flags & COMMENT) != 0) {
939 while (get_char() != 0) /* null */
943 header_bytes = inptr + 2 * sizeof(long); /* include crc and size */
952 fprintf(stderr, "\nnot in gzip format\n");
956 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
961 /* ========================================================================
962 * Signal and error handler.
964 RETSIGTYPE abort_gzip()
969 /* unzip.c -- decompress files in gzip or pkzip format.
970 * Copyright (C) 1992-1993 Jean-loup Gailly
971 * This is free software; you can redistribute it and/or modify it under the
972 * terms of the GNU General Public License, see the file COPYING.
974 * The code in this file is derived from the file funzip.c written
975 * and put in the public domain by Mark Adler.
979 This version can extract files in gzip or pkzip format.
980 For the latter, only the first entry is extracted, and it has to be
981 either deflated or stored.
984 /* #include "crypt.h" */
986 /* crypt.h (dummy version) -- do not perform encryption
987 * Hardly worth copyrighting :-)
991 # undef CRYPT /* dummy version */
994 #define RAND_HEAD_LEN 12 /* length of encryption random header */
999 /* PKZIP header definitions */
1000 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
1001 #define LOCFLG 6 /* offset of bit flag */
1002 #define CRPFLG 1 /* bit for encrypted entry */
1003 #define EXTFLG 8 /* bit for extended local header */
1004 #define LOCHOW 8 /* offset of compression method */
1005 #define LOCTIM 10 /* file mod time (for decryption) */
1006 #define LOCCRC 14 /* offset of crc */
1007 #define LOCSIZ 18 /* offset of compressed size */
1008 #define LOCLEN 22 /* offset of uncompressed length */
1009 #define LOCFIL 26 /* offset of file name field length */
1010 #define LOCEXT 28 /* offset of extra field length */
1011 #define LOCHDR 30 /* size of local header, including sig */
1012 #define EXTHDR 16 /* size of extended local header, inc sig */
1017 char *key; /* not used--needed to link crypt.c */
1018 int pkzip = 0; /* set for a pkzip file */
1019 int ext_header = 0; /* set if extended local header */
1021 /* ===========================================================================
1022 * Unzip in to out. This routine works on both gzip and pkzip files.
1024 * IN assertions: the buffer inbuf contains already the beginning of
1025 * the compressed data, from offsets inptr to insize-1 included.
1026 * The magic header has already been checked. The output buffer is cleared.
1029 int in, out; /* input and output file descriptors */
1031 ulg orig_crc = 0; /* original crc */
1032 ulg orig_len = 0; /* original uncompressed length */
1034 uch buf[EXTHDR]; /* extended local header */
1038 method = get_method(ifd);
1040 exit(exit_code); /* error message already emitted */
1043 updcrc(NULL, 0); /* initialize crc */
1045 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
1046 orig_crc = LG(inbuf + LOCCRC);
1047 orig_len = LG(inbuf + LOCLEN);
1051 if (method == DEFLATED) {
1053 int res = inflate();
1056 errorMsg("out of memory");
1057 } else if (res != 0) {
1058 errorMsg("invalid compressed data--format violated");
1062 errorMsg("internal error, invalid method");
1065 /* Get the crc and original length */
1067 /* crc32 (see algorithm.doc)
1068 * uncompressed input size modulo 2^32
1070 for (n = 0; n < 8; n++) {
1071 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
1074 orig_len = LG(buf + 4);
1076 } else if (ext_header) { /* If extended header, check it */
1077 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
1079 * compressed size 4-bytes
1080 * uncompressed size 4-bytes
1082 for (n = 0; n < EXTHDR; n++) {
1083 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
1085 orig_crc = LG(buf + 4);
1086 orig_len = LG(buf + 12);
1089 /* Validate decompression */
1090 if (orig_crc != updcrc(outbuf, 0)) {
1091 errorMsg("invalid compressed data--crc error");
1093 if (orig_len != (ulg) bytes_out) {
1094 errorMsg("invalid compressed data--length error");
1097 /* Check if there are more entries in a pkzip file */
1098 if (pkzip && inptr + 4 < insize && LG(inbuf + inptr) == LOCSIG) {
1099 WARN((stderr, "has more than one entry--rest ignored\n"));
1101 ext_header = pkzip = 0; /* for next file */
1105 /* util.c -- utility functions for gzip support
1106 * Copyright (C) 1992-1993 Jean-loup Gailly
1107 * This is free software; you can redistribute it and/or modify it under the
1108 * terms of the GNU General Public License, see the file COPYING.
1113 #include <sys/types.h>
1115 #ifdef HAVE_UNISTD_H
1116 # include <unistd.h>
1122 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1123 # include <stdlib.h>
1128 static const ulg crc_32_tab[]; /* crc table, defined below */
1130 /* ===========================================================================
1131 * Run a set of bytes through the crc shift register. If s is a NULL
1132 * pointer, then initialize the crc shift register contents instead.
1133 * Return the current crc in either case.
1136 uch *s; /* pointer to bytes to pump through */
1137 unsigned n; /* number of bytes in s[] */
1139 register ulg c; /* temporary variable */
1141 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
1149 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
1153 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1156 /* ===========================================================================
1157 * Clear input and output buffers
1163 bytes_in = bytes_out = 0L;
1166 /* ===========================================================================
1167 * Fill the input buffer. This is called only when the buffer is empty.
1169 int fill_inbuf(eof_ok)
1170 int eof_ok; /* set if EOF acceptable as a result */
1174 /* Read as much as possible */
1178 len = read(ifd, (char *) inbuf + insize, INBUFSIZ - insize);
1179 if (len == 0 || len == EOF)
1182 } while (insize < INBUFSIZ);
1189 bytes_in += (ulg) insize;
1194 /* ===========================================================================
1195 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1196 * (used for the compressed data only)
1204 write_buf(ofd, (char *) outbuf, outcnt);
1205 bytes_out += (ulg) outcnt;
1209 /* ===========================================================================
1210 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1211 * (Used for the decompressed data only.)
1217 updcrc(window, outcnt);
1220 write_buf(ofd, (char *) window, outcnt);
1221 bytes_out += (ulg) outcnt;
1225 /* ===========================================================================
1226 * Does the same as write(), but also handles partial pipe writes and checks
1229 void write_buf(fd, buf, cnt)
1236 while ((n = write(fd, buf, cnt)) != cnt) {
1237 if (n == (unsigned) (-1)) {
1241 buf = (void *) ((char *) buf + n);
1245 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1247 /* Provide missing strspn and strcspn functions. */
1253 int strspn (const char *s, const char *accept);
1254 int strcspn (const char *s, const char *reject);
1256 /* ========================================================================
1257 * Return the length of the maximum initial segment
1258 * of s which contains only characters in accept.
1260 int strspn(s, accept)
1264 register const char *p;
1265 register const char *a;
1266 register int count = 0;
1268 for (p = s; *p != '\0'; ++p) {
1269 for (a = accept; *a != '\0'; ++a) {
1280 /* ========================================================================
1281 * Return the length of the maximum inital segment of s
1282 * which contains no characters from reject.
1284 int strcspn(s, reject)
1288 register int count = 0;
1290 while (*s != '\0') {
1291 if (strchr(reject, *s++) != NULL)
1298 #endif /* NO_STRING_H */
1301 /* ========================================================================
1306 fprintf(stderr, "\n");
1310 fprintf(stderr, "unexpected end of file\n");
1317 fprintf(stderr, "\n");
1323 /* ========================================================================
1324 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1326 static const ulg crc_32_tab[] = {
1327 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1328 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1329 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1330 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1331 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1332 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1333 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1334 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1335 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1336 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1337 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1338 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1339 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1340 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1341 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1342 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1343 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1344 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1345 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1346 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1347 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1348 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1349 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1350 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1351 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1352 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1353 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1354 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1355 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1356 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1357 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1358 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1359 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1360 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1361 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1362 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1363 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1364 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1365 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1366 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1367 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1368 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1369 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1370 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1371 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1372 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1373 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1374 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1375 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1376 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1377 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1381 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1382 version c10p1, 10 January 1993 */
1384 /* You can do whatever you like with this source file, though I would
1385 prefer that if you modify it and redistribute it that you include
1386 comments to that effect with your name and the date. Thank you.
1387 [The history has been moved to the file ChangeLog.]
1391 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1392 method searches for as much of the current string of bytes (up to a
1393 length of 258) in the previous 32K bytes. If it doesn't find any
1394 matches (of at least length 3), it codes the next byte. Otherwise, it
1395 codes the length of the matched string and its distance backwards from
1396 the current position. There is a single Huffman code that codes both
1397 single bytes (called "literals") and match lengths. A second Huffman
1398 code codes the distance information, which follows a length code. Each
1399 length or distance code actually represents a base value and a number
1400 of "extra" (sometimes zero) bits to get to add to the base value. At
1401 the end of each deflated block is a special end-of-block (EOB) literal/
1402 length code. The decoding process is basically: get a literal/length
1403 code; if EOB then done; if a literal, emit the decoded byte; if a
1404 length then get the distance and emit the referred-to bytes from the
1405 sliding window of previously emitted data.
1407 There are (currently) three kinds of inflate blocks: stored, fixed, and
1408 dynamic. The compressor deals with some chunk of data at a time, and
1409 decides which method to use on a chunk-by-chunk basis. A chunk might
1410 typically be 32K or 64K. If the chunk is uncompressible, then the
1411 "stored" method is used. In this case, the bytes are simply stored as
1412 is, eight bits per byte, with none of the above coding. The bytes are
1413 preceded by a count, since there is no longer an EOB code.
1415 If the data is compressible, then either the fixed or dynamic methods
1416 are used. In the dynamic method, the compressed data is preceded by
1417 an encoding of the literal/length and distance Huffman codes that are
1418 to be used to decode this block. The representation is itself Huffman
1419 coded, and so is preceded by a description of that code. These code
1420 descriptions take up a little space, and so for small blocks, there is
1421 a predefined set of codes, called the fixed codes. The fixed method is
1422 used if the block codes up smaller that way (usually for quite small
1423 chunks), otherwise the dynamic method is used. In the latter case, the
1424 codes are customized to the probabilities in the current block, and so
1425 can code it much better than the pre-determined fixed codes.
1427 The Huffman codes themselves are decoded using a mutli-level table
1428 lookup, in order to maximize the speed of decoding plus the speed of
1429 building the decoding tables. See the comments below that precede the
1430 lbits and dbits tuning parameters.
1435 Notes beyond the 1.93a appnote.txt:
1437 1. Distance pointers never point before the beginning of the output
1439 2. Distance pointers can point back across blocks, up to 32k away.
1440 3. There is an implied maximum of 7 bits for the bit length table and
1441 15 bits for the actual data.
1442 4. If only one code exists, then it is encoded using one bit. (Zero
1443 would be more efficient, but perhaps a little confusing.) If two
1444 codes exist, they are coded using one bit each (0 and 1).
1445 5. There is no way of sending zero distance codes--a dummy must be
1446 sent if there are none. (History: a pre 2.0 version of PKZIP would
1447 store blocks with no distance codes, but this was discovered to be
1448 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1449 zero distance codes, which is sent as one code of zero bits in
1451 6. There are up to 286 literal/length codes. Code 256 represents the
1452 end-of-block. Note however that the static length tree defines
1453 288 codes just to fill out the Huffman codes. Codes 286 and 287
1454 cannot be used though, since there is no length base or extra bits
1455 defined for them. Similarly, there are up to 30 distance codes.
1456 However, static trees define 32 codes (all 5 bits) to fill out the
1457 Huffman codes, but the last two had better not show up in the data.
1458 7. Unzip can check dynamic Huffman blocks for complete code sets.
1459 The exception is that a single code would not be complete (see #4).
1460 8. The five bits following the block type is really the number of
1461 literal codes sent minus 257.
1462 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1463 (1+6+6). Therefore, to output three times the length, you output
1464 three codes (1+1+1), whereas to output four times the same length,
1465 you only need two codes (1+3). Hmm.
1466 10. In the tree reconstruction algorithm, Code = Code + Increment
1467 only if BitLength(i) is not zero. (Pretty obvious.)
1468 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1469 12. Note: length code 284 can represent 227-258, but length code 285
1470 really is 258. The last length deserves its own, short code
1471 since it gets used a lot in very redundant files. The length
1472 258 is special since 258 - 3 (the min match length) is 255.
1473 13. The literal/length and distance code bit lengths are read as a
1474 single stream of lengths. It is possible (and advantageous) for
1475 a repeat code (16, 17, or 18) to go across the boundary between
1476 the two sets of lengths.
1479 #include <sys/types.h>
1481 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1482 # include <stdlib.h>
1486 #define slide window
1488 /* Huffman code lookup table entry--this entry is four bytes for machines
1489 that have 16-bit pointers (e.g. PC's in the small or medium model).
1490 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1491 means that v is a literal, 16 < e < 32 means that v is a pointer to
1492 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1493 an unused code. If a code with e == 99 is looked up, this implies an
1494 error in the data. */
1496 uch e; /* number of extra bits or operation */
1497 uch b; /* number of bits in this code or subcode */
1499 ush n; /* literal, length base, or distance base */
1500 struct huft *t; /* pointer to next level of table */
1505 /* Function prototypes */
1506 int huft_build (unsigned *, unsigned, unsigned, ush *, ush *,
1507 struct huft **, int *);
1508 int huft_free (struct huft *);
1509 int inflate_codes (struct huft *, struct huft *, int, int);
1510 int inflate_stored (void);
1511 int inflate_fixed (void);
1512 int inflate_dynamic (void);
1513 int inflate_block (int *);
1517 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1518 stream to find repeated byte strings. This is implemented here as a
1519 circular buffer. The index is updated simply by incrementing and then
1520 and'ing with 0x7fff (32K-1). */
1521 /* It is left to other modules to supply the 32K area. It is assumed
1522 to be usable as if it were declared "uch slide[32768];" or as just
1523 "uch *slide;" and then malloc'ed in the latter case. The definition
1524 must be in unzip.h, included above. */
1525 /* unsigned wp; current position in slide */
1527 #define flush_output(w) (wp=(w),flush_window())
1529 /* Tables for deflate from PKZIP's appnote.txt. */
1530 static unsigned border[] = { /* Order of the bit length code lengths */
1531 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15
1533 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1534 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1535 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0
1538 /* note: see note #13 above about the 258 in this list. */
1539 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1540 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1541 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99
1542 }; /* 99==invalid */
1543 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1544 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1545 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1546 8193, 12289, 16385, 24577
1548 static ush cpdext[] = { /* Extra bits for distance codes */
1549 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1550 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1556 /* Macros for inflate() bit peeking and grabbing.
1560 x = b & mask_bits[j];
1563 where NEEDBITS makes sure that b has at least j bits in it, and
1564 DUMPBITS removes the bits from b. The macros use the variable k
1565 for the number of bits in b. Normally, b and k are register
1566 variables for speed, and are initialized at the beginning of a
1567 routine that uses these macros from a global bit buffer and count.
1569 If we assume that EOB will be the longest code, then we will never
1570 ask for bits with NEEDBITS that are beyond the end of the stream.
1571 So, NEEDBITS should not read any more bytes than are needed to
1572 meet the request. Then no bytes need to be "returned" to the buffer
1573 at the end of the last block.
1575 However, this assumption is not true for fixed blocks--the EOB code
1576 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1577 (The EOB code is shorter than other codes because fixed blocks are
1578 generally short. So, while a block always has an EOB, many other
1579 literal/length codes have a significantly lower probability of
1580 showing up at all.) However, by making the first table have a
1581 lookup of seven bits, the EOB code will be found in that first
1582 lookup, and so will not require that too many bits be pulled from
1586 ulg bb; /* bit buffer */
1587 unsigned bk; /* bits in bit buffer */
1591 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1592 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1598 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1600 # define NEXTBYTE() (uch)get_byte()
1602 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1603 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1607 Huffman code decoding is performed using a multi-level table lookup.
1608 The fastest way to decode is to simply build a lookup table whose
1609 size is determined by the longest code. However, the time it takes
1610 to build this table can also be a factor if the data being decoded
1611 is not very long. The most common codes are necessarily the
1612 shortest codes, so those codes dominate the decoding time, and hence
1613 the speed. The idea is you can have a shorter table that decodes the
1614 shorter, more probable codes, and then point to subsidiary tables for
1615 the longer codes. The time it costs to decode the longer codes is
1616 then traded against the time it takes to make longer tables.
1618 This results of this trade are in the variables lbits and dbits
1619 below. lbits is the number of bits the first level table for literal/
1620 length codes can decode in one step, and dbits is the same thing for
1621 the distance codes. Subsequent tables are also less than or equal to
1622 those sizes. These values may be adjusted either when all of the
1623 codes are shorter than that, in which case the longest code length in
1624 bits is used, or when the shortest code is *longer* than the requested
1625 table size, in which case the length of the shortest code in bits is
1628 There are two different values for the two tables, since they code a
1629 different number of possibilities each. The literal/length table
1630 codes 286 possible values, or in a flat code, a little over eight
1631 bits. The distance table codes 30 possible values, or a little less
1632 than five bits, flat. The optimum values for speed end up being
1633 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1634 The optimum values may differ though from machine to machine, and
1635 possibly even between compilers. Your mileage may vary.
1639 int lbits = 9; /* bits in base literal/length lookup table */
1640 int dbits = 6; /* bits in base distance lookup table */
1643 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1644 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1645 #define N_MAX 288 /* maximum number of codes in any set */
1648 unsigned hufts; /* track memory usage */
1651 int huft_build(b, n, s, d, e, t, m)
1652 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1653 unsigned n; /* number of codes (assumed <= N_MAX) */
1654 unsigned s; /* number of simple-valued codes (0..s-1) */
1655 ush *d; /* list of base values for non-simple codes */
1656 ush *e; /* list of extra bits for non-simple codes */
1657 struct huft **t; /* result: starting table */
1658 int *m; /* maximum lookup bits, returns actual */
1660 /* Given a list of code lengths and a maximum table size, make a set of
1661 tables to decode that set of codes. Return zero on success, one if
1662 the given code set is incomplete (the tables are still built in this
1663 case), two if the input is invalid (all zero length codes or an
1664 oversubscribed set of lengths), and three if not enough memory. */
1666 unsigned a; /* counter for codes of length k */
1667 unsigned c[BMAX + 1]; /* bit length count table */
1668 unsigned f; /* i repeats in table every f entries */
1669 int g; /* maximum code length */
1670 int h; /* table level */
1671 register unsigned i; /* counter, current code */
1672 register unsigned j; /* counter */
1673 register int k; /* number of bits in current code */
1674 int l; /* bits per table (returned in m) */
1675 register unsigned *p; /* pointer into c[], b[], or v[] */
1676 register struct huft *q; /* points to current table */
1677 struct huft r; /* table entry for structure assignment */
1678 struct huft *u[BMAX]; /* table stack */
1679 unsigned v[N_MAX]; /* values in order of bit length */
1680 register int w; /* bits before this table == (l * h) */
1681 unsigned x[BMAX + 1]; /* bit offsets, then code stack */
1682 unsigned *xp; /* pointer into x */
1683 int y; /* number of dummy codes added */
1684 unsigned z; /* number of entries in current table */
1687 /* Generate counts for each bit length */
1688 memzero(c, sizeof(c));
1695 && n - i <= '~' ? "%c %d\n" : "0x%x %d\n"), n - i, *p));
1696 c[*p]++; /* assume all entries <= BMAX */
1697 p++; /* Can't combine with above line (Solaris bug) */
1699 if (c[0] == n) { /* null input--all zero length codes */
1700 *t = (struct huft *) NULL;
1706 /* Find minimum and maximum length, bound *m by those */
1708 for (j = 1; j <= BMAX; j++)
1711 k = j; /* minimum code length */
1712 if ((unsigned) l < j)
1714 for (i = BMAX; i; i--)
1717 g = i; /* maximum code length */
1718 if ((unsigned) l > i)
1723 /* Adjust last length count to fill out codes, if needed */
1724 for (y = 1 << j; j < i; j++, y <<= 1)
1725 if ((y -= c[j]) < 0)
1726 return 2; /* bad input: more codes than bits */
1727 if ((y -= c[i]) < 0)
1732 /* Generate starting offsets into the value table for each length */
1736 while (--i) { /* note that i == g from above */
1737 *xp++ = (j += *p++);
1741 /* Make a table of values in order of bit lengths */
1745 if ((j = *p++) != 0)
1750 /* Generate the Huffman codes and for each, make the table entries */
1751 x[0] = i = 0; /* first Huffman code is zero */
1752 p = v; /* grab values in bit order */
1753 h = -1; /* no tables yet--level -1 */
1754 w = -l; /* bits decoded == (l * h) */
1755 u[0] = (struct huft *) NULL; /* just to keep compilers happy */
1756 q = (struct huft *) NULL; /* ditto */
1759 /* go through the bit lengths (k already is bits in shortest code) */
1760 for (; k <= g; k++) {
1763 /* here i is the Huffman code of length k bits for value *p */
1764 /* make tables up to required level */
1767 w += l; /* previous table always l bits */
1769 /* compute minimum size table less than or equal to l bits */
1770 z = (z = g - w) > (unsigned) l ? l : z; /* upper limit on table size */
1771 if ((f = 1 << (j = k - w)) > a + 1) { /* try a k-w bit table *//* too few codes for k-w bit table */
1772 f -= a + 1; /* deduct codes from patterns left */
1774 while (++j < z) { /* try smaller tables up to z bits */
1775 if ((f <<= 1) <= *++xp)
1776 break; /* enough codes to use up j bits */
1777 f -= *xp; /* else deduct codes from patterns */
1780 z = 1 << j; /* table entries for j-bit table */
1782 /* allocate and link in new table */
1785 (struct huft *) malloc((z + 1) *
1786 sizeof(struct huft))) ==
1787 (struct huft *) NULL) {
1790 return 3; /* not enough memory */
1792 hufts += z + 1; /* track memory usage */
1793 *t = q + 1; /* link to list for huft_free() */
1794 *(t = &(q->v.t)) = (struct huft *) NULL;
1795 u[h] = ++q; /* table starts after link */
1797 /* connect to last table, if there is one */
1799 x[h] = i; /* save pattern for backing up */
1800 r.b = (uch) l; /* bits to dump before this table */
1801 r.e = (uch) (16 + j); /* bits in this table */
1802 r.v.t = q; /* pointer to this table */
1803 j = i >> (w - l); /* (get around Turbo C bug) */
1804 u[h - 1][j] = r; /* connect to last table */
1808 /* set up table entry in r */
1809 r.b = (uch) (k - w);
1811 r.e = 99; /* out of values--invalid code */
1813 r.e = (uch) (*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1814 r.v.n = (ush) (*p); /* simple code is just the value */
1815 p++; /* one compiler does not like *p++ */
1817 r.e = (uch) e[*p - s]; /* non-simple--look up in lists */
1818 r.v.n = d[*p++ - s];
1821 /* fill code-like entries with r */
1823 for (j = i >> w; j < z; j += f)
1826 /* backwards increment the k-bit code i */
1827 for (j = 1 << (k - 1); i & j; j >>= 1)
1831 /* backup over finished tables */
1832 while ((i & ((1 << w) - 1)) != x[h]) {
1833 h--; /* don't need to update q */
1840 /* Return true (1) if we were given an incomplete table */
1841 return y != 0 && g != 1;
1847 struct huft *t; /* table to free */
1849 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1850 list of the tables it made, with the links in a dummy first entry of
1853 register struct huft *p, *q;
1856 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1858 while (p != (struct huft *) NULL) {
1867 int inflate_codes(tl, td, bl, bd)
1868 struct huft *tl, *td; /* literal/length and distance decoder tables */
1869 int bl, bd; /* number of bits decoded by tl[] and td[] */
1871 /* inflate (decompress) the codes in a deflated (compressed) block.
1872 Return an error code or zero if it all goes ok. */
1874 register unsigned e; /* table entry flag/number of extra bits */
1875 unsigned n, d; /* length and index for copy */
1876 unsigned w; /* current window position */
1877 struct huft *t; /* pointer to table entry */
1878 unsigned ml, md; /* masks for bl and bd bits */
1879 register ulg b; /* bit buffer */
1880 register unsigned k; /* number of bits in bit buffer */
1883 /* make local copies of globals */
1884 b = bb; /* initialize bit buffer */
1886 w = wp; /* initialize window position */
1888 /* inflate the coded data */
1889 ml = mask_bits[bl]; /* precompute masks for speed */
1891 for (;;) { /* do until end of block */
1892 NEEDBITS((unsigned) bl)
1893 if ((e = (t = tl + ((unsigned) b & ml))->e) > 16)
1900 } while ((e = (t = t->v.t + ((unsigned) b & mask_bits[e]))->e)
1903 if (e == 16) { /* then it's a literal */
1904 slide[w++] = (uch) t->v.n;
1905 Tracevv((stderr, "%c", slide[w - 1]));
1910 } else { /* it's an EOB or a length */
1912 /* exit if end of block */
1916 /* get length of block to copy */
1918 n = t->v.n + ((unsigned) b & mask_bits[e]);
1921 /* decode distance of block to copy */
1922 NEEDBITS((unsigned) bd)
1923 if ((e = (t = td + ((unsigned) b & md))->e) > 16)
1934 t->v.t + ((unsigned) b & mask_bits[e]))->e) >
1938 d = w - t->v.n - ((unsigned) b & mask_bits[e]);
1940 Tracevv((stderr, "\\[%d,%d]", w - d, n));
1946 WSIZE - ((d &= WSIZE - 1) > w ? d : w)) >
1948 #if !defined(NOMEMCPY) && !defined(DEBUG)
1949 if (w - d >= e) { /* (this test assumes unsigned comparison) */
1950 memcpy(slide + w, slide + d, e);
1953 } else /* do it slow to avoid memcpy() overlap */
1954 #endif /* !NOMEMCPY */
1956 slide[w++] = slide[d++];
1957 Tracevv((stderr, "%c", slide[w - 1]));
1968 /* restore the globals from the locals */
1969 wp = w; /* restore global window pointer */
1970 bb = b; /* restore global bit buffer */
1979 int inflate_stored()
1980 /* "decompress" an inflated type 0 (stored) block. */
1982 unsigned n; /* number of bytes in block */
1983 unsigned w; /* current window position */
1984 register ulg b; /* bit buffer */
1985 register unsigned k; /* number of bits in bit buffer */
1988 /* make local copies of globals */
1989 b = bb; /* initialize bit buffer */
1991 w = wp; /* initialize window position */
1994 /* go to byte boundary */
1999 /* get the length and its complement */
2001 n = ((unsigned) b & 0xffff);
2004 if (n != (unsigned) ((~b) & 0xffff))
2005 return 1; /* error in compressed data */
2009 /* read and output the compressed data */
2012 slide[w++] = (uch) b;
2021 /* restore the globals from the locals */
2022 wp = w; /* restore global window pointer */
2023 bb = b; /* restore global bit buffer */
2031 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
2032 either replace this with a custom decoder, or at least precompute the
2035 int i; /* temporary variable */
2036 struct huft *tl; /* literal/length code table */
2037 struct huft *td; /* distance code table */
2038 int bl; /* lookup bits for tl */
2039 int bd; /* lookup bits for td */
2040 unsigned l[288]; /* length list for huft_build */
2043 /* set up literal table */
2044 for (i = 0; i < 144; i++)
2046 for (; i < 256; i++)
2048 for (; i < 280; i++)
2050 for (; i < 288; i++) /* make a complete, but wrong code set */
2053 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
2057 /* set up distance table */
2058 for (i = 0; i < 30; i++) /* make an incomplete code set */
2061 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1) {
2067 /* decompress until an end-of-block code */
2068 if (inflate_codes(tl, td, bl, bd))
2072 /* free the decoding tables, return */
2080 int inflate_dynamic()
2081 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
2083 int i; /* temporary variables */
2085 unsigned l; /* last length */
2086 unsigned m; /* mask for bit lengths table */
2087 unsigned n; /* number of lengths to get */
2088 struct huft *tl; /* literal/length code table */
2089 struct huft *td; /* distance code table */
2090 int bl; /* lookup bits for tl */
2091 int bd; /* lookup bits for td */
2092 unsigned nb; /* number of bit length codes */
2093 unsigned nl; /* number of literal/length codes */
2094 unsigned nd; /* number of distance codes */
2096 #ifdef PKZIP_BUG_WORKAROUND
2097 unsigned ll[288 + 32]; /* literal/length and distance code lengths */
2099 unsigned ll[286 + 30]; /* literal/length and distance code lengths */
2101 register ulg b; /* bit buffer */
2102 register unsigned k; /* number of bits in bit buffer */
2105 /* make local bit buffer */
2110 /* read in table lengths */
2112 nl = 257 + ((unsigned) b & 0x1f); /* number of literal/length codes */
2115 nd = 1 + ((unsigned) b & 0x1f); /* number of distance codes */
2118 nb = 4 + ((unsigned) b & 0xf); /* number of bit length codes */
2120 #ifdef PKZIP_BUG_WORKAROUND
2121 if (nl > 288 || nd > 32)
2123 if (nl > 286 || nd > 30)
2125 return 1; /* bad lengths */
2128 /* read in bit-length-code lengths */
2129 for (j = 0; j < nb; j++) {
2131 ll[border[j]] = (unsigned) b & 7;
2138 /* build decoding table for trees--single level, 7 bit lookup */
2140 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0) {
2143 return i; /* incomplete code set */
2147 /* read in literal and distance code lengths */
2151 while ((unsigned) i < n) {
2152 NEEDBITS((unsigned) bl)
2153 j = (td = tl + ((unsigned) b & m))->b;
2156 if (j < 16) /* length of code in bits (0..15) */
2157 ll[i++] = l = j; /* save last length in l */
2158 else if (j == 16) { /* repeat last length 3 to 6 times */
2160 j = 3 + ((unsigned) b & 3);
2162 if ((unsigned) i + j > n)
2166 } else if (j == 17) { /* 3 to 10 zero length codes */
2168 j = 3 + ((unsigned) b & 7);
2170 if ((unsigned) i + j > n)
2175 } else { /* j == 18: 11 to 138 zero length codes */
2178 j = 11 + ((unsigned) b & 0x7f);
2180 if ((unsigned) i + j > n)
2189 /* free decoding table for trees */
2193 /* restore the global bit buffer */
2198 /* build the decoding tables for literal/length and distance codes */
2200 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0) {
2202 fprintf(stderr, " incomplete literal tree\n");
2205 return i; /* incomplete code set */
2208 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0) {
2210 fprintf(stderr, " incomplete distance tree\n");
2211 #ifdef PKZIP_BUG_WORKAROUND
2218 return i; /* incomplete code set */
2223 /* decompress until an end-of-block code */
2224 if (inflate_codes(tl, td, bl, bd))
2228 /* free the decoding tables, return */
2236 int inflate_block(e)
2237 int *e; /* last block flag */
2239 /* decompress an inflated block */
2241 unsigned t; /* block type */
2242 register ulg b; /* bit buffer */
2243 register unsigned k; /* number of bits in bit buffer */
2246 /* make local bit buffer */
2251 /* read in last block bit */
2257 /* read in block type */
2259 t = (unsigned) b & 3;
2263 /* restore the global bit buffer */
2268 /* inflate that block type */
2270 return inflate_dynamic();
2272 return inflate_stored();
2274 return inflate_fixed();
2277 /* bad block type */
2284 /* decompress an inflated entry */
2286 int e; /* last block flag */
2287 int r; /* result code */
2288 unsigned h; /* maximum struct huft's malloc'ed */
2291 /* initialize window, bit buffer */
2297 /* decompress until the last block */
2301 if ((r = inflate_block(&e)) != 0)
2307 /* Undo too much lookahead. The next read will be byte aligned so we
2308 * can discard unused bits in the last meaningful byte.
2315 /* flush out slide */
2319 /* return success */
2321 fprintf(stderr, "<%u> ", h);