1 /* zcat : stripped version based on gzip sources
2 Sven Rudolph <sr1@inf.tu-dresden.de>
7 static const char zcat_usage[] = "zcat\n"
9 "\tuncompress gzipped data from stdin to stdout\n";
13 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
14 * Copyright (C) 1992-1993 Jean-loup Gailly
15 * The unzip code was written and put in the public domain by Mark Adler.
16 * Portions of the lzw code are derived from the public domain 'compress'
17 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
18 * Ken Turkowski, Dave Mack and Peter Jannesen.
20 * See the license_msg below and the file COPYING for the software license.
21 * See the file algorithm.doc for the compression algorithms and file formats.
25 static char *license_msg[] = {
26 " Copyright (C) 1992-1993 Jean-loup Gailly",
27 " This program is free software; you can redistribute it and/or modify",
28 " it under the terms of the GNU General Public License as published by",
29 " the Free Software Foundation; either version 2, or (at your option)",
30 " any later version.",
32 " This program is distributed in the hope that it will be useful,",
33 " but WITHOUT ANY WARRANTY; without even the implied warranty of",
34 " MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the",
35 " GNU General Public License for more details.",
37 " You should have received a copy of the GNU General Public License",
38 " along with this program; if not, write to the Free Software",
39 " Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.",
43 /* Compress files with zip algorithm and 'compress' interface.
44 * See usage() and help() functions below for all options.
46 * file.gz: compressed file with same mode, owner, and utimes
47 * or stdout with -c option or if stdin used as input.
48 * If the output file name had to be truncated, the original name is kept
49 * in the compressed file.
50 * On MSDOS, file.tmp -> file.tmz. On VMS, file.tmp -> file.tmp-gz.
52 * Using gz on MSDOS would create too many file name conflicts. For
53 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
54 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
55 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
56 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
58 * For the meaning of all compilation flags, see comments in Makefile.in.
62 #include <sys/types.h>
67 /* #include "tailor.h" */
69 /* tailor.h -- target dependent definitions
70 * Copyright (C) 1992-1993 Jean-loup Gailly.
71 * This is free software; you can redistribute it and/or modify it under the
72 * terms of the GNU General Public License, see the file COPYING.
75 /* The target dependent definitions should be defined here only.
76 * The target dependent functions should be defined in tailor.c.
81 #define get_char() get_byte()
82 #define put_char(c) put_byte(c)
84 /* #include "gzip.h" */
86 /* gzip.h -- common declarations for all gzip modules
87 * Copyright (C) 1992-1993 Jean-loup Gailly.
88 * This is free software; you can redistribute it and/or modify it under the
89 * terms of the GNU General Public License, see the file COPYING.
92 #if defined(__STDC__) || defined(PROTO)
93 # define OF(args) args
104 /* I don't like nested includes, but the string and io functions are used
108 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
110 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
113 # define memzero(s, n) memset ((voidp)(s), 0, (n))
115 # include <strings.h>
116 # define strchr index
117 # define strrchr rindex
118 # define memcpy(d, s, n) bcopy((s), (d), (n))
119 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
120 # define memzero(s, n) bzero((s), (n))
124 # define RETSIGTYPE void
129 typedef unsigned char uch;
130 typedef unsigned short ush;
131 typedef unsigned long ulg;
133 /* Return codes from gzip */
138 /* Compression methods (see algorithm.doc) */
141 extern int method; /* compression method */
143 /* To save memory for 16 bit systems, some arrays are overlaid between
144 * the various modules:
145 * deflate: prev+head window d_buf l_buf outbuf
146 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
147 * inflate: window inbuf
148 * unpack: window inbuf prefix_len
149 * unlzh: left+right window c_table inbuf c_len
150 * For compression, input is done in window[]. For decompression, output
151 * is done in window except for unlzw.
156 # define INBUFSIZ 0x2000 /* input buffer size */
158 # define INBUFSIZ 0x8000 /* input buffer size */
161 #define INBUF_EXTRA 64 /* required by unlzw() */
165 # define OUTBUFSIZ 8192 /* output buffer size */
167 # define OUTBUFSIZ 16384 /* output buffer size */
170 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
176 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
178 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
182 /*#define DYN_ALLOC*/
185 # define EXTERN(type, array) extern type * array
186 # define DECLARE(type, array, size) type * array
187 # define ALLOC(type, array, size) { \
188 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
189 if (array == NULL) error("insufficient memory"); \
191 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
193 # define EXTERN(type, array) extern type array[]
194 # define DECLARE(type, array, size) type array[size]
195 # define ALLOC(type, array, size)
199 EXTERN(uch, inbuf); /* input buffer */
200 EXTERN(uch, outbuf); /* output buffer */
201 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
202 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
203 #define tab_suffix window
205 # define tab_prefix prev /* hash link (see deflate.c) */
206 # define head (prev+WSIZE) /* hash head (see deflate.c) */
207 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
209 # define tab_prefix0 prev
210 # define head tab_prefix1
211 EXTERN(ush, tab_prefix0); /* prefix for even codes */
212 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
215 extern unsigned insize; /* valid bytes in inbuf */
216 extern unsigned inptr; /* index of next byte to be processed in inbuf */
217 extern unsigned outcnt; /* bytes in output buffer */
219 extern long bytes_in; /* number of input bytes */
220 extern long bytes_out; /* number of output bytes */
221 extern long header_bytes;/* number of bytes in gzip header */
223 extern int ifd; /* input file descriptor */
224 extern int ofd; /* output file descriptor */
226 extern long ifile_size; /* input file size, -1 for devices (debug only) */
228 typedef int file_t; /* Do not use stdio */
229 #define NO_FILE (-1) /* in memory compression */
232 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
235 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
236 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
237 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
238 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
239 #define COMMENT 0x10 /* bit 4 set: file comment present */
240 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
241 #define RESERVED 0xC0 /* bit 6,7: reserved */
244 # define WSIZE 0x8000 /* window size--must be a power of two, and */
245 #endif /* at least 32K for zip's deflate method */
248 #define MAX_MATCH 258
249 /* The minimum and maximum match lengths */
251 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
252 /* Minimum amount of lookahead, except at the end of the input file.
253 * See deflate.c for comments about the MIN_MATCH+1.
256 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
257 /* In order to simplify the code, particularly on 16 bit machines, match
258 * distances are limited to MAX_DIST instead of WSIZE.
261 extern int exit_code; /* program exit code */
262 extern int verbose; /* be verbose (-v) */
263 extern int level; /* compression level */
264 extern int test; /* check .z file integrity */
265 extern int to_stdout; /* output to stdout (-c) */
266 extern int save_orig_name; /* set if original name must be saved */
268 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
269 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
271 /* put_byte is used for the compressed output, put_ubyte for the
272 * uncompressed output. However unlzw() uses window for its
273 * suffix table instead of its output buffer, so it does not use put_ubyte
274 * (to be cleaned up).
276 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
278 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
281 /* Output a 16 bit value, lsb first */
282 #define put_short(w) \
283 { if (outcnt < OUTBUFSIZ-2) { \
284 outbuf[outcnt++] = (uch) ((w) & 0xff); \
285 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
287 put_byte((uch)((w) & 0xff)); \
288 put_byte((uch)((ush)(w) >> 8)); \
292 /* Output a 32 bit value to the bit stream, lsb first */
293 #define put_long(n) { \
294 put_short((n) & 0xffff); \
295 put_short(((ulg)(n)) >> 16); \
298 #define seekable() 0 /* force sequential output */
299 #define translate_eol 0 /* no option -a yet */
301 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
303 /* Macros for getting two-byte and four-byte header values */
304 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
305 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
307 /* Diagnostic functions */
309 # define Assert(cond,msg) {if(!(cond)) error(msg);}
310 # define Trace(x) fprintf x
311 # define Tracev(x) {if (verbose) fprintf x ;}
312 # define Tracevv(x) {if (verbose>1) fprintf x ;}
313 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
314 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
316 # define Assert(cond,msg)
321 # define Tracecv(c,x)
324 #define WARN(msg) {fprintf msg ; \
325 if (exit_code == OK) exit_code = WARNING;}
328 extern int unzip OF((int in, int out));
331 RETSIGTYPE abort_gzip OF((void));
334 void lm_init OF((int pack_level, ush *flags));
335 ulg deflate OF((void));
338 void ct_init OF((ush *attr, int *method));
339 int ct_tally OF((int dist, int lc));
340 ulg flush_block OF((char *buf, ulg stored_len, int eof));
343 void bi_init OF((file_t zipfile));
344 void send_bits OF((int value, int length));
345 unsigned bi_reverse OF((unsigned value, int length));
346 void bi_windup OF((void));
347 void copy_block OF((char *buf, unsigned len, int header));
348 extern int (*read_buf) OF((char *buf, unsigned size));
351 extern int copy OF((int in, int out));
352 extern ulg updcrc OF((uch *s, unsigned n));
353 extern void clear_bufs OF((void));
354 extern int fill_inbuf OF((int eof_ok));
355 extern void flush_outbuf OF((void));
356 extern void flush_window OF((void));
357 extern void write_buf OF((int fd, voidp buf, unsigned cnt));
359 extern char *basename OF((char *fname));
360 #endif /* not __linux__ */
361 extern void error OF((char *m));
362 extern void warn OF((char *a, char *b));
363 extern void read_error OF((void));
364 extern void write_error OF((void));
365 extern voidp xmalloc OF((unsigned int size));
368 extern int inflate OF((void));
370 /* #include "lzw.h" */
372 /* lzw.h -- define the lzw functions.
373 * Copyright (C) 1992-1993 Jean-loup Gailly.
374 * This is free software; you can redistribute it and/or modify it under the
375 * terms of the GNU General Public License, see the file COPYING.
378 #if !defined(OF) && defined(lint)
385 #define INIT_BITS 9 /* Initial number of bits per code */
387 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
389 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
390 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
391 * It's a pity that old uncompress does not check bit 0x20. That makes
392 * extension of the format actually undesirable because old compress
393 * would just crash on the new format instead of giving a meaningful
394 * error message. It does check the number of bits, but it's more
395 * helpful to say "unsupported format, get a new version" than
396 * "can only handle 16 bits".
399 #define BLOCK_MODE 0x80
400 /* Block compression: if table is full and compression rate is dropping,
401 * clear the dictionary.
404 #define LZW_RESERVED 0x60 /* reserved bits */
406 #define CLEAR 256 /* flush the dictionary */
407 #define FIRST (CLEAR+1) /* first free entry */
409 extern int maxbits; /* max bits per code for LZW */
410 extern int block_mode; /* block compress mode -C compatible with 2.0 */
412 extern int lzw OF((int in, int out));
413 extern int unlzw OF((int in, int out));
416 /* #include "revision.h" */
418 /* revision.h -- define the version number
419 * Copyright (C) 1992-1993 Jean-loup Gailly.
420 * This is free software; you can redistribute it and/or modify it under the
421 * terms of the GNU General Public License, see the file COPYING.
424 #define VERSION "1.2.4"
426 #define REVDATE "18 Aug 93"
428 /* This version does not support compression into old compress format: */
433 /* #include "getopt.h" */
435 /* Declarations for getopt.
436 Copyright (C) 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc.
438 This program is free software; you can redistribute it and/or modify it
439 under the terms of the GNU General Public License as published by the
440 Free Software Foundation; either version 2, or (at your option) any
443 This program is distributed in the hope that it will be useful,
444 but WITHOUT ANY WARRANTY; without even the implied warranty of
445 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
446 GNU General Public License for more details.
448 You should have received a copy of the GNU General Public License
449 along with this program; if not, write to the Free Software
450 Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
459 /* For communication from `getopt' to the caller.
460 When `getopt' finds an option that takes an argument,
461 the argument value is returned here.
462 Also, when `ordering' is RETURN_IN_ORDER,
463 each non-option ARGV-element is returned here. */
467 /* Index in ARGV of the next element to be scanned.
468 This is used for communication to and from the caller
469 and for communication between successive calls to `getopt'.
471 On entry to `getopt', zero means this is the first call; initialize.
473 When `getopt' returns EOF, this is the index of the first of the
474 non-option elements that the caller should itself scan.
476 Otherwise, `optind' communicates from one call to the next
477 how much of ARGV has been scanned so far. */
481 /* Callers store zero here to inhibit the error message `getopt' prints
482 for unrecognized options. */
486 /* Set to an option character which was unrecognized. */
490 /* Describe the long-named options requested by the application.
491 The LONG_OPTIONS argument to getopt_long or getopt_long_only is a vector
492 of `struct option' terminated by an element containing a name which is
495 The field `has_arg' is:
496 no_argument (or 0) if the option does not take an argument,
497 required_argument (or 1) if the option requires an argument,
498 optional_argument (or 2) if the option takes an optional argument.
500 If the field `flag' is not NULL, it points to a variable that is set
501 to the value given in the field `val' when the option is found, but
502 left unchanged if the option is not found.
504 To have a long-named option do something other than set an `int' to
505 a compiled-in constant, such as set a value from `optarg', set the
506 option's `flag' field to zero and its `val' field to a nonzero
507 value (the equivalent single-letter option character, if there is
508 one). For long options that have a zero `flag' field, `getopt'
509 returns the contents of the `val' field. */
518 /* has_arg can't be an enum because some compilers complain about
519 type mismatches in all the code that assumes it is an int. */
525 /* Names for the values of the `has_arg' field of `struct option'. */
527 #define no_argument 0
528 #define required_argument 1
529 #define optional_argument 2
531 #if __STDC__ || defined(PROTO)
532 #if defined(__GNU_LIBRARY__)
533 /* Many other libraries have conflicting prototypes for getopt, with
534 differences in the consts, in stdlib.h. To avoid compilation
535 errors, only prototype getopt for the GNU C library. */
536 extern int getopt (int argc, char *const *argv, const char *shortopts);
537 #endif /* not __GNU_LIBRARY__ */
538 extern int getopt_long (int argc, char *const *argv, const char *shortopts,
539 const struct option *longopts, int *longind);
540 extern int getopt_long_only (int argc, char *const *argv,
541 const char *shortopts,
542 const struct option *longopts, int *longind);
544 /* Internal only. Users should not call this directly. */
545 extern int _getopt_internal (int argc, char *const *argv,
546 const char *shortopts,
547 const struct option *longopts, int *longind,
549 #else /* not __STDC__ */
550 extern int getopt ();
551 extern int getopt_long ();
552 extern int getopt_long_only ();
554 extern int _getopt_internal ();
555 #endif /* not __STDC__ */
561 #endif /* _GETOPT_H */
572 typedef struct dirent dir_type;
573 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
574 # define DIR_OPT "DIRENT"
576 # define NLENGTH(dirent) ((dirent)->d_namlen)
578 # include <sys/dir.h>
579 typedef struct direct dir_type;
580 # define DIR_OPT "SYSDIR"
583 # include <sys/ndir.h>
584 typedef struct direct dir_type;
585 # define DIR_OPT "SYSNDIR"
589 typedef struct direct dir_type;
590 # define DIR_OPT "NDIR"
593 # define DIR_OPT "NO_DIR"
599 #if !defined(S_ISDIR) && defined(S_IFDIR)
600 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
602 #if !defined(S_ISREG) && defined(S_IFREG)
603 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
606 typedef RETSIGTYPE (*sig_type) OF((int));
609 # define O_BINARY 0 /* creation mode for open() */
613 /* Pure BSD system? */
614 # include <sys/file.h>
616 # define O_CREAT FCREAT
619 # define O_EXCL FEXCL
624 # define S_IRUSR 0400
627 # define S_IWUSR 0200
629 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
632 # define MAX_PATH_LEN 1024 /* max pathname length */
641 off_t lseek OF((int fd, off_t offset, int whence));
647 DECLARE(uch, inbuf, INBUFSIZ +INBUF_EXTRA);
648 DECLARE(uch, outbuf, OUTBUFSIZ+OUTBUF_EXTRA);
649 DECLARE(ush, d_buf, DIST_BUFSIZE);
650 DECLARE(uch, window, 2L*WSIZE);
652 DECLARE(ush, tab_prefix, 1L<<BITS);
654 DECLARE(ush, tab_prefix0, 1L<<(BITS-1));
655 DECLARE(ush, tab_prefix1, 1L<<(BITS-1));
658 /* local variables */
660 int force = 0; /* don't ask questions, compress links (-f) */
661 int foreground; /* set if program run in foreground */
662 int maxbits = BITS; /* max bits per code for LZW */
663 int method = DEFLATED;/* compression method */
664 int exit_code = OK; /* program exit code */
665 int last_member; /* set for .zip and .Z files */
666 int part_nb; /* number of parts in .gz file */
667 long ifile_size; /* input file size, -1 for devices (debug only) */
669 long bytes_in; /* number of input bytes */
670 long bytes_out; /* number of output bytes */
671 long total_in = 0; /* input bytes for all files */
672 long total_out = 0; /* output bytes for all files */
673 struct stat istat; /* status for input file */
674 int ifd; /* input file descriptor */
675 int ofd; /* output file descriptor */
676 unsigned insize; /* valid bytes in inbuf */
677 unsigned inptr; /* index of next byte to be processed in inbuf */
678 unsigned outcnt; /* bytes in output buffer */
680 long header_bytes; /* number of bytes in gzip header */
682 /* local functions */
684 local void treat_stdin OF((void));
685 local int get_method OF((int in));
686 local void do_exit OF((int exitcode));
687 int main OF((int argc, char **argv));
688 int (*work) OF((int infile, int outfile)) = unzip; /* function to call */
690 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
692 /* ======================================================================== */
693 int zcat_main (int argc, char * * argv)
695 int file_count; /* number of files to precess */
697 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
699 (void) signal (SIGINT, (sig_type)abort_gzip);
702 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
703 (void) signal(SIGTERM, (sig_type)abort_gzip);
707 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
708 (void) signal(SIGHUP, (sig_type)abort_gzip);
712 file_count = argc - optind;
714 /* Allocate all global buffers (for DYN_ALLOC option) */
715 ALLOC(uch, inbuf, INBUFSIZ +INBUF_EXTRA);
716 ALLOC(uch, outbuf, OUTBUFSIZ+OUTBUF_EXTRA);
717 ALLOC(ush, d_buf, DIST_BUFSIZE);
718 ALLOC(uch, window, 2L*WSIZE);
720 ALLOC(ush, tab_prefix, 1L<<BITS);
722 ALLOC(ush, tab_prefix0, 1L<<(BITS-1));
723 ALLOC(ush, tab_prefix1, 1L<<(BITS-1));
726 /* And get to work */
730 return exit_code; /* just to avoid lint warning */
733 /* ========================================================================
734 * Compress or decompress stdin
736 local void treat_stdin()
738 ifile_size = -1L; /* convention for unknown size */
740 clear_bufs(); /* clear input and output buffers */
743 method = get_method(ifd);
746 do_exit(exit_code); /* error message already emitted */
749 (*work)(fileno(stdin), fileno(stdout));
754 /* ========================================================================
755 * Check the magic number of the input file and update ofname if an
756 * original name was given and to_stdout is not set.
757 * Return the compression method, -1 for error, -2 for warning.
758 * Set inptr to the offset of the next byte to be processed.
759 * Updates time_stamp if there is one and --no-time is not used.
760 * This function may be called repeatedly for an input file consisting
761 * of several contiguous gzip'ed members.
762 * IN assertions: there is at least one remaining compressed member.
763 * If the member is a zip file, it must be the only one.
765 local int get_method(in)
766 int in; /* input file descriptor */
768 uch flags; /* compression flags */
769 char magic[2]; /* magic header */
771 /* If --force and --stdout, zcat == cat, so do not complain about
772 * premature end of file: use try_byte instead of get_byte.
775 magic[0] = (char)try_byte();
776 magic[1] = (char)try_byte();
777 /* If try_byte returned EOF, magic[1] == 0xff */
779 magic[0] = (char)get_byte();
780 magic[1] = (char)get_byte();
782 method = -1; /* unknown yet */
783 part_nb++; /* number of parts in gzip file */
785 last_member = RECORD_IO;
786 /* assume multiple members in gzip file except for record oriented I/O */
788 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
790 method = (int)get_byte();
791 if (method != DEFLATED) {
793 "unknown method %d -- get newer version of gzip\n",
799 flags = (uch)get_byte();
801 (ulg)get_byte(); /* Ignore time stamp */
806 (void)get_byte(); /* Ignore extra flags for the moment */
807 (void)get_byte(); /* Ignore OS type for the moment */
809 if ((flags & EXTRA_FIELD) != 0) {
810 unsigned len = (unsigned)get_byte();
811 len |= ((unsigned)get_byte())<<8;
813 while (len--) (void)get_byte();
816 /* Discard original name if any */
817 if ((flags & ORIG_NAME) != 0) {
818 while (get_char() != 0) /* null */ ;
821 /* Discard file comment if any */
822 if ((flags & COMMENT) != 0) {
823 while (get_char() != 0) /* null */ ;
826 header_bytes = inptr + 2*sizeof(long); /* include crc and size */
831 if (method >= 0) return method;
834 fprintf(stderr, "\nnot in gzip format\n");
838 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
844 /* ========================================================================
845 * Free all dynamically allocated variables and exit with the given code.
847 local void do_exit(exitcode)
850 static int in_exit = 0;
852 if (in_exit) exit(exitcode);
867 /* ========================================================================
868 * Signal and error handler.
870 RETSIGTYPE abort_gzip()
874 /* unzip.c -- decompress files in gzip or pkzip format.
875 * Copyright (C) 1992-1993 Jean-loup Gailly
876 * This is free software; you can redistribute it and/or modify it under the
877 * terms of the GNU General Public License, see the file COPYING.
879 * The code in this file is derived from the file funzip.c written
880 * and put in the public domain by Mark Adler.
884 This version can extract files in gzip or pkzip format.
885 For the latter, only the first entry is extracted, and it has to be
886 either deflated or stored.
889 /* #include "crypt.h" */
891 /* crypt.h (dummy version) -- do not perform encryption
892 * Hardly worth copyrighting :-)
896 # undef CRYPT /* dummy version */
899 #define RAND_HEAD_LEN 12 /* length of encryption random header */
904 /* PKZIP header definitions */
905 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
906 #define LOCFLG 6 /* offset of bit flag */
907 #define CRPFLG 1 /* bit for encrypted entry */
908 #define EXTFLG 8 /* bit for extended local header */
909 #define LOCHOW 8 /* offset of compression method */
910 #define LOCTIM 10 /* file mod time (for decryption) */
911 #define LOCCRC 14 /* offset of crc */
912 #define LOCSIZ 18 /* offset of compressed size */
913 #define LOCLEN 22 /* offset of uncompressed length */
914 #define LOCFIL 26 /* offset of file name field length */
915 #define LOCEXT 28 /* offset of extra field length */
916 #define LOCHDR 30 /* size of local header, including sig */
917 #define EXTHDR 16 /* size of extended local header, inc sig */
922 char *key; /* not used--needed to link crypt.c */
923 int pkzip = 0; /* set for a pkzip file */
924 int ext_header = 0; /* set if extended local header */
926 /* ===========================================================================
927 * Unzip in to out. This routine works on both gzip and pkzip files.
929 * IN assertions: the buffer inbuf contains already the beginning of
930 * the compressed data, from offsets inptr to insize-1 included.
931 * The magic header has already been checked. The output buffer is cleared.
934 int in, out; /* input and output file descriptors */
936 ulg orig_crc = 0; /* original crc */
937 ulg orig_len = 0; /* original uncompressed length */
939 uch buf[EXTHDR]; /* extended local header */
944 updcrc(NULL, 0); /* initialize crc */
946 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
947 orig_crc = LG(inbuf + LOCCRC);
948 orig_len = LG(inbuf + LOCLEN);
952 if (method == DEFLATED) {
957 error("out of memory");
958 } else if (res != 0) {
959 error("invalid compressed data--format violated");
963 error("internal error, invalid method");
966 /* Get the crc and original length */
968 /* crc32 (see algorithm.doc)
969 * uncompressed input size modulo 2^32
971 for (n = 0; n < 8; n++) {
972 buf[n] = (uch)get_byte(); /* may cause an error if EOF */
975 orig_len = LG(buf+4);
977 } else if (ext_header) { /* If extended header, check it */
978 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
980 * compressed size 4-bytes
981 * uncompressed size 4-bytes
983 for (n = 0; n < EXTHDR; n++) {
984 buf[n] = (uch)get_byte(); /* may cause an error if EOF */
986 orig_crc = LG(buf+4);
987 orig_len = LG(buf+12);
990 /* Validate decompression */
991 if (orig_crc != updcrc(outbuf, 0)) {
992 error("invalid compressed data--crc error");
994 if (orig_len != (ulg)bytes_out) {
995 error("invalid compressed data--length error");
998 /* Check if there are more entries in a pkzip file */
999 if (pkzip && inptr + 4 < insize && LG(inbuf+inptr) == LOCSIG) {
1000 WARN((stderr,"has more than one entry--rest ignored\n"));
1002 ext_header = pkzip = 0; /* for next file */
1005 /* util.c -- utility functions for gzip support
1006 * Copyright (C) 1992-1993 Jean-loup Gailly
1007 * This is free software; you can redistribute it and/or modify it under the
1008 * terms of the GNU General Public License, see the file COPYING.
1013 #include <sys/types.h>
1015 #ifdef HAVE_UNISTD_H
1016 # include <unistd.h>
1022 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1023 # include <stdlib.h>
1028 static const ulg crc_32_tab[]; /* crc table, defined below */
1030 /* ===========================================================================
1031 * Run a set of bytes through the crc shift register. If s is a NULL
1032 * pointer, then initialize the crc shift register contents instead.
1033 * Return the current crc in either case.
1036 uch *s; /* pointer to bytes to pump through */
1037 unsigned n; /* number of bytes in s[] */
1039 register ulg c; /* temporary variable */
1041 static ulg crc = (ulg)0xffffffffL; /* shift register contents */
1048 c = crc_32_tab[((int)c ^ (*s++)) & 0xff] ^ (c >> 8);
1052 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1055 /* ===========================================================================
1056 * Clear input and output buffers
1062 bytes_in = bytes_out = 0L;
1065 /* ===========================================================================
1066 * Fill the input buffer. This is called only when the buffer is empty.
1068 int fill_inbuf(eof_ok)
1069 int eof_ok; /* set if EOF acceptable as a result */
1073 /* Read as much as possible */
1077 len = read(ifd, (char*)inbuf+insize, INBUFSIZ-insize);
1078 if (len == 0 || len == EOF) break;
1080 } while (insize < INBUFSIZ);
1083 if (eof_ok) return EOF;
1086 bytes_in += (ulg)insize;
1091 /* ===========================================================================
1092 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1093 * (used for the compressed data only)
1097 if (outcnt == 0) return;
1099 write_buf(ofd, (char *)outbuf, outcnt);
1100 bytes_out += (ulg)outcnt;
1104 /* ===========================================================================
1105 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1106 * (Used for the decompressed data only.)
1110 if (outcnt == 0) return;
1111 updcrc(window, outcnt);
1113 write_buf(ofd, (char *)window, outcnt);
1115 bytes_out += (ulg)outcnt;
1119 /* ===========================================================================
1120 * Does the same as write(), but also handles partial pipe writes and checks
1123 void write_buf(fd, buf, cnt)
1130 while ((n = write(fd, buf, cnt)) != cnt) {
1131 if (n == (unsigned)(-1)) {
1135 buf = (voidp)((char*)buf+n);
1139 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1141 /* Provide missing strspn and strcspn functions. */
1147 int strspn OF((const char *s, const char *accept));
1148 int strcspn OF((const char *s, const char *reject));
1150 /* ========================================================================
1151 * Return the length of the maximum initial segment
1152 * of s which contains only characters in accept.
1154 int strspn(s, accept)
1158 register const char *p;
1159 register const char *a;
1160 register int count = 0;
1162 for (p = s; *p != '\0'; ++p) {
1163 for (a = accept; *a != '\0'; ++a) {
1164 if (*p == *a) break;
1166 if (*a == '\0') return count;
1172 /* ========================================================================
1173 * Return the length of the maximum inital segment of s
1174 * which contains no characters from reject.
1176 int strcspn(s, reject)
1180 register int count = 0;
1182 while (*s != '\0') {
1183 if (strchr(reject, *s++) != NULL) return count;
1189 #endif /* NO_STRING_H */
1192 /* ========================================================================
1198 fprintf(stderr, "\n%s\n", m);
1203 char *a, *b; /* message strings juxtaposed in output */
1205 WARN((stderr, "warning: %s%s\n", a, b));
1210 fprintf(stderr, "\n");
1214 fprintf(stderr, "unexpected end of file\n");
1221 fprintf(stderr, "\n");
1227 /* ========================================================================
1228 * Semi-safe malloc -- never returns NULL.
1230 voidp xmalloc (size)
1233 voidp cp = (voidp)malloc (size);
1235 if (cp == NULL) error("out of memory");
1239 /* ========================================================================
1240 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1242 static const ulg crc_32_tab[] = {
1243 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1244 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1245 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1246 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1247 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1248 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1249 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1250 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1251 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1252 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1253 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1254 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1255 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1256 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1257 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1258 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1259 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1260 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1261 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1262 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1263 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1264 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1265 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1266 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1267 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1268 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1269 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1270 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1271 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1272 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1273 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1274 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1275 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1276 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1277 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1278 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1279 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1280 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1281 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1282 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1283 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1284 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1285 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1286 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1287 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1288 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1289 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1290 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1291 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1292 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1293 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1296 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1297 version c10p1, 10 January 1993 */
1299 /* You can do whatever you like with this source file, though I would
1300 prefer that if you modify it and redistribute it that you include
1301 comments to that effect with your name and the date. Thank you.
1302 [The history has been moved to the file ChangeLog.]
1306 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1307 method searches for as much of the current string of bytes (up to a
1308 length of 258) in the previous 32K bytes. If it doesn't find any
1309 matches (of at least length 3), it codes the next byte. Otherwise, it
1310 codes the length of the matched string and its distance backwards from
1311 the current position. There is a single Huffman code that codes both
1312 single bytes (called "literals") and match lengths. A second Huffman
1313 code codes the distance information, which follows a length code. Each
1314 length or distance code actually represents a base value and a number
1315 of "extra" (sometimes zero) bits to get to add to the base value. At
1316 the end of each deflated block is a special end-of-block (EOB) literal/
1317 length code. The decoding process is basically: get a literal/length
1318 code; if EOB then done; if a literal, emit the decoded byte; if a
1319 length then get the distance and emit the referred-to bytes from the
1320 sliding window of previously emitted data.
1322 There are (currently) three kinds of inflate blocks: stored, fixed, and
1323 dynamic. The compressor deals with some chunk of data at a time, and
1324 decides which method to use on a chunk-by-chunk basis. A chunk might
1325 typically be 32K or 64K. If the chunk is uncompressible, then the
1326 "stored" method is used. In this case, the bytes are simply stored as
1327 is, eight bits per byte, with none of the above coding. The bytes are
1328 preceded by a count, since there is no longer an EOB code.
1330 If the data is compressible, then either the fixed or dynamic methods
1331 are used. In the dynamic method, the compressed data is preceded by
1332 an encoding of the literal/length and distance Huffman codes that are
1333 to be used to decode this block. The representation is itself Huffman
1334 coded, and so is preceded by a description of that code. These code
1335 descriptions take up a little space, and so for small blocks, there is
1336 a predefined set of codes, called the fixed codes. The fixed method is
1337 used if the block codes up smaller that way (usually for quite small
1338 chunks), otherwise the dynamic method is used. In the latter case, the
1339 codes are customized to the probabilities in the current block, and so
1340 can code it much better than the pre-determined fixed codes.
1342 The Huffman codes themselves are decoded using a mutli-level table
1343 lookup, in order to maximize the speed of decoding plus the speed of
1344 building the decoding tables. See the comments below that precede the
1345 lbits and dbits tuning parameters.
1350 Notes beyond the 1.93a appnote.txt:
1352 1. Distance pointers never point before the beginning of the output
1354 2. Distance pointers can point back across blocks, up to 32k away.
1355 3. There is an implied maximum of 7 bits for the bit length table and
1356 15 bits for the actual data.
1357 4. If only one code exists, then it is encoded using one bit. (Zero
1358 would be more efficient, but perhaps a little confusing.) If two
1359 codes exist, they are coded using one bit each (0 and 1).
1360 5. There is no way of sending zero distance codes--a dummy must be
1361 sent if there are none. (History: a pre 2.0 version of PKZIP would
1362 store blocks with no distance codes, but this was discovered to be
1363 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1364 zero distance codes, which is sent as one code of zero bits in
1366 6. There are up to 286 literal/length codes. Code 256 represents the
1367 end-of-block. Note however that the static length tree defines
1368 288 codes just to fill out the Huffman codes. Codes 286 and 287
1369 cannot be used though, since there is no length base or extra bits
1370 defined for them. Similarly, there are up to 30 distance codes.
1371 However, static trees define 32 codes (all 5 bits) to fill out the
1372 Huffman codes, but the last two had better not show up in the data.
1373 7. Unzip can check dynamic Huffman blocks for complete code sets.
1374 The exception is that a single code would not be complete (see #4).
1375 8. The five bits following the block type is really the number of
1376 literal codes sent minus 257.
1377 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1378 (1+6+6). Therefore, to output three times the length, you output
1379 three codes (1+1+1), whereas to output four times the same length,
1380 you only need two codes (1+3). Hmm.
1381 10. In the tree reconstruction algorithm, Code = Code + Increment
1382 only if BitLength(i) is not zero. (Pretty obvious.)
1383 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1384 12. Note: length code 284 can represent 227-258, but length code 285
1385 really is 258. The last length deserves its own, short code
1386 since it gets used a lot in very redundant files. The length
1387 258 is special since 258 - 3 (the min match length) is 255.
1388 13. The literal/length and distance code bit lengths are read as a
1389 single stream of lengths. It is possible (and advantageous) for
1390 a repeat code (16, 17, or 18) to go across the boundary between
1391 the two sets of lengths.
1394 #include <sys/types.h>
1396 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1397 # include <stdlib.h>
1401 #define slide window
1403 /* Huffman code lookup table entry--this entry is four bytes for machines
1404 that have 16-bit pointers (e.g. PC's in the small or medium model).
1405 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1406 means that v is a literal, 16 < e < 32 means that v is a pointer to
1407 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1408 an unused code. If a code with e == 99 is looked up, this implies an
1409 error in the data. */
1411 uch e; /* number of extra bits or operation */
1412 uch b; /* number of bits in this code or subcode */
1414 ush n; /* literal, length base, or distance base */
1415 struct huft *t; /* pointer to next level of table */
1420 /* Function prototypes */
1421 int huft_build OF((unsigned *, unsigned, unsigned, ush *, ush *,
1422 struct huft **, int *));
1423 int huft_free OF((struct huft *));
1424 int inflate_codes OF((struct huft *, struct huft *, int, int));
1425 int inflate_stored OF((void));
1426 int inflate_fixed OF((void));
1427 int inflate_dynamic OF((void));
1428 int inflate_block OF((int *));
1429 int inflate OF((void));
1432 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1433 stream to find repeated byte strings. This is implemented here as a
1434 circular buffer. The index is updated simply by incrementing and then
1435 and'ing with 0x7fff (32K-1). */
1436 /* It is left to other modules to supply the 32K area. It is assumed
1437 to be usable as if it were declared "uch slide[32768];" or as just
1438 "uch *slide;" and then malloc'ed in the latter case. The definition
1439 must be in unzip.h, included above. */
1440 /* unsigned wp; current position in slide */
1442 #define flush_output(w) (wp=(w),flush_window())
1444 /* Tables for deflate from PKZIP's appnote.txt. */
1445 static unsigned border[] = { /* Order of the bit length code lengths */
1446 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
1447 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1448 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1449 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0};
1450 /* note: see note #13 above about the 258 in this list. */
1451 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1452 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1453 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99}; /* 99==invalid */
1454 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1455 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1456 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1457 8193, 12289, 16385, 24577};
1458 static ush cpdext[] = { /* Extra bits for distance codes */
1459 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1460 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1465 /* Macros for inflate() bit peeking and grabbing.
1469 x = b & mask_bits[j];
1472 where NEEDBITS makes sure that b has at least j bits in it, and
1473 DUMPBITS removes the bits from b. The macros use the variable k
1474 for the number of bits in b. Normally, b and k are register
1475 variables for speed, and are initialized at the beginning of a
1476 routine that uses these macros from a global bit buffer and count.
1478 If we assume that EOB will be the longest code, then we will never
1479 ask for bits with NEEDBITS that are beyond the end of the stream.
1480 So, NEEDBITS should not read any more bytes than are needed to
1481 meet the request. Then no bytes need to be "returned" to the buffer
1482 at the end of the last block.
1484 However, this assumption is not true for fixed blocks--the EOB code
1485 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1486 (The EOB code is shorter than other codes because fixed blocks are
1487 generally short. So, while a block always has an EOB, many other
1488 literal/length codes have a significantly lower probability of
1489 showing up at all.) However, by making the first table have a
1490 lookup of seven bits, the EOB code will be found in that first
1491 lookup, and so will not require that too many bits be pulled from
1495 ulg bb; /* bit buffer */
1496 unsigned bk; /* bits in bit buffer */
1500 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1501 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1506 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1508 # define NEXTBYTE() (uch)get_byte()
1510 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1511 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1515 Huffman code decoding is performed using a multi-level table lookup.
1516 The fastest way to decode is to simply build a lookup table whose
1517 size is determined by the longest code. However, the time it takes
1518 to build this table can also be a factor if the data being decoded
1519 is not very long. The most common codes are necessarily the
1520 shortest codes, so those codes dominate the decoding time, and hence
1521 the speed. The idea is you can have a shorter table that decodes the
1522 shorter, more probable codes, and then point to subsidiary tables for
1523 the longer codes. The time it costs to decode the longer codes is
1524 then traded against the time it takes to make longer tables.
1526 This results of this trade are in the variables lbits and dbits
1527 below. lbits is the number of bits the first level table for literal/
1528 length codes can decode in one step, and dbits is the same thing for
1529 the distance codes. Subsequent tables are also less than or equal to
1530 those sizes. These values may be adjusted either when all of the
1531 codes are shorter than that, in which case the longest code length in
1532 bits is used, or when the shortest code is *longer* than the requested
1533 table size, in which case the length of the shortest code in bits is
1536 There are two different values for the two tables, since they code a
1537 different number of possibilities each. The literal/length table
1538 codes 286 possible values, or in a flat code, a little over eight
1539 bits. The distance table codes 30 possible values, or a little less
1540 than five bits, flat. The optimum values for speed end up being
1541 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1542 The optimum values may differ though from machine to machine, and
1543 possibly even between compilers. Your mileage may vary.
1547 int lbits = 9; /* bits in base literal/length lookup table */
1548 int dbits = 6; /* bits in base distance lookup table */
1551 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1552 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1553 #define N_MAX 288 /* maximum number of codes in any set */
1556 unsigned hufts; /* track memory usage */
1559 int huft_build(b, n, s, d, e, t, m)
1560 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1561 unsigned n; /* number of codes (assumed <= N_MAX) */
1562 unsigned s; /* number of simple-valued codes (0..s-1) */
1563 ush *d; /* list of base values for non-simple codes */
1564 ush *e; /* list of extra bits for non-simple codes */
1565 struct huft **t; /* result: starting table */
1566 int *m; /* maximum lookup bits, returns actual */
1567 /* Given a list of code lengths and a maximum table size, make a set of
1568 tables to decode that set of codes. Return zero on success, one if
1569 the given code set is incomplete (the tables are still built in this
1570 case), two if the input is invalid (all zero length codes or an
1571 oversubscribed set of lengths), and three if not enough memory. */
1573 unsigned a; /* counter for codes of length k */
1574 unsigned c[BMAX+1]; /* bit length count table */
1575 unsigned f; /* i repeats in table every f entries */
1576 int g; /* maximum code length */
1577 int h; /* table level */
1578 register unsigned i; /* counter, current code */
1579 register unsigned j; /* counter */
1580 register int k; /* number of bits in current code */
1581 int l; /* bits per table (returned in m) */
1582 register unsigned *p; /* pointer into c[], b[], or v[] */
1583 register struct huft *q; /* points to current table */
1584 struct huft r; /* table entry for structure assignment */
1585 struct huft *u[BMAX]; /* table stack */
1586 unsigned v[N_MAX]; /* values in order of bit length */
1587 register int w; /* bits before this table == (l * h) */
1588 unsigned x[BMAX+1]; /* bit offsets, then code stack */
1589 unsigned *xp; /* pointer into x */
1590 int y; /* number of dummy codes added */
1591 unsigned z; /* number of entries in current table */
1594 /* Generate counts for each bit length */
1595 memzero(c, sizeof(c));
1598 Tracecv(*p, (stderr, (n-i >= ' ' && n-i <= '~' ? "%c %d\n" : "0x%x %d\n"),
1600 c[*p]++; /* assume all entries <= BMAX */
1601 p++; /* Can't combine with above line (Solaris bug) */
1603 if (c[0] == n) /* null input--all zero length codes */
1605 *t = (struct huft *)NULL;
1611 /* Find minimum and maximum length, bound *m by those */
1613 for (j = 1; j <= BMAX; j++)
1616 k = j; /* minimum code length */
1617 if ((unsigned)l < j)
1619 for (i = BMAX; i; i--)
1622 g = i; /* maximum code length */
1623 if ((unsigned)l > i)
1628 /* Adjust last length count to fill out codes, if needed */
1629 for (y = 1 << j; j < i; j++, y <<= 1)
1630 if ((y -= c[j]) < 0)
1631 return 2; /* bad input: more codes than bits */
1632 if ((y -= c[i]) < 0)
1637 /* Generate starting offsets into the value table for each length */
1639 p = c + 1; xp = x + 2;
1640 while (--i) { /* note that i == g from above */
1641 *xp++ = (j += *p++);
1645 /* Make a table of values in order of bit lengths */
1648 if ((j = *p++) != 0)
1653 /* Generate the Huffman codes and for each, make the table entries */
1654 x[0] = i = 0; /* first Huffman code is zero */
1655 p = v; /* grab values in bit order */
1656 h = -1; /* no tables yet--level -1 */
1657 w = -l; /* bits decoded == (l * h) */
1658 u[0] = (struct huft *)NULL; /* just to keep compilers happy */
1659 q = (struct huft *)NULL; /* ditto */
1662 /* go through the bit lengths (k already is bits in shortest code) */
1668 /* here i is the Huffman code of length k bits for value *p */
1669 /* make tables up to required level */
1673 w += l; /* previous table always l bits */
1675 /* compute minimum size table less than or equal to l bits */
1676 z = (z = g - w) > (unsigned)l ? l : z; /* upper limit on table size */
1677 if ((f = 1 << (j = k - w)) > a + 1) /* try a k-w bit table */
1678 { /* too few codes for k-w bit table */
1679 f -= a + 1; /* deduct codes from patterns left */
1681 while (++j < z) /* try smaller tables up to z bits */
1683 if ((f <<= 1) <= *++xp)
1684 break; /* enough codes to use up j bits */
1685 f -= *xp; /* else deduct codes from patterns */
1688 z = 1 << j; /* table entries for j-bit table */
1690 /* allocate and link in new table */
1691 if ((q = (struct huft *)malloc((z + 1)*sizeof(struct huft))) ==
1692 (struct huft *)NULL)
1696 return 3; /* not enough memory */
1698 hufts += z + 1; /* track memory usage */
1699 *t = q + 1; /* link to list for huft_free() */
1700 *(t = &(q->v.t)) = (struct huft *)NULL;
1701 u[h] = ++q; /* table starts after link */
1703 /* connect to last table, if there is one */
1706 x[h] = i; /* save pattern for backing up */
1707 r.b = (uch)l; /* bits to dump before this table */
1708 r.e = (uch)(16 + j); /* bits in this table */
1709 r.v.t = q; /* pointer to this table */
1710 j = i >> (w - l); /* (get around Turbo C bug) */
1711 u[h-1][j] = r; /* connect to last table */
1715 /* set up table entry in r */
1718 r.e = 99; /* out of values--invalid code */
1721 r.e = (uch)(*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1722 r.v.n = (ush)(*p); /* simple code is just the value */
1723 p++; /* one compiler does not like *p++ */
1727 r.e = (uch)e[*p - s]; /* non-simple--look up in lists */
1728 r.v.n = d[*p++ - s];
1731 /* fill code-like entries with r */
1733 for (j = i >> w; j < z; j += f)
1736 /* backwards increment the k-bit code i */
1737 for (j = 1 << (k - 1); i & j; j >>= 1)
1741 /* backup over finished tables */
1742 while ((i & ((1 << w) - 1)) != x[h])
1744 h--; /* don't need to update q */
1751 /* Return true (1) if we were given an incomplete table */
1752 return y != 0 && g != 1;
1758 struct huft *t; /* table to free */
1759 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1760 list of the tables it made, with the links in a dummy first entry of
1763 register struct huft *p, *q;
1766 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1768 while (p != (struct huft *)NULL)
1778 int inflate_codes(tl, td, bl, bd)
1779 struct huft *tl, *td; /* literal/length and distance decoder tables */
1780 int bl, bd; /* number of bits decoded by tl[] and td[] */
1781 /* inflate (decompress) the codes in a deflated (compressed) block.
1782 Return an error code or zero if it all goes ok. */
1784 register unsigned e; /* table entry flag/number of extra bits */
1785 unsigned n, d; /* length and index for copy */
1786 unsigned w; /* current window position */
1787 struct huft *t; /* pointer to table entry */
1788 unsigned ml, md; /* masks for bl and bd bits */
1789 register ulg b; /* bit buffer */
1790 register unsigned k; /* number of bits in bit buffer */
1793 /* make local copies of globals */
1794 b = bb; /* initialize bit buffer */
1796 w = wp; /* initialize window position */
1798 /* inflate the coded data */
1799 ml = mask_bits[bl]; /* precompute masks for speed */
1801 for (;;) /* do until end of block */
1803 NEEDBITS((unsigned)bl)
1804 if ((e = (t = tl + ((unsigned)b & ml))->e) > 16)
1811 } while ((e = (t = t->v.t + ((unsigned)b & mask_bits[e]))->e) > 16);
1813 if (e == 16) /* then it's a literal */
1815 slide[w++] = (uch)t->v.n;
1816 Tracevv((stderr, "%c", slide[w-1]));
1823 else /* it's an EOB or a length */
1825 /* exit if end of block */
1829 /* get length of block to copy */
1831 n = t->v.n + ((unsigned)b & mask_bits[e]);
1834 /* decode distance of block to copy */
1835 NEEDBITS((unsigned)bd)
1836 if ((e = (t = td + ((unsigned)b & md))->e) > 16)
1843 } while ((e = (t = t->v.t + ((unsigned)b & mask_bits[e]))->e) > 16);
1846 d = w - t->v.n - ((unsigned)b & mask_bits[e]);
1848 Tracevv((stderr,"\\[%d,%d]", w-d, n));
1852 n -= (e = (e = WSIZE - ((d &= WSIZE-1) > w ? d : w)) > n ? n : e);
1853 #if !defined(NOMEMCPY) && !defined(DEBUG)
1854 if (w - d >= e) /* (this test assumes unsigned comparison) */
1856 memcpy(slide + w, slide + d, e);
1860 else /* do it slow to avoid memcpy() overlap */
1861 #endif /* !NOMEMCPY */
1863 slide[w++] = slide[d++];
1864 Tracevv((stderr, "%c", slide[w-1]));
1876 /* restore the globals from the locals */
1877 wp = w; /* restore global window pointer */
1878 bb = b; /* restore global bit buffer */
1887 int inflate_stored()
1888 /* "decompress" an inflated type 0 (stored) block. */
1890 unsigned n; /* number of bytes in block */
1891 unsigned w; /* current window position */
1892 register ulg b; /* bit buffer */
1893 register unsigned k; /* number of bits in bit buffer */
1896 /* make local copies of globals */
1897 b = bb; /* initialize bit buffer */
1899 w = wp; /* initialize window position */
1902 /* go to byte boundary */
1907 /* get the length and its complement */
1909 n = ((unsigned)b & 0xffff);
1912 if (n != (unsigned)((~b) & 0xffff))
1913 return 1; /* error in compressed data */
1917 /* read and output the compressed data */
1921 slide[w++] = (uch)b;
1931 /* restore the globals from the locals */
1932 wp = w; /* restore global window pointer */
1933 bb = b; /* restore global bit buffer */
1941 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
1942 either replace this with a custom decoder, or at least precompute the
1945 int i; /* temporary variable */
1946 struct huft *tl; /* literal/length code table */
1947 struct huft *td; /* distance code table */
1948 int bl; /* lookup bits for tl */
1949 int bd; /* lookup bits for td */
1950 unsigned l[288]; /* length list for huft_build */
1953 /* set up literal table */
1954 for (i = 0; i < 144; i++)
1956 for (; i < 256; i++)
1958 for (; i < 280; i++)
1960 for (; i < 288; i++) /* make a complete, but wrong code set */
1963 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
1967 /* set up distance table */
1968 for (i = 0; i < 30; i++) /* make an incomplete code set */
1971 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1)
1978 /* decompress until an end-of-block code */
1979 if (inflate_codes(tl, td, bl, bd))
1983 /* free the decoding tables, return */
1991 int inflate_dynamic()
1992 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
1994 int i; /* temporary variables */
1996 unsigned l; /* last length */
1997 unsigned m; /* mask for bit lengths table */
1998 unsigned n; /* number of lengths to get */
1999 struct huft *tl; /* literal/length code table */
2000 struct huft *td; /* distance code table */
2001 int bl; /* lookup bits for tl */
2002 int bd; /* lookup bits for td */
2003 unsigned nb; /* number of bit length codes */
2004 unsigned nl; /* number of literal/length codes */
2005 unsigned nd; /* number of distance codes */
2006 #ifdef PKZIP_BUG_WORKAROUND
2007 unsigned ll[288+32]; /* literal/length and distance code lengths */
2009 unsigned ll[286+30]; /* literal/length and distance code lengths */
2011 register ulg b; /* bit buffer */
2012 register unsigned k; /* number of bits in bit buffer */
2015 /* make local bit buffer */
2020 /* read in table lengths */
2022 nl = 257 + ((unsigned)b & 0x1f); /* number of literal/length codes */
2025 nd = 1 + ((unsigned)b & 0x1f); /* number of distance codes */
2028 nb = 4 + ((unsigned)b & 0xf); /* number of bit length codes */
2030 #ifdef PKZIP_BUG_WORKAROUND
2031 if (nl > 288 || nd > 32)
2033 if (nl > 286 || nd > 30)
2035 return 1; /* bad lengths */
2038 /* read in bit-length-code lengths */
2039 for (j = 0; j < nb; j++)
2042 ll[border[j]] = (unsigned)b & 7;
2049 /* build decoding table for trees--single level, 7 bit lookup */
2051 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0)
2055 return i; /* incomplete code set */
2059 /* read in literal and distance code lengths */
2063 while ((unsigned)i < n)
2065 NEEDBITS((unsigned)bl)
2066 j = (td = tl + ((unsigned)b & m))->b;
2069 if (j < 16) /* length of code in bits (0..15) */
2070 ll[i++] = l = j; /* save last length in l */
2071 else if (j == 16) /* repeat last length 3 to 6 times */
2074 j = 3 + ((unsigned)b & 3);
2076 if ((unsigned)i + j > n)
2081 else if (j == 17) /* 3 to 10 zero length codes */
2084 j = 3 + ((unsigned)b & 7);
2086 if ((unsigned)i + j > n)
2092 else /* j == 18: 11 to 138 zero length codes */
2095 j = 11 + ((unsigned)b & 0x7f);
2097 if ((unsigned)i + j > n)
2106 /* free decoding table for trees */
2110 /* restore the global bit buffer */
2115 /* build the decoding tables for literal/length and distance codes */
2117 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0)
2120 fprintf(stderr, " incomplete literal tree\n");
2123 return i; /* incomplete code set */
2126 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0)
2129 fprintf(stderr, " incomplete distance tree\n");
2130 #ifdef PKZIP_BUG_WORKAROUND
2137 return i; /* incomplete code set */
2142 /* decompress until an end-of-block code */
2143 if (inflate_codes(tl, td, bl, bd))
2147 /* free the decoding tables, return */
2155 int inflate_block(e)
2156 int *e; /* last block flag */
2157 /* decompress an inflated block */
2159 unsigned t; /* block type */
2160 register ulg b; /* bit buffer */
2161 register unsigned k; /* number of bits in bit buffer */
2164 /* make local bit buffer */
2169 /* read in last block bit */
2175 /* read in block type */
2177 t = (unsigned)b & 3;
2181 /* restore the global bit buffer */
2186 /* inflate that block type */
2188 return inflate_dynamic();
2190 return inflate_stored();
2192 return inflate_fixed();
2195 /* bad block type */
2202 /* decompress an inflated entry */
2204 int e; /* last block flag */
2205 int r; /* result code */
2206 unsigned h; /* maximum struct huft's malloc'ed */
2209 /* initialize window, bit buffer */
2215 /* decompress until the last block */
2219 if ((r = inflate_block(&e)) != 0)
2225 /* Undo too much lookahead. The next read will be byte aligned so we
2226 * can discard unused bits in the last meaningful byte.
2233 /* flush out slide */
2237 /* return success */
2239 fprintf(stderr, "<%u> ", h);