1 /* zcat : stripped version based on gzip sources
2 Sven Rudolph <sr1@inf.tu-dresden.de>
7 static const char gunzip_usage[] =
8 "gunzip [OPTION]... FILE\n\n"
9 "Uncompress FILE (or standard input if FILE is '-').\n\n"
11 "\t-c\tWrite output to standard output\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 long ifile_size; /* input file size, -1 for devices (debug only) */
225 typedef int file_t; /* Do not use stdio */
226 #define NO_FILE (-1) /* in memory compression */
229 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
232 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
233 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
234 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
235 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
236 #define COMMENT 0x10 /* bit 4 set: file comment present */
237 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
238 #define RESERVED 0xC0 /* bit 6,7: reserved */
241 # define WSIZE 0x8000 /* window size--must be a power of two, and */
242 #endif /* at least 32K for zip's deflate method */
245 #define MAX_MATCH 258
246 /* The minimum and maximum match lengths */
248 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
249 /* Minimum amount of lookahead, except at the end of the input file.
250 * See deflate.c for comments about the MIN_MATCH+1.
253 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
254 /* In order to simplify the code, particularly on 16 bit machines, match
255 * distances are limited to MAX_DIST instead of WSIZE.
258 extern int exit_code; /* program exit code */
259 extern int verbose; /* be verbose (-v) */
260 extern int level; /* compression level */
261 extern int test; /* check .z file integrity */
262 extern int save_orig_name; /* set if original name must be saved */
264 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
265 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
267 /* put_byte is used for the compressed output, put_ubyte for the
268 * uncompressed output. However unlzw() uses window for its
269 * suffix table instead of its output buffer, so it does not use put_ubyte
270 * (to be cleaned up).
272 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
274 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
277 /* Output a 16 bit value, lsb first */
278 #define put_short(w) \
279 { if (outcnt < OUTBUFSIZ-2) { \
280 outbuf[outcnt++] = (uch) ((w) & 0xff); \
281 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
283 put_byte((uch)((w) & 0xff)); \
284 put_byte((uch)((ush)(w) >> 8)); \
288 /* Output a 32 bit value to the bit stream, lsb first */
289 #define put_long(n) { \
290 put_short((n) & 0xffff); \
291 put_short(((ulg)(n)) >> 16); \
294 #define seekable() 0 /* force sequential output */
295 #define translate_eol 0 /* no option -a yet */
297 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
299 /* Macros for getting two-byte and four-byte header values */
300 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
301 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
303 /* Diagnostic functions */
305 # define Assert(cond,msg) {if(!(cond)) error(msg);}
306 # define Trace(x) fprintf x
307 # define Tracev(x) {if (verbose) fprintf x ;}
308 # define Tracevv(x) {if (verbose>1) fprintf x ;}
309 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
310 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
312 # define Assert(cond,msg)
317 # define Tracecv(c,x)
320 #define WARN(msg) {fprintf msg ; \
321 if (exit_code == OK) exit_code = WARNING;}
324 extern int unzip OF((int in, int out));
327 RETSIGTYPE abort_gzip OF((void));
330 void lm_init OF((int pack_level, ush *flags));
331 ulg deflate OF((void));
334 void ct_init OF((ush *attr, int *method));
335 int ct_tally OF((int dist, int lc));
336 ulg flush_block OF((char *buf, ulg stored_len, int eof));
339 void bi_init OF((file_t zipfile));
340 void send_bits OF((int value, int length));
341 unsigned bi_reverse OF((unsigned value, int length));
342 void bi_windup OF((void));
343 void copy_block OF((char *buf, unsigned len, int header));
344 extern int (*read_buf) OF((char *buf, unsigned size));
347 extern int copy OF((int in, int out));
348 extern ulg updcrc OF((uch *s, unsigned n));
349 extern void clear_bufs OF((void));
350 extern int fill_inbuf OF((int eof_ok));
351 extern void flush_outbuf OF((void));
352 extern void flush_window OF((void));
353 extern void write_buf OF((int fd, voidp buf, unsigned cnt));
355 extern char *basename OF((char *fname));
356 #endif /* not __linux__ */
357 extern void error OF((char *m));
358 extern void warn OF((char *a, char *b));
359 extern void read_error OF((void));
360 extern void write_error OF((void));
361 extern voidp xmalloc OF((unsigned int size));
364 extern int inflate OF((void));
366 /* #include "lzw.h" */
368 /* lzw.h -- define the lzw functions.
369 * Copyright (C) 1992-1993 Jean-loup Gailly.
370 * This is free software; you can redistribute it and/or modify it under the
371 * terms of the GNU General Public License, see the file COPYING.
374 #if !defined(OF) && defined(lint)
381 #define INIT_BITS 9 /* Initial number of bits per code */
383 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
385 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
386 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
387 * It's a pity that old uncompress does not check bit 0x20. That makes
388 * extension of the format actually undesirable because old compress
389 * would just crash on the new format instead of giving a meaningful
390 * error message. It does check the number of bits, but it's more
391 * helpful to say "unsupported format, get a new version" than
392 * "can only handle 16 bits".
395 #define BLOCK_MODE 0x80
396 /* Block compression: if table is full and compression rate is dropping,
397 * clear the dictionary.
400 #define LZW_RESERVED 0x60 /* reserved bits */
402 #define CLEAR 256 /* flush the dictionary */
403 #define FIRST (CLEAR+1) /* first free entry */
405 extern int maxbits; /* max bits per code for LZW */
406 extern int block_mode; /* block compress mode -C compatible with 2.0 */
408 extern int lzw OF((int in, int out));
409 extern int unlzw OF((int in, int out));
412 /* #include "revision.h" */
414 /* revision.h -- define the version number
415 * Copyright (C) 1992-1993 Jean-loup Gailly.
416 * This is free software; you can redistribute it and/or modify it under the
417 * terms of the GNU General Public License, see the file COPYING.
420 #define VERSION "1.2.4"
422 #define REVDATE "18 Aug 93"
424 /* This version does not support compression into old compress format: */
429 /* #include "getopt.h" */
431 /* Declarations for getopt.
432 Copyright (C) 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc.
434 This program is free software; you can redistribute it and/or modify it
435 under the terms of the GNU General Public License as published by the
436 Free Software Foundation; either version 2, or (at your option) any
439 This program is distributed in the hope that it will be useful,
440 but WITHOUT ANY WARRANTY; without even the implied warranty of
441 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
442 GNU General Public License for more details.
444 You should have received a copy of the GNU General Public License
445 along with this program; if not, write to the Free Software
446 Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
455 /* For communication from `getopt' to the caller.
456 When `getopt' finds an option that takes an argument,
457 the argument value is returned here.
458 Also, when `ordering' is RETURN_IN_ORDER,
459 each non-option ARGV-element is returned here. */
463 /* Index in ARGV of the next element to be scanned.
464 This is used for communication to and from the caller
465 and for communication between successive calls to `getopt'.
467 On entry to `getopt', zero means this is the first call; initialize.
469 When `getopt' returns EOF, this is the index of the first of the
470 non-option elements that the caller should itself scan.
472 Otherwise, `optind' communicates from one call to the next
473 how much of ARGV has been scanned so far. */
477 /* Callers store zero here to inhibit the error message `getopt' prints
478 for unrecognized options. */
482 /* Set to an option character which was unrecognized. */
486 /* Describe the long-named options requested by the application.
487 The LONG_OPTIONS argument to getopt_long or getopt_long_only is a vector
488 of `struct option' terminated by an element containing a name which is
491 The field `has_arg' is:
492 no_argument (or 0) if the option does not take an argument,
493 required_argument (or 1) if the option requires an argument,
494 optional_argument (or 2) if the option takes an optional argument.
496 If the field `flag' is not NULL, it points to a variable that is set
497 to the value given in the field `val' when the option is found, but
498 left unchanged if the option is not found.
500 To have a long-named option do something other than set an `int' to
501 a compiled-in constant, such as set a value from `optarg', set the
502 option's `flag' field to zero and its `val' field to a nonzero
503 value (the equivalent single-letter option character, if there is
504 one). For long options that have a zero `flag' field, `getopt'
505 returns the contents of the `val' field. */
514 /* has_arg can't be an enum because some compilers complain about
515 type mismatches in all the code that assumes it is an int. */
521 /* Names for the values of the `has_arg' field of `struct option'. */
523 #define no_argument 0
524 #define required_argument 1
525 #define optional_argument 2
527 #if __STDC__ || defined(PROTO)
528 #if defined(__GNU_LIBRARY__)
529 /* Many other libraries have conflicting prototypes for getopt, with
530 differences in the consts, in stdlib.h. To avoid compilation
531 errors, only prototype getopt for the GNU C library. */
532 extern int getopt (int argc, char *const *argv, const char *shortopts);
533 #endif /* not __GNU_LIBRARY__ */
534 extern int getopt_long (int argc, char *const *argv, const char *shortopts,
535 const struct option *longopts, int *longind);
536 extern int getopt_long_only (int argc, char *const *argv,
537 const char *shortopts,
538 const struct option *longopts, int *longind);
540 /* Internal only. Users should not call this directly. */
541 extern int _getopt_internal (int argc, char *const *argv,
542 const char *shortopts,
543 const struct option *longopts, int *longind,
545 #else /* not __STDC__ */
546 extern int getopt ();
547 extern int getopt_long ();
548 extern int getopt_long_only ();
550 extern int _getopt_internal ();
551 #endif /* not __STDC__ */
557 #endif /* _GETOPT_H */
568 typedef struct dirent dir_type;
569 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
570 # define DIR_OPT "DIRENT"
572 # define NLENGTH(dirent) ((dirent)->d_namlen)
574 # include <sys/dir.h>
575 typedef struct direct dir_type;
576 # define DIR_OPT "SYSDIR"
579 # include <sys/ndir.h>
580 typedef struct direct dir_type;
581 # define DIR_OPT "SYSNDIR"
585 typedef struct direct dir_type;
586 # define DIR_OPT "NDIR"
589 # define DIR_OPT "NO_DIR"
595 #if !defined(S_ISDIR) && defined(S_IFDIR)
596 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
598 #if !defined(S_ISREG) && defined(S_IFREG)
599 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
602 typedef RETSIGTYPE (*sig_type) OF((int));
605 # define O_BINARY 0 /* creation mode for open() */
609 /* Pure BSD system? */
610 # include <sys/file.h>
612 # define O_CREAT FCREAT
615 # define O_EXCL FEXCL
620 # define S_IRUSR 0400
623 # define S_IWUSR 0200
625 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
628 # define MAX_PATH_LEN 1024 /* max pathname length */
637 off_t lseek OF((int fd, off_t offset, int whence));
643 DECLARE(uch, inbuf, INBUFSIZ +INBUF_EXTRA);
644 DECLARE(uch, outbuf, OUTBUFSIZ+OUTBUF_EXTRA);
645 DECLARE(ush, d_buf, DIST_BUFSIZE);
646 DECLARE(uch, window, 2L*WSIZE);
648 DECLARE(ush, tab_prefix, 1L<<BITS);
650 DECLARE(ush, tab_prefix0, 1L<<(BITS-1));
651 DECLARE(ush, tab_prefix1, 1L<<(BITS-1));
654 /* local variables */
656 int force = 0; /* don't ask questions, compress links (-f) */
657 int foreground; /* set if program run in foreground */
658 int maxbits = BITS; /* max bits per code for LZW */
659 int method = DEFLATED;/* compression method */
660 int exit_code = OK; /* program exit code */
661 int last_member; /* set for .zip and .Z files */
662 int part_nb; /* number of parts in .gz file */
663 long ifile_size; /* input file size, -1 for devices (debug only) */
665 long bytes_in; /* number of input bytes */
666 long bytes_out; /* number of output bytes */
667 long total_in = 0; /* input bytes for all files */
668 long total_out = 0; /* output bytes for all files */
669 struct stat istat; /* status for input file */
670 int ifd; /* input file descriptor */
671 int ofd; /* output file descriptor */
672 unsigned insize; /* valid bytes in inbuf */
673 unsigned inptr; /* index of next byte to be processed in inbuf */
674 unsigned outcnt; /* bytes in output buffer */
676 long header_bytes; /* number of bytes in gzip header */
678 /* local functions */
680 local int get_method OF((int in));
681 local void do_exit(int exitcode) __attribute__ ((noreturn));
683 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
685 /* ======================================================================== */
686 int gunzip_main (int argc, char** argv)
688 int file_count; /* number of files to precess */
697 char ifname[MAX_PATH_LEN]; /* input file name */
698 char ofname[MAX_PATH_LEN]; /* output file name */
703 if (strcmp(*argv, "zcat")==0)
706 /* Parse any options */
707 while (--argc > 0 && **(++argv) == '-') {
708 if (*((*argv)+1) == '\0') {
712 while (*(++(*argv))) {
723 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
725 (void) signal (SIGINT, (sig_type)abort_gzip);
728 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
729 (void) signal(SIGTERM, (sig_type)abort_gzip);
733 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
734 (void) signal(SIGHUP, (sig_type)abort_gzip);
738 file_count = argc - optind;
740 /* Allocate all global buffers (for DYN_ALLOC option) */
741 ALLOC(uch, inbuf, INBUFSIZ +INBUF_EXTRA);
742 ALLOC(uch, outbuf, OUTBUFSIZ+OUTBUF_EXTRA);
743 ALLOC(ush, d_buf, DIST_BUFSIZE);
744 ALLOC(uch, window, 2L*WSIZE);
746 ALLOC(ush, tab_prefix, 1L<<BITS);
748 ALLOC(ush, tab_prefix0, 1L<<(BITS-1));
749 ALLOC(ush, tab_prefix1, 1L<<(BITS-1));
753 strcpy(ofname, "stdin");
755 inFileNum=fileno(stdin);
756 ifile_size = -1L; /* convention for unknown size */
758 /* Open up the input file */
761 strncpy(ifname, *argv, MAX_PATH_LEN);
763 /* Open input fille */
764 inFileNum=open( ifname, O_RDONLY);
769 /* Get the time stamp on the input file. */
770 result = stat(ifname, &statBuf);
775 ifile_size = statBuf.st_size;
779 /* And get to work */
780 strcpy(ofname, "stdout");
781 outFileNum=fileno(stdout);
783 clear_bufs(); /* clear input and output buffers */
786 /* Actually do the compression/decompression. */
787 unzip(inFileNum, outFileNum);
792 /* And get to work */
793 strncpy(ofname, ifname, MAX_PATH_LEN-4);
794 pos=strstr(ofname, ".gz");
799 pos=strstr(ofname, ".tgz");
802 strcat( pos, ".tar");
807 /* Open output fille */
808 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
809 outFileNum=open( ofname, O_RDWR|O_CREAT|O_EXCL|O_NOFOLLOW);
811 outFileNum=open( ofname, O_RDWR|O_CREAT|O_EXCL);
813 if (outFileNum < 0) {
817 /* Set permissions on the file */
818 fchmod(outFileNum, statBuf.st_mode);
820 clear_bufs(); /* clear input and output buffers */
823 /* Actually do the compression/decompression. */
824 result=unzip(inFileNum, outFileNum);
828 /* Delete the original file */
834 if (delInputFile == 1 && unlink (delFileName) < 0) {
835 perror (delFileName);
843 /* ========================================================================
844 * Check the magic number of the input file and update ofname if an
845 * original name was given and to_stdout is not set.
846 * Return the compression method, -1 for error, -2 for warning.
847 * Set inptr to the offset of the next byte to be processed.
848 * Updates time_stamp if there is one and --no-time is not used.
849 * This function may be called repeatedly for an input file consisting
850 * of several contiguous gzip'ed members.
851 * IN assertions: there is at least one remaining compressed member.
852 * If the member is a zip file, it must be the only one.
854 local int get_method(in)
855 int in; /* input file descriptor */
857 uch flags; /* compression flags */
858 char magic[2]; /* magic header */
860 /* If --force and --stdout, zcat == cat, so do not complain about
861 * premature end of file: use try_byte instead of get_byte.
864 magic[0] = (char)try_byte();
865 magic[1] = (char)try_byte();
866 /* If try_byte returned EOF, magic[1] == 0xff */
868 magic[0] = (char)get_byte();
869 magic[1] = (char)get_byte();
871 method = -1; /* unknown yet */
872 part_nb++; /* number of parts in gzip file */
874 last_member = RECORD_IO;
875 /* assume multiple members in gzip file except for record oriented I/O */
877 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
879 method = (int)get_byte();
880 if (method != DEFLATED) {
882 "unknown method %d -- get newer version of gzip\n",
887 flags = (uch)get_byte();
889 (ulg)get_byte(); /* Ignore time stamp */
894 (void)get_byte(); /* Ignore extra flags for the moment */
895 (void)get_byte(); /* Ignore OS type for the moment */
897 if ((flags & EXTRA_FIELD) != 0) {
898 unsigned len = (unsigned)get_byte();
899 len |= ((unsigned)get_byte())<<8;
901 while (len--) (void)get_byte();
904 /* Discard original name if any */
905 if ((flags & ORIG_NAME) != 0) {
906 while (get_char() != 0) /* null */ ;
909 /* Discard file comment if any */
910 if ((flags & COMMENT) != 0) {
911 while (get_char() != 0) /* null */ ;
914 header_bytes = inptr + 2*sizeof(long); /* include crc and size */
919 if (method >= 0) return method;
922 fprintf(stderr, "\nnot in gzip format\n");
926 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
932 /* ========================================================================
933 * Free all dynamically allocated variables and exit with the given code.
935 local void do_exit(exitcode)
938 static int in_exit = 0;
940 if (in_exit) exit(exitcode);
955 /* ========================================================================
956 * Signal and error handler.
958 RETSIGTYPE abort_gzip()
962 /* unzip.c -- decompress files in gzip or pkzip format.
963 * Copyright (C) 1992-1993 Jean-loup Gailly
964 * This is free software; you can redistribute it and/or modify it under the
965 * terms of the GNU General Public License, see the file COPYING.
967 * The code in this file is derived from the file funzip.c written
968 * and put in the public domain by Mark Adler.
972 This version can extract files in gzip or pkzip format.
973 For the latter, only the first entry is extracted, and it has to be
974 either deflated or stored.
977 /* #include "crypt.h" */
979 /* crypt.h (dummy version) -- do not perform encryption
980 * Hardly worth copyrighting :-)
984 # undef CRYPT /* dummy version */
987 #define RAND_HEAD_LEN 12 /* length of encryption random header */
992 /* PKZIP header definitions */
993 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
994 #define LOCFLG 6 /* offset of bit flag */
995 #define CRPFLG 1 /* bit for encrypted entry */
996 #define EXTFLG 8 /* bit for extended local header */
997 #define LOCHOW 8 /* offset of compression method */
998 #define LOCTIM 10 /* file mod time (for decryption) */
999 #define LOCCRC 14 /* offset of crc */
1000 #define LOCSIZ 18 /* offset of compressed size */
1001 #define LOCLEN 22 /* offset of uncompressed length */
1002 #define LOCFIL 26 /* offset of file name field length */
1003 #define LOCEXT 28 /* offset of extra field length */
1004 #define LOCHDR 30 /* size of local header, including sig */
1005 #define EXTHDR 16 /* size of extended local header, inc sig */
1010 char *key; /* not used--needed to link crypt.c */
1011 int pkzip = 0; /* set for a pkzip file */
1012 int ext_header = 0; /* set if extended local header */
1014 /* ===========================================================================
1015 * Unzip in to out. This routine works on both gzip and pkzip files.
1017 * IN assertions: the buffer inbuf contains already the beginning of
1018 * the compressed data, from offsets inptr to insize-1 included.
1019 * The magic header has already been checked. The output buffer is cleared.
1022 int in, out; /* input and output file descriptors */
1024 ulg orig_crc = 0; /* original crc */
1025 ulg orig_len = 0; /* original uncompressed length */
1027 uch buf[EXTHDR]; /* extended local header */
1031 method = get_method(ifd);
1033 do_exit(exit_code); /* error message already emitted */
1036 updcrc(NULL, 0); /* initialize crc */
1038 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
1039 orig_crc = LG(inbuf + LOCCRC);
1040 orig_len = LG(inbuf + LOCLEN);
1044 if (method == DEFLATED) {
1046 int res = inflate();
1049 error("out of memory");
1050 } else if (res != 0) {
1051 error("invalid compressed data--format violated");
1055 error("internal error, invalid method");
1058 /* Get the crc and original length */
1060 /* crc32 (see algorithm.doc)
1061 * uncompressed input size modulo 2^32
1063 for (n = 0; n < 8; n++) {
1064 buf[n] = (uch)get_byte(); /* may cause an error if EOF */
1067 orig_len = LG(buf+4);
1069 } else if (ext_header) { /* If extended header, check it */
1070 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
1072 * compressed size 4-bytes
1073 * uncompressed size 4-bytes
1075 for (n = 0; n < EXTHDR; n++) {
1076 buf[n] = (uch)get_byte(); /* may cause an error if EOF */
1078 orig_crc = LG(buf+4);
1079 orig_len = LG(buf+12);
1082 /* Validate decompression */
1083 if (orig_crc != updcrc(outbuf, 0)) {
1084 error("invalid compressed data--crc error");
1086 if (orig_len != (ulg)bytes_out) {
1087 error("invalid compressed data--length error");
1090 /* Check if there are more entries in a pkzip file */
1091 if (pkzip && inptr + 4 < insize && LG(inbuf+inptr) == LOCSIG) {
1092 WARN((stderr,"has more than one entry--rest ignored\n"));
1094 ext_header = pkzip = 0; /* for next file */
1097 /* util.c -- utility functions for gzip support
1098 * Copyright (C) 1992-1993 Jean-loup Gailly
1099 * This is free software; you can redistribute it and/or modify it under the
1100 * terms of the GNU General Public License, see the file COPYING.
1105 #include <sys/types.h>
1107 #ifdef HAVE_UNISTD_H
1108 # include <unistd.h>
1114 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1115 # include <stdlib.h>
1120 static const ulg crc_32_tab[]; /* crc table, defined below */
1122 /* ===========================================================================
1123 * Run a set of bytes through the crc shift register. If s is a NULL
1124 * pointer, then initialize the crc shift register contents instead.
1125 * Return the current crc in either case.
1128 uch *s; /* pointer to bytes to pump through */
1129 unsigned n; /* number of bytes in s[] */
1131 register ulg c; /* temporary variable */
1133 static ulg crc = (ulg)0xffffffffL; /* shift register contents */
1140 c = crc_32_tab[((int)c ^ (*s++)) & 0xff] ^ (c >> 8);
1144 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1147 /* ===========================================================================
1148 * Clear input and output buffers
1154 bytes_in = bytes_out = 0L;
1157 /* ===========================================================================
1158 * Fill the input buffer. This is called only when the buffer is empty.
1160 int fill_inbuf(eof_ok)
1161 int eof_ok; /* set if EOF acceptable as a result */
1165 /* Read as much as possible */
1169 len = read(ifd, (char*)inbuf+insize, INBUFSIZ-insize);
1170 if (len == 0 || len == EOF) break;
1172 } while (insize < INBUFSIZ);
1175 if (eof_ok) return EOF;
1178 bytes_in += (ulg)insize;
1183 /* ===========================================================================
1184 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1185 * (used for the compressed data only)
1189 if (outcnt == 0) return;
1191 write_buf(ofd, (char *)outbuf, outcnt);
1192 bytes_out += (ulg)outcnt;
1196 /* ===========================================================================
1197 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1198 * (Used for the decompressed data only.)
1202 if (outcnt == 0) return;
1203 updcrc(window, outcnt);
1205 write_buf(ofd, (char *)window, outcnt);
1207 bytes_out += (ulg)outcnt;
1211 /* ===========================================================================
1212 * Does the same as write(), but also handles partial pipe writes and checks
1215 void write_buf(fd, buf, cnt)
1222 while ((n = write(fd, buf, cnt)) != cnt) {
1223 if (n == (unsigned)(-1)) {
1227 buf = (voidp)((char*)buf+n);
1231 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1233 /* Provide missing strspn and strcspn functions. */
1239 int strspn OF((const char *s, const char *accept));
1240 int strcspn OF((const char *s, const char *reject));
1242 /* ========================================================================
1243 * Return the length of the maximum initial segment
1244 * of s which contains only characters in accept.
1246 int strspn(s, accept)
1250 register const char *p;
1251 register const char *a;
1252 register int count = 0;
1254 for (p = s; *p != '\0'; ++p) {
1255 for (a = accept; *a != '\0'; ++a) {
1256 if (*p == *a) break;
1258 if (*a == '\0') return count;
1264 /* ========================================================================
1265 * Return the length of the maximum inital segment of s
1266 * which contains no characters from reject.
1268 int strcspn(s, reject)
1272 register int count = 0;
1274 while (*s != '\0') {
1275 if (strchr(reject, *s++) != NULL) return count;
1281 #endif /* NO_STRING_H */
1284 /* ========================================================================
1290 fprintf(stderr, "\n%s\n", m);
1295 char *a, *b; /* message strings juxtaposed in output */
1297 WARN((stderr, "warning: %s%s\n", a, b));
1302 fprintf(stderr, "\n");
1306 fprintf(stderr, "unexpected end of file\n");
1313 fprintf(stderr, "\n");
1319 /* ========================================================================
1320 * Semi-safe malloc -- never returns NULL.
1322 voidp xmalloc (size)
1325 voidp cp = (voidp)malloc (size);
1327 if (cp == NULL) error("out of memory");
1331 /* ========================================================================
1332 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1334 static const ulg crc_32_tab[] = {
1335 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1336 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1337 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1338 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1339 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1340 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1341 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1342 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1343 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1344 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1345 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1346 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1347 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1348 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1349 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1350 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1351 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1352 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1353 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1354 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1355 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1356 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1357 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1358 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1359 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1360 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1361 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1362 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1363 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1364 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1365 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1366 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1367 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1368 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1369 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1370 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1371 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1372 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1373 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1374 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1375 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1376 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1377 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1378 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1379 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1380 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1381 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1382 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1383 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1384 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1385 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1388 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1389 version c10p1, 10 January 1993 */
1391 /* You can do whatever you like with this source file, though I would
1392 prefer that if you modify it and redistribute it that you include
1393 comments to that effect with your name and the date. Thank you.
1394 [The history has been moved to the file ChangeLog.]
1398 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1399 method searches for as much of the current string of bytes (up to a
1400 length of 258) in the previous 32K bytes. If it doesn't find any
1401 matches (of at least length 3), it codes the next byte. Otherwise, it
1402 codes the length of the matched string and its distance backwards from
1403 the current position. There is a single Huffman code that codes both
1404 single bytes (called "literals") and match lengths. A second Huffman
1405 code codes the distance information, which follows a length code. Each
1406 length or distance code actually represents a base value and a number
1407 of "extra" (sometimes zero) bits to get to add to the base value. At
1408 the end of each deflated block is a special end-of-block (EOB) literal/
1409 length code. The decoding process is basically: get a literal/length
1410 code; if EOB then done; if a literal, emit the decoded byte; if a
1411 length then get the distance and emit the referred-to bytes from the
1412 sliding window of previously emitted data.
1414 There are (currently) three kinds of inflate blocks: stored, fixed, and
1415 dynamic. The compressor deals with some chunk of data at a time, and
1416 decides which method to use on a chunk-by-chunk basis. A chunk might
1417 typically be 32K or 64K. If the chunk is uncompressible, then the
1418 "stored" method is used. In this case, the bytes are simply stored as
1419 is, eight bits per byte, with none of the above coding. The bytes are
1420 preceded by a count, since there is no longer an EOB code.
1422 If the data is compressible, then either the fixed or dynamic methods
1423 are used. In the dynamic method, the compressed data is preceded by
1424 an encoding of the literal/length and distance Huffman codes that are
1425 to be used to decode this block. The representation is itself Huffman
1426 coded, and so is preceded by a description of that code. These code
1427 descriptions take up a little space, and so for small blocks, there is
1428 a predefined set of codes, called the fixed codes. The fixed method is
1429 used if the block codes up smaller that way (usually for quite small
1430 chunks), otherwise the dynamic method is used. In the latter case, the
1431 codes are customized to the probabilities in the current block, and so
1432 can code it much better than the pre-determined fixed codes.
1434 The Huffman codes themselves are decoded using a mutli-level table
1435 lookup, in order to maximize the speed of decoding plus the speed of
1436 building the decoding tables. See the comments below that precede the
1437 lbits and dbits tuning parameters.
1442 Notes beyond the 1.93a appnote.txt:
1444 1. Distance pointers never point before the beginning of the output
1446 2. Distance pointers can point back across blocks, up to 32k away.
1447 3. There is an implied maximum of 7 bits for the bit length table and
1448 15 bits for the actual data.
1449 4. If only one code exists, then it is encoded using one bit. (Zero
1450 would be more efficient, but perhaps a little confusing.) If two
1451 codes exist, they are coded using one bit each (0 and 1).
1452 5. There is no way of sending zero distance codes--a dummy must be
1453 sent if there are none. (History: a pre 2.0 version of PKZIP would
1454 store blocks with no distance codes, but this was discovered to be
1455 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1456 zero distance codes, which is sent as one code of zero bits in
1458 6. There are up to 286 literal/length codes. Code 256 represents the
1459 end-of-block. Note however that the static length tree defines
1460 288 codes just to fill out the Huffman codes. Codes 286 and 287
1461 cannot be used though, since there is no length base or extra bits
1462 defined for them. Similarly, there are up to 30 distance codes.
1463 However, static trees define 32 codes (all 5 bits) to fill out the
1464 Huffman codes, but the last two had better not show up in the data.
1465 7. Unzip can check dynamic Huffman blocks for complete code sets.
1466 The exception is that a single code would not be complete (see #4).
1467 8. The five bits following the block type is really the number of
1468 literal codes sent minus 257.
1469 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1470 (1+6+6). Therefore, to output three times the length, you output
1471 three codes (1+1+1), whereas to output four times the same length,
1472 you only need two codes (1+3). Hmm.
1473 10. In the tree reconstruction algorithm, Code = Code + Increment
1474 only if BitLength(i) is not zero. (Pretty obvious.)
1475 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1476 12. Note: length code 284 can represent 227-258, but length code 285
1477 really is 258. The last length deserves its own, short code
1478 since it gets used a lot in very redundant files. The length
1479 258 is special since 258 - 3 (the min match length) is 255.
1480 13. The literal/length and distance code bit lengths are read as a
1481 single stream of lengths. It is possible (and advantageous) for
1482 a repeat code (16, 17, or 18) to go across the boundary between
1483 the two sets of lengths.
1486 #include <sys/types.h>
1488 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1489 # include <stdlib.h>
1493 #define slide window
1495 /* Huffman code lookup table entry--this entry is four bytes for machines
1496 that have 16-bit pointers (e.g. PC's in the small or medium model).
1497 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1498 means that v is a literal, 16 < e < 32 means that v is a pointer to
1499 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1500 an unused code. If a code with e == 99 is looked up, this implies an
1501 error in the data. */
1503 uch e; /* number of extra bits or operation */
1504 uch b; /* number of bits in this code or subcode */
1506 ush n; /* literal, length base, or distance base */
1507 struct huft *t; /* pointer to next level of table */
1512 /* Function prototypes */
1513 int huft_build OF((unsigned *, unsigned, unsigned, ush *, ush *,
1514 struct huft **, int *));
1515 int huft_free OF((struct huft *));
1516 int inflate_codes OF((struct huft *, struct huft *, int, int));
1517 int inflate_stored OF((void));
1518 int inflate_fixed OF((void));
1519 int inflate_dynamic OF((void));
1520 int inflate_block OF((int *));
1521 int inflate OF((void));
1524 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1525 stream to find repeated byte strings. This is implemented here as a
1526 circular buffer. The index is updated simply by incrementing and then
1527 and'ing with 0x7fff (32K-1). */
1528 /* It is left to other modules to supply the 32K area. It is assumed
1529 to be usable as if it were declared "uch slide[32768];" or as just
1530 "uch *slide;" and then malloc'ed in the latter case. The definition
1531 must be in unzip.h, included above. */
1532 /* unsigned wp; current position in slide */
1534 #define flush_output(w) (wp=(w),flush_window())
1536 /* Tables for deflate from PKZIP's appnote.txt. */
1537 static unsigned border[] = { /* Order of the bit length code lengths */
1538 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
1539 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1540 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1541 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0};
1542 /* note: see note #13 above about the 258 in this list. */
1543 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1544 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1545 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99}; /* 99==invalid */
1546 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1547 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1548 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1549 8193, 12289, 16385, 24577};
1550 static ush cpdext[] = { /* Extra bits for distance codes */
1551 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1552 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1557 /* Macros for inflate() bit peeking and grabbing.
1561 x = b & mask_bits[j];
1564 where NEEDBITS makes sure that b has at least j bits in it, and
1565 DUMPBITS removes the bits from b. The macros use the variable k
1566 for the number of bits in b. Normally, b and k are register
1567 variables for speed, and are initialized at the beginning of a
1568 routine that uses these macros from a global bit buffer and count.
1570 If we assume that EOB will be the longest code, then we will never
1571 ask for bits with NEEDBITS that are beyond the end of the stream.
1572 So, NEEDBITS should not read any more bytes than are needed to
1573 meet the request. Then no bytes need to be "returned" to the buffer
1574 at the end of the last block.
1576 However, this assumption is not true for fixed blocks--the EOB code
1577 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1578 (The EOB code is shorter than other codes because fixed blocks are
1579 generally short. So, while a block always has an EOB, many other
1580 literal/length codes have a significantly lower probability of
1581 showing up at all.) However, by making the first table have a
1582 lookup of seven bits, the EOB code will be found in that first
1583 lookup, and so will not require that too many bits be pulled from
1587 ulg bb; /* bit buffer */
1588 unsigned bk; /* bits in bit buffer */
1592 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1593 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 */
1659 /* Given a list of code lengths and a maximum table size, make a set of
1660 tables to decode that set of codes. Return zero on success, one if
1661 the given code set is incomplete (the tables are still built in this
1662 case), two if the input is invalid (all zero length codes or an
1663 oversubscribed set of lengths), and three if not enough memory. */
1665 unsigned a; /* counter for codes of length k */
1666 unsigned c[BMAX+1]; /* bit length count table */
1667 unsigned f; /* i repeats in table every f entries */
1668 int g; /* maximum code length */
1669 int h; /* table level */
1670 register unsigned i; /* counter, current code */
1671 register unsigned j; /* counter */
1672 register int k; /* number of bits in current code */
1673 int l; /* bits per table (returned in m) */
1674 register unsigned *p; /* pointer into c[], b[], or v[] */
1675 register struct huft *q; /* points to current table */
1676 struct huft r; /* table entry for structure assignment */
1677 struct huft *u[BMAX]; /* table stack */
1678 unsigned v[N_MAX]; /* values in order of bit length */
1679 register int w; /* bits before this table == (l * h) */
1680 unsigned x[BMAX+1]; /* bit offsets, then code stack */
1681 unsigned *xp; /* pointer into x */
1682 int y; /* number of dummy codes added */
1683 unsigned z; /* number of entries in current table */
1686 /* Generate counts for each bit length */
1687 memzero(c, sizeof(c));
1690 Tracecv(*p, (stderr, (n-i >= ' ' && n-i <= '~' ? "%c %d\n" : "0x%x %d\n"),
1692 c[*p]++; /* assume all entries <= BMAX */
1693 p++; /* Can't combine with above line (Solaris bug) */
1695 if (c[0] == n) /* null input--all zero length codes */
1697 *t = (struct huft *)NULL;
1703 /* Find minimum and maximum length, bound *m by those */
1705 for (j = 1; j <= BMAX; j++)
1708 k = j; /* minimum code length */
1709 if ((unsigned)l < j)
1711 for (i = BMAX; i; i--)
1714 g = i; /* maximum code length */
1715 if ((unsigned)l > i)
1720 /* Adjust last length count to fill out codes, if needed */
1721 for (y = 1 << j; j < i; j++, y <<= 1)
1722 if ((y -= c[j]) < 0)
1723 return 2; /* bad input: more codes than bits */
1724 if ((y -= c[i]) < 0)
1729 /* Generate starting offsets into the value table for each length */
1731 p = c + 1; xp = x + 2;
1732 while (--i) { /* note that i == g from above */
1733 *xp++ = (j += *p++);
1737 /* Make a table of values in order of bit lengths */
1740 if ((j = *p++) != 0)
1745 /* Generate the Huffman codes and for each, make the table entries */
1746 x[0] = i = 0; /* first Huffman code is zero */
1747 p = v; /* grab values in bit order */
1748 h = -1; /* no tables yet--level -1 */
1749 w = -l; /* bits decoded == (l * h) */
1750 u[0] = (struct huft *)NULL; /* just to keep compilers happy */
1751 q = (struct huft *)NULL; /* ditto */
1754 /* go through the bit lengths (k already is bits in shortest code) */
1760 /* here i is the Huffman code of length k bits for value *p */
1761 /* make tables up to required level */
1765 w += l; /* previous table always l bits */
1767 /* compute minimum size table less than or equal to l bits */
1768 z = (z = g - w) > (unsigned)l ? l : z; /* upper limit on table size */
1769 if ((f = 1 << (j = k - w)) > a + 1) /* try a k-w bit table */
1770 { /* too few codes for k-w bit table */
1771 f -= a + 1; /* deduct codes from patterns left */
1773 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 */
1783 if ((q = (struct huft *)malloc((z + 1)*sizeof(struct huft))) ==
1784 (struct huft *)NULL)
1788 return 3; /* not enough memory */
1790 hufts += z + 1; /* track memory usage */
1791 *t = q + 1; /* link to list for huft_free() */
1792 *(t = &(q->v.t)) = (struct huft *)NULL;
1793 u[h] = ++q; /* table starts after link */
1795 /* connect to last table, if there is one */
1798 x[h] = i; /* save pattern for backing up */
1799 r.b = (uch)l; /* bits to dump before this table */
1800 r.e = (uch)(16 + j); /* bits in this table */
1801 r.v.t = q; /* pointer to this table */
1802 j = i >> (w - l); /* (get around Turbo C bug) */
1803 u[h-1][j] = r; /* connect to last table */
1807 /* set up table entry in r */
1810 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++ */
1819 r.e = (uch)e[*p - s]; /* non-simple--look up in lists */
1820 r.v.n = d[*p++ - s];
1823 /* fill code-like entries with r */
1825 for (j = i >> w; j < z; j += f)
1828 /* backwards increment the k-bit code i */
1829 for (j = 1 << (k - 1); i & j; j >>= 1)
1833 /* backup over finished tables */
1834 while ((i & ((1 << w) - 1)) != x[h])
1836 h--; /* don't need to update q */
1843 /* Return true (1) if we were given an incomplete table */
1844 return y != 0 && g != 1;
1850 struct huft *t; /* table to free */
1851 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1852 list of the tables it made, with the links in a dummy first entry of
1855 register struct huft *p, *q;
1858 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1860 while (p != (struct huft *)NULL)
1870 int inflate_codes(tl, td, bl, bd)
1871 struct huft *tl, *td; /* literal/length and distance decoder tables */
1872 int bl, bd; /* number of bits decoded by tl[] and td[] */
1873 /* inflate (decompress) the codes in a deflated (compressed) block.
1874 Return an error code or zero if it all goes ok. */
1876 register unsigned e; /* table entry flag/number of extra bits */
1877 unsigned n, d; /* length and index for copy */
1878 unsigned w; /* current window position */
1879 struct huft *t; /* pointer to table entry */
1880 unsigned ml, md; /* masks for bl and bd bits */
1881 register ulg b; /* bit buffer */
1882 register unsigned k; /* number of bits in bit buffer */
1885 /* make local copies of globals */
1886 b = bb; /* initialize bit buffer */
1888 w = wp; /* initialize window position */
1890 /* inflate the coded data */
1891 ml = mask_bits[bl]; /* precompute masks for speed */
1893 for (;;) /* do until end of block */
1895 NEEDBITS((unsigned)bl)
1896 if ((e = (t = tl + ((unsigned)b & ml))->e) > 16)
1903 } while ((e = (t = t->v.t + ((unsigned)b & mask_bits[e]))->e) > 16);
1905 if (e == 16) /* then it's a literal */
1907 slide[w++] = (uch)t->v.n;
1908 Tracevv((stderr, "%c", slide[w-1]));
1915 else /* it's an EOB or a length */
1917 /* exit if end of block */
1921 /* get length of block to copy */
1923 n = t->v.n + ((unsigned)b & mask_bits[e]);
1926 /* decode distance of block to copy */
1927 NEEDBITS((unsigned)bd)
1928 if ((e = (t = td + ((unsigned)b & md))->e) > 16)
1935 } while ((e = (t = t->v.t + ((unsigned)b & mask_bits[e]))->e) > 16);
1938 d = w - t->v.n - ((unsigned)b & mask_bits[e]);
1940 Tracevv((stderr,"\\[%d,%d]", w-d, n));
1944 n -= (e = (e = WSIZE - ((d &= WSIZE-1) > w ? d : w)) > n ? n : e);
1945 #if !defined(NOMEMCPY) && !defined(DEBUG)
1946 if (w - d >= e) /* (this test assumes unsigned comparison) */
1948 memcpy(slide + w, slide + d, e);
1952 else /* do it slow to avoid memcpy() overlap */
1953 #endif /* !NOMEMCPY */
1955 slide[w++] = slide[d++];
1956 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 */
2013 slide[w++] = (uch)b;
2023 /* restore the globals from the locals */
2024 wp = w; /* restore global window pointer */
2025 bb = b; /* restore global bit buffer */
2033 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
2034 either replace this with a custom decoder, or at least precompute the
2037 int i; /* temporary variable */
2038 struct huft *tl; /* literal/length code table */
2039 struct huft *td; /* distance code table */
2040 int bl; /* lookup bits for tl */
2041 int bd; /* lookup bits for td */
2042 unsigned l[288]; /* length list for huft_build */
2045 /* set up literal table */
2046 for (i = 0; i < 144; i++)
2048 for (; i < 256; i++)
2050 for (; i < 280; i++)
2052 for (; i < 288; i++) /* make a complete, but wrong code set */
2055 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
2059 /* set up distance table */
2060 for (i = 0; i < 30; i++) /* make an incomplete code set */
2063 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1)
2070 /* decompress until an end-of-block code */
2071 if (inflate_codes(tl, td, bl, bd))
2075 /* free the decoding tables, return */
2083 int inflate_dynamic()
2084 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
2086 int i; /* temporary variables */
2088 unsigned l; /* last length */
2089 unsigned m; /* mask for bit lengths table */
2090 unsigned n; /* number of lengths to get */
2091 struct huft *tl; /* literal/length code table */
2092 struct huft *td; /* distance code table */
2093 int bl; /* lookup bits for tl */
2094 int bd; /* lookup bits for td */
2095 unsigned nb; /* number of bit length codes */
2096 unsigned nl; /* number of literal/length codes */
2097 unsigned nd; /* number of distance codes */
2098 #ifdef PKZIP_BUG_WORKAROUND
2099 unsigned ll[288+32]; /* literal/length and distance code lengths */
2101 unsigned ll[286+30]; /* literal/length and distance code lengths */
2103 register ulg b; /* bit buffer */
2104 register unsigned k; /* number of bits in bit buffer */
2107 /* make local bit buffer */
2112 /* read in table lengths */
2114 nl = 257 + ((unsigned)b & 0x1f); /* number of literal/length codes */
2117 nd = 1 + ((unsigned)b & 0x1f); /* number of distance codes */
2120 nb = 4 + ((unsigned)b & 0xf); /* number of bit length codes */
2122 #ifdef PKZIP_BUG_WORKAROUND
2123 if (nl > 288 || nd > 32)
2125 if (nl > 286 || nd > 30)
2127 return 1; /* bad lengths */
2130 /* read in bit-length-code lengths */
2131 for (j = 0; j < nb; j++)
2134 ll[border[j]] = (unsigned)b & 7;
2141 /* build decoding table for trees--single level, 7 bit lookup */
2143 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0)
2147 return i; /* incomplete code set */
2151 /* read in literal and distance code lengths */
2155 while ((unsigned)i < n)
2157 NEEDBITS((unsigned)bl)
2158 j = (td = tl + ((unsigned)b & m))->b;
2161 if (j < 16) /* length of code in bits (0..15) */
2162 ll[i++] = l = j; /* save last length in l */
2163 else if (j == 16) /* repeat last length 3 to 6 times */
2166 j = 3 + ((unsigned)b & 3);
2168 if ((unsigned)i + j > n)
2173 else if (j == 17) /* 3 to 10 zero length codes */
2176 j = 3 + ((unsigned)b & 7);
2178 if ((unsigned)i + j > n)
2184 else /* j == 18: 11 to 138 zero length codes */
2187 j = 11 + ((unsigned)b & 0x7f);
2189 if ((unsigned)i + j > n)
2198 /* free decoding table for trees */
2202 /* restore the global bit buffer */
2207 /* build the decoding tables for literal/length and distance codes */
2209 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0)
2212 fprintf(stderr, " incomplete literal tree\n");
2215 return i; /* incomplete code set */
2218 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0)
2221 fprintf(stderr, " incomplete distance tree\n");
2222 #ifdef PKZIP_BUG_WORKAROUND
2229 return i; /* incomplete code set */
2234 /* decompress until an end-of-block code */
2235 if (inflate_codes(tl, td, bl, bd))
2239 /* free the decoding tables, return */
2247 int inflate_block(e)
2248 int *e; /* last block flag */
2249 /* decompress an inflated block */
2251 unsigned t; /* block type */
2252 register ulg b; /* bit buffer */
2253 register unsigned k; /* number of bits in bit buffer */
2256 /* make local bit buffer */
2261 /* read in last block bit */
2267 /* read in block type */
2269 t = (unsigned)b & 3;
2273 /* restore the global bit buffer */
2278 /* inflate that block type */
2280 return inflate_dynamic();
2282 return inflate_stored();
2284 return inflate_fixed();
2287 /* bad block type */
2294 /* decompress an inflated entry */
2296 int e; /* last block flag */
2297 int r; /* result code */
2298 unsigned h; /* maximum struct huft's malloc'ed */
2301 /* initialize window, bit buffer */
2307 /* decompress until the last block */
2311 if ((r = inflate_block(&e)) != 0)
2317 /* Undo too much lookahead. The next read will be byte aligned so we
2318 * can discard unused bits in the last meaningful byte.
2325 /* flush out slide */
2329 /* return success */
2331 fprintf(stderr, "<%u> ", h);