1 /* vi: set sw=4 ts=4: */
3 * Gzip implementation for busybox
5 * Based on GNU gzip Copyright (C) 1992-1993 Jean-loup Gailly.
7 * Originally adjusted for busybox by Sven Rudolph <sr1@inf.tu-dresden.de>
8 * based on gzip sources
10 * Adjusted further by Erik Andersen <andersen@lineo.com>, <andersee@debian.org>
11 * to support files as well as stdin/stdout, and to generally behave itself wrt
12 * command line handling.
14 * This program is free software; you can redistribute it and/or modify
15 * it under the terms of the GNU General Public License as published by
16 * the Free Software Foundation; either version 2 of the License, or
17 * (at your option) any later version.
19 * This program is distributed in the hope that it will be useful,
20 * but WITHOUT ANY WARRANTY; without even the implied warranty of
21 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22 * General Public License for more details.
24 * You should have received a copy of the GNU General Public License
25 * along with this program; if not, write to the Free Software
26 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
31 static const char gunzip_usage[] =
32 "gunzip [OPTION]... FILE\n\n"
33 "Uncompress FILE (or standard input if FILE is '-').\n\n"
36 "\t-c\tWrite output to standard output\n"
37 "\t-t\tTest compressed file integrity\n";
40 /* These defines are very important for BusyBox. Without these,
41 * huge chunks of ram are pre-allocated making the BusyBox bss
42 * size Freaking Huge(tm), which is a bad thing.*/
46 #define bb_need_name_too_long
47 #define BB_DECLARE_EXTERN
51 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
52 * Copyright (C) 1992-1993 Jean-loup Gailly
53 * The unzip code was written and put in the public domain by Mark Adler.
54 * Portions of the lzw code are derived from the public domain 'compress'
55 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
56 * Ken Turkowski, Dave Mack and Peter Jannesen.
58 * See the license_msg below and the file COPYING for the software license.
59 * See the file algorithm.doc for the compression algorithms and file formats.
63 static char *license_msg[] = {
64 " Copyright (C) 1992-1993 Jean-loup Gailly",
65 " This program is free software; you can redistribute it and/or modify",
66 " it under the terms of the GNU General Public License as published by",
67 " the Free Software Foundation; either version 2, or (at your option)",
68 " any later version.",
70 " This program is distributed in the hope that it will be useful,",
71 " but WITHOUT ANY WARRANTY; without even the implied warranty of",
72 " MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the",
73 " GNU General Public License for more details.",
75 " You should have received a copy of the GNU General Public License",
76 " along with this program; if not, write to the Free Software",
77 " Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.",
82 /* Compress files with zip algorithm and 'compress' interface.
83 * See usage() and help() functions below for all options.
85 * file.gz: compressed file with same mode, owner, and utimes
86 * or stdout with -c option or if stdin used as input.
87 * If the output file name had to be truncated, the original name is kept
88 * in the compressed file.
89 * On MSDOS, file.tmp -> file.tmz. On VMS, file.tmp -> file.tmp-gz.
91 * Using gz on MSDOS would create too many file name conflicts. For
92 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
93 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
94 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
95 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
97 * For the meaning of all compilation flags, see comments in Makefile.in.
101 #include <sys/types.h>
103 #include <sys/stat.h>
106 /* #include "tailor.h" */
108 /* tailor.h -- target dependent definitions
109 * Copyright (C) 1992-1993 Jean-loup Gailly.
110 * This is free software; you can redistribute it and/or modify it under the
111 * terms of the GNU General Public License, see the file COPYING.
114 /* The target dependent definitions should be defined here only.
115 * The target dependent functions should be defined in tailor.c.
120 #define get_char() get_byte()
121 #define put_char(c) put_byte(c)
124 /* I don't like nested includes, but the string and io functions are used
128 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
130 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
133 # define memzero(s, n) memset ((void *)(s), 0, (n))
135 # include <strings.h>
136 # define strchr index
137 # define strrchr rindex
138 # define memcpy(d, s, n) bcopy((s), (d), (n))
139 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
140 # define memzero(s, n) bzero((s), (n))
144 # define RETSIGTYPE void
149 typedef unsigned char uch;
150 typedef unsigned short ush;
151 typedef unsigned long ulg;
153 /* Return codes from gzip */
158 /* Compression methods (see algorithm.doc) */
161 extern int method; /* compression method */
163 /* To save memory for 16 bit systems, some arrays are overlaid between
164 * the various modules:
165 * deflate: prev+head window d_buf l_buf outbuf
166 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
167 * inflate: window inbuf
168 * unpack: window inbuf prefix_len
169 * unlzh: left+right window c_table inbuf c_len
170 * For compression, input is done in window[]. For decompression, output
171 * is done in window except for unlzw.
176 # define INBUFSIZ 0x2000 /* input buffer size */
178 # define INBUFSIZ 0x8000 /* input buffer size */
181 #define INBUF_EXTRA 64 /* required by unlzw() */
185 # define OUTBUFSIZ 8192 /* output buffer size */
187 # define OUTBUFSIZ 16384 /* output buffer size */
190 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
196 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
198 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
202 /*#define DYN_ALLOC*/
205 # define EXTERN(type, array) extern type * array
206 # define DECLARE(type, array, size) type * array
207 # define ALLOC(type, array, size) { \
208 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
209 if (array == NULL) errorMsg("insufficient memory"); \
211 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
213 # define EXTERN(type, array) extern type array[]
214 # define DECLARE(type, array, size) type array[size]
215 # define ALLOC(type, array, size)
219 EXTERN(uch, inbuf); /* input buffer */
220 EXTERN(uch, outbuf); /* output buffer */
221 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
222 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
223 #define tab_suffix window
225 # define tab_prefix prev /* hash link (see deflate.c) */
226 # define head (prev+WSIZE) /* hash head (see deflate.c) */
227 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
229 # define tab_prefix0 prev
230 # define head tab_prefix1
231 EXTERN(ush, tab_prefix0); /* prefix for even codes */
232 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
235 extern unsigned insize; /* valid bytes in inbuf */
236 extern unsigned inptr; /* index of next byte to be processed in inbuf */
237 extern unsigned outcnt; /* bytes in output buffer */
239 extern long bytes_in; /* number of input bytes */
240 extern long bytes_out; /* number of output bytes */
241 extern long header_bytes; /* number of bytes in gzip header */
243 extern long ifile_size; /* input file size, -1 for devices (debug only) */
245 typedef int file_t; /* Do not use stdio */
247 #define NO_FILE (-1) /* in memory compression */
250 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
253 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
254 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
255 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
256 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
257 #define COMMENT 0x10 /* bit 4 set: file comment present */
258 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
259 #define RESERVED 0xC0 /* bit 6,7: reserved */
262 # define WSIZE 0x8000 /* window size--must be a power of two, and */
263 #endif /* at least 32K for zip's deflate method */
266 #define MAX_MATCH 258
267 /* The minimum and maximum match lengths */
269 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
270 /* Minimum amount of lookahead, except at the end of the input file.
271 * See deflate.c for comments about the MIN_MATCH+1.
274 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
275 /* In order to simplify the code, particularly on 16 bit machines, match
276 * distances are limited to MAX_DIST instead of WSIZE.
279 extern int exit_code; /* program exit code */
280 extern int verbose; /* be verbose (-v) */
281 extern int level; /* compression level */
282 extern int test; /* check .z file integrity */
283 extern int save_orig_name; /* set if original name must be saved */
285 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
286 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
288 /* put_byte is used for the compressed output, put_ubyte for the
289 * uncompressed output. However unlzw() uses window for its
290 * suffix table instead of its output buffer, so it does not use put_ubyte
291 * (to be cleaned up).
293 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
295 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
298 /* Output a 16 bit value, lsb first */
299 #define put_short(w) \
300 { if (outcnt < OUTBUFSIZ-2) { \
301 outbuf[outcnt++] = (uch) ((w) & 0xff); \
302 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
304 put_byte((uch)((w) & 0xff)); \
305 put_byte((uch)((ush)(w) >> 8)); \
309 /* Output a 32 bit value to the bit stream, lsb first */
310 #define put_long(n) { \
311 put_short((n) & 0xffff); \
312 put_short(((ulg)(n)) >> 16); \
315 #define seekable() 0 /* force sequential output */
316 #define translate_eol 0 /* no option -a yet */
318 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
320 /* Macros for getting two-byte and four-byte header values */
321 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
322 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
324 /* Diagnostic functions */
326 # define Assert(cond,msg) {if(!(cond)) errorMsg(msg);}
327 # define Trace(x) fprintf x
328 # define Tracev(x) {if (verbose) fprintf x ;}
329 # define Tracevv(x) {if (verbose>1) fprintf x ;}
330 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
331 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
333 # define Assert(cond,msg)
338 # define Tracecv(c,x)
341 #define WARN(msg) {fprintf msg ; \
342 if (exit_code == OK) exit_code = WARNING;}
345 extern int unzip (int in, int out);
348 RETSIGTYPE abort_gzip (void);
351 void lm_init (int pack_level, ush * flags);
355 void ct_init (ush * attr, int *method);
356 int ct_tally (int dist, int lc);
357 ulg flush_block (char *buf, ulg stored_len, int eof);
360 void bi_init (file_t zipfile);
361 void send_bits (int value, int length);
362 unsigned bi_reverse (unsigned value, int length);
363 void bi_windup (void);
364 void copy_block (char *buf, unsigned len, int header);
365 extern int (*read_buf) (char *buf, unsigned size);
368 extern int copy (int in, int out);
369 extern ulg updcrc (uch * s, unsigned n);
370 extern void clear_bufs (void);
371 extern int fill_inbuf (int eof_ok);
372 extern void flush_outbuf (void);
373 extern void flush_window (void);
374 extern void write_buf (int fd, void * buf, unsigned cnt);
377 extern char *basename (char *fname);
378 #endif /* not __linux__ */
379 extern void read_error (void);
380 extern void write_error (void);
383 extern int inflate (void);
385 /* #include "lzw.h" */
387 /* lzw.h -- define the lzw functions.
388 * Copyright (C) 1992-1993 Jean-loup Gailly.
389 * This is free software; you can redistribute it and/or modify it under the
390 * terms of the GNU General Public License, see the file COPYING.
393 #if !defined(OF) && defined(lint)
400 #define INIT_BITS 9 /* Initial number of bits per code */
402 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
404 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
405 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
406 * It's a pity that old uncompress does not check bit 0x20. That makes
407 * extension of the format actually undesirable because old compress
408 * would just crash on the new format instead of giving a meaningful
409 * error message. It does check the number of bits, but it's more
410 * helpful to say "unsupported format, get a new version" than
411 * "can only handle 16 bits".
414 #define BLOCK_MODE 0x80
415 /* Block compression: if table is full and compression rate is dropping,
416 * clear the dictionary.
419 #define LZW_RESERVED 0x60 /* reserved bits */
421 #define CLEAR 256 /* flush the dictionary */
422 #define FIRST (CLEAR+1) /* first free entry */
424 extern int maxbits; /* max bits per code for LZW */
425 extern int block_mode; /* block compress mode -C compatible with 2.0 */
427 extern int lzw (int in, int out);
428 extern int unlzw (int in, int out);
431 /* #include "revision.h" */
433 /* revision.h -- define the version number
434 * Copyright (C) 1992-1993 Jean-loup Gailly.
435 * This is free software; you can redistribute it and/or modify it under the
436 * terms of the GNU General Public License, see the file COPYING.
439 #define VERSION "1.2.4"
441 #define REVDATE "18 Aug 93"
443 /* This version does not support compression into old compress format: */
448 /* #include "getopt.h" */
450 /* Declarations for getopt.
451 Copyright (C) 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc.
453 This program is free software; you can redistribute it and/or modify it
454 under the terms of the GNU General Public License as published by the
455 Free Software Foundation; either version 2, or (at your option) any
458 This program is distributed in the hope that it will be useful,
459 but WITHOUT ANY WARRANTY; without even the implied warranty of
460 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
461 GNU General Public License for more details.
463 You should have received a copy of the GNU General Public License
464 along with this program; if not, write to the Free Software
465 Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
473 /* For communication from `getopt' to the caller.
474 When `getopt' finds an option that takes an argument,
475 the argument value is returned here.
476 Also, when `ordering' is RETURN_IN_ORDER,
477 each non-option ARGV-element is returned here. */
480 /* Index in ARGV of the next element to be scanned.
481 This is used for communication to and from the caller
482 and for communication between successive calls to `getopt'.
484 On entry to `getopt', zero means this is the first call; initialize.
486 When `getopt' returns EOF, this is the index of the first of the
487 non-option elements that the caller should itself scan.
489 Otherwise, `optind' communicates from one call to the next
490 how much of ARGV has been scanned so far. */
494 /* Callers store zero here to inhibit the error message `getopt' prints
495 for unrecognized options. */
499 /* Set to an option character which was unrecognized. */
503 /* Describe the long-named options requested by the application.
504 The LONG_OPTIONS argument to getopt_long or getopt_long_only is a vector
505 of `struct option' terminated by an element containing a name which is
508 The field `has_arg' is:
509 no_argument (or 0) if the option does not take an argument,
510 required_argument (or 1) if the option requires an argument,
511 optional_argument (or 2) if the option takes an optional argument.
513 If the field `flag' is not NULL, it points to a variable that is set
514 to the value given in the field `val' when the option is found, but
515 left unchanged if the option is not found.
517 To have a long-named option do something other than set an `int' to
518 a compiled-in constant, such as set a value from `optarg', set the
519 option's `flag' field to zero and its `val' field to a nonzero
520 value (the equivalent single-letter option character, if there is
521 one). For long options that have a zero `flag' field, `getopt'
522 returns the contents of the `val' field. */
530 /* has_arg can't be an enum because some compilers complain about
531 type mismatches in all the code that assumes it is an int. */
537 /* Names for the values of the `has_arg' field of `struct option'. */
539 #define no_argument 0
540 #define required_argument 1
541 #define optional_argument 2
543 #if __STDC__ || defined(PROTO)
544 #if defined(__GNU_LIBRARY__)
545 /* Many other libraries have conflicting prototypes for getopt, with
546 differences in the consts, in stdlib.h. To avoid compilation
547 errors, only prototype getopt for the GNU C library. */
548 extern int getopt(int argc, char *const *argv, const char *shortopts);
549 #endif /* not __GNU_LIBRARY__ */
550 extern int getopt_long(int argc, char *const *argv,
551 const char *shortopts,
552 const struct option *longopts, int *longind);
553 extern int getopt_long_only(int argc, char *const *argv,
554 const char *shortopts,
555 const struct option *longopts,
558 /* Internal only. Users should not call this directly. */
559 extern int _getopt_internal(int argc, char *const *argv,
560 const char *shortopts,
561 const struct option *longopts,
562 int *longind, int long_only);
563 #else /* not __STDC__ */
565 extern int getopt_long();
566 extern int getopt_long_only();
568 extern int _getopt_internal();
569 #endif /* not __STDC__ */
574 #endif /* _GETOPT_H */
581 typedef struct dirent dir_type;
583 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
584 # define DIR_OPT "DIRENT"
586 # define NLENGTH(dirent) ((dirent)->d_namlen)
588 # include <sys/dir.h>
589 typedef struct direct dir_type;
591 # define DIR_OPT "SYSDIR"
594 # include <sys/ndir.h>
595 typedef struct direct dir_type;
597 # define DIR_OPT "SYSNDIR"
601 typedef struct direct dir_type;
603 # define DIR_OPT "NDIR"
606 # define DIR_OPT "NO_DIR"
611 #if !defined(S_ISDIR) && defined(S_IFDIR)
612 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
614 #if !defined(S_ISREG) && defined(S_IFREG)
615 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
617 typedef RETSIGTYPE(*sig_type) (int);
620 # define O_BINARY 0 /* creation mode for open() */
624 /* Pure BSD system? */
625 # include <sys/file.h>
627 # define O_CREAT FCREAT
630 # define O_EXCL FEXCL
635 # define S_IRUSR 0400
638 # define S_IWUSR 0200
640 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
642 #ifndef MAX_PATH_LEN /* max pathname length */
644 # define MAX_PATH_LEN BUFSIZ
646 # define MAX_PATH_LEN 1024
656 off_t lseek (int fd, off_t offset, int whence);
662 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
663 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
664 DECLARE(ush, d_buf, DIST_BUFSIZE);
665 DECLARE(uch, window, 2L * WSIZE);
667 DECLARE(ush, tab_prefix, 1L << BITS);
669 DECLARE(ush, tab_prefix0, 1L << (BITS - 1));
670 DECLARE(ush, tab_prefix1, 1L << (BITS - 1));
673 /* local variables */
675 int test_mode = 0; /* check file integrity option */
676 int foreground; /* set if program run in foreground */
677 int maxbits = BITS; /* max bits per code for LZW */
678 int method = DEFLATED; /* compression method */
679 int exit_code = OK; /* program exit code */
680 int last_member; /* set for .zip and .Z files */
681 int part_nb; /* number of parts in .gz file */
682 long ifile_size; /* input file size, -1 for devices (debug only) */
684 long bytes_in; /* number of input bytes */
685 long bytes_out; /* number of output bytes */
686 long total_in = 0; /* input bytes for all files */
687 long total_out = 0; /* output bytes for all files */
688 struct stat istat; /* status for input file */
689 int ifd; /* input file descriptor */
690 int ofd; /* output file descriptor */
691 unsigned insize; /* valid bytes in inbuf */
692 unsigned inptr; /* index of next byte to be processed in inbuf */
693 unsigned outcnt; /* bytes in output buffer */
695 long header_bytes; /* number of bytes in gzip header */
697 /* local functions */
699 local int get_method (int in);
701 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
703 /* ======================================================================== */
704 int gunzip_main(int argc, char **argv)
706 int file_count; /* number of files to precess */
712 int delInputFile = 0;
715 char ifname[MAX_PATH_LEN + 1]; /* input file name */
716 char ofname[MAX_PATH_LEN + 1]; /* output file name */
721 if (strcmp(*argv, "zcat") == 0)
724 /* Parse any options */
725 while (--argc > 0 && **(++argv) == '-') {
726 if (*((*argv) + 1) == '\0') {
730 while (*(++(*argv))) {
745 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
747 (void) signal(SIGINT, (sig_type) abort_gzip);
750 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
751 (void) signal(SIGTERM, (sig_type) abort_gzip);
755 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
756 (void) signal(SIGHUP, (sig_type) abort_gzip);
760 file_count = argc - optind;
762 /* Allocate all global buffers (for DYN_ALLOC option) */
763 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
764 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
765 ALLOC(ush, d_buf, DIST_BUFSIZE);
766 ALLOC(uch, window, 2L * WSIZE);
768 ALLOC(ush, tab_prefix, 1L << BITS);
770 ALLOC(ush, tab_prefix0, 1L << (BITS - 1));
771 ALLOC(ush, tab_prefix1, 1L << (BITS - 1));
774 if (fromstdin == 1) {
775 strcpy(ofname, "stdin");
777 inFileNum = fileno(stdin);
778 ifile_size = -1L; /* convention for unknown size */
780 /* Open up the input file */
783 if (strlen(*argv) > MAX_PATH_LEN) {
784 fprintf(stderr, name_too_long, "gunzip");
787 strcpy(ifname, *argv);
789 /* Open input fille */
790 inFileNum = open(ifname, O_RDONLY);
795 /* Get the time stamp on the input file. */
796 result = stat(ifname, &statBuf);
801 ifile_size = statBuf.st_size;
804 if (to_stdout == 1) {
805 /* And get to work */
806 strcpy(ofname, "stdout");
807 outFileNum = fileno(stdout);
809 clear_bufs(); /* clear input and output buffers */
812 /* Actually do the compression/decompression. */
813 unzip(inFileNum, outFileNum);
815 } else if (test_mode) {
816 /* Actually do the compression/decompression. */
821 /* And get to work */
822 if (strlen(ifname) > MAX_PATH_LEN - 4) {
823 fprintf(stderr, name_too_long, "gunzip");
826 strcpy(ofname, ifname);
827 pos = strstr(ofname, ".gz");
832 pos = strstr(ofname, ".tgz");
840 /* Open output fille */
841 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
842 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
844 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL);
846 if (outFileNum < 0) {
850 /* Set permissions on the file */
851 fchmod(outFileNum, statBuf.st_mode);
853 clear_bufs(); /* clear input and output buffers */
856 /* Actually do the compression/decompression. */
857 result = unzip(inFileNum, outFileNum);
861 /* Delete the original file */
863 delFileName = ifname;
865 delFileName = ofname;
867 if (delInputFile == 1 && unlink(delFileName) < 0) {
876 /* ========================================================================
877 * Check the magic number of the input file and update ofname if an
878 * original name was given and to_stdout is not set.
879 * Return the compression method, -1 for error, -2 for warning.
880 * Set inptr to the offset of the next byte to be processed.
881 * Updates time_stamp if there is one and --no-time is not used.
882 * This function may be called repeatedly for an input file consisting
883 * of several contiguous gzip'ed members.
884 * IN assertions: there is at least one remaining compressed member.
885 * If the member is a zip file, it must be the only one.
887 local int get_method(in)
888 int in; /* input file descriptor */
890 uch flags; /* compression flags */
891 char magic[2]; /* magic header */
893 magic[0] = (char) get_byte();
894 magic[1] = (char) get_byte();
895 method = -1; /* unknown yet */
896 part_nb++; /* number of parts in gzip file */
898 last_member = RECORD_IO;
899 /* assume multiple members in gzip file except for record oriented I/O */
901 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
903 method = (int) get_byte();
904 if (method != DEFLATED) {
906 "unknown method %d -- get newer version of gzip\n",
911 flags = (uch) get_byte();
913 (ulg) get_byte(); /* Ignore time stamp */
918 (void) get_byte(); /* Ignore extra flags for the moment */
919 (void) get_byte(); /* Ignore OS type for the moment */
921 if ((flags & EXTRA_FIELD) != 0) {
922 unsigned len = (unsigned) get_byte();
924 len |= ((unsigned) get_byte()) << 8;
930 /* Discard original name if any */
931 if ((flags & ORIG_NAME) != 0) {
932 while (get_char() != 0) /* null */
936 /* Discard file comment if any */
937 if ((flags & COMMENT) != 0) {
938 while (get_char() != 0) /* null */
942 header_bytes = inptr + 2 * sizeof(long); /* include crc and size */
951 fprintf(stderr, "\nnot in gzip format\n");
955 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
960 /* ========================================================================
961 * Signal and error handler.
963 RETSIGTYPE abort_gzip()
968 /* unzip.c -- decompress files in gzip or pkzip format.
969 * Copyright (C) 1992-1993 Jean-loup Gailly
970 * This is free software; you can redistribute it and/or modify it under the
971 * terms of the GNU General Public License, see the file COPYING.
973 * The code in this file is derived from the file funzip.c written
974 * and put in the public domain by Mark Adler.
978 This version can extract files in gzip or pkzip format.
979 For the latter, only the first entry is extracted, and it has to be
980 either deflated or stored.
983 /* #include "crypt.h" */
985 /* crypt.h (dummy version) -- do not perform encryption
986 * Hardly worth copyrighting :-)
990 # undef CRYPT /* dummy version */
993 #define RAND_HEAD_LEN 12 /* length of encryption random header */
998 /* PKZIP header definitions */
999 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
1000 #define LOCFLG 6 /* offset of bit flag */
1001 #define CRPFLG 1 /* bit for encrypted entry */
1002 #define EXTFLG 8 /* bit for extended local header */
1003 #define LOCHOW 8 /* offset of compression method */
1004 #define LOCTIM 10 /* file mod time (for decryption) */
1005 #define LOCCRC 14 /* offset of crc */
1006 #define LOCSIZ 18 /* offset of compressed size */
1007 #define LOCLEN 22 /* offset of uncompressed length */
1008 #define LOCFIL 26 /* offset of file name field length */
1009 #define LOCEXT 28 /* offset of extra field length */
1010 #define LOCHDR 30 /* size of local header, including sig */
1011 #define EXTHDR 16 /* size of extended local header, inc sig */
1016 char *key; /* not used--needed to link crypt.c */
1017 int pkzip = 0; /* set for a pkzip file */
1018 int ext_header = 0; /* set if extended local header */
1020 /* ===========================================================================
1021 * Unzip in to out. This routine works on both gzip and pkzip files.
1023 * IN assertions: the buffer inbuf contains already the beginning of
1024 * the compressed data, from offsets inptr to insize-1 included.
1025 * The magic header has already been checked. The output buffer is cleared.
1028 int in, out; /* input and output file descriptors */
1030 ulg orig_crc = 0; /* original crc */
1031 ulg orig_len = 0; /* original uncompressed length */
1033 uch buf[EXTHDR]; /* extended local header */
1037 method = get_method(ifd);
1039 exit(exit_code); /* error message already emitted */
1042 updcrc(NULL, 0); /* initialize crc */
1044 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
1045 orig_crc = LG(inbuf + LOCCRC);
1046 orig_len = LG(inbuf + LOCLEN);
1050 if (method == DEFLATED) {
1052 int res = inflate();
1055 errorMsg("out of memory");
1056 } else if (res != 0) {
1057 errorMsg("invalid compressed data--format violated");
1061 errorMsg("internal error, invalid method");
1064 /* Get the crc and original length */
1066 /* crc32 (see algorithm.doc)
1067 * uncompressed input size modulo 2^32
1069 for (n = 0; n < 8; n++) {
1070 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
1073 orig_len = LG(buf + 4);
1075 } else if (ext_header) { /* If extended header, check it */
1076 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
1078 * compressed size 4-bytes
1079 * uncompressed size 4-bytes
1081 for (n = 0; n < EXTHDR; n++) {
1082 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
1084 orig_crc = LG(buf + 4);
1085 orig_len = LG(buf + 12);
1088 /* Validate decompression */
1089 if (orig_crc != updcrc(outbuf, 0)) {
1090 errorMsg("invalid compressed data--crc error");
1092 if (orig_len != (ulg) bytes_out) {
1093 errorMsg("invalid compressed data--length error");
1096 /* Check if there are more entries in a pkzip file */
1097 if (pkzip && inptr + 4 < insize && LG(inbuf + inptr) == LOCSIG) {
1098 WARN((stderr, "has more than one entry--rest ignored\n"));
1100 ext_header = pkzip = 0; /* for next file */
1104 /* util.c -- utility functions for gzip support
1105 * Copyright (C) 1992-1993 Jean-loup Gailly
1106 * This is free software; you can redistribute it and/or modify it under the
1107 * terms of the GNU General Public License, see the file COPYING.
1112 #include <sys/types.h>
1114 #ifdef HAVE_UNISTD_H
1115 # include <unistd.h>
1121 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1122 # include <stdlib.h>
1127 static const ulg crc_32_tab[]; /* crc table, defined below */
1129 /* ===========================================================================
1130 * Run a set of bytes through the crc shift register. If s is a NULL
1131 * pointer, then initialize the crc shift register contents instead.
1132 * Return the current crc in either case.
1135 uch *s; /* pointer to bytes to pump through */
1136 unsigned n; /* number of bytes in s[] */
1138 register ulg c; /* temporary variable */
1140 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
1148 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
1152 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1155 /* ===========================================================================
1156 * Clear input and output buffers
1162 bytes_in = bytes_out = 0L;
1165 /* ===========================================================================
1166 * Fill the input buffer. This is called only when the buffer is empty.
1168 int fill_inbuf(eof_ok)
1169 int eof_ok; /* set if EOF acceptable as a result */
1173 /* Read as much as possible */
1177 len = read(ifd, (char *) inbuf + insize, INBUFSIZ - insize);
1178 if (len == 0 || len == EOF)
1181 } while (insize < INBUFSIZ);
1188 bytes_in += (ulg) insize;
1193 /* ===========================================================================
1194 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1195 * (used for the compressed data only)
1203 write_buf(ofd, (char *) outbuf, outcnt);
1204 bytes_out += (ulg) outcnt;
1208 /* ===========================================================================
1209 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1210 * (Used for the decompressed data only.)
1216 updcrc(window, outcnt);
1219 write_buf(ofd, (char *) window, outcnt);
1220 bytes_out += (ulg) outcnt;
1224 /* ===========================================================================
1225 * Does the same as write(), but also handles partial pipe writes and checks
1228 void write_buf(fd, buf, cnt)
1235 while ((n = write(fd, buf, cnt)) != cnt) {
1236 if (n == (unsigned) (-1)) {
1240 buf = (void *) ((char *) buf + n);
1244 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1246 /* Provide missing strspn and strcspn functions. */
1252 int strspn (const char *s, const char *accept);
1253 int strcspn (const char *s, const char *reject);
1255 /* ========================================================================
1256 * Return the length of the maximum initial segment
1257 * of s which contains only characters in accept.
1259 int strspn(s, accept)
1263 register const char *p;
1264 register const char *a;
1265 register int count = 0;
1267 for (p = s; *p != '\0'; ++p) {
1268 for (a = accept; *a != '\0'; ++a) {
1279 /* ========================================================================
1280 * Return the length of the maximum inital segment of s
1281 * which contains no characters from reject.
1283 int strcspn(s, reject)
1287 register int count = 0;
1289 while (*s != '\0') {
1290 if (strchr(reject, *s++) != NULL)
1297 #endif /* NO_STRING_H */
1300 /* ========================================================================
1305 fprintf(stderr, "\n");
1309 fprintf(stderr, "unexpected end of file\n");
1316 fprintf(stderr, "\n");
1322 /* ========================================================================
1323 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1325 static const ulg crc_32_tab[] = {
1326 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1327 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1328 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1329 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1330 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1331 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1332 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1333 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1334 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1335 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1336 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1337 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1338 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1339 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1340 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1341 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1342 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1343 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1344 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1345 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1346 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1347 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1348 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1349 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1350 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1351 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1352 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1353 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1354 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1355 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1356 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1357 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1358 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1359 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1360 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1361 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1362 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1363 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1364 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1365 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1366 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1367 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1368 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1369 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1370 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1371 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1372 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1373 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1374 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1375 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1376 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1380 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1381 version c10p1, 10 January 1993 */
1383 /* You can do whatever you like with this source file, though I would
1384 prefer that if you modify it and redistribute it that you include
1385 comments to that effect with your name and the date. Thank you.
1386 [The history has been moved to the file ChangeLog.]
1390 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1391 method searches for as much of the current string of bytes (up to a
1392 length of 258) in the previous 32K bytes. If it doesn't find any
1393 matches (of at least length 3), it codes the next byte. Otherwise, it
1394 codes the length of the matched string and its distance backwards from
1395 the current position. There is a single Huffman code that codes both
1396 single bytes (called "literals") and match lengths. A second Huffman
1397 code codes the distance information, which follows a length code. Each
1398 length or distance code actually represents a base value and a number
1399 of "extra" (sometimes zero) bits to get to add to the base value. At
1400 the end of each deflated block is a special end-of-block (EOB) literal/
1401 length code. The decoding process is basically: get a literal/length
1402 code; if EOB then done; if a literal, emit the decoded byte; if a
1403 length then get the distance and emit the referred-to bytes from the
1404 sliding window of previously emitted data.
1406 There are (currently) three kinds of inflate blocks: stored, fixed, and
1407 dynamic. The compressor deals with some chunk of data at a time, and
1408 decides which method to use on a chunk-by-chunk basis. A chunk might
1409 typically be 32K or 64K. If the chunk is uncompressible, then the
1410 "stored" method is used. In this case, the bytes are simply stored as
1411 is, eight bits per byte, with none of the above coding. The bytes are
1412 preceded by a count, since there is no longer an EOB code.
1414 If the data is compressible, then either the fixed or dynamic methods
1415 are used. In the dynamic method, the compressed data is preceded by
1416 an encoding of the literal/length and distance Huffman codes that are
1417 to be used to decode this block. The representation is itself Huffman
1418 coded, and so is preceded by a description of that code. These code
1419 descriptions take up a little space, and so for small blocks, there is
1420 a predefined set of codes, called the fixed codes. The fixed method is
1421 used if the block codes up smaller that way (usually for quite small
1422 chunks), otherwise the dynamic method is used. In the latter case, the
1423 codes are customized to the probabilities in the current block, and so
1424 can code it much better than the pre-determined fixed codes.
1426 The Huffman codes themselves are decoded using a mutli-level table
1427 lookup, in order to maximize the speed of decoding plus the speed of
1428 building the decoding tables. See the comments below that precede the
1429 lbits and dbits tuning parameters.
1434 Notes beyond the 1.93a appnote.txt:
1436 1. Distance pointers never point before the beginning of the output
1438 2. Distance pointers can point back across blocks, up to 32k away.
1439 3. There is an implied maximum of 7 bits for the bit length table and
1440 15 bits for the actual data.
1441 4. If only one code exists, then it is encoded using one bit. (Zero
1442 would be more efficient, but perhaps a little confusing.) If two
1443 codes exist, they are coded using one bit each (0 and 1).
1444 5. There is no way of sending zero distance codes--a dummy must be
1445 sent if there are none. (History: a pre 2.0 version of PKZIP would
1446 store blocks with no distance codes, but this was discovered to be
1447 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1448 zero distance codes, which is sent as one code of zero bits in
1450 6. There are up to 286 literal/length codes. Code 256 represents the
1451 end-of-block. Note however that the static length tree defines
1452 288 codes just to fill out the Huffman codes. Codes 286 and 287
1453 cannot be used though, since there is no length base or extra bits
1454 defined for them. Similarly, there are up to 30 distance codes.
1455 However, static trees define 32 codes (all 5 bits) to fill out the
1456 Huffman codes, but the last two had better not show up in the data.
1457 7. Unzip can check dynamic Huffman blocks for complete code sets.
1458 The exception is that a single code would not be complete (see #4).
1459 8. The five bits following the block type is really the number of
1460 literal codes sent minus 257.
1461 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1462 (1+6+6). Therefore, to output three times the length, you output
1463 three codes (1+1+1), whereas to output four times the same length,
1464 you only need two codes (1+3). Hmm.
1465 10. In the tree reconstruction algorithm, Code = Code + Increment
1466 only if BitLength(i) is not zero. (Pretty obvious.)
1467 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1468 12. Note: length code 284 can represent 227-258, but length code 285
1469 really is 258. The last length deserves its own, short code
1470 since it gets used a lot in very redundant files. The length
1471 258 is special since 258 - 3 (the min match length) is 255.
1472 13. The literal/length and distance code bit lengths are read as a
1473 single stream of lengths. It is possible (and advantageous) for
1474 a repeat code (16, 17, or 18) to go across the boundary between
1475 the two sets of lengths.
1478 #include <sys/types.h>
1480 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1481 # include <stdlib.h>
1485 #define slide window
1487 /* Huffman code lookup table entry--this entry is four bytes for machines
1488 that have 16-bit pointers (e.g. PC's in the small or medium model).
1489 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1490 means that v is a literal, 16 < e < 32 means that v is a pointer to
1491 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1492 an unused code. If a code with e == 99 is looked up, this implies an
1493 error in the data. */
1495 uch e; /* number of extra bits or operation */
1496 uch b; /* number of bits in this code or subcode */
1498 ush n; /* literal, length base, or distance base */
1499 struct huft *t; /* pointer to next level of table */
1504 /* Function prototypes */
1505 int huft_build (unsigned *, unsigned, unsigned, ush *, ush *,
1506 struct huft **, int *);
1507 int huft_free (struct huft *);
1508 int inflate_codes (struct huft *, struct huft *, int, int);
1509 int inflate_stored (void);
1510 int inflate_fixed (void);
1511 int inflate_dynamic (void);
1512 int inflate_block (int *);
1516 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1517 stream to find repeated byte strings. This is implemented here as a
1518 circular buffer. The index is updated simply by incrementing and then
1519 and'ing with 0x7fff (32K-1). */
1520 /* It is left to other modules to supply the 32K area. It is assumed
1521 to be usable as if it were declared "uch slide[32768];" or as just
1522 "uch *slide;" and then malloc'ed in the latter case. The definition
1523 must be in unzip.h, included above. */
1524 /* unsigned wp; current position in slide */
1526 #define flush_output(w) (wp=(w),flush_window())
1528 /* Tables for deflate from PKZIP's appnote.txt. */
1529 static unsigned border[] = { /* Order of the bit length code lengths */
1530 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15
1532 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1533 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1534 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0
1537 /* note: see note #13 above about the 258 in this list. */
1538 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1539 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1540 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99
1541 }; /* 99==invalid */
1542 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1543 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1544 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1545 8193, 12289, 16385, 24577
1547 static ush cpdext[] = { /* Extra bits for distance codes */
1548 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1549 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1555 /* Macros for inflate() bit peeking and grabbing.
1559 x = b & mask_bits[j];
1562 where NEEDBITS makes sure that b has at least j bits in it, and
1563 DUMPBITS removes the bits from b. The macros use the variable k
1564 for the number of bits in b. Normally, b and k are register
1565 variables for speed, and are initialized at the beginning of a
1566 routine that uses these macros from a global bit buffer and count.
1568 If we assume that EOB will be the longest code, then we will never
1569 ask for bits with NEEDBITS that are beyond the end of the stream.
1570 So, NEEDBITS should not read any more bytes than are needed to
1571 meet the request. Then no bytes need to be "returned" to the buffer
1572 at the end of the last block.
1574 However, this assumption is not true for fixed blocks--the EOB code
1575 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1576 (The EOB code is shorter than other codes because fixed blocks are
1577 generally short. So, while a block always has an EOB, many other
1578 literal/length codes have a significantly lower probability of
1579 showing up at all.) However, by making the first table have a
1580 lookup of seven bits, the EOB code will be found in that first
1581 lookup, and so will not require that too many bits be pulled from
1585 ulg bb; /* bit buffer */
1586 unsigned bk; /* bits in bit buffer */
1590 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1591 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1597 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1599 # define NEXTBYTE() (uch)get_byte()
1601 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1602 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1606 Huffman code decoding is performed using a multi-level table lookup.
1607 The fastest way to decode is to simply build a lookup table whose
1608 size is determined by the longest code. However, the time it takes
1609 to build this table can also be a factor if the data being decoded
1610 is not very long. The most common codes are necessarily the
1611 shortest codes, so those codes dominate the decoding time, and hence
1612 the speed. The idea is you can have a shorter table that decodes the
1613 shorter, more probable codes, and then point to subsidiary tables for
1614 the longer codes. The time it costs to decode the longer codes is
1615 then traded against the time it takes to make longer tables.
1617 This results of this trade are in the variables lbits and dbits
1618 below. lbits is the number of bits the first level table for literal/
1619 length codes can decode in one step, and dbits is the same thing for
1620 the distance codes. Subsequent tables are also less than or equal to
1621 those sizes. These values may be adjusted either when all of the
1622 codes are shorter than that, in which case the longest code length in
1623 bits is used, or when the shortest code is *longer* than the requested
1624 table size, in which case the length of the shortest code in bits is
1627 There are two different values for the two tables, since they code a
1628 different number of possibilities each. The literal/length table
1629 codes 286 possible values, or in a flat code, a little over eight
1630 bits. The distance table codes 30 possible values, or a little less
1631 than five bits, flat. The optimum values for speed end up being
1632 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1633 The optimum values may differ though from machine to machine, and
1634 possibly even between compilers. Your mileage may vary.
1638 int lbits = 9; /* bits in base literal/length lookup table */
1639 int dbits = 6; /* bits in base distance lookup table */
1642 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1643 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1644 #define N_MAX 288 /* maximum number of codes in any set */
1647 unsigned hufts; /* track memory usage */
1650 int huft_build(b, n, s, d, e, t, m)
1651 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1652 unsigned n; /* number of codes (assumed <= N_MAX) */
1653 unsigned s; /* number of simple-valued codes (0..s-1) */
1654 ush *d; /* list of base values for non-simple codes */
1655 ush *e; /* list of extra bits for non-simple codes */
1656 struct huft **t; /* result: starting table */
1657 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));
1694 && n - i <= '~' ? "%c %d\n" : "0x%x %d\n"), n - i, *p));
1695 c[*p]++; /* assume all entries <= BMAX */
1696 p++; /* Can't combine with above line (Solaris bug) */
1698 if (c[0] == n) { /* null input--all zero length codes */
1699 *t = (struct huft *) NULL;
1705 /* Find minimum and maximum length, bound *m by those */
1707 for (j = 1; j <= BMAX; j++)
1710 k = j; /* minimum code length */
1711 if ((unsigned) l < j)
1713 for (i = BMAX; i; i--)
1716 g = i; /* maximum code length */
1717 if ((unsigned) l > i)
1722 /* Adjust last length count to fill out codes, if needed */
1723 for (y = 1 << j; j < i; j++, y <<= 1)
1724 if ((y -= c[j]) < 0)
1725 return 2; /* bad input: more codes than bits */
1726 if ((y -= c[i]) < 0)
1731 /* Generate starting offsets into the value table for each length */
1735 while (--i) { /* note that i == g from above */
1736 *xp++ = (j += *p++);
1740 /* Make a table of values in order of bit lengths */
1744 if ((j = *p++) != 0)
1749 /* Generate the Huffman codes and for each, make the table entries */
1750 x[0] = i = 0; /* first Huffman code is zero */
1751 p = v; /* grab values in bit order */
1752 h = -1; /* no tables yet--level -1 */
1753 w = -l; /* bits decoded == (l * h) */
1754 u[0] = (struct huft *) NULL; /* just to keep compilers happy */
1755 q = (struct huft *) NULL; /* ditto */
1758 /* go through the bit lengths (k already is bits in shortest code) */
1759 for (; k <= g; k++) {
1762 /* here i is the Huffman code of length k bits for value *p */
1763 /* make tables up to required level */
1766 w += l; /* previous table always l bits */
1768 /* compute minimum size table less than or equal to l bits */
1769 z = (z = g - w) > (unsigned) l ? l : z; /* upper limit on table size */
1770 if ((f = 1 << (j = k - w)) > a + 1) { /* try a k-w bit table *//* 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 */
1774 if ((f <<= 1) <= *++xp)
1775 break; /* enough codes to use up j bits */
1776 f -= *xp; /* else deduct codes from patterns */
1779 z = 1 << j; /* table entries for j-bit table */
1781 /* allocate and link in new table */
1784 (struct huft *) malloc((z + 1) *
1785 sizeof(struct huft))) ==
1786 (struct huft *) NULL) {
1789 return 3; /* not enough memory */
1791 hufts += z + 1; /* track memory usage */
1792 *t = q + 1; /* link to list for huft_free() */
1793 *(t = &(q->v.t)) = (struct huft *) NULL;
1794 u[h] = ++q; /* table starts after link */
1796 /* 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 */
1808 r.b = (uch) (k - w);
1810 r.e = 99; /* out of values--invalid code */
1812 r.e = (uch) (*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1813 r.v.n = (ush) (*p); /* simple code is just the value */
1814 p++; /* one compiler does not like *p++ */
1816 r.e = (uch) e[*p - s]; /* non-simple--look up in lists */
1817 r.v.n = d[*p++ - s];
1820 /* fill code-like entries with r */
1822 for (j = i >> w; j < z; j += f)
1825 /* backwards increment the k-bit code i */
1826 for (j = 1 << (k - 1); i & j; j >>= 1)
1830 /* backup over finished tables */
1831 while ((i & ((1 << w) - 1)) != x[h]) {
1832 h--; /* don't need to update q */
1839 /* Return true (1) if we were given an incomplete table */
1840 return y != 0 && g != 1;
1846 struct huft *t; /* table to free */
1848 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1849 list of the tables it made, with the links in a dummy first entry of
1852 register struct huft *p, *q;
1855 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1857 while (p != (struct huft *) NULL) {
1866 int inflate_codes(tl, td, bl, bd)
1867 struct huft *tl, *td; /* literal/length and distance decoder tables */
1868 int bl, bd; /* number of bits decoded by tl[] and td[] */
1870 /* inflate (decompress) the codes in a deflated (compressed) block.
1871 Return an error code or zero if it all goes ok. */
1873 register unsigned e; /* table entry flag/number of extra bits */
1874 unsigned n, d; /* length and index for copy */
1875 unsigned w; /* current window position */
1876 struct huft *t; /* pointer to table entry */
1877 unsigned ml, md; /* masks for bl and bd bits */
1878 register ulg b; /* bit buffer */
1879 register unsigned k; /* number of bits in bit buffer */
1882 /* make local copies of globals */
1883 b = bb; /* initialize bit buffer */
1885 w = wp; /* initialize window position */
1887 /* inflate the coded data */
1888 ml = mask_bits[bl]; /* precompute masks for speed */
1890 for (;;) { /* do until end of block */
1891 NEEDBITS((unsigned) bl)
1892 if ((e = (t = tl + ((unsigned) b & ml))->e) > 16)
1899 } while ((e = (t = t->v.t + ((unsigned) b & mask_bits[e]))->e)
1902 if (e == 16) { /* then it's a literal */
1903 slide[w++] = (uch) t->v.n;
1904 Tracevv((stderr, "%c", slide[w - 1]));
1909 } else { /* it's an EOB or a length */
1911 /* exit if end of block */
1915 /* get length of block to copy */
1917 n = t->v.n + ((unsigned) b & mask_bits[e]);
1920 /* decode distance of block to copy */
1921 NEEDBITS((unsigned) bd)
1922 if ((e = (t = td + ((unsigned) b & md))->e) > 16)
1933 t->v.t + ((unsigned) b & mask_bits[e]))->e) >
1937 d = w - t->v.n - ((unsigned) b & mask_bits[e]);
1939 Tracevv((stderr, "\\[%d,%d]", w - d, n));
1945 WSIZE - ((d &= WSIZE - 1) > w ? d : w)) >
1947 #if !defined(NOMEMCPY) && !defined(DEBUG)
1948 if (w - d >= e) { /* (this test assumes unsigned comparison) */
1949 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]));
1967 /* restore the globals from the locals */
1968 wp = w; /* restore global window pointer */
1969 bb = b; /* restore global bit buffer */
1978 int inflate_stored()
1979 /* "decompress" an inflated type 0 (stored) block. */
1981 unsigned n; /* number of bytes in block */
1982 unsigned w; /* current window position */
1983 register ulg b; /* bit buffer */
1984 register unsigned k; /* number of bits in bit buffer */
1987 /* make local copies of globals */
1988 b = bb; /* initialize bit buffer */
1990 w = wp; /* initialize window position */
1993 /* go to byte boundary */
1998 /* get the length and its complement */
2000 n = ((unsigned) b & 0xffff);
2003 if (n != (unsigned) ((~b) & 0xffff))
2004 return 1; /* error in compressed data */
2008 /* read and output the compressed data */
2011 slide[w++] = (uch) b;
2020 /* restore the globals from the locals */
2021 wp = w; /* restore global window pointer */
2022 bb = b; /* restore global bit buffer */
2030 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
2031 either replace this with a custom decoder, or at least precompute the
2034 int i; /* temporary variable */
2035 struct huft *tl; /* literal/length code table */
2036 struct huft *td; /* distance code table */
2037 int bl; /* lookup bits for tl */
2038 int bd; /* lookup bits for td */
2039 unsigned l[288]; /* length list for huft_build */
2042 /* set up literal table */
2043 for (i = 0; i < 144; i++)
2045 for (; i < 256; i++)
2047 for (; i < 280; i++)
2049 for (; i < 288; i++) /* make a complete, but wrong code set */
2052 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
2056 /* set up distance table */
2057 for (i = 0; i < 30; i++) /* make an incomplete code set */
2060 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1) {
2066 /* decompress until an end-of-block code */
2067 if (inflate_codes(tl, td, bl, bd))
2071 /* free the decoding tables, return */
2079 int inflate_dynamic()
2080 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
2082 int i; /* temporary variables */
2084 unsigned l; /* last length */
2085 unsigned m; /* mask for bit lengths table */
2086 unsigned n; /* number of lengths to get */
2087 struct huft *tl; /* literal/length code table */
2088 struct huft *td; /* distance code table */
2089 int bl; /* lookup bits for tl */
2090 int bd; /* lookup bits for td */
2091 unsigned nb; /* number of bit length codes */
2092 unsigned nl; /* number of literal/length codes */
2093 unsigned nd; /* number of distance codes */
2095 #ifdef PKZIP_BUG_WORKAROUND
2096 unsigned ll[288 + 32]; /* literal/length and distance code lengths */
2098 unsigned ll[286 + 30]; /* literal/length and distance code lengths */
2100 register ulg b; /* bit buffer */
2101 register unsigned k; /* number of bits in bit buffer */
2104 /* make local bit buffer */
2109 /* read in table lengths */
2111 nl = 257 + ((unsigned) b & 0x1f); /* number of literal/length codes */
2114 nd = 1 + ((unsigned) b & 0x1f); /* number of distance codes */
2117 nb = 4 + ((unsigned) b & 0xf); /* number of bit length codes */
2119 #ifdef PKZIP_BUG_WORKAROUND
2120 if (nl > 288 || nd > 32)
2122 if (nl > 286 || nd > 30)
2124 return 1; /* bad lengths */
2127 /* read in bit-length-code lengths */
2128 for (j = 0; j < nb; j++) {
2130 ll[border[j]] = (unsigned) b & 7;
2137 /* build decoding table for trees--single level, 7 bit lookup */
2139 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0) {
2142 return i; /* incomplete code set */
2146 /* read in literal and distance code lengths */
2150 while ((unsigned) i < n) {
2151 NEEDBITS((unsigned) bl)
2152 j = (td = tl + ((unsigned) b & m))->b;
2155 if (j < 16) /* length of code in bits (0..15) */
2156 ll[i++] = l = j; /* save last length in l */
2157 else if (j == 16) { /* repeat last length 3 to 6 times */
2159 j = 3 + ((unsigned) b & 3);
2161 if ((unsigned) i + j > n)
2165 } else if (j == 17) { /* 3 to 10 zero length codes */
2167 j = 3 + ((unsigned) b & 7);
2169 if ((unsigned) i + j > n)
2174 } else { /* j == 18: 11 to 138 zero length codes */
2177 j = 11 + ((unsigned) b & 0x7f);
2179 if ((unsigned) i + j > n)
2188 /* free decoding table for trees */
2192 /* restore the global bit buffer */
2197 /* build the decoding tables for literal/length and distance codes */
2199 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0) {
2201 fprintf(stderr, " incomplete literal tree\n");
2204 return i; /* incomplete code set */
2207 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0) {
2209 fprintf(stderr, " incomplete distance tree\n");
2210 #ifdef PKZIP_BUG_WORKAROUND
2217 return i; /* incomplete code set */
2222 /* decompress until an end-of-block code */
2223 if (inflate_codes(tl, td, bl, bd))
2227 /* free the decoding tables, return */
2235 int inflate_block(e)
2236 int *e; /* last block flag */
2238 /* decompress an inflated block */
2240 unsigned t; /* block type */
2241 register ulg b; /* bit buffer */
2242 register unsigned k; /* number of bits in bit buffer */
2245 /* make local bit buffer */
2250 /* read in last block bit */
2256 /* read in block type */
2258 t = (unsigned) b & 3;
2262 /* restore the global bit buffer */
2267 /* inflate that block type */
2269 return inflate_dynamic();
2271 return inflate_stored();
2273 return inflate_fixed();
2276 /* bad block type */
2283 /* decompress an inflated entry */
2285 int e; /* last block flag */
2286 int r; /* result code */
2287 unsigned h; /* maximum struct huft's malloc'ed */
2290 /* initialize window, bit buffer */
2296 /* decompress until the last block */
2300 if ((r = inflate_block(&e)) != 0)
2306 /* Undo too much lookahead. The next read will be byte aligned so we
2307 * can discard unused bits in the last meaningful byte.
2314 /* flush out slide */
2318 /* return success */
2320 fprintf(stderr, "<%u> ", h);