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
33 /* These defines are very important for BusyBox. Without these,
34 * huge chunks of ram are pre-allocated making the BusyBox bss
35 * size Freaking Huge(tm), which is a bad thing.*/
39 #define BB_DECLARE_EXTERN
40 #define bb_need_memory_exhausted
41 #define bb_need_name_too_long
45 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
46 * Copyright (C) 1992-1993 Jean-loup Gailly
47 * The unzip code was written and put in the public domain by Mark Adler.
48 * Portions of the lzw code are derived from the public domain 'compress'
49 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
50 * Ken Turkowski, Dave Mack and Peter Jannesen.
52 * See the license_msg below and the file COPYING for the software license.
53 * See the file algorithm.doc for the compression algorithms and file formats.
57 static char *license_msg[] = {
58 " Copyright (C) 1992-1993 Jean-loup Gailly",
59 " This program is free software; you can redistribute it and/or modify",
60 " it under the terms of the GNU General Public License as published by",
61 " the Free Software Foundation; either version 2, or (at your option)",
62 " any later version.",
64 " This program is distributed in the hope that it will be useful,",
65 " but WITHOUT ANY WARRANTY; without even the implied warranty of",
66 " MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the",
67 " GNU General Public License for more details.",
69 " You should have received a copy of the GNU General Public License",
70 " along with this program; if not, write to the Free Software",
71 " Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.",
76 /* Compress files with zip algorithm and 'compress' interface.
77 * See usage() and help() functions below for all options.
79 * file.gz: compressed file with same mode, owner, and utimes
80 * or stdout with -c option or if stdin used as input.
81 * If the output file name had to be truncated, the original name is kept
82 * in the compressed file.
83 * On MSDOS, file.tmp -> file.tmz. On VMS, file.tmp -> file.tmp-gz.
85 * Using gz on MSDOS would create too many file name conflicts. For
86 * example, foo.txt -> foo.tgz (.tgz must be reserved as shorthand for
87 * tar.gz). Similarly, foo.dir and foo.doc would both be mapped to foo.dgz.
88 * I also considered 12345678.txt -> 12345txt.gz but this truncates the name
89 * too heavily. There is no ideal solution given the MSDOS 8+3 limitation.
91 * For the meaning of all compilation flags, see comments in Makefile.in.
95 #include <sys/types.h>
99 /* #include "tailor.h" */
101 /* tailor.h -- target dependent definitions
102 * Copyright (C) 1992-1993 Jean-loup Gailly.
103 * This is free software; you can redistribute it and/or modify it under the
104 * terms of the GNU General Public License, see the file COPYING.
107 /* The target dependent definitions should be defined here only.
108 * The target dependent functions should be defined in tailor.c.
113 #define get_char() get_byte()
114 #define put_char(c) put_byte(c)
117 /* I don't like nested includes, but the string and io functions are used
121 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
123 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
126 # define memzero(s, n) memset ((void *)(s), 0, (n))
128 # include <strings.h>
129 # define strchr index
130 # define strrchr rindex
131 # define memcpy(d, s, n) bcopy((s), (d), (n))
132 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
133 # define memzero(s, n) bzero((s), (n))
137 # define RETSIGTYPE void
142 typedef unsigned char uch;
143 typedef unsigned short ush;
144 typedef unsigned long ulg;
146 /* Return codes from gzip */
151 /* Compression methods (see algorithm.doc) */
154 extern int method; /* compression method */
156 /* To save memory for 16 bit systems, some arrays are overlaid between
157 * the various modules:
158 * deflate: prev+head window d_buf l_buf outbuf
159 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
160 * inflate: window inbuf
161 * unpack: window inbuf prefix_len
162 * unlzh: left+right window c_table inbuf c_len
163 * For compression, input is done in window[]. For decompression, output
164 * is done in window except for unlzw.
169 # define INBUFSIZ 0x2000 /* input buffer size */
171 # define INBUFSIZ 0x8000 /* input buffer size */
174 #define INBUF_EXTRA 64 /* required by unlzw() */
178 # define OUTBUFSIZ 8192 /* output buffer size */
180 # define OUTBUFSIZ 16384 /* output buffer size */
183 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
189 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
191 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
195 /*#define DYN_ALLOC*/
198 # define EXTERN(type, array) extern type * array
199 # define DECLARE(type, array, size) type * array
200 # define ALLOC(type, array, size) { \
201 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
202 if (array == NULL) errorMsg(memory_exhausted); \
204 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
206 # define EXTERN(type, array) extern type array[]
207 # define DECLARE(type, array, size) type array[size]
208 # define ALLOC(type, array, size)
212 EXTERN(uch, inbuf); /* input buffer */
213 EXTERN(uch, outbuf); /* output buffer */
214 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
215 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
216 #define tab_suffix window
218 # define tab_prefix prev /* hash link (see deflate.c) */
219 # define head (prev+WSIZE) /* hash head (see deflate.c) */
220 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
222 # define tab_prefix0 prev
223 # define head tab_prefix1
224 EXTERN(ush, tab_prefix0); /* prefix for even codes */
225 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
228 extern unsigned insize; /* valid bytes in inbuf */
229 extern unsigned inptr; /* index of next byte to be processed in inbuf */
230 extern unsigned outcnt; /* bytes in output buffer */
232 extern long bytes_in; /* number of input bytes */
233 extern long bytes_out; /* number of output bytes */
234 extern long header_bytes; /* number of bytes in gzip header */
236 extern long ifile_size; /* input file size, -1 for devices (debug only) */
238 typedef int file_t; /* Do not use stdio */
240 #define NO_FILE (-1) /* in memory compression */
243 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
246 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
247 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
248 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
249 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
250 #define COMMENT 0x10 /* bit 4 set: file comment present */
251 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
252 #define RESERVED 0xC0 /* bit 6,7: reserved */
255 # define WSIZE 0x8000 /* window size--must be a power of two, and */
256 #endif /* at least 32K for zip's deflate method */
259 #define MAX_MATCH 258
260 /* The minimum and maximum match lengths */
262 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
263 /* Minimum amount of lookahead, except at the end of the input file.
264 * See deflate.c for comments about the MIN_MATCH+1.
267 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
268 /* In order to simplify the code, particularly on 16 bit machines, match
269 * distances are limited to MAX_DIST instead of WSIZE.
272 extern int exit_code; /* program exit code */
273 extern int verbose; /* be verbose (-v) */
274 extern int level; /* compression level */
275 extern int test; /* check .z file integrity */
276 extern int save_orig_name; /* set if original name must be saved */
278 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
279 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
281 /* put_byte is used for the compressed output, put_ubyte for the
282 * uncompressed output. However unlzw() uses window for its
283 * suffix table instead of its output buffer, so it does not use put_ubyte
284 * (to be cleaned up).
286 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
288 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
291 /* Output a 16 bit value, lsb first */
292 #define put_short(w) \
293 { if (outcnt < OUTBUFSIZ-2) { \
294 outbuf[outcnt++] = (uch) ((w) & 0xff); \
295 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
297 put_byte((uch)((w) & 0xff)); \
298 put_byte((uch)((ush)(w) >> 8)); \
302 /* Output a 32 bit value to the bit stream, lsb first */
303 #define put_long(n) { \
304 put_short((n) & 0xffff); \
305 put_short(((ulg)(n)) >> 16); \
308 #define seekable() 0 /* force sequential output */
309 #define translate_eol 0 /* no option -a yet */
311 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
313 /* Macros for getting two-byte and four-byte header values */
314 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
315 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
317 /* Diagnostic functions */
319 # define Assert(cond,msg) {if(!(cond)) errorMsg(msg);}
320 # define Trace(x) fprintf x
321 # define Tracev(x) {if (verbose) fprintf x ;}
322 # define Tracevv(x) {if (verbose>1) fprintf x ;}
323 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
324 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
326 # define Assert(cond,msg)
331 # define Tracecv(c,x)
334 #define WARN(msg) {fprintf msg ; \
335 if (exit_code == OK) exit_code = WARNING;}
338 extern int unzip (int in, int out);
341 RETSIGTYPE abort_gzip (void);
344 void lm_init (int pack_level, ush * flags);
348 void ct_init (ush * attr, int *method);
349 int ct_tally (int dist, int lc);
350 ulg flush_block (char *buf, ulg stored_len, int eof);
353 void bi_init (file_t zipfile);
354 void send_bits (int value, int length);
355 unsigned bi_reverse (unsigned value, int length);
356 void bi_windup (void);
357 void copy_block (char *buf, unsigned len, int header);
360 extern ulg updcrc (uch * s, unsigned n);
361 extern void clear_bufs (void);
362 static int fill_inbuf (int eof_ok);
363 extern void flush_outbuf (void);
364 static void flush_window (void);
365 extern void write_buf (int fd, void * buf, unsigned cnt);
368 static char *basename (char *fname);
369 #endif /* not __linux__ */
370 void read_error_msg (void);
371 void write_error_msg (void);
374 static int inflate (void);
376 /* #include "lzw.h" */
378 /* lzw.h -- define the lzw functions.
379 * Copyright (C) 1992-1993 Jean-loup Gailly.
380 * This is free software; you can redistribute it and/or modify it under the
381 * terms of the GNU General Public License, see the file COPYING.
384 #if !defined(OF) && defined(lint)
391 #define INIT_BITS 9 /* Initial number of bits per code */
393 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
395 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
396 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
397 * It's a pity that old uncompress does not check bit 0x20. That makes
398 * extension of the format actually undesirable because old compress
399 * would just crash on the new format instead of giving a meaningful
400 * error message. It does check the number of bits, but it's more
401 * helpful to say "unsupported format, get a new version" than
402 * "can only handle 16 bits".
405 #define BLOCK_MODE 0x80
406 /* Block compression: if table is full and compression rate is dropping,
407 * clear the dictionary.
410 #define LZW_RESERVED 0x60 /* reserved bits */
412 #define CLEAR 256 /* flush the dictionary */
413 #define FIRST (CLEAR+1) /* first free entry */
415 extern int maxbits; /* max bits per code for LZW */
416 extern int block_mode; /* block compress mode -C compatible with 2.0 */
418 extern int lzw (int in, int out);
419 extern int unlzw (int in, int out);
422 /* #include "revision.h" */
424 /* revision.h -- define the version number
425 * Copyright (C) 1992-1993 Jean-loup Gailly.
426 * This is free software; you can redistribute it and/or modify it under the
427 * terms of the GNU General Public License, see the file COPYING.
430 #define VERSION "1.2.4"
432 #define REVDATE "18 Aug 93"
434 /* This version does not support compression into old compress format: */
445 typedef struct dirent dir_type;
447 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
448 # define DIR_OPT "DIRENT"
450 # define NLENGTH(dirent) ((dirent)->d_namlen)
452 # include <sys/dir.h>
453 typedef struct direct dir_type;
455 # define DIR_OPT "SYSDIR"
458 # include <sys/ndir.h>
459 typedef struct direct dir_type;
461 # define DIR_OPT "SYSNDIR"
465 typedef struct direct dir_type;
467 # define DIR_OPT "NDIR"
470 # define DIR_OPT "NO_DIR"
475 #if !defined(S_ISDIR) && defined(S_IFDIR)
476 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
478 #if !defined(S_ISREG) && defined(S_IFREG)
479 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
481 typedef RETSIGTYPE(*sig_type) (int);
484 # define O_BINARY 0 /* creation mode for open() */
488 /* Pure BSD system? */
489 # include <sys/file.h>
491 # define O_CREAT FCREAT
494 # define O_EXCL FEXCL
499 # define S_IRUSR 0400
502 # define S_IWUSR 0200
504 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
506 #ifndef MAX_PATH_LEN /* max pathname length */
508 # define MAX_PATH_LEN BUFSIZ
510 # define MAX_PATH_LEN 1024
520 off_t lseek (int fd, off_t offset, int whence);
526 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
527 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
528 DECLARE(ush, d_buf, DIST_BUFSIZE);
529 DECLARE(uch, window, 2L * WSIZE);
531 DECLARE(ush, tab_prefix, 1L << BITS);
533 DECLARE(ush, tab_prefix0, 1L << (BITS - 1));
534 DECLARE(ush, tab_prefix1, 1L << (BITS - 1));
537 /* local variables */
539 int test_mode = 0; /* check file integrity option */
540 int foreground; /* set if program run in foreground */
541 int maxbits = BITS; /* max bits per code for LZW */
542 int method = DEFLATED; /* compression method */
543 int exit_code = OK; /* program exit code */
544 int last_member; /* set for .zip and .Z files */
545 int part_nb; /* number of parts in .gz file */
546 long ifile_size; /* input file size, -1 for devices (debug only) */
548 long bytes_in; /* number of input bytes */
549 long bytes_out; /* number of output bytes */
550 long total_in = 0; /* input bytes for all files */
551 long total_out = 0; /* output bytes for all files */
552 struct stat istat; /* status for input file */
553 int ifd; /* input file descriptor */
554 int ofd; /* output file descriptor */
555 unsigned insize; /* valid bytes in inbuf */
556 unsigned inptr; /* index of next byte to be processed in inbuf */
557 unsigned outcnt; /* bytes in output buffer */
559 long header_bytes; /* number of bytes in gzip header */
561 /* local functions */
563 local int get_method (int in);
565 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
567 /* ======================================================================== */
568 int gunzip_main(int argc, char **argv)
570 int file_count; /* number of files to precess */
576 int delInputFile = 0;
580 char ifname[MAX_PATH_LEN + 1]; /* input file name */
581 char ofname[MAX_PATH_LEN + 1]; /* output file name */
583 if (strcmp(applet_name, "zcat") == 0) {
587 /* Parse any options */
588 while (--argc > 0 && **(++argv) == '-') {
589 if (*((*argv) + 1) == '\0') {
592 while (*(++(*argv))) {
614 if (isatty(fileno(stdin)) && fromstdin==1 && force==0)
615 fatalError( "data not read from terminal. Use -f to force it.\n");
616 if (isatty(fileno(stdout)) && tostdout==1 && force==0)
617 fatalError( "data not written to terminal. Use -f to force it.\n");
620 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
622 (void) signal(SIGINT, (sig_type) abort_gzip);
625 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
626 (void) signal(SIGTERM, (sig_type) abort_gzip);
630 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
631 (void) signal(SIGHUP, (sig_type) abort_gzip);
635 file_count = argc - optind;
637 /* Allocate all global buffers (for DYN_ALLOC option) */
638 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
639 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
640 ALLOC(ush, d_buf, DIST_BUFSIZE);
641 ALLOC(uch, window, 2L * WSIZE);
643 ALLOC(ush, tab_prefix, 1L << BITS);
645 ALLOC(ush, tab_prefix0, 1L << (BITS - 1));
646 ALLOC(ush, tab_prefix1, 1L << (BITS - 1));
649 if (fromstdin == 1) {
650 strcpy(ofname, "stdin");
652 inFileNum = fileno(stdin);
653 ifile_size = -1L; /* convention for unknown size */
655 /* Open up the input file */
658 if (strlen(*argv) > MAX_PATH_LEN) {
659 errorMsg(name_too_long);
662 strcpy(ifname, *argv);
664 /* Open input fille */
665 inFileNum = open(ifname, O_RDONLY);
670 /* Get the time stamp on the input file. */
671 result = stat(ifname, &statBuf);
676 ifile_size = statBuf.st_size;
680 /* And get to work */
681 strcpy(ofname, "stdout");
682 outFileNum = fileno(stdout);
684 clear_bufs(); /* clear input and output buffers */
687 /* Actually do the compression/decompression. */
688 unzip(inFileNum, outFileNum);
690 } else if (test_mode) {
691 /* Actually do the compression/decompression. */
696 /* And get to work */
697 if (strlen(ifname) > MAX_PATH_LEN - 4) {
698 errorMsg(name_too_long);
701 strcpy(ofname, ifname);
702 pos = strstr(ofname, ".gz");
707 pos = strstr(ofname, ".tgz");
715 /* Open output fille */
716 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
717 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
719 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL);
721 if (outFileNum < 0) {
725 /* Set permissions on the file */
726 fchmod(outFileNum, statBuf.st_mode);
728 clear_bufs(); /* clear input and output buffers */
731 /* Actually do the compression/decompression. */
732 result = unzip(inFileNum, outFileNum);
736 /* Delete the original file */
738 delFileName = ifname;
740 delFileName = ofname;
742 if (delInputFile == 1 && unlink(delFileName) < 0) {
751 /* ========================================================================
752 * Check the magic number of the input file and update ofname if an
753 * original name was given and tostdout is not set.
754 * Return the compression method, -1 for error, -2 for warning.
755 * Set inptr to the offset of the next byte to be processed.
756 * Updates time_stamp if there is one and --no-time is not used.
757 * This function may be called repeatedly for an input file consisting
758 * of several contiguous gzip'ed members.
759 * IN assertions: there is at least one remaining compressed member.
760 * If the member is a zip file, it must be the only one.
762 local int get_method(in)
763 int in; /* input file descriptor */
765 uch flags; /* compression flags */
766 char magic[2]; /* magic header */
768 magic[0] = (char) get_byte();
769 magic[1] = (char) get_byte();
770 method = -1; /* unknown yet */
771 part_nb++; /* number of parts in gzip file */
773 last_member = RECORD_IO;
774 /* assume multiple members in gzip file except for record oriented I/O */
776 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
778 method = (int) get_byte();
779 if (method != DEFLATED) {
780 errorMsg("unknown method %d -- get newer version of gzip\n",
785 flags = (uch) get_byte();
787 (ulg) get_byte(); /* Ignore time stamp */
792 (void) get_byte(); /* Ignore extra flags for the moment */
793 (void) get_byte(); /* Ignore OS type for the moment */
795 if ((flags & EXTRA_FIELD) != 0) {
796 unsigned len = (unsigned) get_byte();
798 len |= ((unsigned) get_byte()) << 8;
804 /* Discard original name if any */
805 if ((flags & ORIG_NAME) != 0) {
806 while (get_char() != 0) /* null */
810 /* Discard file comment if any */
811 if ((flags & COMMENT) != 0) {
812 while (get_char() != 0) /* null */
816 header_bytes = inptr + 2 * sizeof(long); /* include crc and size */
825 fprintf(stderr, "\nnot in gzip format\n");
829 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
834 /* ========================================================================
835 * Signal and error handler.
837 RETSIGTYPE abort_gzip()
842 /* unzip.c -- decompress files in gzip or pkzip format.
843 * Copyright (C) 1992-1993 Jean-loup Gailly
844 * This is free software; you can redistribute it and/or modify it under the
845 * terms of the GNU General Public License, see the file COPYING.
847 * The code in this file is derived from the file funzip.c written
848 * and put in the public domain by Mark Adler.
852 This version can extract files in gzip or pkzip format.
853 For the latter, only the first entry is extracted, and it has to be
854 either deflated or stored.
857 /* #include "crypt.h" */
859 /* crypt.h (dummy version) -- do not perform encryption
860 * Hardly worth copyrighting :-)
864 # undef CRYPT /* dummy version */
867 #define RAND_HEAD_LEN 12 /* length of encryption random header */
872 /* PKZIP header definitions */
873 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
874 #define LOCFLG 6 /* offset of bit flag */
875 #define CRPFLG 1 /* bit for encrypted entry */
876 #define EXTFLG 8 /* bit for extended local header */
877 #define LOCHOW 8 /* offset of compression method */
878 #define LOCTIM 10 /* file mod time (for decryption) */
879 #define LOCCRC 14 /* offset of crc */
880 #define LOCSIZ 18 /* offset of compressed size */
881 #define LOCLEN 22 /* offset of uncompressed length */
882 #define LOCFIL 26 /* offset of file name field length */
883 #define LOCEXT 28 /* offset of extra field length */
884 #define LOCHDR 30 /* size of local header, including sig */
885 #define EXTHDR 16 /* size of extended local header, inc sig */
890 char *key; /* not used--needed to link crypt.c */
891 int pkzip = 0; /* set for a pkzip file */
892 int ext_header = 0; /* set if extended local header */
894 /* ===========================================================================
895 * Unzip in to out. This routine works on both gzip and pkzip files.
897 * IN assertions: the buffer inbuf contains already the beginning of
898 * the compressed data, from offsets inptr to insize-1 included.
899 * The magic header has already been checked. The output buffer is cleared.
902 int in, out; /* input and output file descriptors */
904 ulg orig_crc = 0; /* original crc */
905 ulg orig_len = 0; /* original uncompressed length */
907 uch buf[EXTHDR]; /* extended local header */
911 method = get_method(ifd);
913 exit(exit_code); /* error message already emitted */
916 updcrc(NULL, 0); /* initialize crc */
918 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
919 orig_crc = LG(inbuf + LOCCRC);
920 orig_len = LG(inbuf + LOCLEN);
924 if (method == DEFLATED) {
929 errorMsg(memory_exhausted);
930 } else if (res != 0) {
931 errorMsg("invalid compressed data--format violated");
935 errorMsg("internal error, invalid method");
938 /* Get the crc and original length */
940 /* crc32 (see algorithm.doc)
941 * uncompressed input size modulo 2^32
943 for (n = 0; n < 8; n++) {
944 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
947 orig_len = LG(buf + 4);
949 } else if (ext_header) { /* If extended header, check it */
950 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
952 * compressed size 4-bytes
953 * uncompressed size 4-bytes
955 for (n = 0; n < EXTHDR; n++) {
956 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
958 orig_crc = LG(buf + 4);
959 orig_len = LG(buf + 12);
962 /* Validate decompression */
963 if (orig_crc != updcrc(outbuf, 0)) {
964 errorMsg("invalid compressed data--crc error");
966 if (orig_len != (ulg) bytes_out) {
967 errorMsg("invalid compressed data--length error");
970 /* Check if there are more entries in a pkzip file */
971 if (pkzip && inptr + 4 < insize && LG(inbuf + inptr) == LOCSIG) {
972 WARN((stderr, "has more than one entry--rest ignored\n"));
974 ext_header = pkzip = 0; /* for next file */
978 /* util.c -- utility functions for gzip support
979 * Copyright (C) 1992-1993 Jean-loup Gailly
980 * This is free software; you can redistribute it and/or modify it under the
981 * terms of the GNU General Public License, see the file COPYING.
986 #include <sys/types.h>
995 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1001 static const ulg crc_32_tab[]; /* crc table, defined below */
1003 /* ===========================================================================
1004 * Run a set of bytes through the crc shift register. If s is a NULL
1005 * pointer, then initialize the crc shift register contents instead.
1006 * Return the current crc in either case.
1009 uch *s; /* pointer to bytes to pump through */
1010 unsigned n; /* number of bytes in s[] */
1012 register ulg c; /* temporary variable */
1014 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
1022 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
1026 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1029 /* ===========================================================================
1030 * Clear input and output buffers
1032 void clear_bufs(void)
1036 bytes_in = bytes_out = 0L;
1039 /* ===========================================================================
1040 * Fill the input buffer. This is called only when the buffer is empty.
1042 int fill_inbuf(eof_ok)
1043 int eof_ok; /* set if EOF acceptable as a result */
1047 /* Read as much as possible */
1051 len = read(ifd, (char *) inbuf + insize, INBUFSIZ - insize);
1052 if (len == 0 || len == EOF)
1055 } while (insize < INBUFSIZ);
1062 bytes_in += (ulg) insize;
1067 /* ===========================================================================
1068 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1069 * (used for the compressed data only)
1077 write_buf(ofd, (char *) outbuf, outcnt);
1078 bytes_out += (ulg) outcnt;
1082 /* ===========================================================================
1083 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1084 * (Used for the decompressed data only.)
1090 updcrc(window, outcnt);
1093 write_buf(ofd, (char *) window, outcnt);
1094 bytes_out += (ulg) outcnt;
1098 /* ===========================================================================
1099 * Does the same as write(), but also handles partial pipe writes and checks
1102 void write_buf(fd, buf, cnt)
1109 while ((n = write(fd, buf, cnt)) != cnt) {
1110 if (n == (unsigned) (-1)) {
1114 buf = (void *) ((char *) buf + n);
1118 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1120 /* Provide missing strspn and strcspn functions. */
1126 int strspn (const char *s, const char *accept);
1127 int strcspn (const char *s, const char *reject);
1129 /* ========================================================================
1130 * Return the length of the maximum initial segment
1131 * of s which contains only characters in accept.
1133 int strspn(s, accept)
1137 register const char *p;
1138 register const char *a;
1139 register int count = 0;
1141 for (p = s; *p != '\0'; ++p) {
1142 for (a = accept; *a != '\0'; ++a) {
1153 /* ========================================================================
1154 * Return the length of the maximum inital segment of s
1155 * which contains no characters from reject.
1157 int strcspn(s, reject)
1161 register int count = 0;
1163 while (*s != '\0') {
1164 if (strchr(reject, *s++) != NULL)
1171 #endif /* NO_STRING_H */
1174 /* ========================================================================
1177 void read_error_msg()
1179 fprintf(stderr, "\n");
1183 fprintf(stderr, "unexpected end of file\n");
1188 void write_error_msg()
1190 fprintf(stderr, "\n");
1196 /* ========================================================================
1197 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1199 static const ulg crc_32_tab[] = {
1200 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1201 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1202 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1203 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1204 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1205 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1206 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1207 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1208 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1209 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1210 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1211 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1212 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1213 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1214 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1215 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1216 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1217 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1218 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1219 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1220 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1221 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1222 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1223 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1224 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1225 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1226 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1227 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1228 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1229 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1230 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1231 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1232 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1233 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1234 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1235 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1236 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1237 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1238 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1239 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1240 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1241 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1242 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1243 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1244 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1245 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1246 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1247 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1248 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1249 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1250 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1254 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1255 version c10p1, 10 January 1993 */
1257 /* You can do whatever you like with this source file, though I would
1258 prefer that if you modify it and redistribute it that you include
1259 comments to that effect with your name and the date. Thank you.
1260 [The history has been moved to the file ChangeLog.]
1264 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1265 method searches for as much of the current string of bytes (up to a
1266 length of 258) in the previous 32K bytes. If it doesn't find any
1267 matches (of at least length 3), it codes the next byte. Otherwise, it
1268 codes the length of the matched string and its distance backwards from
1269 the current position. There is a single Huffman code that codes both
1270 single bytes (called "literals") and match lengths. A second Huffman
1271 code codes the distance information, which follows a length code. Each
1272 length or distance code actually represents a base value and a number
1273 of "extra" (sometimes zero) bits to get to add to the base value. At
1274 the end of each deflated block is a special end-of-block (EOB) literal/
1275 length code. The decoding process is basically: get a literal/length
1276 code; if EOB then done; if a literal, emit the decoded byte; if a
1277 length then get the distance and emit the referred-to bytes from the
1278 sliding window of previously emitted data.
1280 There are (currently) three kinds of inflate blocks: stored, fixed, and
1281 dynamic. The compressor deals with some chunk of data at a time, and
1282 decides which method to use on a chunk-by-chunk basis. A chunk might
1283 typically be 32K or 64K. If the chunk is uncompressible, then the
1284 "stored" method is used. In this case, the bytes are simply stored as
1285 is, eight bits per byte, with none of the above coding. The bytes are
1286 preceded by a count, since there is no longer an EOB code.
1288 If the data is compressible, then either the fixed or dynamic methods
1289 are used. In the dynamic method, the compressed data is preceded by
1290 an encoding of the literal/length and distance Huffman codes that are
1291 to be used to decode this block. The representation is itself Huffman
1292 coded, and so is preceded by a description of that code. These code
1293 descriptions take up a little space, and so for small blocks, there is
1294 a predefined set of codes, called the fixed codes. The fixed method is
1295 used if the block codes up smaller that way (usually for quite small
1296 chunks), otherwise the dynamic method is used. In the latter case, the
1297 codes are customized to the probabilities in the current block, and so
1298 can code it much better than the pre-determined fixed codes.
1300 The Huffman codes themselves are decoded using a mutli-level table
1301 lookup, in order to maximize the speed of decoding plus the speed of
1302 building the decoding tables. See the comments below that precede the
1303 lbits and dbits tuning parameters.
1308 Notes beyond the 1.93a appnote.txt:
1310 1. Distance pointers never point before the beginning of the output
1312 2. Distance pointers can point back across blocks, up to 32k away.
1313 3. There is an implied maximum of 7 bits for the bit length table and
1314 15 bits for the actual data.
1315 4. If only one code exists, then it is encoded using one bit. (Zero
1316 would be more efficient, but perhaps a little confusing.) If two
1317 codes exist, they are coded using one bit each (0 and 1).
1318 5. There is no way of sending zero distance codes--a dummy must be
1319 sent if there are none. (History: a pre 2.0 version of PKZIP would
1320 store blocks with no distance codes, but this was discovered to be
1321 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1322 zero distance codes, which is sent as one code of zero bits in
1324 6. There are up to 286 literal/length codes. Code 256 represents the
1325 end-of-block. Note however that the static length tree defines
1326 288 codes just to fill out the Huffman codes. Codes 286 and 287
1327 cannot be used though, since there is no length base or extra bits
1328 defined for them. Similarly, there are up to 30 distance codes.
1329 However, static trees define 32 codes (all 5 bits) to fill out the
1330 Huffman codes, but the last two had better not show up in the data.
1331 7. Unzip can check dynamic Huffman blocks for complete code sets.
1332 The exception is that a single code would not be complete (see #4).
1333 8. The five bits following the block type is really the number of
1334 literal codes sent minus 257.
1335 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1336 (1+6+6). Therefore, to output three times the length, you output
1337 three codes (1+1+1), whereas to output four times the same length,
1338 you only need two codes (1+3). Hmm.
1339 10. In the tree reconstruction algorithm, Code = Code + Increment
1340 only if BitLength(i) is not zero. (Pretty obvious.)
1341 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1342 12. Note: length code 284 can represent 227-258, but length code 285
1343 really is 258. The last length deserves its own, short code
1344 since it gets used a lot in very redundant files. The length
1345 258 is special since 258 - 3 (the min match length) is 255.
1346 13. The literal/length and distance code bit lengths are read as a
1347 single stream of lengths. It is possible (and advantageous) for
1348 a repeat code (16, 17, or 18) to go across the boundary between
1349 the two sets of lengths.
1352 #include <sys/types.h>
1354 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1355 # include <stdlib.h>
1359 #define slide window
1361 /* Huffman code lookup table entry--this entry is four bytes for machines
1362 that have 16-bit pointers (e.g. PC's in the small or medium model).
1363 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1364 means that v is a literal, 16 < e < 32 means that v is a pointer to
1365 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1366 an unused code. If a code with e == 99 is looked up, this implies an
1367 error in the data. */
1369 uch e; /* number of extra bits or operation */
1370 uch b; /* number of bits in this code or subcode */
1372 ush n; /* literal, length base, or distance base */
1373 struct huft *t; /* pointer to next level of table */
1378 /* Function prototypes */
1379 int huft_build (unsigned *, unsigned, unsigned, ush *, ush *,
1380 struct huft **, int *);
1381 int huft_free (struct huft *);
1382 int inflate_codes (struct huft *, struct huft *, int, int);
1383 int inflate_stored (void);
1384 int inflate_fixed (void);
1385 int inflate_dynamic (void);
1386 int inflate_block (int *);
1390 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1391 stream to find repeated byte strings. This is implemented here as a
1392 circular buffer. The index is updated simply by incrementing and then
1393 and'ing with 0x7fff (32K-1). */
1394 /* It is left to other modules to supply the 32K area. It is assumed
1395 to be usable as if it were declared "uch slide[32768];" or as just
1396 "uch *slide;" and then malloc'ed in the latter case. The definition
1397 must be in unzip.h, included above. */
1398 /* unsigned wp; current position in slide */
1400 #define flush_output(w) (wp=(w),flush_window())
1402 /* Tables for deflate from PKZIP's appnote.txt. */
1403 static unsigned border[] = { /* Order of the bit length code lengths */
1404 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15
1406 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1407 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1408 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0
1411 /* note: see note #13 above about the 258 in this list. */
1412 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1413 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1414 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99
1415 }; /* 99==invalid */
1416 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1417 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1418 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1419 8193, 12289, 16385, 24577
1421 static ush cpdext[] = { /* Extra bits for distance codes */
1422 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1423 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1429 /* Macros for inflate() bit peeking and grabbing.
1433 x = b & mask_bits[j];
1436 where NEEDBITS makes sure that b has at least j bits in it, and
1437 DUMPBITS removes the bits from b. The macros use the variable k
1438 for the number of bits in b. Normally, b and k are register
1439 variables for speed, and are initialized at the beginning of a
1440 routine that uses these macros from a global bit buffer and count.
1442 If we assume that EOB will be the longest code, then we will never
1443 ask for bits with NEEDBITS that are beyond the end of the stream.
1444 So, NEEDBITS should not read any more bytes than are needed to
1445 meet the request. Then no bytes need to be "returned" to the buffer
1446 at the end of the last block.
1448 However, this assumption is not true for fixed blocks--the EOB code
1449 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1450 (The EOB code is shorter than other codes because fixed blocks are
1451 generally short. So, while a block always has an EOB, many other
1452 literal/length codes have a significantly lower probability of
1453 showing up at all.) However, by making the first table have a
1454 lookup of seven bits, the EOB code will be found in that first
1455 lookup, and so will not require that too many bits be pulled from
1459 ulg bb; /* bit buffer */
1460 unsigned bk; /* bits in bit buffer */
1464 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1465 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1471 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1473 # define NEXTBYTE() (uch)get_byte()
1475 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1476 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1480 Huffman code decoding is performed using a multi-level table lookup.
1481 The fastest way to decode is to simply build a lookup table whose
1482 size is determined by the longest code. However, the time it takes
1483 to build this table can also be a factor if the data being decoded
1484 is not very long. The most common codes are necessarily the
1485 shortest codes, so those codes dominate the decoding time, and hence
1486 the speed. The idea is you can have a shorter table that decodes the
1487 shorter, more probable codes, and then point to subsidiary tables for
1488 the longer codes. The time it costs to decode the longer codes is
1489 then traded against the time it takes to make longer tables.
1491 This results of this trade are in the variables lbits and dbits
1492 below. lbits is the number of bits the first level table for literal/
1493 length codes can decode in one step, and dbits is the same thing for
1494 the distance codes. Subsequent tables are also less than or equal to
1495 those sizes. These values may be adjusted either when all of the
1496 codes are shorter than that, in which case the longest code length in
1497 bits is used, or when the shortest code is *longer* than the requested
1498 table size, in which case the length of the shortest code in bits is
1501 There are two different values for the two tables, since they code a
1502 different number of possibilities each. The literal/length table
1503 codes 286 possible values, or in a flat code, a little over eight
1504 bits. The distance table codes 30 possible values, or a little less
1505 than five bits, flat. The optimum values for speed end up being
1506 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1507 The optimum values may differ though from machine to machine, and
1508 possibly even between compilers. Your mileage may vary.
1512 int lbits = 9; /* bits in base literal/length lookup table */
1513 int dbits = 6; /* bits in base distance lookup table */
1516 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1517 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1518 #define N_MAX 288 /* maximum number of codes in any set */
1521 unsigned hufts; /* track memory usage */
1524 int huft_build(b, n, s, d, e, t, m)
1525 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1526 unsigned n; /* number of codes (assumed <= N_MAX) */
1527 unsigned s; /* number of simple-valued codes (0..s-1) */
1528 ush *d; /* list of base values for non-simple codes */
1529 ush *e; /* list of extra bits for non-simple codes */
1530 struct huft **t; /* result: starting table */
1531 int *m; /* maximum lookup bits, returns actual */
1533 /* Given a list of code lengths and a maximum table size, make a set of
1534 tables to decode that set of codes. Return zero on success, one if
1535 the given code set is incomplete (the tables are still built in this
1536 case), two if the input is invalid (all zero length codes or an
1537 oversubscribed set of lengths), and three if not enough memory. */
1539 unsigned a; /* counter for codes of length k */
1540 unsigned c[BMAX + 1]; /* bit length count table */
1541 unsigned f; /* i repeats in table every f entries */
1542 int g; /* maximum code length */
1543 int h; /* table level */
1544 register unsigned i; /* counter, current code */
1545 register unsigned j; /* counter */
1546 register int k; /* number of bits in current code */
1547 int l; /* bits per table (returned in m) */
1548 register unsigned *p; /* pointer into c[], b[], or v[] */
1549 register struct huft *q; /* points to current table */
1550 struct huft r; /* table entry for structure assignment */
1551 struct huft *u[BMAX]; /* table stack */
1552 unsigned v[N_MAX]; /* values in order of bit length */
1553 register int w; /* bits before this table == (l * h) */
1554 unsigned x[BMAX + 1]; /* bit offsets, then code stack */
1555 unsigned *xp; /* pointer into x */
1556 int y; /* number of dummy codes added */
1557 unsigned z; /* number of entries in current table */
1560 /* Generate counts for each bit length */
1561 memzero(c, sizeof(c));
1568 && n - i <= '~' ? "%c %d\n" : "0x%x %d\n"), n - i, *p));
1569 c[*p]++; /* assume all entries <= BMAX */
1570 p++; /* Can't combine with above line (Solaris bug) */
1572 if (c[0] == n) { /* null input--all zero length codes */
1573 *t = (struct huft *) NULL;
1579 /* Find minimum and maximum length, bound *m by those */
1581 for (j = 1; j <= BMAX; j++)
1584 k = j; /* minimum code length */
1585 if ((unsigned) l < j)
1587 for (i = BMAX; i; i--)
1590 g = i; /* maximum code length */
1591 if ((unsigned) l > i)
1596 /* Adjust last length count to fill out codes, if needed */
1597 for (y = 1 << j; j < i; j++, y <<= 1)
1598 if ((y -= c[j]) < 0)
1599 return 2; /* bad input: more codes than bits */
1600 if ((y -= c[i]) < 0)
1605 /* Generate starting offsets into the value table for each length */
1609 while (--i) { /* note that i == g from above */
1610 *xp++ = (j += *p++);
1614 /* Make a table of values in order of bit lengths */
1618 if ((j = *p++) != 0)
1623 /* Generate the Huffman codes and for each, make the table entries */
1624 x[0] = i = 0; /* first Huffman code is zero */
1625 p = v; /* grab values in bit order */
1626 h = -1; /* no tables yet--level -1 */
1627 w = -l; /* bits decoded == (l * h) */
1628 u[0] = (struct huft *) NULL; /* just to keep compilers happy */
1629 q = (struct huft *) NULL; /* ditto */
1632 /* go through the bit lengths (k already is bits in shortest code) */
1633 for (; k <= g; k++) {
1636 /* here i is the Huffman code of length k bits for value *p */
1637 /* make tables up to required level */
1640 w += l; /* previous table always l bits */
1642 /* compute minimum size table less than or equal to l bits */
1643 z = (z = g - w) > (unsigned) l ? l : z; /* upper limit on table size */
1644 if ((f = 1 << (j = k - w)) > a + 1) { /* try a k-w bit table *//* too few codes for k-w bit table */
1645 f -= a + 1; /* deduct codes from patterns left */
1647 while (++j < z) { /* try smaller tables up to z bits */
1648 if ((f <<= 1) <= *++xp)
1649 break; /* enough codes to use up j bits */
1650 f -= *xp; /* else deduct codes from patterns */
1653 z = 1 << j; /* table entries for j-bit table */
1655 /* allocate and link in new table */
1658 (struct huft *) malloc((z + 1) *
1659 sizeof(struct huft))) ==
1660 (struct huft *) NULL) {
1663 return 3; /* not enough memory */
1665 hufts += z + 1; /* track memory usage */
1666 *t = q + 1; /* link to list for huft_free() */
1667 *(t = &(q->v.t)) = (struct huft *) NULL;
1668 u[h] = ++q; /* table starts after link */
1670 /* connect to last table, if there is one */
1672 x[h] = i; /* save pattern for backing up */
1673 r.b = (uch) l; /* bits to dump before this table */
1674 r.e = (uch) (16 + j); /* bits in this table */
1675 r.v.t = q; /* pointer to this table */
1676 j = i >> (w - l); /* (get around Turbo C bug) */
1677 u[h - 1][j] = r; /* connect to last table */
1681 /* set up table entry in r */
1682 r.b = (uch) (k - w);
1684 r.e = 99; /* out of values--invalid code */
1686 r.e = (uch) (*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1687 r.v.n = (ush) (*p); /* simple code is just the value */
1688 p++; /* one compiler does not like *p++ */
1690 r.e = (uch) e[*p - s]; /* non-simple--look up in lists */
1691 r.v.n = d[*p++ - s];
1694 /* fill code-like entries with r */
1696 for (j = i >> w; j < z; j += f)
1699 /* backwards increment the k-bit code i */
1700 for (j = 1 << (k - 1); i & j; j >>= 1)
1704 /* backup over finished tables */
1705 while ((i & ((1 << w) - 1)) != x[h]) {
1706 h--; /* don't need to update q */
1713 /* Return true (1) if we were given an incomplete table */
1714 return y != 0 && g != 1;
1720 struct huft *t; /* table to free */
1722 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1723 list of the tables it made, with the links in a dummy first entry of
1726 register struct huft *p, *q;
1729 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1731 while (p != (struct huft *) NULL) {
1740 int inflate_codes(tl, td, bl, bd)
1741 struct huft *tl, *td; /* literal/length and distance decoder tables */
1742 int bl, bd; /* number of bits decoded by tl[] and td[] */
1744 /* inflate (decompress) the codes in a deflated (compressed) block.
1745 Return an error code or zero if it all goes ok. */
1747 register unsigned e; /* table entry flag/number of extra bits */
1748 unsigned n, d; /* length and index for copy */
1749 unsigned w; /* current window position */
1750 struct huft *t; /* pointer to table entry */
1751 unsigned ml, md; /* masks for bl and bd bits */
1752 register ulg b; /* bit buffer */
1753 register unsigned k; /* number of bits in bit buffer */
1756 /* make local copies of globals */
1757 b = bb; /* initialize bit buffer */
1759 w = wp; /* initialize window position */
1761 /* inflate the coded data */
1762 ml = mask_bits[bl]; /* precompute masks for speed */
1764 for (;;) { /* do until end of block */
1765 NEEDBITS((unsigned) bl)
1766 if ((e = (t = tl + ((unsigned) b & ml))->e) > 16)
1773 } while ((e = (t = t->v.t + ((unsigned) b & mask_bits[e]))->e)
1776 if (e == 16) { /* then it's a literal */
1777 slide[w++] = (uch) t->v.n;
1778 Tracevv((stderr, "%c", slide[w - 1]));
1783 } else { /* it's an EOB or a length */
1785 /* exit if end of block */
1789 /* get length of block to copy */
1791 n = t->v.n + ((unsigned) b & mask_bits[e]);
1794 /* decode distance of block to copy */
1795 NEEDBITS((unsigned) bd)
1796 if ((e = (t = td + ((unsigned) b & md))->e) > 16)
1807 t->v.t + ((unsigned) b & mask_bits[e]))->e) >
1811 d = w - t->v.n - ((unsigned) b & mask_bits[e]);
1813 Tracevv((stderr, "\\[%d,%d]", w - d, n));
1819 WSIZE - ((d &= WSIZE - 1) > w ? d : w)) >
1821 #if !defined(NOMEMCPY) && !defined(DEBUG)
1822 if (w - d >= e) { /* (this test assumes unsigned comparison) */
1823 memcpy(slide + w, slide + d, e);
1826 } else /* do it slow to avoid memcpy() overlap */
1827 #endif /* !NOMEMCPY */
1829 slide[w++] = slide[d++];
1830 Tracevv((stderr, "%c", slide[w - 1]));
1841 /* restore the globals from the locals */
1842 wp = w; /* restore global window pointer */
1843 bb = b; /* restore global bit buffer */
1852 int inflate_stored()
1853 /* "decompress" an inflated type 0 (stored) block. */
1855 unsigned n; /* number of bytes in block */
1856 unsigned w; /* current window position */
1857 register ulg b; /* bit buffer */
1858 register unsigned k; /* number of bits in bit buffer */
1861 /* make local copies of globals */
1862 b = bb; /* initialize bit buffer */
1864 w = wp; /* initialize window position */
1867 /* go to byte boundary */
1872 /* get the length and its complement */
1874 n = ((unsigned) b & 0xffff);
1877 if (n != (unsigned) ((~b) & 0xffff))
1878 return 1; /* error in compressed data */
1882 /* read and output the compressed data */
1885 slide[w++] = (uch) b;
1894 /* restore the globals from the locals */
1895 wp = w; /* restore global window pointer */
1896 bb = b; /* restore global bit buffer */
1904 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
1905 either replace this with a custom decoder, or at least precompute the
1908 int i; /* temporary variable */
1909 struct huft *tl; /* literal/length code table */
1910 struct huft *td; /* distance code table */
1911 int bl; /* lookup bits for tl */
1912 int bd; /* lookup bits for td */
1913 unsigned l[288]; /* length list for huft_build */
1916 /* set up literal table */
1917 for (i = 0; i < 144; i++)
1919 for (; i < 256; i++)
1921 for (; i < 280; i++)
1923 for (; i < 288; i++) /* make a complete, but wrong code set */
1926 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
1930 /* set up distance table */
1931 for (i = 0; i < 30; i++) /* make an incomplete code set */
1934 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1) {
1940 /* decompress until an end-of-block code */
1941 if (inflate_codes(tl, td, bl, bd))
1945 /* free the decoding tables, return */
1953 int inflate_dynamic()
1954 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
1956 int i; /* temporary variables */
1958 unsigned l; /* last length */
1959 unsigned m; /* mask for bit lengths table */
1960 unsigned n; /* number of lengths to get */
1961 struct huft *tl; /* literal/length code table */
1962 struct huft *td; /* distance code table */
1963 int bl; /* lookup bits for tl */
1964 int bd; /* lookup bits for td */
1965 unsigned nb; /* number of bit length codes */
1966 unsigned nl; /* number of literal/length codes */
1967 unsigned nd; /* number of distance codes */
1969 #ifdef PKZIP_BUG_WORKAROUND
1970 unsigned ll[288 + 32]; /* literal/length and distance code lengths */
1972 unsigned ll[286 + 30]; /* literal/length and distance code lengths */
1974 register ulg b; /* bit buffer */
1975 register unsigned k; /* number of bits in bit buffer */
1978 /* make local bit buffer */
1983 /* read in table lengths */
1985 nl = 257 + ((unsigned) b & 0x1f); /* number of literal/length codes */
1988 nd = 1 + ((unsigned) b & 0x1f); /* number of distance codes */
1991 nb = 4 + ((unsigned) b & 0xf); /* number of bit length codes */
1993 #ifdef PKZIP_BUG_WORKAROUND
1994 if (nl > 288 || nd > 32)
1996 if (nl > 286 || nd > 30)
1998 return 1; /* bad lengths */
2001 /* read in bit-length-code lengths */
2002 for (j = 0; j < nb; j++) {
2004 ll[border[j]] = (unsigned) b & 7;
2011 /* build decoding table for trees--single level, 7 bit lookup */
2013 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0) {
2016 return i; /* incomplete code set */
2020 /* read in literal and distance code lengths */
2024 while ((unsigned) i < n) {
2025 NEEDBITS((unsigned) bl)
2026 j = (td = tl + ((unsigned) b & m))->b;
2029 if (j < 16) /* length of code in bits (0..15) */
2030 ll[i++] = l = j; /* save last length in l */
2031 else if (j == 16) { /* repeat last length 3 to 6 times */
2033 j = 3 + ((unsigned) b & 3);
2035 if ((unsigned) i + j > n)
2039 } else if (j == 17) { /* 3 to 10 zero length codes */
2041 j = 3 + ((unsigned) b & 7);
2043 if ((unsigned) i + j > n)
2048 } else { /* j == 18: 11 to 138 zero length codes */
2051 j = 11 + ((unsigned) b & 0x7f);
2053 if ((unsigned) i + j > n)
2062 /* free decoding table for trees */
2066 /* restore the global bit buffer */
2071 /* build the decoding tables for literal/length and distance codes */
2073 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0) {
2075 fprintf(stderr, " incomplete literal tree\n");
2078 return i; /* incomplete code set */
2081 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0) {
2083 fprintf(stderr, " incomplete distance tree\n");
2084 #ifdef PKZIP_BUG_WORKAROUND
2091 return i; /* incomplete code set */
2096 /* decompress until an end-of-block code */
2097 if (inflate_codes(tl, td, bl, bd))
2101 /* free the decoding tables, return */
2109 int inflate_block(e)
2110 int *e; /* last block flag */
2112 /* decompress an inflated block */
2114 unsigned t; /* block type */
2115 register ulg b; /* bit buffer */
2116 register unsigned k; /* number of bits in bit buffer */
2119 /* make local bit buffer */
2124 /* read in last block bit */
2130 /* read in block type */
2132 t = (unsigned) b & 3;
2136 /* restore the global bit buffer */
2141 /* inflate that block type */
2143 return inflate_dynamic();
2145 return inflate_stored();
2147 return inflate_fixed();
2150 /* bad block type */
2157 /* decompress an inflated entry */
2159 int e; /* last block flag */
2160 int r; /* result code */
2161 unsigned h; /* maximum struct huft's malloc'ed */
2164 /* initialize window, bit buffer */
2170 /* decompress until the last block */
2174 if ((r = inflate_block(&e)) != 0)
2180 /* Undo too much lookahead. The next read will be byte aligned so we
2181 * can discard unused bits in the last meaningful byte.
2188 /* flush out slide */
2192 /* return success */
2194 fprintf(stderr, "<%u> ", h);