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>
100 /* #include "tailor.h" */
102 /* tailor.h -- target dependent definitions
103 * Copyright (C) 1992-1993 Jean-loup Gailly.
104 * This is free software; you can redistribute it and/or modify it under the
105 * terms of the GNU General Public License, see the file COPYING.
108 /* The target dependent definitions should be defined here only.
109 * The target dependent functions should be defined in tailor.c.
114 #define get_char() get_byte()
115 #define put_char(c) put_byte(c)
118 /* I don't like nested includes, but the string and io functions are used
122 #if !defined(NO_STRING_H) || defined(STDC_HEADERS)
124 # if !defined(STDC_HEADERS) && !defined(NO_MEMORY_H) && !defined(__GNUC__)
127 # define memzero(s, n) memset ((void *)(s), 0, (n))
129 # include <strings.h>
130 # define strchr index
131 # define strrchr rindex
132 # define memcpy(d, s, n) bcopy((s), (d), (n))
133 # define memcmp(s1, s2, n) bcmp((s1), (s2), (n))
134 # define memzero(s, n) bzero((s), (n))
138 # define RETSIGTYPE void
143 typedef unsigned char uch;
144 typedef unsigned short ush;
145 typedef unsigned long ulg;
147 /* Return codes from gzip */
152 /* Compression methods (see algorithm.doc) */
155 extern int method; /* compression method */
157 /* To save memory for 16 bit systems, some arrays are overlaid between
158 * the various modules:
159 * deflate: prev+head window d_buf l_buf outbuf
160 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
161 * inflate: window inbuf
162 * unpack: window inbuf prefix_len
163 * unlzh: left+right window c_table inbuf c_len
164 * For compression, input is done in window[]. For decompression, output
165 * is done in window except for unlzw.
170 # define INBUFSIZ 0x2000 /* input buffer size */
172 # define INBUFSIZ 0x8000 /* input buffer size */
175 #define INBUF_EXTRA 64 /* required by unlzw() */
179 # define OUTBUFSIZ 8192 /* output buffer size */
181 # define OUTBUFSIZ 16384 /* output buffer size */
184 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
190 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
192 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
196 /*#define DYN_ALLOC*/
199 # define EXTERN(type, array) extern type * array
200 # define DECLARE(type, array, size) type * array
201 # define ALLOC(type, array, size) { \
202 array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
203 if (array == NULL) errorMsg(memory_exhausted); \
205 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
207 # define EXTERN(type, array) extern type array[]
208 # define DECLARE(type, array, size) type array[size]
209 # define ALLOC(type, array, size)
213 EXTERN(uch, inbuf); /* input buffer */
214 EXTERN(uch, outbuf); /* output buffer */
215 EXTERN(ush, d_buf); /* buffer for distances, see trees.c */
216 EXTERN(uch, window); /* Sliding window and suffix table (unlzw) */
217 #define tab_suffix window
219 # define tab_prefix prev /* hash link (see deflate.c) */
220 # define head (prev+WSIZE) /* hash head (see deflate.c) */
221 EXTERN(ush, tab_prefix); /* prefix code (see unlzw.c) */
223 # define tab_prefix0 prev
224 # define head tab_prefix1
225 EXTERN(ush, tab_prefix0); /* prefix for even codes */
226 EXTERN(ush, tab_prefix1); /* prefix for odd codes */
229 extern unsigned insize; /* valid bytes in inbuf */
230 extern unsigned inptr; /* index of next byte to be processed in inbuf */
231 extern unsigned outcnt; /* bytes in output buffer */
233 extern long bytes_in; /* number of input bytes */
234 extern long bytes_out; /* number of output bytes */
235 extern long header_bytes; /* number of bytes in gzip header */
237 extern long ifile_size; /* input file size, -1 for devices (debug only) */
239 typedef int file_t; /* Do not use stdio */
241 #define NO_FILE (-1) /* in memory compression */
244 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
247 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
248 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
249 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
250 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
251 #define COMMENT 0x10 /* bit 4 set: file comment present */
252 #define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
253 #define RESERVED 0xC0 /* bit 6,7: reserved */
256 # define WSIZE 0x8000 /* window size--must be a power of two, and */
257 #endif /* at least 32K for zip's deflate method */
260 #define MAX_MATCH 258
261 /* The minimum and maximum match lengths */
263 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
264 /* Minimum amount of lookahead, except at the end of the input file.
265 * See deflate.c for comments about the MIN_MATCH+1.
268 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
269 /* In order to simplify the code, particularly on 16 bit machines, match
270 * distances are limited to MAX_DIST instead of WSIZE.
273 extern int exit_code; /* program exit code */
274 extern int verbose; /* be verbose (-v) */
275 extern int level; /* compression level */
276 extern int test; /* check .z file integrity */
277 extern int save_orig_name; /* set if original name must be saved */
279 #define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(0))
280 #define try_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf(1))
282 /* put_byte is used for the compressed output, put_ubyte for the
283 * uncompressed output. However unlzw() uses window for its
284 * suffix table instead of its output buffer, so it does not use put_ubyte
285 * (to be cleaned up).
287 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
289 #define put_ubyte(c) {window[outcnt++]=(uch)(c); if (outcnt==WSIZE)\
292 /* Output a 16 bit value, lsb first */
293 #define put_short(w) \
294 { if (outcnt < OUTBUFSIZ-2) { \
295 outbuf[outcnt++] = (uch) ((w) & 0xff); \
296 outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \
298 put_byte((uch)((w) & 0xff)); \
299 put_byte((uch)((ush)(w) >> 8)); \
303 /* Output a 32 bit value to the bit stream, lsb first */
304 #define put_long(n) { \
305 put_short((n) & 0xffff); \
306 put_short(((ulg)(n)) >> 16); \
309 #define seekable() 0 /* force sequential output */
310 #define translate_eol 0 /* no option -a yet */
312 #define tolow(c) (isupper(c) ? (c)-'A'+'a' : (c)) /* force to lower case */
314 /* Macros for getting two-byte and four-byte header values */
315 #define SH(p) ((ush)(uch)((p)[0]) | ((ush)(uch)((p)[1]) << 8))
316 #define LG(p) ((ulg)(SH(p)) | ((ulg)(SH((p)+2)) << 16))
318 /* Diagnostic functions */
320 # define Assert(cond,msg) {if(!(cond)) errorMsg(msg);}
321 # define Trace(x) fprintf x
322 # define Tracev(x) {if (verbose) fprintf x ;}
323 # define Tracevv(x) {if (verbose>1) fprintf x ;}
324 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
325 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
327 # define Assert(cond,msg)
332 # define Tracecv(c,x)
335 #define WARN(msg) {fprintf msg ; \
336 if (exit_code == OK) exit_code = WARNING;}
339 extern int unzip (int in, int out);
342 RETSIGTYPE abort_gzip (void);
345 void lm_init (int pack_level, ush * flags);
349 void ct_init (ush * attr, int *method);
350 int ct_tally (int dist, int lc);
351 ulg flush_block (char *buf, ulg stored_len, int eof);
354 void bi_init (file_t zipfile);
355 void send_bits (int value, int length);
356 unsigned bi_reverse (unsigned value, int length);
357 void bi_windup (void);
358 void copy_block (char *buf, unsigned len, int header);
361 extern ulg updcrc (uch * s, unsigned n);
362 extern void clear_bufs (void);
363 static int fill_inbuf (int eof_ok);
364 extern void flush_outbuf (void);
365 static void flush_window (void);
366 extern void write_buf (int fd, void * buf, unsigned cnt);
369 static char *basename (char *fname);
370 #endif /* not __linux__ */
371 void read_error_msg (void);
372 void write_error_msg (void);
375 static int inflate (void);
377 /* #include "lzw.h" */
379 /* lzw.h -- define the lzw functions.
380 * Copyright (C) 1992-1993 Jean-loup Gailly.
381 * This is free software; you can redistribute it and/or modify it under the
382 * terms of the GNU General Public License, see the file COPYING.
385 #if !defined(OF) && defined(lint)
392 #define INIT_BITS 9 /* Initial number of bits per code */
394 #define LZW_MAGIC "\037\235" /* Magic header for lzw files, 1F 9D */
396 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
397 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
398 * It's a pity that old uncompress does not check bit 0x20. That makes
399 * extension of the format actually undesirable because old compress
400 * would just crash on the new format instead of giving a meaningful
401 * error message. It does check the number of bits, but it's more
402 * helpful to say "unsupported format, get a new version" than
403 * "can only handle 16 bits".
406 #define BLOCK_MODE 0x80
407 /* Block compression: if table is full and compression rate is dropping,
408 * clear the dictionary.
411 #define LZW_RESERVED 0x60 /* reserved bits */
413 #define CLEAR 256 /* flush the dictionary */
414 #define FIRST (CLEAR+1) /* first free entry */
416 extern int maxbits; /* max bits per code for LZW */
417 extern int block_mode; /* block compress mode -C compatible with 2.0 */
419 extern int lzw (int in, int out);
420 extern int unlzw (int in, int out);
423 /* #include "revision.h" */
425 /* revision.h -- define the version number
426 * Copyright (C) 1992-1993 Jean-loup Gailly.
427 * This is free software; you can redistribute it and/or modify it under the
428 * terms of the GNU General Public License, see the file COPYING.
431 #define VERSION "1.2.4"
433 #define REVDATE "18 Aug 93"
435 /* This version does not support compression into old compress format: */
446 typedef struct dirent dir_type;
448 # define NLENGTH(dirent) ((int)strlen((dirent)->d_name))
449 # define DIR_OPT "DIRENT"
451 # define NLENGTH(dirent) ((dirent)->d_namlen)
453 # include <sys/dir.h>
454 typedef struct direct dir_type;
456 # define DIR_OPT "SYSDIR"
459 # include <sys/ndir.h>
460 typedef struct direct dir_type;
462 # define DIR_OPT "SYSNDIR"
466 typedef struct direct dir_type;
468 # define DIR_OPT "NDIR"
471 # define DIR_OPT "NO_DIR"
476 #if !defined(S_ISDIR) && defined(S_IFDIR)
477 # define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
479 #if !defined(S_ISREG) && defined(S_IFREG)
480 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
482 typedef RETSIGTYPE(*sig_type) (int);
485 # define O_BINARY 0 /* creation mode for open() */
489 /* Pure BSD system? */
490 # include <sys/file.h>
492 # define O_CREAT FCREAT
495 # define O_EXCL FEXCL
500 # define S_IRUSR 0400
503 # define S_IWUSR 0200
505 #define RW_USER (S_IRUSR | S_IWUSR) /* creation mode for open() */
507 #ifndef MAX_PATH_LEN /* max pathname length */
509 # define MAX_PATH_LEN BUFSIZ
511 # define MAX_PATH_LEN 1024
521 off_t lseek (int fd, off_t offset, int whence);
527 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
528 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
529 DECLARE(ush, d_buf, DIST_BUFSIZE);
530 DECLARE(uch, window, 2L * WSIZE);
532 DECLARE(ush, tab_prefix, 1L << BITS);
534 DECLARE(ush, tab_prefix0, 1L << (BITS - 1));
535 DECLARE(ush, tab_prefix1, 1L << (BITS - 1));
538 /* local variables */
540 int test_mode = 0; /* check file integrity option */
541 int foreground; /* set if program run in foreground */
542 int maxbits = BITS; /* max bits per code for LZW */
543 int method = DEFLATED; /* compression method */
544 int exit_code = OK; /* program exit code */
545 int last_member; /* set for .zip and .Z files */
546 int part_nb; /* number of parts in .gz file */
547 long ifile_size; /* input file size, -1 for devices (debug only) */
549 long bytes_in; /* number of input bytes */
550 long bytes_out; /* number of output bytes */
551 long total_in = 0; /* input bytes for all files */
552 long total_out = 0; /* output bytes for all files */
553 struct stat istat; /* status for input file */
554 int ifd; /* input file descriptor */
555 int ofd; /* output file descriptor */
556 unsigned insize; /* valid bytes in inbuf */
557 unsigned inptr; /* index of next byte to be processed in inbuf */
558 unsigned outcnt; /* bytes in output buffer */
560 long header_bytes; /* number of bytes in gzip header */
562 /* local functions */
564 local int get_method (int in);
566 #define strequ(s1, s2) (strcmp((s1),(s2)) == 0)
568 /* ======================================================================== */
569 int gunzip_main(int argc, char **argv)
571 int file_count; /* number of files to precess */
577 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) {
590 /* Parse any options */
591 while (--argc > 0 && **(++argv) == '-') {
592 if (*((*argv) + 1) == '\0') {
596 while (*(++(*argv))) {
611 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
613 (void) signal(SIGINT, (sig_type) abort_gzip);
616 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
617 (void) signal(SIGTERM, (sig_type) abort_gzip);
621 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
622 (void) signal(SIGHUP, (sig_type) abort_gzip);
626 file_count = argc - optind;
628 /* Allocate all global buffers (for DYN_ALLOC option) */
629 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
630 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
631 ALLOC(ush, d_buf, DIST_BUFSIZE);
632 ALLOC(uch, window, 2L * WSIZE);
634 ALLOC(ush, tab_prefix, 1L << BITS);
636 ALLOC(ush, tab_prefix0, 1L << (BITS - 1));
637 ALLOC(ush, tab_prefix1, 1L << (BITS - 1));
640 if (fromstdin == 1) {
641 strcpy(ofname, "stdin");
643 inFileNum = fileno(stdin);
644 ifile_size = -1L; /* convention for unknown size */
646 /* Open up the input file */
649 if (strlen(*argv) > MAX_PATH_LEN) {
650 errorMsg(name_too_long);
653 strcpy(ifname, *argv);
655 /* Open input fille */
656 inFileNum = open(ifname, O_RDONLY);
661 /* Get the time stamp on the input file. */
662 result = stat(ifname, &statBuf);
667 ifile_size = statBuf.st_size;
670 if (to_stdout == 1) {
671 /* And get to work */
672 strcpy(ofname, "stdout");
673 outFileNum = fileno(stdout);
675 clear_bufs(); /* clear input and output buffers */
678 /* Actually do the compression/decompression. */
679 unzip(inFileNum, outFileNum);
681 } else if (test_mode) {
682 /* Actually do the compression/decompression. */
687 /* And get to work */
688 if (strlen(ifname) > MAX_PATH_LEN - 4) {
689 errorMsg(name_too_long);
692 strcpy(ofname, ifname);
693 pos = strstr(ofname, ".gz");
698 pos = strstr(ofname, ".tgz");
706 /* Open output fille */
707 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
708 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
710 outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL);
712 if (outFileNum < 0) {
716 /* Set permissions on the file */
717 fchmod(outFileNum, statBuf.st_mode);
719 clear_bufs(); /* clear input and output buffers */
722 /* Actually do the compression/decompression. */
723 result = unzip(inFileNum, outFileNum);
727 /* Delete the original file */
729 delFileName = ifname;
731 delFileName = ofname;
733 if (delInputFile == 1 && unlink(delFileName) < 0) {
742 /* ========================================================================
743 * Check the magic number of the input file and update ofname if an
744 * original name was given and to_stdout is not set.
745 * Return the compression method, -1 for error, -2 for warning.
746 * Set inptr to the offset of the next byte to be processed.
747 * Updates time_stamp if there is one and --no-time is not used.
748 * This function may be called repeatedly for an input file consisting
749 * of several contiguous gzip'ed members.
750 * IN assertions: there is at least one remaining compressed member.
751 * If the member is a zip file, it must be the only one.
753 local int get_method(in)
754 int in; /* input file descriptor */
756 uch flags; /* compression flags */
757 char magic[2]; /* magic header */
759 magic[0] = (char) get_byte();
760 magic[1] = (char) get_byte();
761 method = -1; /* unknown yet */
762 part_nb++; /* number of parts in gzip file */
764 last_member = RECORD_IO;
765 /* assume multiple members in gzip file except for record oriented I/O */
767 if (memcmp(magic, GZIP_MAGIC, 2) == 0) {
769 method = (int) get_byte();
770 if (method != DEFLATED) {
771 errorMsg("unknown method %d -- get newer version of gzip\n",
776 flags = (uch) get_byte();
778 (ulg) get_byte(); /* Ignore time stamp */
783 (void) get_byte(); /* Ignore extra flags for the moment */
784 (void) get_byte(); /* Ignore OS type for the moment */
786 if ((flags & EXTRA_FIELD) != 0) {
787 unsigned len = (unsigned) get_byte();
789 len |= ((unsigned) get_byte()) << 8;
795 /* Discard original name if any */
796 if ((flags & ORIG_NAME) != 0) {
797 while (get_char() != 0) /* null */
801 /* Discard file comment if any */
802 if ((flags & COMMENT) != 0) {
803 while (get_char() != 0) /* null */
807 header_bytes = inptr + 2 * sizeof(long); /* include crc and size */
816 fprintf(stderr, "\nnot in gzip format\n");
820 WARN((stderr, "\ndecompression OK, trailing garbage ignored\n"));
825 /* ========================================================================
826 * Signal and error handler.
828 RETSIGTYPE abort_gzip()
833 /* unzip.c -- decompress files in gzip or pkzip format.
834 * Copyright (C) 1992-1993 Jean-loup Gailly
835 * This is free software; you can redistribute it and/or modify it under the
836 * terms of the GNU General Public License, see the file COPYING.
838 * The code in this file is derived from the file funzip.c written
839 * and put in the public domain by Mark Adler.
843 This version can extract files in gzip or pkzip format.
844 For the latter, only the first entry is extracted, and it has to be
845 either deflated or stored.
848 /* #include "crypt.h" */
850 /* crypt.h (dummy version) -- do not perform encryption
851 * Hardly worth copyrighting :-)
855 # undef CRYPT /* dummy version */
858 #define RAND_HEAD_LEN 12 /* length of encryption random header */
863 /* PKZIP header definitions */
864 #define LOCSIG 0x04034b50L /* four-byte lead-in (lsb first) */
865 #define LOCFLG 6 /* offset of bit flag */
866 #define CRPFLG 1 /* bit for encrypted entry */
867 #define EXTFLG 8 /* bit for extended local header */
868 #define LOCHOW 8 /* offset of compression method */
869 #define LOCTIM 10 /* file mod time (for decryption) */
870 #define LOCCRC 14 /* offset of crc */
871 #define LOCSIZ 18 /* offset of compressed size */
872 #define LOCLEN 22 /* offset of uncompressed length */
873 #define LOCFIL 26 /* offset of file name field length */
874 #define LOCEXT 28 /* offset of extra field length */
875 #define LOCHDR 30 /* size of local header, including sig */
876 #define EXTHDR 16 /* size of extended local header, inc sig */
881 char *key; /* not used--needed to link crypt.c */
882 int pkzip = 0; /* set for a pkzip file */
883 int ext_header = 0; /* set if extended local header */
885 /* ===========================================================================
886 * Unzip in to out. This routine works on both gzip and pkzip files.
888 * IN assertions: the buffer inbuf contains already the beginning of
889 * the compressed data, from offsets inptr to insize-1 included.
890 * The magic header has already been checked. The output buffer is cleared.
893 int in, out; /* input and output file descriptors */
895 ulg orig_crc = 0; /* original crc */
896 ulg orig_len = 0; /* original uncompressed length */
898 uch buf[EXTHDR]; /* extended local header */
902 method = get_method(ifd);
904 exit(exit_code); /* error message already emitted */
907 updcrc(NULL, 0); /* initialize crc */
909 if (pkzip && !ext_header) { /* crc and length at the end otherwise */
910 orig_crc = LG(inbuf + LOCCRC);
911 orig_len = LG(inbuf + LOCLEN);
915 if (method == DEFLATED) {
920 errorMsg(memory_exhausted);
921 } else if (res != 0) {
922 errorMsg("invalid compressed data--format violated");
926 errorMsg("internal error, invalid method");
929 /* Get the crc and original length */
931 /* crc32 (see algorithm.doc)
932 * uncompressed input size modulo 2^32
934 for (n = 0; n < 8; n++) {
935 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
938 orig_len = LG(buf + 4);
940 } else if (ext_header) { /* If extended header, check it */
941 /* signature - 4bytes: 0x50 0x4b 0x07 0x08
943 * compressed size 4-bytes
944 * uncompressed size 4-bytes
946 for (n = 0; n < EXTHDR; n++) {
947 buf[n] = (uch) get_byte(); /* may cause an error if EOF */
949 orig_crc = LG(buf + 4);
950 orig_len = LG(buf + 12);
953 /* Validate decompression */
954 if (orig_crc != updcrc(outbuf, 0)) {
955 errorMsg("invalid compressed data--crc error");
957 if (orig_len != (ulg) bytes_out) {
958 errorMsg("invalid compressed data--length error");
961 /* Check if there are more entries in a pkzip file */
962 if (pkzip && inptr + 4 < insize && LG(inbuf + inptr) == LOCSIG) {
963 WARN((stderr, "has more than one entry--rest ignored\n"));
965 ext_header = pkzip = 0; /* for next file */
969 /* util.c -- utility functions for gzip support
970 * Copyright (C) 1992-1993 Jean-loup Gailly
971 * This is free software; you can redistribute it and/or modify it under the
972 * terms of the GNU General Public License, see the file COPYING.
977 #include <sys/types.h>
986 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
992 static const ulg crc_32_tab[]; /* crc table, defined below */
994 /* ===========================================================================
995 * Run a set of bytes through the crc shift register. If s is a NULL
996 * pointer, then initialize the crc shift register contents instead.
997 * Return the current crc in either case.
1000 uch *s; /* pointer to bytes to pump through */
1001 unsigned n; /* number of bytes in s[] */
1003 register ulg c; /* temporary variable */
1005 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
1013 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
1017 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
1020 /* ===========================================================================
1021 * Clear input and output buffers
1023 void clear_bufs(void)
1027 bytes_in = bytes_out = 0L;
1030 /* ===========================================================================
1031 * Fill the input buffer. This is called only when the buffer is empty.
1033 int fill_inbuf(eof_ok)
1034 int eof_ok; /* set if EOF acceptable as a result */
1038 /* Read as much as possible */
1042 len = read(ifd, (char *) inbuf + insize, INBUFSIZ - insize);
1043 if (len == 0 || len == EOF)
1046 } while (insize < INBUFSIZ);
1053 bytes_in += (ulg) insize;
1058 /* ===========================================================================
1059 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
1060 * (used for the compressed data only)
1068 write_buf(ofd, (char *) outbuf, outcnt);
1069 bytes_out += (ulg) outcnt;
1073 /* ===========================================================================
1074 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
1075 * (Used for the decompressed data only.)
1081 updcrc(window, outcnt);
1084 write_buf(ofd, (char *) window, outcnt);
1085 bytes_out += (ulg) outcnt;
1089 /* ===========================================================================
1090 * Does the same as write(), but also handles partial pipe writes and checks
1093 void write_buf(fd, buf, cnt)
1100 while ((n = write(fd, buf, cnt)) != cnt) {
1101 if (n == (unsigned) (-1)) {
1105 buf = (void *) ((char *) buf + n);
1109 #if defined(NO_STRING_H) && !defined(STDC_HEADERS)
1111 /* Provide missing strspn and strcspn functions. */
1117 int strspn (const char *s, const char *accept);
1118 int strcspn (const char *s, const char *reject);
1120 /* ========================================================================
1121 * Return the length of the maximum initial segment
1122 * of s which contains only characters in accept.
1124 int strspn(s, accept)
1128 register const char *p;
1129 register const char *a;
1130 register int count = 0;
1132 for (p = s; *p != '\0'; ++p) {
1133 for (a = accept; *a != '\0'; ++a) {
1144 /* ========================================================================
1145 * Return the length of the maximum inital segment of s
1146 * which contains no characters from reject.
1148 int strcspn(s, reject)
1152 register int count = 0;
1154 while (*s != '\0') {
1155 if (strchr(reject, *s++) != NULL)
1162 #endif /* NO_STRING_H */
1165 /* ========================================================================
1168 void read_error_msg()
1170 fprintf(stderr, "\n");
1174 fprintf(stderr, "unexpected end of file\n");
1179 void write_error_msg()
1181 fprintf(stderr, "\n");
1187 /* ========================================================================
1188 * Table of CRC-32's of all single-byte values (made by makecrc.c)
1190 static const ulg crc_32_tab[] = {
1191 0x00000000L, 0x77073096L, 0xee0e612cL, 0x990951baL, 0x076dc419L,
1192 0x706af48fL, 0xe963a535L, 0x9e6495a3L, 0x0edb8832L, 0x79dcb8a4L,
1193 0xe0d5e91eL, 0x97d2d988L, 0x09b64c2bL, 0x7eb17cbdL, 0xe7b82d07L,
1194 0x90bf1d91L, 0x1db71064L, 0x6ab020f2L, 0xf3b97148L, 0x84be41deL,
1195 0x1adad47dL, 0x6ddde4ebL, 0xf4d4b551L, 0x83d385c7L, 0x136c9856L,
1196 0x646ba8c0L, 0xfd62f97aL, 0x8a65c9ecL, 0x14015c4fL, 0x63066cd9L,
1197 0xfa0f3d63L, 0x8d080df5L, 0x3b6e20c8L, 0x4c69105eL, 0xd56041e4L,
1198 0xa2677172L, 0x3c03e4d1L, 0x4b04d447L, 0xd20d85fdL, 0xa50ab56bL,
1199 0x35b5a8faL, 0x42b2986cL, 0xdbbbc9d6L, 0xacbcf940L, 0x32d86ce3L,
1200 0x45df5c75L, 0xdcd60dcfL, 0xabd13d59L, 0x26d930acL, 0x51de003aL,
1201 0xc8d75180L, 0xbfd06116L, 0x21b4f4b5L, 0x56b3c423L, 0xcfba9599L,
1202 0xb8bda50fL, 0x2802b89eL, 0x5f058808L, 0xc60cd9b2L, 0xb10be924L,
1203 0x2f6f7c87L, 0x58684c11L, 0xc1611dabL, 0xb6662d3dL, 0x76dc4190L,
1204 0x01db7106L, 0x98d220bcL, 0xefd5102aL, 0x71b18589L, 0x06b6b51fL,
1205 0x9fbfe4a5L, 0xe8b8d433L, 0x7807c9a2L, 0x0f00f934L, 0x9609a88eL,
1206 0xe10e9818L, 0x7f6a0dbbL, 0x086d3d2dL, 0x91646c97L, 0xe6635c01L,
1207 0x6b6b51f4L, 0x1c6c6162L, 0x856530d8L, 0xf262004eL, 0x6c0695edL,
1208 0x1b01a57bL, 0x8208f4c1L, 0xf50fc457L, 0x65b0d9c6L, 0x12b7e950L,
1209 0x8bbeb8eaL, 0xfcb9887cL, 0x62dd1ddfL, 0x15da2d49L, 0x8cd37cf3L,
1210 0xfbd44c65L, 0x4db26158L, 0x3ab551ceL, 0xa3bc0074L, 0xd4bb30e2L,
1211 0x4adfa541L, 0x3dd895d7L, 0xa4d1c46dL, 0xd3d6f4fbL, 0x4369e96aL,
1212 0x346ed9fcL, 0xad678846L, 0xda60b8d0L, 0x44042d73L, 0x33031de5L,
1213 0xaa0a4c5fL, 0xdd0d7cc9L, 0x5005713cL, 0x270241aaL, 0xbe0b1010L,
1214 0xc90c2086L, 0x5768b525L, 0x206f85b3L, 0xb966d409L, 0xce61e49fL,
1215 0x5edef90eL, 0x29d9c998L, 0xb0d09822L, 0xc7d7a8b4L, 0x59b33d17L,
1216 0x2eb40d81L, 0xb7bd5c3bL, 0xc0ba6cadL, 0xedb88320L, 0x9abfb3b6L,
1217 0x03b6e20cL, 0x74b1d29aL, 0xead54739L, 0x9dd277afL, 0x04db2615L,
1218 0x73dc1683L, 0xe3630b12L, 0x94643b84L, 0x0d6d6a3eL, 0x7a6a5aa8L,
1219 0xe40ecf0bL, 0x9309ff9dL, 0x0a00ae27L, 0x7d079eb1L, 0xf00f9344L,
1220 0x8708a3d2L, 0x1e01f268L, 0x6906c2feL, 0xf762575dL, 0x806567cbL,
1221 0x196c3671L, 0x6e6b06e7L, 0xfed41b76L, 0x89d32be0L, 0x10da7a5aL,
1222 0x67dd4accL, 0xf9b9df6fL, 0x8ebeeff9L, 0x17b7be43L, 0x60b08ed5L,
1223 0xd6d6a3e8L, 0xa1d1937eL, 0x38d8c2c4L, 0x4fdff252L, 0xd1bb67f1L,
1224 0xa6bc5767L, 0x3fb506ddL, 0x48b2364bL, 0xd80d2bdaL, 0xaf0a1b4cL,
1225 0x36034af6L, 0x41047a60L, 0xdf60efc3L, 0xa867df55L, 0x316e8eefL,
1226 0x4669be79L, 0xcb61b38cL, 0xbc66831aL, 0x256fd2a0L, 0x5268e236L,
1227 0xcc0c7795L, 0xbb0b4703L, 0x220216b9L, 0x5505262fL, 0xc5ba3bbeL,
1228 0xb2bd0b28L, 0x2bb45a92L, 0x5cb36a04L, 0xc2d7ffa7L, 0xb5d0cf31L,
1229 0x2cd99e8bL, 0x5bdeae1dL, 0x9b64c2b0L, 0xec63f226L, 0x756aa39cL,
1230 0x026d930aL, 0x9c0906a9L, 0xeb0e363fL, 0x72076785L, 0x05005713L,
1231 0x95bf4a82L, 0xe2b87a14L, 0x7bb12baeL, 0x0cb61b38L, 0x92d28e9bL,
1232 0xe5d5be0dL, 0x7cdcefb7L, 0x0bdbdf21L, 0x86d3d2d4L, 0xf1d4e242L,
1233 0x68ddb3f8L, 0x1fda836eL, 0x81be16cdL, 0xf6b9265bL, 0x6fb077e1L,
1234 0x18b74777L, 0x88085ae6L, 0xff0f6a70L, 0x66063bcaL, 0x11010b5cL,
1235 0x8f659effL, 0xf862ae69L, 0x616bffd3L, 0x166ccf45L, 0xa00ae278L,
1236 0xd70dd2eeL, 0x4e048354L, 0x3903b3c2L, 0xa7672661L, 0xd06016f7L,
1237 0x4969474dL, 0x3e6e77dbL, 0xaed16a4aL, 0xd9d65adcL, 0x40df0b66L,
1238 0x37d83bf0L, 0xa9bcae53L, 0xdebb9ec5L, 0x47b2cf7fL, 0x30b5ffe9L,
1239 0xbdbdf21cL, 0xcabac28aL, 0x53b39330L, 0x24b4a3a6L, 0xbad03605L,
1240 0xcdd70693L, 0x54de5729L, 0x23d967bfL, 0xb3667a2eL, 0xc4614ab8L,
1241 0x5d681b02L, 0x2a6f2b94L, 0xb40bbe37L, 0xc30c8ea1L, 0x5a05df1bL,
1245 /* inflate.c -- Not copyrighted 1992 by Mark Adler
1246 version c10p1, 10 January 1993 */
1248 /* You can do whatever you like with this source file, though I would
1249 prefer that if you modify it and redistribute it that you include
1250 comments to that effect with your name and the date. Thank you.
1251 [The history has been moved to the file ChangeLog.]
1255 Inflate deflated (PKZIP's method 8 compressed) data. The compression
1256 method searches for as much of the current string of bytes (up to a
1257 length of 258) in the previous 32K bytes. If it doesn't find any
1258 matches (of at least length 3), it codes the next byte. Otherwise, it
1259 codes the length of the matched string and its distance backwards from
1260 the current position. There is a single Huffman code that codes both
1261 single bytes (called "literals") and match lengths. A second Huffman
1262 code codes the distance information, which follows a length code. Each
1263 length or distance code actually represents a base value and a number
1264 of "extra" (sometimes zero) bits to get to add to the base value. At
1265 the end of each deflated block is a special end-of-block (EOB) literal/
1266 length code. The decoding process is basically: get a literal/length
1267 code; if EOB then done; if a literal, emit the decoded byte; if a
1268 length then get the distance and emit the referred-to bytes from the
1269 sliding window of previously emitted data.
1271 There are (currently) three kinds of inflate blocks: stored, fixed, and
1272 dynamic. The compressor deals with some chunk of data at a time, and
1273 decides which method to use on a chunk-by-chunk basis. A chunk might
1274 typically be 32K or 64K. If the chunk is uncompressible, then the
1275 "stored" method is used. In this case, the bytes are simply stored as
1276 is, eight bits per byte, with none of the above coding. The bytes are
1277 preceded by a count, since there is no longer an EOB code.
1279 If the data is compressible, then either the fixed or dynamic methods
1280 are used. In the dynamic method, the compressed data is preceded by
1281 an encoding of the literal/length and distance Huffman codes that are
1282 to be used to decode this block. The representation is itself Huffman
1283 coded, and so is preceded by a description of that code. These code
1284 descriptions take up a little space, and so for small blocks, there is
1285 a predefined set of codes, called the fixed codes. The fixed method is
1286 used if the block codes up smaller that way (usually for quite small
1287 chunks), otherwise the dynamic method is used. In the latter case, the
1288 codes are customized to the probabilities in the current block, and so
1289 can code it much better than the pre-determined fixed codes.
1291 The Huffman codes themselves are decoded using a mutli-level table
1292 lookup, in order to maximize the speed of decoding plus the speed of
1293 building the decoding tables. See the comments below that precede the
1294 lbits and dbits tuning parameters.
1299 Notes beyond the 1.93a appnote.txt:
1301 1. Distance pointers never point before the beginning of the output
1303 2. Distance pointers can point back across blocks, up to 32k away.
1304 3. There is an implied maximum of 7 bits for the bit length table and
1305 15 bits for the actual data.
1306 4. If only one code exists, then it is encoded using one bit. (Zero
1307 would be more efficient, but perhaps a little confusing.) If two
1308 codes exist, they are coded using one bit each (0 and 1).
1309 5. There is no way of sending zero distance codes--a dummy must be
1310 sent if there are none. (History: a pre 2.0 version of PKZIP would
1311 store blocks with no distance codes, but this was discovered to be
1312 too harsh a criterion.) Valid only for 1.93a. 2.04c does allow
1313 zero distance codes, which is sent as one code of zero bits in
1315 6. There are up to 286 literal/length codes. Code 256 represents the
1316 end-of-block. Note however that the static length tree defines
1317 288 codes just to fill out the Huffman codes. Codes 286 and 287
1318 cannot be used though, since there is no length base or extra bits
1319 defined for them. Similarly, there are up to 30 distance codes.
1320 However, static trees define 32 codes (all 5 bits) to fill out the
1321 Huffman codes, but the last two had better not show up in the data.
1322 7. Unzip can check dynamic Huffman blocks for complete code sets.
1323 The exception is that a single code would not be complete (see #4).
1324 8. The five bits following the block type is really the number of
1325 literal codes sent minus 257.
1326 9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
1327 (1+6+6). Therefore, to output three times the length, you output
1328 three codes (1+1+1), whereas to output four times the same length,
1329 you only need two codes (1+3). Hmm.
1330 10. In the tree reconstruction algorithm, Code = Code + Increment
1331 only if BitLength(i) is not zero. (Pretty obvious.)
1332 11. Correction: 4 Bits: # of Bit Length codes - 4 (4 - 19)
1333 12. Note: length code 284 can represent 227-258, but length code 285
1334 really is 258. The last length deserves its own, short code
1335 since it gets used a lot in very redundant files. The length
1336 258 is special since 258 - 3 (the min match length) is 255.
1337 13. The literal/length and distance code bit lengths are read as a
1338 single stream of lengths. It is possible (and advantageous) for
1339 a repeat code (16, 17, or 18) to go across the boundary between
1340 the two sets of lengths.
1343 #include <sys/types.h>
1345 #if defined(STDC_HEADERS) || !defined(NO_STDLIB_H)
1346 # include <stdlib.h>
1350 #define slide window
1352 /* Huffman code lookup table entry--this entry is four bytes for machines
1353 that have 16-bit pointers (e.g. PC's in the small or medium model).
1354 Valid extra bits are 0..13. e == 15 is EOB (end of block), e == 16
1355 means that v is a literal, 16 < e < 32 means that v is a pointer to
1356 the next table, which codes e - 16 bits, and lastly e == 99 indicates
1357 an unused code. If a code with e == 99 is looked up, this implies an
1358 error in the data. */
1360 uch e; /* number of extra bits or operation */
1361 uch b; /* number of bits in this code or subcode */
1363 ush n; /* literal, length base, or distance base */
1364 struct huft *t; /* pointer to next level of table */
1369 /* Function prototypes */
1370 int huft_build (unsigned *, unsigned, unsigned, ush *, ush *,
1371 struct huft **, int *);
1372 int huft_free (struct huft *);
1373 int inflate_codes (struct huft *, struct huft *, int, int);
1374 int inflate_stored (void);
1375 int inflate_fixed (void);
1376 int inflate_dynamic (void);
1377 int inflate_block (int *);
1381 /* The inflate algorithm uses a sliding 32K byte window on the uncompressed
1382 stream to find repeated byte strings. This is implemented here as a
1383 circular buffer. The index is updated simply by incrementing and then
1384 and'ing with 0x7fff (32K-1). */
1385 /* It is left to other modules to supply the 32K area. It is assumed
1386 to be usable as if it were declared "uch slide[32768];" or as just
1387 "uch *slide;" and then malloc'ed in the latter case. The definition
1388 must be in unzip.h, included above. */
1389 /* unsigned wp; current position in slide */
1391 #define flush_output(w) (wp=(w),flush_window())
1393 /* Tables for deflate from PKZIP's appnote.txt. */
1394 static unsigned border[] = { /* Order of the bit length code lengths */
1395 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15
1397 static ush cplens[] = { /* Copy lengths for literal codes 257..285 */
1398 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
1399 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0
1402 /* note: see note #13 above about the 258 in this list. */
1403 static ush cplext[] = { /* Extra bits for literal codes 257..285 */
1404 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
1405 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 99, 99
1406 }; /* 99==invalid */
1407 static ush cpdist[] = { /* Copy offsets for distance codes 0..29 */
1408 1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
1409 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
1410 8193, 12289, 16385, 24577
1412 static ush cpdext[] = { /* Extra bits for distance codes */
1413 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
1414 7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
1420 /* Macros for inflate() bit peeking and grabbing.
1424 x = b & mask_bits[j];
1427 where NEEDBITS makes sure that b has at least j bits in it, and
1428 DUMPBITS removes the bits from b. The macros use the variable k
1429 for the number of bits in b. Normally, b and k are register
1430 variables for speed, and are initialized at the beginning of a
1431 routine that uses these macros from a global bit buffer and count.
1433 If we assume that EOB will be the longest code, then we will never
1434 ask for bits with NEEDBITS that are beyond the end of the stream.
1435 So, NEEDBITS should not read any more bytes than are needed to
1436 meet the request. Then no bytes need to be "returned" to the buffer
1437 at the end of the last block.
1439 However, this assumption is not true for fixed blocks--the EOB code
1440 is 7 bits, but the other literal/length codes can be 8 or 9 bits.
1441 (The EOB code is shorter than other codes because fixed blocks are
1442 generally short. So, while a block always has an EOB, many other
1443 literal/length codes have a significantly lower probability of
1444 showing up at all.) However, by making the first table have a
1445 lookup of seven bits, the EOB code will be found in that first
1446 lookup, and so will not require that too many bits be pulled from
1450 ulg bb; /* bit buffer */
1451 unsigned bk; /* bits in bit buffer */
1455 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
1456 0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
1462 # define NEXTBYTE() (cc = get_byte(), zdecode(cc), cc)
1464 # define NEXTBYTE() (uch)get_byte()
1466 #define NEEDBITS(n) {while(k<(n)){b|=((ulg)NEXTBYTE())<<k;k+=8;}}
1467 #define DUMPBITS(n) {b>>=(n);k-=(n);}
1471 Huffman code decoding is performed using a multi-level table lookup.
1472 The fastest way to decode is to simply build a lookup table whose
1473 size is determined by the longest code. However, the time it takes
1474 to build this table can also be a factor if the data being decoded
1475 is not very long. The most common codes are necessarily the
1476 shortest codes, so those codes dominate the decoding time, and hence
1477 the speed. The idea is you can have a shorter table that decodes the
1478 shorter, more probable codes, and then point to subsidiary tables for
1479 the longer codes. The time it costs to decode the longer codes is
1480 then traded against the time it takes to make longer tables.
1482 This results of this trade are in the variables lbits and dbits
1483 below. lbits is the number of bits the first level table for literal/
1484 length codes can decode in one step, and dbits is the same thing for
1485 the distance codes. Subsequent tables are also less than or equal to
1486 those sizes. These values may be adjusted either when all of the
1487 codes are shorter than that, in which case the longest code length in
1488 bits is used, or when the shortest code is *longer* than the requested
1489 table size, in which case the length of the shortest code in bits is
1492 There are two different values for the two tables, since they code a
1493 different number of possibilities each. The literal/length table
1494 codes 286 possible values, or in a flat code, a little over eight
1495 bits. The distance table codes 30 possible values, or a little less
1496 than five bits, flat. The optimum values for speed end up being
1497 about one bit more than those, so lbits is 8+1 and dbits is 5+1.
1498 The optimum values may differ though from machine to machine, and
1499 possibly even between compilers. Your mileage may vary.
1503 int lbits = 9; /* bits in base literal/length lookup table */
1504 int dbits = 6; /* bits in base distance lookup table */
1507 /* If BMAX needs to be larger than 16, then h and x[] should be ulg. */
1508 #define BMAX 16 /* maximum bit length of any code (16 for explode) */
1509 #define N_MAX 288 /* maximum number of codes in any set */
1512 unsigned hufts; /* track memory usage */
1515 int huft_build(b, n, s, d, e, t, m)
1516 unsigned *b; /* code lengths in bits (all assumed <= BMAX) */
1517 unsigned n; /* number of codes (assumed <= N_MAX) */
1518 unsigned s; /* number of simple-valued codes (0..s-1) */
1519 ush *d; /* list of base values for non-simple codes */
1520 ush *e; /* list of extra bits for non-simple codes */
1521 struct huft **t; /* result: starting table */
1522 int *m; /* maximum lookup bits, returns actual */
1524 /* Given a list of code lengths and a maximum table size, make a set of
1525 tables to decode that set of codes. Return zero on success, one if
1526 the given code set is incomplete (the tables are still built in this
1527 case), two if the input is invalid (all zero length codes or an
1528 oversubscribed set of lengths), and three if not enough memory. */
1530 unsigned a; /* counter for codes of length k */
1531 unsigned c[BMAX + 1]; /* bit length count table */
1532 unsigned f; /* i repeats in table every f entries */
1533 int g; /* maximum code length */
1534 int h; /* table level */
1535 register unsigned i; /* counter, current code */
1536 register unsigned j; /* counter */
1537 register int k; /* number of bits in current code */
1538 int l; /* bits per table (returned in m) */
1539 register unsigned *p; /* pointer into c[], b[], or v[] */
1540 register struct huft *q; /* points to current table */
1541 struct huft r; /* table entry for structure assignment */
1542 struct huft *u[BMAX]; /* table stack */
1543 unsigned v[N_MAX]; /* values in order of bit length */
1544 register int w; /* bits before this table == (l * h) */
1545 unsigned x[BMAX + 1]; /* bit offsets, then code stack */
1546 unsigned *xp; /* pointer into x */
1547 int y; /* number of dummy codes added */
1548 unsigned z; /* number of entries in current table */
1551 /* Generate counts for each bit length */
1552 memzero(c, sizeof(c));
1559 && n - i <= '~' ? "%c %d\n" : "0x%x %d\n"), n - i, *p));
1560 c[*p]++; /* assume all entries <= BMAX */
1561 p++; /* Can't combine with above line (Solaris bug) */
1563 if (c[0] == n) { /* null input--all zero length codes */
1564 *t = (struct huft *) NULL;
1570 /* Find minimum and maximum length, bound *m by those */
1572 for (j = 1; j <= BMAX; j++)
1575 k = j; /* minimum code length */
1576 if ((unsigned) l < j)
1578 for (i = BMAX; i; i--)
1581 g = i; /* maximum code length */
1582 if ((unsigned) l > i)
1587 /* Adjust last length count to fill out codes, if needed */
1588 for (y = 1 << j; j < i; j++, y <<= 1)
1589 if ((y -= c[j]) < 0)
1590 return 2; /* bad input: more codes than bits */
1591 if ((y -= c[i]) < 0)
1596 /* Generate starting offsets into the value table for each length */
1600 while (--i) { /* note that i == g from above */
1601 *xp++ = (j += *p++);
1605 /* Make a table of values in order of bit lengths */
1609 if ((j = *p++) != 0)
1614 /* Generate the Huffman codes and for each, make the table entries */
1615 x[0] = i = 0; /* first Huffman code is zero */
1616 p = v; /* grab values in bit order */
1617 h = -1; /* no tables yet--level -1 */
1618 w = -l; /* bits decoded == (l * h) */
1619 u[0] = (struct huft *) NULL; /* just to keep compilers happy */
1620 q = (struct huft *) NULL; /* ditto */
1623 /* go through the bit lengths (k already is bits in shortest code) */
1624 for (; k <= g; k++) {
1627 /* here i is the Huffman code of length k bits for value *p */
1628 /* make tables up to required level */
1631 w += l; /* previous table always l bits */
1633 /* compute minimum size table less than or equal to l bits */
1634 z = (z = g - w) > (unsigned) l ? l : z; /* upper limit on table size */
1635 if ((f = 1 << (j = k - w)) > a + 1) { /* try a k-w bit table *//* too few codes for k-w bit table */
1636 f -= a + 1; /* deduct codes from patterns left */
1638 while (++j < z) { /* try smaller tables up to z bits */
1639 if ((f <<= 1) <= *++xp)
1640 break; /* enough codes to use up j bits */
1641 f -= *xp; /* else deduct codes from patterns */
1644 z = 1 << j; /* table entries for j-bit table */
1646 /* allocate and link in new table */
1649 (struct huft *) malloc((z + 1) *
1650 sizeof(struct huft))) ==
1651 (struct huft *) NULL) {
1654 return 3; /* not enough memory */
1656 hufts += z + 1; /* track memory usage */
1657 *t = q + 1; /* link to list for huft_free() */
1658 *(t = &(q->v.t)) = (struct huft *) NULL;
1659 u[h] = ++q; /* table starts after link */
1661 /* connect to last table, if there is one */
1663 x[h] = i; /* save pattern for backing up */
1664 r.b = (uch) l; /* bits to dump before this table */
1665 r.e = (uch) (16 + j); /* bits in this table */
1666 r.v.t = q; /* pointer to this table */
1667 j = i >> (w - l); /* (get around Turbo C bug) */
1668 u[h - 1][j] = r; /* connect to last table */
1672 /* set up table entry in r */
1673 r.b = (uch) (k - w);
1675 r.e = 99; /* out of values--invalid code */
1677 r.e = (uch) (*p < 256 ? 16 : 15); /* 256 is end-of-block code */
1678 r.v.n = (ush) (*p); /* simple code is just the value */
1679 p++; /* one compiler does not like *p++ */
1681 r.e = (uch) e[*p - s]; /* non-simple--look up in lists */
1682 r.v.n = d[*p++ - s];
1685 /* fill code-like entries with r */
1687 for (j = i >> w; j < z; j += f)
1690 /* backwards increment the k-bit code i */
1691 for (j = 1 << (k - 1); i & j; j >>= 1)
1695 /* backup over finished tables */
1696 while ((i & ((1 << w) - 1)) != x[h]) {
1697 h--; /* don't need to update q */
1704 /* Return true (1) if we were given an incomplete table */
1705 return y != 0 && g != 1;
1711 struct huft *t; /* table to free */
1713 /* Free the malloc'ed tables built by huft_build(), which makes a linked
1714 list of the tables it made, with the links in a dummy first entry of
1717 register struct huft *p, *q;
1720 /* Go through linked list, freeing from the malloced (t[-1]) address. */
1722 while (p != (struct huft *) NULL) {
1731 int inflate_codes(tl, td, bl, bd)
1732 struct huft *tl, *td; /* literal/length and distance decoder tables */
1733 int bl, bd; /* number of bits decoded by tl[] and td[] */
1735 /* inflate (decompress) the codes in a deflated (compressed) block.
1736 Return an error code or zero if it all goes ok. */
1738 register unsigned e; /* table entry flag/number of extra bits */
1739 unsigned n, d; /* length and index for copy */
1740 unsigned w; /* current window position */
1741 struct huft *t; /* pointer to table entry */
1742 unsigned ml, md; /* masks for bl and bd bits */
1743 register ulg b; /* bit buffer */
1744 register unsigned k; /* number of bits in bit buffer */
1747 /* make local copies of globals */
1748 b = bb; /* initialize bit buffer */
1750 w = wp; /* initialize window position */
1752 /* inflate the coded data */
1753 ml = mask_bits[bl]; /* precompute masks for speed */
1755 for (;;) { /* do until end of block */
1756 NEEDBITS((unsigned) bl)
1757 if ((e = (t = tl + ((unsigned) b & ml))->e) > 16)
1764 } while ((e = (t = t->v.t + ((unsigned) b & mask_bits[e]))->e)
1767 if (e == 16) { /* then it's a literal */
1768 slide[w++] = (uch) t->v.n;
1769 Tracevv((stderr, "%c", slide[w - 1]));
1774 } else { /* it's an EOB or a length */
1776 /* exit if end of block */
1780 /* get length of block to copy */
1782 n = t->v.n + ((unsigned) b & mask_bits[e]);
1785 /* decode distance of block to copy */
1786 NEEDBITS((unsigned) bd)
1787 if ((e = (t = td + ((unsigned) b & md))->e) > 16)
1798 t->v.t + ((unsigned) b & mask_bits[e]))->e) >
1802 d = w - t->v.n - ((unsigned) b & mask_bits[e]);
1804 Tracevv((stderr, "\\[%d,%d]", w - d, n));
1810 WSIZE - ((d &= WSIZE - 1) > w ? d : w)) >
1812 #if !defined(NOMEMCPY) && !defined(DEBUG)
1813 if (w - d >= e) { /* (this test assumes unsigned comparison) */
1814 memcpy(slide + w, slide + d, e);
1817 } else /* do it slow to avoid memcpy() overlap */
1818 #endif /* !NOMEMCPY */
1820 slide[w++] = slide[d++];
1821 Tracevv((stderr, "%c", slide[w - 1]));
1832 /* restore the globals from the locals */
1833 wp = w; /* restore global window pointer */
1834 bb = b; /* restore global bit buffer */
1843 int inflate_stored()
1844 /* "decompress" an inflated type 0 (stored) block. */
1846 unsigned n; /* number of bytes in block */
1847 unsigned w; /* current window position */
1848 register ulg b; /* bit buffer */
1849 register unsigned k; /* number of bits in bit buffer */
1852 /* make local copies of globals */
1853 b = bb; /* initialize bit buffer */
1855 w = wp; /* initialize window position */
1858 /* go to byte boundary */
1863 /* get the length and its complement */
1865 n = ((unsigned) b & 0xffff);
1868 if (n != (unsigned) ((~b) & 0xffff))
1869 return 1; /* error in compressed data */
1873 /* read and output the compressed data */
1876 slide[w++] = (uch) b;
1885 /* restore the globals from the locals */
1886 wp = w; /* restore global window pointer */
1887 bb = b; /* restore global bit buffer */
1895 /* decompress an inflated type 1 (fixed Huffman codes) block. We should
1896 either replace this with a custom decoder, or at least precompute the
1899 int i; /* temporary variable */
1900 struct huft *tl; /* literal/length code table */
1901 struct huft *td; /* distance code table */
1902 int bl; /* lookup bits for tl */
1903 int bd; /* lookup bits for td */
1904 unsigned l[288]; /* length list for huft_build */
1907 /* set up literal table */
1908 for (i = 0; i < 144; i++)
1910 for (; i < 256; i++)
1912 for (; i < 280; i++)
1914 for (; i < 288; i++) /* make a complete, but wrong code set */
1917 if ((i = huft_build(l, 288, 257, cplens, cplext, &tl, &bl)) != 0)
1921 /* set up distance table */
1922 for (i = 0; i < 30; i++) /* make an incomplete code set */
1925 if ((i = huft_build(l, 30, 0, cpdist, cpdext, &td, &bd)) > 1) {
1931 /* decompress until an end-of-block code */
1932 if (inflate_codes(tl, td, bl, bd))
1936 /* free the decoding tables, return */
1944 int inflate_dynamic()
1945 /* decompress an inflated type 2 (dynamic Huffman codes) block. */
1947 int i; /* temporary variables */
1949 unsigned l; /* last length */
1950 unsigned m; /* mask for bit lengths table */
1951 unsigned n; /* number of lengths to get */
1952 struct huft *tl; /* literal/length code table */
1953 struct huft *td; /* distance code table */
1954 int bl; /* lookup bits for tl */
1955 int bd; /* lookup bits for td */
1956 unsigned nb; /* number of bit length codes */
1957 unsigned nl; /* number of literal/length codes */
1958 unsigned nd; /* number of distance codes */
1960 #ifdef PKZIP_BUG_WORKAROUND
1961 unsigned ll[288 + 32]; /* literal/length and distance code lengths */
1963 unsigned ll[286 + 30]; /* literal/length and distance code lengths */
1965 register ulg b; /* bit buffer */
1966 register unsigned k; /* number of bits in bit buffer */
1969 /* make local bit buffer */
1974 /* read in table lengths */
1976 nl = 257 + ((unsigned) b & 0x1f); /* number of literal/length codes */
1979 nd = 1 + ((unsigned) b & 0x1f); /* number of distance codes */
1982 nb = 4 + ((unsigned) b & 0xf); /* number of bit length codes */
1984 #ifdef PKZIP_BUG_WORKAROUND
1985 if (nl > 288 || nd > 32)
1987 if (nl > 286 || nd > 30)
1989 return 1; /* bad lengths */
1992 /* read in bit-length-code lengths */
1993 for (j = 0; j < nb; j++) {
1995 ll[border[j]] = (unsigned) b & 7;
2002 /* build decoding table for trees--single level, 7 bit lookup */
2004 if ((i = huft_build(ll, 19, 19, NULL, NULL, &tl, &bl)) != 0) {
2007 return i; /* incomplete code set */
2011 /* read in literal and distance code lengths */
2015 while ((unsigned) i < n) {
2016 NEEDBITS((unsigned) bl)
2017 j = (td = tl + ((unsigned) b & m))->b;
2020 if (j < 16) /* length of code in bits (0..15) */
2021 ll[i++] = l = j; /* save last length in l */
2022 else if (j == 16) { /* repeat last length 3 to 6 times */
2024 j = 3 + ((unsigned) b & 3);
2026 if ((unsigned) i + j > n)
2030 } else if (j == 17) { /* 3 to 10 zero length codes */
2032 j = 3 + ((unsigned) b & 7);
2034 if ((unsigned) i + j > n)
2039 } else { /* j == 18: 11 to 138 zero length codes */
2042 j = 11 + ((unsigned) b & 0x7f);
2044 if ((unsigned) i + j > n)
2053 /* free decoding table for trees */
2057 /* restore the global bit buffer */
2062 /* build the decoding tables for literal/length and distance codes */
2064 if ((i = huft_build(ll, nl, 257, cplens, cplext, &tl, &bl)) != 0) {
2066 fprintf(stderr, " incomplete literal tree\n");
2069 return i; /* incomplete code set */
2072 if ((i = huft_build(ll + nl, nd, 0, cpdist, cpdext, &td, &bd)) != 0) {
2074 fprintf(stderr, " incomplete distance tree\n");
2075 #ifdef PKZIP_BUG_WORKAROUND
2082 return i; /* incomplete code set */
2087 /* decompress until an end-of-block code */
2088 if (inflate_codes(tl, td, bl, bd))
2092 /* free the decoding tables, return */
2100 int inflate_block(e)
2101 int *e; /* last block flag */
2103 /* decompress an inflated block */
2105 unsigned t; /* block type */
2106 register ulg b; /* bit buffer */
2107 register unsigned k; /* number of bits in bit buffer */
2110 /* make local bit buffer */
2115 /* read in last block bit */
2121 /* read in block type */
2123 t = (unsigned) b & 3;
2127 /* restore the global bit buffer */
2132 /* inflate that block type */
2134 return inflate_dynamic();
2136 return inflate_stored();
2138 return inflate_fixed();
2141 /* bad block type */
2148 /* decompress an inflated entry */
2150 int e; /* last block flag */
2151 int r; /* result code */
2152 unsigned h; /* maximum struct huft's malloc'ed */
2155 /* initialize window, bit buffer */
2161 /* decompress until the last block */
2165 if ((r = inflate_block(&e)) != 0)
2171 /* Undo too much lookahead. The next read will be byte aligned so we
2172 * can discard unused bits in the last meaningful byte.
2179 /* flush out slide */
2183 /* return success */
2185 fprintf(stderr, "<%u> ", h);