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 Charles P. Wright <cpw@unix.asb.com>
8 * "this is a stripped down version of gzip I put into busybox, it does
9 * only standard in to standard out with -9 compression. It also requires
10 * the zcat module for some important functions."
12 * Adjusted further by Erik Andersen <andersen@lineo.com>, <andersee@debian.org>
13 * to support files as well as stdin/stdout, and to generally behave itself wrt
14 * command line handling.
16 * This program is free software; you can redistribute it and/or modify
17 * it under the terms of the GNU General Public License as published by
18 * the Free Software Foundation; either version 2 of the License, or
19 * (at your option) any later version.
21 * This program is distributed in the hope that it will be useful,
22 * but WITHOUT ANY WARRANTY; without even the implied warranty of
23 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
24 * General Public License for more details.
26 * You should have received a copy of the GNU General Public License
27 * along with this program; if not, write to the Free Software
28 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
32 /* These defines are very important for BusyBox. Without these,
33 * huge chunks of ram are pre-allocated making the BusyBox bss
34 * size Freaking Huge(tm), which is a bad thing.*/
43 #include <sys/types.h>
47 #include <sys/types.h>
54 #define memzero(s, n) memset ((void *)(s), 0, (n))
57 # define RETSIGTYPE void
60 typedef unsigned char uch;
61 typedef unsigned short ush;
62 typedef unsigned long ulg;
64 /* Return codes from gzip */
69 /* Compression methods (see algorithm.doc) */
70 /* Only STORED and DEFLATED are supported by this BusyBox module */
72 /* methods 4 to 7 reserved */
74 static int method; /* compression method */
76 /* To save memory for 16 bit systems, some arrays are overlaid between
77 * the various modules:
78 * deflate: prev+head window d_buf l_buf outbuf
79 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
80 * For compression, input is done in window[]. For decompression, output
81 * is done in window except for unlzw.
86 # define INBUFSIZ 0x2000 /* input buffer size */
88 # define INBUFSIZ 0x8000 /* input buffer size */
91 #define INBUF_EXTRA 64 /* required by unlzw() */
95 # define OUTBUFSIZ 8192 /* output buffer size */
97 # define OUTBUFSIZ 16384 /* output buffer size */
100 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
104 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
106 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
111 # define DECLARE(type, array, size) static type * array
112 # define ALLOC(type, array, size) { \
113 array = (type*)xcalloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \
115 # define FREE(array) {if (array != NULL) free(array), array=NULL;}
117 # define DECLARE(type, array, size) static type array[size]
118 # define ALLOC(type, array, size)
122 #define tab_suffix window
123 #define tab_prefix prev /* hash link (see deflate.c) */
124 #define head (prev+WSIZE) /* hash head (see deflate.c) */
126 static long bytes_in; /* number of input bytes */
128 #define isize bytes_in
129 /* for compatibility with old zip sources (to be cleaned) */
131 typedef int file_t; /* Do not use stdio */
133 #define NO_FILE (-1) /* in memory compression */
136 #define PACK_MAGIC "\037\036" /* Magic header for packed files */
137 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
138 #define OLD_GZIP_MAGIC "\037\236" /* Magic header for gzip 0.5 = freeze 1.x */
139 #define LZH_MAGIC "\037\240" /* Magic header for SCO LZH Compress files */
140 #define PKZIP_MAGIC "\120\113\003\004" /* Magic header for pkzip files */
143 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
144 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
145 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
146 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
147 #define COMMENT 0x10 /* bit 4 set: file comment present */
148 #define RESERVED 0xC0 /* bit 6,7: reserved */
150 /* internal file attribute */
151 #define UNKNOWN 0xffff
156 # define WSIZE 0x8000 /* window size--must be a power of two, and */
157 #endif /* at least 32K for zip's deflate method */
160 #define MAX_MATCH 258
161 /* The minimum and maximum match lengths */
163 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
164 /* Minimum amount of lookahead, except at the end of the input file.
165 * See deflate.c for comments about the MIN_MATCH+1.
168 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
169 /* In order to simplify the code, particularly on 16 bit machines, match
170 * distances are limited to MAX_DIST instead of WSIZE.
173 /* put_byte is used for the compressed output */
174 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
178 /* Output a 32 bit value to the bit stream, lsb first */
180 #define put_long(n) { \
181 put_short((n) & 0xffff); \
182 put_short(((ulg)(n)) >> 16); \
186 #define seekable() 0 /* force sequential output */
187 #define translate_eol 0 /* no option -a yet */
189 /* Diagnostic functions */
191 # define Assert(cond,msg) {if(!(cond)) error_msg(msg);}
192 # define Trace(x) fprintf x
193 # define Tracev(x) {if (verbose) fprintf x ;}
194 # define Tracevv(x) {if (verbose>1) fprintf x ;}
195 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
196 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
198 # define Assert(cond,msg)
203 # define Tracecv(c,x)
206 #define WARN(msg) {if (!quiet) fprintf msg ; \
207 if (exit_code == OK) exit_code = WARNING;}
210 # define MAX_PATH_LEN 1024 /* max pathname length */
215 static int zip (int in, int out);
216 static int file_read (char *buf, unsigned size);
219 static RETSIGTYPE abort_gzip (void);
222 static void lm_init (ush * flags);
223 static ulg deflate (void);
226 static void ct_init (ush * attr, int *methodp);
227 static int ct_tally (int dist, int lc);
228 static ulg flush_block (char *buf, ulg stored_len, int eof);
231 static void bi_init (file_t zipfile);
232 static void send_bits (int value, int length);
233 static unsigned bi_reverse (unsigned value, int length);
234 static void bi_windup (void);
235 static void copy_block (char *buf, unsigned len, int header);
236 static int (*read_buf) (char *buf, unsigned size);
239 static void flush_outbuf (void);
241 /* lzw.h -- define the lzw functions.
242 * Copyright (C) 1992-1993 Jean-loup Gailly.
243 * This is free software; you can redistribute it and/or modify it under the
244 * terms of the GNU General Public License, see the file COPYING.
247 #if !defined(OF) && defined(lint)
254 #define INIT_BITS 9 /* Initial number of bits per code */
256 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
257 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
258 * It's a pity that old uncompress does not check bit 0x20. That makes
259 * extension of the format actually undesirable because old compress
260 * would just crash on the new format instead of giving a meaningful
261 * error message. It does check the number of bits, but it's more
262 * helpful to say "unsupported format, get a new version" than
263 * "can only handle 16 bits".
266 /* tailor.h -- target dependent definitions
267 * Copyright (C) 1992-1993 Jean-loup Gailly.
268 * This is free software; you can redistribute it and/or modify it under the
269 * terms of the GNU General Public License, see the file COPYING.
272 /* The target dependent definitions should be defined here only.
273 * The target dependent functions should be defined in tailor.c.
277 /* Common defaults */
280 # define OS_CODE 0x03 /* assume Unix */
284 # define PATH_SEP '/'
288 # define OPTIONS_VAR "GZIP"
292 # define Z_SUFFIX ".gz"
296 # define MAX_SUFFIX MAX_EXT_CHARS
298 # define MAX_SUFFIX 30
303 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
304 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
305 DECLARE(ush, d_buf, DIST_BUFSIZE);
306 DECLARE(uch, window, 2L * WSIZE);
307 DECLARE(ush, tab_prefix, 1L << BITS);
309 static int crc_table_empty = 1;
311 static int foreground; /* set if program run in foreground */
312 static int method = DEFLATED; /* compression method */
313 static int exit_code = OK; /* program exit code */
314 static int part_nb; /* number of parts in .gz file */
315 static long time_stamp; /* original time stamp (modification time) */
316 static long ifile_size; /* input file size, -1 for devices (debug only) */
317 static char z_suffix[MAX_SUFFIX + 1]; /* default suffix (can be set with --suffix) */
318 static int z_len; /* strlen(z_suffix) */
320 static int ifd; /* input file descriptor */
321 static int ofd; /* output file descriptor */
322 static unsigned insize; /* valid bytes in inbuf */
323 static unsigned outcnt; /* bytes in output buffer */
326 /* Output a 16 bit value, lsb first */
327 static void put_short(ush w)
329 if (outcnt < OUTBUFSIZ-2) {
330 outbuf[outcnt++] = (uch) ((w) & 0xff);
331 outbuf[outcnt++] = (uch) ((ush)(w) >> 8);
333 put_byte((uch)((w) & 0xff));
334 put_byte((uch)((ush)(w) >> 8));
338 /* ========================================================================
339 * Signal and error handler.
341 static void abort_gzip()
346 /* ===========================================================================
347 * Clear input and output buffers
349 static void clear_bufs(void)
356 static void write_error_msg(void)
358 fprintf(stderr, "\n");
363 /* ===========================================================================
364 * Does the same as write(), but also handles partial pipe writes and checks
367 static void write_buf(int fd, void *buf, unsigned cnt)
371 while ((n = write(fd, buf, cnt)) != cnt) {
372 if (n == (unsigned) (-1)) {
376 buf = (void *) ((char *) buf + n);
380 /* ===========================================================================
381 * Run a set of bytes through the crc shift register. If s is a NULL
382 * pointer, then initialize the crc shift register contents instead.
383 * Return the current crc in either case.
385 static ulg updcrc(uch *s, unsigned n)
387 static ulg crc = (ulg) 0xffffffffL; /* shift register contents */
388 register ulg c; /* temporary variable */
389 static unsigned long crc_32_tab[256];
390 if (crc_table_empty) {
391 unsigned long csr; /* crc shift register */
392 const unsigned long e = 0xedb88320L; /* polynomial exclusive-or pattern */
393 int i; /* counter for all possible eight bit values */
394 int k; /* byte being shifted into crc apparatus */
396 /* Compute table of CRC's. */
397 crc_32_tab[0] = 0x00000000L;
398 for (i = 1; i < 256; i++) {
400 /* The idea to initialize the register with the byte instead of
401 * zero was stolen from Haruhiko Okumura's ar002
404 csr = csr & 1 ? (csr >> 1) ^ e : csr >> 1;
415 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
419 return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */
422 /* bits.c -- output variable-length bit strings
423 * Copyright (C) 1992-1993 Jean-loup Gailly
424 * This is free software; you can redistribute it and/or modify it under the
425 * terms of the GNU General Public License, see the file COPYING.
432 * Output variable-length bit strings. Compression can be done
433 * to a file or to memory. (The latter is not supported in this version.)
437 * The PKZIP "deflate" file format interprets compressed file data
438 * as a sequence of bits. Multi-bit strings in the file may cross
439 * byte boundaries without restriction.
441 * The first bit of each byte is the low-order bit.
443 * The routines in this file allow a variable-length bit value to
444 * be output right-to-left (useful for literal values). For
445 * left-to-right output (useful for code strings from the tree routines),
446 * the bits must have been reversed first with bi_reverse().
448 * For in-memory compression, the compressed bit stream goes directly
449 * into the requested output buffer. The input data is read in blocks
450 * by the mem_read() function. The buffer is limited to 64K on 16 bit
455 * void bi_init (FILE *zipfile)
456 * Initialize the bit string routines.
458 * void send_bits (int value, int length)
459 * Write out a bit string, taking the source bits right to
462 * int bi_reverse (int value, int length)
463 * Reverse the bits of a bit string, taking the source bits left to
464 * right and emitting them right to left.
466 * void bi_windup (void)
467 * Write out any remaining bits in an incomplete byte.
469 * void copy_block(char *buf, unsigned len, int header)
470 * Copy a stored block to the zip file, storing first the length and
471 * its one's complement if requested.
475 /* ===========================================================================
476 * Local data used by the "bit string" routines.
479 static file_t zfile; /* output gzip file */
481 static unsigned short bi_buf;
483 /* Output buffer. bits are inserted starting at the bottom (least significant
487 #define Buf_size (8 * 2*sizeof(char))
488 /* Number of bits used within bi_buf. (bi_buf might be implemented on
489 * more than 16 bits on some systems.)
494 /* Current input function. Set to mem_read for in-memory compression */
497 ulg bits_sent; /* bit length of the compressed data */
500 /* ===========================================================================
501 * Initialize the bit string routines.
503 static void bi_init(file_t zipfile)
512 /* Set the defaults for file compression. They are set by memcompress
513 * for in-memory compression.
515 if (zfile != NO_FILE) {
516 read_buf = file_read;
520 /* ===========================================================================
521 * Send a value on a given number of bits.
522 * IN assertion: length <= 16 and value fits in length bits.
524 static void send_bits(int value, int length)
527 Tracev((stderr, " l %2d v %4x ", length, value));
528 Assert(length > 0 && length <= 15, "invalid length");
529 bits_sent += (ulg) length;
531 /* If not enough room in bi_buf, use (valid) bits from bi_buf and
532 * (16 - bi_valid) bits from value, leaving (width - (16-bi_valid))
533 * unused bits in value.
535 if (bi_valid > (int) Buf_size - length) {
536 bi_buf |= (value << bi_valid);
538 bi_buf = (ush) value >> (Buf_size - bi_valid);
539 bi_valid += length - Buf_size;
541 bi_buf |= value << bi_valid;
546 /* ===========================================================================
547 * Reverse the first len bits of a code, using straightforward code (a faster
548 * method would use a table)
549 * IN assertion: 1 <= len <= 15
551 static unsigned bi_reverse(unsigned code, int len)
553 register unsigned res = 0;
557 code >>= 1, res <<= 1;
562 /* ===========================================================================
563 * Write out any remaining bits in an incomplete byte.
565 static void bi_windup()
569 } else if (bi_valid > 0) {
575 bits_sent = (bits_sent + 7) & ~7;
579 /* ===========================================================================
580 * Copy a stored block to the zip file, storing first the length and its
581 * one's complement if requested.
583 static void copy_block(char *buf, unsigned len, int header)
585 bi_windup(); /* align on byte boundary */
588 put_short((ush) len);
589 put_short((ush) ~ len);
595 bits_sent += (ulg) len << 3;
602 /* deflate.c -- compress data using the deflation algorithm
603 * Copyright (C) 1992-1993 Jean-loup Gailly
604 * This is free software; you can redistribute it and/or modify it under the
605 * terms of the GNU General Public License, see the file COPYING.
611 * Identify new text as repetitions of old text within a fixed-
612 * length sliding window trailing behind the new text.
616 * The "deflation" process depends on being able to identify portions
617 * of the input text which are identical to earlier input (within a
618 * sliding window trailing behind the input currently being processed).
620 * The most straightforward technique turns out to be the fastest for
621 * most input files: try all possible matches and select the longest.
622 * The key feature of this algorithm is that insertions into the string
623 * dictionary are very simple and thus fast, and deletions are avoided
624 * completely. Insertions are performed at each input character, whereas
625 * string matches are performed only when the previous match ends. So it
626 * is preferable to spend more time in matches to allow very fast string
627 * insertions and avoid deletions. The matching algorithm for small
628 * strings is inspired from that of Rabin & Karp. A brute force approach
629 * is used to find longer strings when a small match has been found.
630 * A similar algorithm is used in comic (by Jan-Mark Wams) and freeze
631 * (by Leonid Broukhis).
632 * A previous version of this file used a more sophisticated algorithm
633 * (by Fiala and Greene) which is guaranteed to run in linear amortized
634 * time, but has a larger average cost, uses more memory and is patented.
635 * However the F&G algorithm may be faster for some highly redundant
636 * files if the parameter max_chain_length (described below) is too large.
640 * The idea of lazy evaluation of matches is due to Jan-Mark Wams, and
641 * I found it in 'freeze' written by Leonid Broukhis.
642 * Thanks to many info-zippers for bug reports and testing.
646 * APPNOTE.TXT documentation file in PKZIP 1.93a distribution.
648 * A description of the Rabin and Karp algorithm is given in the book
649 * "Algorithms" by R. Sedgewick, Addison-Wesley, p252.
651 * Fiala,E.R., and Greene,D.H.
652 * Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595
656 * void lm_init (int pack_level, ush *flags)
657 * Initialize the "longest match" routines for a new file
660 * Processes a new input file and return its compressed length. Sets
661 * the compressed length, crc, deflate flags and internal file
666 /* ===========================================================================
667 * Configuration parameters
670 /* Compile with MEDIUM_MEM to reduce the memory requirements or
671 * with SMALL_MEM to use as little memory as possible. Use BIG_MEM if the
672 * entire input file can be held in memory (not possible on 16 bit systems).
673 * Warning: defining these symbols affects HASH_BITS (see below) and thus
674 * affects the compression ratio. The compressed output
675 * is still correct, and might even be smaller in some cases.
679 # define HASH_BITS 13 /* Number of bits used to hash strings */
682 # define HASH_BITS 14
685 # define HASH_BITS 15
686 /* For portability to 16 bit machines, do not use values above 15. */
689 /* To save space (see unlzw.c), we overlay prev+head with tab_prefix and
690 * window with tab_suffix. Check that we can do this:
692 #if (WSIZE<<1) > (1<<BITS)
693 # error cannot overlay window with tab_suffix and prev with tab_prefix0
695 #if HASH_BITS > BITS-1
696 # error cannot overlay head with tab_prefix1
698 #define HASH_SIZE (unsigned)(1<<HASH_BITS)
699 #define HASH_MASK (HASH_SIZE-1)
700 #define WMASK (WSIZE-1)
701 /* HASH_SIZE and WSIZE must be powers of two */
703 /* Tail of hash chains */
706 /* speed options for the general purpose bit flag */
708 # define TOO_FAR 4096
710 /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */
711 /* ===========================================================================
712 * Local data used by the "longest match" routines.
715 typedef unsigned IPos;
717 /* A Pos is an index in the character window. We use short instead of int to
718 * save space in the various tables. IPos is used only for parameter passing.
721 /* DECLARE(uch, window, 2L*WSIZE); */
722 /* Sliding window. Input bytes are read into the second half of the window,
723 * and move to the first half later to keep a dictionary of at least WSIZE
724 * bytes. With this organization, matches are limited to a distance of
725 * WSIZE-MAX_MATCH bytes, but this ensures that IO is always
726 * performed with a length multiple of the block size. Also, it limits
727 * the window size to 64K, which is quite useful on MSDOS.
728 * To do: limit the window size to WSIZE+BSZ if SMALL_MEM (the code would
729 * be less efficient).
732 /* DECLARE(Pos, prev, WSIZE); */
733 /* Link to older string with same hash index. To limit the size of this
734 * array to 64K, this link is maintained only for the last 32K strings.
735 * An index in this array is thus a window index modulo 32K.
738 /* DECLARE(Pos, head, 1<<HASH_BITS); */
739 /* Heads of the hash chains or NIL. */
741 static const ulg window_size = (ulg) 2 * WSIZE;
743 /* window size, 2*WSIZE except for MMAP or BIG_MEM, where it is the
744 * input file length plus MIN_LOOKAHEAD.
747 static long block_start;
749 /* window position at the beginning of the current output block. Gets
750 * negative when the window is moved backwards.
753 static unsigned ins_h; /* hash index of string to be inserted */
755 #define H_SHIFT ((HASH_BITS+MIN_MATCH-1)/MIN_MATCH)
756 /* Number of bits by which ins_h and del_h must be shifted at each
757 * input step. It must be such that after MIN_MATCH steps, the oldest
758 * byte no longer takes part in the hash key, that is:
759 * H_SHIFT * MIN_MATCH >= HASH_BITS
762 static unsigned int prev_length;
764 /* Length of the best match at previous step. Matches not greater than this
765 * are discarded. This is used in the lazy match evaluation.
768 static unsigned strstart; /* start of string to insert */
769 static unsigned match_start; /* start of matching string */
770 static int eofile; /* flag set at end of input file */
771 static unsigned lookahead; /* number of valid bytes ahead in window */
773 static const unsigned max_chain_length=4096;
775 /* To speed up deflation, hash chains are never searched beyond this length.
776 * A higher limit improves compression ratio but degrades the speed.
779 static const unsigned int max_lazy_match=258;
781 /* Attempt to find a better match only when the current match is strictly
782 * smaller than this value. This mechanism is used only for compression
785 #define max_insert_length max_lazy_match
786 /* Insert new strings in the hash table only if the match length
787 * is not greater than this length. This saves time but degrades compression.
788 * max_insert_length is used only for compression levels <= 3.
791 static const unsigned good_match=32;
793 /* Use a faster search when the previous match is longer than this */
796 /* Values for max_lazy_match, good_match and max_chain_length, depending on
797 * the desired pack level (0..9). The values given below have been tuned to
798 * exclude worst case performance for pathological files. Better values may be
799 * found for specific files.
802 static const int nice_match=258; /* Stop searching when current match exceeds this */
804 /* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4
805 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different
810 /* result of memcmp for equal strings */
812 /* ===========================================================================
813 * Prototypes for local functions.
815 static void fill_window (void);
817 static int longest_match (IPos cur_match);
820 static void check_match (IPos start, IPos match, int length);
823 /* ===========================================================================
824 * Update a hash value with the given input byte
825 * IN assertion: all calls to to UPDATE_HASH are made with consecutive
826 * input characters, so that a running hash key can be computed from the
827 * previous key instead of complete recalculation each time.
829 #define UPDATE_HASH(h,c) (h = (((h)<<H_SHIFT) ^ (c)) & HASH_MASK)
831 /* ===========================================================================
832 * Insert string s in the dictionary and set match_head to the previous head
833 * of the hash chain (the most recent string with same hash key). Return
834 * the previous length of the hash chain.
835 * IN assertion: all calls to to INSERT_STRING are made with consecutive
836 * input characters and the first MIN_MATCH bytes of s are valid
837 * (except for the last MIN_MATCH-1 bytes of the input file).
839 #define INSERT_STRING(s, match_head) \
840 (UPDATE_HASH(ins_h, window[(s) + MIN_MATCH-1]), \
841 prev[(s) & WMASK] = match_head = head[ins_h], \
844 /* ===========================================================================
845 * Initialize the "longest match" routines for a new file
847 static void lm_init(ush *flags)
851 /* Initialize the hash table. */
852 memzero((char *) head, HASH_SIZE * sizeof(*head));
853 /* prev will be initialized on the fly */
856 /* ??? reduce max_chain_length for binary files */
861 lookahead = read_buf((char *) window,
862 sizeof(int) <= 2 ? (unsigned) WSIZE : 2 * WSIZE);
864 if (lookahead == 0 || lookahead == (unsigned) EOF) {
865 eofile = 1, lookahead = 0;
869 /* Make sure that we always have enough lookahead. This is important
870 * if input comes from a device such as a tty.
872 while (lookahead < MIN_LOOKAHEAD && !eofile)
876 for (j = 0; j < MIN_MATCH - 1; j++)
877 UPDATE_HASH(ins_h, window[j]);
878 /* If lookahead < MIN_MATCH, ins_h is garbage, but this is
879 * not important since only literal bytes will be emitted.
883 /* ===========================================================================
884 * Set match_start to the longest match starting at the given string and
885 * return its length. Matches shorter or equal to prev_length are discarded,
886 * in which case the result is equal to prev_length and match_start is
888 * IN assertions: cur_match is the head of the hash chain for the current
889 * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1
892 /* For MSDOS, OS/2 and 386 Unix, an optimized version is in match.asm or
893 * match.s. The code is functionally equivalent, so you can use the C version
896 static int longest_match(IPos cur_match)
898 unsigned chain_length = max_chain_length; /* max hash chain length */
899 register uch *scan = window + strstart; /* current string */
900 register uch *match; /* matched string */
901 register int len; /* length of current match */
902 int best_len = prev_length; /* best match length so far */
905 strstart > (IPos) MAX_DIST ? strstart - (IPos) MAX_DIST : NIL;
906 /* Stop when cur_match becomes <= limit. To simplify the code,
907 * we prevent matches with the string of window index 0.
910 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
911 * It is easy to get rid of this optimization if necessary.
913 #if HASH_BITS < 8 || MAX_MATCH != 258
914 # error Code too clever
916 register uch *strend = window + strstart + MAX_MATCH;
917 register uch scan_end1 = scan[best_len - 1];
918 register uch scan_end = scan[best_len];
920 /* Do not waste too much time if we already have a good match: */
921 if (prev_length >= good_match) {
924 Assert(strstart <= window_size - MIN_LOOKAHEAD,
925 "insufficient lookahead");
928 Assert(cur_match < strstart, "no future");
929 match = window + cur_match;
931 /* Skip to next match if the match length cannot increase
932 * or if the match length is less than 2:
934 if (match[best_len] != scan_end ||
935 match[best_len - 1] != scan_end1 ||
936 *match != *scan || *++match != scan[1])
939 /* The check at best_len-1 can be removed because it will be made
940 * again later. (This heuristic is not always a win.)
941 * It is not necessary to compare scan[2] and match[2] since they
942 * are always equal when the other bytes match, given that
943 * the hash keys are equal and that HASH_BITS >= 8.
947 /* We check for insufficient lookahead only every 8th comparison;
948 * the 256th check will be made at strstart+258.
951 } while (*++scan == *++match && *++scan == *++match &&
952 *++scan == *++match && *++scan == *++match &&
953 *++scan == *++match && *++scan == *++match &&
954 *++scan == *++match && *++scan == *++match &&
957 len = MAX_MATCH - (int) (strend - scan);
958 scan = strend - MAX_MATCH;
960 if (len > best_len) {
961 match_start = cur_match;
963 if (len >= nice_match)
965 scan_end1 = scan[best_len - 1];
966 scan_end = scan[best_len];
968 } while ((cur_match = prev[cur_match & WMASK]) > limit
969 && --chain_length != 0);
975 /* ===========================================================================
976 * Check that the match at match_start is indeed a match.
978 static void check_match(IPos start, IPos match, int length)
980 /* check that the match is indeed a match */
981 if (memcmp((char *) window + match,
982 (char *) window + start, length) != EQUAL) {
984 " start %d, match %d, length %d\n", start, match, length);
985 error_msg("invalid match");
988 fprintf(stderr, "\\[%d,%d]", start - match, length);
990 putc(window[start++], stderr);
991 } while (--length != 0);
995 # define check_match(start, match, length)
998 /* ===========================================================================
999 * Fill the window when the lookahead becomes insufficient.
1000 * Updates strstart and lookahead, and sets eofile if end of input file.
1001 * IN assertion: lookahead < MIN_LOOKAHEAD && strstart + lookahead > 0
1002 * OUT assertions: at least one byte has been read, or eofile is set;
1003 * file reads are performed for at least two bytes (required for the
1004 * translate_eol option).
1006 static void fill_window()
1008 register unsigned n, m;
1011 (unsigned) (window_size - (ulg) lookahead - (ulg) strstart);
1012 /* Amount of free space at the end of the window. */
1014 /* If the window is almost full and there is insufficient lookahead,
1015 * move the upper half to the lower one to make room in the upper half.
1017 if (more == (unsigned) EOF) {
1018 /* Very unlikely, but possible on 16 bit machine if strstart == 0
1019 * and lookahead == 1 (input done one byte at time)
1022 } else if (strstart >= WSIZE + MAX_DIST) {
1023 /* By the IN assertion, the window is not empty so we can't confuse
1024 * more == 0 with more == 64K on a 16 bit machine.
1026 Assert(window_size == (ulg) 2 * WSIZE, "no sliding with BIG_MEM");
1028 memcpy((char *) window, (char *) window + WSIZE, (unsigned) WSIZE);
1029 match_start -= WSIZE;
1030 strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */
1032 block_start -= (long) WSIZE;
1034 for (n = 0; n < HASH_SIZE; n++) {
1036 head[n] = (Pos) (m >= WSIZE ? m - WSIZE : NIL);
1038 for (n = 0; n < WSIZE; n++) {
1040 prev[n] = (Pos) (m >= WSIZE ? m - WSIZE : NIL);
1041 /* If n is not on any hash chain, prev[n] is garbage but
1042 * its value will never be used.
1047 /* At this point, more >= 2 */
1049 n = read_buf((char *) window + strstart + lookahead, more);
1050 if (n == 0 || n == (unsigned) EOF) {
1058 /* ===========================================================================
1059 * Flush the current block, with given end-of-file flag.
1060 * IN assertion: strstart is set to the end of the current match.
1062 #define FLUSH_BLOCK(eof) \
1063 flush_block(block_start >= 0L ? (char*)&window[(unsigned)block_start] : \
1064 (char*)NULL, (long)strstart - block_start, (eof))
1066 /* ===========================================================================
1067 * Same as above, but achieves better compression. We use a lazy
1068 * evaluation for matches: a match is finally adopted only if there is
1069 * no better match at the next window position.
1071 static ulg deflate()
1073 IPos hash_head; /* head of hash chain */
1074 IPos prev_match; /* previous match */
1075 int flush; /* set if current block must be flushed */
1076 int match_available = 0; /* set if previous match exists */
1077 register unsigned match_length = MIN_MATCH - 1; /* length of best match */
1079 /* Process the input block. */
1080 while (lookahead != 0) {
1081 /* Insert the string window[strstart .. strstart+2] in the
1082 * dictionary, and set hash_head to the head of the hash chain:
1084 INSERT_STRING(strstart, hash_head);
1086 /* Find the longest match, discarding those <= prev_length.
1088 prev_length = match_length, prev_match = match_start;
1089 match_length = MIN_MATCH - 1;
1091 if (hash_head != NIL && prev_length < max_lazy_match &&
1092 strstart - hash_head <= MAX_DIST) {
1093 /* To simplify the code, we prevent matches with the string
1094 * of window index 0 (in particular we have to avoid a match
1095 * of the string with itself at the start of the input file).
1097 match_length = longest_match(hash_head);
1098 /* longest_match() sets match_start */
1099 if (match_length > lookahead)
1100 match_length = lookahead;
1102 /* Ignore a length 3 match if it is too distant: */
1103 if (match_length == MIN_MATCH
1104 && strstart - match_start > TOO_FAR) {
1105 /* If prev_match is also MIN_MATCH, match_start is garbage
1106 * but we will ignore the current match anyway.
1111 /* If there was a match at the previous step and the current
1112 * match is not better, output the previous match:
1114 if (prev_length >= MIN_MATCH && match_length <= prev_length) {
1116 check_match(strstart - 1, prev_match, prev_length);
1119 ct_tally(strstart - 1 - prev_match,
1120 prev_length - MIN_MATCH);
1122 /* Insert in hash table all strings up to the end of the match.
1123 * strstart-1 and strstart are already inserted.
1125 lookahead -= prev_length - 1;
1129 INSERT_STRING(strstart, hash_head);
1130 /* strstart never exceeds WSIZE-MAX_MATCH, so there are
1131 * always MIN_MATCH bytes ahead. If lookahead < MIN_MATCH
1132 * these bytes are garbage, but it does not matter since the
1133 * next lookahead bytes will always be emitted as literals.
1135 } while (--prev_length != 0);
1136 match_available = 0;
1137 match_length = MIN_MATCH - 1;
1140 FLUSH_BLOCK(0), block_start = strstart;
1142 } else if (match_available) {
1143 /* If there was no match at the previous position, output a
1144 * single literal. If there was a match but the current match
1145 * is longer, truncate the previous match to a single literal.
1147 Tracevv((stderr, "%c", window[strstart - 1]));
1148 if (ct_tally(0, window[strstart - 1])) {
1149 FLUSH_BLOCK(0), block_start = strstart;
1154 /* There is no previous match to compare with, wait for
1155 * the next step to decide.
1157 match_available = 1;
1161 Assert(strstart <= isize && lookahead <= isize, "a bit too far");
1163 /* Make sure that we always have enough lookahead, except
1164 * at the end of the input file. We need MAX_MATCH bytes
1165 * for the next match, plus MIN_MATCH bytes to insert the
1166 * string following the next match.
1168 while (lookahead < MIN_LOOKAHEAD && !eofile)
1171 if (match_available)
1172 ct_tally(0, window[strstart - 1]);
1174 return FLUSH_BLOCK(1); /* eof */
1177 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
1178 * Copyright (C) 1992-1993 Jean-loup Gailly
1179 * The unzip code was written and put in the public domain by Mark Adler.
1180 * Portions of the lzw code are derived from the public domain 'compress'
1181 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
1182 * Ken Turkowski, Dave Mack and Peter Jannesen.
1184 * See the license_msg below and the file COPYING for the software license.
1185 * See the file algorithm.doc for the compression algorithms and file formats.
1188 /* Compress files with zip algorithm and 'compress' interface.
1189 * See usage() and help() functions below for all options.
1191 * file.gz: compressed file with same mode, owner, and utimes
1192 * or stdout with -c option or if stdin used as input.
1193 * If the output file name had to be truncated, the original name is kept
1194 * in the compressed file.
1199 typedef struct dirent dir_type;
1201 typedef RETSIGTYPE(*sig_type) (int);
1203 /* ======================================================================== */
1204 // int main (argc, argv)
1207 int gzip_main(int argc, char **argv)
1212 struct stat statBuf;
1218 while ((opt = getopt(argc, argv, "cf123456789dq")) != -1) {
1226 /* Ignore 1-9 (compression level) options */
1227 case '1': case '2': case '3': case '4': case '5':
1228 case '6': case '7': case '8': case '9':
1232 #ifdef CONFIG_GUNZIP
1235 return gunzip_main(argc, argv);
1242 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
1244 (void) signal(SIGINT, (sig_type) abort_gzip);
1247 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
1248 (void) signal(SIGTERM, (sig_type) abort_gzip);
1252 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
1253 (void) signal(SIGHUP, (sig_type) abort_gzip);
1257 strncpy(z_suffix, Z_SUFFIX, sizeof(z_suffix) - 1);
1258 z_len = strlen(z_suffix);
1260 /* Allocate all global buffers (for DYN_ALLOC option) */
1261 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
1262 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
1263 ALLOC(ush, d_buf, DIST_BUFSIZE);
1264 ALLOC(uch, window, 2L * WSIZE);
1265 ALLOC(ush, tab_prefix, 1L << BITS);
1270 if (optind == argc) {
1273 zip(STDIN_FILENO, STDOUT_FILENO);
1277 for (i = optind; i < argc; i++) {
1280 if (strcmp(argv[i], "-") == 0) {
1283 inFileNum = STDIN_FILENO;
1284 outFileNum = STDOUT_FILENO;
1286 inFileNum = open(argv[i], O_RDONLY);
1287 if (inFileNum < 0 || fstat (inFileNum, &statBuf) < 0)
1288 perror_msg_and_die("%s", argv[i]);
1289 time_stamp = statBuf.st_ctime;
1290 ifile_size = statBuf.st_size;
1293 path = xmalloc(strlen(argv[i]) + 4);
1294 strcpy(path, argv[i]);
1295 strcat(path, ".gz");
1297 /* Open output file */
1298 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
1299 outFileNum = open(path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
1301 outFileNum = open(path, O_RDWR | O_CREAT | O_EXCL);
1303 if (outFileNum < 0) {
1304 perror_msg("%s", path);
1309 /* Set permissions on the file */
1310 fchmod(outFileNum, statBuf.st_mode);
1312 outFileNum = STDOUT_FILENO;
1315 if (path == NULL && isatty(outFileNum) && force == 0) {
1316 error_msg("compressed data not written to a terminal. Use -f to force compression.");
1321 result = zip(inFileNum, outFileNum);
1327 /* Delete the original file */
1329 delFileName = argv[i];
1333 if (unlink(delFileName) < 0)
1334 perror_msg("%s", delFileName);
1344 /* trees.c -- output deflated data using Huffman coding
1345 * Copyright (C) 1992-1993 Jean-loup Gailly
1346 * This is free software; you can redistribute it and/or modify it under the
1347 * terms of the GNU General Public License, see the file COPYING.
1353 * Encode various sets of source values using variable-length
1354 * binary code trees.
1358 * The PKZIP "deflation" process uses several Huffman trees. The more
1359 * common source values are represented by shorter bit sequences.
1361 * Each code tree is stored in the ZIP file in a compressed form
1362 * which is itself a Huffman encoding of the lengths of
1363 * all the code strings (in ascending order by source values).
1364 * The actual code strings are reconstructed from the lengths in
1365 * the UNZIP process, as described in the "application note"
1366 * (APPNOTE.TXT) distributed as part of PKWARE's PKZIP program.
1371 * Data Compression: Techniques and Applications, pp. 53-55.
1372 * Lifetime Learning Publications, 1985. ISBN 0-534-03418-7.
1375 * Data Compression: Methods and Theory, pp. 49-50.
1376 * Computer Science Press, 1988. ISBN 0-7167-8156-5.
1380 * Addison-Wesley, 1983. ISBN 0-201-06672-6.
1384 * void ct_init (ush *attr, int *methodp)
1385 * Allocate the match buffer, initialize the various tables and save
1386 * the location of the internal file attribute (ascii/binary) and
1387 * method (DEFLATE/STORE)
1389 * void ct_tally (int dist, int lc);
1390 * Save the match info and tally the frequency counts.
1392 * long flush_block (char *buf, ulg stored_len, int eof)
1393 * Determine the best encoding for the current block: dynamic trees,
1394 * static trees or store, and output the encoded block to the zip
1395 * file. Returns the total compressed length for the file so far.
1399 /* ===========================================================================
1404 /* All codes must not exceed MAX_BITS bits */
1406 #define MAX_BL_BITS 7
1407 /* Bit length codes must not exceed MAX_BL_BITS bits */
1409 #define LENGTH_CODES 29
1410 /* number of length codes, not counting the special END_BLOCK code */
1412 #define LITERALS 256
1413 /* number of literal bytes 0..255 */
1415 #define END_BLOCK 256
1416 /* end of block literal code */
1418 #define L_CODES (LITERALS+1+LENGTH_CODES)
1419 /* number of Literal or Length codes, including the END_BLOCK code */
1422 /* number of distance codes */
1425 /* number of codes used to transfer the bit lengths */
1427 typedef uch extra_bits_t;
1429 /* extra bits for each length code */
1430 static const extra_bits_t extra_lbits[LENGTH_CODES]
1431 = { 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4,
1432 4, 4, 5, 5, 5, 5, 0 };
1434 /* extra bits for each distance code */
1435 static const extra_bits_t extra_dbits[D_CODES]
1436 = { 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9,
1437 10, 10, 11, 11, 12, 12, 13, 13 };
1439 /* extra bits for each bit length code */
1440 static const extra_bits_t extra_blbits[BL_CODES]
1441 = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 3, 7 };
1443 #define STORED_BLOCK 0
1444 #define STATIC_TREES 1
1446 /* The three kinds of block type */
1450 # define LIT_BUFSIZE 0x2000
1453 # define LIT_BUFSIZE 0x4000
1455 # define LIT_BUFSIZE 0x8000
1459 #ifndef DIST_BUFSIZE
1460 # define DIST_BUFSIZE LIT_BUFSIZE
1462 /* Sizes of match buffers for literals/lengths and distances. There are
1463 * 4 reasons for limiting LIT_BUFSIZE to 64K:
1464 * - frequencies can be kept in 16 bit counters
1465 * - if compression is not successful for the first block, all input data is
1466 * still in the window so we can still emit a stored block even when input
1467 * comes from standard input. (This can also be done for all blocks if
1468 * LIT_BUFSIZE is not greater than 32K.)
1469 * - if compression is not successful for a file smaller than 64K, we can
1470 * even emit a stored file instead of a stored block (saving 5 bytes).
1471 * - creating new Huffman trees less frequently may not provide fast
1472 * adaptation to changes in the input data statistics. (Take for
1473 * example a binary file with poorly compressible code followed by
1474 * a highly compressible string table.) Smaller buffer sizes give
1475 * fast adaptation but have of course the overhead of transmitting trees
1477 * - I can't count above 4
1478 * The current code is general and allows DIST_BUFSIZE < LIT_BUFSIZE (to save
1479 * memory at the expense of compression). Some optimizations would be possible
1480 * if we rely on DIST_BUFSIZE == LIT_BUFSIZE.
1482 #if LIT_BUFSIZE > INBUFSIZ
1483 #error cannot overlay l_buf and inbuf
1486 /* repeat previous bit length 3-6 times (2 bits of repeat count) */
1487 #define REPZ_3_10 17
1488 /* repeat a zero length 3-10 times (3 bits of repeat count) */
1489 #define REPZ_11_138 18
1490 /* repeat a zero length 11-138 times (7 bits of repeat count) *//* ===========================================================================
1492 *//* Data structure describing a single value and its code string. */ typedef struct ct_data {
1494 ush freq; /* frequency count */
1495 ush code; /* bit string */
1498 ush dad; /* father node in Huffman tree */
1499 ush len; /* length of bit string */
1503 #define Freq fc.freq
1504 #define Code fc.code
1508 #define HEAP_SIZE (2*L_CODES+1)
1509 /* maximum heap size */
1511 static ct_data dyn_ltree[HEAP_SIZE]; /* literal and length tree */
1512 static ct_data dyn_dtree[2 * D_CODES + 1]; /* distance tree */
1514 static ct_data static_ltree[L_CODES + 2];
1516 /* The static literal tree. Since the bit lengths are imposed, there is no
1517 * need for the L_CODES extra codes used during heap construction. However
1518 * The codes 286 and 287 are needed to build a canonical tree (see ct_init
1522 static ct_data static_dtree[D_CODES];
1524 /* The static distance tree. (Actually a trivial tree since all codes use
1528 static ct_data bl_tree[2 * BL_CODES + 1];
1530 /* Huffman tree for the bit lengths */
1532 typedef struct tree_desc {
1533 ct_data *dyn_tree; /* the dynamic tree */
1534 ct_data *static_tree; /* corresponding static tree or NULL */
1535 const extra_bits_t *extra_bits; /* extra bits for each code or NULL */
1536 int extra_base; /* base index for extra_bits */
1537 int elems; /* max number of elements in the tree */
1538 int max_length; /* max bit length for the codes */
1539 int max_code; /* largest code with non zero frequency */
1542 static tree_desc l_desc =
1543 { dyn_ltree, static_ltree, extra_lbits, LITERALS + 1, L_CODES,
1546 static tree_desc d_desc =
1547 { dyn_dtree, static_dtree, extra_dbits, 0, D_CODES, MAX_BITS, 0 };
1549 static tree_desc bl_desc =
1550 { bl_tree, (ct_data *) 0, extra_blbits, 0, BL_CODES, MAX_BL_BITS,
1554 static ush bl_count[MAX_BITS + 1];
1556 /* number of codes at each bit length for an optimal tree */
1558 static const uch bl_order[BL_CODES]
1559 = { 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15 };
1561 /* The lengths of the bit length codes are sent in order of decreasing
1562 * probability, to avoid transmitting the lengths for unused bit length codes.
1565 static int heap[2 * L_CODES + 1]; /* heap used to build the Huffman trees */
1566 static int heap_len; /* number of elements in the heap */
1567 static int heap_max; /* element of largest frequency */
1569 /* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used.
1570 * The same heap array is used to build all trees.
1573 static uch depth[2 * L_CODES + 1];
1575 /* Depth of each subtree used as tie breaker for trees of equal frequency */
1577 static uch length_code[MAX_MATCH - MIN_MATCH + 1];
1579 /* length code for each normalized match length (0 == MIN_MATCH) */
1581 static uch dist_code[512];
1583 /* distance codes. The first 256 values correspond to the distances
1584 * 3 .. 258, the last 256 values correspond to the top 8 bits of
1585 * the 15 bit distances.
1588 static int base_length[LENGTH_CODES];
1590 /* First normalized length for each code (0 = MIN_MATCH) */
1592 static int base_dist[D_CODES];
1594 /* First normalized distance for each code (0 = distance of 1) */
1597 /* DECLARE(uch, l_buf, LIT_BUFSIZE); buffer for literals or lengths */
1599 /* DECLARE(ush, d_buf, DIST_BUFSIZE); buffer for distances */
1601 static uch flag_buf[(LIT_BUFSIZE / 8)];
1603 /* flag_buf is a bit array distinguishing literals from lengths in
1604 * l_buf, thus indicating the presence or absence of a distance.
1607 static unsigned last_lit; /* running index in l_buf */
1608 static unsigned last_dist; /* running index in d_buf */
1609 static unsigned last_flags; /* running index in flag_buf */
1610 static uch flags; /* current flags not yet saved in flag_buf */
1611 static uch flag_bit; /* current bit used in flags */
1613 /* bits are filled in flags starting at bit 0 (least significant).
1614 * Note: these flags are overkill in the current code since we don't
1615 * take advantage of DIST_BUFSIZE == LIT_BUFSIZE.
1618 static ulg opt_len; /* bit length of current block with optimal trees */
1619 static ulg static_len; /* bit length of current block with static trees */
1621 static ulg compressed_len; /* total bit length of compressed file */
1624 static ush *file_type; /* pointer to UNKNOWN, BINARY or ASCII */
1625 static int *file_method; /* pointer to DEFLATE or STORE */
1627 /* ===========================================================================
1628 * Local (static) routines in this file.
1631 static void init_block (void);
1632 static void pqdownheap (ct_data * tree, int k);
1633 static void gen_bitlen (tree_desc * desc);
1634 static void gen_codes (ct_data * tree, int max_code);
1635 static void build_tree (tree_desc * desc);
1636 static void scan_tree (ct_data * tree, int max_code);
1637 static void send_tree (ct_data * tree, int max_code);
1638 static int build_bl_tree (void);
1639 static void send_all_trees (int lcodes, int dcodes, int blcodes);
1640 static void compress_block (ct_data * ltree, ct_data * dtree);
1641 static void set_file_type (void);
1645 # define send_code(c, tree) send_bits(tree[c].Code, tree[c].Len)
1646 /* Send a code of the given tree. c and tree must not have side effects */
1649 # define send_code(c, tree) \
1650 { if (verbose>1) fprintf(stderr,"\ncd %3d ",(c)); \
1651 send_bits(tree[c].Code, tree[c].Len); }
1654 #define d_code(dist) \
1655 ((dist) < 256 ? dist_code[dist] : dist_code[256+((dist)>>7)])
1656 /* Mapping from a distance to a distance code. dist is the distance - 1 and
1657 * must not have side effects. dist_code[256] and dist_code[257] are never
1661 /* the arguments must not have side effects */
1663 /* ===========================================================================
1664 * Allocate the match buffer, initialize the various tables and save the
1665 * location of the internal file attribute (ascii/binary) and method
1668 static void ct_init(ush *attr, int *methodp)
1670 int n; /* iterates over tree elements */
1671 int bits; /* bit counter */
1672 int length; /* length value */
1673 int code; /* code value */
1674 int dist; /* distance index */
1677 file_method = methodp;
1678 compressed_len = 0L;
1680 if (static_dtree[0].Len != 0)
1681 return; /* ct_init already called */
1683 /* Initialize the mapping length (0..255) -> length code (0..28) */
1685 for (code = 0; code < LENGTH_CODES - 1; code++) {
1686 base_length[code] = length;
1687 for (n = 0; n < (1 << extra_lbits[code]); n++) {
1688 length_code[length++] = (uch) code;
1691 Assert(length == 256, "ct_init: length != 256");
1692 /* Note that the length 255 (match length 258) can be represented
1693 * in two different ways: code 284 + 5 bits or code 285, so we
1694 * overwrite length_code[255] to use the best encoding:
1696 length_code[length - 1] = (uch) code;
1698 /* Initialize the mapping dist (0..32K) -> dist code (0..29) */
1700 for (code = 0; code < 16; code++) {
1701 base_dist[code] = dist;
1702 for (n = 0; n < (1 << extra_dbits[code]); n++) {
1703 dist_code[dist++] = (uch) code;
1706 Assert(dist == 256, "ct_init: dist != 256");
1707 dist >>= 7; /* from now on, all distances are divided by 128 */
1708 for (; code < D_CODES; code++) {
1709 base_dist[code] = dist << 7;
1710 for (n = 0; n < (1 << (extra_dbits[code] - 7)); n++) {
1711 dist_code[256 + dist++] = (uch) code;
1714 Assert(dist == 256, "ct_init: 256+dist != 512");
1716 /* Construct the codes of the static literal tree */
1717 for (bits = 0; bits <= MAX_BITS; bits++)
1721 static_ltree[n++].Len = 8, bl_count[8]++;
1723 static_ltree[n++].Len = 9, bl_count[9]++;
1725 static_ltree[n++].Len = 7, bl_count[7]++;
1727 static_ltree[n++].Len = 8, bl_count[8]++;
1728 /* Codes 286 and 287 do not exist, but we must include them in the
1729 * tree construction to get a canonical Huffman tree (longest code
1732 gen_codes((ct_data *) static_ltree, L_CODES + 1);
1734 /* The static distance tree is trivial: */
1735 for (n = 0; n < D_CODES; n++) {
1736 static_dtree[n].Len = 5;
1737 static_dtree[n].Code = bi_reverse(n, 5);
1740 /* Initialize the first block of the first file: */
1744 /* ===========================================================================
1745 * Initialize a new block.
1747 static void init_block()
1749 int n; /* iterates over tree elements */
1751 /* Initialize the trees. */
1752 for (n = 0; n < L_CODES; n++)
1753 dyn_ltree[n].Freq = 0;
1754 for (n = 0; n < D_CODES; n++)
1755 dyn_dtree[n].Freq = 0;
1756 for (n = 0; n < BL_CODES; n++)
1757 bl_tree[n].Freq = 0;
1759 dyn_ltree[END_BLOCK].Freq = 1;
1760 opt_len = static_len = 0L;
1761 last_lit = last_dist = last_flags = 0;
1767 /* Index within the heap array of least frequent node in the Huffman tree */
1770 /* ===========================================================================
1771 * Remove the smallest element from the heap and recreate the heap with
1772 * one less element. Updates heap and heap_len.
1774 #define pqremove(tree, top) \
1776 top = heap[SMALLEST]; \
1777 heap[SMALLEST] = heap[heap_len--]; \
1778 pqdownheap(tree, SMALLEST); \
1781 /* ===========================================================================
1782 * Compares to subtrees, using the tree depth as tie breaker when
1783 * the subtrees have equal frequency. This minimizes the worst case length.
1785 #define smaller(tree, n, m) \
1786 (tree[n].Freq < tree[m].Freq || \
1787 (tree[n].Freq == tree[m].Freq && depth[n] <= depth[m]))
1789 /* ===========================================================================
1790 * Restore the heap property by moving down the tree starting at node k,
1791 * exchanging a node with the smallest of its two sons if necessary, stopping
1792 * when the heap property is re-established (each father smaller than its
1795 static void pqdownheap(ct_data *tree, int k)
1798 int j = k << 1; /* left son of k */
1800 while (j <= heap_len) {
1801 /* Set j to the smallest of the two sons: */
1802 if (j < heap_len && smaller(tree, heap[j + 1], heap[j]))
1805 /* Exit if v is smaller than both sons */
1806 if (smaller(tree, v, heap[j]))
1809 /* Exchange v with the smallest son */
1813 /* And continue down the tree, setting j to the left son of k */
1819 /* ===========================================================================
1820 * Compute the optimal bit lengths for a tree and update the total bit length
1821 * for the current block.
1822 * IN assertion: the fields freq and dad are set, heap[heap_max] and
1823 * above are the tree nodes sorted by increasing frequency.
1824 * OUT assertions: the field len is set to the optimal bit length, the
1825 * array bl_count contains the frequencies for each bit length.
1826 * The length opt_len is updated; static_len is also updated if stree is
1829 static void gen_bitlen(tree_desc *desc)
1831 ct_data *tree = desc->dyn_tree;
1832 const extra_bits_t *extra = desc->extra_bits;
1833 int base = desc->extra_base;
1834 int max_code = desc->max_code;
1835 int max_length = desc->max_length;
1836 ct_data *stree = desc->static_tree;
1837 int h; /* heap index */
1838 int n, m; /* iterate over the tree elements */
1839 int bits; /* bit length */
1840 int xbits; /* extra bits */
1841 ush f; /* frequency */
1842 int overflow = 0; /* number of elements with bit length too large */
1844 for (bits = 0; bits <= MAX_BITS; bits++)
1847 /* In a first pass, compute the optimal bit lengths (which may
1848 * overflow in the case of the bit length tree).
1850 tree[heap[heap_max]].Len = 0; /* root of the heap */
1852 for (h = heap_max + 1; h < HEAP_SIZE; h++) {
1854 bits = tree[tree[n].Dad].Len + 1;
1855 if (bits > max_length)
1856 bits = max_length, overflow++;
1857 tree[n].Len = (ush) bits;
1858 /* We overwrite tree[n].Dad which is no longer needed */
1861 continue; /* not a leaf node */
1866 xbits = extra[n - base];
1868 opt_len += (ulg) f *(bits + xbits);
1871 static_len += (ulg) f *(stree[n].Len + xbits);
1876 Trace((stderr, "\nbit length overflow\n"));
1877 /* This happens for example on obj2 and pic of the Calgary corpus */
1879 /* Find the first bit length which could increase: */
1881 bits = max_length - 1;
1882 while (bl_count[bits] == 0)
1884 bl_count[bits]--; /* move one leaf down the tree */
1885 bl_count[bits + 1] += 2; /* move one overflow item as its brother */
1886 bl_count[max_length]--;
1887 /* The brother of the overflow item also moves one step up,
1888 * but this does not affect bl_count[max_length]
1891 } while (overflow > 0);
1893 /* Now recompute all bit lengths, scanning in increasing frequency.
1894 * h is still equal to HEAP_SIZE. (It is simpler to reconstruct all
1895 * lengths instead of fixing only the wrong ones. This idea is taken
1896 * from 'ar' written by Haruhiko Okumura.)
1898 for (bits = max_length; bits != 0; bits--) {
1904 if (tree[m].Len != (unsigned) bits) {
1906 (stderr, "code %d bits %d->%d\n", m, tree[m].Len,
1910 (long) tree[m].Len) * (long) tree[m].Freq;
1911 tree[m].Len = (ush) bits;
1918 /* ===========================================================================
1919 * Generate the codes for a given tree and bit counts (which need not be
1921 * IN assertion: the array bl_count contains the bit length statistics for
1922 * the given tree and the field len is set for all tree elements.
1923 * OUT assertion: the field code is set for all tree elements of non
1926 static void gen_codes(ct_data *tree, int max_code)
1928 ush next_code[MAX_BITS + 1]; /* next code value for each bit length */
1929 ush code = 0; /* running code value */
1930 int bits; /* bit index */
1931 int n; /* code index */
1933 /* The distribution counts are first used to generate the code values
1934 * without bit reversal.
1936 for (bits = 1; bits <= MAX_BITS; bits++) {
1937 next_code[bits] = code = (code + bl_count[bits - 1]) << 1;
1939 /* Check that the bit counts in bl_count are consistent. The last code
1942 Assert(code + bl_count[MAX_BITS] - 1 == (1 << MAX_BITS) - 1,
1943 "inconsistent bit counts");
1944 Tracev((stderr, "\ngen_codes: max_code %d ", max_code));
1946 for (n = 0; n <= max_code; n++) {
1947 int len = tree[n].Len;
1951 /* Now reverse the bits */
1952 tree[n].Code = bi_reverse(next_code[len]++, len);
1954 Tracec(tree != static_ltree,
1955 (stderr, "\nn %3d %c l %2d c %4x (%x) ", n,
1956 (isgraph(n) ? n : ' '), len, tree[n].Code,
1957 next_code[len] - 1));
1961 /* ===========================================================================
1962 * Construct one Huffman tree and assigns the code bit strings and lengths.
1963 * Update the total bit length for the current block.
1964 * IN assertion: the field freq is set for all tree elements.
1965 * OUT assertions: the fields len and code are set to the optimal bit length
1966 * and corresponding code. The length opt_len is updated; static_len is
1967 * also updated if stree is not null. The field max_code is set.
1969 static void build_tree(tree_desc *desc)
1971 ct_data *tree = desc->dyn_tree;
1972 ct_data *stree = desc->static_tree;
1973 int elems = desc->elems;
1974 int n, m; /* iterate over heap elements */
1975 int max_code = -1; /* largest code with non zero frequency */
1976 int node = elems; /* next internal node of the tree */
1978 /* Construct the initial heap, with least frequent element in
1979 * heap[SMALLEST]. The sons of heap[n] are heap[2*n] and heap[2*n+1].
1980 * heap[0] is not used.
1982 heap_len = 0, heap_max = HEAP_SIZE;
1984 for (n = 0; n < elems; n++) {
1985 if (tree[n].Freq != 0) {
1986 heap[++heap_len] = max_code = n;
1993 /* The pkzip format requires that at least one distance code exists,
1994 * and that at least one bit should be sent even if there is only one
1995 * possible code. So to avoid special checks later on we force at least
1996 * two codes of non zero frequency.
1998 while (heap_len < 2) {
1999 int new = heap[++heap_len] = (max_code < 2 ? ++max_code : 0);
2005 static_len -= stree[new].Len;
2006 /* new is 0 or 1 so it does not have extra bits */
2008 desc->max_code = max_code;
2010 /* The elements heap[heap_len/2+1 .. heap_len] are leaves of the tree,
2011 * establish sub-heaps of increasing lengths:
2013 for (n = heap_len / 2; n >= 1; n--)
2014 pqdownheap(tree, n);
2016 /* Construct the Huffman tree by repeatedly combining the least two
2020 pqremove(tree, n); /* n = node of least frequency */
2021 m = heap[SMALLEST]; /* m = node of next least frequency */
2023 heap[--heap_max] = n; /* keep the nodes sorted by frequency */
2024 heap[--heap_max] = m;
2026 /* Create a new node father of n and m */
2027 tree[node].Freq = tree[n].Freq + tree[m].Freq;
2028 depth[node] = (uch) (MAX(depth[n], depth[m]) + 1);
2029 tree[n].Dad = tree[m].Dad = (ush) node;
2031 if (tree == bl_tree) {
2032 fprintf(stderr, "\nnode %d(%d), sons %d(%d) %d(%d)",
2033 node, tree[node].Freq, n, tree[n].Freq, m,
2037 /* and insert the new node in the heap */
2038 heap[SMALLEST] = node++;
2039 pqdownheap(tree, SMALLEST);
2041 } while (heap_len >= 2);
2043 heap[--heap_max] = heap[SMALLEST];
2045 /* At this point, the fields freq and dad are set. We can now
2046 * generate the bit lengths.
2048 gen_bitlen((tree_desc *) desc);
2050 /* The field len is now set, we can generate the bit codes */
2051 gen_codes((ct_data *) tree, max_code);
2054 /* ===========================================================================
2055 * Scan a literal or distance tree to determine the frequencies of the codes
2056 * in the bit length tree. Updates opt_len to take into account the repeat
2057 * counts. (The contribution of the bit length codes will be added later
2058 * during the construction of bl_tree.)
2060 static void scan_tree(ct_data *tree, int max_code)
2062 int n; /* iterates over all tree elements */
2063 int prevlen = -1; /* last emitted length */
2064 int curlen; /* length of current code */
2065 int nextlen = tree[0].Len; /* length of next code */
2066 int count = 0; /* repeat count of the current code */
2067 int max_count = 7; /* max repeat count */
2068 int min_count = 4; /* min repeat count */
2071 max_count = 138, min_count = 3;
2072 tree[max_code + 1].Len = (ush) 0xffff; /* guard */
2074 for (n = 0; n <= max_code; n++) {
2076 nextlen = tree[n + 1].Len;
2077 if (++count < max_count && curlen == nextlen) {
2079 } else if (count < min_count) {
2080 bl_tree[curlen].Freq += count;
2081 } else if (curlen != 0) {
2082 if (curlen != prevlen)
2083 bl_tree[curlen].Freq++;
2084 bl_tree[REP_3_6].Freq++;
2085 } else if (count <= 10) {
2086 bl_tree[REPZ_3_10].Freq++;
2088 bl_tree[REPZ_11_138].Freq++;
2093 max_count = 138, min_count = 3;
2094 } else if (curlen == nextlen) {
2095 max_count = 6, min_count = 3;
2097 max_count = 7, min_count = 4;
2102 /* ===========================================================================
2103 * Send a literal or distance tree in compressed form, using the codes in
2106 static void send_tree(ct_data *tree, int max_code)
2108 int n; /* iterates over all tree elements */
2109 int prevlen = -1; /* last emitted length */
2110 int curlen; /* length of current code */
2111 int nextlen = tree[0].Len; /* length of next code */
2112 int count = 0; /* repeat count of the current code */
2113 int max_count = 7; /* max repeat count */
2114 int min_count = 4; /* min repeat count */
2116 /* tree[max_code+1].Len = -1; *//* guard already set */
2118 max_count = 138, min_count = 3;
2120 for (n = 0; n <= max_code; n++) {
2122 nextlen = tree[n + 1].Len;
2123 if (++count < max_count && curlen == nextlen) {
2125 } else if (count < min_count) {
2127 send_code(curlen, bl_tree);
2128 } while (--count != 0);
2130 } else if (curlen != 0) {
2131 if (curlen != prevlen) {
2132 send_code(curlen, bl_tree);
2135 Assert(count >= 3 && count <= 6, " 3_6?");
2136 send_code(REP_3_6, bl_tree);
2137 send_bits(count - 3, 2);
2139 } else if (count <= 10) {
2140 send_code(REPZ_3_10, bl_tree);
2141 send_bits(count - 3, 3);
2144 send_code(REPZ_11_138, bl_tree);
2145 send_bits(count - 11, 7);
2150 max_count = 138, min_count = 3;
2151 } else if (curlen == nextlen) {
2152 max_count = 6, min_count = 3;
2154 max_count = 7, min_count = 4;
2159 /* ===========================================================================
2160 * Construct the Huffman tree for the bit lengths and return the index in
2161 * bl_order of the last bit length code to send.
2163 static const int build_bl_tree()
2165 int max_blindex; /* index of last bit length code of non zero freq */
2167 /* Determine the bit length frequencies for literal and distance trees */
2168 scan_tree((ct_data *) dyn_ltree, l_desc.max_code);
2169 scan_tree((ct_data *) dyn_dtree, d_desc.max_code);
2171 /* Build the bit length tree: */
2172 build_tree((tree_desc *) (&bl_desc));
2173 /* opt_len now includes the length of the tree representations, except
2174 * the lengths of the bit lengths codes and the 5+5+4 bits for the counts.
2177 /* Determine the number of bit length codes to send. The pkzip format
2178 * requires that at least 4 bit length codes be sent. (appnote.txt says
2179 * 3 but the actual value used is 4.)
2181 for (max_blindex = BL_CODES - 1; max_blindex >= 3; max_blindex--) {
2182 if (bl_tree[bl_order[max_blindex]].Len != 0)
2185 /* Update opt_len to include the bit length tree and counts */
2186 opt_len += 3 * (max_blindex + 1) + 5 + 5 + 4;
2188 (stderr, "\ndyn trees: dyn %ld, stat %ld", opt_len,
2194 /* ===========================================================================
2195 * Send the header for a block using dynamic Huffman trees: the counts, the
2196 * lengths of the bit length codes, the literal tree and the distance tree.
2197 * IN assertion: lcodes >= 257, dcodes >= 1, blcodes >= 4.
2199 static void send_all_trees(int lcodes, int dcodes, int blcodes)
2201 int rank; /* index in bl_order */
2203 Assert(lcodes >= 257 && dcodes >= 1
2204 && blcodes >= 4, "not enough codes");
2205 Assert(lcodes <= L_CODES && dcodes <= D_CODES
2206 && blcodes <= BL_CODES, "too many codes");
2207 Tracev((stderr, "\nbl counts: "));
2208 send_bits(lcodes - 257, 5); /* not +255 as stated in appnote.txt */
2209 send_bits(dcodes - 1, 5);
2210 send_bits(blcodes - 4, 4); /* not -3 as stated in appnote.txt */
2211 for (rank = 0; rank < blcodes; rank++) {
2212 Tracev((stderr, "\nbl code %2d ", bl_order[rank]));
2213 send_bits(bl_tree[bl_order[rank]].Len, 3);
2215 Tracev((stderr, "\nbl tree: sent %ld", bits_sent));
2217 send_tree((ct_data *) dyn_ltree, lcodes - 1); /* send the literal tree */
2218 Tracev((stderr, "\nlit tree: sent %ld", bits_sent));
2220 send_tree((ct_data *) dyn_dtree, dcodes - 1); /* send the distance tree */
2221 Tracev((stderr, "\ndist tree: sent %ld", bits_sent));
2224 /* ===========================================================================
2225 * Determine the best encoding for the current block: dynamic trees, static
2226 * trees or store, and output the encoded block to the zip file. This function
2227 * returns the total compressed length for the file so far.
2229 static ulg flush_block(char *buf, ulg stored_len, int eof)
2231 ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */
2232 int max_blindex; /* index of last bit length code of non zero freq */
2234 flag_buf[last_flags] = flags; /* Save the flags for the last 8 items */
2236 /* Check if the file is ascii or binary */
2237 if (*file_type == (ush) UNKNOWN)
2240 /* Construct the literal and distance trees */
2241 build_tree((tree_desc *) (&l_desc));
2242 Tracev((stderr, "\nlit data: dyn %ld, stat %ld", opt_len, static_len));
2244 build_tree((tree_desc *) (&d_desc));
2246 (stderr, "\ndist data: dyn %ld, stat %ld", opt_len,
2248 /* At this point, opt_len and static_len are the total bit lengths of
2249 * the compressed block data, excluding the tree representations.
2252 /* Build the bit length tree for the above two trees, and get the index
2253 * in bl_order of the last bit length code to send.
2255 max_blindex = build_bl_tree();
2257 /* Determine the best encoding. Compute first the block length in bytes */
2258 opt_lenb = (opt_len + 3 + 7) >> 3;
2259 static_lenb = (static_len + 3 + 7) >> 3;
2263 "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u dist %u ",
2264 opt_lenb, opt_len, static_lenb, static_len, stored_len,
2265 last_lit, last_dist));
2267 if (static_lenb <= opt_lenb)
2268 opt_lenb = static_lenb;
2270 /* If compression failed and this is the first and last block,
2271 * and if the zip file can be seeked (to rewrite the local header),
2272 * the whole file is transformed into a stored file:
2274 if (stored_len <= opt_lenb && eof && compressed_len == 0L
2276 /* Since LIT_BUFSIZE <= 2*WSIZE, the input data must be there: */
2277 if (buf == (char *) 0)
2278 error_msg("block vanished");
2280 copy_block(buf, (unsigned) stored_len, 0); /* without header */
2281 compressed_len = stored_len << 3;
2282 *file_method = STORED;
2284 } else if (stored_len + 4 <= opt_lenb && buf != (char *) 0) {
2285 /* 4: two words for the lengths */
2286 /* The test buf != NULL is only necessary if LIT_BUFSIZE > WSIZE.
2287 * Otherwise we can't have processed more than WSIZE input bytes since
2288 * the last block flush, because compression would have been
2289 * successful. If LIT_BUFSIZE <= WSIZE, it is never too late to
2290 * transform a block into a stored block.
2292 send_bits((STORED_BLOCK << 1) + eof, 3); /* send block type */
2293 compressed_len = (compressed_len + 3 + 7) & ~7L;
2294 compressed_len += (stored_len + 4) << 3;
2296 copy_block(buf, (unsigned) stored_len, 1); /* with header */
2298 } else if (static_lenb == opt_lenb) {
2299 send_bits((STATIC_TREES << 1) + eof, 3);
2300 compress_block((ct_data *) static_ltree,
2301 (ct_data *) static_dtree);
2302 compressed_len += 3 + static_len;
2304 send_bits((DYN_TREES << 1) + eof, 3);
2305 send_all_trees(l_desc.max_code + 1, d_desc.max_code + 1,
2307 compress_block((ct_data *) dyn_ltree,
2308 (ct_data *) dyn_dtree);
2309 compressed_len += 3 + opt_len;
2311 Assert(compressed_len == bits_sent, "bad compressed size");
2316 compressed_len += 7; /* align on byte boundary */
2318 Tracev((stderr, "\ncomprlen %lu(%lu) ", compressed_len >> 3,
2319 compressed_len - 7 * eof));
2321 return compressed_len >> 3;
2324 /* ===========================================================================
2325 * Save the match info and tally the frequency counts. Return true if
2326 * the current block must be flushed.
2328 static int ct_tally(int dist, int lc)
2330 l_buf[last_lit++] = (uch) lc;
2332 /* lc is the unmatched char */
2333 dyn_ltree[lc].Freq++;
2335 /* Here, lc is the match length - MIN_MATCH */
2336 dist--; /* dist = match distance - 1 */
2337 Assert((ush) dist < (ush) MAX_DIST &&
2338 (ush) lc <= (ush) (MAX_MATCH - MIN_MATCH) &&
2339 (ush) d_code(dist) < (ush) D_CODES, "ct_tally: bad match");
2341 dyn_ltree[length_code[lc] + LITERALS + 1].Freq++;
2342 dyn_dtree[d_code(dist)].Freq++;
2344 d_buf[last_dist++] = (ush) dist;
2349 /* Output the flags if they fill a byte: */
2350 if ((last_lit & 7) == 0) {
2351 flag_buf[last_flags++] = flags;
2352 flags = 0, flag_bit = 1;
2354 /* Try to guess if it is profitable to stop the current block here */
2355 if ((last_lit & 0xfff) == 0) {
2356 /* Compute an upper bound for the compressed length */
2357 ulg out_length = (ulg) last_lit * 8L;
2358 ulg in_length = (ulg) strstart - block_start;
2361 for (dcode = 0; dcode < D_CODES; dcode++) {
2363 (ulg) dyn_dtree[dcode].Freq * (5L + extra_dbits[dcode]);
2368 "\nlast_lit %u, last_dist %u, in %ld, out ~%ld(%ld%%) ",
2369 last_lit, last_dist, in_length, out_length,
2370 100L - out_length * 100L / in_length));
2371 if (last_dist < last_lit / 2 && out_length < in_length / 2)
2374 return (last_lit == LIT_BUFSIZE - 1 || last_dist == DIST_BUFSIZE);
2375 /* We avoid equality with LIT_BUFSIZE because of wraparound at 64K
2376 * on 16 bit machines and because stored blocks are restricted to
2381 /* ===========================================================================
2382 * Send the block data compressed using the given Huffman trees
2384 static void compress_block(ct_data *ltree, ct_data *dtree)
2386 unsigned dist; /* distance of matched string */
2387 int lc; /* match length or unmatched char (if dist == 0) */
2388 unsigned lx = 0; /* running index in l_buf */
2389 unsigned dx = 0; /* running index in d_buf */
2390 unsigned fx = 0; /* running index in flag_buf */
2391 uch flag = 0; /* current flags */
2392 unsigned code; /* the code to send */
2393 int extra; /* number of extra bits to send */
2398 flag = flag_buf[fx++];
2400 if ((flag & 1) == 0) {
2401 send_code(lc, ltree); /* send a literal byte */
2402 Tracecv(isgraph(lc), (stderr, " '%c' ", lc));
2404 /* Here, lc is the match length - MIN_MATCH */
2405 code = length_code[lc];
2406 send_code(code + LITERALS + 1, ltree); /* send the length code */
2407 extra = extra_lbits[code];
2409 lc -= base_length[code];
2410 send_bits(lc, extra); /* send the extra length bits */
2413 /* Here, dist is the match distance - 1 */
2414 code = d_code(dist);
2415 Assert(code < D_CODES, "bad d_code");
2417 send_code(code, dtree); /* send the distance code */
2418 extra = extra_dbits[code];
2420 dist -= base_dist[code];
2421 send_bits(dist, extra); /* send the extra distance bits */
2423 } /* literal or match pair ? */
2425 } while (lx < last_lit);
2427 send_code(END_BLOCK, ltree);
2430 /* ===========================================================================
2431 * Set the file type to ASCII or BINARY, using a crude approximation:
2432 * binary if more than 20% of the bytes are <= 6 or >= 128, ascii otherwise.
2433 * IN assertion: the fields freq of dyn_ltree are set and the total of all
2434 * frequencies does not exceed 64K (to fit in an int on 16 bit machines).
2436 static void set_file_type()
2439 unsigned ascii_freq = 0;
2440 unsigned bin_freq = 0;
2443 bin_freq += dyn_ltree[n++].Freq;
2445 ascii_freq += dyn_ltree[n++].Freq;
2446 while (n < LITERALS)
2447 bin_freq += dyn_ltree[n++].Freq;
2448 *file_type = bin_freq > (ascii_freq >> 2) ? BINARY : ASCII;
2449 if (*file_type == BINARY && translate_eol) {
2450 error_msg("-l used on binary file");
2454 /* zip.c -- compress files to the gzip or pkzip format
2455 * Copyright (C) 1992-1993 Jean-loup Gailly
2456 * This is free software; you can redistribute it and/or modify it under the
2457 * terms of the GNU General Public License, see the file COPYING.
2461 static ulg crc; /* crc on uncompressed file data */
2462 static long header_bytes; /* number of bytes in gzip header */
2464 static void put_long(ulg n)
2466 put_short((n) & 0xffff);
2467 put_short(((ulg)(n)) >> 16);
2470 /* put_header_byte is used for the compressed output
2471 * - for the initial 4 bytes that can't overflow the buffer.
2473 #define put_header_byte(c) {outbuf[outcnt++]=(uch)(c);}
2475 /* ===========================================================================
2476 * Deflate in to out.
2477 * IN assertions: the input and output buffers are cleared.
2478 * The variables time_stamp and save_orig_name are initialized.
2480 static int zip(int in, int out)
2482 uch my_flags = 0; /* general purpose bit flags */
2483 ush attr = 0; /* ascii/binary flag */
2484 ush deflate_flags = 0; /* pkzip -es, -en or -ex equivalent */
2490 /* Write the header to the gzip file. See algorithm.doc for the format */
2494 put_header_byte(GZIP_MAGIC[0]); /* magic header */
2495 put_header_byte(GZIP_MAGIC[1]);
2496 put_header_byte(DEFLATED); /* compression method */
2498 put_header_byte(my_flags); /* general flags */
2499 put_long(time_stamp);
2501 /* Write deflated file to zip file */
2505 ct_init(&attr, &method);
2506 lm_init(&deflate_flags);
2508 put_byte((uch) deflate_flags); /* extra flags */
2509 put_byte(OS_CODE); /* OS identifier */
2511 header_bytes = (long) outcnt;
2515 /* Write the crc and uncompressed size */
2518 header_bytes += 2 * sizeof(long);
2525 /* ===========================================================================
2526 * Read a new buffer from the current input file, perform end-of-line
2527 * translation, and update the crc and input file size.
2528 * IN assertion: size >= 2 (for end-of-line translation)
2530 static int file_read(char *buf, unsigned size)
2534 Assert(insize == 0, "inbuf not empty");
2536 len = read(ifd, buf, size);
2537 if (len == (unsigned) (-1) || len == 0)
2540 crc = updcrc((uch *) buf, len);
2545 /* ===========================================================================
2546 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
2547 * (used for the compressed data only)
2549 static void flush_outbuf()
2554 write_buf(ofd, (char *) outbuf, outcnt);