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@codepoet.org> to support
13 * files as well as stdin/stdout, and to generally behave itself wrt
14 * command line handling.
16 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
19 // TODO: full support for -v for DESKTOP
28 #include <sys/types.h>
32 #include <sys/types.h>
39 typedef unsigned char uch;
40 typedef unsigned short ush;
41 typedef unsigned long ulg;
43 /* Return codes from gzip */
48 /* Compression methods (see algorithm.doc) */
49 /* Only STORED and DEFLATED are supported by this BusyBox module */
51 /* methods 4 to 7 reserved */
54 /* To save memory for 16 bit systems, some arrays are overlaid between
55 * the various modules:
56 * deflate: prev+head window d_buf l_buf outbuf
57 * unlzw: tab_prefix tab_suffix stack inbuf outbuf
58 * For compression, input is done in window[]. For decompression, output
59 * is done in window except for unlzw.
64 # define INBUFSIZ 0x2000 /* input buffer size */
66 # define INBUFSIZ 0x8000 /* input buffer size */
69 #define INBUF_EXTRA 64 /* required by unlzw() */
73 # define OUTBUFSIZ 8192 /* output buffer size */
75 # define OUTBUFSIZ 16384 /* output buffer size */
78 #define OUTBUF_EXTRA 2048 /* required by unlzw() */
82 # define DIST_BUFSIZE 0x2000 /* buffer for distances, see trees.c */
84 # define DIST_BUFSIZE 0x8000 /* buffer for distances, see trees.c */
88 # define DECLARE(type, array, size) static type * array
89 # define ALLOC(type, array, size) { \
90 array = (type*)xzalloc((size_t)(((size)+1L)/2) * 2*sizeof(type)); \
92 # define FREE(array) {free(array), array=NULL;}
94 #define tab_suffix window
95 #define tab_prefix prev /* hash link (see deflate.c) */
96 #define head (prev+WSIZE) /* hash head (see deflate.c) */
98 static long bytes_in; /* number of input bytes */
100 #define isize bytes_in
101 /* for compatibility with old zip sources (to be cleaned) */
103 typedef int file_t; /* Do not use stdio */
105 #define NO_FILE (-1) /* in memory compression */
108 #define PACK_MAGIC "\037\036" /* Magic header for packed files */
109 #define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */
110 #define OLD_GZIP_MAGIC "\037\236" /* Magic header for gzip 0.5 = freeze 1.x */
111 #define LZH_MAGIC "\037\240" /* Magic header for SCO LZH Compress files */
112 #define PKZIP_MAGIC "\120\113\003\004" /* Magic header for pkzip files */
115 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
116 #define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */
117 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
118 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
119 #define COMMENT 0x10 /* bit 4 set: file comment present */
120 #define RESERVED 0xC0 /* bit 6,7: reserved */
122 /* internal file attribute */
123 #define UNKNOWN 0xffff
128 # define WSIZE 0x8000 /* window size--must be a power of two, and */
129 #endif /* at least 32K for zip's deflate method */
132 #define MAX_MATCH 258
133 /* The minimum and maximum match lengths */
135 #define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1)
136 /* Minimum amount of lookahead, except at the end of the input file.
137 * See deflate.c for comments about the MIN_MATCH+1.
140 #define MAX_DIST (WSIZE-MIN_LOOKAHEAD)
141 /* In order to simplify the code, particularly on 16 bit machines, match
142 * distances are limited to MAX_DIST instead of WSIZE.
145 /* put_byte is used for the compressed output */
146 #define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\
149 #define seekable() 0 /* force sequential output */
150 #define translate_eol 0 /* no option -a yet */
152 /* Diagnostic functions */
154 # define Assert(cond,msg) {if(!(cond)) bb_error_msg(msg);}
155 # define Trace(x) fprintf x
156 # define Tracev(x) {if (verbose) fprintf x ;}
157 # define Tracevv(x) {if (verbose>1) fprintf x ;}
158 # define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
159 # define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
161 # define Assert(cond,msg)
166 # define Tracecv(c,x)
169 #define WARN(msg) {if (!quiet) fprintf msg ; \
170 if (exit_code == OK) exit_code = WARNING;}
173 # define MAX_PATH_LEN 1024 /* max pathname length */
178 static int zip(int in, int out);
179 static int file_read(char *buf, unsigned size);
182 static void lm_init(ush * flags);
183 static ulg deflate(void);
186 static void ct_init(ush * attr, int *methodp);
187 static int ct_tally(int dist, int lc);
188 static ulg flush_block(char *buf, ulg stored_len, int eof);
191 static void bi_init(file_t zipfile);
192 static void send_bits(int value, int length);
193 static unsigned bi_reverse(unsigned value, int length);
194 static void bi_windup(void);
195 static void copy_block(char *buf, unsigned len, int header);
196 static int (*read_buf) (char *buf, unsigned size);
199 static void flush_outbuf(void);
201 /* lzw.h -- define the lzw functions.
202 * Copyright (C) 1992-1993 Jean-loup Gailly.
203 * This is free software; you can redistribute it and/or modify it under the
204 * terms of the GNU General Public License, see the file COPYING.
210 #define INIT_BITS 9 /* Initial number of bits per code */
212 #define BIT_MASK 0x1f /* Mask for 'number of compression bits' */
213 /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free.
214 * It's a pity that old uncompress does not check bit 0x20. That makes
215 * extension of the format actually undesirable because old compress
216 * would just crash on the new format instead of giving a meaningful
217 * error message. It does check the number of bits, but it's more
218 * helpful to say "unsupported format, get a new version" than
219 * "can only handle 16 bits".
222 /* tailor.h -- target dependent definitions
223 * Copyright (C) 1992-1993 Jean-loup Gailly.
224 * This is free software; you can redistribute it and/or modify it under the
225 * terms of the GNU General Public License, see the file COPYING.
228 /* The target dependent definitions should be defined here only.
229 * The target dependent functions should be defined in tailor.c.
233 /* Common defaults */
236 # define OS_CODE 0x03 /* assume Unix */
240 # define PATH_SEP '/'
244 # define OPTIONS_VAR "GZIP"
248 # define Z_SUFFIX ".gz"
252 # define MAX_SUFFIX MAX_EXT_CHARS
254 # define MAX_SUFFIX 30
259 DECLARE(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
260 DECLARE(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
261 DECLARE(ush, d_buf, DIST_BUFSIZE);
262 DECLARE(uch, window, 2L * WSIZE);
263 DECLARE(ush, tab_prefix, 1L << BITS);
265 static int foreground; /* set if program run in foreground */
266 static int method = DEFLATED; /* compression method */
267 static int exit_code = OK; /* program exit code */
268 static long time_stamp; /* original time stamp (modification time) */
269 static char z_suffix[MAX_SUFFIX + 1]; /* default suffix (can be set with --suffix) */
271 static int ifd; /* input file descriptor */
272 static int ofd; /* output file descriptor */
274 static unsigned insize; /* valid bytes in inbuf */
276 static unsigned outcnt; /* bytes in output buffer */
278 static uint32_t *crc_32_tab;
280 /* Output a 16 bit value, lsb first */
281 static void put_short(ush w)
283 if (outcnt < OUTBUFSIZ - 2) {
284 outbuf[outcnt++] = (uch) ((w) & 0xff);
285 outbuf[outcnt++] = (uch) ((ush) (w) >> 8);
287 put_byte((uch) ((w) & 0xff));
288 put_byte((uch) ((ush) (w) >> 8));
292 /* ========================================================================
293 * Signal and error handler.
295 static void abort_gzip(int ATTRIBUTE_UNUSED ignored)
300 /* ===========================================================================
301 * Clear input and output buffers
303 static void clear_bufs(void)
312 /* ===========================================================================
313 * Does the same as write(), but also handles partial pipe writes and checks
316 static void write_buf(int fd, void *buf, unsigned cnt)
320 while ((n = write(fd, buf, cnt)) != cnt) {
321 if (n == (unsigned) (-1)) bb_error_msg_and_die(bb_msg_write_error);
323 buf = (void *) ((char *) buf + n);
327 /* ===========================================================================
328 * Run a set of bytes through the crc shift register. If s is a NULL
329 * pointer, then initialize the crc shift register contents instead.
330 * Return the current crc in either case.
332 static uint32_t updcrc(uch * s, unsigned n)
334 static uint32_t crc = ~0; /* shift register contents */
335 uint32_t c; /* temporary variable */
343 c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8);
350 /* bits.c -- output variable-length bit strings
351 * Copyright (C) 1992-1993 Jean-loup Gailly
352 * This is free software; you can redistribute it and/or modify it under the
353 * terms of the GNU General Public License, see the file COPYING.
360 * Output variable-length bit strings. Compression can be done
361 * to a file or to memory. (The latter is not supported in this version.)
365 * The PKZIP "deflate" file format interprets compressed file data
366 * as a sequence of bits. Multi-bit strings in the file may cross
367 * byte boundaries without restriction.
369 * The first bit of each byte is the low-order bit.
371 * The routines in this file allow a variable-length bit value to
372 * be output right-to-left (useful for literal values). For
373 * left-to-right output (useful for code strings from the tree routines),
374 * the bits must have been reversed first with bi_reverse().
376 * For in-memory compression, the compressed bit stream goes directly
377 * into the requested output buffer. The input data is read in blocks
378 * by the mem_read() function. The buffer is limited to 64K on 16 bit
383 * void bi_init (FILE *zipfile)
384 * Initialize the bit string routines.
386 * void send_bits (int value, int length)
387 * Write out a bit string, taking the source bits right to
390 * int bi_reverse (int value, int length)
391 * Reverse the bits of a bit string, taking the source bits left to
392 * right and emitting them right to left.
394 * void bi_windup (void)
395 * Write out any remaining bits in an incomplete byte.
397 * void copy_block(char *buf, unsigned len, int header)
398 * Copy a stored block to the zip file, storing first the length and
399 * its one's complement if requested.
403 /* ===========================================================================
404 * Local data used by the "bit string" routines.
407 static file_t zfile; /* output gzip file */
409 static unsigned short bi_buf;
411 /* Output buffer. bits are inserted starting at the bottom (least significant
415 #define Buf_size (8 * 2*sizeof(char))
416 /* Number of bits used within bi_buf. (bi_buf might be implemented on
417 * more than 16 bits on some systems.)
422 /* Current input function. Set to mem_read for in-memory compression */
425 ulg bits_sent; /* bit length of the compressed data */
428 /* ===========================================================================
429 * Initialize the bit string routines.
431 static void bi_init(file_t zipfile)
440 /* Set the defaults for file compression. They are set by memcompress
441 * for in-memory compression.
443 if (zfile != NO_FILE) {
444 read_buf = file_read;
448 /* ===========================================================================
449 * Send a value on a given number of bits.
450 * IN assertion: length <= 16 and value fits in length bits.
452 static void send_bits(int value, int length)
455 Tracev((stderr, " l %2d v %4x ", length, value));
456 Assert(length > 0 && length <= 15, "invalid length");
457 bits_sent += (ulg) length;
459 /* If not enough room in bi_buf, use (valid) bits from bi_buf and
460 * (16 - bi_valid) bits from value, leaving (width - (16-bi_valid))
461 * unused bits in value.
463 if (bi_valid > (int) Buf_size - length) {
464 bi_buf |= (value << bi_valid);
466 bi_buf = (ush) value >> (Buf_size - bi_valid);
467 bi_valid += length - Buf_size;
469 bi_buf |= value << bi_valid;
474 /* ===========================================================================
475 * Reverse the first len bits of a code, using straightforward code (a faster
476 * method would use a table)
477 * IN assertion: 1 <= len <= 15
479 static unsigned bi_reverse(unsigned code, int len)
485 code >>= 1, res <<= 1;
490 /* ===========================================================================
491 * Write out any remaining bits in an incomplete byte.
493 static void bi_windup(void)
497 } else if (bi_valid > 0) {
503 bits_sent = (bits_sent + 7) & ~7;
507 /* ===========================================================================
508 * Copy a stored block to the zip file, storing first the length and its
509 * one's complement if requested.
511 static void copy_block(char *buf, unsigned len, int header)
513 bi_windup(); /* align on byte boundary */
516 put_short((ush) len);
517 put_short((ush) ~ len);
523 bits_sent += (ulg) len << 3;
530 /* deflate.c -- compress data using the deflation algorithm
531 * Copyright (C) 1992-1993 Jean-loup Gailly
532 * This is free software; you can redistribute it and/or modify it under the
533 * terms of the GNU General Public License, see the file COPYING.
539 * Identify new text as repetitions of old text within a fixed-
540 * length sliding window trailing behind the new text.
544 * The "deflation" process depends on being able to identify portions
545 * of the input text which are identical to earlier input (within a
546 * sliding window trailing behind the input currently being processed).
548 * The most straightforward technique turns out to be the fastest for
549 * most input files: try all possible matches and select the longest.
550 * The key feature of this algorithm is that insertions into the string
551 * dictionary are very simple and thus fast, and deletions are avoided
552 * completely. Insertions are performed at each input character, whereas
553 * string matches are performed only when the previous match ends. So it
554 * is preferable to spend more time in matches to allow very fast string
555 * insertions and avoid deletions. The matching algorithm for small
556 * strings is inspired from that of Rabin & Karp. A brute force approach
557 * is used to find longer strings when a small match has been found.
558 * A similar algorithm is used in comic (by Jan-Mark Wams) and freeze
559 * (by Leonid Broukhis).
560 * A previous version of this file used a more sophisticated algorithm
561 * (by Fiala and Greene) which is guaranteed to run in linear amortized
562 * time, but has a larger average cost, uses more memory and is patented.
563 * However the F&G algorithm may be faster for some highly redundant
564 * files if the parameter max_chain_length (described below) is too large.
568 * The idea of lazy evaluation of matches is due to Jan-Mark Wams, and
569 * I found it in 'freeze' written by Leonid Broukhis.
570 * Thanks to many info-zippers for bug reports and testing.
574 * APPNOTE.TXT documentation file in PKZIP 1.93a distribution.
576 * A description of the Rabin and Karp algorithm is given in the book
577 * "Algorithms" by R. Sedgewick, Addison-Wesley, p252.
579 * Fiala,E.R., and Greene,D.H.
580 * Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595
584 * void lm_init (int pack_level, ush *flags)
585 * Initialize the "longest match" routines for a new file
588 * Processes a new input file and return its compressed length. Sets
589 * the compressed length, crc, deflate flags and internal file
594 /* ===========================================================================
595 * Configuration parameters
598 /* Compile with MEDIUM_MEM to reduce the memory requirements or
599 * with SMALL_MEM to use as little memory as possible. Use BIG_MEM if the
600 * entire input file can be held in memory (not possible on 16 bit systems).
601 * Warning: defining these symbols affects HASH_BITS (see below) and thus
602 * affects the compression ratio. The compressed output
603 * is still correct, and might even be smaller in some cases.
607 # define HASH_BITS 13 /* Number of bits used to hash strings */
610 # define HASH_BITS 14
613 # define HASH_BITS 15
614 /* For portability to 16 bit machines, do not use values above 15. */
617 /* To save space (see unlzw.c), we overlay prev+head with tab_prefix and
618 * window with tab_suffix. Check that we can do this:
620 #if (WSIZE<<1) > (1<<BITS)
621 # error cannot overlay window with tab_suffix and prev with tab_prefix0
623 #if HASH_BITS > BITS-1
624 # error cannot overlay head with tab_prefix1
626 #define HASH_SIZE (unsigned)(1<<HASH_BITS)
627 #define HASH_MASK (HASH_SIZE-1)
628 #define WMASK (WSIZE-1)
629 /* HASH_SIZE and WSIZE must be powers of two */
631 /* Tail of hash chains */
634 /* speed options for the general purpose bit flag */
636 # define TOO_FAR 4096
638 /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */
639 /* ===========================================================================
640 * Local data used by the "longest match" routines.
643 typedef unsigned IPos;
645 /* A Pos is an index in the character window. We use short instead of int to
646 * save space in the various tables. IPos is used only for parameter passing.
649 /* DECLARE(uch, window, 2L*WSIZE); */
650 /* Sliding window. Input bytes are read into the second half of the window,
651 * and move to the first half later to keep a dictionary of at least WSIZE
652 * bytes. With this organization, matches are limited to a distance of
653 * WSIZE-MAX_MATCH bytes, but this ensures that IO is always
654 * performed with a length multiple of the block size. Also, it limits
655 * the window size to 64K, which is quite useful on MSDOS.
656 * To do: limit the window size to WSIZE+BSZ if SMALL_MEM (the code would
657 * be less efficient).
660 /* DECLARE(Pos, prev, WSIZE); */
661 /* Link to older string with same hash index. To limit the size of this
662 * array to 64K, this link is maintained only for the last 32K strings.
663 * An index in this array is thus a window index modulo 32K.
666 /* DECLARE(Pos, head, 1<<HASH_BITS); */
667 /* Heads of the hash chains or NIL. */
669 static const ulg window_size = (ulg) 2 * WSIZE;
671 /* window size, 2*WSIZE except for MMAP or BIG_MEM, where it is the
672 * input file length plus MIN_LOOKAHEAD.
675 static long block_start;
677 /* window position at the beginning of the current output block. Gets
678 * negative when the window is moved backwards.
681 static unsigned ins_h; /* hash index of string to be inserted */
683 #define H_SHIFT ((HASH_BITS+MIN_MATCH-1)/MIN_MATCH)
684 /* Number of bits by which ins_h and del_h must be shifted at each
685 * input step. It must be such that after MIN_MATCH steps, the oldest
686 * byte no longer takes part in the hash key, that is:
687 * H_SHIFT * MIN_MATCH >= HASH_BITS
690 static unsigned int prev_length;
692 /* Length of the best match at previous step. Matches not greater than this
693 * are discarded. This is used in the lazy match evaluation.
696 static unsigned strstart; /* start of string to insert */
697 static unsigned match_start; /* start of matching string */
698 static int eofile; /* flag set at end of input file */
699 static unsigned lookahead; /* number of valid bytes ahead in window */
702 max_chain_length = 4096,
704 /* To speed up deflation, hash chains are never searched beyond this length.
705 * A higher limit improves compression ratio but degrades the speed.
708 max_lazy_match = 258,
710 /* Attempt to find a better match only when the current match is strictly
711 * smaller than this value. This mechanism is used only for compression
714 max_insert_length = max_lazy_match,
715 /* Insert new strings in the hash table only if the match length
716 * is not greater than this length. This saves time but degrades compression.
717 * max_insert_length is used only for compression levels <= 3.
722 /* Use a faster search when the previous match is longer than this */
725 /* Values for max_lazy_match, good_match and max_chain_length, depending on
726 * the desired pack level (0..9). The values given below have been tuned to
727 * exclude worst case performance for pathological files. Better values may be
728 * found for specific files.
731 nice_match = 258 /* Stop searching when current match exceeds this */
733 /* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4
734 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different
740 /* result of memcmp for equal strings */
742 /* ===========================================================================
743 * Prototypes for local functions.
745 static void fill_window(void);
747 static int longest_match(IPos cur_match);
750 static void check_match(IPos start, IPos match, int length);
753 /* ===========================================================================
754 * Update a hash value with the given input byte
755 * IN assertion: all calls to to UPDATE_HASH are made with consecutive
756 * input characters, so that a running hash key can be computed from the
757 * previous key instead of complete recalculation each time.
759 #define UPDATE_HASH(h,c) (h = (((h)<<H_SHIFT) ^ (c)) & HASH_MASK)
761 /* ===========================================================================
762 * Insert string s in the dictionary and set match_head to the previous head
763 * of the hash chain (the most recent string with same hash key). Return
764 * the previous length of the hash chain.
765 * IN assertion: all calls to to INSERT_STRING are made with consecutive
766 * input characters and the first MIN_MATCH bytes of s are valid
767 * (except for the last MIN_MATCH-1 bytes of the input file).
769 #define INSERT_STRING(s, match_head) \
770 (UPDATE_HASH(ins_h, window[(s) + MIN_MATCH-1]), \
771 prev[(s) & WMASK] = match_head = head[ins_h], \
774 /* ===========================================================================
775 * Initialize the "longest match" routines for a new file
777 static void lm_init(ush * flags)
781 /* Initialize the hash table. */
782 memset(head, 0, HASH_SIZE * sizeof(*head));
783 /* prev will be initialized on the fly */
786 /* ??? reduce max_chain_length for binary files */
791 lookahead = read_buf((char *) window,
792 sizeof(int) <= 2 ? (unsigned) WSIZE : 2 * WSIZE);
794 if (lookahead == 0 || lookahead == (unsigned) EOF) {
795 eofile = 1, lookahead = 0;
799 /* Make sure that we always have enough lookahead. This is important
800 * if input comes from a device such as a tty.
802 while (lookahead < MIN_LOOKAHEAD && !eofile)
806 for (j = 0; j < MIN_MATCH - 1; j++)
807 UPDATE_HASH(ins_h, window[j]);
808 /* If lookahead < MIN_MATCH, ins_h is garbage, but this is
809 * not important since only literal bytes will be emitted.
813 /* ===========================================================================
814 * Set match_start to the longest match starting at the given string and
815 * return its length. Matches shorter or equal to prev_length are discarded,
816 * in which case the result is equal to prev_length and match_start is
818 * IN assertions: cur_match is the head of the hash chain for the current
819 * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1
822 /* For MSDOS, OS/2 and 386 Unix, an optimized version is in match.asm or
823 * match.s. The code is functionally equivalent, so you can use the C version
826 static int longest_match(IPos cur_match)
828 unsigned chain_length = max_chain_length; /* max hash chain length */
829 uch *scan = window + strstart; /* current string */
830 uch *match; /* matched string */
831 int len; /* length of current match */
832 int best_len = prev_length; /* best match length so far */
834 strstart > (IPos) MAX_DIST ? strstart - (IPos) MAX_DIST : NIL;
835 /* Stop when cur_match becomes <= limit. To simplify the code,
836 * we prevent matches with the string of window index 0.
839 /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16.
840 * It is easy to get rid of this optimization if necessary.
842 #if HASH_BITS < 8 || MAX_MATCH != 258
843 # error Code too clever
845 uch *strend = window + strstart + MAX_MATCH;
846 uch scan_end1 = scan[best_len - 1];
847 uch scan_end = scan[best_len];
849 /* Do not waste too much time if we already have a good match: */
850 if (prev_length >= good_match) {
853 Assert(strstart <= window_size - MIN_LOOKAHEAD, "insufficient lookahead");
856 Assert(cur_match < strstart, "no future");
857 match = window + cur_match;
859 /* Skip to next match if the match length cannot increase
860 * or if the match length is less than 2:
862 if (match[best_len] != scan_end ||
863 match[best_len - 1] != scan_end1 ||
864 *match != *scan || *++match != scan[1])
867 /* The check at best_len-1 can be removed because it will be made
868 * again later. (This heuristic is not always a win.)
869 * It is not necessary to compare scan[2] and match[2] since they
870 * are always equal when the other bytes match, given that
871 * the hash keys are equal and that HASH_BITS >= 8.
875 /* We check for insufficient lookahead only every 8th comparison;
876 * the 256th check will be made at strstart+258.
879 } while (*++scan == *++match && *++scan == *++match &&
880 *++scan == *++match && *++scan == *++match &&
881 *++scan == *++match && *++scan == *++match &&
882 *++scan == *++match && *++scan == *++match && scan < strend);
884 len = MAX_MATCH - (int) (strend - scan);
885 scan = strend - MAX_MATCH;
887 if (len > best_len) {
888 match_start = cur_match;
890 if (len >= nice_match)
892 scan_end1 = scan[best_len - 1];
893 scan_end = scan[best_len];
895 } while ((cur_match = prev[cur_match & WMASK]) > limit
896 && --chain_length != 0);
902 /* ===========================================================================
903 * Check that the match at match_start is indeed a match.
905 static void check_match(IPos start, IPos match, int length)
907 /* check that the match is indeed a match */
908 if (memcmp((char *) window + match,
909 (char *) window + start, length) != EQUAL) {
910 bb_error_msg(" start %d, match %d, length %d", start, match, length);
911 bb_error_msg("invalid match");
914 bb_error_msg("\\[%d,%d]", start - match, length);
916 putc(window[start++], stderr);
917 } while (--length != 0);
921 # define check_match(start, match, length)
924 /* ===========================================================================
925 * Fill the window when the lookahead becomes insufficient.
926 * Updates strstart and lookahead, and sets eofile if end of input file.
927 * IN assertion: lookahead < MIN_LOOKAHEAD && strstart + lookahead > 0
928 * OUT assertions: at least one byte has been read, or eofile is set;
929 * file reads are performed for at least two bytes (required for the
930 * translate_eol option).
932 static void fill_window(void)
936 (unsigned) (window_size - (ulg) lookahead - (ulg) strstart);
937 /* Amount of free space at the end of the window. */
939 /* If the window is almost full and there is insufficient lookahead,
940 * move the upper half to the lower one to make room in the upper half.
942 if (more == (unsigned) EOF) {
943 /* Very unlikely, but possible on 16 bit machine if strstart == 0
944 * and lookahead == 1 (input done one byte at time)
947 } else if (strstart >= WSIZE + MAX_DIST) {
948 /* By the IN assertion, the window is not empty so we can't confuse
949 * more == 0 with more == 64K on a 16 bit machine.
951 Assert(window_size == (ulg) 2 * WSIZE, "no sliding with BIG_MEM");
953 memcpy((char *) window, (char *) window + WSIZE, (unsigned) WSIZE);
954 match_start -= WSIZE;
955 strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */
957 block_start -= (long) WSIZE;
959 for (n = 0; n < HASH_SIZE; n++) {
961 head[n] = (Pos) (m >= WSIZE ? m - WSIZE : NIL);
963 for (n = 0; n < WSIZE; n++) {
965 prev[n] = (Pos) (m >= WSIZE ? m - WSIZE : NIL);
966 /* If n is not on any hash chain, prev[n] is garbage but
967 * its value will never be used.
972 /* At this point, more >= 2 */
974 n = read_buf((char *) window + strstart + lookahead, more);
975 if (n == 0 || n == (unsigned) EOF) {
983 /* ===========================================================================
984 * Flush the current block, with given end-of-file flag.
985 * IN assertion: strstart is set to the end of the current match.
987 #define FLUSH_BLOCK(eof) \
988 flush_block(block_start >= 0L ? (char*)&window[(unsigned)block_start] : \
989 (char*)NULL, (long)strstart - block_start, (eof))
991 /* ===========================================================================
992 * Same as above, but achieves better compression. We use a lazy
993 * evaluation for matches: a match is finally adopted only if there is
994 * no better match at the next window position.
996 static ulg deflate(void)
998 IPos hash_head; /* head of hash chain */
999 IPos prev_match; /* previous match */
1000 int flush; /* set if current block must be flushed */
1001 int match_available = 0; /* set if previous match exists */
1002 unsigned match_length = MIN_MATCH - 1; /* length of best match */
1004 /* Process the input block. */
1005 while (lookahead != 0) {
1006 /* Insert the string window[strstart .. strstart+2] in the
1007 * dictionary, and set hash_head to the head of the hash chain:
1009 INSERT_STRING(strstart, hash_head);
1011 /* Find the longest match, discarding those <= prev_length.
1013 prev_length = match_length, prev_match = match_start;
1014 match_length = MIN_MATCH - 1;
1016 if (hash_head != NIL && prev_length < max_lazy_match &&
1017 strstart - hash_head <= MAX_DIST) {
1018 /* To simplify the code, we prevent matches with the string
1019 * of window index 0 (in particular we have to avoid a match
1020 * of the string with itself at the start of the input file).
1022 match_length = longest_match(hash_head);
1023 /* longest_match() sets match_start */
1024 if (match_length > lookahead)
1025 match_length = lookahead;
1027 /* Ignore a length 3 match if it is too distant: */
1028 if (match_length == MIN_MATCH && strstart - match_start > TOO_FAR) {
1029 /* If prev_match is also MIN_MATCH, match_start is garbage
1030 * but we will ignore the current match anyway.
1035 /* If there was a match at the previous step and the current
1036 * match is not better, output the previous match:
1038 if (prev_length >= MIN_MATCH && match_length <= prev_length) {
1040 check_match(strstart - 1, prev_match, prev_length);
1043 ct_tally(strstart - 1 - prev_match, prev_length - MIN_MATCH);
1045 /* Insert in hash table all strings up to the end of the match.
1046 * strstart-1 and strstart are already inserted.
1048 lookahead -= prev_length - 1;
1052 INSERT_STRING(strstart, hash_head);
1053 /* strstart never exceeds WSIZE-MAX_MATCH, so there are
1054 * always MIN_MATCH bytes ahead. If lookahead < MIN_MATCH
1055 * these bytes are garbage, but it does not matter since the
1056 * next lookahead bytes will always be emitted as literals.
1058 } while (--prev_length != 0);
1059 match_available = 0;
1060 match_length = MIN_MATCH - 1;
1063 FLUSH_BLOCK(0), block_start = strstart;
1065 } else if (match_available) {
1066 /* If there was no match at the previous position, output a
1067 * single literal. If there was a match but the current match
1068 * is longer, truncate the previous match to a single literal.
1070 Tracevv((stderr, "%c", window[strstart - 1]));
1071 if (ct_tally(0, window[strstart - 1])) {
1072 FLUSH_BLOCK(0), block_start = strstart;
1077 /* There is no previous match to compare with, wait for
1078 * the next step to decide.
1080 match_available = 1;
1084 Assert(strstart <= isize && lookahead <= isize, "a bit too far");
1086 /* Make sure that we always have enough lookahead, except
1087 * at the end of the input file. We need MAX_MATCH bytes
1088 * for the next match, plus MIN_MATCH bytes to insert the
1089 * string following the next match.
1091 while (lookahead < MIN_LOOKAHEAD && !eofile)
1094 if (match_available)
1095 ct_tally(0, window[strstart - 1]);
1097 return FLUSH_BLOCK(1); /* eof */
1100 /* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface
1101 * Copyright (C) 1992-1993 Jean-loup Gailly
1102 * The unzip code was written and put in the public domain by Mark Adler.
1103 * Portions of the lzw code are derived from the public domain 'compress'
1104 * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies,
1105 * Ken Turkowski, Dave Mack and Peter Jannesen.
1107 * See the license_msg below and the file COPYING for the software license.
1108 * See the file algorithm.doc for the compression algorithms and file formats.
1111 /* Compress files with zip algorithm and 'compress' interface.
1112 * See usage() and help() functions below for all options.
1114 * file.gz: compressed file with same mode, owner, and utimes
1115 * or stdout with -c option or if stdin used as input.
1116 * If the output file name had to be truncated, the original name is kept
1117 * in the compressed file.
1122 typedef struct dirent dir_type;
1124 /* ======================================================================== */
1125 int gzip_main(int argc, char **argv)
1136 struct stat statBuf;
1139 opt = bb_getopt_ulflags(argc, argv, "cf123456789qv" USE_GUNZIP("d"));
1140 //if (opt & 0x1) // -c
1141 //if (opt & 0x2) // -f
1142 /* Ignore 1-9 (compression level) options */
1143 //if (opt & 0x4) // -1
1144 //if (opt & 0x8) // -2
1145 //if (opt & 0x10) // -3
1146 //if (opt & 0x20) // -4
1147 //if (opt & 0x40) // -5
1148 //if (opt & 0x80) // -6
1149 //if (opt & 0x100) // -7
1150 //if (opt & 0x200) // -8
1151 //if (opt & 0x400) // -9
1152 //if (opt & 0x800) // -q
1153 //if (opt & 0x1000) // -v
1154 if (ENABLE_GUNZIP && (opt & 0x2000)) { // -d
1155 /* FIXME: bb_getopt_ulflags should not depend on optind */
1157 return gunzip_main(argc, argv);
1160 foreground = signal(SIGINT, SIG_IGN) != SIG_IGN;
1162 (void) signal(SIGINT, abort_gzip);
1165 if (signal(SIGTERM, SIG_IGN) != SIG_IGN) {
1166 (void) signal(SIGTERM, abort_gzip);
1170 if (signal(SIGHUP, SIG_IGN) != SIG_IGN) {
1171 (void) signal(SIGHUP, abort_gzip);
1175 strncpy(z_suffix, Z_SUFFIX, sizeof(z_suffix) - 1);
1177 /* Allocate all global buffers (for DYN_ALLOC option) */
1178 ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA);
1179 ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA);
1180 ALLOC(ush, d_buf, DIST_BUFSIZE);
1181 ALLOC(uch, window, 2L * WSIZE);
1182 ALLOC(ush, tab_prefix, 1L << BITS);
1184 /* Initialise the CRC32 table */
1185 crc_32_tab = crc32_filltable(0);
1189 if (optind == argc) {
1191 zip(STDIN_FILENO, STDOUT_FILENO);
1195 for (i = optind; i < argc; i++) {
1199 if (strcmp(argv[i], "-") == 0) {
1201 inFileNum = STDIN_FILENO;
1202 outFileNum = STDOUT_FILENO;
1204 inFileNum = xopen(argv[i], O_RDONLY);
1205 if (fstat(inFileNum, &statBuf) < 0)
1206 bb_perror_msg_and_die("%s", argv[i]);
1207 time_stamp = statBuf.st_ctime;
1209 if (!(opt & OPT_tostdout)) {
1210 path = xasprintf("%s.gz", argv[i]);
1212 /* Open output file */
1213 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1) && defined O_NOFOLLOW
1215 open(path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW);
1217 outFileNum = open(path, O_RDWR | O_CREAT | O_EXCL);
1219 if (outFileNum < 0) {
1220 bb_perror_msg("%s", path);
1225 /* Set permissions on the file */
1226 fchmod(outFileNum, statBuf.st_mode);
1228 outFileNum = STDOUT_FILENO;
1231 if (path == NULL && isatty(outFileNum) && !(opt & OPT_force)) {
1233 ("compressed data not written to a terminal. Use -f to force compression.");
1238 result = zip(inFileNum, outFileNum);
1244 /* Delete the original file */
1246 delFileName = argv[i];
1250 if (unlink(delFileName) < 0)
1251 bb_perror_msg("%s", delFileName);
1261 /* trees.c -- output deflated data using Huffman coding
1262 * Copyright (C) 1992-1993 Jean-loup Gailly
1263 * This is free software; you can redistribute it and/or modify it under the
1264 * terms of the GNU General Public License, see the file COPYING.
1270 * Encode various sets of source values using variable-length
1271 * binary code trees.
1275 * The PKZIP "deflation" process uses several Huffman trees. The more
1276 * common source values are represented by shorter bit sequences.
1278 * Each code tree is stored in the ZIP file in a compressed form
1279 * which is itself a Huffman encoding of the lengths of
1280 * all the code strings (in ascending order by source values).
1281 * The actual code strings are reconstructed from the lengths in
1282 * the UNZIP process, as described in the "application note"
1283 * (APPNOTE.TXT) distributed as part of PKWARE's PKZIP program.
1288 * Data Compression: Techniques and Applications, pp. 53-55.
1289 * Lifetime Learning Publications, 1985. ISBN 0-534-03418-7.
1292 * Data Compression: Methods and Theory, pp. 49-50.
1293 * Computer Science Press, 1988. ISBN 0-7167-8156-5.
1297 * Addison-Wesley, 1983. ISBN 0-201-06672-6.
1301 * void ct_init (ush *attr, int *methodp)
1302 * Allocate the match buffer, initialize the various tables and save
1303 * the location of the internal file attribute (ascii/binary) and
1304 * method (DEFLATE/STORE)
1306 * void ct_tally (int dist, int lc);
1307 * Save the match info and tally the frequency counts.
1309 * long flush_block (char *buf, ulg stored_len, int eof)
1310 * Determine the best encoding for the current block: dynamic trees,
1311 * static trees or store, and output the encoded block to the zip
1312 * file. Returns the total compressed length for the file so far.
1316 /* ===========================================================================
1321 /* All codes must not exceed MAX_BITS bits */
1323 #define MAX_BL_BITS 7
1324 /* Bit length codes must not exceed MAX_BL_BITS bits */
1326 #define LENGTH_CODES 29
1327 /* number of length codes, not counting the special END_BLOCK code */
1329 #define LITERALS 256
1330 /* number of literal bytes 0..255 */
1332 #define END_BLOCK 256
1333 /* end of block literal code */
1335 #define L_CODES (LITERALS+1+LENGTH_CODES)
1336 /* number of Literal or Length codes, including the END_BLOCK code */
1339 /* number of distance codes */
1342 /* number of codes used to transfer the bit lengths */
1344 typedef uch extra_bits_t;
1346 /* extra bits for each length code */
1347 static const extra_bits_t extra_lbits[LENGTH_CODES]
1348 = { 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4,
1352 /* extra bits for each distance code */
1353 static const extra_bits_t extra_dbits[D_CODES]
1354 = { 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9,
1355 10, 10, 11, 11, 12, 12, 13, 13
1358 /* extra bits for each bit length code */
1359 static const extra_bits_t extra_blbits[BL_CODES]
1360 = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 3, 7 };
1362 #define STORED_BLOCK 0
1363 #define STATIC_TREES 1
1365 /* The three kinds of block type */
1369 # define LIT_BUFSIZE 0x2000
1372 # define LIT_BUFSIZE 0x4000
1374 # define LIT_BUFSIZE 0x8000
1378 #ifndef DIST_BUFSIZE
1379 # define DIST_BUFSIZE LIT_BUFSIZE
1381 /* Sizes of match buffers for literals/lengths and distances. There are
1382 * 4 reasons for limiting LIT_BUFSIZE to 64K:
1383 * - frequencies can be kept in 16 bit counters
1384 * - if compression is not successful for the first block, all input data is
1385 * still in the window so we can still emit a stored block even when input
1386 * comes from standard input. (This can also be done for all blocks if
1387 * LIT_BUFSIZE is not greater than 32K.)
1388 * - if compression is not successful for a file smaller than 64K, we can
1389 * even emit a stored file instead of a stored block (saving 5 bytes).
1390 * - creating new Huffman trees less frequently may not provide fast
1391 * adaptation to changes in the input data statistics. (Take for
1392 * example a binary file with poorly compressible code followed by
1393 * a highly compressible string table.) Smaller buffer sizes give
1394 * fast adaptation but have of course the overhead of transmitting trees
1396 * - I can't count above 4
1397 * The current code is general and allows DIST_BUFSIZE < LIT_BUFSIZE (to save
1398 * memory at the expense of compression). Some optimizations would be possible
1399 * if we rely on DIST_BUFSIZE == LIT_BUFSIZE.
1401 #if LIT_BUFSIZE > INBUFSIZ
1402 #error cannot overlay l_buf and inbuf
1405 /* repeat previous bit length 3-6 times (2 bits of repeat count) */
1406 #define REPZ_3_10 17
1407 /* repeat a zero length 3-10 times (3 bits of repeat count) */
1408 #define REPZ_11_138 18
1409 /* repeat a zero length 11-138 times (7 bits of repeat count) */
1411 /* ===========================================================================
1415 /* Data structure describing a single value and its code string. */
1416 typedef struct ct_data {
1418 ush freq; /* frequency count */
1419 ush code; /* bit string */
1422 ush dad; /* father node in Huffman tree */
1423 ush len; /* length of bit string */
1427 #define Freq fc.freq
1428 #define Code fc.code
1432 #define HEAP_SIZE (2*L_CODES+1)
1433 /* maximum heap size */
1435 static ct_data dyn_ltree[HEAP_SIZE]; /* literal and length tree */
1436 static ct_data dyn_dtree[2 * D_CODES + 1]; /* distance tree */
1438 static ct_data static_ltree[L_CODES + 2];
1440 /* The static literal tree. Since the bit lengths are imposed, there is no
1441 * need for the L_CODES extra codes used during heap construction. However
1442 * The codes 286 and 287 are needed to build a canonical tree (see ct_init
1446 static ct_data static_dtree[D_CODES];
1448 /* The static distance tree. (Actually a trivial tree since all codes use
1452 static ct_data bl_tree[2 * BL_CODES + 1];
1454 /* Huffman tree for the bit lengths */
1456 typedef struct tree_desc {
1457 ct_data *dyn_tree; /* the dynamic tree */
1458 ct_data *static_tree; /* corresponding static tree or NULL */
1459 const extra_bits_t *extra_bits; /* extra bits for each code or NULL */
1460 int extra_base; /* base index for extra_bits */
1461 int elems; /* max number of elements in the tree */
1462 int max_length; /* max bit length for the codes */
1463 int max_code; /* largest code with non zero frequency */
1466 static tree_desc l_desc =
1467 { dyn_ltree, static_ltree, extra_lbits, LITERALS + 1, L_CODES,
1471 static tree_desc d_desc =
1472 { dyn_dtree, static_dtree, extra_dbits, 0, D_CODES, MAX_BITS, 0 };
1474 static tree_desc bl_desc =
1475 { bl_tree, (ct_data *) 0, extra_blbits, 0, BL_CODES, MAX_BL_BITS,
1480 static ush bl_count[MAX_BITS + 1];
1482 /* number of codes at each bit length for an optimal tree */
1484 static const uch bl_order[BL_CODES]
1485 = { 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15 };
1487 /* The lengths of the bit length codes are sent in order of decreasing
1488 * probability, to avoid transmitting the lengths for unused bit length codes.
1491 static int heap[2 * L_CODES + 1]; /* heap used to build the Huffman trees */
1492 static int heap_len; /* number of elements in the heap */
1493 static int heap_max; /* element of largest frequency */
1495 /* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used.
1496 * The same heap array is used to build all trees.
1499 static uch depth[2 * L_CODES + 1];
1501 /* Depth of each subtree used as tie breaker for trees of equal frequency */
1503 static uch length_code[MAX_MATCH - MIN_MATCH + 1];
1505 /* length code for each normalized match length (0 == MIN_MATCH) */
1507 static uch dist_code[512];
1509 /* distance codes. The first 256 values correspond to the distances
1510 * 3 .. 258, the last 256 values correspond to the top 8 bits of
1511 * the 15 bit distances.
1514 static int base_length[LENGTH_CODES];
1516 /* First normalized length for each code (0 = MIN_MATCH) */
1518 static int base_dist[D_CODES];
1520 /* First normalized distance for each code (0 = distance of 1) */
1523 /* DECLARE(uch, l_buf, LIT_BUFSIZE); buffer for literals or lengths */
1525 /* DECLARE(ush, d_buf, DIST_BUFSIZE); buffer for distances */
1527 static uch flag_buf[(LIT_BUFSIZE / 8)];
1529 /* flag_buf is a bit array distinguishing literals from lengths in
1530 * l_buf, thus indicating the presence or absence of a distance.
1533 static unsigned last_lit; /* running index in l_buf */
1534 static unsigned last_dist; /* running index in d_buf */
1535 static unsigned last_flags; /* running index in flag_buf */
1536 static uch flags; /* current flags not yet saved in flag_buf */
1537 static uch flag_bit; /* current bit used in flags */
1539 /* bits are filled in flags starting at bit 0 (least significant).
1540 * Note: these flags are overkill in the current code since we don't
1541 * take advantage of DIST_BUFSIZE == LIT_BUFSIZE.
1544 static ulg opt_len; /* bit length of current block with optimal trees */
1545 static ulg static_len; /* bit length of current block with static trees */
1547 static ulg compressed_len; /* total bit length of compressed file */
1550 static ush *file_type; /* pointer to UNKNOWN, BINARY or ASCII */
1551 static int *file_method; /* pointer to DEFLATE or STORE */
1553 /* ===========================================================================
1554 * Local (static) routines in this file.
1557 static void init_block(void);
1558 static void pqdownheap(ct_data * tree, int k);
1559 static void gen_bitlen(tree_desc * desc);
1560 static void gen_codes(ct_data * tree, int max_code);
1561 static void build_tree(tree_desc * desc);
1562 static void scan_tree(ct_data * tree, int max_code);
1563 static void send_tree(ct_data * tree, int max_code);
1564 static int build_bl_tree(void);
1565 static void send_all_trees(int lcodes, int dcodes, int blcodes);
1566 static void compress_block(ct_data * ltree, ct_data * dtree);
1567 static void set_file_type(void);
1571 # define send_code(c, tree) send_bits(tree[c].Code, tree[c].Len)
1572 /* Send a code of the given tree. c and tree must not have side effects */
1575 # define send_code(c, tree) \
1576 { if (verbose>1) bb_error_msg("\ncd %3d ",(c)); \
1577 send_bits(tree[c].Code, tree[c].Len); }
1580 #define d_code(dist) \
1581 ((dist) < 256 ? dist_code[dist] : dist_code[256+((dist)>>7)])
1582 /* Mapping from a distance to a distance code. dist is the distance - 1 and
1583 * must not have side effects. dist_code[256] and dist_code[257] are never
1587 /* the arguments must not have side effects */
1589 /* ===========================================================================
1590 * Allocate the match buffer, initialize the various tables and save the
1591 * location of the internal file attribute (ascii/binary) and method
1594 static void ct_init(ush * attr, int *methodp)
1596 int n; /* iterates over tree elements */
1597 int bits; /* bit counter */
1598 int length; /* length value */
1599 int code; /* code value */
1600 int dist; /* distance index */
1603 file_method = methodp;
1604 compressed_len = 0L;
1606 if (static_dtree[0].Len != 0)
1607 return; /* ct_init already called */
1609 /* Initialize the mapping length (0..255) -> length code (0..28) */
1611 for (code = 0; code < LENGTH_CODES - 1; code++) {
1612 base_length[code] = length;
1613 for (n = 0; n < (1 << extra_lbits[code]); n++) {
1614 length_code[length++] = (uch) code;
1617 Assert(length == 256, "ct_init: length != 256");
1618 /* Note that the length 255 (match length 258) can be represented
1619 * in two different ways: code 284 + 5 bits or code 285, so we
1620 * overwrite length_code[255] to use the best encoding:
1622 length_code[length - 1] = (uch) code;
1624 /* Initialize the mapping dist (0..32K) -> dist code (0..29) */
1626 for (code = 0; code < 16; code++) {
1627 base_dist[code] = dist;
1628 for (n = 0; n < (1 << extra_dbits[code]); n++) {
1629 dist_code[dist++] = (uch) code;
1632 Assert(dist == 256, "ct_init: dist != 256");
1633 dist >>= 7; /* from now on, all distances are divided by 128 */
1634 for (; code < D_CODES; code++) {
1635 base_dist[code] = dist << 7;
1636 for (n = 0; n < (1 << (extra_dbits[code] - 7)); n++) {
1637 dist_code[256 + dist++] = (uch) code;
1640 Assert(dist == 256, "ct_init: 256+dist != 512");
1642 /* Construct the codes of the static literal tree */
1643 for (bits = 0; bits <= MAX_BITS; bits++)
1647 static_ltree[n++].Len = 8, bl_count[8]++;
1649 static_ltree[n++].Len = 9, bl_count[9]++;
1651 static_ltree[n++].Len = 7, bl_count[7]++;
1653 static_ltree[n++].Len = 8, bl_count[8]++;
1654 /* Codes 286 and 287 do not exist, but we must include them in the
1655 * tree construction to get a canonical Huffman tree (longest code
1658 gen_codes((ct_data *) static_ltree, L_CODES + 1);
1660 /* The static distance tree is trivial: */
1661 for (n = 0; n < D_CODES; n++) {
1662 static_dtree[n].Len = 5;
1663 static_dtree[n].Code = bi_reverse(n, 5);
1666 /* Initialize the first block of the first file: */
1670 /* ===========================================================================
1671 * Initialize a new block.
1673 static void init_block(void)
1675 int n; /* iterates over tree elements */
1677 /* Initialize the trees. */
1678 for (n = 0; n < L_CODES; n++)
1679 dyn_ltree[n].Freq = 0;
1680 for (n = 0; n < D_CODES; n++)
1681 dyn_dtree[n].Freq = 0;
1682 for (n = 0; n < BL_CODES; n++)
1683 bl_tree[n].Freq = 0;
1685 dyn_ltree[END_BLOCK].Freq = 1;
1686 opt_len = static_len = 0L;
1687 last_lit = last_dist = last_flags = 0;
1693 /* Index within the heap array of least frequent node in the Huffman tree */
1696 /* ===========================================================================
1697 * Remove the smallest element from the heap and recreate the heap with
1698 * one less element. Updates heap and heap_len.
1700 #define pqremove(tree, top) \
1702 top = heap[SMALLEST]; \
1703 heap[SMALLEST] = heap[heap_len--]; \
1704 pqdownheap(tree, SMALLEST); \
1707 /* ===========================================================================
1708 * Compares to subtrees, using the tree depth as tie breaker when
1709 * the subtrees have equal frequency. This minimizes the worst case length.
1711 #define smaller(tree, n, m) \
1712 (tree[n].Freq < tree[m].Freq || \
1713 (tree[n].Freq == tree[m].Freq && depth[n] <= depth[m]))
1715 /* ===========================================================================
1716 * Restore the heap property by moving down the tree starting at node k,
1717 * exchanging a node with the smallest of its two sons if necessary, stopping
1718 * when the heap property is re-established (each father smaller than its
1721 static void pqdownheap(ct_data * tree, int k)
1724 int j = k << 1; /* left son of k */
1726 while (j <= heap_len) {
1727 /* Set j to the smallest of the two sons: */
1728 if (j < heap_len && smaller(tree, heap[j + 1], heap[j]))
1731 /* Exit if v is smaller than both sons */
1732 if (smaller(tree, v, heap[j]))
1735 /* Exchange v with the smallest son */
1739 /* And continue down the tree, setting j to the left son of k */
1745 /* ===========================================================================
1746 * Compute the optimal bit lengths for a tree and update the total bit length
1747 * for the current block.
1748 * IN assertion: the fields freq and dad are set, heap[heap_max] and
1749 * above are the tree nodes sorted by increasing frequency.
1750 * OUT assertions: the field len is set to the optimal bit length, the
1751 * array bl_count contains the frequencies for each bit length.
1752 * The length opt_len is updated; static_len is also updated if stree is
1755 static void gen_bitlen(tree_desc * desc)
1757 ct_data *tree = desc->dyn_tree;
1758 const extra_bits_t *extra = desc->extra_bits;
1759 int base = desc->extra_base;
1760 int max_code = desc->max_code;
1761 int max_length = desc->max_length;
1762 ct_data *stree = desc->static_tree;
1763 int h; /* heap index */
1764 int n, m; /* iterate over the tree elements */
1765 int bits; /* bit length */
1766 int xbits; /* extra bits */
1767 ush f; /* frequency */
1768 int overflow = 0; /* number of elements with bit length too large */
1770 for (bits = 0; bits <= MAX_BITS; bits++)
1773 /* In a first pass, compute the optimal bit lengths (which may
1774 * overflow in the case of the bit length tree).
1776 tree[heap[heap_max]].Len = 0; /* root of the heap */
1778 for (h = heap_max + 1; h < HEAP_SIZE; h++) {
1780 bits = tree[tree[n].Dad].Len + 1;
1781 if (bits > max_length)
1782 bits = max_length, overflow++;
1783 tree[n].Len = (ush) bits;
1784 /* We overwrite tree[n].Dad which is no longer needed */
1787 continue; /* not a leaf node */
1792 xbits = extra[n - base];
1794 opt_len += (ulg) f *(bits + xbits);
1797 static_len += (ulg) f *(stree[n].Len + xbits);
1802 Trace((stderr, "\nbit length overflow\n"));
1803 /* This happens for example on obj2 and pic of the Calgary corpus */
1805 /* Find the first bit length which could increase: */
1807 bits = max_length - 1;
1808 while (bl_count[bits] == 0)
1810 bl_count[bits]--; /* move one leaf down the tree */
1811 bl_count[bits + 1] += 2; /* move one overflow item as its brother */
1812 bl_count[max_length]--;
1813 /* The brother of the overflow item also moves one step up,
1814 * but this does not affect bl_count[max_length]
1817 } while (overflow > 0);
1819 /* Now recompute all bit lengths, scanning in increasing frequency.
1820 * h is still equal to HEAP_SIZE. (It is simpler to reconstruct all
1821 * lengths instead of fixing only the wrong ones. This idea is taken
1822 * from 'ar' written by Haruhiko Okumura.)
1824 for (bits = max_length; bits != 0; bits--) {
1830 if (tree[m].Len != (unsigned) bits) {
1831 Trace((stderr, "code %d bits %d->%d\n", m, tree[m].Len,
1834 ((long) bits - (long) tree[m].Len) * (long) tree[m].Freq;
1835 tree[m].Len = (ush) bits;
1842 /* ===========================================================================
1843 * Generate the codes for a given tree and bit counts (which need not be
1845 * IN assertion: the array bl_count contains the bit length statistics for
1846 * the given tree and the field len is set for all tree elements.
1847 * OUT assertion: the field code is set for all tree elements of non
1850 static void gen_codes(ct_data * tree, int max_code)
1852 ush next_code[MAX_BITS + 1]; /* next code value for each bit length */
1853 ush code = 0; /* running code value */
1854 int bits; /* bit index */
1855 int n; /* code index */
1857 /* The distribution counts are first used to generate the code values
1858 * without bit reversal.
1860 for (bits = 1; bits <= MAX_BITS; bits++) {
1861 next_code[bits] = code = (code + bl_count[bits - 1]) << 1;
1863 /* Check that the bit counts in bl_count are consistent. The last code
1866 Assert(code + bl_count[MAX_BITS] - 1 == (1 << MAX_BITS) - 1,
1867 "inconsistent bit counts");
1868 Tracev((stderr, "\ngen_codes: max_code %d ", max_code));
1870 for (n = 0; n <= max_code; n++) {
1871 int len = tree[n].Len;
1875 /* Now reverse the bits */
1876 tree[n].Code = bi_reverse(next_code[len]++, len);
1878 Tracec(tree != static_ltree,
1879 (stderr, "\nn %3d %c l %2d c %4x (%x) ", n,
1880 (isgraph(n) ? n : ' '), len, tree[n].Code,
1881 next_code[len] - 1));
1885 /* ===========================================================================
1886 * Construct one Huffman tree and assigns the code bit strings and lengths.
1887 * Update the total bit length for the current block.
1888 * IN assertion: the field freq is set for all tree elements.
1889 * OUT assertions: the fields len and code are set to the optimal bit length
1890 * and corresponding code. The length opt_len is updated; static_len is
1891 * also updated if stree is not null. The field max_code is set.
1893 static void build_tree(tree_desc * desc)
1895 ct_data *tree = desc->dyn_tree;
1896 ct_data *stree = desc->static_tree;
1897 int elems = desc->elems;
1898 int n, m; /* iterate over heap elements */
1899 int max_code = -1; /* largest code with non zero frequency */
1900 int node = elems; /* next internal node of the tree */
1902 /* Construct the initial heap, with least frequent element in
1903 * heap[SMALLEST]. The sons of heap[n] are heap[2*n] and heap[2*n+1].
1904 * heap[0] is not used.
1906 heap_len = 0, heap_max = HEAP_SIZE;
1908 for (n = 0; n < elems; n++) {
1909 if (tree[n].Freq != 0) {
1910 heap[++heap_len] = max_code = n;
1917 /* The pkzip format requires that at least one distance code exists,
1918 * and that at least one bit should be sent even if there is only one
1919 * possible code. So to avoid special checks later on we force at least
1920 * two codes of non zero frequency.
1922 while (heap_len < 2) {
1923 int new = heap[++heap_len] = (max_code < 2 ? ++max_code : 0);
1929 static_len -= stree[new].Len;
1930 /* new is 0 or 1 so it does not have extra bits */
1932 desc->max_code = max_code;
1934 /* The elements heap[heap_len/2+1 .. heap_len] are leaves of the tree,
1935 * establish sub-heaps of increasing lengths:
1937 for (n = heap_len / 2; n >= 1; n--)
1938 pqdownheap(tree, n);
1940 /* Construct the Huffman tree by repeatedly combining the least two
1944 pqremove(tree, n); /* n = node of least frequency */
1945 m = heap[SMALLEST]; /* m = node of next least frequency */
1947 heap[--heap_max] = n; /* keep the nodes sorted by frequency */
1948 heap[--heap_max] = m;
1950 /* Create a new node father of n and m */
1951 tree[node].Freq = tree[n].Freq + tree[m].Freq;
1952 depth[node] = (uch) (MAX(depth[n], depth[m]) + 1);
1953 tree[n].Dad = tree[m].Dad = (ush) node;
1955 if (tree == bl_tree) {
1956 bb_error_msg("\nnode %d(%d), sons %d(%d) %d(%d)",
1957 node, tree[node].Freq, n, tree[n].Freq, m, tree[m].Freq);
1960 /* and insert the new node in the heap */
1961 heap[SMALLEST] = node++;
1962 pqdownheap(tree, SMALLEST);
1964 } while (heap_len >= 2);
1966 heap[--heap_max] = heap[SMALLEST];
1968 /* At this point, the fields freq and dad are set. We can now
1969 * generate the bit lengths.
1971 gen_bitlen((tree_desc *) desc);
1973 /* The field len is now set, we can generate the bit codes */
1974 gen_codes((ct_data *) tree, max_code);
1977 /* ===========================================================================
1978 * Scan a literal or distance tree to determine the frequencies of the codes
1979 * in the bit length tree. Updates opt_len to take into account the repeat
1980 * counts. (The contribution of the bit length codes will be added later
1981 * during the construction of bl_tree.)
1983 static void scan_tree(ct_data * tree, int max_code)
1985 int n; /* iterates over all tree elements */
1986 int prevlen = -1; /* last emitted length */
1987 int curlen; /* length of current code */
1988 int nextlen = tree[0].Len; /* length of next code */
1989 int count = 0; /* repeat count of the current code */
1990 int max_count = 7; /* max repeat count */
1991 int min_count = 4; /* min repeat count */
1994 max_count = 138, min_count = 3;
1995 tree[max_code + 1].Len = (ush) 0xffff; /* guard */
1997 for (n = 0; n <= max_code; n++) {
1999 nextlen = tree[n + 1].Len;
2000 if (++count < max_count && curlen == nextlen) {
2002 } else if (count < min_count) {
2003 bl_tree[curlen].Freq += count;
2004 } else if (curlen != 0) {
2005 if (curlen != prevlen)
2006 bl_tree[curlen].Freq++;
2007 bl_tree[REP_3_6].Freq++;
2008 } else if (count <= 10) {
2009 bl_tree[REPZ_3_10].Freq++;
2011 bl_tree[REPZ_11_138].Freq++;
2016 max_count = 138, min_count = 3;
2017 } else if (curlen == nextlen) {
2018 max_count = 6, min_count = 3;
2020 max_count = 7, min_count = 4;
2025 /* ===========================================================================
2026 * Send a literal or distance tree in compressed form, using the codes in
2029 static void send_tree(ct_data * tree, int max_code)
2031 int n; /* iterates over all tree elements */
2032 int prevlen = -1; /* last emitted length */
2033 int curlen; /* length of current code */
2034 int nextlen = tree[0].Len; /* length of next code */
2035 int count = 0; /* repeat count of the current code */
2036 int max_count = 7; /* max repeat count */
2037 int min_count = 4; /* min repeat count */
2039 /* tree[max_code+1].Len = -1; *//* guard already set */
2041 max_count = 138, min_count = 3;
2043 for (n = 0; n <= max_code; n++) {
2045 nextlen = tree[n + 1].Len;
2046 if (++count < max_count && curlen == nextlen) {
2048 } else if (count < min_count) {
2050 send_code(curlen, bl_tree);
2051 } while (--count != 0);
2053 } else if (curlen != 0) {
2054 if (curlen != prevlen) {
2055 send_code(curlen, bl_tree);
2058 Assert(count >= 3 && count <= 6, " 3_6?");
2059 send_code(REP_3_6, bl_tree);
2060 send_bits(count - 3, 2);
2062 } else if (count <= 10) {
2063 send_code(REPZ_3_10, bl_tree);
2064 send_bits(count - 3, 3);
2067 send_code(REPZ_11_138, bl_tree);
2068 send_bits(count - 11, 7);
2073 max_count = 138, min_count = 3;
2074 } else if (curlen == nextlen) {
2075 max_count = 6, min_count = 3;
2077 max_count = 7, min_count = 4;
2082 /* ===========================================================================
2083 * Construct the Huffman tree for the bit lengths and return the index in
2084 * bl_order of the last bit length code to send.
2086 static int build_bl_tree(void)
2088 int max_blindex; /* index of last bit length code of non zero freq */
2090 /* Determine the bit length frequencies for literal and distance trees */
2091 scan_tree((ct_data *) dyn_ltree, l_desc.max_code);
2092 scan_tree((ct_data *) dyn_dtree, d_desc.max_code);
2094 /* Build the bit length tree: */
2095 build_tree((tree_desc *) (&bl_desc));
2096 /* opt_len now includes the length of the tree representations, except
2097 * the lengths of the bit lengths codes and the 5+5+4 bits for the counts.
2100 /* Determine the number of bit length codes to send. The pkzip format
2101 * requires that at least 4 bit length codes be sent. (appnote.txt says
2102 * 3 but the actual value used is 4.)
2104 for (max_blindex = BL_CODES - 1; max_blindex >= 3; max_blindex--) {
2105 if (bl_tree[bl_order[max_blindex]].Len != 0)
2108 /* Update opt_len to include the bit length tree and counts */
2109 opt_len += 3 * (max_blindex + 1) + 5 + 5 + 4;
2110 Tracev((stderr, "\ndyn trees: dyn %ld, stat %ld", opt_len, static_len));
2115 /* ===========================================================================
2116 * Send the header for a block using dynamic Huffman trees: the counts, the
2117 * lengths of the bit length codes, the literal tree and the distance tree.
2118 * IN assertion: lcodes >= 257, dcodes >= 1, blcodes >= 4.
2120 static void send_all_trees(int lcodes, int dcodes, int blcodes)
2122 int rank; /* index in bl_order */
2124 Assert(lcodes >= 257 && dcodes >= 1 && blcodes >= 4, "not enough codes");
2125 Assert(lcodes <= L_CODES && dcodes <= D_CODES
2126 && blcodes <= BL_CODES, "too many codes");
2127 Tracev((stderr, "\nbl counts: "));
2128 send_bits(lcodes - 257, 5); /* not +255 as stated in appnote.txt */
2129 send_bits(dcodes - 1, 5);
2130 send_bits(blcodes - 4, 4); /* not -3 as stated in appnote.txt */
2131 for (rank = 0; rank < blcodes; rank++) {
2132 Tracev((stderr, "\nbl code %2d ", bl_order[rank]));
2133 send_bits(bl_tree[bl_order[rank]].Len, 3);
2135 Tracev((stderr, "\nbl tree: sent %ld", bits_sent));
2137 send_tree((ct_data *) dyn_ltree, lcodes - 1); /* send the literal tree */
2138 Tracev((stderr, "\nlit tree: sent %ld", bits_sent));
2140 send_tree((ct_data *) dyn_dtree, dcodes - 1); /* send the distance tree */
2141 Tracev((stderr, "\ndist tree: sent %ld", bits_sent));
2144 /* ===========================================================================
2145 * Determine the best encoding for the current block: dynamic trees, static
2146 * trees or store, and output the encoded block to the zip file. This function
2147 * returns the total compressed length for the file so far.
2149 static ulg flush_block(char *buf, ulg stored_len, int eof)
2151 ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */
2152 int max_blindex; /* index of last bit length code of non zero freq */
2154 flag_buf[last_flags] = flags; /* Save the flags for the last 8 items */
2156 /* Check if the file is ascii or binary */
2157 if (*file_type == (ush) UNKNOWN)
2160 /* Construct the literal and distance trees */
2161 build_tree((tree_desc *) (&l_desc));
2162 Tracev((stderr, "\nlit data: dyn %ld, stat %ld", opt_len, static_len));
2164 build_tree((tree_desc *) (&d_desc));
2165 Tracev((stderr, "\ndist data: dyn %ld, stat %ld", opt_len, static_len));
2166 /* At this point, opt_len and static_len are the total bit lengths of
2167 * the compressed block data, excluding the tree representations.
2170 /* Build the bit length tree for the above two trees, and get the index
2171 * in bl_order of the last bit length code to send.
2173 max_blindex = build_bl_tree();
2175 /* Determine the best encoding. Compute first the block length in bytes */
2176 opt_lenb = (opt_len + 3 + 7) >> 3;
2177 static_lenb = (static_len + 3 + 7) >> 3;
2180 "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u dist %u ",
2181 opt_lenb, opt_len, static_lenb, static_len, stored_len,
2182 last_lit, last_dist));
2184 if (static_lenb <= opt_lenb)
2185 opt_lenb = static_lenb;
2187 /* If compression failed and this is the first and last block,
2188 * and if the zip file can be seeked (to rewrite the local header),
2189 * the whole file is transformed into a stored file:
2191 if (stored_len <= opt_lenb && eof && compressed_len == 0L && seekable()) {
2192 /* Since LIT_BUFSIZE <= 2*WSIZE, the input data must be there: */
2193 if (buf == (char *) 0)
2194 bb_error_msg("block vanished");
2196 copy_block(buf, (unsigned) stored_len, 0); /* without header */
2197 compressed_len = stored_len << 3;
2198 *file_method = STORED;
2200 } else if (stored_len + 4 <= opt_lenb && buf != (char *) 0) {
2201 /* 4: two words for the lengths */
2202 /* The test buf != NULL is only necessary if LIT_BUFSIZE > WSIZE.
2203 * Otherwise we can't have processed more than WSIZE input bytes since
2204 * the last block flush, because compression would have been
2205 * successful. If LIT_BUFSIZE <= WSIZE, it is never too late to
2206 * transform a block into a stored block.
2208 send_bits((STORED_BLOCK << 1) + eof, 3); /* send block type */
2209 compressed_len = (compressed_len + 3 + 7) & ~7L;
2210 compressed_len += (stored_len + 4) << 3;
2212 copy_block(buf, (unsigned) stored_len, 1); /* with header */
2214 } else if (static_lenb == opt_lenb) {
2215 send_bits((STATIC_TREES << 1) + eof, 3);
2216 compress_block((ct_data *) static_ltree, (ct_data *) static_dtree);
2217 compressed_len += 3 + static_len;
2219 send_bits((DYN_TREES << 1) + eof, 3);
2220 send_all_trees(l_desc.max_code + 1, d_desc.max_code + 1,
2222 compress_block((ct_data *) dyn_ltree, (ct_data *) dyn_dtree);
2223 compressed_len += 3 + opt_len;
2225 Assert(compressed_len == bits_sent, "bad compressed size");
2230 compressed_len += 7; /* align on byte boundary */
2232 Tracev((stderr, "\ncomprlen %lu(%lu) ", compressed_len >> 3,
2233 compressed_len - 7 * eof));
2235 return compressed_len >> 3;
2238 /* ===========================================================================
2239 * Save the match info and tally the frequency counts. Return true if
2240 * the current block must be flushed.
2242 static int ct_tally(int dist, int lc)
2244 l_buf[last_lit++] = (uch) lc;
2246 /* lc is the unmatched char */
2247 dyn_ltree[lc].Freq++;
2249 /* Here, lc is the match length - MIN_MATCH */
2250 dist--; /* dist = match distance - 1 */
2251 Assert((ush) dist < (ush) MAX_DIST &&
2252 (ush) lc <= (ush) (MAX_MATCH - MIN_MATCH) &&
2253 (ush) d_code(dist) < (ush) D_CODES, "ct_tally: bad match");
2255 dyn_ltree[length_code[lc] + LITERALS + 1].Freq++;
2256 dyn_dtree[d_code(dist)].Freq++;
2258 d_buf[last_dist++] = (ush) dist;
2263 /* Output the flags if they fill a byte: */
2264 if ((last_lit & 7) == 0) {
2265 flag_buf[last_flags++] = flags;
2266 flags = 0, flag_bit = 1;
2268 /* Try to guess if it is profitable to stop the current block here */
2269 if ((last_lit & 0xfff) == 0) {
2270 /* Compute an upper bound for the compressed length */
2271 ulg out_length = (ulg) last_lit * 8L;
2272 ulg in_length = (ulg) strstart - block_start;
2275 for (dcode = 0; dcode < D_CODES; dcode++) {
2277 (ulg) dyn_dtree[dcode].Freq * (5L + extra_dbits[dcode]);
2281 "\nlast_lit %u, last_dist %u, in %ld, out ~%ld(%ld%%) ",
2282 last_lit, last_dist, in_length, out_length,
2283 100L - out_length * 100L / in_length));
2284 if (last_dist < last_lit / 2 && out_length < in_length / 2)
2287 return (last_lit == LIT_BUFSIZE - 1 || last_dist == DIST_BUFSIZE);
2288 /* We avoid equality with LIT_BUFSIZE because of wraparound at 64K
2289 * on 16 bit machines and because stored blocks are restricted to
2294 /* ===========================================================================
2295 * Send the block data compressed using the given Huffman trees
2297 static void compress_block(ct_data * ltree, ct_data * dtree)
2299 unsigned dist; /* distance of matched string */
2300 int lc; /* match length or unmatched char (if dist == 0) */
2301 unsigned lx = 0; /* running index in l_buf */
2302 unsigned dx = 0; /* running index in d_buf */
2303 unsigned fx = 0; /* running index in flag_buf */
2304 uch flag = 0; /* current flags */
2305 unsigned code; /* the code to send */
2306 int extra; /* number of extra bits to send */
2311 flag = flag_buf[fx++];
2313 if ((flag & 1) == 0) {
2314 send_code(lc, ltree); /* send a literal byte */
2315 Tracecv(isgraph(lc), (stderr, " '%c' ", lc));
2317 /* Here, lc is the match length - MIN_MATCH */
2318 code = length_code[lc];
2319 send_code(code + LITERALS + 1, ltree); /* send the length code */
2320 extra = extra_lbits[code];
2322 lc -= base_length[code];
2323 send_bits(lc, extra); /* send the extra length bits */
2326 /* Here, dist is the match distance - 1 */
2327 code = d_code(dist);
2328 Assert(code < D_CODES, "bad d_code");
2330 send_code(code, dtree); /* send the distance code */
2331 extra = extra_dbits[code];
2333 dist -= base_dist[code];
2334 send_bits(dist, extra); /* send the extra distance bits */
2336 } /* literal or match pair ? */
2338 } while (lx < last_lit);
2340 send_code(END_BLOCK, ltree);
2343 /* ===========================================================================
2344 * Set the file type to ASCII or BINARY, using a crude approximation:
2345 * binary if more than 20% of the bytes are <= 6 or >= 128, ascii otherwise.
2346 * IN assertion: the fields freq of dyn_ltree are set and the total of all
2347 * frequencies does not exceed 64K (to fit in an int on 16 bit machines).
2349 static void set_file_type(void)
2352 unsigned ascii_freq = 0;
2353 unsigned bin_freq = 0;
2356 bin_freq += dyn_ltree[n++].Freq;
2358 ascii_freq += dyn_ltree[n++].Freq;
2359 while (n < LITERALS)
2360 bin_freq += dyn_ltree[n++].Freq;
2361 *file_type = bin_freq > (ascii_freq >> 2) ? BINARY : ASCII;
2362 if (*file_type == BINARY && translate_eol) {
2363 bb_error_msg("-l used on binary file");
2367 /* zip.c -- compress files to the gzip or pkzip format
2368 * Copyright (C) 1992-1993 Jean-loup Gailly
2369 * This is free software; you can redistribute it and/or modify it under the
2370 * terms of the GNU General Public License, see the file COPYING.
2374 static uint32_t crc; /* crc on uncompressed file data */
2375 static long header_bytes; /* number of bytes in gzip header */
2377 static void put_long(ulg n)
2379 put_short((n) & 0xffff);
2380 put_short(((ulg) (n)) >> 16);
2383 /* put_header_byte is used for the compressed output
2384 * - for the initial 4 bytes that can't overflow the buffer.
2386 #define put_header_byte(c) {outbuf[outcnt++]=(uch)(c);}
2388 /* ===========================================================================
2389 * Deflate in to out.
2390 * IN assertions: the input and output buffers are cleared.
2391 * The variables time_stamp and save_orig_name are initialized.
2393 static int zip(int in, int out)
2395 uch my_flags = 0; /* general purpose bit flags */
2396 ush attr = 0; /* ascii/binary flag */
2397 ush deflate_flags = 0; /* pkzip -es, -en or -ex equivalent */
2403 /* Write the header to the gzip file. See algorithm.doc for the format */
2407 put_header_byte(GZIP_MAGIC[0]); /* magic header */
2408 put_header_byte(GZIP_MAGIC[1]);
2409 put_header_byte(DEFLATED); /* compression method */
2411 put_header_byte(my_flags); /* general flags */
2412 put_long(time_stamp);
2414 /* Write deflated file to zip file */
2418 ct_init(&attr, &method);
2419 lm_init(&deflate_flags);
2421 put_byte((uch) deflate_flags); /* extra flags */
2422 put_byte(OS_CODE); /* OS identifier */
2424 header_bytes = (long) outcnt;
2428 /* Write the crc and uncompressed size */
2431 header_bytes += 2 * sizeof(long);
2438 /* ===========================================================================
2439 * Read a new buffer from the current input file, perform end-of-line
2440 * translation, and update the crc and input file size.
2441 * IN assertion: size >= 2 (for end-of-line translation)
2443 static int file_read(char *buf, unsigned size)
2447 Assert(insize == 0, "inbuf not empty");
2449 len = read(ifd, buf, size);
2450 if (len == (unsigned) (-1) || len == 0)
2453 crc = updcrc((uch *) buf, len);
2458 /* ===========================================================================
2459 * Write the output buffer outbuf[0..outcnt-1] and update bytes_out.
2460 * (used for the compressed data only)
2462 static void flush_outbuf(void)
2467 write_buf(ofd, (char *) outbuf, outcnt);