X-Git-Url: https://git.librecmc.org/?a=blobdiff_plain;f=archival%2Fgzip.c;h=43804b2e4060fd2a0cd447e501595e6338d0df51;hb=8b814b4a349e2262c0ad25793b05206a14651ebb;hp=cdf22688924ebd9dfefddc1b4ac434dd39d4d56c;hpb=3073dfbf30da26ac60196319866c8e0a931c2789;p=oweals%2Fbusybox.git diff --git a/archival/gzip.c b/archival/gzip.c index cdf226889..43804b2e4 100644 --- a/archival/gzip.c +++ b/archival/gzip.c @@ -5,81 +5,68 @@ * Based on GNU gzip Copyright (C) 1992-1993 Jean-loup Gailly. * * Originally adjusted for busybox by Charles P. Wright - * "this is a stripped down version of gzip I put into busybox, it does - * only standard in to standard out with -9 compression. It also requires - * the zcat module for some important functions." + * "this is a stripped down version of gzip I put into busybox, it does + * only standard in to standard out with -9 compression. It also requires + * the zcat module for some important functions." * - * Adjusted further by Erik Andersen , - * to support files as well as stdin/stdout, and to generally behave itself wrt + * Adjusted further by Erik Andersen to support + * files as well as stdin/stdout, and to generally behave itself wrt * command line handling. * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA - * + * Licensed under GPLv2 or later, see file LICENSE in this tarball for details. */ -/* These defines are very important for BusyBox. Without these, - * huge chunks of ram are pre-allocated making the BusyBox bss - * size Freaking Huge(tm), which is a bad thing.*/ -#define SMALL_MEM -#define DYN_ALLOC - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include "busybox.h" - -#define memzero(s, n) memset ((void *)(s), 0, (n)) - -#ifndef RETSIGTYPE -# define RETSIGTYPE void -#endif +/* big objects in bss: + * 00000020 b bl_count + * 00000074 b base_length + * 00000078 b base_dist + * 00000078 b static_dtree + * 0000009c b bl_tree + * 000000f4 b dyn_dtree + * 00000100 b length_code + * 00000200 b dist_code + * 0000023d b depth + * 00000400 b flag_buf + * 0000047a b heap + * 00000480 b static_ltree + * 000008f4 b dyn_ltree + */ + +/* TODO: full support for -v for DESKTOP + * "/usr/bin/gzip -v a bogus aa" should say: +a: 85.1% -- replaced with a.gz +gzip: bogus: No such file or directory +aa: 85.1% -- replaced with aa.gz +*/ + +#include "libbb.h" +#include "unarchive.h" -typedef unsigned char uch; -typedef unsigned short ush; -typedef unsigned long ulg; -/* Return codes from gzip */ -#define OK 0 -#define ERROR 1 -#define WARNING 2 +/* =========================================================================== + */ +//#define DEBUG 1 +/* Diagnostic functions */ +#ifdef DEBUG +# define Assert(cond,msg) { if (!(cond)) bb_error_msg(msg); } +# define Trace(x) fprintf x +# define Tracev(x) {if (verbose) fprintf x; } +# define Tracevv(x) {if (verbose > 1) fprintf x; } +# define Tracec(c,x) {if (verbose && (c)) fprintf x; } +# define Tracecv(c,x) {if (verbose > 1 && (c)) fprintf x; } +#else +# define Assert(cond,msg) +# define Trace(x) +# define Tracev(x) +# define Tracevv(x) +# define Tracec(c,x) +# define Tracecv(c,x) +#endif -/* Compression methods (see algorithm.doc) */ -/* Only STORED and DEFLATED are supported by this BusyBox module */ -#define STORED 0 -/* methods 4 to 7 reserved */ -#define DEFLATED 8 -static int method; /* compression method */ -/* To save memory for 16 bit systems, some arrays are overlaid between - * the various modules: - * deflate: prev+head window d_buf l_buf outbuf - * unlzw: tab_prefix tab_suffix stack inbuf outbuf - * For compression, input is done in window[]. For decompression, output - * is done in window except for unlzw. +/* =========================================================================== */ +#define SMALL_MEM #ifndef INBUFSIZ # ifdef SMALL_MEM @@ -88,7 +75,6 @@ static int method; /* compression method */ # define INBUFSIZ 0x8000 /* input buffer size */ # endif #endif -#define INBUF_EXTRA 64 /* required by unlzw() */ #ifndef OUTBUFSIZ # ifdef SMALL_MEM @@ -97,7 +83,6 @@ static int method; /* compression method */ # define OUTBUFSIZ 16384 /* output buffer size */ # endif #endif -#define OUTBUF_EXTRA 2048 /* required by unlzw() */ #ifndef DIST_BUFSIZE # ifdef SMALL_MEM @@ -107,46 +92,13 @@ static int method; /* compression method */ # endif #endif -#ifdef DYN_ALLOC -# define DECLARE(type, array, size) static type * array -# define ALLOC(type, array, size) { \ - array = (type*)calloc((size_t)(((size)+1L)/2), 2*sizeof(type)); \ - if (array == NULL) error_msg(memory_exhausted); \ - } -# define FREE(array) {if (array != NULL) free(array), array=NULL;} -#else -# define DECLARE(type, array, size) static type array[size] -# define ALLOC(type, array, size) -# define FREE(array) -#endif - -#define tab_suffix window -#define tab_prefix prev /* hash link (see deflate.c) */ -#define head (prev+WSIZE) /* hash head (see deflate.c) */ - -static long bytes_in; /* number of input bytes */ - -#define isize bytes_in -/* for compatibility with old zip sources (to be cleaned) */ - -typedef int file_t; /* Do not use stdio */ - -#define NO_FILE (-1) /* in memory compression */ - - -#define PACK_MAGIC "\037\036" /* Magic header for packed files */ -#define GZIP_MAGIC "\037\213" /* Magic header for gzip files, 1F 8B */ -#define OLD_GZIP_MAGIC "\037\236" /* Magic header for gzip 0.5 = freeze 1.x */ -#define LZH_MAGIC "\037\240" /* Magic header for SCO LZH Compress files */ -#define PKZIP_MAGIC "\120\113\003\004" /* Magic header for pkzip files */ - /* gzip flag byte */ -#define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */ -#define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */ -#define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */ -#define ORIG_NAME 0x08 /* bit 3 set: original file name present */ -#define COMMENT 0x10 /* bit 4 set: file comment present */ -#define RESERVED 0xC0 /* bit 6,7: reserved */ +#define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */ +#define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */ +#define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */ +#define ORIG_NAME 0x08 /* bit 3 set: original file name present */ +#define COMMENT 0x10 /* bit 4 set: file comment present */ +#define RESERVED 0xC0 /* bit 6,7: reserved */ /* internal file attribute */ #define UNKNOWN 0xffff @@ -154,8 +106,8 @@ typedef int file_t; /* Do not use stdio */ #define ASCII 1 #ifndef WSIZE -# define WSIZE 0x8000 /* window size--must be a power of two, and */ -#endif /* at least 32K for zip's deflate method */ +# define WSIZE 0x8000 /* window size--must be a power of two, and */ +#endif /* at least 32K for zip's deflate method */ #define MIN_MATCH 3 #define MAX_MATCH 258 @@ -171,99 +123,19 @@ typedef int file_t; /* Do not use stdio */ * distances are limited to MAX_DIST instead of WSIZE. */ -/* put_byte is used for the compressed output */ -#define put_byte(c) {outbuf[outcnt++]=(uch)(c); if (outcnt==OUTBUFSIZ)\ - flush_outbuf();} - -/* Output a 16 bit value, lsb first */ -#define put_short(w) \ -{ if (outcnt < OUTBUFSIZ-2) { \ - outbuf[outcnt++] = (uch) ((w) & 0xff); \ - outbuf[outcnt++] = (uch) ((ush)(w) >> 8); \ - } else { \ - put_byte((uch)((w) & 0xff)); \ - put_byte((uch)((ush)(w) >> 8)); \ - } \ -} - -/* Output a 32 bit value to the bit stream, lsb first */ -#define put_long(n) { \ - put_short((n) & 0xffff); \ - put_short(((ulg)(n)) >> 16); \ -} - -#define seekable() 0 /* force sequential output */ -#define translate_eol 0 /* no option -a yet */ - -/* Diagnostic functions */ -#ifdef DEBUG -# define Assert(cond,msg) {if(!(cond)) error_msg(msg);} -# define Trace(x) fprintf x -# define Tracev(x) {if (verbose) fprintf x ;} -# define Tracevv(x) {if (verbose>1) fprintf x ;} -# define Tracec(c,x) {if (verbose && (c)) fprintf x ;} -# define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;} -#else -# define Assert(cond,msg) -# define Trace(x) -# define Tracev(x) -# define Tracevv(x) -# define Tracec(c,x) -# define Tracecv(c,x) -#endif - -#define WARN(msg) {if (!quiet) fprintf msg ; \ - if (exit_code == OK) exit_code = WARNING;} - #ifndef MAX_PATH_LEN # define MAX_PATH_LEN 1024 /* max pathname length */ #endif - - - /* from zip.c: */ -static int zip (int in, int out); -static int file_read (char *buf, unsigned size); - - /* from gzip.c */ -static RETSIGTYPE abort_gzip (void); - - /* from deflate.c */ -static void lm_init (ush * flags); -static ulg deflate (void); - - /* from trees.c */ -static void ct_init (ush * attr, int *methodp); -static int ct_tally (int dist, int lc); -static ulg flush_block (char *buf, ulg stored_len, int eof); - - /* from bits.c */ -static void bi_init (file_t zipfile); -static void send_bits (int value, int length); -static unsigned bi_reverse (unsigned value, int length); -static void bi_windup (void); -static void copy_block (char *buf, unsigned len, int header); -static int (*read_buf) (char *buf, unsigned size); - - /* from util.c: */ -static void flush_outbuf (void); - -/* lzw.h -- define the lzw functions. - * Copyright (C) 1992-1993 Jean-loup Gailly. - * This is free software; you can redistribute it and/or modify it under the - * terms of the GNU General Public License, see the file COPYING. - */ - -#if !defined(OF) && defined(lint) -# include "gzip.h" -#endif +#define seekable() 0 /* force sequential output */ +#define translate_eol 0 /* no option -a yet */ #ifndef BITS # define BITS 16 #endif -#define INIT_BITS 9 /* Initial number of bits per code */ +#define INIT_BITS 9 /* Initial number of bits per code */ -#define BIT_MASK 0x1f /* Mask for 'number of compression bits' */ +#define BIT_MASK 0x1f /* Mask for 'number of compression bits' */ /* Mask 0x20 is reserved to mean a fourth header byte, and 0x40 is free. * It's a pity that old uncompress does not check bit 0x20. That makes * extension of the format actually undesirable because old compress @@ -273,1082 +145,556 @@ static void flush_outbuf (void); * "can only handle 16 bits". */ -/* tailor.h -- target dependent definitions - * Copyright (C) 1992-1993 Jean-loup Gailly. - * This is free software; you can redistribute it and/or modify it under the - * terms of the GNU General Public License, see the file COPYING. - */ - -/* The target dependent definitions should be defined here only. - * The target dependent functions should be defined in tailor.c. - */ +#ifdef MAX_EXT_CHARS +# define MAX_SUFFIX MAX_EXT_CHARS +#else +# define MAX_SUFFIX 30 +#endif - /* Common defaults */ +/* =========================================================================== + * Compile with MEDIUM_MEM to reduce the memory requirements or + * with SMALL_MEM to use as little memory as possible. Use BIG_MEM if the + * entire input file can be held in memory (not possible on 16 bit systems). + * Warning: defining these symbols affects HASH_BITS (see below) and thus + * affects the compression ratio. The compressed output + * is still correct, and might even be smaller in some cases. + */ -#ifndef OS_CODE -# define OS_CODE 0x03 /* assume Unix */ +#ifdef SMALL_MEM +# define HASH_BITS 13 /* Number of bits used to hash strings */ #endif - -#ifndef PATH_SEP -# define PATH_SEP '/' +#ifdef MEDIUM_MEM +# define HASH_BITS 14 #endif - -#ifndef OPTIONS_VAR -# define OPTIONS_VAR "GZIP" +#ifndef HASH_BITS +# define HASH_BITS 15 + /* For portability to 16 bit machines, do not use values above 15. */ #endif -#ifndef Z_SUFFIX -# define Z_SUFFIX ".gz" +#define HASH_SIZE (unsigned)(1<= 4. + */ -/* ======================================================================== - * Signal and error handler. + max_insert_length = max_lazy_match, +/* Insert new strings in the hash table only if the match length + * is not greater than this length. This saves time but degrades compression. + * max_insert_length is used only for compression levels <= 3. */ -static void abort_gzip() -{ - exit(ERROR); -} -/* =========================================================================== - * Clear input and output buffers + good_match = 32, +/* Use a faster search when the previous match is longer than this */ + +/* Values for max_lazy_match, good_match and max_chain_length, depending on + * the desired pack level (0..9). The values given below have been tuned to + * exclude worst case performance for pathological files. Better values may be + * found for specific files. */ -static void clear_bufs(void) -{ - outcnt = 0; - insize = 0; - bytes_in = 0L; -} -static void write_error_msg() -{ - fprintf(stderr, "\n"); - perror(""); - abort_gzip(); -} + nice_match = 258, /* Stop searching when current match exceeds this */ +/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 + * For deflate_fast() (levels <= 3) good is ignored and lazy has a different + * meaning. + */ +}; -/* =========================================================================== - * Does the same as write(), but also handles partial pipe writes and checks - * for error return. + +struct globals { + + lng block_start; + +/* window position at the beginning of the current output block. Gets + * negative when the window is moved backwards. */ -static void write_buf(int fd, void *buf, unsigned cnt) -{ - unsigned n; + unsigned ins_h; /* hash index of string to be inserted */ - while ((n = write(fd, buf, cnt)) != cnt) { - if (n == (unsigned) (-1)) { - write_error_msg(); - } - cnt -= n; - buf = (void *) ((char *) buf + n); - } -} +#define H_SHIFT ((HASH_BITS+MIN_MATCH-1) / MIN_MATCH) +/* Number of bits by which ins_h and del_h must be shifted at each + * input step. It must be such that after MIN_MATCH steps, the oldest + * byte no longer takes part in the hash key, that is: + * H_SHIFT * MIN_MATCH >= HASH_BITS + */ -/* =========================================================================== - * Run a set of bytes through the crc shift register. If s is a NULL - * pointer, then initialize the crc shift register contents instead. - * Return the current crc in either case. + unsigned prev_length; + +/* Length of the best match at previous step. Matches not greater than this + * are discarded. This is used in the lazy match evaluation. */ -static ulg updcrc(uch *s, unsigned n) -{ - static ulg crc = (ulg) 0xffffffffL; /* shift register contents */ - register ulg c; /* temporary variable */ - static unsigned long crc_32_tab[256]; - if (crc_table_empty) { - unsigned long csr; /* crc shift register */ - unsigned long e=0; /* polynomial exclusive-or pattern */ - int i; /* counter for all possible eight bit values */ - int k; /* byte being shifted into crc apparatus */ - - /* terms of polynomial defining this crc (except x^32): */ - static const int p[] = {0,1,2,4,5,7,8,10,11,12,16,22,23,26}; - - /* Make exclusive-or pattern from polynomial (0xedb88320) */ - for (i = 0; i < sizeof(p)/sizeof(int); i++) - e |= 1L << (31 - p[i]); - - /* Compute and print table of CRC's, five per line */ - crc_32_tab[0] = 0x00000000L; - for (i = 1; i < 256; i++) { - csr = i; - /* The idea to initialize the register with the byte instead of - * zero was stolen from Haruhiko Okumura's ar002 - */ - for (k = 8; k; k--) - csr = csr & 1 ? (csr >> 1) ^ e : csr >> 1; - crc_32_tab[i]=csr; - } - } - if (s == NULL) { - c = 0xffffffffL; - } else { - c = crc; - if (n) - do { - c = crc_32_tab[((int) c ^ (*s++)) & 0xff] ^ (c >> 8); - } while (--n); - } - crc = c; - return c ^ 0xffffffffL; /* (instead of ~c for 64-bit machines) */ -} + unsigned strstart; /* start of string to insert */ + unsigned match_start; /* start of matching string */ + unsigned lookahead; /* number of valid bytes ahead in window */ -/* bits.c -- output variable-length bit strings - * Copyright (C) 1992-1993 Jean-loup Gailly - * This is free software; you can redistribute it and/or modify it under the - * terms of the GNU General Public License, see the file COPYING. +/* =========================================================================== */ +#define DECLARE(type, array, size) \ + type * array +#define ALLOC(type, array, size) \ + array = xzalloc((size_t)(((size)+1L)/2) * 2*sizeof(type)); +#define FREE(array) \ + do { free(array); array = NULL; } while (0) + /* global buffers */ -/* - * PURPOSE - * - * Output variable-length bit strings. Compression can be done - * to a file or to memory. (The latter is not supported in this version.) - * - * DISCUSSION - * - * The PKZIP "deflate" file format interprets compressed file data - * as a sequence of bits. Multi-bit strings in the file may cross - * byte boundaries without restriction. - * - * The first bit of each byte is the low-order bit. - * - * The routines in this file allow a variable-length bit value to - * be output right-to-left (useful for literal values). For - * left-to-right output (useful for code strings from the tree routines), - * the bits must have been reversed first with bi_reverse(). - * - * For in-memory compression, the compressed bit stream goes directly - * into the requested output buffer. The input data is read in blocks - * by the mem_read() function. The buffer is limited to 64K on 16 bit - * machines. - * - * INTERFACE - * - * void bi_init (FILE *zipfile) - * Initialize the bit string routines. - * - * void send_bits (int value, int length) - * Write out a bit string, taking the source bits right to - * left. - * - * int bi_reverse (int value, int length) - * Reverse the bits of a bit string, taking the source bits left to - * right and emitting them right to left. - * - * void bi_windup (void) - * Write out any remaining bits in an incomplete byte. - * - * void copy_block(char *buf, unsigned len, int header) - * Copy a stored block to the zip file, storing first the length and - * its one's complement if requested. - * + /* buffer for literals or lengths */ + /* DECLARE(uch, l_buf, LIT_BUFSIZE); */ + DECLARE(uch, l_buf, INBUFSIZ); + + DECLARE(ush, d_buf, DIST_BUFSIZE); + DECLARE(uch, outbuf, OUTBUFSIZ); + +/* Sliding window. Input bytes are read into the second half of the window, + * and move to the first half later to keep a dictionary of at least WSIZE + * bytes. With this organization, matches are limited to a distance of + * WSIZE-MAX_MATCH bytes, but this ensures that IO is always + * performed with a length multiple of the block size. Also, it limits + * the window size to 64K, which is quite useful on MSDOS. + * To do: limit the window size to WSIZE+BSZ if SMALL_MEM (the code would + * be less efficient). */ + DECLARE(uch, window, 2L * WSIZE); + +/* Link to older string with same hash index. To limit the size of this + * array to 64K, this link is maintained only for the last 32K strings. + * An index in this array is thus a window index modulo 32K. + */ + /* DECLARE(Pos, prev, WSIZE); */ + DECLARE(ush, prev, 1L << BITS); + +/* Heads of the hash chains or 0. */ + /* DECLARE(Pos, head, 1< 0 && length <= 15, "invalid length"); - bits_sent += (ulg) length; -#endif - /* If not enough room in bi_buf, use (valid) bits from bi_buf and - * (16 - bi_valid) bits from value, leaving (width - (16-bi_valid)) - * unused bits in value. - */ - if (bi_valid > (int) Buf_size - length) { - bi_buf |= (value << bi_valid); - put_short(bi_buf); - bi_buf = (ush) value >> (Buf_size - bi_valid); - bi_valid += length - Buf_size; + if (G1.outcnt < OUTBUFSIZ - 2) { + G1.outbuf[G1.outcnt++] = w; + G1.outbuf[G1.outcnt++] = w >> 8; } else { - bi_buf |= value << bi_valid; - bi_valid += length; + put_8bit(w); + put_8bit(w >> 8); } } -/* =========================================================================== - * Reverse the first len bits of a code, using straightforward code (a faster - * method would use a table) - * IN assertion: 1 <= len <= 15 - */ -static unsigned bi_reverse(unsigned code, int len) +static void put_32bit(ulg n) { - register unsigned res = 0; - - do { - res |= code & 1; - code >>= 1, res <<= 1; - } while (--len > 0); - return res >> 1; + put_16bit(n); + put_16bit(n >> 16); } /* =========================================================================== - * Write out any remaining bits in an incomplete byte. + * Clear input and output buffers */ -static void bi_windup() +static void clear_bufs(void) { - if (bi_valid > 8) { - put_short(bi_buf); - } else if (bi_valid > 0) { - put_byte(bi_buf); - } - bi_buf = 0; - bi_valid = 0; + G1.outcnt = 0; #ifdef DEBUG - bits_sent = (bits_sent + 7) & ~7; + G1.insize = 0; #endif + G1.isize = 0; } + /* =========================================================================== - * Copy a stored block to the zip file, storing first the length and its - * one's complement if requested. - */ -static void copy_block(char *buf, unsigned len, int header) + * Run a set of bytes through the crc shift register. If s is a NULL + * pointer, then initialize the crc shift register contents instead. + * Return the current crc in either case. + */ +static uint32_t updcrc(uch * s, unsigned n) { - bi_windup(); /* align on byte boundary */ - - if (header) { - put_short((ush) len); - put_short((ush) ~ len); -#ifdef DEBUG - bits_sent += 2 * 16; -#endif - } -#ifdef DEBUG - bits_sent += (ulg) len << 3; -#endif - while (len--) { - put_byte(*buf++); + uint32_t c = G1.crc; + while (n) { + c = G1.crc_32_tab[(uch)(c ^ *s++)] ^ (c >> 8); + n--; } + G1.crc = c; + return c; } -/* deflate.c -- compress data using the deflation algorithm - * Copyright (C) 1992-1993 Jean-loup Gailly - * This is free software; you can redistribute it and/or modify it under the - * terms of the GNU General Public License, see the file COPYING. - */ - -/* - * PURPOSE - * - * Identify new text as repetitions of old text within a fixed- - * length sliding window trailing behind the new text. - * - * DISCUSSION - * - * The "deflation" process depends on being able to identify portions - * of the input text which are identical to earlier input (within a - * sliding window trailing behind the input currently being processed). - * - * The most straightforward technique turns out to be the fastest for - * most input files: try all possible matches and select the longest. - * The key feature of this algorithm is that insertions into the string - * dictionary are very simple and thus fast, and deletions are avoided - * completely. Insertions are performed at each input character, whereas - * string matches are performed only when the previous match ends. So it - * is preferable to spend more time in matches to allow very fast string - * insertions and avoid deletions. The matching algorithm for small - * strings is inspired from that of Rabin & Karp. A brute force approach - * is used to find longer strings when a small match has been found. - * A similar algorithm is used in comic (by Jan-Mark Wams) and freeze - * (by Leonid Broukhis). - * A previous version of this file used a more sophisticated algorithm - * (by Fiala and Greene) which is guaranteed to run in linear amortized - * time, but has a larger average cost, uses more memory and is patented. - * However the F&G algorithm may be faster for some highly redundant - * files if the parameter max_chain_length (described below) is too large. - * - * ACKNOWLEDGEMENTS - * - * The idea of lazy evaluation of matches is due to Jan-Mark Wams, and - * I found it in 'freeze' written by Leonid Broukhis. - * Thanks to many info-zippers for bug reports and testing. - * - * REFERENCES - * - * APPNOTE.TXT documentation file in PKZIP 1.93a distribution. - * - * A description of the Rabin and Karp algorithm is given in the book - * "Algorithms" by R. Sedgewick, Addison-Wesley, p252. - * - * Fiala,E.R., and Greene,D.H. - * Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595 - * - * INTERFACE - * - * void lm_init (int pack_level, ush *flags) - * Initialize the "longest match" routines for a new file - * - * ulg deflate (void) - * Processes a new input file and return its compressed length. Sets - * the compressed length, crc, deflate flags and internal file - * attributes. - */ - /* =========================================================================== - * Configuration parameters - */ - -/* Compile with MEDIUM_MEM to reduce the memory requirements or - * with SMALL_MEM to use as little memory as possible. Use BIG_MEM if the - * entire input file can be held in memory (not possible on 16 bit systems). - * Warning: defining these symbols affects HASH_BITS (see below) and thus - * affects the compression ratio. The compressed output - * is still correct, and might even be smaller in some cases. - */ - -#ifdef SMALL_MEM -# define HASH_BITS 13 /* Number of bits used to hash strings */ -#endif -#ifdef MEDIUM_MEM -# define HASH_BITS 14 -#endif -#ifndef HASH_BITS -# define HASH_BITS 15 - /* For portability to 16 bit machines, do not use values above 15. */ -#endif - -/* To save space (see unlzw.c), we overlay prev+head with tab_prefix and - * window with tab_suffix. Check that we can do this: - */ -#if (WSIZE<<1) > (1< BITS-1 -# error cannot overlay head with tab_prefix1 -#endif -#define HASH_SIZE (unsigned)(1<= HASH_BITS - */ - -static unsigned int prev_length; - -/* Length of the best match at previous step. Matches not greater than this - * are discarded. This is used in the lazy match evaluation. - */ - -static unsigned strstart; /* start of string to insert */ -static unsigned match_start; /* start of matching string */ -static int eofile; /* flag set at end of input file */ -static unsigned lookahead; /* number of valid bytes ahead in window */ - -static const unsigned max_chain_length=4096; - -/* To speed up deflation, hash chains are never searched beyond this length. - * A higher limit improves compression ratio but degrades the speed. - */ - -static const unsigned int max_lazy_match=258; - -/* Attempt to find a better match only when the current match is strictly - * smaller than this value. This mechanism is used only for compression - * levels >= 4. - */ -#define max_insert_length max_lazy_match -/* Insert new strings in the hash table only if the match length - * is not greater than this length. This saves time but degrades compression. - * max_insert_length is used only for compression levels <= 3. + * Read a new buffer from the current input file, perform end-of-line + * translation, and update the crc and input file size. + * IN assertion: size >= 2 (for end-of-line translation) */ +static unsigned file_read(void *buf, unsigned size) +{ + unsigned len; -static const unsigned good_match=32; - -/* Use a faster search when the previous match is longer than this */ - - -/* Values for max_lazy_match, good_match and max_chain_length, depending on - * the desired pack level (0..9). The values given below have been tuned to - * exclude worst case performance for pathological files. Better values may be - * found for specific files. - */ + Assert(G1.insize == 0, "l_buf not empty"); -static const int nice_match=258; /* Stop searching when current match exceeds this */ + len = safe_read(ifd, buf, size); + if (len == (unsigned)(-1) || len == 0) + return len; -/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 - * For deflate_fast() (levels <= 3) good is ignored and lazy has a different - * meaning. - */ + updcrc(buf, len); + G1.isize += len; + return len; +} -#define EQUAL 0 -/* result of memcmp for equal strings */ /* =========================================================================== - * Prototypes for local functions. + * Send a value on a given number of bits. + * IN assertion: length <= 16 and value fits in length bits. */ -static void fill_window (void); - -static int longest_match (IPos cur_match); - +static void send_bits(int value, int length) +{ #ifdef DEBUG -static void check_match (IPos start, IPos match, int length); + Tracev((stderr, " l %2d v %4x ", length, value)); + Assert(length > 0 && length <= 15, "invalid length"); + G1.bits_sent += length; #endif - -/* =========================================================================== - * Update a hash value with the given input byte - * IN assertion: all calls to to UPDATE_HASH are made with consecutive - * input characters, so that a running hash key can be computed from the - * previous key instead of complete recalculation each time. - */ -#define UPDATE_HASH(h,c) (h = (((h)< (int) BUF_SIZE - length) { + G1.bi_buf |= (value << G1.bi_valid); + put_16bit(G1.bi_buf); + G1.bi_buf = (ush) value >> (BUF_SIZE - G1.bi_valid); + G1.bi_valid += length - BUF_SIZE; + } else { + G1.bi_buf |= value << G1.bi_valid; + G1.bi_valid += length; + } } -/* =========================================================================== - * Set match_start to the longest match starting at the given string and - * return its length. Matches shorter or equal to prev_length are discarded, - * in which case the result is equal to prev_length and match_start is - * garbage. - * IN assertions: cur_match is the head of the hash chain for the current - * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 - */ -/* For MSDOS, OS/2 and 386 Unix, an optimized version is in match.asm or - * match.s. The code is functionally equivalent, so you can use the C version - * if desired. +/* =========================================================================== + * Reverse the first len bits of a code, using straightforward code (a faster + * method would use a table) + * IN assertion: 1 <= len <= 15 */ -static int longest_match(IPos cur_match) +static unsigned bi_reverse(unsigned code, int len) { - unsigned chain_length = max_chain_length; /* max hash chain length */ - register uch *scan = window + strstart; /* current string */ - register uch *match; /* matched string */ - register int len; /* length of current match */ - int best_len = prev_length; /* best match length so far */ - IPos limit = - - strstart > (IPos) MAX_DIST ? strstart - (IPos) MAX_DIST : NIL; - /* Stop when cur_match becomes <= limit. To simplify the code, - * we prevent matches with the string of window index 0. - */ + unsigned res = 0; -/* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. - * It is easy to get rid of this optimization if necessary. - */ -#if HASH_BITS < 8 || MAX_MATCH != 258 -# error Code too clever -#endif - register uch *strend = window + strstart + MAX_MATCH; - register uch scan_end1 = scan[best_len - 1]; - register uch scan_end = scan[best_len]; - - /* Do not waste too much time if we already have a good match: */ - if (prev_length >= good_match) { - chain_length >>= 2; + while (1) { + res |= code & 1; + if (--len <= 0) return res; + code >>= 1; + res <<= 1; } - Assert(strstart <= window_size - MIN_LOOKAHEAD, - "insufficient lookahead"); - - do { - Assert(cur_match < strstart, "no future"); - match = window + cur_match; - - /* Skip to next match if the match length cannot increase - * or if the match length is less than 2: - */ - if (match[best_len] != scan_end || - match[best_len - 1] != scan_end1 || - *match != *scan || *++match != scan[1]) - continue; - - /* The check at best_len-1 can be removed because it will be made - * again later. (This heuristic is not always a win.) - * It is not necessary to compare scan[2] and match[2] since they - * are always equal when the other bytes match, given that - * the hash keys are equal and that HASH_BITS >= 8. - */ - scan += 2, match++; - - /* We check for insufficient lookahead only every 8th comparison; - * the 256th check will be made at strstart+258. - */ - do { - } while (*++scan == *++match && *++scan == *++match && - *++scan == *++match && *++scan == *++match && - *++scan == *++match && *++scan == *++match && - *++scan == *++match && *++scan == *++match && - scan < strend); - - len = MAX_MATCH - (int) (strend - scan); - scan = strend - MAX_MATCH; - - if (len > best_len) { - match_start = cur_match; - best_len = len; - if (len >= nice_match) - break; - scan_end1 = scan[best_len - 1]; - scan_end = scan[best_len]; - } - } while ((cur_match = prev[cur_match & WMASK]) > limit - && --chain_length != 0); - - return best_len; } -#ifdef DEBUG -/* =========================================================================== - * Check that the match at match_start is indeed a match. - */ -static void check_match(IPos start, IPos match, int length) -{ - /* check that the match is indeed a match */ - if (memcmp((char *) window + match, - (char *) window + start, length) != EQUAL) { - fprintf(stderr, - " start %d, match %d, length %d\n", start, match, length); - error_msg("invalid match"); - } - if (verbose > 1) { - fprintf(stderr, "\\[%d,%d]", start - match, length); - do { - putc(window[start++], stderr); - } while (--length != 0); - } -} -#else -# define check_match(start, match, length) -#endif /* =========================================================================== - * Fill the window when the lookahead becomes insufficient. - * Updates strstart and lookahead, and sets eofile if end of input file. - * IN assertion: lookahead < MIN_LOOKAHEAD && strstart + lookahead > 0 - * OUT assertions: at least one byte has been read, or eofile is set; - * file reads are performed for at least two bytes (required for the - * translate_eol option). + * Write out any remaining bits in an incomplete byte. */ -static void fill_window() +static void bi_windup(void) { - register unsigned n, m; - unsigned more = - - (unsigned) (window_size - (ulg) lookahead - (ulg) strstart); - /* Amount of free space at the end of the window. */ - - /* If the window is almost full and there is insufficient lookahead, - * move the upper half to the lower one to make room in the upper half. - */ - if (more == (unsigned) EOF) { - /* Very unlikely, but possible on 16 bit machine if strstart == 0 - * and lookahead == 1 (input done one byte at time) - */ - more--; - } else if (strstart >= WSIZE + MAX_DIST) { - /* By the IN assertion, the window is not empty so we can't confuse - * more == 0 with more == 64K on a 16 bit machine. - */ - Assert(window_size == (ulg) 2 * WSIZE, "no sliding with BIG_MEM"); - - memcpy((char *) window, (char *) window + WSIZE, (unsigned) WSIZE); - match_start -= WSIZE; - strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */ - - block_start -= (long) WSIZE; - - for (n = 0; n < HASH_SIZE; n++) { - m = head[n]; - head[n] = (Pos) (m >= WSIZE ? m - WSIZE : NIL); - } - for (n = 0; n < WSIZE; n++) { - m = prev[n]; - prev[n] = (Pos) (m >= WSIZE ? m - WSIZE : NIL); - /* If n is not on any hash chain, prev[n] is garbage but - * its value will never be used. - */ - } - more += WSIZE; - } - /* At this point, more >= 2 */ - if (!eofile) { - n = read_buf((char *) window + strstart + lookahead, more); - if (n == 0 || n == (unsigned) EOF) { - eofile = 1; - } else { - lookahead += n; - } + if (G1.bi_valid > 8) { + put_16bit(G1.bi_buf); + } else if (G1.bi_valid > 0) { + put_8bit(G1.bi_buf); } + G1.bi_buf = 0; + G1.bi_valid = 0; +#ifdef DEBUG + G1.bits_sent = (G1.bits_sent + 7) & ~7; +#endif } -/* =========================================================================== - * Flush the current block, with given end-of-file flag. - * IN assertion: strstart is set to the end of the current match. - */ -#define FLUSH_BLOCK(eof) \ - flush_block(block_start >= 0L ? (char*)&window[(unsigned)block_start] : \ - (char*)NULL, (long)strstart - block_start, (eof)) /* =========================================================================== - * Same as above, but achieves better compression. We use a lazy - * evaluation for matches: a match is finally adopted only if there is - * no better match at the next window position. + * Copy a stored block to the zip file, storing first the length and its + * one's complement if requested. */ -static ulg deflate() +static void copy_block(char *buf, unsigned len, int header) { - IPos hash_head; /* head of hash chain */ - IPos prev_match; /* previous match */ - int flush; /* set if current block must be flushed */ - int match_available = 0; /* set if previous match exists */ - register unsigned match_length = MIN_MATCH - 1; /* length of best match */ + bi_windup(); /* align on byte boundary */ - /* Process the input block. */ - while (lookahead != 0) { - /* Insert the string window[strstart .. strstart+2] in the - * dictionary, and set hash_head to the head of the hash chain: - */ - INSERT_STRING(strstart, hash_head); - - /* Find the longest match, discarding those <= prev_length. - */ - prev_length = match_length, prev_match = match_start; - match_length = MIN_MATCH - 1; - - if (hash_head != NIL && prev_length < max_lazy_match && - strstart - hash_head <= MAX_DIST) { - /* To simplify the code, we prevent matches with the string - * of window index 0 (in particular we have to avoid a match - * of the string with itself at the start of the input file). - */ - match_length = longest_match(hash_head); - /* longest_match() sets match_start */ - if (match_length > lookahead) - match_length = lookahead; - - /* Ignore a length 3 match if it is too distant: */ - if (match_length == MIN_MATCH - && strstart - match_start > TOO_FAR) { - /* If prev_match is also MIN_MATCH, match_start is garbage - * but we will ignore the current match anyway. - */ - match_length--; - } - } - /* If there was a match at the previous step and the current - * match is not better, output the previous match: - */ - if (prev_length >= MIN_MATCH && match_length <= prev_length) { - - check_match(strstart - 1, prev_match, prev_length); - - flush = - ct_tally(strstart - 1 - prev_match, - prev_length - MIN_MATCH); - - /* Insert in hash table all strings up to the end of the match. - * strstart-1 and strstart are already inserted. - */ - lookahead -= prev_length - 1; - prev_length -= 2; - do { - strstart++; - INSERT_STRING(strstart, hash_head); - /* strstart never exceeds WSIZE-MAX_MATCH, so there are - * always MIN_MATCH bytes ahead. If lookahead < MIN_MATCH - * these bytes are garbage, but it does not matter since the - * next lookahead bytes will always be emitted as literals. - */ - } while (--prev_length != 0); - match_available = 0; - match_length = MIN_MATCH - 1; - strstart++; - if (flush) - FLUSH_BLOCK(0), block_start = strstart; - - } else if (match_available) { - /* If there was no match at the previous position, output a - * single literal. If there was a match but the current match - * is longer, truncate the previous match to a single literal. - */ - Tracevv((stderr, "%c", window[strstart - 1])); - if (ct_tally(0, window[strstart - 1])) { - FLUSH_BLOCK(0), block_start = strstart; - } - strstart++; - lookahead--; - } else { - /* There is no previous match to compare with, wait for - * the next step to decide. - */ - match_available = 1; - strstart++; - lookahead--; - } - Assert(strstart <= isize && lookahead <= isize, "a bit too far"); - - /* Make sure that we always have enough lookahead, except - * at the end of the input file. We need MAX_MATCH bytes - * for the next match, plus MIN_MATCH bytes to insert the - * string following the next match. - */ - while (lookahead < MIN_LOOKAHEAD && !eofile) - fill_window(); + if (header) { + put_16bit(len); + put_16bit(~len); +#ifdef DEBUG + G1.bits_sent += 2 * 16; +#endif + } +#ifdef DEBUG + G1.bits_sent += (ulg) len << 3; +#endif + while (len--) { + put_8bit(*buf++); } - if (match_available) - ct_tally(0, window[strstart - 1]); - - return FLUSH_BLOCK(1); /* eof */ } -/* gzip (GNU zip) -- compress files with zip algorithm and 'compress' interface - * Copyright (C) 1992-1993 Jean-loup Gailly - * The unzip code was written and put in the public domain by Mark Adler. - * Portions of the lzw code are derived from the public domain 'compress' - * written by Spencer Thomas, Joe Orost, James Woods, Jim McKie, Steve Davies, - * Ken Turkowski, Dave Mack and Peter Jannesen. - * - * See the license_msg below and the file COPYING for the software license. - * See the file algorithm.doc for the compression algorithms and file formats. - */ - -/* Compress files with zip algorithm and 'compress' interface. - * See usage() and help() functions below for all options. - * Outputs: - * file.gz: compressed file with same mode, owner, and utimes - * or stdout with -c option or if stdin used as input. - * If the output file name had to be truncated, the original name is kept - * in the compressed file. - */ - /* configuration */ +/* =========================================================================== + * Fill the window when the lookahead becomes insufficient. + * Updates strstart and lookahead, and sets eofile if end of input file. + * IN assertion: lookahead < MIN_LOOKAHEAD && strstart + lookahead > 0 + * OUT assertions: at least one byte has been read, or eofile is set; + * file reads are performed for at least two bytes (required for the + * translate_eol option). + */ +static void fill_window(void) +{ + unsigned n, m; + unsigned more = WINDOW_SIZE - G1.lookahead - G1.strstart; + /* Amount of free space at the end of the window. */ -typedef struct dirent dir_type; + /* If the window is almost full and there is insufficient lookahead, + * move the upper half to the lower one to make room in the upper half. + */ + if (more == (unsigned) -1) { + /* Very unlikely, but possible on 16 bit machine if strstart == 0 + * and lookahead == 1 (input done one byte at time) + */ + more--; + } else if (G1.strstart >= WSIZE + MAX_DIST) { + /* By the IN assertion, the window is not empty so we can't confuse + * more == 0 with more == 64K on a 16 bit machine. + */ + Assert(WINDOW_SIZE == 2 * WSIZE, "no sliding with BIG_MEM"); -typedef RETSIGTYPE(*sig_type) (int); + memcpy(G1.window, G1.window + WSIZE, WSIZE); + G1.match_start -= WSIZE; + G1.strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */ + G1.block_start -= WSIZE; -/* ======================================================================== */ -// int main (argc, argv) -// int argc; -// char **argv; -int gzip_main(int argc, char **argv) -{ - int result; - int inFileNum; - int outFileNum; - struct stat statBuf; - char *delFileName; - int tostdout = 0; - int fromstdin = 0; - int force = 0; - int opt; - - while ((opt = getopt(argc, argv, "cf123456789dq")) != -1) { - switch (opt) { - case 'c': - tostdout = 1; - break; - case 'f': - force = 1; - break; - /* Ignore 1-9 (compression level) options */ - case '1': case '2': case '3': case '4': case '5': - case '6': case '7': case '8': case '9': - break; - case 'q': - break; -#ifdef BB_GUNZIP - case 'd': - optind = 1; - return gunzip_main(argc, argv); -#endif - default: - show_usage(); + for (n = 0; n < HASH_SIZE; n++) { + m = head[n]; + head[n] = (Pos) (m >= WSIZE ? m - WSIZE : 0); + } + for (n = 0; n < WSIZE; n++) { + m = G1.prev[n]; + G1.prev[n] = (Pos) (m >= WSIZE ? m - WSIZE : 0); + /* If n is not on any hash chain, prev[n] is garbage but + * its value will never be used. + */ } + more += WSIZE; } - if (optind == argc) { - fromstdin = 1; - tostdout = 1; + /* At this point, more >= 2 */ + if (!G1.eofile) { + n = file_read(G1.window + G1.strstart + G1.lookahead, more); + if (n == 0 || n == (unsigned) -1) { + G1.eofile = 1; + } else { + G1.lookahead += n; + } } +} - if (isatty(fileno(stdout)) && tostdout==1 && force==0) - error_msg_and_die( "compressed data not written to terminal. Use -f to force it."); - foreground = signal(SIGINT, SIG_IGN) != SIG_IGN; - if (foreground) { - (void) signal(SIGINT, (sig_type) abort_gzip); - } -#ifdef SIGTERM - if (signal(SIGTERM, SIG_IGN) != SIG_IGN) { - (void) signal(SIGTERM, (sig_type) abort_gzip); - } -#endif -#ifdef SIGHUP - if (signal(SIGHUP, SIG_IGN) != SIG_IGN) { - (void) signal(SIGHUP, (sig_type) abort_gzip); - } -#endif +/* =========================================================================== + * Set match_start to the longest match starting at the given string and + * return its length. Matches shorter or equal to prev_length are discarded, + * in which case the result is equal to prev_length and match_start is + * garbage. + * IN assertions: cur_match is the head of the hash chain for the current + * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 + */ + +/* For MSDOS, OS/2 and 386 Unix, an optimized version is in match.asm or + * match.s. The code is functionally equivalent, so you can use the C version + * if desired. + */ +static int longest_match(IPos cur_match) +{ + unsigned chain_length = max_chain_length; /* max hash chain length */ + uch *scan = G1.window + G1.strstart; /* current string */ + uch *match; /* matched string */ + int len; /* length of current match */ + int best_len = G1.prev_length; /* best match length so far */ + IPos limit = G1.strstart > (IPos) MAX_DIST ? G1.strstart - (IPos) MAX_DIST : 0; + /* Stop when cur_match becomes <= limit. To simplify the code, + * we prevent matches with the string of window index 0. + */ - strncpy(z_suffix, Z_SUFFIX, sizeof(z_suffix) - 1); - z_len = strlen(z_suffix); +/* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. + * It is easy to get rid of this optimization if necessary. + */ +#if HASH_BITS < 8 || MAX_MATCH != 258 +# error Code too clever +#endif + uch *strend = G1.window + G1.strstart + MAX_MATCH; + uch scan_end1 = scan[best_len - 1]; + uch scan_end = scan[best_len]; - /* Allocate all global buffers (for DYN_ALLOC option) */ - ALLOC(uch, inbuf, INBUFSIZ + INBUF_EXTRA); - ALLOC(uch, outbuf, OUTBUFSIZ + OUTBUF_EXTRA); - ALLOC(ush, d_buf, DIST_BUFSIZE); - ALLOC(uch, window, 2L * WSIZE); - ALLOC(ush, tab_prefix, 1L << BITS); - - if (fromstdin == 1) { - strcpy(ofname, "stdin"); - - inFileNum = fileno(stdin); - time_stamp = 0; /* time unknown by default */ - ifile_size = -1L; /* convention for unknown size */ - } else { - /* Open up the input file */ - strncpy(ifname, argv[optind], MAX_PATH_LEN); - - /* Open input file */ - inFileNum = open(ifname, O_RDONLY); - if (inFileNum < 0) - perror_msg_and_die("%s", ifname); - /* Get the time stamp on the input file. */ - if (stat(ifname, &statBuf) < 0) - perror_msg_and_die("%s", ifname); - time_stamp = statBuf.st_ctime; - ifile_size = statBuf.st_size; + /* Do not waste too much time if we already have a good match: */ + if (G1.prev_length >= good_match) { + chain_length >>= 2; } + Assert(G1.strstart <= WINDOW_SIZE - MIN_LOOKAHEAD, "insufficient lookahead"); + + do { + Assert(cur_match < G1.strstart, "no future"); + match = G1.window + cur_match; + /* Skip to next match if the match length cannot increase + * or if the match length is less than 2: + */ + if (match[best_len] != scan_end || + match[best_len - 1] != scan_end1 || + *match != *scan || *++match != scan[1]) + continue; - if (tostdout == 1) { - /* And get to work */ - strcpy(ofname, "stdout"); - outFileNum = fileno(stdout); + /* The check at best_len-1 can be removed because it will be made + * again later. (This heuristic is not always a win.) + * It is not necessary to compare scan[2] and match[2] since they + * are always equal when the other bytes match, given that + * the hash keys are equal and that HASH_BITS >= 8. + */ + scan += 2, match++; - clear_bufs(); /* clear input and output buffers */ - part_nb = 0; + /* We check for insufficient lookahead only every 8th comparison; + * the 256th check will be made at strstart+258. + */ + do { + } while (*++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && scan < strend); - /* Actually do the compression/decompression. */ - zip(inFileNum, outFileNum); + len = MAX_MATCH - (int) (strend - scan); + scan = strend - MAX_MATCH; - } else { + if (len > best_len) { + G1.match_start = cur_match; + best_len = len; + if (len >= nice_match) + break; + scan_end1 = scan[best_len - 1]; + scan_end = scan[best_len]; + } + } while ((cur_match = G1.prev[cur_match & WMASK]) > limit + && --chain_length != 0); - /* And get to work */ - strncpy(ofname, ifname, MAX_PATH_LEN - 4); - strcat(ofname, ".gz"); + return best_len; +} - /* Open output fille */ -#if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1) - outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW); +#ifdef DEBUG +/* =========================================================================== + * Check that the match at match_start is indeed a match. + */ +static void check_match(IPos start, IPos match, int length) +{ + /* check that the match is indeed a match */ + if (memcmp(G1.window + match, G1.window + start, length) != 0) { + bb_error_msg(" start %d, match %d, length %d", start, match, length); + bb_error_msg("invalid match"); + } + if (verbose > 1) { + bb_error_msg("\\[%d,%d]", start - match, length); + do { + fputc(G1.window[start++], stderr); + } while (--length != 0); + } +} #else - outFileNum = open(ofname, O_RDWR | O_CREAT | O_EXCL); +# define check_match(start, match, length) ((void)0) #endif - if (outFileNum < 0) - perror_msg_and_die("%s", ofname); - /* Set permissions on the file */ - fchmod(outFileNum, statBuf.st_mode); - - clear_bufs(); /* clear input and output buffers */ - part_nb = 0; - - /* Actually do the compression/decompression. */ - result = zip(inFileNum, outFileNum); - close(outFileNum); - close(inFileNum); - /* Delete the original file */ - if (result == OK) - delFileName = ifname; - else - delFileName = ofname; - - if (unlink(delFileName) < 0) - perror_msg_and_die("%s", delFileName); - } - return(exit_code); -} /* trees.c -- output deflated data using Huffman coding * Copyright (C) 1992-1993 Jean-loup Gailly @@ -1356,14 +702,11 @@ int gzip_main(int argc, char **argv) * terms of the GNU General Public License, see the file COPYING. */ -/* - * PURPOSE - * +/* PURPOSE * Encode various sets of source values using variable-length * binary code trees. * * DISCUSSION - * * The PKZIP "deflation" process uses several Huffman trees. The more * common source values are represented by shorter bit sequences. * @@ -1375,7 +718,6 @@ int gzip_main(int argc, char **argv) * (APPNOTE.TXT) distributed as part of PKWARE's PKZIP program. * * REFERENCES - * * Lynch, Thomas J. * Data Compression: Techniques and Applications, pp. 53-55. * Lifetime Learning Publications, 1985. ISBN 0-534-03418-7. @@ -1389,24 +731,18 @@ int gzip_main(int argc, char **argv) * Addison-Wesley, 1983. ISBN 0-201-06672-6. * * INTERFACE - * - * void ct_init (ush *attr, int *methodp) - * Allocate the match buffer, initialize the various tables and save + * void ct_init() + * Allocate the match buffer, initialize the various tables [and save * the location of the internal file attribute (ascii/binary) and - * method (DEFLATE/STORE) + * method (DEFLATE/STORE) -- deleted in bbox] * - * void ct_tally (int dist, int lc); + * void ct_tally(int dist, int lc); * Save the match info and tally the frequency counts. * - * long flush_block (char *buf, ulg stored_len, int eof) + * ulg flush_block(char *buf, ulg stored_len, int eof) * Determine the best encoding for the current block: dynamic trees, * static trees or store, and output the encoded block to the zip * file. Returns the total compressed length for the file so far. - * - */ - -/* =========================================================================== - * Constants */ #define MAX_BITS 15 @@ -1433,17 +769,25 @@ int gzip_main(int argc, char **argv) #define BL_CODES 19 /* number of codes used to transfer the bit lengths */ +/* extra bits for each length code */ +static const uint8_t extra_lbits[LENGTH_CODES] ALIGN1 = { + 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, + 4, 4, 5, 5, 5, 5, 0 +}; -static const int extra_lbits[LENGTH_CODES] /* extra bits for each length code */ - = { 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, - 4, 4, 5, 5, 5, 5, 0 }; +/* extra bits for each distance code */ +static const uint8_t extra_dbits[D_CODES] ALIGN1 = { + 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, + 10, 10, 11, 11, 12, 12, 13, 13 +}; -static const int extra_dbits[D_CODES] /* extra bits for each distance code */ - = { 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, - 10, 10, 11, 11, 12, 12, 13, 13 }; +/* extra bits for each bit length code */ +static const uint8_t extra_blbits[BL_CODES] ALIGN1 = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 3, 7 }; -static const int extra_blbits[BL_CODES] /* extra bits for each bit length code */ -= { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 3, 7 }; +/* number of codes at each bit length for an optimal tree */ +static const uint8_t bl_order[BL_CODES] ALIGN1 = { + 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15 }; #define STORED_BLOCK 0 #define STATIC_TREES 1 @@ -1484,24 +828,24 @@ static const int extra_blbits[BL_CODES] /* extra bits for each bit length code * * memory at the expense of compression). Some optimizations would be possible * if we rely on DIST_BUFSIZE == LIT_BUFSIZE. */ -#if LIT_BUFSIZE > INBUFSIZ -error cannot overlay l_buf and inbuf -#endif #define REP_3_6 16 /* repeat previous bit length 3-6 times (2 bits of repeat count) */ #define REPZ_3_10 17 /* repeat a zero length 3-10 times (3 bits of repeat count) */ #define REPZ_11_138 18 -/* repeat a zero length 11-138 times (7 bits of repeat count) *//* =========================================================================== - * Local data - *//* Data structure describing a single value and its code string. */ typedef struct ct_data { +/* repeat a zero length 11-138 times (7 bits of repeat count) */ + +/* =========================================================================== +*/ +/* Data structure describing a single value and its code string. */ +typedef struct ct_data { union { - ush freq; /* frequency count */ - ush code; /* bit string */ + ush freq; /* frequency count */ + ush code; /* bit string */ } fc; union { - ush dad; /* father node in Huffman tree */ - ush len; /* length of bit string */ + ush dad; /* father node in Huffman tree */ + ush len; /* length of bit string */ } dl; } ct_data; @@ -1510,13 +854,33 @@ error cannot overlay l_buf and inbuf #define Dad dl.dad #define Len dl.len -#define HEAP_SIZE (2*L_CODES+1) +#define HEAP_SIZE (2*L_CODES + 1) /* maximum heap size */ -static ct_data dyn_ltree[HEAP_SIZE]; /* literal and length tree */ -static ct_data dyn_dtree[2 * D_CODES + 1]; /* distance tree */ +typedef struct tree_desc { + ct_data *dyn_tree; /* the dynamic tree */ + ct_data *static_tree; /* corresponding static tree or NULL */ + const uint8_t *extra_bits; /* extra bits for each code or NULL */ + int extra_base; /* base index for extra_bits */ + int elems; /* max number of elements in the tree */ + int max_length; /* max bit length for the codes */ + int max_code; /* largest code with non zero frequency */ +} tree_desc; + +struct globals2 { + + ush heap[HEAP_SIZE]; /* heap used to build the Huffman trees */ + int heap_len; /* number of elements in the heap */ + int heap_max; /* element of largest frequency */ -static ct_data static_ltree[L_CODES + 2]; +/* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used. + * The same heap array is used to build all trees. + */ + + ct_data dyn_ltree[HEAP_SIZE]; /* literal and length tree */ + ct_data dyn_dtree[2 * D_CODES + 1]; /* distance tree */ + + ct_data static_ltree[L_CODES + 2]; /* The static literal tree. Since the bit lengths are imposed, there is no * need for the L_CODES extra codes used during heap construction. However @@ -1524,272 +888,129 @@ static ct_data static_ltree[L_CODES + 2]; * below). */ -static ct_data static_dtree[D_CODES]; + ct_data static_dtree[D_CODES]; /* The static distance tree. (Actually a trivial tree since all codes use * 5 bits.) */ -static ct_data bl_tree[2 * BL_CODES + 1]; + ct_data bl_tree[2 * BL_CODES + 1]; /* Huffman tree for the bit lengths */ -typedef struct tree_desc { - ct_data *dyn_tree; /* the dynamic tree */ - ct_data *static_tree; /* corresponding static tree or NULL */ - const int *extra_bits; /* extra bits for each code or NULL */ - int extra_base; /* base index for extra_bits */ - int elems; /* max number of elements in the tree */ - int max_length; /* max bit length for the codes */ - int max_code; /* largest code with non zero frequency */ -} tree_desc; - -static tree_desc l_desc = - { dyn_ltree, static_ltree, extra_lbits, LITERALS + 1, L_CODES, - MAX_BITS, 0 }; - -static tree_desc d_desc = - { dyn_dtree, static_dtree, extra_dbits, 0, D_CODES, MAX_BITS, 0 }; - -static tree_desc bl_desc = - { bl_tree, (ct_data *) 0, extra_blbits, 0, BL_CODES, MAX_BL_BITS, - 0 }; - + tree_desc l_desc; + tree_desc d_desc; + tree_desc bl_desc; -static ush bl_count[MAX_BITS + 1]; - -/* number of codes at each bit length for an optimal tree */ - -static const uch bl_order[BL_CODES] -= { 16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15 }; + ush bl_count[MAX_BITS + 1]; /* The lengths of the bit length codes are sent in order of decreasing * probability, to avoid transmitting the lengths for unused bit length codes. */ -static int heap[2 * L_CODES + 1]; /* heap used to build the Huffman trees */ -static int heap_len; /* number of elements in the heap */ -static int heap_max; /* element of largest frequency */ - -/* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used. - * The same heap array is used to build all trees. - */ - -static uch depth[2 * L_CODES + 1]; + uch depth[2 * L_CODES + 1]; /* Depth of each subtree used as tie breaker for trees of equal frequency */ -static uch length_code[MAX_MATCH - MIN_MATCH + 1]; + uch length_code[MAX_MATCH - MIN_MATCH + 1]; /* length code for each normalized match length (0 == MIN_MATCH) */ -static uch dist_code[512]; + uch dist_code[512]; /* distance codes. The first 256 values correspond to the distances * 3 .. 258, the last 256 values correspond to the top 8 bits of * the 15 bit distances. */ -static int base_length[LENGTH_CODES]; + int base_length[LENGTH_CODES]; /* First normalized length for each code (0 = MIN_MATCH) */ -static int base_dist[D_CODES]; + int base_dist[D_CODES]; /* First normalized distance for each code (0 = distance of 1) */ -#define l_buf inbuf -/* DECLARE(uch, l_buf, LIT_BUFSIZE); buffer for literals or lengths */ - -/* DECLARE(ush, d_buf, DIST_BUFSIZE); buffer for distances */ - -static uch flag_buf[(LIT_BUFSIZE / 8)]; + uch flag_buf[LIT_BUFSIZE / 8]; /* flag_buf is a bit array distinguishing literals from lengths in * l_buf, thus indicating the presence or absence of a distance. */ -static unsigned last_lit; /* running index in l_buf */ -static unsigned last_dist; /* running index in d_buf */ -static unsigned last_flags; /* running index in flag_buf */ -static uch flags; /* current flags not yet saved in flag_buf */ -static uch flag_bit; /* current bit used in flags */ + unsigned last_lit; /* running index in l_buf */ + unsigned last_dist; /* running index in d_buf */ + unsigned last_flags; /* running index in flag_buf */ + uch flags; /* current flags not yet saved in flag_buf */ + uch flag_bit; /* current bit used in flags */ /* bits are filled in flags starting at bit 0 (least significant). * Note: these flags are overkill in the current code since we don't * take advantage of DIST_BUFSIZE == LIT_BUFSIZE. */ -static ulg opt_len; /* bit length of current block with optimal trees */ -static ulg static_len; /* bit length of current block with static trees */ + ulg opt_len; /* bit length of current block with optimal trees */ + ulg static_len; /* bit length of current block with static trees */ -static ulg compressed_len; /* total bit length of compressed file */ + ulg compressed_len; /* total bit length of compressed file */ +}; +#define G2ptr ((struct globals2*)(ptr_to_globals)) +#define G2 (*G2ptr) -static ush *file_type; /* pointer to UNKNOWN, BINARY or ASCII */ -static int *file_method; /* pointer to DEFLATE or STORE */ /* =========================================================================== - * Local (static) routines in this file. */ - -static void init_block (void); -static void pqdownheap (ct_data * tree, int k); -static void gen_bitlen (tree_desc * desc); -static void gen_codes (ct_data * tree, int max_code); -static void build_tree (tree_desc * desc); -static void scan_tree (ct_data * tree, int max_code); -static void send_tree (ct_data * tree, int max_code); -static int build_bl_tree (void); -static void send_all_trees (int lcodes, int dcodes, int blcodes); -static void compress_block (ct_data * ltree, ct_data * dtree); -static void set_file_type (void); +static void gen_codes(ct_data * tree, int max_code); +static void build_tree(tree_desc * desc); +static void scan_tree(ct_data * tree, int max_code); +static void send_tree(ct_data * tree, int max_code); +static int build_bl_tree(void); +static void send_all_trees(int lcodes, int dcodes, int blcodes); +static void compress_block(ct_data * ltree, ct_data * dtree); #ifndef DEBUG -# define send_code(c, tree) send_bits(tree[c].Code, tree[c].Len) - /* Send a code of the given tree. c and tree must not have side effects */ - -#else /* DEBUG */ -# define send_code(c, tree) \ - { if (verbose>1) fprintf(stderr,"\ncd %3d ",(c)); \ - send_bits(tree[c].Code, tree[c].Len); } +/* Send a code of the given tree. c and tree must not have side effects */ +# define SEND_CODE(c, tree) send_bits(tree[c].Code, tree[c].Len) +#else +# define SEND_CODE(c, tree) \ +{ \ + if (verbose > 1) bb_error_msg("\ncd %3d ",(c)); \ + send_bits(tree[c].Code, tree[c].Len); \ +} #endif -#define d_code(dist) \ - ((dist) < 256 ? dist_code[dist] : dist_code[256+((dist)>>7)]) +#define D_CODE(dist) \ + ((dist) < 256 ? G2.dist_code[dist] : G2.dist_code[256 + ((dist)>>7)]) /* Mapping from a distance to a distance code. dist is the distance - 1 and * must not have side effects. dist_code[256] and dist_code[257] are never * used. + * The arguments must not have side effects. */ -/* the arguments must not have side effects */ - -/* =========================================================================== - * Allocate the match buffer, initialize the various tables and save the - * location of the internal file attribute (ascii/binary) and method - * (DEFLATE/STORE). - */ -static void ct_init(ush *attr, int *methodp) -{ - int n; /* iterates over tree elements */ - int bits; /* bit counter */ - int length; /* length value */ - int code; /* code value */ - int dist; /* distance index */ - - file_type = attr; - file_method = methodp; - compressed_len = 0L; - - if (static_dtree[0].Len != 0) - return; /* ct_init already called */ - - /* Initialize the mapping length (0..255) -> length code (0..28) */ - length = 0; - for (code = 0; code < LENGTH_CODES - 1; code++) { - base_length[code] = length; - for (n = 0; n < (1 << extra_lbits[code]); n++) { - length_code[length++] = (uch) code; - } - } - Assert(length == 256, "ct_init: length != 256"); - /* Note that the length 255 (match length 258) can be represented - * in two different ways: code 284 + 5 bits or code 285, so we - * overwrite length_code[255] to use the best encoding: - */ - length_code[length - 1] = (uch) code; - - /* Initialize the mapping dist (0..32K) -> dist code (0..29) */ - dist = 0; - for (code = 0; code < 16; code++) { - base_dist[code] = dist; - for (n = 0; n < (1 << extra_dbits[code]); n++) { - dist_code[dist++] = (uch) code; - } - } - Assert(dist == 256, "ct_init: dist != 256"); - dist >>= 7; /* from now on, all distances are divided by 128 */ - for (; code < D_CODES; code++) { - base_dist[code] = dist << 7; - for (n = 0; n < (1 << (extra_dbits[code] - 7)); n++) { - dist_code[256 + dist++] = (uch) code; - } - } - Assert(dist == 256, "ct_init: 256+dist != 512"); - - /* Construct the codes of the static literal tree */ - for (bits = 0; bits <= MAX_BITS; bits++) - bl_count[bits] = 0; - n = 0; - while (n <= 143) - static_ltree[n++].Len = 8, bl_count[8]++; - while (n <= 255) - static_ltree[n++].Len = 9, bl_count[9]++; - while (n <= 279) - static_ltree[n++].Len = 7, bl_count[7]++; - while (n <= 287) - static_ltree[n++].Len = 8, bl_count[8]++; - /* Codes 286 and 287 do not exist, but we must include them in the - * tree construction to get a canonical Huffman tree (longest code - * all ones) - */ - gen_codes((ct_data *) static_ltree, L_CODES + 1); - - /* The static distance tree is trivial: */ - for (n = 0; n < D_CODES; n++) { - static_dtree[n].Len = 5; - static_dtree[n].Code = bi_reverse(n, 5); - } - - /* Initialize the first block of the first file: */ - init_block(); -} /* =========================================================================== * Initialize a new block. */ -static void init_block() +static void init_block(void) { - int n; /* iterates over tree elements */ + int n; /* iterates over tree elements */ /* Initialize the trees. */ for (n = 0; n < L_CODES; n++) - dyn_ltree[n].Freq = 0; + G2.dyn_ltree[n].Freq = 0; for (n = 0; n < D_CODES; n++) - dyn_dtree[n].Freq = 0; + G2.dyn_dtree[n].Freq = 0; for (n = 0; n < BL_CODES; n++) - bl_tree[n].Freq = 0; - - dyn_ltree[END_BLOCK].Freq = 1; - opt_len = static_len = 0L; - last_lit = last_dist = last_flags = 0; - flags = 0; - flag_bit = 1; -} - -#define SMALLEST 1 -/* Index within the heap array of least frequent node in the Huffman tree */ - + G2.bl_tree[n].Freq = 0; -/* =========================================================================== - * Remove the smallest element from the heap and recreate the heap with - * one less element. Updates heap and heap_len. - */ -#define pqremove(tree, top) \ -{\ - top = heap[SMALLEST]; \ - heap[SMALLEST] = heap[heap_len--]; \ - pqdownheap(tree, SMALLEST); \ + G2.dyn_ltree[END_BLOCK].Freq = 1; + G2.opt_len = G2.static_len = 0; + G2.last_lit = G2.last_dist = G2.last_flags = 0; + G2.flags = 0; + G2.flag_bit = 1; } -/* =========================================================================== - * Compares to subtrees, using the tree depth as tie breaker when - * the subtrees have equal frequency. This minimizes the worst case length. - */ -#define smaller(tree, n, m) \ - (tree[n].Freq < tree[m].Freq || \ - (tree[n].Freq == tree[m].Freq && depth[n] <= depth[m])) /* =========================================================================== * Restore the heap property by moving down the tree starting at node k, @@ -1797,30 +1018,38 @@ static void init_block() * when the heap property is re-established (each father smaller than its * two sons). */ -static void pqdownheap(ct_data *tree, int k) + +/* Compares to subtrees, using the tree depth as tie breaker when + * the subtrees have equal frequency. This minimizes the worst case length. */ +#define SMALLER(tree, n, m) \ + (tree[n].Freq < tree[m].Freq \ + || (tree[n].Freq == tree[m].Freq && G2.depth[n] <= G2.depth[m])) + +static void pqdownheap(ct_data * tree, int k) { - int v = heap[k]; - int j = k << 1; /* left son of k */ + int v = G2.heap[k]; + int j = k << 1; /* left son of k */ - while (j <= heap_len) { + while (j <= G2.heap_len) { /* Set j to the smallest of the two sons: */ - if (j < heap_len && smaller(tree, heap[j + 1], heap[j])) + if (j < G2.heap_len && SMALLER(tree, G2.heap[j + 1], G2.heap[j])) j++; /* Exit if v is smaller than both sons */ - if (smaller(tree, v, heap[j])) + if (SMALLER(tree, v, G2.heap[j])) break; /* Exchange v with the smallest son */ - heap[k] = heap[j]; + G2.heap[k] = G2.heap[j]; k = j; /* And continue down the tree, setting j to the left son of k */ j <<= 1; } - heap[k] = v; + G2.heap[k] = v; } + /* =========================================================================== * Compute the optimal bit lengths for a tree and update the total bit length * for the current block. @@ -1831,49 +1060,51 @@ static void pqdownheap(ct_data *tree, int k) * The length opt_len is updated; static_len is also updated if stree is * not null. */ -static void gen_bitlen(tree_desc *desc) +static void gen_bitlen(tree_desc * desc) { ct_data *tree = desc->dyn_tree; - const int *extra = desc->extra_bits; + const uint8_t *extra = desc->extra_bits; int base = desc->extra_base; int max_code = desc->max_code; int max_length = desc->max_length; ct_data *stree = desc->static_tree; - int h; /* heap index */ - int n, m; /* iterate over the tree elements */ - int bits; /* bit length */ - int xbits; /* extra bits */ - ush f; /* frequency */ - int overflow = 0; /* number of elements with bit length too large */ + int h; /* heap index */ + int n, m; /* iterate over the tree elements */ + int bits; /* bit length */ + int xbits; /* extra bits */ + ush f; /* frequency */ + int overflow = 0; /* number of elements with bit length too large */ for (bits = 0; bits <= MAX_BITS; bits++) - bl_count[bits] = 0; + G2.bl_count[bits] = 0; /* In a first pass, compute the optimal bit lengths (which may * overflow in the case of the bit length tree). */ - tree[heap[heap_max]].Len = 0; /* root of the heap */ + tree[G2.heap[G2.heap_max]].Len = 0; /* root of the heap */ - for (h = heap_max + 1; h < HEAP_SIZE; h++) { - n = heap[h]; + for (h = G2.heap_max + 1; h < HEAP_SIZE; h++) { + n = G2.heap[h]; bits = tree[tree[n].Dad].Len + 1; - if (bits > max_length) - bits = max_length, overflow++; + if (bits > max_length) { + bits = max_length; + overflow++; + } tree[n].Len = (ush) bits; /* We overwrite tree[n].Dad which is no longer needed */ if (n > max_code) - continue; /* not a leaf node */ + continue; /* not a leaf node */ - bl_count[bits]++; + G2.bl_count[bits]++; xbits = 0; if (n >= base) xbits = extra[n - base]; f = tree[n].Freq; - opt_len += (ulg) f *(bits + xbits); + G2.opt_len += (ulg) f *(bits + xbits); if (stree) - static_len += (ulg) f *(stree[n].Len + xbits); + G2.static_len += (ulg) f * (stree[n].Len + xbits); } if (overflow == 0) return; @@ -1884,11 +1115,11 @@ static void gen_bitlen(tree_desc *desc) /* Find the first bit length which could increase: */ do { bits = max_length - 1; - while (bl_count[bits] == 0) + while (G2.bl_count[bits] == 0) bits--; - bl_count[bits]--; /* move one leaf down the tree */ - bl_count[bits + 1] += 2; /* move one overflow item as its brother */ - bl_count[max_length]--; + G2.bl_count[bits]--; /* move one leaf down the tree */ + G2.bl_count[bits + 1] += 2; /* move one overflow item as its brother */ + G2.bl_count[max_length]--; /* The brother of the overflow item also moves one step up, * but this does not affect bl_count[max_length] */ @@ -1901,25 +1132,22 @@ static void gen_bitlen(tree_desc *desc) * from 'ar' written by Haruhiko Okumura.) */ for (bits = max_length; bits != 0; bits--) { - n = bl_count[bits]; + n = G2.bl_count[bits]; while (n != 0) { - m = heap[--h]; + m = G2.heap[--h]; if (m > max_code) continue; if (tree[m].Len != (unsigned) bits) { - Trace( - (stderr, "code %d bits %d->%d\n", m, tree[m].Len, - bits)); - opt_len += - ((long) bits - - (long) tree[m].Len) * (long) tree[m].Freq; - tree[m].Len = (ush) bits; + Trace((stderr, "code %d bits %d->%d\n", m, tree[m].Len, bits)); + G2.opt_len += ((int32_t) bits - tree[m].Len) * tree[m].Freq; + tree[m].Len = bits; } n--; } } } + /* =========================================================================== * Generate the codes for a given tree and bit counts (which need not be * optimal). @@ -1928,23 +1156,23 @@ static void gen_bitlen(tree_desc *desc) * OUT assertion: the field code is set for all tree elements of non * zero code length. */ -static void gen_codes(ct_data *tree, int max_code) +static void gen_codes(ct_data * tree, int max_code) { ush next_code[MAX_BITS + 1]; /* next code value for each bit length */ - ush code = 0; /* running code value */ - int bits; /* bit index */ - int n; /* code index */ + ush code = 0; /* running code value */ + int bits; /* bit index */ + int n; /* code index */ /* The distribution counts are first used to generate the code values * without bit reversal. */ for (bits = 1; bits <= MAX_BITS; bits++) { - next_code[bits] = code = (code + bl_count[bits - 1]) << 1; + next_code[bits] = code = (code + G2.bl_count[bits - 1]) << 1; } /* Check that the bit counts in bl_count are consistent. The last code * must be all ones. */ - Assert(code + bl_count[MAX_BITS] - 1 == (1 << MAX_BITS) - 1, + Assert(code + G2.bl_count[MAX_BITS] - 1 == (1 << MAX_BITS) - 1, "inconsistent bit counts"); Tracev((stderr, "\ngen_codes: max_code %d ", max_code)); @@ -1956,13 +1184,14 @@ static void gen_codes(ct_data *tree, int max_code) /* Now reverse the bits */ tree[n].Code = bi_reverse(next_code[len]++, len); - Tracec(tree != static_ltree, + Tracec(tree != G2.static_ltree, (stderr, "\nn %3d %c l %2d c %4x (%x) ", n, (isgraph(n) ? n : ' '), len, tree[n].Code, next_code[len] - 1)); } } + /* =========================================================================== * Construct one Huffman tree and assigns the code bit strings and lengths. * Update the total bit length for the current block. @@ -1971,25 +1200,40 @@ static void gen_codes(ct_data *tree, int max_code) * and corresponding code. The length opt_len is updated; static_len is * also updated if stree is not null. The field max_code is set. */ -static void build_tree(tree_desc *desc) + +/* Remove the smallest element from the heap and recreate the heap with + * one less element. Updates heap and heap_len. */ + +#define SMALLEST 1 +/* Index within the heap array of least frequent node in the Huffman tree */ + +#define PQREMOVE(tree, top) \ +do { \ + top = G2.heap[SMALLEST]; \ + G2.heap[SMALLEST] = G2.heap[G2.heap_len--]; \ + pqdownheap(tree, SMALLEST); \ +} while (0) + +static void build_tree(tree_desc * desc) { ct_data *tree = desc->dyn_tree; ct_data *stree = desc->static_tree; int elems = desc->elems; - int n, m; /* iterate over heap elements */ - int max_code = -1; /* largest code with non zero frequency */ - int node = elems; /* next internal node of the tree */ + int n, m; /* iterate over heap elements */ + int max_code = -1; /* largest code with non zero frequency */ + int node = elems; /* next internal node of the tree */ /* Construct the initial heap, with least frequent element in * heap[SMALLEST]. The sons of heap[n] are heap[2*n] and heap[2*n+1]. * heap[0] is not used. */ - heap_len = 0, heap_max = HEAP_SIZE; + G2.heap_len = 0; + G2.heap_max = HEAP_SIZE; for (n = 0; n < elems; n++) { if (tree[n].Freq != 0) { - heap[++heap_len] = max_code = n; - depth[n] = 0; + G2.heap[++G2.heap_len] = max_code = n; + G2.depth[n] = 0; } else { tree[n].Len = 0; } @@ -2000,14 +1244,14 @@ static void build_tree(tree_desc *desc) * possible code. So to avoid special checks later on we force at least * two codes of non zero frequency. */ - while (heap_len < 2) { - int new = heap[++heap_len] = (max_code < 2 ? ++max_code : 0); + while (G2.heap_len < 2) { + int new = G2.heap[++G2.heap_len] = (max_code < 2 ? ++max_code : 0); tree[new].Freq = 1; - depth[new] = 0; - opt_len--; + G2.depth[new] = 0; + G2.opt_len--; if (stree) - static_len -= stree[new].Len; + G2.static_len -= stree[new].Len; /* new is 0 or 1 so it does not have extra bits */ } desc->max_code = max_code; @@ -2015,37 +1259,36 @@ static void build_tree(tree_desc *desc) /* The elements heap[heap_len/2+1 .. heap_len] are leaves of the tree, * establish sub-heaps of increasing lengths: */ - for (n = heap_len / 2; n >= 1; n--) + for (n = G2.heap_len / 2; n >= 1; n--) pqdownheap(tree, n); /* Construct the Huffman tree by repeatedly combining the least two * frequent nodes. */ do { - pqremove(tree, n); /* n = node of least frequency */ - m = heap[SMALLEST]; /* m = node of next least frequency */ + PQREMOVE(tree, n); /* n = node of least frequency */ + m = G2.heap[SMALLEST]; /* m = node of next least frequency */ - heap[--heap_max] = n; /* keep the nodes sorted by frequency */ - heap[--heap_max] = m; + G2.heap[--G2.heap_max] = n; /* keep the nodes sorted by frequency */ + G2.heap[--G2.heap_max] = m; /* Create a new node father of n and m */ tree[node].Freq = tree[n].Freq + tree[m].Freq; - depth[node] = (uch) (MAX(depth[n], depth[m]) + 1); + G2.depth[node] = MAX(G2.depth[n], G2.depth[m]) + 1; tree[n].Dad = tree[m].Dad = (ush) node; #ifdef DUMP_BL_TREE - if (tree == bl_tree) { - fprintf(stderr, "\nnode %d(%d), sons %d(%d) %d(%d)", - node, tree[node].Freq, n, tree[n].Freq, m, - tree[m].Freq); + if (tree == G2.bl_tree) { + bb_error_msg("\nnode %d(%d), sons %d(%d) %d(%d)", + node, tree[node].Freq, n, tree[n].Freq, m, tree[m].Freq); } #endif /* and insert the new node in the heap */ - heap[SMALLEST] = node++; + G2.heap[SMALLEST] = node++; pqdownheap(tree, SMALLEST); - } while (heap_len >= 2); + } while (G2.heap_len >= 2); - heap[--heap_max] = heap[SMALLEST]; + G2.heap[--G2.heap_max] = G2.heap[SMALLEST]; /* At this point, the fields freq and dad are set. We can now * generate the bit lengths. @@ -2056,67 +1299,75 @@ static void build_tree(tree_desc *desc) gen_codes((ct_data *) tree, max_code); } + /* =========================================================================== * Scan a literal or distance tree to determine the frequencies of the codes * in the bit length tree. Updates opt_len to take into account the repeat * counts. (The contribution of the bit length codes will be added later * during the construction of bl_tree.) */ -static void scan_tree(ct_data *tree, int max_code) +static void scan_tree(ct_data * tree, int max_code) { - int n; /* iterates over all tree elements */ - int prevlen = -1; /* last emitted length */ - int curlen; /* length of current code */ + int n; /* iterates over all tree elements */ + int prevlen = -1; /* last emitted length */ + int curlen; /* length of current code */ int nextlen = tree[0].Len; /* length of next code */ - int count = 0; /* repeat count of the current code */ - int max_count = 7; /* max repeat count */ - int min_count = 4; /* min repeat count */ + int count = 0; /* repeat count of the current code */ + int max_count = 7; /* max repeat count */ + int min_count = 4; /* min repeat count */ - if (nextlen == 0) - max_count = 138, min_count = 3; - tree[max_code + 1].Len = (ush) 0xffff; /* guard */ + if (nextlen == 0) { + max_count = 138; + min_count = 3; + } + tree[max_code + 1].Len = 0xffff; /* guard */ for (n = 0; n <= max_code; n++) { curlen = nextlen; nextlen = tree[n + 1].Len; - if (++count < max_count && curlen == nextlen) { + if (++count < max_count && curlen == nextlen) continue; - } else if (count < min_count) { - bl_tree[curlen].Freq += count; + + if (count < min_count) { + G2.bl_tree[curlen].Freq += count; } else if (curlen != 0) { if (curlen != prevlen) - bl_tree[curlen].Freq++; - bl_tree[REP_3_6].Freq++; + G2.bl_tree[curlen].Freq++; + G2.bl_tree[REP_3_6].Freq++; } else if (count <= 10) { - bl_tree[REPZ_3_10].Freq++; + G2.bl_tree[REPZ_3_10].Freq++; } else { - bl_tree[REPZ_11_138].Freq++; + G2.bl_tree[REPZ_11_138].Freq++; } count = 0; prevlen = curlen; + + max_count = 7; + min_count = 4; if (nextlen == 0) { - max_count = 138, min_count = 3; + max_count = 138; + min_count = 3; } else if (curlen == nextlen) { - max_count = 6, min_count = 3; - } else { - max_count = 7, min_count = 4; + max_count = 6; + min_count = 3; } } } + /* =========================================================================== * Send a literal or distance tree in compressed form, using the codes in * bl_tree. */ -static void send_tree(ct_data *tree, int max_code) +static void send_tree(ct_data * tree, int max_code) { - int n; /* iterates over all tree elements */ - int prevlen = -1; /* last emitted length */ - int curlen; /* length of current code */ + int n; /* iterates over all tree elements */ + int prevlen = -1; /* last emitted length */ + int curlen; /* length of current code */ int nextlen = tree[0].Len; /* length of next code */ - int count = 0; /* repeat count of the current code */ - int max_count = 7; /* max repeat count */ - int min_count = 4; /* min repeat count */ + int count = 0; /* repeat count of the current code */ + int max_count = 7; /* max repeat count */ + int min_count = 4; /* min repeat count */ /* tree[max_code+1].Len = -1; *//* guard already set */ if (nextlen == 0) @@ -2129,52 +1380,53 @@ static void send_tree(ct_data *tree, int max_code) continue; } else if (count < min_count) { do { - send_code(curlen, bl_tree); - } while (--count != 0); - + SEND_CODE(curlen, G2.bl_tree); + } while (--count); } else if (curlen != 0) { if (curlen != prevlen) { - send_code(curlen, bl_tree); + SEND_CODE(curlen, G2.bl_tree); count--; } Assert(count >= 3 && count <= 6, " 3_6?"); - send_code(REP_3_6, bl_tree); + SEND_CODE(REP_3_6, G2.bl_tree); send_bits(count - 3, 2); - } else if (count <= 10) { - send_code(REPZ_3_10, bl_tree); + SEND_CODE(REPZ_3_10, G2.bl_tree); send_bits(count - 3, 3); - } else { - send_code(REPZ_11_138, bl_tree); + SEND_CODE(REPZ_11_138, G2.bl_tree); send_bits(count - 11, 7); } count = 0; prevlen = curlen; if (nextlen == 0) { - max_count = 138, min_count = 3; + max_count = 138; + min_count = 3; } else if (curlen == nextlen) { - max_count = 6, min_count = 3; + max_count = 6; + min_count = 3; } else { - max_count = 7, min_count = 4; + max_count = 7; + min_count = 4; } } } + /* =========================================================================== * Construct the Huffman tree for the bit lengths and return the index in * bl_order of the last bit length code to send. */ -static const int build_bl_tree() +static int build_bl_tree(void) { - int max_blindex; /* index of last bit length code of non zero freq */ + int max_blindex; /* index of last bit length code of non zero freq */ /* Determine the bit length frequencies for literal and distance trees */ - scan_tree((ct_data *) dyn_ltree, l_desc.max_code); - scan_tree((ct_data *) dyn_dtree, d_desc.max_code); + scan_tree(G2.dyn_ltree, G2.l_desc.max_code); + scan_tree(G2.dyn_dtree, G2.d_desc.max_code); /* Build the bit length tree: */ - build_tree((tree_desc *) (&bl_desc)); + build_tree(&G2.bl_desc); /* opt_len now includes the length of the tree representations, except * the lengths of the bit lengths codes and the 5+5+4 bits for the counts. */ @@ -2184,18 +1436,17 @@ static const int build_bl_tree() * 3 but the actual value used is 4.) */ for (max_blindex = BL_CODES - 1; max_blindex >= 3; max_blindex--) { - if (bl_tree[bl_order[max_blindex]].Len != 0) + if (G2.bl_tree[bl_order[max_blindex]].Len != 0) break; } /* Update opt_len to include the bit length tree and counts */ - opt_len += 3 * (max_blindex + 1) + 5 + 5 + 4; - Tracev( - (stderr, "\ndyn trees: dyn %ld, stat %ld", opt_len, - static_len)); + G2.opt_len += 3 * (max_blindex + 1) + 5 + 5 + 4; + Tracev((stderr, "\ndyn trees: dyn %ld, stat %ld", G2.opt_len, G2.static_len)); return max_blindex; } + /* =========================================================================== * Send the header for a block using dynamic Huffman trees: the counts, the * lengths of the bit length codes, the literal tree and the distance tree. @@ -2203,10 +1454,9 @@ static const int build_bl_tree() */ static void send_all_trees(int lcodes, int dcodes, int blcodes) { - int rank; /* index in bl_order */ + int rank; /* index in bl_order */ - Assert(lcodes >= 257 && dcodes >= 1 - && blcodes >= 4, "not enough codes"); + Assert(lcodes >= 257 && dcodes >= 1 && blcodes >= 4, "not enough codes"); Assert(lcodes <= L_CODES && dcodes <= D_CODES && blcodes <= BL_CODES, "too many codes"); Tracev((stderr, "\nbl counts: ")); @@ -2215,17 +1465,124 @@ static void send_all_trees(int lcodes, int dcodes, int blcodes) send_bits(blcodes - 4, 4); /* not -3 as stated in appnote.txt */ for (rank = 0; rank < blcodes; rank++) { Tracev((stderr, "\nbl code %2d ", bl_order[rank])); - send_bits(bl_tree[bl_order[rank]].Len, 3); + send_bits(G2.bl_tree[bl_order[rank]].Len, 3); } - Tracev((stderr, "\nbl tree: sent %ld", bits_sent)); + Tracev((stderr, "\nbl tree: sent %ld", G1.bits_sent)); - send_tree((ct_data *) dyn_ltree, lcodes - 1); /* send the literal tree */ - Tracev((stderr, "\nlit tree: sent %ld", bits_sent)); + send_tree((ct_data *) G2.dyn_ltree, lcodes - 1); /* send the literal tree */ + Tracev((stderr, "\nlit tree: sent %ld", G1.bits_sent)); - send_tree((ct_data *) dyn_dtree, dcodes - 1); /* send the distance tree */ - Tracev((stderr, "\ndist tree: sent %ld", bits_sent)); + send_tree((ct_data *) G2.dyn_dtree, dcodes - 1); /* send the distance tree */ + Tracev((stderr, "\ndist tree: sent %ld", G1.bits_sent)); } + +/* =========================================================================== + * Save the match info and tally the frequency counts. Return true if + * the current block must be flushed. + */ +static int ct_tally(int dist, int lc) +{ + G1.l_buf[G2.last_lit++] = lc; + if (dist == 0) { + /* lc is the unmatched char */ + G2.dyn_ltree[lc].Freq++; + } else { + /* Here, lc is the match length - MIN_MATCH */ + dist--; /* dist = match distance - 1 */ + Assert((ush) dist < (ush) MAX_DIST + && (ush) lc <= (ush) (MAX_MATCH - MIN_MATCH) + && (ush) D_CODE(dist) < (ush) D_CODES, "ct_tally: bad match" + ); + + G2.dyn_ltree[G2.length_code[lc] + LITERALS + 1].Freq++; + G2.dyn_dtree[D_CODE(dist)].Freq++; + + G1.d_buf[G2.last_dist++] = dist; + G2.flags |= G2.flag_bit; + } + G2.flag_bit <<= 1; + + /* Output the flags if they fill a byte: */ + if ((G2.last_lit & 7) == 0) { + G2.flag_buf[G2.last_flags++] = G2.flags; + G2.flags = 0; + G2.flag_bit = 1; + } + /* Try to guess if it is profitable to stop the current block here */ + if ((G2.last_lit & 0xfff) == 0) { + /* Compute an upper bound for the compressed length */ + ulg out_length = G2.last_lit * 8L; + ulg in_length = (ulg) G1.strstart - G1.block_start; + int dcode; + + for (dcode = 0; dcode < D_CODES; dcode++) { + out_length += G2.dyn_dtree[dcode].Freq * (5L + extra_dbits[dcode]); + } + out_length >>= 3; + Trace((stderr, + "\nlast_lit %u, last_dist %u, in %ld, out ~%ld(%ld%%) ", + G2.last_lit, G2.last_dist, in_length, out_length, + 100L - out_length * 100L / in_length)); + if (G2.last_dist < G2.last_lit / 2 && out_length < in_length / 2) + return 1; + } + return (G2.last_lit == LIT_BUFSIZE - 1 || G2.last_dist == DIST_BUFSIZE); + /* We avoid equality with LIT_BUFSIZE because of wraparound at 64K + * on 16 bit machines and because stored blocks are restricted to + * 64K-1 bytes. + */ +} + +/* =========================================================================== + * Send the block data compressed using the given Huffman trees + */ +static void compress_block(ct_data * ltree, ct_data * dtree) +{ + unsigned dist; /* distance of matched string */ + int lc; /* match length or unmatched char (if dist == 0) */ + unsigned lx = 0; /* running index in l_buf */ + unsigned dx = 0; /* running index in d_buf */ + unsigned fx = 0; /* running index in flag_buf */ + uch flag = 0; /* current flags */ + unsigned code; /* the code to send */ + int extra; /* number of extra bits to send */ + + if (G2.last_lit != 0) do { + if ((lx & 7) == 0) + flag = G2.flag_buf[fx++]; + lc = G1.l_buf[lx++]; + if ((flag & 1) == 0) { + SEND_CODE(lc, ltree); /* send a literal byte */ + Tracecv(isgraph(lc), (stderr, " '%c' ", lc)); + } else { + /* Here, lc is the match length - MIN_MATCH */ + code = G2.length_code[lc]; + SEND_CODE(code + LITERALS + 1, ltree); /* send the length code */ + extra = extra_lbits[code]; + if (extra != 0) { + lc -= G2.base_length[code]; + send_bits(lc, extra); /* send the extra length bits */ + } + dist = G1.d_buf[dx++]; + /* Here, dist is the match distance - 1 */ + code = D_CODE(dist); + Assert(code < D_CODES, "bad d_code"); + + SEND_CODE(code, dtree); /* send the distance code */ + extra = extra_dbits[code]; + if (extra != 0) { + dist -= G2.base_dist[code]; + send_bits(dist, extra); /* send the extra distance bits */ + } + } /* literal or match pair ? */ + flag >>= 1; + } while (lx < G2.last_lit); + + SEND_CODE(END_BLOCK, ltree); +} + + /* =========================================================================== * Determine the best encoding for the current block: dynamic trees, static * trees or store, and output the encoded block to the zip file. This function @@ -2233,23 +1590,17 @@ static void send_all_trees(int lcodes, int dcodes, int blcodes) */ static ulg flush_block(char *buf, ulg stored_len, int eof) { - ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */ - int max_blindex; /* index of last bit length code of non zero freq */ - - flag_buf[last_flags] = flags; /* Save the flags for the last 8 items */ + ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */ + int max_blindex; /* index of last bit length code of non zero freq */ - /* Check if the file is ascii or binary */ - if (*file_type == (ush) UNKNOWN) - set_file_type(); + G2.flag_buf[G2.last_flags] = G2.flags; /* Save the flags for the last 8 items */ /* Construct the literal and distance trees */ - build_tree((tree_desc *) (&l_desc)); - Tracev((stderr, "\nlit data: dyn %ld, stat %ld", opt_len, static_len)); + build_tree(&G2.l_desc); + Tracev((stderr, "\nlit data: dyn %ld, stat %ld", G2.opt_len, G2.static_len)); - build_tree((tree_desc *) (&d_desc)); - Tracev( - (stderr, "\ndist data: dyn %ld, stat %ld", opt_len, - static_len)); + build_tree(&G2.d_desc); + Tracev((stderr, "\ndist data: dyn %ld, stat %ld", G2.opt_len, G2.static_len)); /* At this point, opt_len and static_len are the total bit lengths of * the compressed block data, excluding the tree representations. */ @@ -2260,14 +1611,13 @@ static ulg flush_block(char *buf, ulg stored_len, int eof) max_blindex = build_bl_tree(); /* Determine the best encoding. Compute first the block length in bytes */ - opt_lenb = (opt_len + 3 + 7) >> 3; - static_lenb = (static_len + 3 + 7) >> 3; + opt_lenb = (G2.opt_len + 3 + 7) >> 3; + static_lenb = (G2.static_len + 3 + 7) >> 3; - Trace( - (stderr, + Trace((stderr, "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u dist %u ", - opt_lenb, opt_len, static_lenb, static_len, stored_len, - last_lit, last_dist)); + opt_lenb, G2.opt_len, static_lenb, G2.static_len, stored_len, + G2.last_lit, G2.last_dist)); if (static_lenb <= opt_lenb) opt_lenb = static_lenb; @@ -2276,17 +1626,15 @@ static ulg flush_block(char *buf, ulg stored_len, int eof) * and if the zip file can be seeked (to rewrite the local header), * the whole file is transformed into a stored file: */ - if (stored_len <= opt_lenb && eof && compressed_len == 0L - && seekable()) { + if (stored_len <= opt_lenb && eof && G2.compressed_len == 0L && seekable()) { /* Since LIT_BUFSIZE <= 2*WSIZE, the input data must be there: */ - if (buf == (char *) 0) - error_msg("block vanished"); + if (buf == NULL) + bb_error_msg("block vanished"); copy_block(buf, (unsigned) stored_len, 0); /* without header */ - compressed_len = stored_len << 3; - *file_method = STORED; + G2.compressed_len = stored_len << 3; - } else if (stored_len + 4 <= opt_lenb && buf != (char *) 0) { + } else if (stored_len + 4 <= opt_lenb && buf != NULL) { /* 4: two words for the lengths */ /* The test buf != NULL is only necessary if LIT_BUFSIZE > WSIZE. * Otherwise we can't have processed more than WSIZE input bytes since @@ -2295,256 +1643,460 @@ static ulg flush_block(char *buf, ulg stored_len, int eof) * transform a block into a stored block. */ send_bits((STORED_BLOCK << 1) + eof, 3); /* send block type */ - compressed_len = (compressed_len + 3 + 7) & ~7L; - compressed_len += (stored_len + 4) << 3; + G2.compressed_len = (G2.compressed_len + 3 + 7) & ~7L; + G2.compressed_len += (stored_len + 4) << 3; copy_block(buf, (unsigned) stored_len, 1); /* with header */ } else if (static_lenb == opt_lenb) { send_bits((STATIC_TREES << 1) + eof, 3); - compress_block((ct_data *) static_ltree, - (ct_data *) static_dtree); - compressed_len += 3 + static_len; + compress_block((ct_data *) G2.static_ltree, (ct_data *) G2.static_dtree); + G2.compressed_len += 3 + G2.static_len; } else { send_bits((DYN_TREES << 1) + eof, 3); - send_all_trees(l_desc.max_code + 1, d_desc.max_code + 1, + send_all_trees(G2.l_desc.max_code + 1, G2.d_desc.max_code + 1, max_blindex + 1); - compress_block((ct_data *) dyn_ltree, - (ct_data *) dyn_dtree); - compressed_len += 3 + opt_len; + compress_block((ct_data *) G2.dyn_ltree, (ct_data *) G2.dyn_dtree); + G2.compressed_len += 3 + G2.opt_len; } - Assert(compressed_len == bits_sent, "bad compressed size"); + Assert(G2.compressed_len == G1.bits_sent, "bad compressed size"); init_block(); if (eof) { bi_windup(); - compressed_len += 7; /* align on byte boundary */ + G2.compressed_len += 7; /* align on byte boundary */ } - Tracev((stderr, "\ncomprlen %lu(%lu) ", compressed_len >> 3, - compressed_len - 7 * eof)); + Tracev((stderr, "\ncomprlen %lu(%lu) ", G2.compressed_len >> 3, + G2.compressed_len - 7 * eof)); - return compressed_len >> 3; + return G2.compressed_len >> 3; } + /* =========================================================================== - * Save the match info and tally the frequency counts. Return true if - * the current block must be flushed. + * Update a hash value with the given input byte + * IN assertion: all calls to to UPDATE_HASH are made with consecutive + * input characters, so that a running hash key can be computed from the + * previous key instead of complete recalculation each time. */ -static int ct_tally(int dist, int lc) +#define UPDATE_HASH(h, c) (h = (((h)<= 0L \ + ? (char*)&G1.window[(unsigned)G1.block_start] \ + : (char*)NULL, \ + (ulg)G1.strstart - G1.block_start, \ + (eof) \ + ) + +/* Insert string s in the dictionary and set match_head to the previous head + * of the hash chain (the most recent string with same hash key). Return + * the previous length of the hash chain. + * IN assertion: all calls to to INSERT_STRING are made with consecutive + * input characters and the first MIN_MATCH bytes of s are valid + * (except for the last MIN_MATCH-1 bytes of the input file). */ +#define INSERT_STRING(s, match_head) \ +do { \ + UPDATE_HASH(G1.ins_h, G1.window[(s) + MIN_MATCH-1]); \ + G1.prev[(s) & WMASK] = match_head = head[G1.ins_h]; \ + head[G1.ins_h] = (s); \ +} while (0) + +static ulg deflate(void) { - l_buf[last_lit++] = (uch) lc; - if (dist == 0) { - /* lc is the unmatched char */ - dyn_ltree[lc].Freq++; - } else { - /* Here, lc is the match length - MIN_MATCH */ - dist--; /* dist = match distance - 1 */ - Assert((ush) dist < (ush) MAX_DIST && - (ush) lc <= (ush) (MAX_MATCH - MIN_MATCH) && - (ush) d_code(dist) < (ush) D_CODES, "ct_tally: bad match"); + IPos hash_head; /* head of hash chain */ + IPos prev_match; /* previous match */ + int flush; /* set if current block must be flushed */ + int match_available = 0; /* set if previous match exists */ + unsigned match_length = MIN_MATCH - 1; /* length of best match */ - dyn_ltree[length_code[lc] + LITERALS + 1].Freq++; - dyn_dtree[d_code(dist)].Freq++; + /* Process the input block. */ + while (G1.lookahead != 0) { + /* Insert the string window[strstart .. strstart+2] in the + * dictionary, and set hash_head to the head of the hash chain: + */ + INSERT_STRING(G1.strstart, hash_head); - d_buf[last_dist++] = (ush) dist; - flags |= flag_bit; - } - flag_bit <<= 1; + /* Find the longest match, discarding those <= prev_length. + */ + G1.prev_length = match_length; + prev_match = G1.match_start; + match_length = MIN_MATCH - 1; - /* Output the flags if they fill a byte: */ - if ((last_lit & 7) == 0) { - flag_buf[last_flags++] = flags; - flags = 0, flag_bit = 1; - } - /* Try to guess if it is profitable to stop the current block here */ - if ((last_lit & 0xfff) == 0) { - /* Compute an upper bound for the compressed length */ - ulg out_length = (ulg) last_lit * 8L; - ulg in_length = (ulg) strstart - block_start; - int dcode; + if (hash_head != 0 && G1.prev_length < max_lazy_match + && G1.strstart - hash_head <= MAX_DIST + ) { + /* To simplify the code, we prevent matches with the string + * of window index 0 (in particular we have to avoid a match + * of the string with itself at the start of the input file). + */ + match_length = longest_match(hash_head); + /* longest_match() sets match_start */ + if (match_length > G1.lookahead) + match_length = G1.lookahead; - for (dcode = 0; dcode < D_CODES; dcode++) { - out_length += - (ulg) dyn_dtree[dcode].Freq * (5L + extra_dbits[dcode]); + /* Ignore a length 3 match if it is too distant: */ + if (match_length == MIN_MATCH && G1.strstart - G1.match_start > TOO_FAR) { + /* If prev_match is also MIN_MATCH, G1.match_start is garbage + * but we will ignore the current match anyway. + */ + match_length--; + } } - out_length >>= 3; - Trace( - (stderr, - "\nlast_lit %u, last_dist %u, in %ld, out ~%ld(%ld%%) ", - last_lit, last_dist, in_length, out_length, - 100L - out_length * 100L / in_length)); - if (last_dist < last_lit / 2 && out_length < in_length / 2) - return 1; + /* If there was a match at the previous step and the current + * match is not better, output the previous match: + */ + if (G1.prev_length >= MIN_MATCH && match_length <= G1.prev_length) { + check_match(G1.strstart - 1, prev_match, G1.prev_length); + flush = ct_tally(G1.strstart - 1 - prev_match, G1.prev_length - MIN_MATCH); + + /* Insert in hash table all strings up to the end of the match. + * strstart-1 and strstart are already inserted. + */ + G1.lookahead -= G1.prev_length - 1; + G1.prev_length -= 2; + do { + G1.strstart++; + INSERT_STRING(G1.strstart, hash_head); + /* strstart never exceeds WSIZE-MAX_MATCH, so there are + * always MIN_MATCH bytes ahead. If lookahead < MIN_MATCH + * these bytes are garbage, but it does not matter since the + * next lookahead bytes will always be emitted as literals. + */ + } while (--G1.prev_length != 0); + match_available = 0; + match_length = MIN_MATCH - 1; + G1.strstart++; + if (flush) { + FLUSH_BLOCK(0); + G1.block_start = G1.strstart; + } + } else if (match_available) { + /* If there was no match at the previous position, output a + * single literal. If there was a match but the current match + * is longer, truncate the previous match to a single literal. + */ + Tracevv((stderr, "%c", G1.window[G1.strstart - 1])); + if (ct_tally(0, G1.window[G1.strstart - 1])) { + FLUSH_BLOCK(0); + G1.block_start = G1.strstart; + } + G1.strstart++; + G1.lookahead--; + } else { + /* There is no previous match to compare with, wait for + * the next step to decide. + */ + match_available = 1; + G1.strstart++; + G1.lookahead--; + } + Assert(G1.strstart <= G1.isize && lookahead <= G1.isize, "a bit too far"); + + /* Make sure that we always have enough lookahead, except + * at the end of the input file. We need MAX_MATCH bytes + * for the next match, plus MIN_MATCH bytes to insert the + * string following the next match. + */ + while (G1.lookahead < MIN_LOOKAHEAD && !G1.eofile) + fill_window(); } - return (last_lit == LIT_BUFSIZE - 1 || last_dist == DIST_BUFSIZE); - /* We avoid equality with LIT_BUFSIZE because of wraparound at 64K - * on 16 bit machines and because stored blocks are restricted to - * 64K-1 bytes. - */ + if (match_available) + ct_tally(0, G1.window[G1.strstart - 1]); + + return FLUSH_BLOCK(1); /* eof */ } + /* =========================================================================== - * Send the block data compressed using the given Huffman trees + * Initialize the bit string routines. */ -static void compress_block(ct_data *ltree, ct_data *dtree) +static void bi_init(void) { - unsigned dist; /* distance of matched string */ - int lc; /* match length or unmatched char (if dist == 0) */ - unsigned lx = 0; /* running index in l_buf */ - unsigned dx = 0; /* running index in d_buf */ - unsigned fx = 0; /* running index in flag_buf */ - uch flag = 0; /* current flags */ - unsigned code; /* the code to send */ - int extra; /* number of extra bits to send */ - - if (last_lit != 0) - do { - if ((lx & 7) == 0) - flag = flag_buf[fx++]; - lc = l_buf[lx++]; - if ((flag & 1) == 0) { - send_code(lc, ltree); /* send a literal byte */ - Tracecv(isgraph(lc), (stderr, " '%c' ", lc)); - } else { - /* Here, lc is the match length - MIN_MATCH */ - code = length_code[lc]; - send_code(code + LITERALS + 1, ltree); /* send the length code */ - extra = extra_lbits[code]; - if (extra != 0) { - lc -= base_length[code]; - send_bits(lc, extra); /* send the extra length bits */ - } - dist = d_buf[dx++]; - /* Here, dist is the match distance - 1 */ - code = d_code(dist); - Assert(code < D_CODES, "bad d_code"); - - send_code(code, dtree); /* send the distance code */ - extra = extra_dbits[code]; - if (extra != 0) { - dist -= base_dist[code]; - send_bits(dist, extra); /* send the extra distance bits */ - } - } /* literal or match pair ? */ - flag >>= 1; - } while (lx < last_lit); - - send_code(END_BLOCK, ltree); + G1.bi_buf = 0; + G1.bi_valid = 0; +#ifdef DEBUG + G1.bits_sent = 0L; +#endif } + /* =========================================================================== - * Set the file type to ASCII or BINARY, using a crude approximation: - * binary if more than 20% of the bytes are <= 6 or >= 128, ascii otherwise. - * IN assertion: the fields freq of dyn_ltree are set and the total of all - * frequencies does not exceed 64K (to fit in an int on 16 bit machines). + * Initialize the "longest match" routines for a new file */ -static void set_file_type() +static void lm_init(ush * flagsp) { - int n = 0; - unsigned ascii_freq = 0; - unsigned bin_freq = 0; - - while (n < 7) - bin_freq += dyn_ltree[n++].Freq; - while (n < 128) - ascii_freq += dyn_ltree[n++].Freq; - while (n < LITERALS) - bin_freq += dyn_ltree[n++].Freq; - *file_type = bin_freq > (ascii_freq >> 2) ? BINARY : ASCII; - if (*file_type == BINARY && translate_eol) { - error_msg("-l used on binary file"); + unsigned j; + + /* Initialize the hash table. */ + memset(head, 0, HASH_SIZE * sizeof(*head)); + /* prev will be initialized on the fly */ + + /* speed options for the general purpose bit flag */ + *flagsp |= 2; /* FAST 4, SLOW 2 */ + /* ??? reduce max_chain_length for binary files */ + + G1.strstart = 0; + G1.block_start = 0L; + + G1.lookahead = file_read(G1.window, + sizeof(int) <= 2 ? (unsigned) WSIZE : 2 * WSIZE); + + if (G1.lookahead == 0 || G1.lookahead == (unsigned) -1) { + G1.eofile = 1; + G1.lookahead = 0; + return; } + G1.eofile = 0; + /* Make sure that we always have enough lookahead. This is important + * if input comes from a device such as a tty. + */ + while (G1.lookahead < MIN_LOOKAHEAD && !G1.eofile) + fill_window(); + + G1.ins_h = 0; + for (j = 0; j < MIN_MATCH - 1; j++) + UPDATE_HASH(G1.ins_h, G1.window[j]); + /* If lookahead < MIN_MATCH, ins_h is garbage, but this is + * not important since only literal bytes will be emitted. + */ } -/* zip.c -- compress files to the gzip or pkzip format - * Copyright (C) 1992-1993 Jean-loup Gailly - * This is free software; you can redistribute it and/or modify it under the - * terms of the GNU General Public License, see the file COPYING. + +/* =========================================================================== + * Allocate the match buffer, initialize the various tables and save the + * location of the internal file attribute (ascii/binary) and method + * (DEFLATE/STORE). + * One callsite in zip() */ +static void ct_init(void) +{ + int n; /* iterates over tree elements */ + int length; /* length value */ + int code; /* code value */ + int dist; /* distance index */ + + G2.compressed_len = 0L; + +#ifdef NOT_NEEDED + if (G2.static_dtree[0].Len != 0) + return; /* ct_init already called */ +#endif + + /* Initialize the mapping length (0..255) -> length code (0..28) */ + length = 0; + for (code = 0; code < LENGTH_CODES - 1; code++) { + G2.base_length[code] = length; + for (n = 0; n < (1 << extra_lbits[code]); n++) { + G2.length_code[length++] = code; + } + } + Assert(length == 256, "ct_init: length != 256"); + /* Note that the length 255 (match length 258) can be represented + * in two different ways: code 284 + 5 bits or code 285, so we + * overwrite length_code[255] to use the best encoding: + */ + G2.length_code[length - 1] = code; + + /* Initialize the mapping dist (0..32K) -> dist code (0..29) */ + dist = 0; + for (code = 0; code < 16; code++) { + G2.base_dist[code] = dist; + for (n = 0; n < (1 << extra_dbits[code]); n++) { + G2.dist_code[dist++] = code; + } + } + Assert(dist == 256, "ct_init: dist != 256"); + dist >>= 7; /* from now on, all distances are divided by 128 */ + for (; code < D_CODES; code++) { + G2.base_dist[code] = dist << 7; + for (n = 0; n < (1 << (extra_dbits[code] - 7)); n++) { + G2.dist_code[256 + dist++] = code; + } + } + Assert(dist == 256, "ct_init: 256+dist != 512"); + + /* Construct the codes of the static literal tree */ + /* already zeroed - it's in bss + for (n = 0; n <= MAX_BITS; n++) + G2.bl_count[n] = 0; */ + + n = 0; + while (n <= 143) { + G2.static_ltree[n++].Len = 8; + G2.bl_count[8]++; + } + while (n <= 255) { + G2.static_ltree[n++].Len = 9; + G2.bl_count[9]++; + } + while (n <= 279) { + G2.static_ltree[n++].Len = 7; + G2.bl_count[7]++; + } + while (n <= 287) { + G2.static_ltree[n++].Len = 8; + G2.bl_count[8]++; + } + /* Codes 286 and 287 do not exist, but we must include them in the + * tree construction to get a canonical Huffman tree (longest code + * all ones) + */ + gen_codes((ct_data *) G2.static_ltree, L_CODES + 1); + + /* The static distance tree is trivial: */ + for (n = 0; n < D_CODES; n++) { + G2.static_dtree[n].Len = 5; + G2.static_dtree[n].Code = bi_reverse(n, 5); + } + /* Initialize the first block of the first file: */ + init_block(); +} -static ulg crc; /* crc on uncompressed file data */ -static long header_bytes; /* number of bytes in gzip header */ /* =========================================================================== * Deflate in to out. * IN assertions: the input and output buffers are cleared. - * The variables time_stamp and save_orig_name are initialized. */ -static int zip(int in, int out) + +static void zip(ulg time_stamp) { - uch my_flags = 0; /* general purpose bit flags */ - ush attr = 0; /* ascii/binary flag */ - ush deflate_flags = 0; /* pkzip -es, -en or -ex equivalent */ + ush deflate_flags = 0; /* pkzip -es, -en or -ex equivalent */ - ifd = in; - ofd = out; - outcnt = 0; + G1.outcnt = 0; /* Write the header to the gzip file. See algorithm.doc for the format */ - - - method = DEFLATED; - put_byte(GZIP_MAGIC[0]); /* magic header */ - put_byte(GZIP_MAGIC[1]); - put_byte(DEFLATED); /* compression method */ - - put_byte(my_flags); /* general flags */ - put_long(time_stamp); + /* magic header for gzip files: 1F 8B */ + /* compression method: 8 (DEFLATED) */ + /* general flags: 0 */ + put_32bit(0x00088b1f); + put_32bit(time_stamp); /* Write deflated file to zip file */ - crc = updcrc(0, 0); + G1.crc = ~0; - bi_init(out); - ct_init(&attr, &method); + bi_init(); + ct_init(); lm_init(&deflate_flags); - put_byte((uch) deflate_flags); /* extra flags */ - put_byte(OS_CODE); /* OS identifier */ - - header_bytes = (long) outcnt; + put_8bit(deflate_flags); /* extra flags */ + put_8bit(3); /* OS identifier = 3 (Unix) */ - (void) deflate(); + deflate(); /* Write the crc and uncompressed size */ - put_long(crc); - put_long(isize); - header_bytes += 2 * sizeof(long); + put_32bit(~G1.crc); + put_32bit(G1.isize); flush_outbuf(); - return OK; } -/* =========================================================================== - * Read a new buffer from the current input file, perform end-of-line - * translation, and update the crc and input file size. - * IN assertion: size >= 2 (for end-of-line translation) - */ -static int file_read(char *buf, unsigned size) +/* ======================================================================== */ +static +char* make_new_name_gzip(char *filename) { - unsigned len; - - Assert(insize == 0, "inbuf not empty"); + return xasprintf("%s.gz", filename); +} - len = read(ifd, buf, size); - if (len == (unsigned) (-1) || len == 0) - return (int) len; +static +USE_DESKTOP(long long) int pack_gzip(unpack_info_t *info UNUSED_PARAM) +{ + struct stat s; - crc = updcrc((uch *) buf, len); - isize += (ulg) len; - return (int) len; + clear_bufs(); + s.st_ctime = 0; + fstat(STDIN_FILENO, &s); + zip(s.st_ctime); + return 0; } -/* =========================================================================== - * Write the output buffer outbuf[0..outcnt-1] and update bytes_out. - * (used for the compressed data only) - */ -static void flush_outbuf() +/* + * Linux kernel build uses gzip -d -n. We accept and ignore it. + * Man page says: + * -n --no-name + * gzip: do not save the original file name and time stamp. + * (The original name is always saved if the name had to be truncated.) + * gunzip: do not restore the original file name/time even if present + * (remove only the gzip suffix from the compressed file name). + * This option is the default when decompressing. + * -N --name + * gzip: always save the original file name and time stamp (this is the default) + * gunzip: restore the original file name and time stamp if present. + */ + +int gzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; +#if ENABLE_GUNZIP +int gzip_main(int argc, char **argv) +#else +int gzip_main(int argc UNUSED_PARAM, char **argv) +#endif { - if (outcnt == 0) - return; + unsigned opt; + + /* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */ + opt = getopt32(argv, "cfv" USE_GUNZIP("dt") "q123456789n"); +#if ENABLE_GUNZIP /* gunzip_main may not be visible... */ + if (opt & 0x18) // -d and/or -t + return gunzip_main(argc, argv); +#endif + option_mask32 &= 0x7; /* ignore -q, -0..9 */ + //if (opt & 0x1) // -c + //if (opt & 0x2) // -f + //if (opt & 0x4) // -v + argv += optind; + + SET_PTR_TO_GLOBALS(xzalloc(sizeof(struct globals) + sizeof(struct globals2)) + + sizeof(struct globals)); + barrier(); + G2.l_desc.dyn_tree = G2.dyn_ltree; + G2.l_desc.static_tree = G2.static_ltree; + G2.l_desc.extra_bits = extra_lbits; + G2.l_desc.extra_base = LITERALS + 1; + G2.l_desc.elems = L_CODES; + G2.l_desc.max_length = MAX_BITS; + //G2.l_desc.max_code = 0; + + G2.d_desc.dyn_tree = G2.dyn_dtree; + G2.d_desc.static_tree = G2.static_dtree; + G2.d_desc.extra_bits = extra_dbits; + //G2.d_desc.extra_base = 0; + G2.d_desc.elems = D_CODES; + G2.d_desc.max_length = MAX_BITS; + //G2.d_desc.max_code = 0; + + G2.bl_desc.dyn_tree = G2.bl_tree; + //G2.bl_desc.static_tree = NULL; + G2.bl_desc.extra_bits = extra_blbits, + //G2.bl_desc.extra_base = 0; + G2.bl_desc.elems = BL_CODES; + G2.bl_desc.max_length = MAX_BL_BITS; + //G2.bl_desc.max_code = 0; + + /* Allocate all global buffers (for DYN_ALLOC option) */ + ALLOC(uch, G1.l_buf, INBUFSIZ); + ALLOC(uch, G1.outbuf, OUTBUFSIZ); + ALLOC(ush, G1.d_buf, DIST_BUFSIZE); + ALLOC(uch, G1.window, 2L * WSIZE); + ALLOC(ush, G1.prev, 1L << BITS); + + /* Initialise the CRC32 table */ + G1.crc_32_tab = crc32_filltable(NULL, 0); - write_buf(ofd, (char *) outbuf, outcnt); - outcnt = 0; + return bbunpack(argv, make_new_name_gzip, pack_gzip); }