Source Code 1994 March

home *** CD-ROM | disk | FTP | other *** search

/ Source Code 1994 March / Source_Code_CD-ROM_Walnut_Creek_March_1994.iso / compsrcs / misc / volume34 / freeze / part02 < prev next >

Wrap

Text File | 1993-01-18 | 54.6 KB | 2,204 lines

Newsgroups: comp.sources.misc From: leo@ipmce.su (Leonid A. Broukhis) Subject: v34i126: freeze - Freeze/melt compression program vers. 2.4, Part02/03 Message-ID: <1993Jan19.043553.29582@sparky.imd.sterling.com> X-Md4-Signature: 2a34590838e1849d96027d7eafecafe3 Date: Tue, 19 Jan 1993 04:35:53 GMT Approved: kent@sparky.imd.sterling.com Submitted-by: leo@ipmce.su (Leonid A. Broukhis) Posting-number: Volume 34, Issue 126 Archive-name: freeze/part02 Environment: ISC, Xenix, SunOS, MS-DOS Supersedes: freeze: Volume 25, Issue 12-13 #! /bin/sh # This is a shell archive. Remove anything before this line, then feed it # into a shell via "sh file" or similar. To overwrite existing files, # type "sh file -c". # Contents: Makefile.in config.h.in config.msc config.tur default.c # freeze.1 freeze.h huf.c lz.c lz.h statist.1 statist.c # Wrapped by kent@sparky on Mon Jan 18 22:27:48 1993 PATH=/bin:/usr/bin:/usr/ucb:/usr/local/bin:/usr/lbin ; export PATH echo If this archive is complete, you will see the following message: echo ' "shar: End of archive 2 (of 3)."' if test -f 'Makefile.in' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'Makefile.in'\" else echo shar: Extracting \"'Makefile.in'\" $3683 characters$ sed "s/^X//" >'Makefile.in' <<'END_OF_FILE' XSHELL = /bin/sh X XCC = @CC@ XCFLAGS = # -O2 # for gcc 2.2.2 X XINSTALL = @INSTALL@ XINSTALL_PROGRAM = @INSTALL_PROGRAM@ -s XINSTALL_DATA = @INSTALL_DATA@ X XLIBS = @LIBS@ XOBJ = o XEXE = X Xdefault: prog X X# Added the prefix macro, so that it was easier to change installation place. Xprefix = /usr/local XDEST = $(prefix)/bin XMANDEST = $(prefix)/man/man1 XSEC = 1 X XHDRS = bitio.h\ X compat.h\ X freeze.h\ X huf.h\ X lz.h\ X patchlevel.h X X# define DEFFILE as a filename with freeze's default Huffman values X# e.g. -DDEFFILE=\"/etc/default/freeze\" X# !!!! NOTE !!!! default is now $(prefix)/lib/freeze.cnf !!!!! X XOPTIONS = -DDEFFILE=\"$(prefix)/lib/freeze.cnf\" X XLINTFLAGS = -DCOMPAT -DDEBUG -DGATHER_STAT -x -h X XMAKEFILE = makefile X XOBJS = bitio.$(OBJ)\ X debug.$(OBJ)\ X decode.$(OBJ)\ X default.$(OBJ)\ X encode.$(OBJ)\ X freeze.$(OBJ)\ X huf.$(OBJ)\ X lz.$(OBJ) X XCATMAN = freeze.man statist.man X XMAN = freeze.1 statist.1 X XSRCS = bitio.c\ X debug.c\ X decode.c\ X default.c\ X encode.c\ X freeze.c\ X huf.c\ X lz.c X X.SUFFIXES: .man .1 .$(suffix) X X.1.man: X nroff -man < $< > $@ X X.c.$(OBJ): X $(CC) -c $(CFLAGS) $(OPTIONS) $< X Xprog: freeze$(EXE) statist$(EXE) showhuf$(EXE) X Xman: $(CATMAN) X Xlint: $(SRCS) X lint $(LINTFLAGS) $(SRCS) > lint.out X Xfreeze$(EXE): $(OBJS) X $(CC) $(LDFLAGS) -o $@ $(OBJS) $(LIBS) X Xstatist$(EXE): statist.$(OBJ) lz.$(OBJ) X $(CC) $(LDFLAGS) -o $@ statist.$(OBJ) lz.$(OBJ) $(LIBS) X Xshowhuf$(EXE): showhuf.$(OBJ) X $(CC) $(LDFLAGS) -o $@ showhuf.$(OBJ) $(LIBS) X Xclobber: clean X rm -f freeze$(EXE) statist$(EXE) showhuf$(EXE) *.man \#* *~ config.h X Xclean:; rm -f *.$(OBJ) *.b .,* core *.out X Xinstall: $(DEST)/freeze $(DEST)/statist $(MANDEST)/freeze.$(SEC) $(MANDEST)/statist.$(SEC) X Xpatch:; rm -f patch.out X -for i in distribution/*.[ch1] Makefile.in configure ; do \ X (diff -c $$i `basename $$i` >> patch.out); \ X done X X$(DEST)/freeze: freeze X $(INSTALL_PROGRAM) freeze $@ X -ln -f $@ $(DEST)/melt X -ln -f $@ $(DEST)/unfreeze X -ln -f $@ $(DEST)/fcat X X$(DEST)/statist: statist X $(INSTALL_PROGRAM) statist $@ X X$(MANDEST)/freeze.$(SEC): freeze.1 X $(INSTALL_DATA) freeze.1 $@ X -ln -f $@ $(MANDEST)/melt.$(SEC) X -ln -f $@ $(MANDEST)/unfreeze.$(SEC) X -ln -f $@ $(MANDEST)/fcat.$(SEC) X# This is much better for places which keep preformated manpages. X# echo ".so man1/freeze.$(SEC)" > $(MANDEST)/melt.$(SEC) X# echo ".so man1/freeze.$(SEC)" > $(MANDEST)/unfreeze.$(SEC) X# echo ".so man1/freeze.$(SEC)" > $(MANDEST)/fcat.$(SEC) X X X$(MANDEST)/statist.$(SEC): statist.1 X $(INSTALL_DATA) statist.1 $@ X Xx286: X $(MAKE) prog "CC=cc -LARGE" CFLAGS="-Ox -Ml2" LDFLAGS="-Ml2" X Xx286install: X $(MAKE) install MANDEST=/usr/man/man.C SEC=C X Xmsc: X if not exist config.h copy config.msc config.h X $(MAKE) prog EXE=.exe OBJ=obj CC="cl" CFLAGS="-Ms -Ox" LDFLAGS="-Ml" LIBS="" X Xturbo: X if not exist config.h copy config.tur config.h X $(MAKE) prog EXE=.exe OBJ=obj CC="tcc" CFLAGS="-O -ms" LIBS="" X Xborland: X if not exist config.h copy config.tur config.h X $(MAKE) prog EXE=.exe OBJ=obj CC="bcc" CFLAGS="-O -ms" LIBS="" X X### Xbitio.$(OBJ): freeze.h compat.h bitio.h Xdebug.$(OBJ): freeze.h compat.h huf.h bitio.h Xdecode.$(OBJ): freeze.h compat.h huf.h bitio.h Xdefault.$(OBJ): freeze.h compat.h Xencode.$(OBJ): freeze.h compat.h lz.h huf.h bitio.h Xfreeze.$(OBJ): freeze.h compat.h patchlev.h Xhuf.$(OBJ): freeze.h compat.h huf.h bitio.h Xlz.$(OBJ): freeze.h compat.h lz.h Xstatist.$(OBJ): freeze.h compat.h lz.h Xshowhuf.$(OBJ): freeze.h huf.h END_OF_FILE if test 3683 -ne `wc -c <'Makefile.in'`; then echo shar: \"'Makefile.in'\" unpacked with wrong size! fi # end of 'Makefile.in' fi if test -f 'config.h.in' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'config.h.in'\" else echo shar: Extracting \"'config.h.in'\" $2522 characters$ sed "s/^X//" >'config.h.in' <<'END_OF_FILE' X/* This is a configuration file prototype, copy it to config.h and */ X/* "#define" appropriate macros if you have problems with "configure". */ X X/* define as "int" or "void"; default (undefined) means "void" */ X#undef RETSIGTYPE X X/* define if your computer/system allows unaligned word access */ X#undef ALLOW_MISALIGN X X/* define if sizeof(int) == 2 */ X#undef INT_16_BITS X X/* define if your computer cannot handle data items of more than 64K */ X#undef SEGMENTED X X/* define if filenames can be of more than 14 chars */ X#undef HAVE_LONG_FILE_NAMES X X/* define no more than one, according to your standard #include's */ X/* if you have <dirent.h> */ X#undef DIRENT X/* if you have <sys/ndir.h> */ X#undef SYSNDIR X/* if you have <sys/dir.h> */ X#undef SYSDIR X X/* define if you have <sys/stdtypes.h> */ X#undef HAVE_SYS_STDTYPES_H X X/* define if you have "rindex" and "setlinebuf" correspondingly */ X#undef HAVE_RINDEX X#undef HAVE_SETLINEBUF X X/* define no more than one, according to your standard #include's */ X/* if you have <utime.h> */ X#undef UTIME X/* if you have <sys/utime.h> */ X#undef SYSUTIME X/* if you have "struct timeval" in <sys/time.h> */ X#undef SYSTIME X X/* define if you want to have freeze compatible with vers. 1.0 */ X#undef COMPAT X X/* define if your system has multibyte NEWLINE (as in MS-DOS) and */ X/* you want to do text conversion by default */ X#undef TEXT_DEFAULT X X/* define if you want to build freeze in small model (64K data) */ X/* (segmented architectures only) */ X#undef TINY X X/* define if you want to decrease the amount of memory but without */ X/* 64K restriction */ X#undef SMALL X X/* define to increase the compression speed by about 10% at the cost */ X/* of some tenths of % compression rate */ X#undef FASTHASH X X/* default Huffman values, define if you don't like the default */ X/* 0,0,1,2,6,19,34,0. These below are reasonably good also. */ X/* #define HUFVALUES 0,1,1,1,4,10,27,18 */ END_OF_FILE if test 2522 -ne `wc -c <'config.h.in'`; then echo shar: \"'config.h.in'\" unpacked with wrong size! fi # end of 'config.h.in' fi if test -f 'config.msc' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'config.msc'\" else echo shar: Extracting \"'config.msc'\" $2490 characters$ sed "s/^X//" >'config.msc' <<'END_OF_FILE' X/* This is a configuration file prototype, copy it to config.h and */ X/* "#define" appropriate macros if you have problems with "configure". */ X X/* define as "int" or "void" */ X#define RETSIGTYPE void X X/* define if your computer/system allows unaligned word access */ X#define ALLOW_MISALIGN X X/* define if sizeof(int) == 2 */ X#define INT_16_BITS X X/* define if your computer cannot handle data items of more than 64K */ X#define SEGMENTED X X/* define if filenames can be of more than 14 chars */ X#undef HAVE_LONG_FILE_NAMES X X/* define no more than one of, according to your standard #include's */ X/* if you have <dirent.h> */ X#undef DIRENT X/* if you have <sys/ndir.h> */ X#undef SYSNDIR X/* if you have <sys/dir.h> */ X#undef SYSDIR X X/* define if you have <sys/stdtypes.h> */ X#undef HAVE_SYS_STDTYPES_H X X/* define if you have "rindex" and "setlinebuf" correspondingly */ X#undef HAVE_RINDEX X#undef HAVE_SETLINEBUF X X/* define no more than one of, according to your standard #include's */ X/* if you have <utime.h> */ X#define UTIME X/* if you have <sys/utime.h> */ X#undef SYSUTIME X/* if you have "struct timeval" in <sys/time.h> */ X#undef SYSTIME X X/* define if you want to have freeze compatible with vers. 1.0 */ X#undef COMPAT X X/* define if your system has multibyte NEWLINE (as in MS-DOS) and */ X/* you want to do text conversion by default */ X#undef TEXT_DEFAULT X X/* define if you want to build freeze in small model (64K data) */ X/* (segmented architectures only) */ X#define TINY X X/* define if you want to decrease the amount of memory but without */ X/* 64K restriction */ X#undef SMALL X X/* define to increase the compression speed by about 10% at the cost */ X/* of some tenths of % compression rate */ X#undef FASTHASH X X/* default Huffman values, define if you don't like the default */ X/* 0,0,1,2,6,19,34,0. These below are reasonably good also. */ X/* #define HUFVALUES 0,1,1,1,4,10,27,18 */ END_OF_FILE if test 2490 -ne `wc -c <'config.msc'`; then echo shar: \"'config.msc'\" unpacked with wrong size! fi # end of 'config.msc' fi if test -f 'config.tur' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'config.tur'\" else echo shar: Extracting \"'config.tur'\" $2584 characters$ sed "s/^X//" >'config.tur' <<'END_OF_FILE' X/* This is a configuration file prototype, copy it to config.h and */ X/* "#define" appropriate macros if you have problems with "configure". */ X X/* define as "int" or "void" */ X#define RETSIGTYPE void X X/* define if your computer/system allows unaligned word access */ X#define ALLOW_MISALIGN X X/* define if sizeof(int) == 2 */ X#define INT_16_BITS X X/* define if your computer cannot handle data items of more than 64K */ X#define SEGMENTED X X/* define if filenames can be of more than 14 chars */ X#undef HAVE_LONG_FILE_NAMES X X/* define no more than one of, according to your standard #include's */ X/* if you have <dirent.h> */ X#undef DIRENT X/* if you have <sys/ndir.h> */ X#undef SYSNDIR X/* if you have <sys/dir.h> */ X#undef SYSDIR X X/* define if you have <sys/stdtypes.h> */ X#undef HAVE_SYS_STDTYPES_H X X/* define if you have "rindex" and "setlinebuf" correspondingly */ X#undef HAVE_RINDEX X#undef HAVE_SETLINEBUF X X/* define no more than one of, according to your standard #include's */ X/* if you have <utime.h> */ X#undef UTIME X/* if you have <sys/utime.h> */ X#undef SYSUTIME X/* if you have "struct timeval" in <sys/time.h> */ X#undef SYSTIME X X/* BORLAND C has no off_t definition in <sys/types.h> */ X#define off_t long X X/* define if you want to have freeze compatible with vers. 1.0 */ X#undef COMPAT X X/* define if your system has multibyte NEWLINE (as in MS-DOS) and */ X/* you want to do text conversion by default */ X#undef TEXT_DEFAULT X X/* define if you want to build freeze in small model (64K data) */ X/* (segmented architectures only) */ X#define TINY X X/* define if you want to decrease the amount of memory but without */ X/* 64K restriction */ X#undef SMALL X X/* define to increase the compression speed by about 10% at the cost */ X/* of some tenths of % compression rate */ X#undef FASTHASH X X/* default Huffman values, define if you don't like the default */ X/* 0,0,1,2,6,19,34,0. These below are reasonably good also. */ X/* #define HUFVALUES 0,1,1,1,4,10,27,18 */ END_OF_FILE if test 2584 -ne `wc -c <'config.tur'`; then echo shar: \"'config.tur'\" unpacked with wrong size! fi # end of 'config.tur' fi if test -f 'default.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'default.c'\" else echo shar: Extracting \"'default.c'\" $1274 characters$ sed "s/^X//" >'default.c' <<'END_OF_FILE' X#include "freeze.h" X#include <errno.h> X X /*-------------------------------*/ X /* DEFAULTS FILE HANDLING */ X /*-------------------------------*/ X X#define OK 0 X#define FAIL NULL X#define NOFILE ((FILE *) 0) X#define MAXLINE 128 X Xchar *strchr(); Xstatic FILE *defd = NOFILE; /* defaults file stream */ X Xint defopen(fname) /* open | reopen | close defaults file */ X char *fname; X{ X register FILE *fd; X X if (!fname) { X if (defd) X (void) fclose(defd); X defd = NOFILE; X return OK; X } X X if (!(fd = fopen(fname, "r"))) X return errno; /* problems opening file */ X X defd = fd; X return OK; X} X Xstatic char defline[MAXLINE + 1]; X Xchar *defread(pattern) X register char *pattern; X{ X register sz_patt; X register char *cp; X X if (!defd) X return FAIL; /* defaults file not opened */ X X rewind(defd); X sz_patt = strlen(pattern); X X while (fgets(defline, MAXLINE, defd)) { X if (!(cp = strchr(defline, '\n'))) X return FAIL; /* line too long */ X if (cp - defline < sz_patt) X continue; /* line too short */ X *cp = '\0'; X if (!strncmp(pattern, defline, sz_patt)) X return defline + sz_patt; /* match found */ X } X X return FAIL; /* no matching lines */ X} END_OF_FILE if test 1274 -ne `wc -c <'default.c'`; then echo shar: \"'default.c'\" unpacked with wrong size! fi # end of 'default.c' fi if test -f 'freeze.1' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'freeze.1'\" else echo shar: Extracting \"'freeze.1'\" $7193 characters$ sed "s/^X//" >'freeze.1' <<'END_OF_FILE' X.PU X.TH FREEZE 1 local X.SH NAME Xfreeze, unfreeze, melt, fcat \- compress and uncompress files X.SH SYNOPSIS X.ll +8 X.B freeze X[ X.B \-cdfvVgx X] [ X.I "filename | type \&..." X] X.ll -8 X.br X.B unfreeze X[ X.B \-cfvV X] [ X.I "filename \&..." X] X.br X.B melt X[ X.B \-cfvV X] [ X.I "filename \&..." X] X.br X.B fcat X[ X.I "filename \&..." X] X.SH DESCRIPTION XCompresses the specified files or standard input. XEach file is replaced by a file with the extension X.B "\&.F," Xbut only if the file got smaller. If no files are specified, Xthe compression is applied to the standard input Xand is written to standard output regardless of the results. XCompressed files can be restored to their original form by specifying the X.B \-d Xoption, or by running X.I melt Xor X.I unfreeze X(both linked to X.IR freeze ), Xon the X.B "\&.F" Xfiles or the standard input. X.PP XIf the output file exists, it will not be overwritten unless the X.B \-f Xflag is given. If X.B \-f Xis not specified and X.I freeze Xis run in the foreground, Xthe user is prompted Xas to whether the file should be overwritten. X.PP XIf the X.B \-g Xflag is given, a slightly less powerful (compression Xrate is 1.5% less), but somewhat faster heuristic is used. This flag can be Xused more than once (this mode is quite useful when freezing bitmaps) for Xadditional speedup. X.PP XIf you want to improve compression rate at the cost of speed, use X.B \-x Xflag. Like X.B \-g, Xthis flag can be used more than once, thus increasing compression rate. XNote that usually it isn't worth to use it more than twice. X.PP XIf the X.B \-f Xflag is given, all files specified are replaced with X.B "\&.F" Xfiles \- even if the file didn't get smaller. X.PP XWhen file names are given, the ownership (if run by root), modes, accessed Xand modified times are maintained between the file and its X.B "\&.F" Xversion. In this respect, X.I freeze Xcan be used for archival purposes, yet can still be used with X.IR make "(1)" Xafter melting. X.PP XThe X.B \-c Xoption causes the results of the freeze/melt operation to be written Xto stdout; no files are changed. The X.I fcat Xprogram is the same as specifying X.B \-c Xto X.I melt X(all files are unpacked and written to stdout). X.PP XThe X.B \-v X(verbose) option causes the diagnostics (at the end of each file processing) Xto be printed to stderr, and the X.B \-vv Xoption causes the progress indicator to be drawn to the same place. X.PP X.I Type Xis a token preceded by a '+' or a '--', which defines the type Xof following files in the command string. An explicite definition Xof the file's type can give up to 2% of additional compression. XThe list of types is stored in file X.IR /usr/local/lib/freeze.cnf . XTypes may be abbreviated while not ambigious. You can also determine Xvalues for the static Huffman table by using a list of 8 numbers Xseparated by commas instead of X.I type. X.PP X.I Freeze Xuses the Lempel-Ziv algorithm on the first pass and the dynamic XHuffman algorithm on the second one. The size of sliding window Xis 8K, and the maximum length of matched string is 256. XThe positions on the window are coded using a static Huffman table. X.PP XA two byte magic number is prepended to the file Xto ensure that neither melting of random text nor refreezing of Xalready frozen text are attempted. In addition, the characteristics Xof the static Huffman table being used during X.I freeze Xis written to the file so that these characteristics may be adapted Xto concrete conditions. X.PP X.ne 8 XThe amount of compression obtained depends on the size of the Xinput file and the distribution of character substrings and their Xprobabilities. XTypically, text files, such as C programs, Xare reduced by 60\-75%, executable files are reduced by 50%. XCompression is generally much better than that achieved by XLZW coding (as used in X.IR compress ), Xor Huffman coding X.RI ( pack ), Xthough takes more time to compute. X.PP XIf the X.B \-V X(version) flag is given, the program's version number and compilation Xoptions are printed. X.PP XThe exit status is normally 0; Xif the last file gets bigger after freezing, the exit status is 2; Xif an error occurs, the exit status is 1. X.SH "SEE ALSO" Xcompact(1), pack(1), compress(1) X.SH "DIAGNOSTICS" XUnknown flag: X.I "\'x\';" XUsage: freeze [-cdfvVg] [file|+type ...] X.in +8 XInvalid options were specified on the command line. X.in -8 X.IR file : Xnot in frozen format X.in +8 XThe specified file has not been frozen. X.in -8 X.IR file : Xalready has .F suffix -- no change X.in +8 XCannot compress a file that has a ".F" suffix. X.IR mv "(1)" Xthe file to a different name and try again. X.in -8 X.IR file : Xfilename too long to tack on .F X.in +8 XThe specified file cannot be compressed because its filename is longer than X12 characters. X.IR mv "(1)" Xthe file to a different name and try again. This message does not occur on X4.XBSD systems. X.in -8 X.I file Xalready exists; do you wish to overwrite (y or n)? X.in +8 XRespond "y" if you want the output file to be replaced; "n" if you want it Xto be left alone. X.in -8 X.IR file : X.IR xx % X.in +8 Xor X.in -8 X.IR xxx K X.in +8 XThese message fragments are written during the processing of a file, if X.B \-vv Xoption was given in the command line (in percents, if the length of file Xbeing processed is known; in Kbytes otherwise). X.in -8 XFreezing: X.I "xx.xx% (y.yy" Xbits) X.in +8 XThis message fragment gives the percentage of the input file that has been Xsaved by freezing and the number of remaining bits per byte of original file. X.in -8 X-- not a regular file: unchanged X.in +8 XThis message fragment is written when the input file is not a regular file. XThe input file is left unchanged. X.in -8 X-- has X.I xx Xother links: unchanged X.in +8 XThis message fragment is written when the input file has links. The input Xfile is left unchanged. See X.IR ln "(1)" Xfor more information. X.in -8 X-- file unchanged X.in +8 XThis message fragment is written when no savings are achieved by Xfreezing. The input file is left unchanged. X.in -8 X-- replaced with X.I file X.in +8 XThis message fragment is written when a file has been sucessfully Xfrozen/melt. X.in -8 XUsing " X.I type X" type X.in +8 XThis message indicates a successful switching to Xposition table for mentioned file type. X.in -8 X" X.I xxx X" - no such file type X.in +8 Xor X.in -8 X.I xxx X- a list of 8 numbers expected X.in +8 XThis message means the given file type does not exist or Xthe given string contains a comma, but is not a valid list Xof values for static Huffman table. X.in -8 Xmelt: corrupt input X.in +8 XThis message fragment is written when an error in header or Xunexpected end of frozen file is detected. Partial X(or empty, is there was an error in the header) file is created. X.in -8 Xalready frozen -- file unchanged X.in +8 XThis message fragment is written when an input file already has XFreeze's magic header. X.in -8 XInvalid position table X.in +8 Xor X.in -8 X" X.I type X" - invalid entry X.in +8 XThese messages appear only if Freeze has been made with incorrect Xdata for static Huffman table. It does never appear when freeze Xis called from a public access directory. X.in -8 XUnknown header format X.in +8 XUnknown values of flag bits were discovered in the header Xof frozen file. X.in -8 X.SH "BUGS" XFound bugs descriptions, incompatibilities, etc. please send to Xleo@ipmce.su. END_OF_FILE if test 7193 -ne `wc -c <'freeze.1'`; then echo shar: \"'freeze.1'\" unpacked with wrong size! fi # end of 'freeze.1' fi if test -f 'freeze.h' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'freeze.h'\" else echo shar: Extracting \"'freeze.h'\" $4101 characters$ sed "s/^X//" >'freeze.h' <<'END_OF_FILE' X#include <stdio.h> X#include "config.h" X X#ifdef HAVE_SYS_STDTYPES_H X# include <sys/stdtypes.h> X#endif X X#ifndef getc X# ifdef m88k /* Green Hill C library bug. */ X# define getc(p) (--(p)->_cnt < 0 ? __filbuf(p) : (int) *(p)->_ptr++) X# else X# define getc(p) (--(p)->_cnt < 0 ? _filbuf(p) : (int) *(p)->_ptr++) X# endif X#endif X#ifndef putc X# ifdef m88k /* Green Hill C library bug. */ X# define putc(x, p) (--(p)->_cnt < 0 ? __flsbuf((unsigned char) (x), (p)) : (int) (*(p)->_ptr++ = (unsigned char) (x))) X# else X# define putc(x, p) (--(p)->_cnt < 0 ? _flsbuf((unsigned char) (x), (p)) : (int) (*(p)->_ptr++ = (unsigned char) (x))) X# endif X#endif X X#if !defined(MSDOS) && defined(__MSDOS__) X# define MSDOS X#endif X X#ifdef MSDOS X# define DOS X# include <fcntl.h> X#endif /* MSDOS */ X X#ifdef TOS X# define DOS X# define O_TEXT 0x01 X# define O_BINARY 0x02 X#endif X X#include <ctype.h> X#include <signal.h> X X#ifndef RETSIGTYPE X# define RETSIGTYPE void X#endif /* RETSIGTYPE */ X X#ifndef TOS X# include <sys/types.h> X# include <sys/stat.h> X#else X# include <tos.h> X# include <types.h> X#endif X X#ifdef SYSTIME X# include <sys/time.h> X# define UTIMES X#else X# ifdef UTIME X# include <utime.h> X# else X# ifdef SYSUTIME X# include <sys/utime.h> X# else X# ifdef unix X/* UNIX without any declaration of utimbuf .... Strange! */ Xstruct utimbuf { X time_t actime; X time_t modtime; X}; Xextern int utime(); X# else X# ifndef DOS X# define BITS no_utimbuf_definition_on_unknown_system X# endif X# endif X# endif X# endif X#endif X X/* for MAXNAMLEN only !!! */ X#ifdef DIRENT X# include <dirent.h> X#else X# ifdef SYSNDIR X# include <sys/ndir.h> X# else X# ifdef SYSDIR X# include <sys/dir.h> X# endif X# endif X#endif X X#ifndef MAXNAMLEN X# define MAXNAMLEN 255 X#endif X X#if MAXNAMLEN < 255 X# undef MAXNAMLEN X# define MAXNAMLEN 255 X#endif X X#ifdef DEBUG X# include <assert.h> X#endif /* DEBUG */ X X#ifdef DOS X# include <stdlib.h> X#endif /* DOS */ X X#ifdef __TURBOC__ X# ifdef MSDOS X# include <io.h> X# include <alloc.h> X# else /* TOS */ X# include <ext.h> X# endif /* MSDOS */ X#endif /* __TURBOC__ */ X Xtypedef unsigned short us_t; Xtypedef unsigned char uc_t; Xtypedef unsigned long ul_t; X X#define LOOKAHEAD 256 /* pre-sence buffer size */ X#define MAXDIST 7936 X#define WINSIZE (MAXDIST + LOOKAHEAD) /* must be a power of 2 */ X#define WINMASK (WINSIZE - 1) X X#define THRESHOLD 2 X X#define N_CHAR2 (256 - THRESHOLD + LOOKAHEAD + 1) /* code: 0 .. N_CHARi - 1 */ X#define T2 (N_CHAR2 * 2 - 1) /* size of table */ X X#define ENDOF 256 /* pseudo-literal */ X Xextern uc_t Table2[]; X Xextern long in_count, bytes_out; Xextern off_t file_length; X Xextern uc_t text_buf[]; X Xextern long indc_threshold, indc_count; X Xextern short do_melt, topipe, greedy, quiet, force; /* useful flags */ X X#define MAGIC1 ((uc_t)'\037') X#define MAGIC2_1 ((uc_t)'\236') /* freeze vers. 1.X */ X#define MAGIC2_2 ((uc_t)'\237') X Xextern int exit_stat; X X#ifdef DEBUG Xextern short debug; Xextern short verbose; Xextern char * pr_char(); X#endif /* DEBUG */ X X#if defined(GATHER_STAT) || defined(DEBUG) Xextern long refers_out, symbols_out; X#endif X Xextern short DecodeChar(), DecodePosition(), GetNBits(); Xextern void melt2(), (*meltfunc)(), writeerr(), prratio(), prbits(), freeze(); X X#ifdef COMPAT X#include "compat.h" X#endif X X#define INDICATOR \ Xif (quiet < 0 && (in_count > indc_count)) {\ X if (ferror(stdout))\ X writeerr();\ X if (file_length) {\ X static int percents, old_percents = -1;\ X if ((percents = ftell(stdin) * 100 / file_length) !=\ X old_percents) {\ X fprintf(stderr, " %2d%%\b\b\b\b", percents);\ X old_percents = percents;\ X }\ X indc_count += indc_threshold;\ X } else {\ X fprintf(stderr, " %5ldK\b\b\b\b\b\b\b", in_count / 1024);\ X indc_count += indc_threshold;\ X indc_threshold += 1024;\ X }\ X fflush (stderr);\ X} X X#ifdef HAVE_RINDEX X#define strchr index X#define strrchr rindex X#endif X Xextern char *strchr(), *strrchr(); X END_OF_FILE if test 4101 -ne `wc -c <'freeze.h'`; then echo shar: \"'freeze.h'\" unpacked with wrong size! fi # end of 'freeze.h' fi if test -f 'huf.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'huf.c'\" else echo shar: Extracting \"'huf.c'\" $7717 characters$ sed "s/^X//" >'huf.c' <<'END_OF_FILE' X#include "freeze.h" X#include "huf.h" X#include "bitio.h" X X/*----------------------------------------------------------------------*/ X/* */ X/* HUFFMAN ENCODING */ X/* */ X/*----------------------------------------------------------------------*/ X X/* TABLES OF ENCODE/DECODE for upper 6 bits position information */ X X/* The contents of `Table' are used for freezing only, so we use X * it freely when melting. X */ X X#ifndef HUFVALUES X#define HUFVALUES 0,0,1,2,6,19,34,0 X#endif X Xuc_t Table2[9] = { 0, HUFVALUES }; X Xuc_t p_len[64]; /* These arrays are built accordingly to values */ Xuc_t d_len[256]; /* of `Table' above which are default, from the */ X /* command line or from the header of frozen file */ X Xuc_t code[256]; X Xus_t freq[T2 + 1]; /* frequency table */ Xshort son[T2]; /* points to son node (son[i],son[i+1]) */ Xshort prnt[T2 + N_CHAR2]; /* points to parent node */ X Xstatic short t, r, chars; X X/* notes : X prnt[Tx .. Tx + N_CHARx - 1] used by X indicates leaf position that corresponding to code. X*/ X X/* Initializes Huffman tree, bit I/O variables, etc. X Static array is initialized with `table', dynamic Huffman tree X has `n_char' leaves. X*/ X Xvoid StartHuff (n_char) X int n_char; X{ X register short i, j; X t = n_char * 2 - 1; X r = t - 1; X chars = n_char; X X/* A priori frequences are 1 */ X X for (i = 0; i < n_char; i++) { X freq[i] = 1; X son[i] = i + t; X prnt[i + t] = i; X } X i = 0; j = n_char; X X/* Building the balanced tree */ X X while (j <= r) { X freq[j] = freq[i] + freq[i + 1]; X son[j] = i; X prnt[i] = prnt[i + 1] = j; X i += 2; j++; X } X freq[t] = 0xffff; X prnt[r] = 0; X in_count = 1; X bytes_out = 5; X#if defined(DEBUG) || defined (GATHER_STAT) X symbols_out = refers_out = 0; X#endif X} X X/* Reconstructs tree with `chars' leaves */ X Xvoid reconst () X{ X register int i, j, k; X register int f; X X#ifdef DEBUG X if (quiet < 0) X fprintf(stderr, X "Reconstructing Huffman tree: symbols: %ld, references: %ld\n", X symbols_out, refers_out); X#endif X X/* correct leaf node into of first half, X and set these freqency to (freq+1)/2 X*/ X j = 0; X for (i = 0; i < t; i++) { X if (son[i] >= t) { X freq[j] = (freq[i] + 1) / 2; X son[j] = son[i]; X j++; X } X } X/* Build tree. Link sons first */ X X for (i = 0, j = chars; j < t; i += 2, j++) { X k = i + 1; X f = freq[j] = freq[i] + freq[k]; X for (k = j - 1; f < freq[k]; k--); X k++; X { register us_t *p, *e; X for (p = &freq[j], e = &freq[k]; p > e; p--) X p[0] = p[-1]; X freq[k] = f; X } X { register short *p, *e; X for (p = &son[j], e = &son[k]; p > e; p--) X p[0] = p[-1]; X son[k] = i; X } X } X X/* Link parents */ X for (i = 0; i < t; i++) { X if ((k = son[i]) >= t) { X prnt[k] = i; X } else { X prnt[k] = prnt[k + 1] = i; X } X } X} X X X/* Updates given code's frequency, and updates tree */ X Xvoid update (c) X register int c; X{ X register us_t *p; X register int i, j, k, l; X X if (freq[r] == MAX_FREQ) { X reconst(); X } X c = prnt[c + t]; X do { X k = ++freq[c]; X X /* swap nodes when become wrong frequency order. */ X if (k > freq[l = c + 1]) { X for (p = freq+l+1; k > *p++; ) ; X l = p - freq - 2; X freq[c] = p[-2]; X p[-2] = k; X X i = son[c]; X prnt[i] = l; X if (i < t) prnt[i + 1] = l; X X j = son[l]; X son[l] = i; X X prnt[j] = c; X if (j < t) prnt[j + 1] = c; X son[c] = j; X X c = l; X } X } while ((c = prnt[c]) != 0); /* loop until reach to root */ X} X X/* Encodes the literal or the length information */ X Xvoid EncodeChar (c) X int c; X{ X ul_t i; X register int j, k; X X i = 0; X j = 0; X k = prnt[c + t]; X X/* trace links from leaf node to root */ X X do { X i >>= 1; X X/* if node index is odd, trace larger of sons */ X if (k & 1) i += 0x80000000; X X j++; X } while ((k = prnt[k]) != r) ; X X/* `j' never reaches the value of 32 ! */ X X if (j > 16) { X Putcode(16, (us_t)(i >> 16)); X Putcode(j - 16, (us_t)i); X } else { X Putcode(j, (us_t)(i >> 16)); X } X update(c); X} X X/* Encodes the position information */ X Xvoid EncodePosition (c) X register int c; X{ X register us_t i; X X /* output upper 6 bit from table */ X i = c >> 7; X Putcode((int)p_len[i], (us_t)(code[i]) << 8); X X /* output lower 7 bit */ X Putcode(7, (us_t)(c & 0x7f) << 9); X} X X X/* Decodes the literal or length info and returns its value. X Returns ENDOF, if the file is corrupt. X*/ X Xshort DecodeChar () X{ X register int c; X c = son[r]; X X /* trace from root to leaf, X got bit is 0 to small(son[]), 1 to large (son[]+1) son node */ X X while (c < t) { X c += GetBit(); X c = son[c]; X } X c -= t; X update(c); X if (crpt_flag) { X crpt_message(); X return ENDOF; X } X crpt_flag = feof(stdin); X return c; X} X X/* Decodes the position info and returns it */ X Xshort DecodePosition () X{ X register us_t i, j, c; X X /* decode upper 6 bits from the table */ X X i = GetByte(); X crpt_flag = feof(stdin); X X c = (us_t)code[i] << 7; X j = d_len[i] - 1; X X /* get lower 7 bits literally */ X X return c | (((i << j) | GetNBits (j)) & 0x7f); X} X X X/* Initializes static Huffman arrays */ X Xvoid init(table) uc_t * table; { X short i, j, k, num; X num = 0; X X/* There are `table[i]' `i'-bits Huffman codes */ X X for(i = 1, j = 0; i <= 8; i++) { X num += table[i] << (8 - i); X for(k = table[i]; k; j++, k--) X p_len[j] = i; X } X if (num != 256) { X fprintf(stderr, "Invalid position table\n"); X exit(1); X } X num = j; X if (do_melt == 0) X X/* Freezing: building the table for encoding */ X X for(i = j = 0;;) { X code[j] = i << (8 - p_len[j]); X i++; X j++; X if (j == num) break; X i <<= p_len[j] - p_len[j-1]; X } X else { X X/* Melting: building the table for decoding */ X X for(k = j = 0; j < num; j ++) X for(i = 1 << (8 - p_len[j]); i--;) X code[k++] = j; X X for(k = j = 0; j < num; j ++) X for(i = 1 << (8 - p_len[j]); i--;) X d_len[k++] = p_len[j]; X } X} X X/* Writes a 3-byte header into the frozen form of file; Table[7] and X Table[8] aren't necessary, see `read_header'. X*/ X Xvoid write_header() { X us_t i; X X i = Table2[5] & 0x1F; i <<= 4; X i |= Table2[4] & 0xF; i <<= 3; X i |= Table2[3] & 7; i <<= 2; X i |= Table2[2] & 3; i <<= 1; X i |= Table2[1] & 1; X X putchar((int)(i & 0xFF)); X putchar((int)((i >> 8))); X putchar((int)(Table2[6] & 0x3F)); X if (ferror(stdout)) X writeerr(); X} X X/* Reconstructs `Table' from the header of the frozen file and checks X its correctness. Returns 0 if OK, EOF otherwise. X*/ X Xint read_header() { X short i, j; X i = getchar() & 0xFF; X i |= (getchar() & 0xFF) << 8; X Table2[1] = i & 1; i >>= 1; X Table2[2] = i & 3; i >>= 2; X Table2[3] = i & 7; i >>= 3; X Table2[4] = i & 0xF; i >>= 4; X Table2[5] = i & 0x1F; i >>= 5; X X if (i & 1 || (i = getchar()) & 0xC0) { X fprintf(stderr, "Unknown header format.\n"); X crpt_message(); X return EOF; X } X X Table2[6] = i & 0x3F; X X i = Table2[1] + Table2[2] + Table2[3] + Table2[4] + X Table2[5] + Table2[6]; X X i = 62 - i; /* free variable length codes for 7 & 8 bits */ X X j = 128 * Table2[1] + 64 * Table2[2] + 32 * Table2[3] + X 16 * Table2[4] + 8 * Table2[5] + 4 * Table2[6]; X X j = 256 - j; /* free byte images for these codes */ X X/* Equation: X Table[7] + Table[8] = i X 2 * Table[7] + Table[8] = j X*/ X j -= i; X if (j < 0 || i < j) { X crpt_message(); X return EOF; X } X Table2[7] = j; X Table2[8] = i - j; X X#ifdef DEBUG X fprintf(stderr, "Codes: %d %d %d %d %d %d %d %d\n", X Table2[1], Table2[2], Table2[3], Table2[4], X Table2[5], Table2[6], Table2[7], Table2[8]); X#endif X return 0; X} X X#ifdef COMPAT X Xuc_t Table1[9] = { 0, 0, 0, 1, 3, 8, 12, 24, 16 }; X X/* Old version of a routine above for handling files made by X the 1st version of Freeze. X*/ X Xshort DecodePOld () X{ X register us_t i, j, c; X X i = GetByte(); X crpt_flag = feof(stdin); X X c = (us_t)code[i] << 6; X j = d_len[i] - 2; X X return c | (((i << j) | GetNBits (j)) & 0x3f); X} X#endif END_OF_FILE if test 7717 -ne `wc -c <'huf.c'`; then echo shar: \"'huf.c'\" unpacked with wrong size! fi # end of 'huf.c' fi if test -f 'lz.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'lz.c'\" else echo shar: Extracting \"'lz.c'\" $3906 characters$ sed "s/^X//" >'lz.c' <<'END_OF_FILE' X#include "freeze.h" X#include "lz.h" X X/*----------------------------------------------------------------------*/ X/* */ X/* LZSS ENCODING */ X/* */ X/*----------------------------------------------------------------------*/ X Xuc_t text_buf[WINSIZE + LOOKAHEAD - 1];/* cyclic buffer with an overlay */ Xint match_position; /* current position of X matched pattern */ Xint chain_length; /* max_chain_length == X CHAIN_THRESHOLD >> greedy */ X X/* next[N+1..] is used as hash table, X the rest of next is a link down, X*/ X Xhash_t hashtab[array_size]; /* a VERY large array :-) */ Xhash_t next[WINSIZE]; X X#ifdef GATHER_STAT Xlong node_matches, node_compares, node_prolongations; X#endif /* GATHER_STAT */ X X/* Initialize the data structures and allocate memory, if needed. X Although there is no more trees in the LZ algorithm X implementation, routine name is kept intact :-) X*/ X Xvoid InitTree () X{ X X#if BITS == 16 X ul_t i; X#else X unsigned i; X#endif X X#ifdef GATHER_STAT X node_matches = node_compares = node_prolongations = 0; X#endif /* GATHER_STAT */ X X for (i = 0; i < array_size; i++ ) X hashtab[i] = 0; X X if (greedy >= 0) X chain_length = ((CHAIN_THRESHOLD - 1) >> greedy) + 1; X else X chain_length = ((CHAIN_THRESHOLD - 1) << -greedy) + 1; X} X X/* Get the longest (longer than `match_length' when entering in function) X nearest match of the string beginning in text_buf[r] X to the cyclic buffer. Result (length & position) is returned X as the result and in global variable X `match_position'). Unchanged `match_length' denotes failure and X `match_position' contains garbage !! X In order to achieve faster operation, `match_length' is shifted X down to LOOKAHEAD. Ideas of Andrew Cadach <kadach@isi.itfs.nsk.su> X have been used (lastbyte). X*/ X Xint get_next_match (match_length, r) X register hash_t r; int match_length; X{ X register int p = r & WINMASK; X register int m; X#ifdef ALLOW_MISALIGN X register us_t lastbyte; X#else X register uc_t lastbyte; X#endif X register uc_t *key FIX_SI, *pattern FIX_DI; X int chain_count = chain_length; X X#ifdef GATHER_STAT X node_matches++; X#endif X key = text_buf + (r & WINMASK) + LOOKAHEAD; X r -= MAXDIST; /* `r' is now a "barrier value" */ X X for(;;) { X lastbyte = FETCH(key, match_length); X do { X if(chain_count <= 0) X /* chain length exceeded, simple return */ X return match_length; X X pattern = text_buf + match_length + LOOKAHEAD; X X do { X if ((p = next[p]) < r) X return match_length; X } while (FETCH(pattern, p &= WINMASK) != lastbyte); X X chain_count--; /* successful lastbyte match, cost = 1 */ X pattern = text_buf + p + LOOKAHEAD; X X#ifdef GATHER_STAT X node_compares++; X#endif X X#ifdef ALLOW_MISALIGN X for (m = -LOOKAHEAD; X *(unsigned*)&key[m] == *(unsigned*)&pattern[m] && X (m += sizeof(unsigned)) < 0;); X#ifndef INT_16_BITS X if (m < 0 && *(us_t*)&key[m] == *(us_t*)&pattern[m]) X m += sizeof(us_t); X#endif X if (m < 0 && key[m] == pattern[m]) X ++m; X#else X for (m = -LOOKAHEAD; key[m] == pattern[m] && ++m < 0;); X#endif X } while (m < match_length); X X match_position = p; /* remember new results */ X if (m == 0) X return 0; X match_length = m; X X#ifdef GATHER_STAT X node_prolongations++; X#endif X chain_count -= 2; /* yet another match found, cost = 2 */ X } X} X Xhash_t Xrehash(r) Xhash_t r; X{ X#if BITS == 16 X ul_t i; X#else X unsigned i; X#endif X r += WINSIZE; X for (i = 0; i < WINSIZE; i++) X /* zero must remain zero */ X if (next[i] && (next[i] += WINSIZE) > r) { X next[i] = 0; X } X for (i = 0; i < array_size; i++) X /* zero must remain zero */ X if (hashtab[i] && (hashtab[i] += WINSIZE) > r) { X hashtab[i] = 0; X } X return r; X} END_OF_FILE if test 3906 -ne `wc -c <'lz.c'`; then echo shar: \"'lz.c'\" unpacked with wrong size! fi # end of 'lz.c' fi if test -f 'lz.h' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'lz.h'\" else echo shar: Extracting \"'lz.h'\" $2916 characters$ sed "s/^X//" >'lz.h' <<'END_OF_FILE' Xextern void InitTree(); X X#ifndef SEGMENTED X# define MAXBITS 16 X#else X# ifdef INT_16_BITS X# define MAXBITS 15 X# else X# define MAXBITS 14 X# endif X#endif X X#ifdef SEGMENTED X# ifdef TINY X# undef MAXBITS X# define MAXBITS 13 X# endif X#endif X X#ifndef BITS X# define BITS MAXBITS X#endif X X#if BITS < 13 X# undef BITS X# define BITS 13 /* 1:1 hash */ X#endif X X#if BITS > 16 X# undef BITS X# define BITS 16 X#endif X X/* The following hash-function isn't optimal but it is very fast: X X HASH = ((first + (second << LEN0) + X (third << LEN1)) & ((1L << BITS) - 1); X X The difference of LENs is no more than one bit. X*/ X X#define LEN0 ((BITS-8)/2) X#define LEN1 (BITS-8) X X/* `array_size' is the size of array `next', which contains X the heads of linked lists and the references to X next members of these lists. X*/ X X#define array_size (1L << BITS) X X/* If native size of integer is 16 bits, don't use longs */ X X#if defined(INT_16_BITS) || defined(SMALL) || defined(TINY) Xtypedef us_t hash_t; X#else Xtypedef ul_t hash_t; X#endif /* INT_16_BITS */ X Xextern int match_position, chain_length; X Xextern hash_t hashtab[], next[]; X X/* Some defines to eliminate function-call overhead */ X X/* Hash function (no more than 16 bits, so we don't need longs */ X X#define hash(p)\ X ((unsigned)(p)[0] + ((unsigned)(p)[1] << LEN0) +\ X ((unsigned)(p)[2] << LEN1)) X X#ifdef FASTHASH X#define hashof(p)\ X (((p)[0] != (p)[1] ? hash(p) : hash(p) + hash((p) + 3)) &\ X ((1L << BITS) - 1)) X#else X#define hashof(p)\ X (hash(p) & ((1L << BITS) - 1)) X#endif X X/* Inserting of a node `r' into hashed linked list: `r' becomes X the head of list. X*/ X X#define InsertNode()\ X{\ X register uc_t *key = &text_buf[r & WINMASK];\ X register unsigned p = hashof(key);\ X if (r < MAXDIST) /* wraparound occured */\ X r = rehash(r);\ X next[r & WINMASK] = hashtab[p];\ X hashtab[p] = r;\ X} X X/* This routine inputs the char from stdin and does some other X actions depending of this char's presence. X*/ X X#define Next_Char(N,F)\ X{\ X if ((c = getchar()) != EOF) {\ X text_buf[s] = c;\ X if (s < F - 1)\ X text_buf[s + N] = c;\ X s = (s + 1) & (N - 1);\ X in_count++;\ X } else\ X len--;\ X r++;\ X InsertNode();\ X} X X#if defined(__GNUC__) X#if defined(__i386__) X/* Optimizer cannot allocate these registers correctly :( (v1.39) */ X#define FIX_SI asm("si") X#define FIX_DI asm("di") X#else X X/* GNU-style register allocations for other processors are welcome! */ X X#define FIX_SI X#define FIX_DI X#endif X#else X X/* Dummy defines for non-GNU compilers */ X X#define FIX_SI X#define FIX_DI X#endif X X/* some heuristic to avoid necessity of "-ggg..." */ X#define CHAIN_THRESHOLD (LOOKAHEAD >> (BITS - 12)) X Xextern int get_next_match(); Xextern hash_t rehash(); X X#ifdef GATHER_STAT Xextern long node_matches, node_compares, node_prolongations; X#endif X X#ifdef ALLOW_MISALIGN X#define FETCH(array,index) *(us_t*)(&array[index]-1) X#else X#define FETCH(array,index) array[index] X#endif END_OF_FILE if test 2916 -ne `wc -c <'lz.h'`; then echo shar: \"'lz.h'\" unpacked with wrong size! fi # end of 'lz.h' fi if test -f 'statist.1' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'statist.1'\" else echo shar: Extracting \"'statist.1'\" $3300 characters$ sed "s/^X//" >'statist.1' <<'END_OF_FILE' X.PU X.TH STATIST 1 local X.SH NAME Xstatist \- calculate Huffman distribution for X.IR freeze "(1)" X.SH SYNOPSIS X.ll +8 X.B statist X[ X.B \-gx... X] X.ll -8 X.br X.SH DESCRIPTION XThe default table is tuned for both C texts and executable files (as in XLHARC). If you will freeze any other files (natural language texts, Xdatabases, images, fonts, etc.) you can calculate the matching Xpositions distribution using the X.B "`statist'" Xprogram, which calculates and displays the mentioned Xdistribution for the given file. It is useful for large (100K or more) Xfiles. X XThough the built-in position table is polyvalent, the tuning can increase Xthe compression rate up to one additional percent. (Observed mainly on Xtext files.) X.SH USAGE X.br X.B statist [\-g...] < sample_file X.in +8 Xor X.in -8 X.B gensample | statist [\-g...] X.br Xwhere X.B "`gensample'" Xis a program generating some sample stream of Xbytes similar to files to be frozen. X.PP XThe X.B \-g Xand X.B \-x Xswitches have the same meaning as for X.IR freeze "(1)" Xand may be repeated. X.PP XYou can also see the intermediate values Xand watch their changes by pressing INTR key when you wish. X.PP XNote: If you use X.B "gensample | statist" X, remember that INTR influence BOTH Xprocesses !! X.br XThe results have the following format: X.br X.I "n1 n2 n3 n4 n5 n6 n7 n8" X(uncertainty = X.I x) X.br XAverage match length: X.I xx.yy X.br XPercentile 99.9: X.I p999 X.br XPercentile 99.5: X.I p995 X.br XPercentile 99.0: X.I p990 X.br XPercentile 97.0: X.I p970 X.br XPercentile 95.0: X.I p950 X.br XPercentile 90.0: X.I p900 X.br XPercentile 80.0: X.I p800 X.br XPercentile 70.0: X.I p700 X.br XPercentile 50.0: X.I p500 X.br XSigma: X.I xx.yy X.br X.PP XHere X.I n1 \- n8 Xare values of the calculated position table elements, Xuncertainty is a number which denotes validity of given results X(non-zero values of uncertainty indicate that the Xresults may be unusable). Other values (average match length, Xpercentiles and sigma) are FYI only. X.PP XYou may create the X.IR /etc/default/freeze Xfile (if you don't like X.IR /etc/default/ Xdirectory, choose another - in MS-DOS it is FREEZE.CNF in Xthe directory of FREEZE.EXE), which has the following format: X.in +8 X.I name X= X.I "n1 n2 n3 n4 n5 n6 n7 n8" X.in -8 X.I (name Xmust start in column 1). For example: X.ll +8 X.br X---------- cut here ----------- X.br X# This is freeze's defaults file X.br Xrussian=0 0 1 2 6 20 31 2 # The sample was mailx.lp (Russian) X.br Xenglish=0 0 1 2 7 16 36 0 # The sample was gcc.lp (English) X.br X# End of file X.br X---------- cut here ----------- X.ll -8 X.PP XIf you find values, which are better THAN DEFAULT both for text (C Xprograms) and binary (executable) files, please send them to me. X XImportant note: statist.c is NOT a part of freeze package, it is an Xaditional feature. X X.SH "SEE ALSO" Xfreeze(1), melt(1), fcat(1) X.SH "DIAGNOSTICS" XHuffman tree has more than 8 levels, reducing... X.in +8 XSelf-explanatory, but sometimes reducing falls into infinite loop. X.in -8 X.IR xxx K X.in +8 XProgress indicator is written after each 4K of a file processed. X.in -8 X.SH "BUGS" XSometimes use of the results with uncertainty = 1 (on a file) Xgives compression rate worse than default but use of the results Xwith uncertainty = 13 (on other file) works quite good. X.PP XFound bugs descriptions, incompatibilities, etc. please send to Xleo@s514.ipmce.su. X END_OF_FILE if test 3300 -ne `wc -c <'statist.1'`; then echo shar: \"'statist.1'\" unpacked with wrong size! fi # end of 'statist.1' fi if test -f 'statist.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'statist.c'\" else echo shar: Extracting \"'statist.c'\" $6376 characters$ sed "s/^X//" >'statist.c' <<'END_OF_FILE' X#include "freeze.h" X#include "lz.h" X X/* This program calculates the distribution of the matched strings' Xpositions and lengths using nearly the same code as `freeze'. X*/ X X#define N_POS 62 X#define T (N_POS * 2 - 1) X#define R (T - 1) X X#define update(c) (/* fprintf(stderr, "%d\n", c), */ freq[c]++) X Xlong in_count, refers = 0; X Xlong indc_count; Xshort reduceflag = 0, greedy = 0; X Xint lens[LOOKAHEAD+1]; X Xus_t bits[9]; X Xshort prnt[T]; Xul_t freq[T]; Xshort used[T]; X Xvoid freeze(), StartHuff(); X XRETSIGTYPE giveres(); X Xint main(argc, argv) char ** argv; { X argv++; X while (argc > 1) { X if (**argv == '-') { X while (*++(*argv) == 'g' || **argv == 'x') X greedy += ((**argv == 'g') << 1) - 1; X if (**argv) X goto usage; X argc--; argv++; X } else X break; X } X usage: X if(argc != 1) { X fprintf(stderr, "Usage: statist [-gx...] < sample_file\n"); X fprintf(stderr, "Press INTR to display current values\n"); X exit(0); X } X signal(SIGINT, giveres); X X#ifdef DOS X setmode(fileno(stdin), O_BINARY); /* Oh this MS-DOS ... */ X#endif /* DOS */ X X freeze(); X giveres(); X return 0; X} X Xul_t isqrt(val) Xul_t val; X{ X ul_t result = 0; X ul_t side = 0; X ul_t left = 0; X int digit = 0; X int i; X for (i=0; i<sizeof(ul_t)*4; i++) X { X left = (left << 2) + (val >> (sizeof(ul_t) * 8 - 2)); X val <<= 2; X if (left >= side*2 + 1) X { X left -= side*2+1; X side = (side+1)*2; X result <<= 1; X result |= 1; X } X else X { X side *= 2; X result <<= 1; X } X } X return result; X} X X X/* Prints the (current) values of tunable parameters. Uncertainty is Xthe number of missequencings (algorithm assumes the probabilities Xof references decrease uniformly when distance increases). Ideally Xit should be 0, but somewhat about 5 or less denotes the given 8 values Xcould improve the compression rate when using them. X*/ X XRETSIGTYPE giveres() { X us_t c; X register int i, j, k, pr, f, average, sum; X ul_t cumul, sigma2; X short r, percent; X signal(SIGINT, giveres); X newtry: X StartHuff(N_POS); X pr = f = 0; X i = N_POS; X r = N_POS * 2 - 2; X while (i <= r) { X j = findmin(i); X k = findmin(i); X freq[i] = freq[j] + freq[k]; X prnt[j] = prnt[k] = i++; X } X X for (c = 1; c <= 6; c++) bits[c] = 0; X X printf("Non-monotonities are in: "); X X for(c = 0; c < N_POS; c++) { X j = 0; X k = c; X do j++; while ((k = prnt[k]) != r); X if (j <= 6) X bits[j]++; X if (j < pr) { X f += pr - j; X printf("%d, ", c); X X } else X pr = j; X } X if(f == 0) X printf("\b\b\b\babsent.\n"); X else X printf("\b\b.\n"); X X k = bits[1] + bits[2] + bits[3] + bits[4] + X bits[5] + bits[6]; X X k = N_POS - k; /* free variable length codes for 7 & 8 bits */ X X j = 128 * bits[1] + 64 * bits[2] + 32 * bits[3] + X 16 * bits[4] + 8 * bits[5] + 4 * bits[6]; X X j = 256 - j; /* free byte images for these codes */ X X/* Equation: X bits[7] + bits[8] = k X 2 * bits[7] + bits[8] = j X*/ X j -= k; X if (j < 0 || k < j) { X printf("Huffman tree has more than 8 levels, reducing...\n"); X for (i = 0; i < N_POS; i++) X if (!freq[i]) X freq[i] = 1; X else if (reduceflag) X freq[i] = (freq[i] + 1) / 2; X reduceflag = 1; X goto newtry; X } else { X bits[7] = j; X bits[8] = k - j; X printf("%d,%d,%d,%d,%d,%d,%d,%d (uncertainty = %d)\n", X bits[1], bits[2], bits[3], bits[4], X bits[5], bits[6], bits[7], bits[8], f); X } X sum = 0; cumul = 0; X for(i = 3; i <= LOOKAHEAD; i++) { X cumul += (ul_t) i * lens[i]; X sum += lens[i]; X } X sum || sum++; X printf("Average match length: %d.%02d\n", X average = cumul / sum, i = cumul * 100 / sum % 100); X if (i >= 50) average++; X j = sum; X percent = 0; X for (i = LOOKAHEAD; i >= 3; i--) { X static pcs[] = { 999, 995, 990, 970, 950, 900, 800, 700, 500 }; X j -= lens[i]; X newpcs: X if (j <= sum * pcs[percent] / 1000) { X printf("Percentile %d.%d: %d\n", X pcs[percent] / 10, pcs[percent] % 10, i); X if (percent == sizeof(pcs)/sizeof(int) - 1) X break; X else { X percent++; X goto newpcs; X } X } X } X for (sigma2 = 0, i = 3; i <= LOOKAHEAD; i++) X sigma2 += (ul_t)(i - average)*(i - average)*lens[i]; X sigma2 = sigma2 * 100 / sum; X j = (int)isqrt(sigma2); X printf("Sigma: %d.%1d\n", j / 10, j % 10); X printf("References: %ld\n", refers); X fflush(stdout); X} X X Xvoid freeze () X{ X register int i, len, s, c; X register hash_t r; X int match_length; X X StartHuff(0); X InitTree(); X r = MAXDIST + 1; X s = (r + LOOKAHEAD) & WINMASK; X for (len = 0; len < LOOKAHEAD && (c = getchar()) != EOF; len++) X text_buf[r + len] = c; X X in_count = len; X for (i = r - LOOKAHEAD; i < MAXDIST; i++) X text_buf[i] = ' '; X for (i = r - LOOKAHEAD; i <= r; i++) { X register uc_t *key = &text_buf[i]; X register unsigned p = hashof(key); X next[i] = hashtab[p]; X hashtab[p] = i; X } X while (len != 0) { X match_length = LOOKAHEAD + get_next_match(THRESHOLD - LOOKAHEAD, r); X if (match_length > len) X match_length = len; X if (match_length <= THRESHOLD) { X match_length = 1; X } else if (match_length >= chain_length) { X lens[match_length] ++; X update((((r - match_position) & WINMASK) - 1) >> 7); X refers ++; X } else { X register us_t orig_length, orig_position; X orig_length = match_length; X orig_position = match_position; X Next_Char(WINSIZE, LOOKAHEAD); X match_length = LOOKAHEAD + get_next_match(match_length - LOOKAHEAD, r); X if (match_length > len) X match_length = len; X if (orig_length >= match_length) { X lens[orig_length] ++; X update((((r - 1 - orig_position) & WINMASK) - 1) >> 7); X match_length = orig_length - 1; X } else { X lens[match_length] ++; X update((((r - match_position) & WINMASK) - 1) >> 7); X } X refers ++; X } X for (i = 0; i < match_length && X (c = getchar()) != EOF; i++) { X text_buf[s] = c; X if (s < LOOKAHEAD - 1) X text_buf[s + WINSIZE] = c; X s = (s + 1) & WINMASK; X r++; X InsertNode(); X } X in_count += i; X if ((in_count > indc_count)) { X fprintf(stderr, "%5dK\b\b\b\b\b\b", in_count / 1024); X fflush (stderr); X indc_count += 4096; X } X while (i++ < match_length) { X len--; X r++; X InsertNode(); X } X } X} X Xvoid StartHuff(beg) { X int i; X for (i = beg; i < N_POS * 2 - 1; i++) X freq[i] = 0; X for (i = 0; i < N_POS * 2 - 1; i++) X used[i] = prnt[i] = 0; X} X Xint findmin(range) { X long min = (1 << 30) - 1, argmin = -1, i; X for (i = 0; i < range; i++) { X if(!used[i] && freq[i] < min) X min = freq[argmin = i]; X } X used[argmin] = 1; X return argmin; X} END_OF_FILE if test 6376 -ne `wc -c <'statist.c'`; then echo shar: \"'statist.c'\" unpacked with wrong size! fi # end of 'statist.c' fi echo shar: End of archive 2 $of 3$. cp /dev/null ark2isdone MISSING="" for I in 1 2 3 ; do if test ! -f ark${I}isdone ; then MISSING="${MISSING} ${I}" fi done if test "${MISSING}" = "" ; then echo You have unpacked all 3 archives. rm -f ark[1-9]isdone else echo You still must unpack the following archives: echo " " ${MISSING} fi exit 0 exit 0 # Just in case...