ANULOWANIE POPRZEDNIEGO COMMITU
Revert "Replacing old implementation with working implementation"
This reverts commit 1e121f45e2.
Please enter the commit message for your changes. Lines starting
with '#' will be ignored, and an empty message aborts the commit.
Committer: tom <tom@lim.(none)>
On branch master
Changes to be committed:
(use "git reset HEAD <file>..." to unstage)
modified: _old/app/Makefile
deleted: _old/app/conf/Makefile
deleted: _old/app/conf/compiledic.conf
deleted: _old/app/conf/cor.conf
deleted: _old/app/conf/dgc.conf
deleted: _old/app/conf/dgp.conf
deleted: _old/app/conf/gph.conf
deleted: _old/app/conf/grp.conf
deleted: _old/app/conf/gue.conf
deleted: _old/app/conf/kor.conf
deleted: _old/app/conf/lem.conf
deleted: _old/app/conf/mar.conf
deleted: _old/app/conf/ser.conf
deleted: _old/app/conf/utt.conf
modified: _old/app/src/common/Makefile
modified: _old/app/src/compiledic/Makefile
modified: _old/app/src/compiledic/aut2fsa.cc
modified: _old/app/src/cor/Makefile
modified: _old/app/src/dgp/Makefile
new file: _old/app/src/dgp/canonize
new file: _old/app/src/dgp/dgc
modified: _old/app/src/dgp/grammar.hh
modified: _old/app/src/dgp/mgraph.hh
modified: _old/app/src/dgp/sgraph.hh
modified: _old/app/src/dgp/thesymbols.hh
new file: _old/app/src/dgp/tre
modified: _old/app/src/gue/Makefile
modified: _old/app/src/gue/guess.cc
modified: _old/app/src/kor/Makefile
modified: _old/app/src/kor/corlist.cc
modified: _old/app/src/kor/corr.cc
new file: _old/app/src/kor/corr.hh
modified: _old/app/src/kor/main.cc
modified: _old/app/src/lem/Makefile
modified: _old/app/src/lem/lem.cc
modified: _old/app/src/lib/Makefile
modified: _old/app/src/lib/auttools.cc
modified: _old/app/src/lib/symtab.cc
modified: _old/app/src/lib/tft.h
modified: _old/app/src/lib/tfti.h
modified: _old/app/src/lib/ttrans.h
modified: _old/app/src/lib/word.cc
modified: _old/app/src/lib/word.h
modified: _old/app/src/tok.c/Makefile
modified: _old/app/src/tok.c/cmdline_tok.ggo
modified: _old/app/src/tok.c/common_tok.cc
modified: _old/app/src/tok/Makefile
modified: _old/nawszelkiwypadek/tools/aut2fsa
modified: _old/nawszelkiwypadek/tools/cor_dic/makeLabels.pl
modified: _old/nawszelkiwypadek/tools/cor_dic/prep.pl
modified: _old/nawszelkiwypadek/tools/fsm2aut
modified: _old/nawszelkiwypadek/tools/gue_dic/canon.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/compile_user_dict.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/count_prefs.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/cut_prefs.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/makeLabels.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/prep.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/prep_user_dict.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/rmDup.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/stat.pl
modified: _old/nawszelkiwypadek/tools/gue_dic/stat_pre.pl
modified: _old/nawszelkiwypadek/tools/lem_dic/makeLabels.pl
modified: _old/nawszelkiwypadek/tools/lem_dic/prep.pl
modified: auto/defaults
modified: auto/options
modified: auto/output/Makefile
modified: auto/output/config_h
modified: auto/summary
modified: configure
This commit is contained in:
tom
parent
1e121f45e2
commit
a6e708f37f
@ -53,7 +53,9 @@ dirs:
|
||||
mkdir -p ${UTT_SHARE_DIR}/man/man3
|
||||
|
||||
.PHONY: components
|
||||
components:
|
||||
components:
|
||||
cd $(SRC_DIR)/lib && make; cd $(CUR_DIR)
|
||||
|
||||
@for cmp in $(COMPONENTS); do\
|
||||
cd $(SRC_DIR)/$$cmp && make && make copy; cd $(CUR_DIR); \
|
||||
done
|
||||
@ -79,7 +81,7 @@ clean_doc:
|
||||
|
||||
.PHONY: clean_dist
|
||||
clean_dist:
|
||||
rm -rf ${UTT_DIR}
|
||||
|
||||
|
||||
# ------------------------------------------------------------------
|
||||
# sekcja przygotowania paczki (programow wraz z dodatkowymi plikami)
|
||||
@ -142,6 +144,7 @@ dist_deb: build
|
||||
#dist: all
|
||||
# tar -czvf $(UTT_DIST_NAME).tgz $(UTT_DIR)
|
||||
|
||||
|
||||
#clean_dist:
|
||||
# if [ -d $(UTT_DIST_DIR) ]; then rm -r $(UTT_DIST_DIR); else true; fi
|
||||
# if [ -f $(UTT_DIST_FILE).tar.bz2 ]; then rm $(UTT_DIST_FILE).tar.bz2; else true; fi
|
||||
|
||||
@ -1,8 +0,0 @@
|
||||
main:
|
||||
|
||||
copy:
|
||||
ifdef UTT_CONF_DIR
|
||||
cp *.conf ${UTT_CONF_DIR}
|
||||
endif
|
||||
|
||||
clean:
|
||||
@ -1,13 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
symbols = PATH_PREFIX/share/utt/pl_PL.ISO-8859-2/pl_PL.ISO-8859-2.sym
|
||||
@ -1,14 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
dictionary-home = PATH_PREFIX/share/utt
|
||||
process = W
|
||||
@ -1,15 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
categories = PATH_PREFIX/share/utt/cats.dgc
|
||||
grammar = PATH_PREFIX/share/utt/gram.dgc
|
||||
outputfile = PATH_PREFIX/share/utt/gram.dgp
|
||||
@ -1,18 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
|
||||
grammar = PATH_PREFIX/share/utt/gram.dgp
|
||||
process = W
|
||||
process = BOS
|
||||
process = EOS
|
||||
info = d
|
||||
@ -1,16 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
process = W
|
||||
process = BOS
|
||||
process = EOS
|
||||
reset = BOS
|
||||
@ -1,15 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
macros = PATH_PREFIX/lib/utt/terms.m4
|
||||
tags=uam
|
||||
|
||||
@ -1,14 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
dictionary-home = PATH_PREFIX/share/utt
|
||||
process = W
|
||||
@ -1,16 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
dictionary-home = PATH_PREFIX/share/utt
|
||||
weights = PATH_PREFIX/share/utt/weights.kor
|
||||
threshold = 1.0
|
||||
process=W
|
||||
@ -1,14 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
dictionary-home = PATH_PREFIX/share/utt
|
||||
process = W
|
||||
@ -1,15 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
macros = PATH_PREFIX/lib/utt/terms.m4
|
||||
tags=uam
|
||||
|
||||
@ -1,15 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
macros = PATH_PREFIX/lib/utt/terms.m4
|
||||
flex-template = PATH_PREFIX/lib/utt/ser.l.template
|
||||
tags=uam
|
||||
@ -1,14 +0,0 @@
|
||||
# ************************************************************
|
||||
# * This file was created automatically during installation. *
|
||||
# * If you don't need do not change it. *
|
||||
# * *
|
||||
# * UAM Text Tools *
|
||||
# * Adam Mickiewicz University, Poland *
|
||||
# * http://utt.amu.edu.pl *
|
||||
# ************************************************************
|
||||
#
|
||||
# All lines must looks like:
|
||||
# parameter_name [=] value
|
||||
#
|
||||
# user locale (dictionary)
|
||||
language = pl_PL.ISO-8859-2
|
||||
@ -1,7 +1,8 @@
|
||||
COMMON_PATH=../common
|
||||
# main: cmdline.c main_template.cc
|
||||
# g++ -o main cmdline.c common.cc main_template.cc
|
||||
|
||||
include ./common.mk
|
||||
# cmdline.c cmdline.h : cmdline.ggo
|
||||
# gengetopt -i cmdline.ggo
|
||||
|
||||
clean:
|
||||
rm -f cmdline.c cmdline.h cmdline.ggo
|
||||
|
||||
# cmdline.ggo: cmdline_common.ggo cmdline_program.ggo
|
||||
# cat cmdline_common.ggo cmdline_program.ggo > cmdline.ggo
|
||||
|
||||
@ -1,13 +1,16 @@
|
||||
CFLAG1 = -Wno-deprecated -O3 -fpermissive
|
||||
CFLAG1 = -m32 -Wno-deprecated -O3 -fpermissive
|
||||
CFLAG_ST = -Wno-deprecated -O3 -fpermissive -static
|
||||
|
||||
all: compiledic aut2fsa
|
||||
|
||||
compiledic:
|
||||
|
||||
|
||||
aut2fsa: aut2fsa.cc
|
||||
#g++ -m32 -Wno-deprecated -O3 -fpermissive -static -o aut2fsa aut2fsa.cc
|
||||
g++ $(CFLAG1) -o aut2fsa aut2fsa.cc
|
||||
|
||||
|
||||
copy:
|
||||
ifdef UTT_BIN_DIR
|
||||
cp compiledic fsm2aut aut2fsa ${UTT_BIN_DIR}
|
||||
|
||||
@ -1,6 +1,6 @@
|
||||
|
||||
#include <iostream>
|
||||
#include <cstdlib>
|
||||
#include <stdlib.h>
|
||||
|
||||
#include "../lib/tfti.h"
|
||||
|
||||
|
||||
@ -1,11 +1,12 @@
|
||||
PAR=-Wno-deprecated -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -fpermissive
|
||||
PAR=-Wno-deprecated -m32 -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -m32 -fpermissive
|
||||
LIB_PATH=../lib
|
||||
COMMON_PATH=../common
|
||||
CMDLINE_FILE='"../cor/cmdline.h"'
|
||||
|
||||
|
||||
cor: main.cc corr.o word.o auttools.o cmdline.c common_cor.o common.o
|
||||
cor: main.cc corr.o $(LIB_PATH)/word.o \
|
||||
$(LIB_PATH)/auttools.o cmdline.c common_cor.o common.o
|
||||
g++ $(PAR) -D _CMDLINE_FILE=$(CMDLINE_FILE) main.cc corr.o common.o \
|
||||
$(LIB_PATH)/word.o $(LIB_PATH)/auttools.o cmdline.c common_cor.o \
|
||||
-o cor
|
||||
@ -13,7 +14,8 @@ cor: main.cc corr.o word.o auttools.o cmdline.c common_cor.o common.o
|
||||
corr.o: corr.cc corr.hh
|
||||
g++ $(PAR2) corr.cc
|
||||
|
||||
common.o: cmdline_common
|
||||
common.o: $(COMMON_PATH)/cmdline_common.ggo $(COMMON_PATH)/common.cc \
|
||||
$(COMMON_PATH)/common.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) $(COMMON_PATH)/common.cc
|
||||
|
||||
common_cor.o: cmdline.h common_cor.cc common_cor.h
|
||||
@ -37,5 +39,3 @@ clean: clean.cmdline
|
||||
clean.cmdline:
|
||||
rm cmdline.* || true
|
||||
|
||||
include $(COMMON_PATH)/common.mk
|
||||
include $(LIB_PATH)/lib.mk
|
||||
|
||||
@ -1,3 +1,5 @@
|
||||
|
||||
|
||||
SHELL = /bin/sh
|
||||
LIB_PATH=../../lib
|
||||
COMMON_PATH=../common
|
||||
@ -6,12 +8,12 @@ CMDLINE_FILE='"../dgp/cmdline.h"'
|
||||
|
||||
#vpath %.o .
|
||||
|
||||
CXXFLAGS = -O2 -static -fpermissive
|
||||
CXXFLAGS = -O2 -static
|
||||
|
||||
sources = main.cc grammar.cc symbol.cc mgraph.cc sgraph.cc dgp0.cc cmdline.cc \
|
||||
$(COMMON_PATH)/common.cc global.cc
|
||||
|
||||
bin = dgp dgc canonize tre
|
||||
bin = dgp
|
||||
|
||||
# plik *.o sa umieszczane w podkatalogu o
|
||||
objs = $(sources:%.cc=%.o)
|
||||
@ -42,7 +44,8 @@ cmdline.ggo: cmdline_dgp.ggo ../common/cmdline_common.ggo
|
||||
|
||||
|
||||
clean:
|
||||
rm ${bin} ${objs} cmdline.cc cmdline.h *.d
|
||||
rm ${bin} ${objs} cmdline.cc cmdline.h
|
||||
rm -rf *.d
|
||||
|
||||
prof: dgp
|
||||
gprof dgp ~/tmp/dgp-pl/gmon.out > dgp.prof
|
||||
|
||||
50
_old/app/src/dgp/canonize
Executable file
50
_old/app/src/dgp/canonize
Executable file
@ -0,0 +1,50 @@
|
||||
#!/usr/bin/perl
|
||||
|
||||
#package: UAM TExt Tools
|
||||
#component: canonize
|
||||
#version: 1.0
|
||||
#author: Tomasz Obrebski
|
||||
|
||||
use lib "/usr/local/lib/utt";
|
||||
use lib "$ENV{'HOME'}/.local/lib/utt";
|
||||
|
||||
use strict;
|
||||
use Getopt::Long;
|
||||
use attr;
|
||||
|
||||
|
||||
my $help;
|
||||
|
||||
GetOptions("help|h" => \$help);
|
||||
|
||||
if($help)
|
||||
{
|
||||
print <<'END'
|
||||
|
||||
Transforms syntactic categories to their canonical form.
|
||||
|
||||
Usage: canonize
|
||||
|
||||
Options:
|
||||
--help -h Help.
|
||||
|
||||
END
|
||||
;
|
||||
exit 0;
|
||||
}
|
||||
|
||||
#$|=1;
|
||||
|
||||
my %tra;
|
||||
|
||||
while(<>)
|
||||
{
|
||||
s/$attr::pos_re\/$attr::avlist_re/trans($&)/ge;
|
||||
print;
|
||||
}
|
||||
|
||||
sub trans
|
||||
{
|
||||
my $cat=shift;
|
||||
exists($tra{$cat}) ? $tra{$cat} : ( $tra{$cat} = attr::canonize $cat );
|
||||
}
|
||||
292
_old/app/src/dgp/dgc
Executable file
292
_old/app/src/dgp/dgc
Executable file
@ -0,0 +1,292 @@
|
||||
#!/usr/bin/perl
|
||||
|
||||
#package: UAM Text Tools
|
||||
#component: dgc (dg compiler)
|
||||
#version: 1.0
|
||||
#author: Tomasz Obrebski
|
||||
|
||||
# wymaga niejawnie programu canonize!!!!
|
||||
use lib "/usr/local/lib/utt";
|
||||
use lib "$ENV{'HOME'}/.local/lib/utt";
|
||||
|
||||
use strict;
|
||||
use Getopt::Long;
|
||||
use Data::Dumper;
|
||||
use attr;
|
||||
use File::HomeDir;
|
||||
|
||||
my $systemconfigfile='/usr/local/etc/utt/dgc.conf';
|
||||
my $userconfigfile=home()."/.utt/dgc.conf";
|
||||
|
||||
Getopt::Long::Configure('no_ignore_case_always');
|
||||
|
||||
my $help=0;
|
||||
my $catfile=0;
|
||||
my $dicfile=0;
|
||||
my $gramfile=0;
|
||||
my $outputfile=0;
|
||||
|
||||
#read configuration files###########################
|
||||
my $file;
|
||||
foreach $file ($systemconfigfile, $userconfigfile){
|
||||
if(open(CONFIG, $file)){
|
||||
while (<CONFIG>) {
|
||||
chomp;
|
||||
s/#.*//;
|
||||
s/^\s+//;
|
||||
s/\s+$//;
|
||||
next unless length;
|
||||
my ($name, $value) = split(/\s*=\s*/, $_, 2);
|
||||
if(($name eq "categories")or($name eq "c")){
|
||||
$catfile=$value;
|
||||
}
|
||||
elsif(($name eq "dictionary")or($name eq "d")){
|
||||
$dicfile=$value;
|
||||
}
|
||||
elsif(($name eq "grammar")or($name eq "g")){
|
||||
$gramfile=$value;
|
||||
}
|
||||
elsif(($name eq "outputfile")or($name eq "o")){
|
||||
$outputfile=$value;
|
||||
}
|
||||
elsif(($name eq "help")or($name eq "h")){
|
||||
$help=1;
|
||||
}
|
||||
|
||||
}
|
||||
close CONFIG;
|
||||
}
|
||||
}
|
||||
#########################################################
|
||||
|
||||
GetOptions("help|h" => \$help,
|
||||
"categories|c=s" => \$catfile,
|
||||
"dictionary|d=s" => \$dicfile,
|
||||
"grammar|g=s" => \$gramfile,
|
||||
"outputfile|o=s" => \$outputfile);
|
||||
|
||||
my $homedir = $ENV{'HOME'};
|
||||
$catfile =~ s/~/$homedir/;
|
||||
$dicfile =~ s/~/$homedir/;
|
||||
$gramfile =~ s/~/$homedir/;
|
||||
$outputfile =~ s/~/$homedir/;
|
||||
|
||||
|
||||
if($help)
|
||||
{
|
||||
print <<'END'
|
||||
Usage: dgc [OPTIONS]
|
||||
|
||||
Options:
|
||||
--categories -c filename List of syntactic categories.
|
||||
--dictionary -d filename Dictionary.
|
||||
--grammar -g filename List of grammar rules.
|
||||
--outputfile -o filename Output file name.
|
||||
--help -h Help.
|
||||
END
|
||||
;
|
||||
exit 0;
|
||||
}
|
||||
|
||||
die("At least one of --cats and --dic must be given.\n") if !$catfile && !$dicfile;
|
||||
|
||||
my $ncat=0;
|
||||
my $nrole=0;
|
||||
my $nsgl=0;
|
||||
my $nleft=0;
|
||||
my $nright=0;
|
||||
my $nreq=0;
|
||||
my $nlink=0;
|
||||
my $nflag=0;
|
||||
|
||||
my %cats;
|
||||
my %roles;
|
||||
my %agr;
|
||||
my %gov;
|
||||
|
||||
if(!$outputfile) {
|
||||
*OUTPUT = *STDOUT;
|
||||
}
|
||||
elsif($outputfile eq "-") {
|
||||
*OUTPUT = *STDOUT;
|
||||
}
|
||||
else {
|
||||
open(OUTPUT, ">$outputfile") or die("Can't open output file: $outputfile!");
|
||||
}
|
||||
|
||||
|
||||
loadcats($catfile) if $catfile;
|
||||
extractcats($dicfile) if $dicfile;
|
||||
|
||||
|
||||
my $cats_re = qr/(?:$attr::cat_re\s*(?:,\s*$attr::cat_re)*)/;
|
||||
|
||||
# class parse_class:
|
||||
# /$attr::cat_re/g;
|
||||
|
||||
|
||||
if(!$gramfile) {
|
||||
*INPUT = *STDIN;
|
||||
}
|
||||
elsif($gramfile eq "-"){
|
||||
*INPUT = *STDIN;
|
||||
}
|
||||
else {
|
||||
open(INPUT, $gramfile) or die("Unable to open: $gramfile!");
|
||||
}
|
||||
|
||||
while(<INPUT>)
|
||||
{
|
||||
s/#.*//;
|
||||
s/^\s+//;
|
||||
s/\s+$//;
|
||||
if(/^AGR\s+(\S+)\s+(\S+)$/)
|
||||
{
|
||||
push @{$agr{$1}}, $2;
|
||||
}
|
||||
elsif(/^GOV\s+(\S+)\s+(\S+)$/)
|
||||
{
|
||||
push @{$gov{$1}}, attr::parse($2);
|
||||
}
|
||||
elsif(/^ROLE\s+\S+$/)
|
||||
{
|
||||
$roles{$_}=1;
|
||||
print OUTPUT "$_\n";
|
||||
}
|
||||
elsif(/^SGL\s+\S+$/)
|
||||
{
|
||||
++$nsgl;
|
||||
print OUTPUT "$_\n";
|
||||
}
|
||||
elsif(/^REQ\s+(\S+)\s+(\S+)$/)
|
||||
{
|
||||
print OUTPUT "#$_\n";
|
||||
my $cat = attr::parse $1;
|
||||
for my $atomcat (keys %cats)
|
||||
{
|
||||
if(attr::match @$cat, @{$cats{$atomcat}})
|
||||
{
|
||||
print OUTPUT "REQ ".$atomcat." $2\n";
|
||||
++$nreq;
|
||||
}
|
||||
}
|
||||
}
|
||||
elsif(/^LEFT\s+\S+$/)
|
||||
{
|
||||
++$nleft;
|
||||
print OUTPUT "$_\n";
|
||||
}
|
||||
elsif(/^RIGHT\s+\S+$/)
|
||||
{
|
||||
++$nright;
|
||||
print OUTPUT "$_\n";
|
||||
}
|
||||
elsif(my ($hs,$ds,$r) = /^LINK\s+($cats_re)\s+($cats_re)\s+(\S+)$/)
|
||||
{
|
||||
print OUTPUT "#$_\n";
|
||||
for my $h ($hs =~ /$attr::cat_re/g)
|
||||
{
|
||||
for my $d ($ds =~ /$attr::cat_re/g)
|
||||
{
|
||||
addlinks($h,$d,$r);
|
||||
}
|
||||
}
|
||||
}
|
||||
elsif(/^FLAG\s+\S+$/)
|
||||
{
|
||||
++$nflag;
|
||||
print OUTPUT "$_\n"
|
||||
}
|
||||
elsif(/^$/) {
|
||||
# pomijamy puste linie oraz komentarze
|
||||
}
|
||||
else
|
||||
{
|
||||
print STDERR "Illegal format: $_\n";
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
sub addlinks
|
||||
{
|
||||
my ($h,$d,$r) = @_;
|
||||
|
||||
for my $a (@{$agr{$r}}) { print OUTPUT "#AGR $r $a\n"; }
|
||||
for my $c (@{$gov{$r}}) { print OUTPUT "#GOV $r ".attr::unparse(@$c)."\n"; }
|
||||
my $head = attr::parse $h;
|
||||
my $dep = attr::parse $d;
|
||||
|
||||
for my $atomhead (keys %cats)
|
||||
{
|
||||
if(attr::match @$head, @{$cats{$atomhead}})
|
||||
{
|
||||
DEP:
|
||||
for my $atomdep (keys %cats)
|
||||
{
|
||||
next DEP if ! attr::match @$dep, @{$cats{$atomdep}};
|
||||
|
||||
for my $a (@{$agr{$r}})
|
||||
{
|
||||
next DEP if ! attr::agree(@{$cats{$atomhead}},@{$cats{$atomdep}},$a);
|
||||
}
|
||||
|
||||
for my $c (@{$gov{$r}})
|
||||
{
|
||||
next DEP if ! attr::match(@$c,@{$cats{$atomdep}});
|
||||
}
|
||||
|
||||
print OUTPUT "LINK ";
|
||||
print OUTPUT $atomhead." ";
|
||||
print OUTPUT $atomdep." $r\n";
|
||||
++$nlink;
|
||||
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
printf STDERR "%6d CAT statements\n", 0+keys(%cats);
|
||||
printf STDERR "%6d ROLE statements\n", 0+keys(%roles);
|
||||
printf STDERR "%6d SGL statements\n", $nsgl;
|
||||
printf STDERR "%6d REQ statements\n", $nreq;
|
||||
printf STDERR "%6d LEFT statements\n", $nleft;
|
||||
printf STDERR "%6d RIGHT statements\n", $nright;
|
||||
printf STDERR "%6d LINK statements\n", $nlink;
|
||||
printf STDERR "%6d FLAG statements\n", $nflag;
|
||||
|
||||
|
||||
sub extractcats
|
||||
{
|
||||
my $file = shift;
|
||||
open DICFILE, "canonize $file |";
|
||||
while(<DICFILE>)
|
||||
{
|
||||
while(/,([^[:space:];]+)/g)
|
||||
{
|
||||
my $cat=$1;
|
||||
next if !$cat || exists $cats{$cat};
|
||||
$ncat++;
|
||||
print OUTPUT "CAT $1\n";
|
||||
$cats{$cat}=attr::parse($cat);
|
||||
}
|
||||
}
|
||||
close DICFILE;
|
||||
}
|
||||
|
||||
|
||||
sub loadcats
|
||||
{
|
||||
my $file = shift;
|
||||
open CATFILE, "canonize $file |";
|
||||
while(<CATFILE>)
|
||||
{
|
||||
tr/ \t\n//d;
|
||||
next if !$_ || exists $cats{$_};
|
||||
print OUTPUT "CAT $_\n";
|
||||
++$ncat;
|
||||
$cats{$_}=attr::parse($_);
|
||||
}
|
||||
close CATFILE;
|
||||
}
|
||||
|
||||
@ -11,6 +11,8 @@
|
||||
#include "sgraph.hh"
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
class Link
|
||||
{
|
||||
Role role;
|
||||
|
||||
@ -7,6 +7,9 @@
|
||||
#include "thesymbols.hh"
|
||||
#include "../common/common.h"
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
class MNode
|
||||
{
|
||||
public:
|
||||
|
||||
@ -11,6 +11,8 @@
|
||||
#include "thesymbols.hh"
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
class MNode;
|
||||
|
||||
|
||||
|
||||
@ -8,6 +8,9 @@
|
||||
#include <set>
|
||||
#include <bitset>
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
typedef Symbol<1> Cat;
|
||||
|
||||
typedef Symbol<2> Role;
|
||||
|
||||
304
_old/app/src/dgp/tre
Executable file
304
_old/app/src/dgp/tre
Executable file
@ -0,0 +1,304 @@
|
||||
#!/usr/bin/ruby -I /usr/local/lib/utt -I $HOME/.local/lib/utt
|
||||
|
||||
$: << "#{ENV['HOME']}/.local/lib/utt"
|
||||
$: << "/usr/local/lib/utt"
|
||||
|
||||
require 'getoptlong'
|
||||
require 'seg.rb'
|
||||
|
||||
opts = GetoptLong.new(
|
||||
[ '--help', '-h', GetoptLong::NO_ARGUMENT ],
|
||||
[ '--debug', '-d', GetoptLong::NO_ARGUMENT ],
|
||||
[ '--format', '-F', GetoptLong::REQUIRED_ARGUMENT ],
|
||||
[ '--info', '-I', GetoptLong::REQUIRED_ARGUMENT ],
|
||||
[ '--only-trees','-t', GetoptLong::NO_ARGUMENT ])
|
||||
|
||||
$helptext=
|
||||
"The program generates trees from the graph output by dgp. dgp must\n"+
|
||||
"must be run with '-i ds' option.\n\n"+
|
||||
"Command: tre [options]\n\n"+
|
||||
"Options:\n"+
|
||||
"--help -h Print help (this text) and exit.\n"+
|
||||
"--debug -d Verbose output. For developers only.\n"+
|
||||
"--format=s -F s Output format. Recognized values:\n"+
|
||||
" a root + list of arcs\n"+
|
||||
" p parenthesized notation\n"+
|
||||
" h human readable indented tree format\n"+
|
||||
" Multiple values are allowed. (default p)\n"+
|
||||
"--info=s -I s Information printed. Recognized values:\n"+
|
||||
" n node identifier\n"+
|
||||
" f surface form\n"+
|
||||
" m morphological information\n"+
|
||||
" l arc labels\n"+
|
||||
"--only-trees -t Do not copy input. Print trees only.\n"
|
||||
|
||||
$DEBUG=false
|
||||
$FORMAT='p'
|
||||
$INFO='DEFAULT'
|
||||
$ONLYTREES=false
|
||||
|
||||
opts.each do |opt, arg|
|
||||
case opt
|
||||
when '--help'
|
||||
print $helptext
|
||||
exit 0
|
||||
when '--debug'
|
||||
$DEBUG=true
|
||||
when '--format'
|
||||
$FORMAT=arg
|
||||
when '--info'
|
||||
$INFO=arg
|
||||
when '--only-trees'
|
||||
$ONLYTREES=true
|
||||
else
|
||||
print "Unknown option #{opt}. Ignored.\n"
|
||||
end
|
||||
end
|
||||
|
||||
if $INFO=='DEFAULT'
|
||||
case $FORMAT
|
||||
when 'p','a'
|
||||
$INFO='nl'
|
||||
when 'h'
|
||||
$INFO='fmnl'
|
||||
end
|
||||
end
|
||||
|
||||
$dgpsep=';'
|
||||
|
||||
def tre(input)
|
||||
$gphid=[]
|
||||
$form=[]
|
||||
$lem=[]
|
||||
nodes=[]
|
||||
count=0
|
||||
seg=Seg.new
|
||||
for line in input
|
||||
print line unless $ONLYTREES
|
||||
seg.set(line)
|
||||
if dgp=seg['dgp']
|
||||
if nodes==[] && seg[3]!='BOS'
|
||||
print "A sentence must start with BOS segment. Aborting.\n"
|
||||
return
|
||||
end
|
||||
|
||||
id=dgp[/^\d+/].to_i
|
||||
|
||||
if gph=seg['gph']
|
||||
$gphid[id]=gph[/^\d+/].to_i
|
||||
else
|
||||
print "No gph field. Aborting.\n"
|
||||
return
|
||||
end
|
||||
|
||||
$form[$gphid[id]]=seg[4]
|
||||
$lem[$gphid[id]]=seg['lem']
|
||||
|
||||
nodes[id] = [seg[1].to_i,dgp]
|
||||
|
||||
if seg[3]=='EOS'
|
||||
$pref = "#{seg[1]} #{seg[2]} SYN *"
|
||||
parsegraph(nodes)
|
||||
printgraph if $DEBUG
|
||||
$thetrees=[]
|
||||
gentrees2
|
||||
for t in $thetrees
|
||||
count += 1
|
||||
t1=ground(t)
|
||||
case $FORMAT
|
||||
when /a/
|
||||
print "#{$pref} tre:#{count} arc:"
|
||||
printarcs(t1[0],t1[1])
|
||||
print "\n"
|
||||
when /p/
|
||||
print "#{$pref} tre:#{count} par:"
|
||||
printpar(t1[0],t1[1])
|
||||
print "\n"
|
||||
when /h/
|
||||
print "#\n# tree #{count}\n# ------\n"
|
||||
printtree(t1[0],t1[1],0)
|
||||
end
|
||||
end
|
||||
nodes=[]
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
|
||||
def nodeinfo(id)
|
||||
info=""
|
||||
if $INFO =~ /n/
|
||||
info += id.to_s
|
||||
info += '.' if $INFO =~ /[fm]/
|
||||
end
|
||||
if $INFO =~ /f/
|
||||
info += $form[id]
|
||||
info += ';' if $INFO =~ /m/
|
||||
end
|
||||
if $INFO =~ /m/
|
||||
info += $lem[id]
|
||||
end
|
||||
info
|
||||
end
|
||||
|
||||
|
||||
def printarcs(root,arcs)
|
||||
print nodeinfo(root)
|
||||
for a in arcs
|
||||
print ';'
|
||||
print "#{a[2]}:" if $INFO =~ /l/
|
||||
print nodeinfo(a[0])+'-'+nodeinfo(a[1])
|
||||
end
|
||||
end
|
||||
|
||||
def printtree(root,arcs,o)
|
||||
if o==0
|
||||
print "# %-16s" % "root: "
|
||||
end
|
||||
print nodeinfo(root),"\n"
|
||||
for arc in arcs.select{ |a| a[0]==root }.sort{|a,b| a[1]<=>b[1] }
|
||||
print '# '," "*(o+1)
|
||||
print "%-16s" % (arc[2]+": ")
|
||||
printtree(arc[1],arcs,o+1)
|
||||
end
|
||||
end
|
||||
|
||||
def printpar(root,arcs)
|
||||
print nodeinfo(root)
|
||||
deps = arcs.select{ |a| a[0]==root }.sort{|a,b| a[1]<=>b[1] }
|
||||
unless deps == []
|
||||
print '('
|
||||
cont=false
|
||||
for arc in deps
|
||||
if cont then print ',' else cont=true end
|
||||
print arc[2],':' if $INFO =~ /l/
|
||||
printpar(arc[1],arcs)
|
||||
end
|
||||
print ')'
|
||||
end
|
||||
end
|
||||
|
||||
|
||||
def parsegraph(nodes)
|
||||
|
||||
$n =nodes.length
|
||||
$sat =[];
|
||||
|
||||
$vis =[];
|
||||
$succ=[];
|
||||
$lhs =[];
|
||||
$arcs=[];
|
||||
$pos=[]
|
||||
|
||||
for dgp in nodes
|
||||
|
||||
parts = dgp[1].split($dgpsep,6)
|
||||
|
||||
if parts[3]==nil || parts[4]==nil || parts[5]==nil
|
||||
$stderr.print "ERR: tre requires dgp be called with '--info s' option. Aborting.\n"
|
||||
exit
|
||||
end
|
||||
|
||||
i = parts[0].to_i
|
||||
$pos[i] = dgp[0].to_i
|
||||
$sat << i if parts[1]=="s"
|
||||
$arcs |= parts[2].split(',').map{ |a| case a
|
||||
when /\-\-(\w+)-(\d+)\/(\d+)/
|
||||
[i, $2.to_i, $1, $3.to_i]
|
||||
when /\+\+(\d+)-(\w+)\/(\d+)/
|
||||
[$1.to_i, i, $2, $3.to_i]
|
||||
end }
|
||||
$succ |= parts[3][1..-2].split(',').map{|x| [x.to_i,i]}
|
||||
$vis |= parts[4][1..-2].split(',').map{|x| [x.to_i,i]}
|
||||
$lhs |= parts[5][1..-2].split(',').map{|x| [x.to_i,i]} + [[i,i]]
|
||||
|
||||
end
|
||||
end
|
||||
|
||||
|
||||
def ground(t)
|
||||
[ $gphid[t[0]] , t[1].map{|a| [$gphid[a[0]],$gphid[a[1]],a[2]]} ]
|
||||
end
|
||||
|
||||
|
||||
def gentrees2()
|
||||
$thetrees=[];
|
||||
bos=0; eos=$n-1;
|
||||
roots = (1...eos).select{|i| $vis.include? [i,eos]}.select{|i| $vis.include? [bos,i]}
|
||||
if $DEBUG then print "ROOTS: #{roots.inspect}\n" end
|
||||
for i in roots
|
||||
$theroot=i
|
||||
for r in buildR(i , eos, [])
|
||||
(rmin,rmax,rtree) = r
|
||||
buildR(bos, rmin, rtree)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
|
||||
def buildR(min, max, tree)
|
||||
if $DEBUG then print "buildR--#{min}--#{max}--#{tree.inspect}\n" end
|
||||
trees=[]
|
||||
for a in $arcs.select{|a| a[0]==max && $vis.include?([min,a[1]]) }
|
||||
if $DEBUG then print "ARC: #{a.inspect}\n" end
|
||||
for r in buildR(a[1],a[3],tree+[a])
|
||||
(rmin,rmax,rarcs) = r
|
||||
for l in buildR(min,rmin,rarcs)
|
||||
(lmin,lmax,larcs) = l
|
||||
trees << [lmin,rmax,larcs]
|
||||
end
|
||||
end
|
||||
end
|
||||
for i in (0...$n).select{|i| $succ.include?([i,max])}.select{|i| $lhs.include?([min,i])}
|
||||
for l in buildL(min,i,tree)
|
||||
(lmin,lmax,larcs) = l
|
||||
trees << [lmin,lmax,larcs]
|
||||
end
|
||||
end
|
||||
trees
|
||||
end
|
||||
|
||||
|
||||
def buildL(min,max,tree)
|
||||
if $DEBUG then print "buildL--#{min}--#{max}--#{tree.inspect}\n" end
|
||||
if $pos[min]==$pos[max]
|
||||
if min==0 && max==0
|
||||
$thetrees.push [$theroot,tree]
|
||||
if $DEBUG then print "adding tree: #{tree.inspect}\n" end
|
||||
end
|
||||
return [[max,max,tree]]
|
||||
end
|
||||
trees=[]
|
||||
for arc in $arcs.select{|a| a[1]==max && $lhs.include?([min,a[0]]) }
|
||||
if $DEBUG then print "ARC: #{arc.inspect}\n" end
|
||||
for r in buildR(arc[3],max,tree+[arc])
|
||||
(rmin,rmax,rarcs) = r
|
||||
for l in buildL(min,rmin,rarcs)
|
||||
(lmin,lmax,larcs) = l
|
||||
trees << [lmin,lmax,larcs]
|
||||
end
|
||||
end
|
||||
end
|
||||
trees
|
||||
end
|
||||
|
||||
|
||||
def printgraph()
|
||||
|
||||
print "N: #{$n}\n"
|
||||
print "SAT: #{set_to_s($sat)}\n"
|
||||
print "SUCC: #{rel_to_s($succ)}\n"
|
||||
print "VIS: #{rel_to_s($vis)}\n"
|
||||
print "LHS: #{rel_to_s($lhs)}\n"
|
||||
print "ARCS: #{arcs_to_s($arcs)}\n"
|
||||
end
|
||||
|
||||
def set_to_s(s) "{#{s.join(',')}}" end
|
||||
def rel_to_s(r) "{#{r.map{|p| "(#{p[0]},#{p[1]})"}.join(',')}}" end
|
||||
def arc_to_s(q) "-#{q[0]}-#{q[2]}-#{q[1]}/#{q[3]}" end
|
||||
def arcs_to_s(a) "{#{a.map{|q| arc_to_s(q)}.join(',')}}" end
|
||||
|
||||
######################################################################
|
||||
|
||||
tre($stdin)
|
||||
@ -1,11 +1,12 @@
|
||||
PAR=-Wno-deprecated -O2 -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -O2 -fpermissive
|
||||
PAR=-Wno-deprecated -O3 -fpermissive -m32 -static
|
||||
PAR2=-c -Wno-deprecated -O3 -fpermissive -m32
|
||||
LIB_PATH=../lib
|
||||
COMMON_PATH=../common
|
||||
CMDLINE_FILE='"../gue/cmdline.h"'
|
||||
|
||||
|
||||
gue: main.cc guess.o auttools.o word.o cmdline.c common_guess.o common.o
|
||||
gue: main.cc guess.o $(LIB_PATH)/auttools.o $(LIB_PATH)/word.o \
|
||||
cmdline.c common_guess.o common.o
|
||||
g++ $(PAR) main.cc guess.o \
|
||||
$(LIB_PATH)/auttools.o $(LIB_PATH)/word.o cmdline.c common.o common_guess.o \
|
||||
-o gue
|
||||
@ -16,7 +17,8 @@ guess.o: guess.h guess.cc
|
||||
common_guess.o: cmdline.h common_guess.cc common_guess.h
|
||||
g++ $(PAR2) common_guess.cc
|
||||
|
||||
common.o: $(COMMON_PATH)/cmdline_common.ggo cmdline_common
|
||||
common.o: $(COMMON_PATH)/cmdline_common.ggo $(COMMON_PATH)/common.cc \
|
||||
$(COMMON_PATH)/common.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) $(COMMON_PATH)/common.cc
|
||||
|
||||
cmdline.c cmdline.h: cmdline.ggo
|
||||
@ -38,6 +40,3 @@ copy:
|
||||
ifdef UTT_BIN_DIR
|
||||
cp gue ${UTT_BIN_DIR}
|
||||
endif
|
||||
|
||||
include $(LIB_PATH)/lib.mk
|
||||
include $(COMMON_PATH)/common.mk
|
||||
|
||||
@ -1,11 +1,11 @@
|
||||
|
||||
#include "guess.h"
|
||||
|
||||
#include <cstring>
|
||||
#include <string.h>
|
||||
#include <iostream>
|
||||
#include <cstdlib>
|
||||
#include <cassert>
|
||||
#include <ctime>
|
||||
#include <stdlib.h>
|
||||
#include <assert.h>
|
||||
#include <time.h>
|
||||
|
||||
#define DICT 1
|
||||
#define COR 2
|
||||
@ -17,8 +17,10 @@
|
||||
|
||||
#define PREF_SIGN '_'
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
|
||||
Guess::Guess(const char* suf_file)
|
||||
: _suf(suf_file) {
|
||||
/* _suf = NULL;
|
||||
|
||||
@ -1,22 +1,26 @@
|
||||
PAR=-Wno-deprecated -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -fpermissive
|
||||
PAR=-Wno-deprecated -m32 -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -m32 -fpermissive
|
||||
LIB_PATH=../lib
|
||||
COMMON_PATH=../common
|
||||
CMDLINE_FILE='"../kor/cmdline.h"'
|
||||
|
||||
|
||||
kor: main.cc corr.o corlist.o cmdline.o word.o auttools.o cmdline.c common_cor.o common.o
|
||||
kor: main.cc corr.o corlist.o cmdline.o $(LIB_PATH)/word.o \
|
||||
$(LIB_PATH)/auttools.o cmdline.c common_cor.o common.o
|
||||
g++ $(PAR) -D _CMDLINE_FILE=$(CMDLINE_FILE) main.cc corlist.o corr.o common.o \
|
||||
$(LIB_PATH)/word.o $(LIB_PATH)/auttools.o cmdline.c common_cor.o \
|
||||
-o kor
|
||||
|
||||
corr.o: corr.cc corr.h cmdline.h
|
||||
corr.o: corr.cc corr.hh cmdline.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) corr.cc
|
||||
|
||||
corlist.o: corlist.cc corlist.h cmdline.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) corlist.cc
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) corlist.cc
|
||||
|
||||
|
||||
common.o: cmdline.h $(COMMON_PATH)/cmdline_common.ggo cmdline_common
|
||||
|
||||
common.o: cmdline.h $(COMMON_PATH)/cmdline_common.ggo $(COMMON_PATH)/common.cc \
|
||||
$(COMMON_PATH)/common.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) $(COMMON_PATH)/common.cc
|
||||
|
||||
common_cor.o: cmdline.h common_cor.cc common_cor.h
|
||||
@ -40,5 +44,3 @@ clean: clean.cmdline
|
||||
clean.cmdline:
|
||||
rm cmdline.* || true
|
||||
|
||||
include $(COMMON_PATH)/common.mk
|
||||
include $(LIB_PATH)/lib.mk
|
||||
|
||||
@ -1,10 +1,9 @@
|
||||
#include <cstdio>
|
||||
//#include <alloc.h>
|
||||
#include <stdio.h>
|
||||
#include <malloc.h>
|
||||
#include "corlist.h"
|
||||
|
||||
#define min(x,y) ((x<y)?(x):(y))
|
||||
|
||||
using namespace std;
|
||||
|
||||
Weight CorList::GetValue(char X[100], char Y[100], Weight (*H2)[100], int i, int j)
|
||||
{
|
||||
|
||||
@ -1,6 +1,6 @@
|
||||
//---------------------------------------------------------------------------
|
||||
#include "common_cor.h"
|
||||
#include "corr.h"
|
||||
#include "corr.hh"
|
||||
|
||||
#define MAXPATH 256
|
||||
|
||||
|
||||
39
_old/app/src/kor/corr.hh
Executable file
39
_old/app/src/kor/corr.hh
Executable file
@ -0,0 +1,39 @@
|
||||
//---------------------------------------------------------------------------
|
||||
#ifndef _corr_hh
|
||||
#define _corr_hh
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
#include "../lib/tfti.h"
|
||||
#include "../lib/word.h"
|
||||
#include "corlist.h"
|
||||
#include "../common/common.h"
|
||||
|
||||
class Corr : public TFTiv<char,char>
|
||||
{
|
||||
private:
|
||||
Weight H[100][100];
|
||||
char X[100]; // misspelled string
|
||||
char Y[100]; // (possibly partial) candidate string
|
||||
int m; // length of X
|
||||
int n; // maximal length of Y
|
||||
|
||||
Weight ed(int,int);
|
||||
Weight cuted(int);
|
||||
void recomputeH(int);
|
||||
|
||||
|
||||
public:
|
||||
Weight (*H2)[100]; // moje: zmiana z int na Weight (float)
|
||||
int t; // threshold
|
||||
CorList CL; // moje
|
||||
|
||||
Corr() : H2((Weight(*)[100])&H[2][2]) {}; // moje (int->float)
|
||||
Corr(const char* a) : TFTiv<char,char>(a), H2((Weight(*)[100])&H[2][2]) { };
|
||||
|
||||
int correct(const char* w, Words& tab);
|
||||
|
||||
int load2(char *Name); // moje
|
||||
};
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
#endif
|
||||
@ -2,7 +2,7 @@
|
||||
#include <ctype.h>
|
||||
#include "../lib/iotools.h"
|
||||
#include "common_cor.h"
|
||||
#include "corr.h"
|
||||
#include "corr.hh"
|
||||
#include <locale.h>
|
||||
|
||||
|
||||
|
||||
@ -1,11 +1,12 @@
|
||||
PAR=-Wno-deprecated -O2 -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -O2 -fpermissive -static
|
||||
PAR=-Wno-deprecated -m32 -O3 -fpermissive -static
|
||||
PAR2=-c -Wno-deprecated -m32 -O3 -fpermissive -static
|
||||
LIB_PATH=../lib
|
||||
COMMON_PATH=../common
|
||||
CMDLINE_FILE='"../lem/cmdline.h"'
|
||||
|
||||
|
||||
lem: main.cc lem.o auttools.o word.o cmdline.c common_lem.o common.o symtab.o
|
||||
lem: main.cc lem.o $(LIB_PATH)/auttools.o $(LIB_PATH)/word.o \
|
||||
cmdline.c common_lem.o common.o symtab.o
|
||||
g++ $(PAR) -D _CMDLINE_FILE=$(CMDLINE_FILE) \
|
||||
main.cc lem.o $(LIB_PATH)/auttools.o \
|
||||
$(LIB_PATH)/word.o cmdline.c common.o common_lem.o \
|
||||
@ -26,7 +27,8 @@ lem.o: lem.h lem.cc
|
||||
# erro.o: $(LIB_PATH)/erro.h $(LIB_PATH)/erro.cc
|
||||
# g++ $(PAR2) $(LIB_PATH)/erro.cc
|
||||
|
||||
common.o: cmdline_common
|
||||
common.o: $(COMMON_PATH)/cmdline_common.ggo $(COMMON_PATH)/common.cc \
|
||||
$(COMMON_PATH)/common.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) $(COMMON_PATH)/common.cc
|
||||
|
||||
common_lem.o: cmdline.h common_lem.h common_lem.cc
|
||||
@ -35,7 +37,7 @@ common_lem.o: cmdline.h common_lem.h common_lem.cc
|
||||
cmdline.c cmdline.h: cmdline.ggo
|
||||
gengetopt -i cmdline.ggo --conf-parser
|
||||
|
||||
cmdline.ggo: cmdline_lem.ggo $(COMMON_PATH)/cmdline_common.ggo
|
||||
cmdline.ggo: cmdline_lem.ggo ../common/cmdline_common.ggo
|
||||
cat cmdline_lem.ggo ../common/cmdline_common.ggo > cmdline.ggo
|
||||
|
||||
symtab.o: $(LIB_PATH)/symtab.h $(LIB_PATH)/symtab.cc
|
||||
@ -52,6 +54,3 @@ copy:
|
||||
ifdef UTT_BIN_DIR
|
||||
cp lem $(UTT_BIN_DIR)
|
||||
endif
|
||||
|
||||
include $(COMMON_PATH)/common.mk
|
||||
include $(LIB_PATH)/lib.mk
|
||||
|
||||
@ -1,12 +1,9 @@
|
||||
#include "lem.h"
|
||||
|
||||
#include <stdlib.h>
|
||||
#include <assert.h>
|
||||
#include <cstdlib>
|
||||
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
/* Znajduje opisy slownikowe dla wyrazu.
|
||||
* Parametry:
|
||||
* form - wyraz,
|
||||
|
||||
@ -1,11 +1,17 @@
|
||||
PAR=-Wno-deprecated -O3
|
||||
PAR2=-c -Wno-deprecated -O3 -fpermissive
|
||||
PAR=-Wno-deprecated -m32 -O3
|
||||
PAR2=-c -Wno-deprecated -m32 -O3 -fpermissive
|
||||
# -static
|
||||
LIB_PATH=../lib
|
||||
COMMON_PATH=../common
|
||||
|
||||
main: auttools.o word.o copy
|
||||
|
||||
auttools.o: auttools.h auttools.cc
|
||||
g++ $(PAR2) auttools.cc
|
||||
|
||||
word.o: word.h word.cc
|
||||
g++ $(PAR2) word.cc
|
||||
|
||||
clean:
|
||||
rm *.o
|
||||
|
||||
@ -13,5 +19,3 @@ copy:
|
||||
#ifdef UTT_LIB_DIR
|
||||
# cp -r perl $(UTT_LIB_DIR)/
|
||||
#endif
|
||||
|
||||
include ./lib.mk
|
||||
|
||||
@ -162,4 +162,3 @@ void despos(const char* des, char* pos)
|
||||
pos[pi]='\0';
|
||||
}
|
||||
|
||||
|
||||
|
||||
@ -1,9 +1,8 @@
|
||||
#include "symtab.h"
|
||||
#include <values.h>
|
||||
#include <cstdio>
|
||||
//#include <alloc.h>
|
||||
#include <cstdlib>
|
||||
using namespace std;
|
||||
#include <stdio.h>
|
||||
#include <malloc.h>
|
||||
#include <stdlib.h>
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
SymbolTable::SymbolTable(int n, int (*h)(const char*,int), const char* filename)
|
||||
@ -158,8 +157,7 @@ int hash2(const char* s, int _sz)
|
||||
strncpy((char*)&i1,s,sizeof(int));
|
||||
strncpy((char*)&i2,s+(l/2-2),sizeof(int));
|
||||
strncpy((char*)&i3,s+(l-4),sizeof(int));
|
||||
//return abs((i1+i2+i3) % _sz);
|
||||
return (i1+i2+i3) % _sz;
|
||||
return abs((i1+i2+i3) % _sz);
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
@ -4,19 +4,20 @@
|
||||
#include <stddef.h>
|
||||
#include <iostream>
|
||||
#include <typeinfo>
|
||||
#include <cstring>
|
||||
#include <string.h>
|
||||
|
||||
#include <cstdio>
|
||||
#include <stdio.h>
|
||||
|
||||
//#include "top.h"
|
||||
#include "ttrans.h"
|
||||
using namespace std;
|
||||
|
||||
using namespace std;
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
/// Klasa bazowa przetwornika sko<EFBFBD>czonego.
|
||||
/// Klasa bazowa przetwornika skończonego.
|
||||
/**
|
||||
\remark Po co ta klasa? Co dotyczy samych przej<EFBFBD><EFBFBD>, przenie<EFBFBD><EFBFBD> do TTrans,
|
||||
reszt<EFBFBD> wcieli<EFBFBD> do TFT.
|
||||
\remark Po co ta klasa? Co dotyczy samych przejść, przenieść do TTrans,
|
||||
resztę wcielić do TFT.
|
||||
*/
|
||||
class FT
|
||||
{
|
||||
@ -24,46 +25,46 @@ public:
|
||||
FT() : copy_default(false), print_mode(OO), ttn(0) {};
|
||||
|
||||
//print mode
|
||||
enum OUTPUT { II, ///< tylko symbole wej<EFBFBD>ciowe
|
||||
OO, ///< tylko symbole wyj<EFBFBD>ciowe
|
||||
IOIO, ///< symbol wyj<EFBFBD>ciowy po wej<65>ciowym
|
||||
OIOI, ///< symbol wyj<EFBFBD>ciowy przed wej<65>ciowym
|
||||
IIOO, ///< ca<EFBFBD>e wej<65>cie, potem ca<63>e wyj<79>cie
|
||||
OOII ///< ca<EFBFBD>e wyj<79>cie, potem ca<63>e wej<65>cie
|
||||
enum OUTPUT { II, ///< tylko symbole wejściowe
|
||||
OO, ///< tylko symbole wyjściowe
|
||||
IOIO, ///< symbol wyjściowy po wejściowym
|
||||
OIOI, ///< symbol wyjściowy przed wejściowym
|
||||
IIOO, ///< całe wejście, potem całe wyjście
|
||||
OOII ///< całe wyjście, potem całe wejście
|
||||
|
||||
};
|
||||
|
||||
/// maks d<EFBFBD>ugo<EFBFBD><EFBFBD> <20>cie<69>ki
|
||||
/// maks długość ścieżki
|
||||
static const unsigned int ftMAXPATH=500;
|
||||
|
||||
/// maks d<EFBFBD>ugo<EFBFBD><EFBFBD> opisu typu symbolu we/wy
|
||||
/// maks długość opisu typu symbolu we/wy
|
||||
/**
|
||||
\remark Przenie<EFBFBD><EFBFBD> do TTrans
|
||||
\remark Przenieść do TTrans
|
||||
*/
|
||||
static const unsigned int ftTYPELEN=32;
|
||||
|
||||
/// specjalny symbol dla warto<EFBFBD>ci 'epsilon'
|
||||
/// specjalny symbol dla wartości 'epsilon'
|
||||
/**
|
||||
\remark Przenie<EFBFBD><EFBFBD> do TTrans
|
||||
\remark Przenieść do TTrans
|
||||
*/
|
||||
static const char ftEPSILON='~';
|
||||
|
||||
/// specialny symbol dla warto<EFBFBD>ci 'default'
|
||||
/// specialny symbol dla wartości 'default'
|
||||
/**
|
||||
\remark Przenie<EFBFBD><EFBFBD> do TTrans
|
||||
\remark Przenieść do TTrans
|
||||
*/
|
||||
static const char ftDEFAULT='@';
|
||||
|
||||
/// domy<EFBFBD>lny symbol wyj<79>ciowy (true-'@', flase-'~')
|
||||
/// domyślny symbol wyjściowy (true-'@', flase-'~')
|
||||
/**
|
||||
\remark Przenie<EFBFBD><EFBFBD> do TTrans(???)
|
||||
\remark Przenieść do TTrans(???)
|
||||
*/
|
||||
bool copy_default;
|
||||
|
||||
/// tryb wyj<EFBFBD>cia
|
||||
/// tryb wyjścia
|
||||
OUTPUT print_mode;
|
||||
|
||||
/// false, je<EFBFBD>li automat nie ma przej<65><6A>
|
||||
/// false, jeśli automat nie ma przejść
|
||||
operator bool() { return (bool)ttn; };
|
||||
|
||||
virtual const char* intype() { return itype; };
|
||||
@ -71,39 +72,39 @@ public:
|
||||
|
||||
protected:
|
||||
|
||||
/// liczba element<EFBFBD>w tablicy tt
|
||||
/// liczba elementów tablicy tt
|
||||
unsigned long ttn;
|
||||
|
||||
/// liczba stan<EFBFBD>w
|
||||
/// liczba stanów
|
||||
unsigned long states;
|
||||
|
||||
/// liczba przej<EFBFBD><EFBFBD>
|
||||
/// liczba przejść
|
||||
unsigned long transitions;
|
||||
|
||||
/// typ symboli wej<EFBFBD>ciowych (napis)
|
||||
/// typ symboli wejściowych (napis)
|
||||
/**
|
||||
\remark Przenie<EFBFBD><EFBFBD> do TTrans(???)
|
||||
\remark Przenieść do TTrans(???)
|
||||
*/
|
||||
char itype[ftTYPELEN];
|
||||
|
||||
/// typ symboli wyj<EFBFBD>ciowych (napis)
|
||||
/// typ symboli wyjściowych (napis)
|
||||
/**
|
||||
\remark Przenie<EFBFBD><EFBFBD> do TTrans(???)
|
||||
\remark Przenieść do TTrans(???)
|
||||
*/
|
||||
char otype[ftTYPELEN];
|
||||
};
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
/// Szablon przetwornika sko<EFBFBD>czonego
|
||||
/// Szablon przetwornika skończonego
|
||||
/**
|
||||
\param I - typ symbolu wej<EFBFBD>ciowego
|
||||
\param Ipass - typ, jaki ma by<EFBFBD> u<EFBFBD>yty przy przekazywaniu symbolu we jako parametru
|
||||
do funkcji (metody), r<EFBFBD>wny \a I lub \a I&
|
||||
\param O - typ symbolu wyj<EFBFBD>ciowego
|
||||
\param Opass - typ, jaki ma by<EFBFBD> u<EFBFBD>yty przy przekazywaniu symbolu wy jako parametru
|
||||
do funkcji (metody), r<EFBFBD>wny \a O lub \a O&
|
||||
\param - typ przej<EFBFBD>cia, musi by<EFBFBD> podklas<EFBFBD> TTrans
|
||||
\param I - typ symbolu wejściowego
|
||||
\param Ipass - typ, jaki ma być użyty przy przekazywaniu symbolu we jako parametru
|
||||
do funkcji (metody), równy \a I lub \a I&
|
||||
\param O - typ symbolu wyjściowego
|
||||
\param Opass - typ, jaki ma być użyty przy przekazywaniu symbolu wy jako parametru
|
||||
do funkcji (metody), równy \a O lub \a O&
|
||||
\param - typ przejścia, musi być podklasą TTrans
|
||||
*/
|
||||
template<class I, class Ipass, class O, class Opass, class TT>
|
||||
class TFT : public FT
|
||||
@ -116,58 +117,58 @@ public:
|
||||
|
||||
/**
|
||||
\name Metody poziomu 1
|
||||
Poziom przej<EFBFBD><EFBFBD>.
|
||||
Poziom przejść.
|
||||
*/
|
||||
|
||||
//@{
|
||||
|
||||
/// Test, czy przej<EFBFBD>cie \a t akceptuje symbol \a in.
|
||||
/// Test, czy przejście \a t akceptuje symbol \a in.
|
||||
bool accepts(long t, Ipass in) const;
|
||||
|
||||
/// Test, czy lista przej<EFBFBD><EFBFBD> dla aktualnego stanu jest kontynuowana po \a t.
|
||||
/// Test, czy lista przejść dla aktualnego stanu jest kontynuowana po \a t.
|
||||
bool continued(long t) const;
|
||||
|
||||
/// Stan, do kt<EFBFBD>rego prowadzi przej<65>cie \a t.
|
||||
/// Stan, do którego prowadzi przejście \a t.
|
||||
/**
|
||||
\pre !empty(t)
|
||||
*/
|
||||
long next(long t) const;
|
||||
|
||||
/// Symbol wej<EFBFBD>ciowy przej<65>cia \a t.
|
||||
/// Symbol wejściowy przejścia \a t.
|
||||
Ipass input(long t) const;
|
||||
|
||||
/// Symbol wyj<EFBFBD>ciowy przej<65>cia \a t.
|
||||
/// Symbol wyjściowy przejścia \a t.
|
||||
Opass output(long t) const;
|
||||
|
||||
/// Zwraca \c true, je<EFBFBD>li symbolem we przej<65>cia \a t jest epsilon.
|
||||
/// Zwraca \c true, jeśli symbolem we przejścia \a t jest epsilon.
|
||||
bool epsi(long t) const;
|
||||
|
||||
/// Zwraca \c true, je<EFBFBD>li symbolem we przej<65>cia \a t jest symbol domy<6D>lny.
|
||||
/// Zwraca \c true, jeśli symbolem we przejścia \a t jest symbol domyślny.
|
||||
bool defi(long t) const;
|
||||
|
||||
/// Zwraca \c true, je<EFBFBD>li symbolem wy przej<65>cia \a t jest epsilon.
|
||||
/// Zwraca \c true, jeśli symbolem wy przejścia \a t jest epsilon.
|
||||
bool epso(long t) const;
|
||||
|
||||
/// Zwraca \c true, je<EFBFBD>li symbolem wy przej<65>cia \a t jest symbol domy<6D>lny.
|
||||
/// Zwraca \c true, jeśli symbolem wy przejścia \a t jest symbol domyślny.
|
||||
bool defo(long t) const;
|
||||
|
||||
/// Indeks przej<EFBFBD>cia przez \a in.
|
||||
/// Indeks przejścia przez \a in.
|
||||
long tra(long t, Ipass in) const;
|
||||
|
||||
/// Indeks przej<EFBFBD>cia przez \a in - non-deterministic.
|
||||
/// Indeks przejścia przez \a in - non-deterministic.
|
||||
long tra_nd(long t, Ipass in, long nth) const;
|
||||
|
||||
//@}
|
||||
|
||||
/**
|
||||
\name Poziom 2
|
||||
Poziom stan<EFBFBD>w. Stan (indeks stanu) = indeks jego pierwszego przej<EFBFBD>cia
|
||||
Poziom stanów. Stan (indeks stanu) = indeks jego pierwszego przejścia
|
||||
*/
|
||||
//@{
|
||||
/// Zwraca \c true je<EFBFBD>li stan \a s jest pusty (nie ma z niego przej<65><6A>).
|
||||
/// Zwraca \c true jeśli stan \a s jest pusty (nie ma z niego przejść).
|
||||
bool empty(long s) const { return tt[s].empty(); }
|
||||
|
||||
/// Zwraca \c true je<EFBFBD>li stan \a s jest stanem ko<6B>cowym.
|
||||
/// Zwraca \c true jeśli stan \a s jest stanem końcowym.
|
||||
bool final(long s) const { return tt[s].final(); }
|
||||
|
||||
long next(long t, Ipass in) const;
|
||||
@ -201,7 +202,7 @@ Poziom ...
|
||||
|
||||
protected:
|
||||
|
||||
TT* tt; // tablica przej<EFBFBD><EFBFBD>
|
||||
TT* tt; // tablica przejść
|
||||
|
||||
long prn(const I* si, long* p, O* so) const;
|
||||
|
||||
@ -209,7 +210,7 @@ protected:
|
||||
|
||||
void sort();
|
||||
|
||||
void setiotypes(); // NIE DZIA<EFBFBD>A (dlaczego???)
|
||||
void setiotypes(); // NIE DZIAŁA (dlaczego???)
|
||||
|
||||
// friend ostream& operator<<(ostream&,const CDFA&);
|
||||
// friend istream& operator>>(istream&,CDFA&);
|
||||
@ -227,12 +228,12 @@ private:
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
/**
|
||||
stan = indeks pierwszego przej<EFBFBD>cia
|
||||
stan = indeks pierwszego przejścia
|
||||
|
||||
state(t) = stan, do kt<EFBFBD>rego nale<EFBFBD>y t
|
||||
state(t) = stan, do którego należy t
|
||||
|
||||
symbol zerowy = symbol s, dla kt<EFBFBD>rego (bool)s zwraca \c false,
|
||||
w przypadku znak<EFBFBD>w - '\0'
|
||||
symbol zerowy = symbol s, dla którego (bool)s zwraca \c false,
|
||||
w przypadku znaków - '\0'
|
||||
*/
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
@ -289,10 +290,10 @@ bool TFT<I,Ipass,O,Opass,TT>::defo(long t) const
|
||||
{ return tt[t].defo(); }
|
||||
|
||||
/**
|
||||
\param +t - indeks przej<EFBFBD>cia
|
||||
\param +t - indeks przejścia
|
||||
\param +in - symbol we
|
||||
\return Indeks przj<EFBFBD>cia (>=\a t) dla bie<EFBFBD><EFBFBD>cego stanu, kt<EFBFBD>re
|
||||
akceptuje symbol we \a in lub -1, je<EFBFBD>li nie ma takiego przej<EFBFBD>cia
|
||||
\return Indeks przjścia (>=\a t) dla bieżącego stanu, które
|
||||
akceptuje symbol we \a in lub -1, jeśli nie ma takiego przejścia
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>::tra(long t, Ipass in) const
|
||||
@ -310,13 +311,13 @@ long TFT<I,Ipass,O,Opass,TT>::tra(long t, Ipass in) const
|
||||
}
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
/// Indeks przej<EFBFBD>cia - wersja dla automatu niedeterministycznego.
|
||||
/// Indeks przejścia - wersja dla automatu niedeterministycznego.
|
||||
/**
|
||||
\param +t - indeks przej<EFBFBD>cia
|
||||
\param +t - indeks przejścia
|
||||
\param +in - symbol we
|
||||
\return Indeks przj<EFBFBD>cia (>=\a t) dla bie<EFBFBD><EFBFBD>cego stanu, kt<EFBFBD>re
|
||||
akceptuje symbol we \a in lub -1, je<EFBFBD>li nie ma takiego przej<EFBFBD>cia
|
||||
Je<EFBFBD>li nth==0, t1>=t, w przeciwnym razie t1>t.
|
||||
\return Indeks przjścia (>=\a t) dla bieżącego stanu, które
|
||||
akceptuje symbol we \a in lub -1, jeśli nie ma takiego przejścia
|
||||
Jeśli nth==0, t1>=t, w przeciwnym razie t1>t.
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>::tra_nd(long t, Ipass in, long nth) const
|
||||
@ -347,12 +348,12 @@ long TFT<I,Ipass,O,Opass,TT>::tra_nd(long t, Ipass in, long nth) const
|
||||
//----------------------------------------------------------------------------
|
||||
|
||||
|
||||
/// Funkcja przej<EFBFBD>cia.
|
||||
/// Funkcja przejścia.
|
||||
/**
|
||||
\param t - stan
|
||||
\param in - symbol we
|
||||
\return Stan, do kt<EFBFBD>rego mo<EFBFBD>na przej<EFBFBD><EFBFBD> z \a t po wp<EFBFBD>ywem symbolu \a in
|
||||
lub -1, je<EFBFBD>li nie ma przej<EFBFBD>cia przez \a in
|
||||
\return Stan, do którego można przejść z \a t po wpływem symbolu \a in
|
||||
lub -1, jeśli nie ma przejścia przez \a in
|
||||
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
@ -375,13 +376,13 @@ long TFT<I,Ipass,O,Opass,TT>::next(long t, Ipass in) const
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
//----------------------------------------------------------------------------
|
||||
/// Uog<EFBFBD>lniona funkcja przejscia.
|
||||
/// Uogólniona funkcja przejscia.
|
||||
/**
|
||||
\param +s - stan
|
||||
\param +w - wska<EFBFBD>nik pierwszego elementu ci<EFBFBD>gu symboli we, zako<EFBFBD>czonego symbolem zerowym
|
||||
\param maxpath maksymalna d<EFBFBD>ugo<EFBFBD><EFBFBD> <EFBFBD>cie<EFBFBD>ki, domy<EFBFBD>lnie ftMAXPATH
|
||||
\return stan osi<EFBFBD>galny z \a s pod wp<EFBFBD>ywem \a w (na <EFBFBD>cie<EFBFBD>ce mog<EFBFBD> si<EFBFBD> pojawi<EFBFBD>
|
||||
epsilon-przej<EFBFBD>cia
|
||||
\param +w - wskaźnik pierwszego elementu ciągu symboli we, zakończonego symbolem zerowym
|
||||
\param maxpath maksymalna długość ścieżki, domyślnie ftMAXPATH
|
||||
\return stan osiągalny z \a s pod wpływem \a w (na ścieżce mogą się pojawić
|
||||
epsilon-przejścia
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>::gtra(long s, const I* w, long maxpath) const
|
||||
@ -410,12 +411,12 @@ long TFT<I,Ipass,O,Opass,TT>::gtra(long s, const I* w, long maxpath) const
|
||||
/// Kontynuacja.
|
||||
/**
|
||||
...
|
||||
\param +s stan, je<EFBFBD>li -1 - poszukiwane jest nast<EFBFBD>pne rozwi<EFBFBD>zanie
|
||||
\param -c ci<EFBFBD>g symboli we ze <EFBFBD>cie<EFBFBD>ki prowadz<EFBFBD>cej z \a s do
|
||||
stanu ko<EFBFBD>cowego
|
||||
\return d<EFBFBD>ugo<EFBFBD><EFBFBD> ci<EFBFBD>gu \a c (= d<EFBFBD>ugo<EFBFBD><EFBFBD> <EFBFBD>cie<EFBFBD>ki)
|
||||
\remark DZIA<EFBFBD>A TYLKO DLA ZNAK<EFBFBD>W!!!
|
||||
EPSILON-PRZEJ<EFBFBD>CIA NIEDOZWOLONE!!!
|
||||
\param +s stan, jeśli -1 - poszukiwane jest następne rozwiązanie
|
||||
\param -c ciąg symboli we ze ścieżki prowadzącej z \a s do
|
||||
stanu końcowego
|
||||
\return długość ciągu \a c (= długość ścieżki)
|
||||
\remark DZIAŁA TYLKO DLA ZNAKÓW!!!
|
||||
EPSILON-PRZEJŚCIA NIEDOZWOLONE!!!
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>::cont(long s, I* c) const
|
||||
@ -481,10 +482,10 @@ long TFT<I,Ipass,O,Opass,TT>::match(const I* w, long* p) const
|
||||
//----------------------------------------------------------------------------
|
||||
/// Dopasowanie niedeterministyczne.
|
||||
/**
|
||||
\param +w - wska<EFBFBD>nik pierwszego elementu ci<EFBFBD>gu symboli we, zako<EFBFBD>czonego symbolem zerowym,
|
||||
je<EFBFBD>li NULL - poszukiwane jest nast<EFBFBD>pne rozwi<EFBFBD>zanie
|
||||
\param -p ci<EFBFBD>g przej<EFBFBD><EFBFBD> zako<EFBFBD>czony -1
|
||||
\return d<EFBFBD>ugo<EFBFBD><EFBFBD> dopasowania (PO CO?)
|
||||
\param +w - wskaźnik pierwszego elementu ciągu symboli we, zakończonego symbolem zerowym,
|
||||
jeśli NULL - poszukiwane jest następne rozwiązanie
|
||||
\param -p ciąg przejść zakończony -1
|
||||
\return długość dopasowania (PO CO?)
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>::match_nd(const I* w, long* p) const
|
||||
@ -525,13 +526,13 @@ long TFT<I,Ipass,O,Opass,TT>::match_nd(const I* w, long* p) const
|
||||
}
|
||||
|
||||
//----------------------------------------------------------------------------
|
||||
/// Najd<EFBFBD>u<EFBFBD>sze dopasowanie.
|
||||
/// Najdłuższe dopasowanie.
|
||||
/**
|
||||
\param +w wska<EFBFBD>nik pierwszego elementu ci<EFBFBD>gu symboli wej<EFBFBD>ciowych
|
||||
\param -p <EFBFBD>cie<EFBFBD>ka
|
||||
\param -plen d<EFBFBD>ugo<EFBFBD><EFBFBD> <EFBFBD>cie<EFBFBD>ki
|
||||
\param +maxpath maks dd<EFBFBD>ugo<EFBFBD><EFBFBD> <EFBFBD>cie<EFBFBD>ki, domy<EFBFBD>lnie FT::ftMAXPATH
|
||||
\return d<EFBFBD>ugo<EFBFBD><EFBFBD> skonsumowanego wej<EFBFBD>cia
|
||||
\param +w wskaźnik pierwszego elementu ciągu symboli wejściowych
|
||||
\param -p ścieżka
|
||||
\param -plen długość ścieżki
|
||||
\param +maxpath maks ddługość ścieżki, domyślnie FT::ftMAXPATH
|
||||
\return długość skonsumowanego wejścia
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>
|
||||
@ -555,13 +556,13 @@ long TFT<I,Ipass,O,Opass,TT>
|
||||
}
|
||||
|
||||
//----------------------------------------------------------------------------
|
||||
/// Najd<EFBFBD>u<EFBFBD>sza <20>cie<69>ka.
|
||||
/// Najdłuższa ścieżka.
|
||||
/**
|
||||
\param +buf wska<EFBFBD>nik pierwszego elementu ci<EFBFBD>gu symboli wej<EFBFBD>ciowych
|
||||
\param +buf wskaźnik pierwszego elementu ciągu symboli wejściowych
|
||||
\param -buf pozycja jeden za skonsumowanym prefiksem
|
||||
\param +path wska<EFBFBD>nik pierwszego elementu wektora przej<EFBFBD><EFBFBD>
|
||||
\param -path wska<EFBFBD>nik jeden za ostatnim przej<EFBFBD>ciem
|
||||
\return d<EFBFBD>ugo<EFBFBD><EFBFBD> skonsumowanego prefiksu (PO CO? LEPIEJ D<EFBFBD> <EFBFBD>CIE<EFBFBD>KI)
|
||||
\param +path wskaźnik pierwszego elementu wektora przejść
|
||||
\param -path wskaźnik jeden za ostatnim przejściem
|
||||
\return długość skonsumowanego prefiksu (PO CO? LEPIEJ DŁ ŚCIEŻKI)
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>
|
||||
@ -581,14 +582,14 @@ long TFT<I,Ipass,O,Opass,TT>
|
||||
}
|
||||
|
||||
//----------------------------------------------------------------------------
|
||||
/// Najd<EFBFBD>u<EFBFBD>szy prefiks.
|
||||
/// Najdłuższy prefiks.
|
||||
/**
|
||||
\param +buf wska<EFBFBD>nik pierwszego elementu ci<EFBFBD>gu symboli wej<EFBFBD>ciowych
|
||||
\param +buf wskaźnik pierwszego elementu ciągu symboli wejściowych
|
||||
\param -buf pozycja jeden za skonsumowanym prefiksem
|
||||
\param +sep separator
|
||||
\return stan po przej<EFBFBD>ciu przez \a sep
|
||||
\remark Dzia<EFBFBD>a tylko dla automat<EFBFBD>w deterministycznych, minimalnych, eps-wolnych,
|
||||
gdzie d<EFBFBD>. <EFBFBD>cie<EFBFBD>ki == d<EFBFBD>. dopasowania.
|
||||
\return stan po przejściu przez \a sep
|
||||
\remark Działa tylko dla automatów deterministycznych, minimalnych, eps-wolnych,
|
||||
gdzie dł. ścieżki == dł. dopasowania.
|
||||
*/
|
||||
template <class I, class Ipass, class O, class Opass, class TT>
|
||||
long TFT<I,Ipass,O,Opass,TT>
|
||||
@ -600,14 +601,14 @@ long TFT<I,Ipass,O,Opass,TT>
|
||||
static bool more;
|
||||
|
||||
long s;
|
||||
if(*buf) // pierwsze wywo<EFBFBD>anie
|
||||
if(*buf) // pierwsze wywołanie
|
||||
{
|
||||
if(!lgstpath(buf,path,start))
|
||||
return -1;
|
||||
--path;
|
||||
more=true;
|
||||
}
|
||||
else // kolejne wywo<EFBFBD>anie
|
||||
else // kolejne wywołanie
|
||||
--buf,--path;
|
||||
while(more)
|
||||
if(path>=pathtab)
|
||||
|
||||
@ -2,16 +2,17 @@
|
||||
#define TFTiH
|
||||
//---------------------------------------------------------------------------
|
||||
#include <fstream>
|
||||
#include <cmath>
|
||||
#include <math.h>
|
||||
#include <iomanip>
|
||||
#include <iostream>
|
||||
//#include <typeinfo.h>
|
||||
|
||||
#include "tft.h"
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
using namespace std;
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
|
||||
template<class I, class Ipass, class O, class Opass>
|
||||
class TFTi : public TFT<I,Ipass,O,Opass,TTrans_i<I,Ipass,O,Opass> >
|
||||
{
|
||||
|
||||
@ -2,9 +2,11 @@
|
||||
#define _TTransi_h
|
||||
//---------------------------------------------------------------------------
|
||||
#include <iostream>
|
||||
using namespace std;
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
|
||||
using namespace std;
|
||||
|
||||
//! The template for a transition with input and output symbols stored internally.
|
||||
/*!
|
||||
A state is identified with the set of its outgoing transitions.
|
||||
|
||||
@ -1,11 +1,11 @@
|
||||
//---------------------------------------------------------------------------
|
||||
#include "word.h"
|
||||
#include "auttools.h"
|
||||
//#include <istream.h>
|
||||
#include <istream>
|
||||
#include <algorithm>
|
||||
using namespace std;
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
using namespace std;
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
void Word::autodescr(const char* fo, const char* de)
|
||||
@ -122,13 +122,13 @@ int Words::next() {
|
||||
}
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
void Words::sort() {
|
||||
std::sort(tab.begin(), tab.end(), Word::cmp_w);
|
||||
void Words::sort() {
|
||||
// sort(tab.begin(), tab.end(), Word::cmp_w); //NIE DZIALA
|
||||
}
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
void Words::sort_rev() {
|
||||
std::sort(tab.begin(), tab.end(), cmp_w_rev_fun);
|
||||
void Words::sort_rev() {
|
||||
// sort(tab.begin(), tab.end(), cmp_w_rev_fun); // NIE DZIALA
|
||||
}
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
@ -143,15 +143,15 @@ int Words::add(const char* fo)
|
||||
if (cnt>=tab.capacity()-1)
|
||||
tab.resize(tab.size()*2);
|
||||
|
||||
|
||||
Word o;
|
||||
o.form(fo);
|
||||
o.w_suf(0.0);
|
||||
tab.push_back(o);
|
||||
|
||||
Word o;
|
||||
o.form(fo);
|
||||
o.w_suf(0.0);
|
||||
tab.push_back(o);
|
||||
// tab[cnt].form(fo);
|
||||
// tab[cnt].w_suf(0.0);
|
||||
|
||||
|
||||
|
||||
// if(cnt<MAX-1) {
|
||||
/* tab.push_back(new Word());
|
||||
tab[cnt]->form(fo);
|
||||
@ -163,8 +163,7 @@ int Words::add(const char* fo)
|
||||
}
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
//TYMCZASOWO TAK(DLA CORA)
|
||||
//TYMCZASOWO TAK(DLA CORA)
|
||||
int Words::add(const char* fo, float weight)
|
||||
{
|
||||
int i = find(fo);
|
||||
@ -175,13 +174,13 @@ int Words::add(const char* fo, float weight)
|
||||
if (cnt>=tab.capacity()-1)
|
||||
tab.resize(tab.size()*2);
|
||||
|
||||
Word o;
|
||||
o.form(fo);
|
||||
o.w_suf(weight);
|
||||
tab.push_back(o);
|
||||
Word o;
|
||||
o.form(fo);
|
||||
o.w_suf(weight);
|
||||
tab.push_back(o);
|
||||
// tab[cnt].form(fo);
|
||||
// tab[cnt].w_suf(weight);
|
||||
|
||||
|
||||
return cnt++;
|
||||
// }
|
||||
//return -1;
|
||||
|
||||
@ -5,17 +5,17 @@
|
||||
//#include "alphabet.h"
|
||||
//#include "erro.h"
|
||||
#include "const.h"
|
||||
//#include <iostream.h>
|
||||
#include <iostream>
|
||||
|
||||
#include <string.h>
|
||||
|
||||
#include <vector>
|
||||
#include <cstring>
|
||||
using namespace std;
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
using namespace std;
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
class Word
|
||||
{
|
||||
public:
|
||||
@ -81,7 +81,7 @@ public:
|
||||
|
||||
};
|
||||
|
||||
|
||||
|
||||
inline Word::Word(const Word& word)
|
||||
{ strcpy(f,word.f); strcpy(l,word.l); strcpy(d,word.d); _len_suf=word._len_suf; _w_suf=word._w_suf; returned = 0; }
|
||||
|
||||
@ -106,14 +106,14 @@ inline int Word::cmp(const Word& w) { return strcmp(f,w.f); }
|
||||
//inline int Word::cmpi(const Word& w) { return PL.cmpi(f,w.f); }
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
|
||||
|
||||
|
||||
bool cmp_w_fun(Word a, Word b);
|
||||
bool cmp_w_rev_fun(Word a, Word b);
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
bool cmp_w_fun(Word a, Word b);
|
||||
bool cmp_w_rev_fun(Word a, Word b);
|
||||
|
||||
|
||||
//---------------------------------------------------------------------------
|
||||
//---------------------------------------------------------------------------
|
||||
|
||||
@ -132,7 +132,7 @@ class Words
|
||||
int count() const { return cnt; }
|
||||
void clear() { cnt=0; tab.clear(); }
|
||||
int add(const char* fo);
|
||||
int add(const char* fo, float weight);
|
||||
int add(const char* fo, float weight);
|
||||
int add(const char* fo, const char* des);
|
||||
|
||||
/* zwraca index nastepnego wyniku, podczas pierwszego wywolania
|
||||
|
||||
@ -1,7 +1,6 @@
|
||||
PAR=-Wno-deprecated -O3 -fpermissive
|
||||
PAR=-Wno-deprecated -O3
|
||||
PAR2=-c -Wno-deprecated -O3
|
||||
LIB_PATH=../lib
|
||||
COMMON_PATH=../common
|
||||
CMDLINE_FILE='"../tok.c/cmdline.h"'
|
||||
|
||||
|
||||
@ -15,11 +14,12 @@ tok.o: tok.c cmdline.h
|
||||
common_tok.o: cmdline.h common_tok.cc common_tok.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) common_tok.cc
|
||||
|
||||
common.o: cmdline_common
|
||||
common.o: $(COMMON_PATH)/cmdline_common.ggo $(COMMON_PATH)/common.cc \
|
||||
$(COMMON_PATH)/common.h
|
||||
g++ $(PAR2) -D _CMDLINE_FILE=$(CMDLINE_FILE) $(COMMON_PATH)/common.cc
|
||||
|
||||
cmdline.ggo: cmdline_tok.ggo $(COMMON_PATH)/cmdline_common.ggo
|
||||
cat cmdline_tok.ggo $(COMMON_PATH)/cmdline_common.ggo > cmdline.ggo
|
||||
cmdline.ggo: cmdline_tok.ggo ../common/cmdline_common.ggo
|
||||
cat cmdline_tok.ggo ../common/cmdline_common.ggo > cmdline.ggo
|
||||
|
||||
cmdline.c cmdline.h: cmdline.ggo
|
||||
gengetopt -i cmdline.ggo --conf-parser
|
||||
@ -31,11 +31,7 @@ endif
|
||||
|
||||
|
||||
clean: clean.cmdline
|
||||
rm *.o || true
|
||||
rm tok || true
|
||||
rm *.o
|
||||
|
||||
clean.cmdline:
|
||||
rm cmdline.*
|
||||
|
||||
include $(COMMON_PATH)/common.mk
|
||||
include $(LIB_PATH)/lib.mk
|
||||
|
||||
@ -1,5 +1,4 @@
|
||||
package "tok"
|
||||
version "0.1"
|
||||
|
||||
option "dictionary" d "Dictionary" string typestr="FILENAME" default="cor.bin" no
|
||||
|
||||
|
||||
@ -1,9 +1,7 @@
|
||||
#include <cstdlib>
|
||||
#include <cstring>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
#include "common_tok.h"
|
||||
|
||||
using namespace std;
|
||||
|
||||
char dictionary[255];
|
||||
|
||||
void process_tok_options(gengetopt_args_info args)
|
||||
|
||||
@ -1,6 +1,6 @@
|
||||
tok:
|
||||
|
||||
copy:
|
||||
copy:
|
||||
ifdef UTT_BIN_DIR
|
||||
cp tok ${UTT_BIN_DIR}
|
||||
endif
|
||||
|
||||
0
_old/nawszelkiwypadek/tools/aut2fsa
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/aut2fsa
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/cor_dic/makeLabels.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/cor_dic/makeLabels.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/cor_dic/prep.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/cor_dic/prep.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/fsm2aut
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/fsm2aut
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/canon.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/canon.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/compile_user_dict.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/compile_user_dict.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/count_prefs.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/count_prefs.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/cut_prefs.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/cut_prefs.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/makeLabels.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/makeLabels.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/prep.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/prep.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/prep_user_dict.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/prep_user_dict.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/rmDup.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/rmDup.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/stat.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/stat.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/stat_pre.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/gue_dic/stat_pre.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/lem_dic/makeLabels.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/lem_dic/makeLabels.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/lem_dic/prep.pl
Normal file → Executable file
0
_old/nawszelkiwypadek/tools/lem_dic/prep.pl
Normal file → Executable file
0
auto/defaults
Normal file → Executable file
0
auto/defaults
Normal file → Executable file
0
auto/options
Normal file → Executable file
0
auto/options
Normal file → Executable file
0
auto/output/Makefile
Normal file → Executable file
0
auto/output/Makefile
Normal file → Executable file
0
auto/output/config_h
Normal file → Executable file
0
auto/output/config_h
Normal file → Executable file
0
auto/summary
Normal file → Executable file
0
auto/summary
Normal file → Executable file
Loading…
Reference in New Issue
Block a user