Wersja niestabilna, zawiera wstepne poprawki.
git-svn-id: svn://atos.wmid.amu.edu.pl/utt@13 e293616e-ec6a-49c2-aa92-f4a8b91c5d16
This commit is contained in:
parent
d593c5e768
commit
20b4e44e75
@ -54,6 +54,8 @@ dirs:
|
|||||||
|
|
||||||
.PHONY: components
|
.PHONY: components
|
||||||
components:
|
components:
|
||||||
|
cd $(SRC_DIR)/lib && make; cd $(CUR_DIR)
|
||||||
|
|
||||||
@for cmp in $(COMPONENTS); do\
|
@for cmp in $(COMPONENTS); do\
|
||||||
cd $(SRC_DIR)/$$cmp && make && make copy; cd $(CUR_DIR); \
|
cd $(SRC_DIR)/$$cmp && make && make copy; cd $(CUR_DIR); \
|
||||||
done
|
done
|
||||||
|
28
app/TODO
28
app/TODO
@ -1,6 +1,28 @@
|
|||||||
* wyprowadzic grp-pre i grp-post z grp do aux?
|
|
||||||
* zamienic kota na lepszego (Kubis)
|
* gue SIE NIE KOMPILUJE !!!
|
||||||
*
|
|
||||||
|
* zamienic kota na lepszego (Kubis) [TO]
|
||||||
|
|
||||||
|
* uniezaleznic mar od attr.pm (ew. wcielic attr.pm) [TO]
|
||||||
|
|
||||||
|
* mar nie obsluguje plikow konfiguracyjnych [JW]
|
||||||
|
|
||||||
|
* compiledic - plik tymczasowe [PK]
|
||||||
|
|
||||||
|
* ser i grp - parametr tags (zeby bral wskazany program tag2re) [TO]
|
||||||
|
|
||||||
|
* lista zaleznosci dla komend system w perlu [PK]
|
||||||
|
|
||||||
|
* polaczyc sen'y [TO]
|
||||||
|
|
||||||
|
* poprawic utt_make_conf (wykrywanie slownikow) [PK]
|
||||||
|
|
||||||
|
* generowanie i sprawdzanie zaleznosci dla tarballa [PK]
|
||||||
|
|
||||||
|
* przygotowanie dystrybujcji slownikow [PK]
|
||||||
|
|
||||||
|
|
||||||
|
STARE
|
||||||
|
|
||||||
1. DONE. Makefile do gph (install).
|
1. DONE. Makefile do gph (install).
|
||||||
2. (zrobione dla ser?) Nazwy pmdb2re -> pmdb.tag2re (grp, ser).
|
2. (zrobione dla ser?) Nazwy pmdb2re -> pmdb.tag2re (grp, ser).
|
||||||
|
4
app/dist/tarball/INSTALL
vendored
4
app/dist/tarball/INSTALL
vendored
@ -1,4 +1,6 @@
|
|||||||
Here you can find some information about how to install utt.
|
Here you can find some information about how to install utt.
|
||||||
|
|
||||||
You should just unpack archive and then
|
You should just unpack archive and then
|
||||||
execute create_utt_conf.pl and remove it.
|
execute
|
||||||
|
bin/utt_make_config.pl
|
||||||
|
|
||||||
|
2
app/dist/tarball/README
vendored
2
app/dist/tarball/README
vendored
@ -3,4 +3,4 @@ This directory contains Makefile, which allows to create tar.gz archive.
|
|||||||
To create archive, just write:
|
To create archive, just write:
|
||||||
make
|
make
|
||||||
|
|
||||||
Warning: you need define variable _UTT_DIST_DIR.
|
Warning: you need define variable UTT_DIST_DIR.
|
||||||
|
Binary file not shown.
@ -1,19 +1,33 @@
|
|||||||
#! /usr/bin/env perl
|
#! /usr/bin/env perl
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: compiledic
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
|
use strict;
|
||||||
|
use locale;
|
||||||
|
use File::HomeDir;
|
||||||
|
use File::Basename;
|
||||||
|
use File::Temp;
|
||||||
|
use Getopt::Long;
|
||||||
|
|
||||||
|
my $systemconfigfile='';
|
||||||
|
my $userconfigfile=home()."/.utt/compiledic.conf";
|
||||||
|
|
||||||
|
Getopt::Long::Configure('no_ignore_case_always');
|
||||||
|
|
||||||
$symfile='~/.utt/pl/pl_PL.iso-8859-2.sym';
|
$symfile='~/.utt/pl/pl_PL.iso-8859-2.sym';
|
||||||
$symfilenoext = $symfile;
|
$symfilenoext = $symfile;
|
||||||
$symfilenoext =~ s/\.sym$//;
|
$symfilenoext =~ s/\.sym$//;
|
||||||
$labfile = $symfilenoext . '.lab';
|
$labfile = $symfilenoext . '.lab';
|
||||||
$sclfile = $symfilenoext . '.scl';
|
$sclfile = $symfilenoext . '.scl';
|
||||||
|
|
||||||
use locale;
|
|
||||||
#use strict;
|
|
||||||
|
|
||||||
##################################################
|
##################################################
|
||||||
$linesPerFile = 20000;
|
$linesPerFile = 20000;
|
||||||
|
|
||||||
if (@ARGV < 1) {
|
if (@ARGV < 1) {
|
||||||
print "usage: prep_user_dict.pl dictionary_file\n";
|
print "usage: ", basename($0), " dictionary_file\n";
|
||||||
exit;
|
exit;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -1,5 +1,11 @@
|
|||||||
#!/usr/bin/perl
|
#!/usr/bin/perl
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: gph
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
|
use strict;
|
||||||
use Getopt::Long;
|
use Getopt::Long;
|
||||||
|
|
||||||
my @process;
|
my @process;
|
||||||
|
@ -1,16 +1,20 @@
|
|||||||
#!/usr/bin/perl
|
#!/usr/bin/perl
|
||||||
|
|
||||||
#package: UAM Text Tools
|
#package: UAM Text Tools
|
||||||
#component name: gre
|
#component name: grp
|
||||||
#author: Tomasz Obrêbski
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
use strict;
|
use strict;
|
||||||
use Getopt::Long;
|
use Getopt::Long;
|
||||||
|
use File::HomeDir;
|
||||||
|
|
||||||
my $LIB_DIR="/usr/local/lib/utt"; # katalog zawierajacy terms.m4
|
# katalog zawierajacy terms.m4
|
||||||
|
my $LIB_DIR="/usr/local/lib/utt";
|
||||||
|
|
||||||
my $systemconfigfile="/usr/local/etc/utt/grp.conf";
|
my $systemconfigfile="/usr/local/etc/utt/grp.conf";
|
||||||
my $userconfigfile="$ENV{'HOME'}/.utt/grp.conf";
|
#my $userconfigfile="$ENV{'HOME'}/.utt/grp.conf";
|
||||||
|
my $userconfigfile=home()."/.utt/grp.conf";
|
||||||
|
|
||||||
Getopt::Long::Configure('no_ignore_case_always');
|
Getopt::Long::Configure('no_ignore_case_always');
|
||||||
|
|
||||||
|
@ -1,31 +1,39 @@
|
|||||||
#!/usr/bin/perl -w
|
#!/usr/bin/perl -w
|
||||||
use strict;
|
|
||||||
use Getopt::Long;
|
#package: UAM Text Tools
|
||||||
use locale;
|
#component: kon (search context)
|
||||||
|
#version: 1.0
|
||||||
Getopt::Long::Configure('no_ignore_case_always');
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
my $l='30c';
|
use strict;
|
||||||
my $r='30c';
|
use Getopt::Long;
|
||||||
my $trim=0;
|
use locale;
|
||||||
my $white=0;
|
use File::HomeDir;
|
||||||
my $bon='[0-9]+ [0-9]+ BOM .*';
|
|
||||||
my $eon='[0-9]+ [0-9]+ EOM .*';
|
Getopt::Long::Configure('no_ignore_case_always');
|
||||||
my $bod='[';
|
|
||||||
my $eod=']';
|
my $l='30c';
|
||||||
my $column=0;
|
my $r='30c';
|
||||||
my $ignore=0;
|
my $trim=0;
|
||||||
my $help=0;
|
my $white=0;
|
||||||
|
my $bon='[0-9]+ [0-9]+ BOM .*';
|
||||||
my $systemconfigfile='/usr/local/etc/utt/kon.conf';
|
my $eon='[0-9]+ [0-9]+ EOM .*';
|
||||||
my $userconfigfile="$ENV{'HOME'}/.utt/kon.conf";
|
my $bod='[';
|
||||||
|
my $eod=']';
|
||||||
#read configuration files###########################
|
my $column=0;
|
||||||
my $file;
|
my $ignore=0;
|
||||||
foreach $file ($systemconfigfile, $userconfigfile){
|
my $help=0;
|
||||||
if(open(CONFIG, $file)){
|
|
||||||
while (<CONFIG>) {
|
my $systemconfigfile='/usr/local/etc/utt/kon.conf';
|
||||||
chomp;
|
#my $userconfigfile="$ENV{'HOME'}/.utt/kon.conf";
|
||||||
|
my $userconfigfile=home()."/.utt/kon.conf";
|
||||||
|
|
||||||
|
#read configuration files###########################
|
||||||
|
my $file;
|
||||||
|
foreach $file ($systemconfigfile, $userconfigfile){
|
||||||
|
if(open(CONFIG, $file)){
|
||||||
|
while (<CONFIG>) {
|
||||||
|
chomp;
|
||||||
s/#.*//;
|
s/#.*//;
|
||||||
s/^\s+//;
|
s/^\s+//;
|
||||||
s/\s+$//;
|
s/\s+$//;
|
||||||
|
@ -1,14 +1,21 @@
|
|||||||
#!/usr/bin/perl
|
#!/usr/bin/perl
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: kot
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
use strict;
|
use strict;
|
||||||
use Getopt::Long;
|
use Getopt::Long;
|
||||||
|
use File::HomeDir;
|
||||||
|
|
||||||
my $help=0;
|
my $help=0;
|
||||||
my $gap_fill="\n-----\n";
|
my $gap_fill="\n-----\n";
|
||||||
my $spaces=0;
|
my $spaces=0;
|
||||||
|
|
||||||
my $systemconfigfile='/usr/local/etc/utt/kot.conf';
|
my $systemconfigfile='/usr/local/etc/utt/kot.conf';
|
||||||
my $userconfigfile="$ENV{'HOME'}/.utt/kot.conf";
|
#my $userconfigfile="$ENV{'HOME'}/.utt/kot.conf";
|
||||||
|
my $userconfigfile=home()."/.utt/kot.conf";
|
||||||
|
|
||||||
#read configuration files###########################
|
#read configuration files###########################
|
||||||
my $file;
|
my $file;
|
||||||
|
@ -1,8 +1,9 @@
|
|||||||
#!/usr/bin/perl
|
#!/usr/bin/perl
|
||||||
|
|
||||||
#package: UAM Text Tools
|
#package: UAM Text Tools
|
||||||
#component name: mrk
|
#component: mar
|
||||||
#author: Marcin Walas
|
#version: 1.0
|
||||||
|
#author: Marcin Walas
|
||||||
|
|
||||||
#this program tags the tokenized file with given tags
|
#this program tags the tokenized file with given tags
|
||||||
#tags can be given in any order and configuration through the expression
|
#tags can be given in any order and configuration through the expression
|
||||||
|
@ -1,3 +1,9 @@
|
|||||||
#!/bin/bash
|
#!/bin/bash
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: rm12
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
|
|
||||||
sed -r '/[0-9]+[ \t]+[0-9]+[ \t]+BOS/! s/[0-9]+[ \t]+[0-9]+[ \t]//'
|
sed -r '/[0-9]+[ \t]+[0-9]+[ \t]+BOS/! s/[0-9]+[ \t]+[0-9]+[ \t]//'
|
||||||
|
@ -2,19 +2,18 @@
|
|||||||
|
|
||||||
#package: UAM Text Tools
|
#package: UAM Text Tools
|
||||||
#component: ser (pattern search tool)
|
#component: ser (pattern search tool)
|
||||||
#author: Tomasz Obrêbski
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
use strict;
|
use strict;
|
||||||
use Getopt::Long;
|
use Getopt::Long;
|
||||||
use File::Temp;
|
use File::Temp;
|
||||||
|
use File::HomeDir;
|
||||||
|
|
||||||
my $LIB_DIR="/usr/local/lib/utt";
|
my $LIB_DIR="/usr/local/lib/utt";
|
||||||
|
|
||||||
my $systemconfigfile='/usr/local/etc/utt/ser.conf';
|
my $systemconfigfile='/usr/local/etc/utt/ser.conf';
|
||||||
my $userconfigfile="$ENV{'HOME'}/.utt/ser.conf";
|
#my $userconfigfile="$ENV{'HOME'}/.utt/ser.conf";
|
||||||
|
my $userconfigfile=home()."/.utt/ser.conf";
|
||||||
#use lib "$ENV{HOME}/.utt/lib/perl";
|
|
||||||
#use attr;
|
|
||||||
|
|
||||||
Getopt::Long::Configure('no_ignore_case_always');
|
Getopt::Long::Configure('no_ignore_case_always');
|
||||||
|
|
||||||
@ -160,13 +159,25 @@ system "m4 \"--define=PATTERN=$flexpattern\" \"--define=DEFAULTACTION=$defaultac
|
|||||||
|
|
||||||
if($flex)
|
if($flex)
|
||||||
{
|
{
|
||||||
system "cat $tmpfile_l";
|
#system "cat $tmpfile_l";
|
||||||
exit 0;
|
if(open(FLEX, $tmpfile_l)) {
|
||||||
|
while(<FLEX>) {
|
||||||
|
print @_;
|
||||||
|
}
|
||||||
|
close FLEX;
|
||||||
|
}
|
||||||
|
else {
|
||||||
|
print "Unable to open file $tmpfile_l\n";
|
||||||
|
}
|
||||||
|
exit 0;
|
||||||
}
|
}
|
||||||
|
|
||||||
system "flex -o$tmpfile_c $tmpfile_l";
|
system "flex -o$tmpfile_c $tmpfile_l";
|
||||||
system "cc -O3 -o $tmpfile_x $tmpfile_c -lfl";
|
system "cc -O3 -o $tmpfile_x $tmpfile_c -lfl";
|
||||||
system "$tmpfile_x";
|
system "$tmpfile_x";
|
||||||
|
|
||||||
system "rm -f $tmpfile_l $tmpfile_c $tmpfile_x";
|
#system "rm -f $tmpfile_l $tmpfile_c $tmpfile_x";
|
||||||
|
unlink $tmpfile_l;
|
||||||
|
unlink $tmpfile_c;
|
||||||
|
unlink $tmpfile_x;
|
||||||
|
|
||||||
|
@ -1,5 +1,11 @@
|
|||||||
#!/usr/bin/perl
|
#!/usr/bin/perl
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: tags for utt
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
|
use strict;
|
||||||
use locale;
|
use locale;
|
||||||
|
|
||||||
my $input = <>;
|
my $input = <>;
|
||||||
|
@ -1,13 +1,21 @@
|
|||||||
#!/usr/bin/perl
|
#!/usr/bin/perl
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: tok (tokenizer)
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
|
use strict;
|
||||||
use locale;
|
use locale;
|
||||||
use Getopt::Long;
|
use Getopt::Long;
|
||||||
|
use File::HomeDir;
|
||||||
|
|
||||||
my $interactive=0;
|
my $interactive=0;
|
||||||
my $help;
|
my $help;
|
||||||
|
|
||||||
my $systemconfigfile='/usr/local/etc/utt/tok.conf';
|
my $systemconfigfile='/usr/local/etc/utt/tok.conf';
|
||||||
my $userconfigfile="$ENV{'HOME'}/.utt/tok.conf";
|
#my $userconfigfile="$ENV{'HOME'}/.utt/tok.conf";
|
||||||
|
my $userconfigfile=home()."/.utt/tok.conf";
|
||||||
|
|
||||||
#read configuration files###########################
|
#read configuration files###########################
|
||||||
my $file;
|
my $file;
|
||||||
|
@ -1 +1,8 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
|
||||||
|
#package: UAM Text Tools
|
||||||
|
#component: unfla
|
||||||
|
#version: 1.0
|
||||||
|
#author: Tomasz Obrebski
|
||||||
|
|
||||||
tr '\014' '\012'
|
tr '\014' '\012'
|
||||||
|
Loading…
Reference in New Issue
Block a user