--- trunk2/all2all.pl 2004/10/26 17:44:18 545 +++ trunk2/all2all.pl 2004/10/26 18:59:38 546 @@ -11,6 +11,7 @@ use YAML; use Carp; use Getopt::Long; +use Text::Unaccent 1.02; use lib './lib'; use WebPAC; @@ -46,6 +47,25 @@ }, }; +## remove accented characters +# +sub unac { + my $string = shift || return; + $string = unac_string($code_page,$string); + $string =~ tr/ðÐ/dD/; + return $string; +} +sub unac_2 { + my $string = shift || return; + if (length($string) > 2) { + my $pr = substr($string,0,2); + $string = unac_string($code_page,substr($string,2)); + $string =~ tr/ðÐ/dD/; + $string = $pr . $string; + } + return lc($string); +} + # create WebPAC object # my $webpac = new WebPAC( @@ -150,7 +170,7 @@ #path => $f, path => $webpac->mfn, headline => $h, - words => $words, + words => unac($words), ); } @@ -162,8 +182,9 @@ foreach my $h (@{$ds->{'index'}}) { $thes->{$ds->{'tag'}}->insert( - headline => $h, + sort_by => unac_2($h), mfn => $webpac->mfn, + headline => $h, ); } }