--- trunk2/all2all.pl 2004/10/26 20:53:03 549
+++ trunk2/all2all.pl 2004/10/30 23:56:57 562
@@ -36,15 +36,25 @@
my $filter = {
'CROVOC' => sub {
my $tmp = shift || return;
- return undef unless ($tmp =~ s/\s*CROVOC.*$/ */);
+ return undef unless ($tmp =~ s/\s*CROVOC.*$/ #C#/);
+ # remove repeating stars
+ # FIXME this should be fixed at right place, not here!
+ $tmp =~ s/(\s*#C#)+/ #C#/g;
return $tmp;
},
'CROVOC_tree' => sub {
my $tmp = shift || return;
- $tmp =~ s/\s*CROVOC.*$/ */;
+ $tmp =~ s/\s*CROVOC.*$/ /;
$tmp =~ s/\s*EUROVOC.*//;
return $tmp;
},
+ # TT filter
+ 'CROVOC_img' => sub {
+ my $tmp = shift;
+ $tmp =~ s/\s*#C#\s*/ /gis;
+ $tmp =~ s/"img\/crovoc.png"/"..\/img\/crovoc.png"/gis;
+ return $tmp;
+ }
};
## remove accented characters
@@ -53,6 +63,7 @@
my $string = shift || return;
$string = unac_string($code_page,$string);
$string =~ tr/ðÐ/dD/;
+ $string = unac_entities($string);
return $string;
}
sub unac_2 {
@@ -63,13 +74,21 @@
$string =~ tr/ðÐ/dD/;
$string = $pr . $string;
}
- $string =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi;
- $string =~ s/ð/d/g;
- $string =~ s/&E[tT][hH];/D/g;
- $string =~ s/&(\w\w)lig;/$1/gi;
+ $string = unac_entities($string);
return lc($string);
}
+sub unac_entities {
+ my $ent = shift || return;
+
+ $ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi;
+ $ent =~ s/ð/d/g;
+ $ent =~ s/&E[tT][hH];/D/g;
+ $ent =~ s/&(\w\w)lig;/$1/gi;
+
+ return $ent;
+}
+
# create WebPAC object
#
my $webpac = new WebPAC(
@@ -87,7 +106,7 @@
my $index = new WebPAC::jsFind(
index_path => $index_path,
- keys => 10,
+ keys => 62,
) || die;
my $thes;