sergiotarxz/burguillos.info
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

More suffixes fixes.

Browse Source
This commit is contained in:
Sergiotarxz 2023-09-05 17:54:23 +02:00
parent e50816eee1
commit bf23b12ea8
1 changed files with 72 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

82
lib/BurguillosInfo/IndexUtils.pm
View File

@ -12,21 +12,83 @@ use Unicode::Normalize qw/NFKD/;
use Moo; use Moo;
sub normalize($self, $text) { sub normalize ( $self, $text ) {
return undef if !defined $text; return undef if !defined $text;
my $decomposed = NFKD( $text ); my $decomposed = NFKD($text);
$decomposed =~ s/\p{NonspacingMark}//g; $decomposed =~ s/\p{NonspacingMark}//g;
$decomposed =~ s/(?<=\w{4})(?: $decomposed =~ s/es\b//g;
ada|ado|aje|cion|diccion|duccion|dura|eccion|epcion|ido|miento| $decomposed =~ s/as\b//g;
ncia|scripcion|sicion|sion|dad|tad|bilidad|edad|era|eria|ez|eza|ia|idad|ismo| $decomposed =~ s/os\b//g;
ncia|ante|ente|ura|dor|dero|ero|ista|ado|ario|ia|ero|eria|able|aceo|aco|al|aneo| $decomposed =~ s/e\b//g;
ante|ario|ente|rgir|ento|errimo|ible|ico|ifico|il|ino|isimo|ivo|izo|oso|ecer| $decomposed =~ s/o\b//g;
ificar|izar|es|as|os|e|o|a $decomposed =~ s/a\b//g;
)\b//xg; $decomposed =~ s/aui\b//g;
$decomposed =~ s/ada\b//g;
$decomposed =~ s/ado\b//g;
$decomposed =~ s/aje\b//g;
$decomposed =~ s/cion\b//g;
$decomposed =~ s/diccion\b//g;
$decomposed =~ s/duccion\b//g;
$decomposed =~ s/dura\b//g;
$decomposed =~ s/eccion\b//g;
$decomposed =~ s/epcion\b//g;
$decomposed =~ s/ido\b//g;
$decomposed =~ s/miento\b//g;
$decomposed =~ s/ncia\b//g;
$decomposed =~ s/scripcion\b//g;
$decomposed =~ s/sicion\b//g;
$decomposed =~ s/sion\b//g;
$decomposed =~ s/dad\b//g;
$decomposed =~ s/tad\b//g;
$decomposed =~ s/bilidad\b//g;
$decomposed =~ s/edad\b//g;
$decomposed =~ s/era\b//g;
$decomposed =~ s/eria\b//g;
$decomposed =~ s/ez\b//g;
$decomposed =~ s/eza\b//g;
$decomposed =~ s/ia\b//g;
$decomposed =~ s/idad\b//g;
$decomposed =~ s/ismo\b//g;
$decomposed =~ s/ncia\b//g;
$decomposed =~ s/ante\b//g;
$decomposed =~ s/ente\b//g;
$decomposed =~ s/ura\b//g;
$decomposed =~ s/dor\b//g;
$decomposed =~ s/dero\b//g;
$decomposed =~ s/ero\b//g;
$decomposed =~ s/ista\b//g;
$decomposed =~ s/ado\b//g;
$decomposed =~ s/ario\b//g;
$decomposed =~ s/ia\b//g;
$decomposed =~ s/ero\b//g;
$decomposed =~ s/eria\b//g;
$decomposed =~ s/able\b//g;
$decomposed =~ s/aceo\b//g;
$decomposed =~ s/aco\b//g;
$decomposed =~ s/al\b//g;
$decomposed =~ s/aneo\b//g;
$decomposed =~ s/ante\b//g;
$decomposed =~ s/ario\b//g;
$decomposed =~ s/ente\b//g;
$decomposed =~ s/rgir\b//g;
$decomposed =~ s/ento\b//g;
$decomposed =~ s/errimo\b//g;
$decomposed =~ s/ible\b//g;
$decomposed =~ s/ico\b//g;
$decomposed =~ s/ifico\b//g;
$decomposed =~ s/il\b//g;
$decomposed =~ s/ino\b//g;
$decomposed =~ s/isimo\b//g;
$decomposed =~ s/ivo\b//g;
$decomposed =~ s/izo\b//g;
$decomposed =~ s/oso\b//g;
$decomposed =~ s/ecer\b//g;
$decomposed =~ s/ificar\b//g;
$decomposed =~ s/izar\b//g;
return $decomposed; return $decomposed;
} }
sub n(@args) { sub n (@args) {
normalize(@args); normalize(@args);
} }
1; 1;