From 9f309be709880281b2f30f19d637e6a866ad4b04 Mon Sep 17 00:00:00 2001
From: sergiotarxz <sergiotarxz@posteo.net>
Date: Tue, 5 Sep 2023 19:29:30 +0200
Subject: [PATCH] Adding a Stemmer insteado of suffixes.

---
 Build.PL                         |  1 +
 lib/BurguillosInfo/IndexUtils.pm | 50 +++++++-------------------------
 2 files changed, 11 insertions(+), 40 deletions(-)

diff --git a/Build.PL b/Build.PL
index 1b7570a..5d0170a 100755
--- a/Build.PL
+++ b/Build.PL
@@ -25,6 +25,7 @@ my $build = Module::Build->new(
         'Moo::Role'                    => 0,
         'Module::Pluggable'            => 0,
         'List::AllUtils'               => 0,
+        'Lingua::Stem::Snowball'       => 0,
     },
 );
 $build->create_build_script;
diff --git a/lib/BurguillosInfo/IndexUtils.pm b/lib/BurguillosInfo/IndexUtils.pm
index 828be48..a6c78ef 100644
--- a/lib/BurguillosInfo/IndexUtils.pm
+++ b/lib/BurguillosInfo/IndexUtils.pm
@@ -12,49 +12,19 @@ use Unicode::Normalize qw/NFKD/;
 
 use Moo;
 
-sub normalize ( $self, $text ) {
+use Lingua::Stem::Snowball;
+
+sub normalize($self, $text) {
     return undef if !defined $text;
     my $decomposed = NFKD($text);
     $decomposed =~ s/\p{NonspacingMark}//g;
-    $decomposed =~ s/es\b//g;
-    $decomposed =~ s/as\b//g;
-    $decomposed =~ s/os\b//g;
-    $decomposed =~ s/e\b//g;
-    $decomposed =~ s/o\b//g;
-    $decomposed =~ s/a\b//g;
-    $decomposed =~ s/i\b//g;
-    $decomposed =~ s/cion\b//g;
-    $decomposed =~ s/diccion\b//g;
-    $decomposed =~ s/duccion\b//g;
-    $decomposed =~ s/dur\b//g;
-    $decomposed =~ s/eccion\b//g;
-    $decomposed =~ s/epcion\b//g;
-    $decomposed =~ s/mient\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/idad\b//g;
-    $decomposed =~ s/ism\b//g;
-    $decomposed =~ s/ant\b//g;
-    $decomposed =~ s/ent\b//g;
-    $decomposed =~ s/dor\b//g;
-    $decomposed =~ s/der\b//g;
-    $decomposed =~ s/ist\b//g;
-    $decomposed =~ s/abl\b//g;
-    $decomposed =~ s/ant\b//g;
-    $decomposed =~ s/ent\b//g;
-    $decomposed =~ s/rgir\b//g;
-    $decomposed =~ s/ent\b//g;
-    $decomposed =~ s/errim\b//g;
-    $decomposed =~ s/ibl\b//g;
-    $decomposed =~ s/ific\b//g;
-    $decomposed =~ s/isim\b//g;
-    $decomposed =~ s/ecer\b//g;
-    $decomposed =~ s/ific\b//g;
+    my @words;
+    while ($decomposed =~ /\b(\w+)\b/g) {
+        push @words, $1;
+    }
+    my $stemmer = Lingua::Stem::Snowball->new( lang => 'es' );
+    $stemmer->stem_in_place(\@words);
+    $decomposed = join " ", @words;
     return $decomposed;
 }