From: Dobrica Pavlinusic <dpavlin@rot13.org>
Date: Tue, 14 Nov 2023 09:43:07 +0000 (+0100)
Subject: similarity 0.9, merge all suggestions
X-Git-Url: http://git.rot13.org/?a=commitdiff_plain;h=543260c582708142468b530844363888ee09c8cf;p=csv-join-similarity

similarity 0.9, merge all suggestions
---

diff --git a/upari.pl b/upari.pl
index 9366a8e..72ef4aa 100755
--- a/upari.pl
+++ b/upari.pl
@@ -49,22 +49,23 @@ foreach my $nr ( 1 .. 4 ) {
 			$stat->{error}->{new_exact}->{$nr}++;
 
 			my @candidates;
+			my $limit_sim = 0.9;
 			foreach my $key ( keys %{ $keys->{ $key_id } } ) {
-				my $s = similarity $num, $key, 0.4;
+				my $s = similarity $num, $key, $limit_sim;
 				#warn "# $num $key $s\n";
-				if ($s > 0.8 ) {
+				if ($s > $limit_sim ) {
 					push @candidates, { key => $key, s => $s };
 				}
 			}
-			if ( $#candidates == 0 ) {
-				$stat->{similarity}->{$nr}++;
-				my $new_num = $candidates[0]->{key};
-				$id = join('-',uc $new_num, $row->[1], $row->[2]);
-				print "SIMILARITY $nr $num -> $new_num\n";
-				push @{ $keys->{ $key_id }->{ $id } }, uc $id;
-			} elsif ( $#candidates > 1 ) {
-				warn "# similarity IGNORED $num = ",dump( @candidates ),$/;
-				$stat->{similarity_multiple}->{$nr}++;
+			if ( @candidates ) {
+				my $multi =  $#candidates > 0 ? 'multi' : '';
+				$stat->{ 'similarity' . $multi }->{$nr}++;
+				foreach my $i ( 0 .. $#candidates ) {
+					my $new_num = $candidates[$i]->{key};
+					$id = join('-',uc $new_num, $row->[1], $row->[2]);
+					print "SIMILARITY$multi $nr $num -> $new_num\n";
+					push @{ $keys->{ $key_id }->{ $id } }, uc $id;
+				}
 			} else {
 				$stat->{similarity_none}->{$nr}++;
 				push @{ $keys->{ $key_id }->{ $num } }, uc $row->[0];