From 7f69041cbb4f26a9d9e280e013673fa41394347d Mon Sep 17 00:00:00 2001 From: dbs Date: Mon, 13 Sep 2010 14:30:10 +0000 Subject: [PATCH] =?utf8?q?Add=20indexing=20normalization=20for=20Polish=20?= =?utf8?q?l=20(=C5=82)?= MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit git-svn-id: svn://svn.open-ils.org/ILS-Contrib/conifer/branches/rel_1_6_1@987 6d9bc8c9-1ec2-4278-b937-99fde70a366f --- src/perlmods/OpenILS/Application/Ingest.pm | 3 +++ src/perlmods/OpenILS/Application/Storage/Driver/Pg/fts.pm | 3 +++ 2 files changed, 6 insertions(+) diff --git a/src/perlmods/OpenILS/Application/Ingest.pm b/src/perlmods/OpenILS/Application/Ingest.pm index 30892b517b..e7e2bf0dfd 100644 --- a/src/perlmods/OpenILS/Application/Ingest.pm +++ b/src/perlmods/OpenILS/Application/Ingest.pm @@ -900,6 +900,9 @@ sub class_index_string_xml { $value =~ s/\pC+//sgo; $value =~ s/\W+$//sgo; + $term =~ s/\x{142}/l/sgo; # Convert Polish l (lowercase) + $term =~ s/\x{141}/l/sgo; # Convert Polish l (uppercase) + # hack to normalize ratio-like strings while ($term =~ /\b\d{1}:[, ]?\d+(?:[ ,]\d+[^:])+/o) { $term = $` . join ('', split(/[, ]/, $&)) . $'; diff --git a/src/perlmods/OpenILS/Application/Storage/Driver/Pg/fts.pm b/src/perlmods/OpenILS/Application/Storage/Driver/Pg/fts.pm index 055266a8fe..42550d54fd 100644 --- a/src/perlmods/OpenILS/Application/Storage/Driver/Pg/fts.pm +++ b/src/perlmods/OpenILS/Application/Storage/Driver/Pg/fts.pm @@ -24,6 +24,9 @@ $term =~ s/(\pM+)//gos; $term =~ s/(\b\.\b)//gos; + $term =~ s/\x{142}/l/sgo; # Convert Polish l (lowercase) + $term =~ s/\x{141}/l/sgo; # Convert Polish l (uppercase) + # hack to normalize ratio-like strings while ($term =~ /\b\d{1}:[, ]?\d+(?:[ ,]\d+[^:])+/o) { $term = $` . join ('', split(/[, ]/, $&)) . $'; -- 2.11.0