From ea51ba559eb1f1831864daf7b14d566708e6c96d Mon Sep 17 00:00:00 2001
From: Colin Kuskie <colink@perldreamer.com>
Date: Fri, 21 Aug 2009 21:29:43 +0000
Subject: [PATCH] Add tests for ideogram searching. ft_min_word_len must be 2
 or less.

---
 docs/gotcha.txt            |  5 +++++
 lib/WebGUI/Search.pm       |  2 +-
 lib/WebGUI/Search/Index.pm |  2 +-
 t/Search.t                 | 30 +++++++++++++++++++++++++++++-
 4 files changed, 36 insertions(+), 3 deletions(-)

diff --git a/docs/gotcha.txt b/docs/gotcha.txt
index a12b567a2..e2a969aa9 100644
--- a/docs/gotcha.txt
+++ b/docs/gotcha.txt
@@ -18,6 +18,11 @@ save you many hours of grief.
       or, to index all of the sites on the server:
         perl search.pl --indexall
 
+    * For searching content with ideograms, ft_min_word_len must be
+      set to 2 in the Mysql config file.  WebGUI will safely pad the
+      characters so that it does not need to be set to 1.  The WRE
+      has ft_min_word_len set to 2 by default.
+
 7.7.18
 --------------------------------------------------------------------
     * The search indexer was not properly indexing non-ASCII content.  The
diff --git a/lib/WebGUI/Search.pm b/lib/WebGUI/Search.pm
index b830b521f..7d401946c 100644
--- a/lib/WebGUI/Search.pm
+++ b/lib/WebGUI/Search.pm
@@ -383,7 +383,7 @@ sub search {
             for my $term (@terms) {
                 # we add padding to ideographic characters to avoid minimum word length limits on indexing
                 if ($term =~ /\p{Ideographic}/) {
-                    $term = qq{''$term''};
+                    $term = q{''}.$term.q{''};
                 }
                 $term .= q{*};
                 next
diff --git a/lib/WebGUI/Search/Index.pm b/lib/WebGUI/Search/Index.pm
index 08605afea..d12612a0e 100644
--- a/lib/WebGUI/Search/Index.pm
+++ b/lib/WebGUI/Search/Index.pm
@@ -193,7 +193,7 @@ sub _filterKeywords {
         $word =~ s/\p{isPunct}+\z//;
         # we add padding to ideographic characters to avoid minimum word length limits on indexing
         if ($word =~ /\p{Ideographic}/) {
-            $word = qq{''$word''};
+            $word = q{''}.$word.q{''};
         }
     }
 
diff --git a/t/Search.t b/t/Search.t
index 31ce4a8ca..cb22ef303 100644
--- a/t/Search.t
+++ b/t/Search.t
@@ -17,6 +17,7 @@ use FindBin;
 use strict;
 use lib "$FindBin::Bin/lib";
 use Test::More;
+use Test::Deep;
 use WebGUI::Test; # Must use this before any other WebGUI modules
 use WebGUI::Session;
 
@@ -28,12 +29,13 @@ my $session         = WebGUI::Test->session;
 #----------------------------------------------------------------------------
 # Tests
 
-plan tests => 7;        # Increment this number for each test you create
+plan tests => 10;        # Increment this number for each test you create
 
 #----------------------------------------------------------------------------
 # put your tests here
 
 use_ok('WebGUI::Search');
+use_ok('WebGUI::Search::Index');
 
 my $search = WebGUI::Search->new($session);
 
@@ -50,6 +52,32 @@ ok(  $search->_isStopword('anybody+'),       '_isStopword: regex metacharacter +
 ok(  $search->_isStopword('maybe?'),         '_isStopword: regex metacharacter ? does not crash the search');
 ok(! $search->_isStopword('private.+'),      '_isStopword: regex metacharacters .+ do not crash the search');
 
+################################################
+#
+# Chinese ideograph handling
+#
+################################################
+{
+    use utf8;
+
+    # Create an article to index
+    my $article         = WebGUI::Asset->getImportNode( $session )->addChild( {
+        className       => 'WebGUI::Asset::Wobject::Article',
+        title           => 'Chinese ideograph experiment',
+        description     => "甲骨文",
+    } );
+    my $tag = WebGUI::VersionTag->getWorking( $session );
+    $tag->commit;
+    WebGUI::Test->tagsToRollback($tag);
+    WebGUI::Search::Index->create( $article );
+    my $searcher = WebGUI::Search->new($session);
+    my $assetIds = $searcher->search({ keywords => "甲", })->getAssetIds;
+    cmp_deeply( $assetIds, [ $article->getId ], 'basic test for search works');
+    my $searcher = WebGUI::Search->new($session);
+    my $assetIds = $searcher->search({ keywords => "Chinese", })->getAssetIds;
+    cmp_deeply( $assetIds, [ $article->getId ], 'ideograph search works');
+}
+
 #----------------------------------------------------------------------------
 # Cleanup
 END {