From 1845a3e1d52a7e6bb9657602b08094911000618d Mon Sep 17 00:00:00 2001
From: JT Smith <jt@plainblack.com>
Date: Thu, 13 Jul 2006 15:41:30 +0000
Subject: [PATCH]  - Added Chinese character support to search engine and
 indexer thanks to Zhou    Xiaopeng.

---
 docs/changelog/7.x.x.txt   |  2 ++
 docs/credits.txt           |  2 ++
 lib/WebGUI/Search.pm       | 11 +++++++++++
 lib/WebGUI/Search/Index.pm | 10 ++++++++++
 4 files changed, 25 insertions(+)

diff --git a/docs/changelog/7.x.x.txt b/docs/changelog/7.x.x.txt
index 893c96ebc..312c686e3 100644
--- a/docs/changelog/7.x.x.txt
+++ b/docs/changelog/7.x.x.txt
@@ -6,6 +6,8 @@
    / Procolix)
  - fix: Insert WebGUI Image inserts image, but does not retain border, spacing
    or alignment.(Martin Kamerbeek / Procolix)
+ - Added Chinese character support to search engine and indexer thanks to Zhou
+   Xiaopeng.
 
 7.0.1
  - fix: User profile field "Department" needs i18n 
diff --git a/docs/credits.txt b/docs/credits.txt
index 6cc333a4c..f40acd269 100644
--- a/docs/credits.txt
+++ b/docs/credits.txt
@@ -45,8 +45,10 @@ Contributing Developers..............Lucas Bartholemy
                                      Steve Swanson / Plain Black
                                      Jeff Szpak / Plain Black
                                      Sean Tu / WDI
+                                     Vladimir Vitkovsky / WebGUI Worldwide
                                      Madsen Wikholm
                                      Matt Wilson / Plain Black
+                                     Zhou Xiaopeng / WebGUI Worldwide
                                      Gerald Young
 
 
diff --git a/lib/WebGUI/Search.pm b/lib/WebGUI/Search.pm
index 1b8578711..88336a188 100644
--- a/lib/WebGUI/Search.pm
+++ b/lib/WebGUI/Search.pm
@@ -270,6 +270,17 @@ sub search {
 		unless ($keywords =~ m/"|\*/) { # do wildcards for people, like they'd expect
         		my @terms = split(' ',$keywords);
         		for (my $i = 0; $i < scalar(@terms); $i++) {
+			#-------------- Edited by zxp for Chinese Word Segment
+				utf8::decode($terms[$i]);
+				my @segs = split /([A-z|\d]+|\S)/, $terms[$i];
+				$terms[$i] = join " ",@segs;
+				$terms[$i] =~ s/\s{2,}/ /g;
+				$terms[$i] =~ s/(^\s|\s$)//g;
+				$terms[$i] =~ s/\s/\'\'/g;
+				if($terms[$i] =~ m/\'/) { # has non-latin latter in terms
+					$terms[$i] = '"' . $terms[$i] . '"';
+				}
+			#-------------- Edited by zxp end
                 		$terms[$i] .= "*";
         		}
         		$keywords = join(" ", @terms);
diff --git a/lib/WebGUI/Search/Index.pm b/lib/WebGUI/Search/Index.pm
index cc03715dc..4d2ca0fbb 100644
--- a/lib/WebGUI/Search/Index.pm
+++ b/lib/WebGUI/Search/Index.pm
@@ -112,6 +112,16 @@ sub create {
 	my $description = WebGUI::HTML::filter($asset->get('description'), "all");
 	my $keywords = WebGUI::HTML::filter(join(" ",$asset->get("title"), $asset->get("menuTitle"), $asset->get("synopsis"), $url, $description), "all");
 	my $synopsis = $asset->get("synopsis") || substr($description,0,255) || substr($keywords,0,255);
+
+#-------------------- added by zxp for chinese word segment
+	utf8::decode($keywords);
+	my @segs = split /([A-z|\d]+|\S)/, $keywords;
+	$keywords = join " ",@segs;
+	$keywords =~ s/\s{2,}/ /g;
+	$keywords =~ s/(^\s|\s$)//g;
+	$keywords =~ s/\s/\'\'/g;
+#-------------------- added by zxp end
+
 	my $add = $self->session->db->prepare("insert into assetIndex (assetId, title, url, creationDate, revisionDate, 
 		ownerUserId, groupIdView, groupIdEdit, lineage, className, synopsis, keywords) values (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ? )");
 	$add->execute([$asset->getId, $asset->get("title"), $asset->get("url"), $asset->get("creationDate"),