oqapi/webgui
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Decode HTML entities sent by rich editors. Add tests for the decoding

Browse source 
inside the search indexer.  Gotcha note.  Fixes #10797.
This commit is contained in:
Colin Kuskie 2009-08-21 16:55:57 +00:00
parent 91b37aae36
commit f68db111aa
4 changed files with 43 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

7
lib/WebGUI/Search/Index.pm
View file

@ -15,6 +15,7 @@ package WebGUI::Search::Index;
=cut
use strict;
use HTML::Entities;
=head1 NAME
@ -82,7 +83,6 @@ sub addKeywords {
my $self = shift;
my $text = join(" ", @_);
$text = WebGUI::HTML::filter($text, "all");
$text = $self->_filterKeywords($text);
my ($keywords) = $self->session->db->quickArray("select keywords from assetIndex where assetId=?",[$self->getId]);
$self->session->db->write("update assetIndex set keywords =? where assetId=?", [$keywords.' '.$text, $self->getId]);
@ -166,7 +166,8 @@ sub DESTROY {
=head2 _filterKeywords ( $keywords )
Perform filtering and cleaning up of the keywords before submitting them.
Perform filtering and cleaning up of the keywords before submitting them. Ideographic characters are padded
so that they are still searchable. HTML entities are decoded.
=head3 $keywords
@ -179,6 +180,8 @@ sub _filterKeywords {
my $keywords = shift;
$keywords = WebGUI::HTML::filter($keywords, "all");
$keywords = HTML::Entities::decode_entities($keywords);
utf8::upgrade($keywords);
# split into 'words'. Ideographic characters (such as Chinese) are
# treated as distinct words. Everything else is space delimited.