Changeset 17110 for gsdl/trunk/perllib/basebuilder.pm

Timestamp:

2008-08-30T07:35:20+12:00 (16 years ago)

Author:

kjdon

Message:

changed way cjk separation is done. Not done in plugins any more, but is now an indexoption. cnseg called from filter_text method. generate_index_options sets up the field in buildproc

File:

• gsdl/trunk/perllib/basebuilder.pm (modified) (4 diffs)

gsdl/trunk/perllib/basebuilder.pm

@@ -148 +148 @@
     
     $self->generate_index_list();
-    $self->generate_index_options();
-
+ 
     # sort out subcollection indexes
     if (defined $self->{'collect_cfg'}->{'indexsubcollections'}) {
@@ -216 +215 @@
     die "$@" if $@;
 
+    
+   $self->generate_index_options();
+
     if (!$self->{'debug'} && !$self->{'keepold'}) {
     # remove any old builds
@@ -234 +236 @@
 }
 
+sub generate_index_options {
+    my $self = shift (@_);
+
+    my $separate_cjk = 0;
+    
+    if (defined($self->{'collect_cfg'}->{'indexoptions'})) {
+    foreach my $option (@{$self->{'collect_cfg'}->{'indexoptions'}}) {
+        if ($option =~ /separate_cjk/) {
+        $separate_cjk = 1;
+        }
+    }
+    }
+    # set this for building
+    $self->{'buildproc'}->set_separate_cjk($separate_cjk);
+    # record it for build.cfg
+    $self->{'separate_cjk'} = $separate_cjk;
+}
+ 
 sub set_sections_index_document_metadata {
     my $self = shift (@_);
@@ -429 +449 @@
     $build_cfg->{'indexstem'} = &util::get_dirsep_tail($self->{'collection'});
     $build_cfg->{'stemindexes'} = $self->{'stemindexes'};
+    if ($self->{'separate_cjk'}) {
+    $build_cfg->{'separate_cjk'} = "true";
+    }
     
     # store the number of documents and number of bytes