Changeset 24460 for main/trunk/greenstone2

Timestamp:

2011-08-24T15:52:02+12:00 (13 years ago)

Author:

davidb

Message:

Code changes to support indexers that are provided through the extension mechanism

Location:

main/trunk/greenstone2/perllib

Files:

• basebuilder.pm (modified) (2 diffs)
• lucenebuilder.pm (modified) (7 diffs)
• lucenebuildproc.pm (modified) (2 diffs)
• mgppbuilder.pm (modified) (5 diffs)

main/trunk/greenstone2/perllib/basebuilder.pm

@@ -329 +329 @@
     my $outhandle = $self->{'outhandle'};
 
+    $self->pre_build_indexes();
+
     my $indexes = [];
     if (defined $indexname && $indexname =~ /\w/) {
@@ -352 +354 @@
     }
 
-    $self->build_indexes_extra();
-
+    $self->post_build_indexes();
+
+}
+
+# implement this in subclass if want to do extra stuff at before building
+# all the indexes
+sub pre_build_indexes {
+    my $self = shift(@_);
+    my ($indexname) = @_; # optional parameter
 }
 
 # implement this in subclass if want to do extra stuff at the end of building
 # all the indexes
-sub build_indexes_extra {
-    my $self = shift(@_);
-    
+sub post_build_indexes {
+    my $self = shift(@_);   
 }
 

main/trunk/greenstone2/perllib/lucenebuilder.pm

@@ -26 +26 @@
 ###########################################################################
 # /*
-#  *  @version 1.0 ?
+#  *  @version 1.0 Automated incremental building
 #  *  @version 2.0 Incremental building assistance added, including
 #  *               remove_document_from_database which implements the granddad's
@@ -33 +33 @@
 #  *               perl script is. John Rowe
 #  *
+#  *  @authro Waikato Digital Library Research Group
 #  *  @author John Rowe, DL Consulting Ltd.
 #  */
@@ -117 +118 @@
     }
     }       
-    
 }
 
@@ -227 +227 @@
     my $outhandle = $self->{'outhandle'};
 
+    $self->pre_build_indexes($indexname);
+
     my $indexes = [];
     if (defined $indexname && $indexname =~ /\w/) {
@@ -233 +235 @@
     $indexes = $self->{'collect_cfg'}->{'indexes'};
     }
-    # have we got para index?
+
+    # Have we got para index?
     foreach my $level (keys %{$self->{'levels'}}) {
     if ($level =~ /paragraph/) {
-        print $outhandle "Warning: Paragraph level indexing not supported by Lucene\n";
+        print $outhandle "Warning: Paragraph level indexing not supported by Lucene/Solr\n";
         last;
     }
     }
-    # create the mapping between the index descriptions
+
+    # Create the mapping between the index descriptions
     # and their directory names (includes subcolls and langs)
     $self->{'index_mapping'} = $self->create_index_mapping ($indexes);
@@ -246 +250 @@
     # build each of the indexes
     foreach my $index (@$indexes) {
+
     if ($self->want_built($index)) {
 
@@ -271 +276 @@
     }
 
-    #define the final field lists
-    $self->make_final_field_list();
+    $self->post_build_indexes();
 }
 

main/trunk/greenstone2/perllib/lucenebuildproc.pm

@@ -239 +239 @@
         } # for each field in this one index
         
+
         # now we add the text and/or metadata into new_text
         if ($section_text ne "" || scalar(@metadata_list)) {
@@ -283 +284 @@
         
     } # foreach field
-
 
     if ($all_metadata_specified) {

main/trunk/greenstone2/perllib/mgppbuilder.pm

@@ -282 +282 @@
 
 
-sub build_indexes_extra {
+sub post_build_indexes {
     my $self = shift(@_);
+
     #define the final field lists
     $self->make_final_field_list();
@@ -617 +618 @@
     my $collection_infodb = shift(@_);
 
-    # define the indexed field mapping if not already done so (ie if infodb called separately from build_index)
+    # define the indexed field mapping if not already done so 
+    # (i.e. if infodb called separately from build_index)
     if (!defined $self->{'build_cfg'}) {
     $self->read_final_field_list();
@@ -703 +705 @@
 # at the end of building, we have an indexfieldmap with all the mappings,
 # plus some extras, and indexmap with any indexes in it that weren't
-# specified in the index definition.  we want to make an ordered list of
-# fields that are indexed, and a list of mappings that are used. this will
+# specified in the index definition.  We want to make an ordered list of
+# fields that are indexed, and a list of mappings that are used. This will
 # be used for the build.cfg file, and for collection meta definition we
 # store these in a build.cfg bit
@@ -763 +765 @@
         push (@indexfields, "$field");
         }
-
-        
     }
     }
@@ -771 +771 @@
     $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
     }
+
     if (scalar @indexfields) {
     $self->{'build_cfg'}->{'indexfields'} = \@indexfields;