Changeset 4768

Timestamp:

2003-06-23T14:49:14+12:00 (21 years ago)

Author:

kjdon

Message:

changed the index specification for mgpp colls. now we only build one index, and the contents are separated by spaces, eg
indexes text Title Subject metadata allfields
these are all built into one index. the field list is used inplace of index list in the query form. the order of teh indexes line specifies the order they should appear in the field list.

File:

• trunk/gsdl/perllib/mgppbuilder.pm (modified) (6 diffs)

trunk/gsdl/perllib/mgppbuilder.pm

@@ -63 +63 @@
 
 # change this so a user can add their own ones in via a file or cfg
+#add AND, OR, NOT NEAR to this list - these cannot be used as field names
 %static_indexfield_map = ('Title'=>'TI',
               'TI'=>1,
@@ -69 +70 @@
               'Creator'=>'CR',
               'CR'=>1,
-              'Organization'=>'OR',
-              'OR'=>1,
+              'Organization'=>'ORG',
+              'ORG'=>1,
               'Source'=>'SO',
               'SO'=>1,
@@ -81 +82 @@
               'People'=>'PE',
               'PE'=>1,
+              'AllFields'=>'ZZ',
+              'ZZ'=>1,
               'TextOnly'=>'TX',
-              'TX'=>1);
+              'TX'=>1,
+              'AND'=>1,
+              'OR'=>1,
+              'NOT'=>1,
+              'NEAR'=>1);
 
 sub new {
@@ -115 +122 @@
     $self->{'collect_cfg'} = &colcfg::read_collect_cfg ($colcfgname);
 
+    # sort out the indexes
+    #indexes are specified with spaces, but we put them into one index
+    my $indexes = $self->{'collect_cfg'}->{'indexes'};
+    $self->{'collect_cfg'}->{'indexes'} = [];
+    push (@{$self->{'collect_cfg'}->{'indexes'}}, join(',', @$indexes));
+     
+    
     # sort out subcollection indexes
     if (defined $self->{'collect_cfg'}->{'indexsubcollections'}) {
@@ -402 +416 @@
 
     # the directory name starts with a processed version of index fields
-    my ($pindex) = $self->process_field($fields);
-    # next comes a processed version of the index
-    $pindex = lc ($pindex);
-
+    #my ($pindex) = $self->process_field($fields);
+    #$pindex = lc ($pindex);
+    # now we only ever have one index, and its called 'idx'
+    $pindex = 'idx';
+    
     # next comes a processed version of the subcollection if there is one.
     my $psub = $self->process_field ($subcollection);
@@ -900 +915 @@
     # store the indexfieldmap information
     my @indexfieldmap = ();
-    #add all fields bit - sort based on keys. text only is put at front
-    if (defined $self->{'buildproc'}->{'indexfields'}->{'TextOnly'}) {
-    push (@indexfieldmap, "TextOnly\-\>TX");
-    }
-    foreach $field (sort keys %{$self->{'buildproc'}->{'indexfields'}}) {
-    next if $field eq "TextOnly";
-    push (@indexfieldmap, "$field\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$field}");
+    my @indexfields = ();
+    my $specifiedfields = {};
+    my @specifiedfieldorder = ();
+    # go through the index definition and add each thing to a map, so we can easily check if it is already specified - when doing the metadata, we print out all the individual fields, but some may already be specified in the index definition, so we dont want to add those again.
+    foreach $field (@{$self->{'collect_cfg'}->{'indexes'}}) {
+    my @fs = split(',', $field);
+    foreach $f(@fs) {
+        $specifiedfields->{$f}=1;
+        push (@specifiedfieldorder, "$f");
+    }
+    }
+    
+    #add all fields bit 
+    foreach $field (@specifiedfieldorder) {
+    if ($field eq "metadata") {
+        foreach $newfield (keys %{$self->{'buildproc'}->{'indexfields'}}) {
+        if (!defined $specifiedfields->{$newfield}) {
+            push (@indexfieldmap, "$newfield\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$newfield}");
+            push (@indexfields, "$newfield");
+        }
+        }
+
+    } elsif ($field eq 'text') {
+        push (@indexfieldmap, "TextOnly\-\>TX");
+        push (@indexfields, "TextOnly");
+    } elsif ($field eq 'allfields') {
+        push (@indexfieldmap, "AllFields\-\>ZZ");
+        push (@indexfields, "AllFields");
+    } else {
+        push (@indexfieldmap, "$field\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$field}");
+        push (@indexfields, "$field");
+        
+    }
+    #if (defined $self->{'buildproc'}->{'indexfields'}->{'TextOnly'}) {
+    #push (@indexfieldmap, "TextOnly\-\>TX");
+    #}
+    #foreach $field (sort keys %{$self->{'buildproc'}->{'indexfields'}}) {
+    #next if $field eq "TextOnly";
+    #push (@indexfieldmap, "$field\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$field}");
     }
 
     $build_cfg->{'indexfieldmap'} = \@indexfieldmap;
+    $build_cfg->{'indexfields'} = \@indexfields;
 
     #store the indexed field information
-    foreach $field (sort keys %{$self->{'buildproc'}->{'indexfields'}}) {
+    #foreach $field (sort keys %{$self->{'buildproc'}->{'indexfields'}}) {
     
-    push (@{$build_cfg->{'indexfields'}}, $field);
-    }
+    #push (@{$build_cfg->{'indexfields'}}, $field);
+    #}
     # write out the build information
     &cfgread::write_cfg_file("$self->{'build_dir'}/build.cfg", $build_cfg,