Changeset 29176

Timestamp:

2014-08-08T16:57:34+12:00 (10 years ago)

Author:

ak19

Message:

For Solr: 1. collConfig.pm now reads any option sub-elements for index elements defined in collectionConfig.xml. 2. The solrfieldtype option sub-element to an index element can now be set by hand in collectionConfig.xml and its value will be used in the solr collection/etc/conf/schema.xml file for that index in that collection, instead of the old default of text_en_splitting for all fields.

Files:

• gs3-extensions/solr/trunk/src/perllib/solrbuilder.pm (modified) (2 diffs)
• main/trunk/greenstone2/perllib/collConfigxml.pm (modified) (4 diffs)

gs3-extensions/solr/trunk/src/perllib/solrbuilder.pm

@@ -282 +282 @@
     foreach my $ifm (@{$self->{'build_cfg'}->{'indexfieldmap'}}) {
 
-        my ($field) = ($ifm =~ m/^.*->(.*)$/);
+        my ($fullfieldname, $field) = ($ifm =~ m/^(.*)->(.*)$/);
 
         $schema_insert_xml .= "    "; # indent
@@ -297 +297 @@
         else
         {
-            $schema_insert_xml .= "type=\"text_en_splitting\" ";
+            #$schema_insert_xml .= "type=\"text_en_splitting\" ";
+
+            # original default solr field type for all fields is text_en_splitting
+            my $solrfieldtype = "text_en_splitting";
+            if(defined $self->{'collect_cfg'}->{'indexfieldoptions'}->{$fullfieldname}->{'solrfieldtype'}) {    
+            $solrfieldtype = $self->{'collect_cfg'}->{'indexfieldoptions'}->{$fullfieldname}->{'solrfieldtype'};
+            #print STDERR "@@@@#### found TYPE: $solrfieldtype\n";
+            }
+            $schema_insert_xml .= "type=\"$solrfieldtype\" ";
+            
         }
         # set termVectors=\"true\" when term vectors info is required, 

main/trunk/greenstone2/perllib/collConfigxml.pm

@@ -53 +53 @@
            "defaultIndexLanguage" => "defaultlanguage",
            "index" => "indexes",
+           "indexfieldoptions" => "indexfieldoptions",
            "sort" => "sortfields",
            "facet" => "facetfields", 
@@ -86 +87 @@
 # ("plugin", "option")
 my $currentIndex = 0;
-my $arrayexp = q/^(index|sort|facet|level|indexOption|indexSubcollection|indexLanguage|orthogonalBuildTypes)$/; 
+
+my $structexp = q/^(index)$/;
+# structexp contains a hashmap of option(name, value) pairs per index name like allfields/ZZ or titles/TI
+# e.g. <index name="allfields">
+#    <displayItem ... />
+#    <option name="solrfieldtype" value="text_ja" />
+#      </index>
+
+my $arrayexp = q/^(sort|facet|level|indexOption|indexSubcollection|indexLanguage|orthogonalBuildTypes)$/; 
+#my $arrayexp = q/^(index|sort|facet|level|indexOption|indexSubcollection|indexLanguage|orthogonalBuildTypes)$/; 
 my $arrayarrayexp = q/^(plugin|classifier)$/; #|buildOption)$/;
 my $hashexp = q/^(subcollection)$/; # add other element names that should be represented by hash expressions here
@@ -199 +209 @@
     #}
  
-    #@ Handle index|sort|facet|level|indexOption|indexSubcollection|indexLanguage 
+    #@ Handle sort|facet|level|indexOption|indexSubcollection|indexLanguage 
     elsif ($element =~ /$arrayexp/) {
       my $key = $nameMap->{$element};   # 
@@ -210 +220 @@
       }
     }
+
+    #@ Handle index which can have options as children to be put in a map: <option name="name" value="value" />
+    elsif ($element =~ /$structexp/) {
+    # find the gs2 mapping name
+        $currentLevel = $element;
+    
+    # for GS2, 'indexes' should be an arrayexp, so maintain that part of the code as it is  
+    my $key = $nameMap->{$element}; # 'indexes'
+    if (!defined $data->{$key}) {
+        $data->{$key} = [];     
+    }
+    
+    if (defined $name) {
+        push (@{$data->{$key}},$name);      
+    }
+    }
+
+    #@ Handling the option elements in each index structure, if any, only for GS2
+    elsif ($currentLevel =~ /$structexp/ && $element eq "option") {
+    # find the gs2 mapping name for classifier and plugin
+    my $key = $nameMap->{$currentLevel."fieldoptions"}; # my $key = $currentLevel."fieldoptions"; # indexfieldoptions 
+
+    # The last element of the 'indexes' array contains the name of the index currently being processed
+    # e.g. "allfields"
+    my $indexKey = $nameMap->{$currentLevel}; # 'indexes'
+    my $arrSize = scalar( @{$data->{$indexKey}} ); # length of 'indexes' array
+    my $indexName = @{$data->{$indexKey}}[$arrSize-1]; # name of index currently being processed in prev elsif
+
+    if (!defined $data->{$key}) {
+        $data->{$key} = {}; # 'indexoptions' is a new hashmap
+    }   
+    if (defined $name and $name =~ /\w/ && defined $value and $value =~ /\w/) {
+        # we have a name and value to this option, add them as options associated with the current index
+        
+        if (!defined $data->{$key}->{$indexName}) {
+        $data->{$key}->{$indexName} = {}; # indexoptions -> allfields is a new hashmap
+        }
+        
+        $data->{$key}->{$indexName}->{$name} = $value;
+        
+        #print STDERR "@@@ Found: Value: data->{'indexfieldoptions'}->{$indexName}->{$name}: " . $data->{'indexfieldoptions'}->{$indexName}->{$name} . "\n";        
+    }
+    }
+
     # importOption and buildOption, just stored at top level, name=value, 
     # as per gs2 version