Changeset 120

Timestamp:

12/19/06 21:54:21 (2 years ago)

Author:

jk

Message:

r2825@monsoon: jk | 2006-12-19 22:54:09 +0100
work arount the annoying stop words-with tokenized-fields-problem - query terms not scoped to a specific field will not be queried against untokenized fields any more

Files:

• trunk/.todo (added)
• trunk/demo/test/unit/comment_test.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/lib/class_methods.rb (modified) (3 diffs)

trunk/demo/test/unit/comment_test.rb

@@ -160 +160 @@
-  end
-
-  # fails due to Ferret bug in 0.10.13, http://pastie.caboo.se/22886
-  def test_stopwords
-    comment = Comment.create( :author => 'john doe', :content => 'Move or shake' )

trunk/plugin/acts_as_ferret/lib/class_methods.rb

@@ -93 +93 @@
-          ferret_configuration = {
-            :or_default => false, 
-
+
+
-            #:max_clauses => 512,
-            #:default_field => '*',
-            #:analyzer => Ferret::Analysis::StandardAnalyzer.new,
-            # :wild_card_downcase => true
@@ -142 +142 @@
-            EOV
-          FerretMixin::Acts::ARFerret::ensure_directory configuration[:index_dir]
+
+          # now that all fields have been added, we can initialize the default
+          # field list to be used by the query parser.
+          # It will include all content fields *not* marked as :untokenized.
+          # This fixes the otherwise failing CommentTest#test_stopwords 
+          ferret_configuration[:default_field] = fields_for_ferret.keys.select do |f| 
+            fields_for_ferret[f][:index] != :untokenized
+          end
+          logger.debug "set default field list to #{ferret_configuration[:default_field].inspect}"
-        end
-        
@@ -399 +408 @@
-          result = []
-          index = self.ferret_index
+          # puts "query: #{index.process_query q}"
-          total_hits = index.search_each(q, options) do |hit, score|
-            # only collect result data if we intend to return it