Support estimating prefix search

lib/mrb/scripts/expression_size_estimator.rb

@@ -73,6 +73,8 @@ def estimate_data(data)
            Operator::GREATER,
            Operator::GREATER_EQUAL
         size = estimate_range(data, index_column)
+      when Operator::PREFIX
+        size = estimate_prefix(data, index_column)
       when Operator::CALL
         size = estimate_call(data, index_column)
       end
@@ -134,6 +136,30 @@ def estimate_range(data, index_column)
       end
     end
 
+    def estimate_prefix(data, index_column)
+      lexicon = index_column.lexicon
+      n_terms = lexicon.size
+      return 0 if n_terms.zero?
+
+      value = data.query.value
+      limit = n_terms / 0.01
+      if limit < 10
+        limit = 10
+      elsif limit > 1000
+        limit = 1000
+      end
+      options = {
+        :min => value,
+        :limit => limit,
+        :flags => TableCursorFlags::PREFIX,
+      }
+      TableCursor.open(lexicon, options) do |cursor|
+        size = index_column.estimate_size(:lexicon_cursor => cursor)
+        size += 1 if cursor.next != ID::NIL
+        size
+      end
+    end
+
     def estimate_call(data, index_column)
       procedure = data.args[0]
       arguments = data.args[1..-1].collect do |arg|

test/mruby/suite/query_optimizer/test_estimate_size.rb

@@ -198,6 +198,37 @@ def test_have_record
     end
   end
 
+  class TestPrefixSearch < self
+    def setup
+      Groonga::Schema.define do |schema|
+        schema.create_table("Logs") do |table|
+          table.short_text("user")
+        end
+
+        schema.create_table("Users",
+                            :type => :patricia_trie,
+                            :key_type => :short_text) do |table|
+          table.index("Logs", "user")
+        end
+      end
+      super
+    end
+
+    def test_no_record
+      assert_equal(0, estimate_size("user @^ 'al'"))
+    end
+
+    def test_have_record
+      @logs.add(:user => "bob")
+      @logs.add(:user => "alice")
+      @logs.add(:user => "alian")
+      @logs.add(:user => "alice")
+      @logs.add(:user => "alfread")
+      @logs.add(:user => "calros")
+      assert_equal(6, estimate_size("user @^ 'al'"))
+    end
+  end
+
   class TestBetweenSearch < self
     def setup
       Groonga::Schema.define do |schema|