RubyGems - rroonga - Versions diffs - 4.0.0 → 4.0.1 - Mend

rroonga 4.0.0 → 4.0.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (28) hide show

checksums.yaml +4 -4
data/.yardopts +6 -0
data/doc/text/news.textile +26 -0
data/ext/groonga/rb-grn-column.c +42 -2
data/ext/groonga/rb-grn-index-column.c +57 -159
data/ext/groonga/rb-grn-object.c +7 -0
data/ext/groonga/rb-grn-table.c +19 -3
data/ext/groonga/rb-grn-utils.c +4 -1
data/ext/groonga/rb-grn-variable-size-column.c +405 -5
data/ext/groonga/rb-grn.h +18 -1
data/lib/groonga/dumper.rb +46 -22
data/lib/groonga/expression-builder.rb +2 -4
data/lib/groonga/schema.rb +16 -234
data/rroonga-build.rb +3 -3
data/rroonga.gemspec +1 -1
data/test/test-array.rb +6 -5
data/test/test-column.rb +0 -14
data/test/test-database.rb +6 -5
data/test/test-double-array-trie.rb +6 -5
data/test/test-expression.rb +2 -2
data/test/test-hash.rb +6 -5
data/test/test-index-column.rb +163 -33
data/test/test-patricia-trie.rb +6 -5
data/test/test-schema-dumper.rb +39 -7
data/test/test-table-dumper.rb +46 -1
data/test/test-table-select-weight.rb +12 -0
data/test/test-variable-size-column.rb +87 -3
metadata +140 -139

data/rroonga-build.rb CHANGED Viewed

@@ -19,15 +19,15 @@ module RroongaBuild
   module RequiredGroongaVersion
     MAJOR = 4
     MINOR = 0
-    MICRO = 0
+    MICRO = 1
     VERSION = [MAJOR, MINOR, MICRO]
-    RELEASED_DATE = Time.utc(2014, 2, 9)
+    RELEASED_DATE = Time.utc(2014, 3, 29)
   end
   module LatestGroongaVersion
     MAJOR = 4
     MINOR = 0
-    MICRO = 0
+    MICRO = 1
     VERSION = [MAJOR, MINOR, MICRO]
   end

data/rroonga.gemspec CHANGED Viewed

@@ -59,7 +59,7 @@ Gem::Specification.new do |s|
   description = clean_white_space.call(entries[entries.index("Description") + 1])
   s.summary, s.description, = description.split(/\n\n+/, 3)
-  s.files = ["README.textile", "AUTHORS", "Rakefile", "Gemfile"]
+  s.files = ["README.textile", "AUTHORS", "Rakefile", "Gemfile", ".yardopts"]
   s.files += Dir.glob("doc/text/*.textile")
   s.files += ["#{s.name}.gemspec"]
   s.files += ["rroonga-build.rb", "extconf.rb"]

data/test/test-array.rb CHANGED Viewed

@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2013  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -117,11 +117,12 @@ class ArrayTest < Test::Unit::TestCase
     users = Groonga::Array.create(:name => "Users")
     users.define_column("name", "ShortText")
     users.define_column("address", "ShortText")
-    1000.times do |i|
-      users.add(:name => "user #{i}" * 1000,
-                :address => "address #{i}" * 1000)
+    large_data = "x" * (2 ** 16)
+    100.times do |i|
+      users.add(:name => "user #{i}" + large_data,
+                :address => "address #{i}" + large_data)
     end
-    assert_equal(3, users.defrag)
+    assert_equal(2, users.defrag)
   end
   def test_rename

data/test/test-column.rb CHANGED Viewed

@@ -93,20 +93,6 @@ class ColumnTest < Test::Unit::TestCase
     assert_equal([@bookmarks_content], @bookmarks_index_content.sources)
   end
-  def test_update_index_column
-    groonga = @bookmarks.add
-    groonga["content"] = "<html><body>groonga</body></html>"
-    ruby = @bookmarks.add
-    ruby["content"] = "<html><body>ruby</body></html>"
-    @bookmarks_index_content[groonga.id] = groonga["content"]
-    @bookmarks_index_content[ruby.id] = ruby["content"]
-    assert_content_search([groonga], "groonga")
-    assert_content_search([groonga, ruby], "html")
-  end
   def test_range
     assert_equal(context[Groonga::Type::SHORT_TEXT], @bookmarks_uri.range)
     assert_equal(context[Groonga::Type::TEXT], @bookmarks_comment.range)

data/test/test-database.rb CHANGED Viewed

@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2013  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -146,11 +146,12 @@ class DatabaseTest < Test::Unit::TestCase
       end
     end
     users = context["Users"]
-    1000.times do |i|
-      users.add(:name => "user #{i}" * 1000,
-                :address => "address #{i}" * 1000)
+    large_data = "x" * (2 ** 16)
+    100.times do |i|
+      users.add(:name => "user #{i}" + large_data,
+                :address => "address #{i}" + large_data)
     end
-    assert_equal(3, @database.defrag)
+    assert_equal(2, @database.defrag)
   end
   def test_tables

data/test/test-double-array-trie.rb CHANGED Viewed

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2011-2012  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2011-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -158,12 +158,13 @@ class DoubleArrayTrieTest < Test::Unit::TestCase
                                             :key_type => "ShortText")
     users.define_column("name", "ShortText")
     users.define_column("address", "ShortText")
-    1000.times do |i|
+    large_data = "x" * (2 ** 16)
+    100.times do |i|
       users.add("user #{i}",
-                :name => "user #{i}" * 1000,
-                :address => "address #{i}" * 1000)
+                :name => "user #{i}" + large_data,
+                :address => "address #{i}" + large_data)
     end
-    assert_equal(3, users.defrag)
+    assert_equal(2, users.defrag)
   end
   def test_rename

data/test/test-expression.rb CHANGED Viewed

@@ -37,7 +37,7 @@ class ExpressionTest < Test::Unit::TestCase
   def test_get_value
     users = Groonga::Hash.create(:name => "Users")
-    name = users.define_column("name", "ShortText")
+    users.define_column("name", "ShortText")
     morita = users.add("morita", :name => "mori daijiro")
@@ -51,7 +51,7 @@ class ExpressionTest < Test::Unit::TestCase
   def test_get_value_with_variable
     users = Groonga::Hash.create(:name => "Users")
-    name = users.define_column("name", "ShortText")
+    users.define_column("name", "ShortText")
     morita = users.add("morita", :name => "mori daijiro")
     gunyara_kun = users.add("gunyara-kun", :name => "Tasuku SUENAGA")

data/test/test-hash.rb CHANGED Viewed

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2009-2013  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -367,12 +367,13 @@ class HashTest < Test::Unit::TestCase
                                  :key_type => "ShortText")
     users.define_column("name", "ShortText")
     users.define_column("address", "ShortText")
-    1000.times do |i|
+    large_data = "x" * (2 ** 16)
+    100.times do |i|
       users.add("user #{i}",
-                :name => "user #{i}" * 1000,
-                :address => "address #{i}" * 1000)
+                :name => "user #{i}" + large_data,
+                :address => "address #{i}" + large_data)
     end
-    assert_equal(3, users.defrag)
+    assert_equal(2, users.defrag)
   end
   def test_rename

data/test/test-index-column.rb CHANGED Viewed

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2009-2013  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -55,40 +55,24 @@ class IndexColumnTest < Test::Unit::TestCase
   class CRUDTest < self
     setup
     def setup_schema
-      @articles = Groonga::Array.create(:name => "Articles")
-      @articles.define_column("content", "Text")
-      terms = Groonga::Hash.create(:name => "Terms",
-                                   :key_type => "ShortText",
-                                   :default_tokenizer => "TokenBigram")
-      @index = terms.define_index_column("content", @articles,
-                                         :with_position => true,
-                                         :with_section => true)
-    end
-    def test_array_set_with_record
-      content = <<-EOC
-      groonga は組み込み型の全文検索エンジンライブラリです。
-      DBMSやスクリプト言語処理系等に組み込むことによって、その
-      全文検索機能を強化することができます。また、リレーショナ
-      ルモデルに基づくデータストア機能を内包しており、groonga
-      単体でも高速なデータストアサーバとして使用することができ
-      ます。
-      ■全文検索方式
-      転置索引型の全文検索エンジンです。転置索引は圧縮されてファ
-      イルに格納され、検索時のディスク読み出し量を小さく、かつ
-      局所的に抑えるように設計されています。用途に応じて以下の
-      索引タイプを選択できます。
-      EOC
-      groonga = @articles.add(:content => content)
+      Groonga::Schema.define do |schema|
+        schema.create_table("Articles") do |table|
+          table.text("content")
+        end
-      content.split(/\n{2,}/).each_with_index do |sentence, i|
-        @index[groonga] = {:value => sentence, :section => i + 1}
+        schema.create_table("Terms",
+                            :type => :hash,
+                            :key_type => "ShortText",
+                            :default_tokenizer => "TokenBigram") do |table|
+          table.index("Articles.content",
+                      :name => "articles_content",
+                      :with_position => true,
+                      :with_section => true)
+        end
       end
-      assert_equal([groonga],
-                   @index.search("エンジン").collect {|record| record.key})
+      @articles = Groonga["Articles"]
+      @index = Groonga["Terms.articles_content"]
     end
     def test_add
@@ -365,4 +349,150 @@ class IndexColumnTest < Test::Unit::TestCase
                    @index.disk_usage)
     end
   end
+  class SearchTest < self
+    class WeightTest < self
+      def setup
+        super
+        setup_schema
+      end
+      def setup_schema
+        Groonga::Schema.define do |schema|
+          schema.create_table("Memos",
+                              :type => :hash,
+                              :key_type => :short_text) do |table|
+            table.short_text("tags",
+                             :type => :vector,
+                             :with_weight => true)
+          end
+          schema.create_table("Tags",
+                              :type => :hash,
+                              :key_type => :short_text) do |table|
+            table.index("Memos.tags",
+                        :name => "memos_index",
+                        :with_weight => true)
+          end
+        end
+        @memos = context["Memos"]
+        @index = context["Tags.memos_index"]
+      end
+      def search(keyword, options={})
+        @index.search(keyword, options).collect do |record|
+          {
+            :key => record._key,
+            :score => record.score,
+          }
+        end
+      end
+      def test_index
+        record = @memos.add("Rroonga is fun!")
+        record.tags = [
+          {
+            :value => "rroonga",
+            :weight => 9,
+          }
+        ]
+        expected = [
+          {
+            :key => "Rroonga is fun!",
+            :score => 10,
+          }
+        ]
+        assert_equal(expected, search("rroonga"))
+      end
+      def test_search
+        record = @memos.add("Rroonga is fun!")
+        record.tags = [
+          {
+            :value => "rroonga",
+          }
+        ]
+        expected = [
+          {
+            :key => "Rroonga is fun!",
+            :score => 5,
+          }
+        ]
+        assert_equal(expected, search("rroonga", :weight => 5))
+      end
+      def test_index_and_search
+        record = @memos.add("Rroonga is fun!")
+        record.tags = [
+          {
+            :value => "rroonga",
+            :weight => 9,
+          }
+        ]
+        expected = [
+          {
+            :key => "Rroonga is fun!",
+            :score => 50
+          }
+        ]
+        assert_equal(expected, search("rroonga", :weight => 5))
+      end
+    end
+    class OperatorTest < self
+      def setup
+        super
+        setup_schema
+      end
+      def setup_schema
+        Groonga::Schema.define do |schema|
+          schema.create_table("Memos",
+                              :type => :hash,
+                              :key_type => :short_text) do |table|
+            table.short_text("tags", :type => :vector)
+          end
+          schema.create_table("Tags",
+                              :type => :hash,
+                              :key_type => :short_text) do |table|
+            table.index("Memos.tags",
+                        :name => "memos_index")
+          end
+        end
+        @memos = context["Memos"]
+        @index = context["Tags.memos_index"]
+      end
+      def test_adjust
+        @memos.add("Rroonga is fun!", :tags => ["rroonga", "groonga"])
+        @memos.add("Groonga is fast!", :tags => ["groonga"])
+        result = @index.search("groonga")
+        @index.search("rroonga", :result => result, :operator => :adjust)
+        expected = [
+          {
+            :key => "Rroonga is fun!",
+            :score => 2,
+          },
+          {
+            :key => "Groonga is fast!",
+            :score => 1,
+          }
+        ]
+        actual = result.collect do |record|
+          {
+            :key => record._key,
+            :score => record.score,
+          }
+        end
+        assert_equal(expected, actual)
+      end
+    end
+  end
 end

data/test/test-patricia-trie.rb CHANGED Viewed

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2009-2012  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -417,12 +417,13 @@ class PatriciaTrieTest < Test::Unit::TestCase
                                          :key_type => "ShortText")
     users.define_column("name", "ShortText")
     users.define_column("address", "ShortText")
-    1000.times do |i|
+    large_data = "x" * (2 ** 16)
+    100.times do |i|
       users.add("user #{i}",
-                :name => "user #{i}" * 1000,
-                :address => "address #{i}" * 1000)
+                :name => "user #{i}" + large_data,
+                :address => "address #{i}" + large_data)
     end
-    assert_equal(3, users.defrag)
+    assert_equal(2, users.defrag)
   end
   def test_rename

data/test/test-schema-dumper.rb CHANGED Viewed

@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2013  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2009-2014  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -96,14 +96,26 @@ class SchemaDumperTest < Test::Unit::TestCase
       schema.create_table("Terms",
                           :type => :patricia_trie,
                           :key_type => "ShortText",
-                          :key_normalize => true,
-                          :default_tokenizer => "TokenBigram") do |table|
+                          :default_tokenizer => "TokenBigram",
+                          :normalizer => "NormalizerAuto") do |table|
         table.index("Items", "_key")
         table.index("Items", "title")
       end
     end
   end
+  def define_weight_vector_schema
+    Groonga::Schema.define do |schema|
+      schema.create_table("Memos",
+                          :type => :patricia_trie,
+                          :key_type => "ShortText") do |table|
+        table.short_text("tags",
+                         :type => :vector,
+                         :with_weight => true)
+      end
+    end
+  end
   def define_double_array_trie_schema
     Groonga::Schema.define do |schema|
       schema.create_table("Accounts",
@@ -208,14 +220,26 @@ end
 create_table("Terms",
              :type => :patricia_trie,
              :key_type => "ShortText",
-             :key_normalize => true,
              :default_tokenizer => "TokenBigram",
+             :normalizer => "NormalizerAuto",
              :force => true) do |table|
 end
 change_table("Terms") do |table|
-  table.index("Items", "_key", :name => "Items__key")
-  table.index("Items", "title", :name => "Items_title")
+  table.index("Items", "_key", :name => "Items__key", :with_position => true)
+  table.index("Items", "title", :name => "Items_title", :with_position => true)
+end
+      SCHEMA
+    end
+    def test_weight_vector
+      define_weight_vector_schema
+      assert_equal(<<-SCHEMA, dump)
+create_table("Memos",
+             :type => :patricia_trie,
+             :key_type => "ShortText",
+             :force => true) do |table|
+  table.short_text("tags", :type => :vector, :with_weight => true)
 end
       SCHEMA
     end
@@ -282,13 +306,21 @@ column_create Comments item COLUMN_SCALAR Items
 table_create Items TABLE_HASH_KEY --key_type ShortText
 column_create Items title COLUMN_SCALAR ShortText
-table_create Terms TABLE_PAT_KEY|KEY_NORMALIZE --key_type ShortText --default_tokenizer TokenBigram
+table_create Terms TABLE_PAT_KEY --key_type ShortText --default_tokenizer TokenBigram --normalizer NormalizerAuto
 column_create Terms Items__key COLUMN_INDEX|WITH_POSITION Items _key
 column_create Terms Items_title COLUMN_INDEX|WITH_POSITION Items title
       SCHEMA
     end
+    def test_weight_vector
+      define_weight_vector_schema
+      assert_equal(<<-SCHEMA, dump)
+table_create Memos TABLE_PAT_KEY --key_type ShortText
+column_create Memos tags COLUMN_VECTOR|WITH_WEIGHT ShortText
+      SCHEMA
+    end
     def test_double_array_trie
       define_double_array_trie_schema
       assert_equal(<<-SCHEMA, dump)