grn_mini 0.3.0 → 0.4.0

data/lib/grn_mini/hash.rb

@@ -1,49 +1,14 @@
-require 'grn_mini/util'
-require 'groonga'
-require 'tmpdir'
+require 'grn_mini/table'
 
 module GrnMini
-  class Hash
-    attr_accessor :grn
-    include Enumerable
-
-    def self.tmpdb
-      Dir.mktmpdir do |dir|
-        # p dir
-        yield self.new(File.join(dir, "tmp.db"))
-      end
-    end
-
-    def initialize(path)
-      unless File.exist?(path)
-        Groonga::Database.create(path: path)
-      else
-        Groonga::Database.open(path)
-      end
-
-      @grn = Groonga["Hash"] || Groonga::Hash.create(name: "Hash", persistent: true)
-      @terms = Groonga["Terms"] || Groonga::PatriciaTrie.create(name: "Terms", key_normalize: true, default_tokenizer: "TokenBigramSplitSymbolAlphaDigit")
+  class Hash < Table
+    def initialize(name = "Hash")
+      super(name,
+            Groonga[name] || Groonga::Hash.create(name: name, persistent: true))
     end
 
     def add(key, values)
-      if @grn.empty?
-        values.each do |key, value|
-          column = key.to_s
-
-          # @todo Need define_index_column ?
-          if value.is_a?(Time)
-            @grn.define_column(column, "Time")
-          elsif value.is_a?(Float)
-            @grn.define_column(column, "Float")
-          elsif value.is_a?(Numeric)
-            @grn.define_column(column, "Int32")
-          else
-            @grn.define_column(column, "ShortText")
-            @terms.define_index_column("array_#{column}", @grn, source: "Hash.#{column}", with_position: true)
-          end
-        end
-      end
-      
+      setup_columns(values) unless @setup_columns_once
       @grn.add(key, values)
     end
 
@@ -55,41 +20,13 @@ module GrnMini
       add(key, value)
     end
 
-    def select(query, options = {default_column: "text"})
-      @grn.select(query, options)
-    end
-
-    def size
-      @grn.size
-    end
-
-    alias length size
-
-    def empty?
-      size == 0
-    end
-    
-    def each
-      @grn.each do |record|
-        yield record
-      end
-    end
-
     def delete(id = nil, &block)
       if block_given?
         @grn.delete(&block)
       else
-        raise IdIsGreaterThanZero if id == 0
         @grn.delete(id)
       end
     end
 
-    def sort(keys, options = {})
-      @grn.sort(keys, options)
-    end
-
-    def group(key, options = {})
-      @grn.group(key, options)
-    end
   end
 end

data/lib/grn_mini/table.rb

@@ -0,0 +1,91 @@
+require 'grn_mini/util'
+require 'groonga'
+
+module GrnMini
+  class NotSupportColumnType < RuntimeError; end
+
+  class Table
+    attr_accessor :grn
+    include Enumerable
+
+    def initialize(name, grn)
+      @name = name
+      @grn  = grn
+      @terms = Groonga["Terms"] || Groonga::PatriciaTrie.create(name: "Terms", key_normalize: true, default_tokenizer: "TokenBigramSplitSymbolAlphaDigit")
+      @setup_columns_once = false
+    end
+
+    def setup_columns(hash)
+      hash.each do |key, value|
+        column = key.to_s
+
+        if value.is_a?(String)
+          @grn.define_column(column, value_type(value))
+          @terms.define_index_column("#{@name}_#{column}", @grn, source: "#{@name}.#{column}", with_position: true)
+        elsif value.is_a?(::Array)
+          @grn.define_column(column, value_type(value), type: :vector)
+
+          elem = value.first
+          
+          if elem.is_a?(GrnMini::Table)
+            elem.grn.define_index_column("index_#{column}", @grn, source: "#{@grn.name}.#{column}")
+          elsif elem.is_a?(Groonga::Table)
+            elem.define_index_column("index_#{column}", @grn, source: "#{@grn.name}.#{column}")
+          end
+        else
+          @grn.define_column(column, value_type(value))
+        end
+      end
+
+      @setup_columns_once = true
+    end
+    
+    def value_type(value)
+      if value.is_a?(TrueClass) || value.is_a?(FalseClass)
+        "Bool"
+      elsif value.is_a?(Integer)
+        "Int32"
+      elsif value.is_a?(Float)
+        "Float"
+      elsif value.is_a?(Time)
+        "Time"
+      elsif value.is_a?(String)
+        "ShortText"
+      elsif value.is_a?(GrnMini::Table)
+        value.grn.name
+      elsif value.is_a?(Groonga::Table)
+        value.name
+      elsif value.is_a?(::Array)
+        value_type(value.first)
+      else
+        raise NotSupportColumnType, value
+      end
+    end
+
+    def select(*args, &block)
+      @grn.select(*args, &block)
+    end
+    
+    def size
+      @grn.size
+    end
+
+    def empty?
+      size == 0
+    end
+
+    def each
+      @grn.each do |record|
+        yield record
+      end
+    end
+
+    def sort(keys, options = {})
+      @grn.sort(keys, options)
+    end
+
+    def group(key, options = {})
+      @grn.group(key, options)
+    end
+  end
+end

data/lib/grn_mini/version.rb

@@ -1,3 +1,3 @@
 module GrnMini
-  VERSION = "0.3.0"
+  VERSION = "0.4.0"
 end

data/test/test_grn_mini_array.rb

@@ -2,13 +2,15 @@ require 'minitest_helper'
 
 class TestGrnMiniArray < MiniTest::Unit::TestCase
   def test_initialize
-    Dir.mktmpdir do |dir|
-      array = GrnMini::Array.new(File.join(dir, "test.db"))
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
     end
   end
 
   def test_add
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array.add(text:"aaa", number:1)
       assert_equal 1, array.size
 
@@ -19,12 +21,14 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_select
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       array << {text:"ccc", number:3}
 
-      results = array.select("bb")
+      results = array.select("text:@bb")
 
       assert_equal 1, results.size
       assert_equal "bbb", results.first.text
@@ -32,13 +36,15 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_select2
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       array << {text:"bbb", number:20}
       array << {text:"ccc", number:3}
 
-      results = array.select("bb number:<10")
+      results = array.select("text:@bb number:<10")
 
       assert_equal 1, results.size
       assert_equal 2, results.first.key.id
@@ -48,19 +54,21 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_size
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       assert_equal 0, array.size
-      assert_equal 0, array.length
       
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       assert_equal 2, array.size
-      assert_equal 2, array.length
     end
   end
 
   def test_empty?
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       assert_equal true, array.empty?
 
       array << {text:"aaa", number:1}
@@ -69,7 +77,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_each
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       array << {text:"ccc", number:3}
@@ -91,7 +101,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_read_by_id
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       array << {text:"ccc", number:3}
@@ -111,7 +123,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_write_by_id
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       array << {text:"ccc", number:3}
@@ -128,7 +142,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_delete_by_id
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaa", number:1}
       array << {text:"bbb", number:2}
       array << {text:"ccc", number:3}
@@ -163,7 +179,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_delete_by_block
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"bbb", number:2}
       array << {text:"ccc", number:3}
       array << {text:"aaa", number:1}
@@ -179,7 +197,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_float_column
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaaa", float: 1.5}
       array << {text:"bbbb", float: 2.5}
       array << {text:"cccc", float: 3.5}
@@ -192,7 +212,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_time_column
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaaa", timestamp: Time.new(2013)} # 2013-01-01
       array << {text:"bbbb", timestamp: Time.new(2014)} # 2014-01-01
       array << {text:"cccc", timestamp: Time.new(2015)} # 2015-01-01
@@ -205,8 +227,29 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
     end    
   end
 
+  def test_bool_column
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
+      array << {bool: true}
+      array << {bool: false}
+      array << {bool: false}
+
+      assert_equal true, array[1].bool
+      assert_equal false, array[2].bool
+
+      results = array.select { |r| r.bool == true }
+      assert_equal 1, results.size
+
+      results = array.select { |r| r.bool == false }
+      assert_equal 2, results.size
+    end    
+  end
+
   def test_record_attributes
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaaa", int: 1}
       array << {text:"bbbb", int: 2}
       array << {text:"cccc", int: 3}
@@ -218,7 +261,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_assign_long_text_to_short_text
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {filename:"a.txt"}
       array << {filename:"a"*4095 + ".txt" } # Over 4095 byte (ShortText limit)
 
@@ -228,7 +273,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_grn_object
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text: "aaaa", filename:"a.txt", int: 1, float: 1.5, time: Time.at(2001)}
 
       raw = array.grn
@@ -251,7 +298,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_sort
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {name:"Tanaka",  age: 11, height: 162.5}
       array << {name:"Suzuki",  age: 31, height: 170.0}
       array << {name:"Hayashi", age: 21, height: 175.4}
@@ -275,7 +324,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_group_from_array
-    GrnMini::Array.tmpdb do |array| 
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaaa.txt", suffix:"txt", type:1}
       array << {text:"aaaa.doc", suffix:"doc", type:2}
       array << {text:"aabb.txt", suffix:"txt", type:2}
@@ -289,13 +340,15 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_group_from_selection_results
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text:"aaaa", suffix:"txt"}
       array << {text:"aaaa", suffix:"doc"}
       array << {text:"aaaa", suffix:"txt"}
       array << {text:"cccc", suffix:"txt"}
 
-      results = array.select("aa")
+      results = array.select("text:@aa")
       groups = GrnMini::Util::group_with_sort(results, "suffix")
 
       assert_equal 2, groups.size
@@ -305,7 +358,9 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
   end
 
   def test_text_snippet_from_selection_results
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text: <<EOF, filename: "aaa.txt"}
 [1] This is a pen pep pea pek pet.
 ------------------------------
@@ -320,7 +375,7 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
 ------------------------------
 EOF
 
-      results = array.select("This pen")
+      results = array.select("This pen", default_column: "text")
       snippet = GrnMini::Util::text_snippet_from_selection_results(results)
 
       record = results.first
@@ -332,14 +387,16 @@ EOF
   end
 
   def test_html_snippet_from_selection_results
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text: <<EOF, filename: "aaa.txt"}
 <html>
   <div>This is a pen pep pea pek pet.</div>
 </html>
 EOF
 
-      results = array.select("This pen")
+      results = array.select("text:@This text:@pen")
       snippet = GrnMini::Util::html_snippet_from_selection_results(results)
 
       record = results.first
@@ -350,7 +407,9 @@ EOF
   end
 
   def test_paginate
-    GrnMini::Array.tmpdb do |array|
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
       array << {text: "aaaa", filename: "1.txt"}
       array << {text: "aaaa aaaa", filename: "2a.txt"}
       array << {text: "aaaa aaaa aaaa", filename: "3.txt"}
@@ -358,7 +417,7 @@ EOF
       array << {text: "aaaa aaaa", filename: "2c.txt"}
       array << {text: "aaaa aaaa", filename: "2d.txt"}
 
-      results = array.select("aaaa")
+      results = array.select { |r| r.text =~ "aaaa" }
 
       # page1
       page_entries = results.paginate([["_score", :desc]], :page => 1, :size => 5)
@@ -375,4 +434,175 @@ EOF
       assert_equal "1.txt", page_entries.first.filename
     end
   end
+
+  def test_multi_table
+    Dir.mktmpdir do |dir|
+      GrnMini::create_or_open(File.join(dir, "test.db"))
+      
+      users = GrnMini::Array.new("Users")
+      users << {name: "aaa", age: 10}
+      users << {name: "bbb", age: 20}
+      users << {name: "ccc", age: 30}
+      
+      articles = GrnMini::Array.new("Articles")
+      articles << {title: "Hello", text: "hello everyone."}
+
+      assert Groonga["Users"]
+      assert_equal 3, users.size
+
+      assert Groonga["Articles"]
+      assert_equal 1, articles.size
+    end
+  end
+
+  def test_multi_table_select
+    GrnMini::tmpdb do
+      users = GrnMini::Array.new("Users")
+      users << {name: "aaa", age: 10, text: ""}
+      users << {name: "bbb", age: 20}
+      users << {name: "ccc", age: 30}
+      
+      articles = GrnMini::Array.new("Articles")
+      articles << {title: "1", text: "111 aaa"}
+      articles << {title: "2", text: "222 bbb"}
+      articles << {title: "3", text: "333 ccc"}
+
+      results = users.select("222 OR ccc", default_column: "text")
+      assert_equal 0, results.size
+      
+      results = articles.select("222 OR ccc", default_column: "text")
+      assert_equal 2, results.size
+      assert_equal "2", results.first.title
+    end
+  end
+
+  def test_setup_columns
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+
+      array.setup_columns(
+        name: "",
+        age:  0,
+      )
+
+      array << {name: "bbb", age: 20 }
+      assert_raises(Groonga::NoSuchColumn) { array << {name: "aaa", age: 10, text: ""} }
+      assert_equal 2, array.size
+    end
+  end
+
+  def test_setup_columns_groonga_table
+    GrnMini::tmpdb do
+      users = GrnMini::Array.new("Users")
+
+      users.setup_columns(
+        parent: users,
+      )
+
+      users << {}
+      users << { parent: users[1] }
+      users << { parent: users[2] }
+
+      # No error ..
+      # users << { parent: 123456 }
+      # users << { parent: 1.5 }
+      # users << { parent: Time.now }
+
+      assert_equal      nil, users[1].parent
+      assert_equal users[1], users[2].parent
+      assert_equal users[1], users[3].parent.parent
+    end
+  end
+
+  def test_setup_columns_vector
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+      array.setup_columns(links:[""])
+
+      array << {}
+      assert_equal [], array[1].links
+
+      # Nothing happens ..
+      array[1].links << "http://ongaeshi.me"
+      assert_equal 0, array[1].links.size
+
+      # Use links=
+      array[1].links = ["http://ongaeshi.me", "http://yahoo.co.jp"]
+      assert_equal 2, array[1].links.size
+
+      array << { links: ["aaa", "bbb", "ccc"]}
+      array << { links: ["AAA"]}
+
+      # array.each do |record|
+      #   p record.attributes
+      # end
+
+      assert_equal 3, array.size
+      assert_equal 3, array[2].links.size
+    end
+  end
+
+  def test_setup_columns_vector2
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+      array.setup_columns(strs:    [""],
+                          numbers: [0],
+                          floats:  [0.1],
+                          times:   [Time.new],
+                          arrays:  [array]
+                          )
+
+      array.add(strs: ["AAA", "BBB"],
+                numbers: [1, 2, 3],
+                floats: [0.1, 0.2, 0.3],
+                times:  [Time.at(0), Time.at(1), Time.at(3)],
+                arrays: []
+                )
+      array.add(arrays: [array[1]])
+      array.add(arrays: [array[1], array[2]])
+
+      # First element is [1..N], Second and subsequent elements is [0..N-1].
+      assert_equal 0.2, array[3].arrays[1].arrays[0].floats[1]
+    end    
+  end
+
+  def test_tweet
+    GrnMini::tmpdb do
+      users = GrnMini::Array.new("Users")
+      tweets = GrnMini::Array.new("Tweets")
+
+      users.setup_columns(name: "A Name"
+                          )
+      tweets.setup_columns(user: users,
+                           time: Time.new,
+                           text: "A Tweet.",
+                           replies: [tweets]
+                           )
+
+      users << { name: "AAA" }
+      users << { name: "BBB" }
+
+      tweets << { user: users[1], time: Time.at(1), text: "AAA Tweet 1" }
+      tweets << { user: users[2], time: Time.at(2), text: "BBB Tweet 1" }
+      tweets << { user: users[1], time: Time.at(3), text: "AAA Tweet 2" }
+      tweets << { user: users[1], time: Time.at(4), text: "Re: BBB Tweet 1" }
+      tweets << { user: users[2], time: Time.at(5), text: "Re: Re: BBB Tweet 1" }
+      tweets[3].replies = [tweets[4], tweets[5]]
+
+      # tweets.each do |record|
+      #   p record.attributes
+      # end
+
+      assert_equal 3, tweets.select("user:1").size
+      assert_equal "AAA Tweet 2", tweets.select("user:1 text:@Tweet text:@2").first.text
+      assert_equal 2, tweets.select("user:2").size
+    end    
+  end
+
+  def test_not_support_column_type
+    GrnMini::tmpdb do
+      array = GrnMini::Array.new
+      assert_raises(GrnMini::NotSupportColumnType) { array.setup_columns(error: Hash) } 
+    end    
+  end
 end