grn_mini 0.2.0 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8b9f025b3d3dfc38fd85b82b8552b9df10244dd8
-  data.tar.gz: 2294e3e7af33ffd32d15253d01572bbeff84bbb3
+  metadata.gz: 954b7ebeb13d40913ed6022a6b96d972193b73ea
+  data.tar.gz: c64cf4de87e4cf59b172fb1b4d5c64d7f8f17e12
 SHA512:
-  metadata.gz: 44faf36a48ca1245ddd3621955b998295a7e608e1f6ddb98229374391374cb18919c4883c0a3c0517fa4cee9eb9034a213e3a87ba271f3f4c2611f3454015c9b
-  data.tar.gz: 68bd08647aa71e471072aec681a547821f06c111fcac909309545986f0898b9046d365fee677bc4334047772f844a286a673a39b738ff0317a2e8001160a619b
+  metadata.gz: 0406c079b5f1fac34c4f35de6e0ca59b1236df413b285ec28d529c054e1e08f1558f03712b24c37d2dba53f843e219db3ecc37e3d3ea33143faf4121c50b6fb1
+  data.tar.gz: 747fe7c71d88f786229ad7461639b9ff4de8cee0bc73e27aff73ce3301c98659311173517434b98b857b91bbbace9aabb084ee3900c968fd828bad8531d43af3

data/README.md

@@ -52,6 +52,25 @@ end
 # Delete temporary database
 ```
 
+## Create Hash
+
+```ruby
+require 'grn_mini'
+
+hash = GrnMini::Hash.new("test.db")
+
+# Add
+hash["a"] = {text:"aaa", number:1}
+hash["b"] = {text:"bbb", number:2}
+hash["c"] = {text:"ccc", number:3}
+
+# Read
+hash["b"].text       #=> "bbb"
+
+# Write
+hash["b"].text = "BBB"
+```
+
 ## Data Type
 
 ```ruby

data/lib/grn_mini.rb

@@ -1,5 +1,6 @@
 require "grn_mini/version"
 require "grn_mini/array"
+require "grn_mini/hash"
 require "grn_mini/util"
 
 module GrnMini

data/lib/grn_mini/hash.rb

@@ -0,0 +1,95 @@
+require 'grn_mini/util'
+require 'groonga'
+require 'tmpdir'
+
+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")
+    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
+      
+      @grn.add(key, values)
+    end
+
+    def [](key)
+      @grn[key]
+    end
+    
+    def []=(key, value)
+      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/version.rb

@@ -1,3 +1,3 @@
 module GrnMini
-  VERSION = "0.2.0"
+  VERSION = "0.3.0"
 end

data/sample/mini-directory-search.rb

@@ -15,7 +15,7 @@ module Input
 
       if File.file? filename
         next if ignore_file?(filename)
-        array << {filename: filename, text: read_file(filename), timestamp: File.stat(filename).mtime, suffix: File.extname(filename).sub('.', ""), index: index } # index is dirty
+        array << {filename: filename, text: read_file(filename), timestamp: File.stat(filename).mtime, suffix: File.extname(filename).sub('.', "")}
         index += 1
       end
     end
@@ -61,7 +61,7 @@ class Search
       elements = []
 
       page_entries.each do |record|
-        element = "<hr>\n<a href=\"/#{record.index}\">#{record.filename}</a>\n"
+        element = "<hr>\n<a href=\"/#{record.value.key.id}\">#{record.filename}</a>\n"
         
         snippet.execute(record.text).each do |segment|
           element += "<pre style=\"border:1px solid #bbb;\">#{segment}</pre>\n"

data/test/test_grn_mini_array.rb

@@ -41,6 +41,7 @@ class TestGrnMiniArray < MiniTest::Unit::TestCase
       results = array.select("bb number:<10")
 
       assert_equal 1, results.size
+      assert_equal 2, results.first.key.id
       assert_equal "bbb", results.first.text
       assert_equal 2, results.first.number
     end 
@@ -325,8 +326,8 @@ EOF
       record = results.first
       segments = snippet.execute(record.text)
       assert_equal 2, segments.size 
-      assert_match /\[1\] <<This>> is a <<pen>> pep pea pek pet./, segments[0]
-      assert_match /\[2\] <<This>> is a <<pen>> pep pea pek pet./, segments[1]
+      assert_match /\[1\]<< This>> is a<< pen>> pep pea pek pet./, segments[0]
+      assert_match /\[2\]<< This>> is a<< pen>> pep pea pek pet./, segments[1]
     end
   end
 
@@ -344,7 +345,7 @@ EOF
       record = results.first
       segments = snippet.execute(record.text)
       assert_equal 1, segments.size
-      assert_equal "&lt;html&gt;\n  &lt;div&gt;<strong>This</strong> is a <strong>pen</strong> pep pea pek pet.&lt;/div&gt;\n&lt;/html&gt;\n", segments.first
+      assert_equal "&lt;html&gt;\n  &lt;div&gt;<strong>This</strong> is a<strong> pen</strong> pep pea pek pet.&lt;/div&gt;\n&lt;/html&gt;\n", segments.first
     end
   end
 

data/test/test_grn_mini_hash.rb

@@ -0,0 +1,260 @@
+require 'minitest_helper'
+
+class TestGrnMiniHash < MiniTest::Unit::TestCase
+  def test_initialize
+    Dir.mktmpdir do |dir|
+      hash = GrnMini::Hash.new(File.join(dir, "test.db"))
+    end
+  end
+
+  def test_add
+    GrnMini::Hash.tmpdb do |hash|
+      hash.add("aaa", text:"aaa", number:1)
+      assert_equal 1, hash.size
+
+      # alias []=
+      hash["bbb"] = {text:"bbb", number:2}
+      assert_equal 2, hash.size
+    end 
+  end
+
+  def test_select
+    GrnMini::Hash.tmpdb do |hash|
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      hash["c"] = {text:"ccc", number:3}
+
+      results = hash.select("bb")
+
+      assert_equal 1, results.size
+      assert_equal "b", results.first.key.key
+      assert_equal "bbb", results.first.text
+    end 
+  end
+
+  def test_select2
+    GrnMini::Hash.tmpdb do |hash|
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      hash["c"] = {text:"bbb", number:20}
+      hash["d"] = {text:"ccc", number:3}
+
+      results = hash.select("bb number:<10")
+
+      assert_equal 1, results.size
+      assert_equal "b", results.first.key.key
+      assert_equal "bbb", results.first.text
+      assert_equal 2, results.first.number
+    end 
+  end
+
+  def test_size
+    GrnMini::Hash.tmpdb do |hash|
+      assert_equal 0, hash.size
+      assert_equal 0, hash.length
+      
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      assert_equal 2, hash.size
+      assert_equal 2, hash.length
+    end
+  end
+
+  def test_empty?
+    GrnMini::Hash.tmpdb do |hash|
+      assert_equal true, hash.empty?
+
+      hash["a"] = {text:"aaa", number:1}
+      assert_equal false, hash.empty?
+    end
+  end
+
+  def test_each
+    GrnMini::Hash.tmpdb do |hash|
+      hash["c"] = {text:"ccc", number:3}
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+
+      total = 0
+
+      hash.each do |v|
+        total += v.number
+      end
+
+      assert_equal 6, total
+    end
+  end
+
+  def test_read_by_key
+    GrnMini::Hash.tmpdb do |hash|
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      hash["c"] = {text:"ccc", number:3}
+
+      assert_equal   nil, hash["not found"]
+      assert_equal "aaa", hash["a"].text
+      assert_equal "bbb", hash["b"].text
+      assert_equal "ccc", hash["c"].text
+    end
+  end
+
+  def test_write_by_key
+    GrnMini::Hash.tmpdb do |hash|
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      hash["c"] = {text:"ccc", number:3}
+
+      assert_equal "bbb", hash["b"].text
+      assert_equal     2, hash["b"].number
+
+      hash["b"].text   = "BBB"
+      hash["b"].number = 22
+
+      assert_equal "BBB", hash["b"].text
+      assert_equal    22, hash["b"].number
+    end
+  end
+
+  def test_delete_by_id
+    GrnMini::Hash.tmpdb do |hash|
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      hash["c"] = {text:"ccc", number:3}
+
+      # Delete from key
+      assert_equal 3, hash.size
+      assert_equal 2, hash["b"].number
+
+      hash.delete("b")
+      
+      assert_equal 2, hash.size
+      assert_nil hash["b"]
+    end
+  end
+
+  def test_delete_by_block
+    GrnMini::Hash.tmpdb do |hash|
+      hash["a"] = {text:"aaa", number:1}
+      hash["b"] = {text:"bbb", number:2}
+      hash["c"] = {text:"ccc", number:3}
+
+      hash.delete do |record|
+        record.number >= 2
+      end
+
+      assert_equal 1, hash.size
+      assert_equal "aaa", hash.first.text
+      assert_equal 1, hash.first.number
+    end
+  end
+
+  def test_float_column
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {text:"aaaa", float: 1.5}
+      hash["b"] = {text:"bbbb", float: 2.5}
+      hash["c"] = {text:"cccc", float: 3.5}
+
+      assert_equal 2.5, hash["b"].float
+
+      results = hash.select("float:>2.6")
+      assert_equal 3.5, results.first.float
+    end
+  end
+
+  def test_time_column
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {text:"aaaa", timestamp: Time.new(2013)} # 2013-01-01
+      hash["b"] = {text:"bbbb", timestamp: Time.new(2014)} # 2014-01-01
+      hash["c"] = {text:"cccc", timestamp: Time.new(2015)} # 2015-01-01
+
+      assert_equal Time.new(2014), hash["b"].timestamp
+
+      results = hash.select("timestamp:<=#{Time.new(2013,12).to_i}")
+      assert_equal 1, results.size
+      assert_equal Time.new(2013), results.first.timestamp
+    end    
+  end
+
+  def test_record_attributes
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {text:"aaaa", int: 1}
+      hash["b"] = {text:"bbbb", int: 2}
+      hash["c"] = {text:"cccc", int: 3}
+
+      assert_equal({"_id"=>1, "_key"=>"a", "int"=>1, "text"=>"aaaa"}, hash["a"].attributes)
+      assert_equal({"_id"=>2, "_key"=>"b", "int"=>2, "text"=>"bbbb"}, hash["b"].attributes)
+      assert_equal({"_id"=>3, "_key"=>"c", "int"=>3, "text"=>"cccc"}, hash["c"].attributes)
+    end
+  end
+
+  def test_assign_long_text_to_short_text
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {filename:"a.txt"}
+      hash["b"] = {filename:"a"*4095 + ".txt" } # Over 4095 byte (ShortText limit)
+
+      results = hash.select("txt", default_column: "filename")
+      assert_equal 2, results.size 
+    end
+  end
+
+  def test_grn_object
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {text: "aaaa", filename:"a.txt", int: 1, float: 1.5, time: Time.at(2001)}
+
+      raw = hash.grn
+
+      assert_equal true, raw.have_column?("filename")
+      assert_equal true, raw.have_column?("int")
+      assert_equal true, raw.have_column?("float")
+      assert_equal true, raw.have_column?("time")
+      assert_equal false, raw.have_column?("timeee")
+
+      assert_equal "ShortText", raw.column("text").range.name
+      assert_equal "ShortText", raw.column("filename").range.name
+      assert_equal "Int32"    , raw.column("int").range.name
+      assert_equal "Float"    , raw.column("float").range.name
+      assert_equal "Time"     , raw.column("time").range.name
+
+      assert_equal true, raw.support_key? # GrnMini::Array is false
+      assert_equal false, raw.support_sub_records?
+    end
+  end
+
+  def test_sort
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {name:"Tanaka",  age: 11, height: 162.5}
+      hash["b"] = {name:"Suzuki",  age: 31, height: 170.0}
+      hash["c"] = {name:"Hayashi", age: 21, height: 175.4}
+      hash["d"] = {name:"Suzuki",  age:  5, height: 110.0}
+
+      sorted_by_age = hash.sort(["age"])
+      sorted_array = sorted_by_age.map {|r| {name: r.name, age: r.age}}
+      assert_equal [{:name=>"Suzuki", :age=>5},
+                    {:name=>"Tanaka", :age=>11},
+                    {:name=>"Hayashi", :age=>21},
+                    {:name=>"Suzuki", :age=>31}], sorted_array
+
+      sorted_by_combination = hash.sort([{key: "name", order: :ascending},
+                                          {key: "age" , order: :descending}])
+      sorted_array = sorted_by_combination.map {|r| {name: r.name, age: r.age}}
+      assert_equal [{:name=>"Hayashi", :age=>21},
+                    {:name=>"Suzuki", :age=>31},
+                    {:name=>"Suzuki", :age=>5},
+                    {:name=>"Tanaka", :age=>11}], sorted_array
+    end
+  end
+
+  def test_group_from_hash
+    GrnMini::Hash.tmpdb do |hash| 
+      hash["a"] = {text:"aaaa.txt", suffix:"txt", type:1}
+      hash["b"] = {text:"aaaa.doc", suffix:"doc", type:2}
+      hash["c"] = {text:"aabb.txt", suffix:"txt", type:2}
+
+      groups = GrnMini::Util::group_with_sort(hash, "suffix")
+
+      assert_equal 2, groups.size
+      assert_equal ["txt", 2], [groups[0].key, groups[0].n_sub_records]
+      assert_equal ["doc", 1], [groups[1].key, groups[1].n_sub_records]
+    end
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: grn_mini
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 0.3.0
 platform: ruby
 authors:
 - ongaeshi
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-01-07 00:00:00.000000000 Z
+date: 2014-01-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -85,12 +85,14 @@ files:
 - grn_mini.gemspec
 - lib/grn_mini.rb
 - lib/grn_mini/array.rb
+- lib/grn_mini/hash.rb
 - lib/grn_mini/util.rb
 - lib/grn_mini/version.rb
 - sample/mini-directory-search.rb
 - test/minitest_helper.rb
 - test/test_grn_mini.rb
 - test/test_grn_mini_array.rb
+- test/test_grn_mini_hash.rb
 homepage: ''
 licenses:
 - MIT
@@ -119,3 +121,4 @@ test_files:
 - test/minitest_helper.rb
 - test/test_grn_mini.rb
 - test/test_grn_mini_array.rb
+- test/test_grn_mini_hash.rb