twitter_cldr 4.0.0 → 4.1.0

data/lib/twitter_cldr/resources/properties/grapheme_break_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class GraphemeBreakPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/auxiliary/GraphemeBreakProperty.txt'
+        DATA_FILE = 'ucd/auxiliary/GraphemeBreakProperty.txt'
         PROPERTY_NAME = 'Grapheme_Cluster_Break'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/hangul_syllable_type_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class HangulSyllableTypePropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/HangulSyllableType.txt'
+        DATA_FILE = 'ucd/HangulSyllableType.txt'
         PROPERTY_NAME = 'Hangul_Syllable_Type'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/indic_positional_category_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class IndicPositionalCategoryPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/IndicPositionalCategory.txt'
+        DATA_FILE = 'ucd/IndicPositionalCategory.txt'
         PROPERTY_NAME = 'Indic_Positional_Category'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/indic_syllabic_category_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class IndicSyllabicCategoryPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/IndicSyllabicCategory.txt'
+        DATA_FILE = 'ucd/IndicSyllabicCategory.txt'
         PROPERTY_NAME = 'Indic_Syllabic_Category'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/jamo_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class JamoPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/Jamo.txt'
+        DATA_FILE = 'ucd/Jamo.txt'
         PROPERTY_NAME = 'Jamo_Short_Name'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/line_break_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class LineBreakPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/LineBreak.txt'
+        DATA_FILE = 'ucd/LineBreak.txt'
         PROPERTY_NAME = 'Line_Break'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/prop_list_importer.rb

@@ -8,20 +8,18 @@ module TwitterCldr
     module Properties
 
       class PropListImporter < PropertyImporter
-        DATA_URL = 'ucd/PropList.txt'
-
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: nil,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
-        end
+        DATA_FILE = 'ucd/PropList.txt'
+
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
 
         private
 
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
+        end
+
         def load
           super do |data, ret|
             code_points = expand_range(data[0])

data/lib/twitter_cldr/resources/properties/property_importer.rb

@@ -3,25 +3,14 @@
 # Copyright 2012 Twitter, Inc
 # http://www.apache.org/licenses/LICENSE-2.0
 
-require 'twitter_cldr/resources/download'
-
 module TwitterCldr
   module Resources
     module Properties
 
-      class PropertyImporter < UnicodeImporter
-        attr_reader :input_path, :property_name
-        attr_reader :data_url, :data_path, :database
-
-        def initialize(options)
-          @input_path = options.fetch(:input_path)
-          @property_name = options.fetch(:property_name)
-          @data_url = options.fetch(:data_url)
-          @data_path = options.fetch(:data_path)
-          @database = options.fetch(:database)
-        end
+      class PropertyImporter < Importer
+        private
 
-        def import
+        def execute
           load.each_pair do |property_name, property_values|
             property_values.each_pair do |property_value, ranges|
               database.store(property_name, property_value, ranges)
@@ -29,7 +18,15 @@ module TwitterCldr
           end
         end
 
-        private
+        def database
+          @database ||= TwitterCldr::Shared::PropertiesDatabase.new(
+            params.fetch(:output_path)
+          )
+        end
+
+        def parse_file(file, &block)
+          UnicodeFileParser.parse_standard_file(file, &block)
+        end
 
         def load
           results = Hash.new do |h, k|
@@ -37,7 +34,7 @@ module TwitterCldr
           end
 
           rangify_hash(
-            parse_standard_file(data_file).each_with_object(results) do |data, ret|
+            parse_file(source_path).each_with_object(results) do |data, ret|
               next unless data[0].size > 0
 
               if block_given?
@@ -70,12 +67,6 @@ module TwitterCldr
         def format_property_value(value)
           value
         end
-
-        def data_file
-          TwitterCldr::Resources.download_unicode_data_if_necessary(
-            File.join(input_path, data_path), data_url
-          )
-        end
       end
 
     end

data/lib/twitter_cldr/resources/properties/script_extensions_property_importer.rb

@@ -8,28 +8,30 @@ module TwitterCldr
     module Properties
 
       class ScriptExtensionsPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/ScriptExtensions.txt'
+        DATA_FILE = 'ucd/ScriptExtensions.txt'
         PROPERTY_NAME = 'Script_Extensions'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
         end
 
         private
 
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
+        end
+
         def load
           super do |data, ret|
             code_points = expand_range(data[0])
             property_values = data[1].split(' ')
 
             property_values.each do |property_value|
-              ret[PROPERTY_NAME][property_value] += code_points
+              ret[property_name][property_value] += code_points
             end
           end
         end

data/lib/twitter_cldr/resources/properties/script_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class ScriptPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/Scripts.txt'
+        DATA_FILE = 'ucd/Scripts.txt'
         PROPERTY_NAME = 'Script'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/sentence_break_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class SentenceBreakPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/auxiliary/SentenceBreakProperty.txt'
+        DATA_FILE = 'ucd/auxiliary/SentenceBreakProperty.txt'
         PROPERTY_NAME = 'Sentence_Break'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end
 

data/lib/twitter_cldr/resources/properties/unicode_data_properties_importer.rb

@@ -8,7 +8,7 @@ module TwitterCldr
     module Properties
 
       class UnicodeDataPropertiesImporter < PropertyImporter
-        DATA_URL = 'ucd/UnicodeData.txt'
+        DATA_FILE = 'ucd/UnicodeData.txt'
 
         PROPERTIES = {
           2  => 'General_Category',
@@ -18,18 +18,20 @@ module TwitterCldr
           9  => 'Bidi_Mirrored',
         }
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: nil,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
         end
 
         private
 
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
+        end
+
         def load
           range_start = nil
 

data/lib/twitter_cldr/resources/properties/word_break_property_importer.rb

@@ -8,17 +8,21 @@ module TwitterCldr
     module Properties
 
       class WordBreakPropertyImporter < PropertyImporter
-        DATA_URL = 'ucd/auxiliary/WordBreakProperty.txt'
+        DATA_FILE = 'ucd/auxiliary/WordBreakProperty.txt'
         PROPERTY_NAME = 'Word_Break'
 
-        def initialize(input_path, database)
-          super(
-            input_path: input_path,
-            property_name: PROPERTY_NAME,
-            data_url: DATA_URL,
-            data_path: File.basename(DATA_URL),
-            database: database
-          )
+        requirement :unicode, Versions.unicode_version, [DATA_FILE]
+        output_path 'unicode_data/properties'
+        ruby_engine :mri
+
+        def property_name
+          PROPERTY_NAME
+        end
+
+        private
+
+        def source_path
+          requirements[:unicode].source_path_for(DATA_FILE)
         end
       end