regexp_property_values 0.3.4-java → 1.2.0-java

data/lib/regexp_property_values/value/shared_methods.rb

@@ -0,0 +1,63 @@
+module RegexpPropertyValues
+  class Value
+    module SharedMethods
+      attr_reader :name
+
+      def initialize(name)
+        @name = name
+      end
+
+      def supported_by_current_ruby?
+        !!regexp rescue false
+      end
+
+      def ==(other)
+        identifier == other.identifier
+      end
+      alias eql? ==
+
+      def hash
+        @hash ||= identifier.hash
+      end
+
+      def identifier
+        @identifier ||= name.to_s.downcase.gsub(/[^0-9a-z=.]/, '')
+      end
+      alias to_s identifier
+
+      def full_name
+        (original = find_original) ? original.name : raise_unknown_error
+      end
+
+      def character_set
+        require 'character_set'
+        CharacterSet.from_ranges(*matched_ranges)
+      end
+
+      private
+
+      def regexp
+        @regexp ||= /\p{#{identifier}}/u
+      rescue RegexpError, SyntaxError
+        raise_unsupported_or_unknown_error
+      end
+
+      def find_original
+        RegexpPropertyValues.all.find { |orig| orig.eql?(self) } ||
+          RegexpPropertyValues.alias_hash[self]
+      end
+
+      def raise_unsupported_or_unknown_error
+        find_original ? raise_unsupported_error : raise_unknown_error
+      end
+
+      def raise_unsupported_error
+        raise Error, "Property name `#{name}` is known, but not in this Ruby"
+      end
+
+      def raise_unknown_error
+        raise Error, "Property name `#{name}` is not known in any Ruby"
+      end
+    end
+  end
+end

data/lib/regexp_property_values/value.rb

@@ -0,0 +1,14 @@
+module RegexpPropertyValues
+  class Value
+    require_relative 'value/shared_methods'
+    include SharedMethods
+
+    if const_defined?(:OnigRegexpPropertyHelper)
+      require_relative 'value/ext_adapter'
+      include ExtAdapter
+    else
+      require_relative 'value/ruby_fallback'
+      include RubyFallback
+    end
+  end
+end

data/lib/regexp_property_values/version.rb

@@ -1,3 +1,3 @@
 module RegexpPropertyValues
-  VERSION = '0.3.4'
+  VERSION = '1.2.0'
 end

data/lib/regexp_property_values.rb

@@ -3,77 +3,35 @@ begin
 rescue LoadError
   warn 'regexp_property_values could not load C extension, using slower Ruby'
 end
-require 'regexp_property_values/extension'
+require 'regexp_property_values/updater'
+require 'regexp_property_values/value'
 require 'regexp_property_values/version'
 
 module RegexpPropertyValues
-  module_function
+  Error = Class.new(StandardError)
 
-  LIST_URL = 'https://raw.githubusercontent.com/k-takata/Onigmo/master/doc/UnicodeProps.txt'
+  VALUES_PATH  = File.join(__dir__, 'values')
+  ALIASES_PATH = File.join(__dir__, 'aliases')
 
-  def update
-    puts "Downloading #{LIST_URL}"
-    require 'open-uri'
-    File.open(file_path, 'w') { |f| IO.copy_stream(open(LIST_URL), f) }
-    puts 'Done!'
+  def self.[](name)
+    Value.new(name)
   end
 
-  def file_path
-    File.expand_path('../UnicodeProps.txt', __FILE__)
+  def self.all_for_current_ruby
+    @all_for_current_ruby ||= all.select(&:supported_by_current_ruby?)
   end
 
-  def all
-    by_category.values.flatten
+  def self.all
+    @all ||= File.readlines(VALUES_PATH).map { |line| Value.new(line.chomp) }
   end
 
-  def all_for_current_ruby
-    all.select(&:supported_by_current_ruby?)
+  def self.alias_hash
+    @alias_hash ||= File.readlines(ALIASES_PATH).map do |line|
+      line.chomp.split(';').map { |name| Value.new(name) }
+    end.to_h
   end
 
-  def by_category
-    result = File.foreach(file_path).each_with_object({}) do |line, hash|
-      if /^\* (?<category>\S.+)/ =~ line
-        @current_category = category
-        hash[@current_category] ||= []
-      elsif /^ {4}(?<value_name>\S.*)/ =~ line
-        hash[@current_category] << value_name.extend(Extension)
-      end
-    end
-    add_oniguruma_properties(result)
-    result
-  end
-
-  def add_oniguruma_properties(props_by_category)
-    props_by_category['Special'] << 'Newline'.extend(Extension)
-  end
-
-  def alias_hash
-    short_names, long_names = short_and_long_names
-    return {} if short_names.empty?
-
-    long_names -= by_category['POSIX brackets']
-    by_matched_codepoints.each_value.each_with_object({}) do |props, hash|
-      next if props.count < 2
-      long_name = (props & long_names)[0] || fail("no long name for #{props}")
-      (props & short_names).each { |short_name| hash[short_name] = long_name }
-    end
-  end
-
-  def short_and_long_names
-    short_name_categories = ['Major and General Categories',
-                             'PropertyAliases',
-                             'PropertyValueAliases (Script)']
-    by_category.each_with_object([[], []]) do |(cat_name, props), (short, long)|
-      (short_name_categories.include?(cat_name) ? short : long).concat(props)
-    end
-  end
-
-  def by_matched_codepoints
-    puts 'Establishing property codepoints, this may take a bit ...'
-    all_for_current_ruby.group_by(&:matched_codepoints)
-  end
-
-  def [](prop)
-    prop.extend(Extension)
+  def self.update(ucd_path: nil, emoji_path: nil)
+    Updater.call(ucd_path: ucd_path, emoji_path: emoji_path)
   end
 end