character.rb in tanuki_emoji-0.11.0

- old
+ new

@@ -6,11 +6,11 @@
   #
   # @see https://www.unicode.org/reports/tr51/
   class Character
     IMAGE_PREFIX = 'emoji_u'
     IMAGE_EXTENSION = '.png'
-    FLAG_REGEXP = /[🇦-🇿]{2}/u
+    FLAG_REGEXP = /[🇦-🇿]{2}|(\u{1F3F4}.+\u{E007F})/u
     ALPHA_CODE_REGEXP = /:(?<alpha_text>[_+\-a-z0-9]+):/
 
     # This denotes a "color" or "emoji" version
     EMOJI_VARIATION_SELECTOR = 0xFE0F
 
@@ -21,11 +21,11 @@
     # Zero Width Joiner is used in sequences to indicate they should all be evaluated and displayed as a single thing
     ZWJ_TAG = 0x200D
 
     attr_reader :name, :codepoints, :codepoints_alternates, :alpha_code, :aliases, :ascii_aliases, :description, :category
 
-    attr_accessor :unicode_version, :sort_key
+    attr_accessor :unicode_version, :sort_key, :noto_image
 
     # @param [String] name
     # @param [String] codepoints
     # @param [String] alpha_code
     # @param [String] description
@@ -43,43 +43,67 @@
 
     # Add alternative codepoints to this character
     #
     # @param [String] codepoints
     def add_codepoints(codepoints)
+      return if @codepoints == codepoints
+      return if codepoints_alternates.include?(codepoints)
+
       codepoints_alternates << codepoints
     end
 
     # Add alternative alpha_codes to this character
     #
     # @param [String] alpha_code
     def add_alias(alpha_code)
-      aliases << self.class.format_alpha_code(alpha_code)
+      formatted_code = self.class.format_alpha_code(alpha_code)
+
+      return if @alpha_code == alpha_code
+      return if aliases.include?(formatted_code)
+
+      aliases << formatted_code
     end
 
     # Add alternative ASCII aliases to this character
     #
     # @param [String] ascii_string
     def add_ascii_alias(ascii_string)
+      return if ascii_aliases.include?(ascii_string)
+
       ascii_aliases << ascii_string
     end
 
     # Return a Hex formatted version of the Unicode code points
     #
     # @return [String] Hex formatted version of the unicode
-    def hex
-      unicode_to_hex(codepoints).join('-')
+    def hex(codepoint = nil)
+      codepoint ? unicode_to_hex(codepoint).join('-') : unicode_to_hex(codepoints).join('-')
     end
 
     # Generate the image name to be used as fallback for this character
     #
     # @return [String] image name with extension
     def image_name
       # Noto doesn't ship flags as part of regular hex-named files
       # Flags are stored in a separate third-party folder and follow ISO-3166-1 codes
       # @see http://en.wikipedia.org/wiki/ISO_3166-1
-      return alpha_code.tr(':', '').sub('flag_', '').upcase + IMAGE_EXTENSION if flag?
+      # also see https://www.unicode.org/reports/tr51/#flag-emoji-tag-sequences for
+      # regional flags.
+      if flag?
+        name = noto_image
 
+        unless name
+          # The two character code is only found in the aliases
+          aliases.each do |item|
+            name = item.tr(':', '').sub('flag_', '')
+            break if name.length == 2
+          end
+        end
+
+        return name.upcase + IMAGE_EXTENSION
+      end
+
       # Noto omits Emoji Variation Selector on their resources file names
       IMAGE_PREFIX + unicode_to_hex(codepoints).reject { |i| i == EMOJI_VARIATION_SELECTOR.to_s(16) }.join('_') + IMAGE_EXTENSION
     end
 
     # Return whether current character represents a flag or not
@@ -92,10 +116,10 @@
     def to_s
       codepoints
     end
 
     def inspect
-      "#<#{self.class.name}:#{name} #{codepoints}(#{hex})>"
+      "#<#{self.class.name}: #{codepoints} (#{hex}) #{alpha_code} aliases: #{aliases}>"
     end
 
     def ==(other)
       name == other.name &&
         codepoints == other.codepoints &&