yosina 0.1.0

data/lib/yosina/transliterators/prolonged_sound_marks.rb

@@ -0,0 +1,206 @@
+# frozen_string_literal: true
+
+module Yosina
+  module Transliterators
+    # Handle prolonged sound marks transliterator
+    module ProlongedSoundMarks
+      # Mix-in for character type checks
+      module CharType
+        # Hyphen-like characters that can be converted to prolonged sound marks
+        HYPHEN_LIKE_CHARS = [
+          0x002d,  # HYPHEN-MINUS
+          0x2010,  # HYPHEN
+          0x2014,  # EM DASH
+          0x2015,  # HORIZONTAL BAR
+          0x2212,  # MINUS SIGN
+          0xff0d,  # FULLWIDTH HYPHEN-MINUS
+          0xff70,  # HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK (already converted)
+          0x30fc   # KATAKANA-HIRAGANA PROLONGED SOUND MARK (already converted)
+        ].freeze
+
+        # Check if character is halfwidth Japanese
+        def halfwidth?(char_code)
+          halfwidth_alphanumeric?(char_code) ||
+            (char_code >= 0xff66 && char_code <= 0xff6f) ||
+            (char_code >= 0xff70 && char_code <= 0xff9f)
+        end
+
+        def hiragana?(char_code)
+          (char_code >= 0x3041 && char_code <= 0x309c &&
+            char_code != 0x3063 && char_code != 0x3093) ||
+            char_code == 0x309f
+        end
+
+        def katakana?(char_code)
+          (char_code >= 0x30a1 && char_code <= 0x30fa && char_code != 0x30c3 && char_code != 0x30f3) ||
+            (char_code >= 0x30fd && char_code <= 0x30ff) ||
+            (char_code >= 0xff70 && char_code <= 0xff9f && char_code != 0xff6f && char_code != 0xff9d)
+        end
+
+        def halfwidth_alphanumeric?(char_code)
+          (char_code >= 0x30 && char_code <= 0x39) ||
+            (char_code >= 0x41 && char_code <= 0x5A) ||
+            (char_code >= 0x61 && char_code <= 0x7A)
+        end
+
+        def fullwidth_alphanumeric?(char_code)
+          (char_code >= 0xff10 && char_code <= 0xff19) ||
+            (char_code >= 0xff21 && char_code <= 0xff3a) ||
+            (char_code >= 0xff41 && char_code <= 0xff5a)
+        end
+
+        # Check if character is fullwidth Japanese
+        def fullwidth?(char_code)
+          char_code == 0x30fc || hiragana?(char_code) || katakana?(char_code) || fullwidth_alphanumeric?(char_code)
+        end
+
+        # Check if character is alphanumeric
+        def alphanumeric?(char_code)
+          halfwidth_alphanumeric?(char_code) || fullwidth_alphanumeric?(char_code)
+        end
+
+        def hatsuon?(char_code)
+          [0x3093, 0x30f3, 0xff9d].include?(char_code)
+        end
+
+        def sokuon?(char_code)
+          [0x3063, 0x30c3, 0xff6f].include?(char_code)
+        end
+
+        def prolonged_sound_mark?(char_code)
+          [0x30fc, 0xff70].include?(char_code)
+        end
+
+        def prolongable?(char_code)
+          prolonged_sound_mark?(char_code) || hiragana?(char_code) || katakana?(char_code)
+        end
+
+        def hyphen_like?(char_code)
+          HYPHEN_LIKE_CHARS.include?(char_code)
+        end
+      end
+
+      # Transliterator for prolonged sound marks
+      class Transliterator < Yosina::BaseTransliterator
+        attr_reader :skip_already_transliterated_chars, :allow_prolonged_hatsuon,
+                    :allow_prolonged_sokuon, :replace_prolonged_marks_following_alnums
+
+        # Initialize the transliterator with options
+        #
+        # @param options [Hash] Configuration options
+        # @option options [Boolean] :skip_already_transliterated_chars Skip chars that were already processed.
+        #     Default: false.
+        # @option options [Boolean] :allow_prolonged_hatsuon Allow prolonging ん/ン/ﾝ.
+        #     Default: false.
+        # @option options [Boolean] :allow_prolonged_sokuon Allow prolonging っ/ッ/ｯ.
+        #     Default: false.
+        # @option options [Boolean] :replace_prolonged_marks_following_alnums Replace prolonged marks after alphanum
+        #    with hyphens. Default: false.
+        def initialize(options = {})
+          super()
+          @skip_already_transliterated_chars = options.fetch(:skip_already_transliterated_chars, false)
+          @allow_prolonged_hatsuon = options.fetch(:allow_prolonged_hatsuon, false)
+          @allow_prolonged_sokuon = options.fetch(:allow_prolonged_sokuon, false)
+          @replace_prolonged_marks_following_alnums = options.fetch(:replace_prolonged_marks_following_alnums, false)
+        end
+
+        # Convert hyphen-like characters to appropriate prolonged sound marks
+        #
+        # @param input_chars [Enumerable<Char>] The characters to transliterate
+        # @return [Enumerable<Char>] The transliterated characters
+        def call(input_chars)
+          offset = 0
+          processed_char_in_lookahead = false
+          lookahead_buf = []
+          last_non_prolonged_char = nil
+
+          Chars.enum do |y|
+            input_chars.each do |char|
+              unless lookahead_buf.empty?
+                if !char.c.empty? && hyphen_like?(char.c.ord)
+                  processed_char_in_lookahead = true unless char.source.nil?
+                  lookahead_buf << char
+                  next
+                end
+                prev_non_prolonged_char = last_non_prolonged_char
+                last_non_prolonged_char = char
+
+                if (prev_non_prolonged_char.nil? || alphanumeric?(prev_non_prolonged_char.c.ord)) && (
+                  !@skip_already_transliterated_chars || !processed_char_in_lookahead
+                )
+                  halfwidth = halfwidth?(
+                    prev_non_prolonged_char.nil? ? last_non_prolonged_char.c.ord : prev_non_prolonged_char.c.ord
+                  )
+                  replacement = halfwidth ? "\u002d" : "\uff0d"
+                  lookahead_buf.each do |buffered_char|
+                    y << Char.new(c: replacement, offset: offset, source: buffered_char)
+                    offset += replacement.length
+                  end
+                else
+                  lookahead_buf.each do |buffered_char|
+                    y << buffered_char.with_offset(offset)
+                    offset += buffered_char.c.length
+                  end
+                end
+
+                lookahead_buf.clear
+                y << char.with_offset(offset)
+                offset += char.c.length
+                last_non_prolonged_char = char
+                processed_char_in_lookahead = false
+                next
+              end
+              if !char.c.empty? && hyphen_like?(char.c.ord)
+                should_procses = !@skip_already_transliterated_chars || !char.transliterated?
+                if should_procses && !last_non_prolonged_char.nil?
+                  if prolongable_char?(last_non_prolonged_char.c.ord)
+                    replacement = halfwidth?(last_non_prolonged_char.c.ord) ? "\uff70" : "\u30fc"
+                    y << Char.new(c: replacement, offset: offset, source: char)
+                    offset += replacement.length
+                    next
+                  elsif @replace_prolonged_marks_following_alnums && alphanumeric?(last_non_prolonged_char.c.ord)
+                    lookahead_buf << char
+                    next
+                  end
+                end
+              else
+                last_non_prolonged_char = char
+              end
+              y << char.with_offset(offset)
+              offset += char.c.length
+            end
+          end
+        end
+
+        private
+
+        include CharType
+
+        # Check if a character can be prolonged
+        #
+        # @param char [String] The character to check
+        # @return [Boolean] True if the character can be prolonged
+        def prolongable_char?(char_code)
+          # Default prolongable characters
+          return true if prolongable?(char_code)
+
+          # Hatsuon (if allowed)
+          return true if @allow_prolonged_hatsuon && hatsuon?(char_code)
+
+          # Sokuon (if allowed)
+          return true if @allow_prolonged_sokuon && sokuon?(char_code)
+
+          false
+        end
+      end
+
+      # Factory method to create a prolonged sound marks transliterator
+      #
+      # @param options [Hash] Configuration options
+      # @return [Transliterator] A new prolonged sound marks transliterator instance
+      def self.call(options = {})
+        Transliterator.new(options)
+      end
+    end
+  end
+end

data/lib/yosina/transliterators/radicals.rb

@@ -0,0 +1,361 @@
+# frozen_string_literal: true
+
+module Yosina
+  module Transliterators
+    # Replace Kangxi radicals with equivalent CJK ideographs
+    module Radicals
+      # Generated mapping data from radicals.json
+      RADICALS_MAPPINGS = {
+        "\u{2f00}" => "\u{4e00}",
+        "\u{2f01}" => "\u{4e28}",
+        "\u{2f02}" => "\u{4e36}",
+        "\u{2f03}" => "\u{4e3f}",
+        "\u{2f04}" => "\u{4e59}",
+        "\u{2f05}" => "\u{4e85}",
+        "\u{2f06}" => "\u{4e8c}",
+        "\u{2f07}" => "\u{4ea0}",
+        "\u{2f08}" => "\u{4eba}",
+        "\u{2f09}" => "\u{513f}",
+        "\u{2f0a}" => "\u{5165}",
+        "\u{2f0b}" => "\u{516b}",
+        "\u{2f0c}" => "\u{5182}",
+        "\u{2f0d}" => "\u{5196}",
+        "\u{2f0e}" => "\u{51ab}",
+        "\u{2f0f}" => "\u{51e0}",
+        "\u{2f10}" => "\u{51f5}",
+        "\u{2f11}" => "\u{5200}",
+        "\u{2f12}" => "\u{529b}",
+        "\u{2f13}" => "\u{52f9}",
+        "\u{2f14}" => "\u{5315}",
+        "\u{2f15}" => "\u{531a}",
+        "\u{2f16}" => "\u{5338}",
+        "\u{2f17}" => "\u{5341}",
+        "\u{2f18}" => "\u{535c}",
+        "\u{2f19}" => "\u{5369}",
+        "\u{2f1a}" => "\u{5382}",
+        "\u{2f1b}" => "\u{53b6}",
+        "\u{2f1c}" => "\u{53c8}",
+        "\u{2f1d}" => "\u{53e3}",
+        "\u{2f1e}" => "\u{56d7}",
+        "\u{2f1f}" => "\u{571f}",
+        "\u{2f20}" => "\u{58eb}",
+        "\u{2f21}" => "\u{5902}",
+        "\u{2f22}" => "\u{590a}",
+        "\u{2f23}" => "\u{5915}",
+        "\u{2f24}" => "\u{5927}",
+        "\u{2f25}" => "\u{5973}",
+        "\u{2f26}" => "\u{5b50}",
+        "\u{2f27}" => "\u{5b80}",
+        "\u{2f28}" => "\u{5bf8}",
+        "\u{2f29}" => "\u{5c0f}",
+        "\u{2f2a}" => "\u{5c22}",
+        "\u{2f2b}" => "\u{5c38}",
+        "\u{2f2c}" => "\u{5c6e}",
+        "\u{2f2d}" => "\u{5c71}",
+        "\u{2f2e}" => "\u{5ddb}",
+        "\u{2f2f}" => "\u{5de5}",
+        "\u{2f30}" => "\u{5df1}",
+        "\u{2f31}" => "\u{5dfe}",
+        "\u{2f32}" => "\u{5e72}",
+        "\u{2f33}" => "\u{5e7a}",
+        "\u{2f34}" => "\u{5e7f}",
+        "\u{2f35}" => "\u{5ef4}",
+        "\u{2f36}" => "\u{5efe}",
+        "\u{2f37}" => "\u{5f0b}",
+        "\u{2f38}" => "\u{5f13}",
+        "\u{2f39}" => "\u{5f50}",
+        "\u{2f3a}" => "\u{5f61}",
+        "\u{2f3b}" => "\u{5f73}",
+        "\u{2f3c}" => "\u{5fc3}",
+        "\u{2f3d}" => "\u{6208}",
+        "\u{2f3e}" => "\u{6236}",
+        "\u{2f3f}" => "\u{624b}",
+        "\u{2f40}" => "\u{652f}",
+        "\u{2f41}" => "\u{6534}",
+        "\u{2f42}" => "\u{6587}",
+        "\u{2f43}" => "\u{6597}",
+        "\u{2f44}" => "\u{65a4}",
+        "\u{2f45}" => "\u{65b9}",
+        "\u{2f46}" => "\u{65e0}",
+        "\u{2f47}" => "\u{65e5}",
+        "\u{2f48}" => "\u{66f0}",
+        "\u{2f49}" => "\u{6708}",
+        "\u{2f4a}" => "\u{6728}",
+        "\u{2f4b}" => "\u{6b20}",
+        "\u{2f4c}" => "\u{6b62}",
+        "\u{2f4d}" => "\u{6b79}",
+        "\u{2f4e}" => "\u{6bb3}",
+        "\u{2f4f}" => "\u{6bcb}",
+        "\u{2f50}" => "\u{6bd4}",
+        "\u{2f51}" => "\u{6bdb}",
+        "\u{2f52}" => "\u{6c0f}",
+        "\u{2f53}" => "\u{6c14}",
+        "\u{2f54}" => "\u{6c34}",
+        "\u{2f55}" => "\u{706b}",
+        "\u{2f56}" => "\u{722a}",
+        "\u{2f57}" => "\u{7236}",
+        "\u{2f58}" => "\u{723b}",
+        "\u{2f59}" => "\u{723f}",
+        "\u{2f5a}" => "\u{7247}",
+        "\u{2f5b}" => "\u{7259}",
+        "\u{2f5c}" => "\u{725b}",
+        "\u{2f5d}" => "\u{72ac}",
+        "\u{2f5e}" => "\u{7384}",
+        "\u{2f5f}" => "\u{7389}",
+        "\u{2f60}" => "\u{74dc}",
+        "\u{2f61}" => "\u{74e6}",
+        "\u{2f62}" => "\u{7518}",
+        "\u{2f63}" => "\u{751f}",
+        "\u{2f64}" => "\u{7528}",
+        "\u{2f65}" => "\u{7530}",
+        "\u{2f66}" => "\u{758b}",
+        "\u{2f67}" => "\u{7592}",
+        "\u{2f68}" => "\u{7676}",
+        "\u{2f69}" => "\u{767d}",
+        "\u{2f6a}" => "\u{76ae}",
+        "\u{2f6b}" => "\u{76bf}",
+        "\u{2f6c}" => "\u{76ee}",
+        "\u{2f6d}" => "\u{77db}",
+        "\u{2f6e}" => "\u{77e2}",
+        "\u{2f6f}" => "\u{77f3}",
+        "\u{2f70}" => "\u{793a}",
+        "\u{2f71}" => "\u{79b8}",
+        "\u{2f72}" => "\u{79be}",
+        "\u{2f73}" => "\u{7a74}",
+        "\u{2f74}" => "\u{7acb}",
+        "\u{2f75}" => "\u{7af9}",
+        "\u{2f76}" => "\u{7c73}",
+        "\u{2f77}" => "\u{7cf8}",
+        "\u{2f78}" => "\u{7f36}",
+        "\u{2f79}" => "\u{7f51}",
+        "\u{2f7a}" => "\u{7f8a}",
+        "\u{2f7b}" => "\u{7fbd}",
+        "\u{2f7c}" => "\u{8001}",
+        "\u{2f7d}" => "\u{800c}",
+        "\u{2f7e}" => "\u{8012}",
+        "\u{2f7f}" => "\u{8033}",
+        "\u{2f80}" => "\u{807f}",
+        "\u{2f81}" => "\u{8089}",
+        "\u{2f82}" => "\u{81e3}",
+        "\u{2f83}" => "\u{81ea}",
+        "\u{2f84}" => "\u{81f3}",
+        "\u{2f85}" => "\u{81fc}",
+        "\u{2f86}" => "\u{820c}",
+        "\u{2f87}" => "\u{821b}",
+        "\u{2f88}" => "\u{821f}",
+        "\u{2f89}" => "\u{826e}",
+        "\u{2f8a}" => "\u{8272}",
+        "\u{2f8b}" => "\u{8278}",
+        "\u{2f8c}" => "\u{864d}",
+        "\u{2f8d}" => "\u{866b}",
+        "\u{2f8e}" => "\u{8840}",
+        "\u{2f8f}" => "\u{884c}",
+        "\u{2f90}" => "\u{8863}",
+        "\u{2f91}" => "\u{897e}",
+        "\u{2f92}" => "\u{898b}",
+        "\u{2f93}" => "\u{89d2}",
+        "\u{2f94}" => "\u{8a00}",
+        "\u{2f95}" => "\u{8c37}",
+        "\u{2f96}" => "\u{8c46}",
+        "\u{2f97}" => "\u{8c55}",
+        "\u{2f98}" => "\u{8c78}",
+        "\u{2f99}" => "\u{8c9d}",
+        "\u{2f9a}" => "\u{8d64}",
+        "\u{2f9b}" => "\u{8d70}",
+        "\u{2f9c}" => "\u{8db3}",
+        "\u{2f9d}" => "\u{8eab}",
+        "\u{2f9e}" => "\u{8eca}",
+        "\u{2f9f}" => "\u{8f9b}",
+        "\u{2fa0}" => "\u{8fb0}",
+        "\u{2fa1}" => "\u{8fb5}",
+        "\u{2fa2}" => "\u{9091}",
+        "\u{2fa3}" => "\u{9149}",
+        "\u{2fa4}" => "\u{91c6}",
+        "\u{2fa5}" => "\u{91cc}",
+        "\u{2fa6}" => "\u{91d1}",
+        "\u{2fa7}" => "\u{9577}",
+        "\u{2fa8}" => "\u{9580}",
+        "\u{2fa9}" => "\u{961c}",
+        "\u{2faa}" => "\u{96b6}",
+        "\u{2fab}" => "\u{96b9}",
+        "\u{2fac}" => "\u{96e8}",
+        "\u{2fad}" => "\u{9751}",
+        "\u{2fae}" => "\u{975e}",
+        "\u{2faf}" => "\u{9762}",
+        "\u{2fb0}" => "\u{9769}",
+        "\u{2fb1}" => "\u{97cb}",
+        "\u{2fb2}" => "\u{97ed}",
+        "\u{2fb3}" => "\u{97f3}",
+        "\u{2fb4}" => "\u{9801}",
+        "\u{2fb5}" => "\u{98a8}",
+        "\u{2fb6}" => "\u{98db}",
+        "\u{2fb7}" => "\u{98df}",
+        "\u{2fb8}" => "\u{9996}",
+        "\u{2fb9}" => "\u{9999}",
+        "\u{2fba}" => "\u{99ac}",
+        "\u{2fbb}" => "\u{9aa8}",
+        "\u{2fbc}" => "\u{9ad8}",
+        "\u{2fbd}" => "\u{9adf}",
+        "\u{2fbe}" => "\u{9b25}",
+        "\u{2fbf}" => "\u{9b2f}",
+        "\u{2fc0}" => "\u{9b32}",
+        "\u{2fc1}" => "\u{9b3c}",
+        "\u{2fc2}" => "\u{9b5a}",
+        "\u{2fc3}" => "\u{9ce5}",
+        "\u{2fc4}" => "\u{9e75}",
+        "\u{2fc5}" => "\u{9e7f}",
+        "\u{2fc6}" => "\u{9ea5}",
+        "\u{2fc7}" => "\u{9ebb}",
+        "\u{2fc8}" => "\u{9ec3}",
+        "\u{2fc9}" => "\u{9ecd}",
+        "\u{2fca}" => "\u{9ed1}",
+        "\u{2fcb}" => "\u{9ef9}",
+        "\u{2fcc}" => "\u{9efd}",
+        "\u{2fcd}" => "\u{9f0e}",
+        "\u{2fce}" => "\u{9f13}",
+        "\u{2fcf}" => "\u{9f20}",
+        "\u{2fd0}" => "\u{9f3b}",
+        "\u{2fd1}" => "\u{9f4a}",
+        "\u{2fd2}" => "\u{9f52}",
+        "\u{2fd3}" => "\u{9f8d}",
+        "\u{2fd4}" => "\u{9f9c}",
+        "\u{2fd5}" => "\u{9fa0}",
+        "\u{2e80}" => "\u{51ab}",
+        "\u{2e81}" => "\u{5382}",
+        "\u{2e82}" => "\u{4e5b}",
+        "\u{2e83}" => "\u{4e5a}",
+        "\u{2e84}" => "\u{4e59}",
+        "\u{2e85}" => "\u{4ebb}",
+        "\u{2e86}" => "\u{5182}",
+        "\u{2e89}" => "\u{5202}",
+        "\u{2e8a}" => "\u{535c}",
+        "\u{2e8b}" => "\u{353e}",
+        "\u{2e8e}" => "\u{5140}",
+        "\u{2e8f}" => "\u{5c23}",
+        "\u{2e90}" => "\u{5c22}",
+        "\u{2e92}" => "\u{5df3}",
+        "\u{2e93}" => "\u{5e7a}",
+        "\u{2e94}" => "\u{5f51}",
+        "\u{2e95}" => "\u{5f50}",
+        "\u{2e96}" => "\u{5fc4}",
+        "\u{2e97}" => "\u{38fa}",
+        "\u{2e98}" => "\u{624c}",
+        "\u{2e99}" => "\u{6535}",
+        "\u{2e9b}" => "\u{65e1}",
+        "\u{2e9d}" => "\u{6708}",
+        "\u{2e9e}" => "\u{6b7a}",
+        "\u{2e9f}" => "\u{6bcd}",
+        "\u{2ea0}" => "\u{6c11}",
+        "\u{2ea1}" => "\u{6c35}",
+        "\u{2ea2}" => "\u{6c3a}",
+        "\u{2ea3}" => "\u{706c}",
+        "\u{2ea5}" => "\u{722b}",
+        "\u{2ea6}" => "\u{4e2c}",
+        "\u{2ea8}" => "\u{72ad}",
+        "\u{2eab}" => "\u{7f52}",
+        "\u{2eac}" => "\u{793a}",
+        "\u{2ead}" => "\u{793b}",
+        "\u{2eaf}" => "\u{7cf9}",
+        "\u{2eb0}" => "\u{7e9f}",
+        "\u{2eb1}" => "\u{7f53}",
+        "\u{2eb3}" => "\u{34c1}",
+        "\u{2eb4}" => "\u{34c1}",
+        "\u{2eb9}" => "\u{8002}",
+        "\u{2eba}" => "\u{8080}",
+        "\u{2ebc}" => "\u{6708}",
+        "\u{2ebd}" => "\u{81fc}",
+        "\u{2ebe}" => "\u{8279}",
+        "\u{2ebf}" => "\u{8279}",
+        "\u{2ec0}" => "\u{8279}",
+        "\u{2ec1}" => "\u{864e}",
+        "\u{2ec2}" => "\u{8864}",
+        "\u{2ec3}" => "\u{8980}",
+        "\u{2ec4}" => "\u{897f}",
+        "\u{2ec5}" => "\u{89c1}",
+        "\u{2ec8}" => "\u{8ba0}",
+        "\u{2ec9}" => "\u{8d1d}",
+        "\u{2ecb}" => "\u{8f66}",
+        "\u{2ecd}" => "\u{8fb6}",
+        "\u{2ecf}" => "\u{961d}",
+        "\u{2ed0}" => "\u{9485}",
+        "\u{2ed1}" => "\u{9577}",
+        "\u{2ed2}" => "\u{9578}",
+        "\u{2ed3}" => "\u{957f}",
+        "\u{2ed6}" => "\u{961d}",
+        "\u{2ed8}" => "\u{9752}",
+        "\u{2ed9}" => "\u{97e6}",
+        "\u{2eda}" => "\u{9875}",
+        "\u{2edb}" => "\u{98ce}",
+        "\u{2edc}" => "\u{98de}",
+        "\u{2edd}" => "\u{98df}",
+        "\u{2edf}" => "\u{98e0}",
+        "\u{2ee0}" => "\u{9963}",
+        "\u{2ee2}" => "\u{9a6c}",
+        "\u{2ee3}" => "\u{9aa8}",
+        "\u{2ee4}" => "\u{9b3c}",
+        "\u{2ee5}" => "\u{9c7c}",
+        "\u{2ee6}" => "\u{9e1f}",
+        "\u{2ee7}" => "\u{5364}",
+        "\u{2ee8}" => "\u{9ea6}",
+        "\u{2ee9}" => "\u{9ec4}",
+        "\u{2eea}" => "\u{9efe}",
+        "\u{2eeb}" => "\u{6589}",
+        "\u{2eec}" => "\u{9f50}",
+        "\u{2eed}" => "\u{6b6f}",
+        "\u{2eee}" => "\u{9f7f}",
+        "\u{2eef}" => "\u{7adc}",
+        "\u{2ef0}" => "\u{9f99}",
+        "\u{2ef1}" => "\u{9f9c}",
+        "\u{2ef2}" => "\u{4e80}",
+        "\u{2ef3}" => "\u{9f9f}"
+      }.freeze
+
+      # Transliterator for radicals
+      class Transliterator < Yosina::BaseTransliterator
+        # Initialize the transliterator with options
+        #
+        # @param _options [Hash] Configuration options (currently unused)
+        def initialize(_options = {})
+          # Options currently unused for radicals transliterator
+          super()
+        end
+
+        # Replace Kangxi radicals with equivalent CJK ideographs
+        #
+        # @param input_chars [Enumerable<Char>] The characters to transliterate
+        # @return [Enumerable<Char>] The transliterated characters
+        def call(input_chars)
+          offset = 0
+
+          result = input_chars.filter_map do |char|
+            replacement = RADICALS_MAPPINGS[char.c]
+            c = if replacement
+                  # Skip empty replacements (character removal)
+                  next if replacement.empty?
+
+                  Char.new(c: replacement, offset: offset, source: char)
+                else
+                  char.with_offset(offset)
+                end
+            offset += c.c.length
+            c
+          end
+
+          class << result
+            include Yosina::Chars
+          end
+
+          result
+        end
+      end
+
+      # Factory method to create a radicals transliterator
+      #
+      # @param options [Hash] Configuration options
+      # @return [Transliterator] A new radicals transliterator instance
+      def self.call(options = {})
+        Transliterator.new(options)
+      end
+    end
+  end
+end

data/lib/yosina/transliterators/spaces.rb

@@ -0,0 +1,79 @@
+# frozen_string_literal: true
+
+module Yosina
+  module Transliterators
+    # Replace various space characters with plain whitespace
+    module Spaces
+      # Generated mapping data from spaces.json
+      SPACES_MAPPINGS = {
+        "\u{a0}" => ' ',
+        "\u{180e}" => '',
+        "\u{2000}" => ' ',
+        "\u{2001}" => ' ',
+        "\u{2002}" => ' ',
+        "\u{2003}" => ' ',
+        "\u{2004}" => ' ',
+        "\u{2005}" => ' ',
+        "\u{2006}" => ' ',
+        "\u{2007}" => ' ',
+        "\u{2008}" => ' ',
+        "\u{2009}" => ' ',
+        "\u{200a}" => ' ',
+        "\u{200b}" => ' ',
+        "\u{202f}" => ' ',
+        "\u{205f}" => ' ',
+        "\u{3000}" => ' ',
+        "\u{3164}" => ' ',
+        "\u{ffa0}" => ' ',
+        "\u{feff}" => ''
+      }.freeze
+
+      # Transliterator for spaces
+      class Transliterator < Yosina::BaseTransliterator
+        # Initialize the transliterator with options
+        #
+        # @param _options [Hash] Configuration options (currently unused)
+        def initialize(_options = {})
+          # Options currently unused for spaces transliterator
+          super()
+        end
+
+        # Replace various space characters with plain whitespace
+        #
+        # @param input_chars [Enumerable<Char>] The characters to transliterate
+        # @return [Enumerable<Char>] The transliterated characters
+        def call(input_chars)
+          offset = 0
+
+          result = input_chars.filter_map do |char|
+            replacement = SPACES_MAPPINGS[char.c]
+            c = if replacement
+                  # Skip empty replacements (character removal)
+                  next if replacement.empty?
+
+                  Char.new(c: replacement, offset: offset, source: char)
+                else
+                  char.with_offset(offset)
+                end
+            offset += c.c.length
+            c
+          end
+
+          class << result
+            include Yosina::Chars
+          end
+
+          result
+        end
+      end
+
+      # Factory method to create a spaces transliterator
+      #
+      # @param options [Hash] Configuration options
+      # @return [Transliterator] A new spaces transliterator instance
+      def self.call(options = {})
+        Transliterator.new(options)
+      end
+    end
+  end
+end