yosina 0.2.0 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4831a722a9b91da965e6aa838c2af758ba86cdc445577cd2ca153614e6f38d7b
-  data.tar.gz: 85961c7d59c4d8e8032f93108c5201a47cfa3b8598bcb686dad1e4da34ac3787
+  metadata.gz: c3683e10291faa325375fe5c3b069cdc6069e1654a7aae8c9cc754d9cf33b17b
+  data.tar.gz: 891d3b94688ae9af9693cd61b8edc30b9c2e0388662586a33de39ebb900bb4bf
 SHA512:
-  metadata.gz: 744642abe6c4cbf69848552953008697a1ddc98765331703800a91c5b504dbd3697a2de80992222c37130fe70aeb2b14d813b88a2ea76fde83a11caf04d901e6
-  data.tar.gz: a307fb1706da402d99c0c865ad5e8ed293dfdf8f8bd69a59aab3aa9c9172b9cd1461b8075a6934e9d7f496cfe971a12d02a89cd00d2af90d78c55d98b7b04fa8
+  metadata.gz: 86a08510acaafae1a2afe8b72b194204839e6aa63a07e44b4ebd1219048722536e38d6bcc5972f3c3a543fd5f94648b6a98dbe9f6ee6c90000eacc0a0a6495d6
+  data.tar.gz: 0ce9a44af3df6dcb1f89bc6e3b2661da834008a0d3c972c69acac5c3a644ffe4455ac9ca70cb5556980456e6e7ce4e50b8beb324ad9134531e936141b4c06f4d

data/codegen/dataset.rb

@@ -14,6 +14,8 @@ DatasetSourceDefs = Struct.new(
   :combined,
   :circled_or_squared,
   :roman_numerals,
+  :archaic_hirakatas,
+  :small_hirakatas,
   keyword_init: true
 )
 
@@ -29,6 +31,8 @@ Dataset = Struct.new(
   :combined,
   :circled_or_squared,
   :roman_numerals,
+  :archaic_hirakatas,
+  :small_hirakatas,
   keyword_init: true
 )
 
@@ -241,6 +245,8 @@ def build_dataset_from_data_root(data_root, defs)
     kanji_old_new: load_kanji_old_new_data(data_root / defs.kanji_old_new),
     combined: load_combined_data(data_root / defs.combined),
     circled_or_squared: load_circled_or_squared_data(data_root / defs.circled_or_squared),
-    roman_numerals: load_roman_numerals_data(data_root / defs.roman_numerals)
+    roman_numerals: load_roman_numerals_data(data_root / defs.roman_numerals),
+    archaic_hirakatas: load_simple_data(data_root / defs.archaic_hirakatas),
+    small_hirakatas: load_simple_data(data_root / defs.small_hirakatas)
   )
 end

data/codegen/main.rb

@@ -32,7 +32,9 @@ def main
     kanji_old_new: 'kanji-old-new-form.json',
     combined: 'combined-chars.json',
     circled_or_squared: 'circled-or-squared.json',
-    roman_numerals: 'roman-numerals.json'
+    roman_numerals: 'roman-numerals.json',
+    archaic_hirakatas: 'archaic-hirakatas.json',
+    small_hirakatas: 'small-hirakatas.json'
   )
 
   # Load the dataset
@@ -64,6 +66,16 @@ def main
       'kanji_old_new',
       'Replace old-style kanji with modern equivalents',
       dataset.kanji_old_new
+    ],
+    [
+      'archaic_hirakatas',
+      'Replaces archaic kana (hentaigana) with their modern equivalents.',
+      dataset.archaic_hirakatas
+    ],
+    [
+      'small_hirakatas',
+      'Replaces small hiragana/katakana with their ordinary-sized equivalents.',
+      dataset.small_hirakatas
     ]
   ]
 

data/lib/yosina/recipes.rb

@@ -51,12 +51,15 @@ module Yosina
   end
 
   # Configuration recipe for building transliterator chains
+  # rubocop:disable Metrics/ClassLength
   class TransliterationRecipe
     attr_accessor :kanji_old_new, :hira_kata, :replace_japanese_iteration_marks,
                   :replace_suspicious_hyphens_to_prolonged_sound_marks,
                   :replace_combined_characters, :replace_circled_or_squared_characters,
                   :replace_ideographic_annotations, :replace_radicals, :replace_spaces,
                   :replace_hyphens, :replace_mathematical_alphanumerics, :replace_roman_numerals,
+                  :replace_archaic_hirakatas, :replace_small_hirakatas,
+                  :convert_historical_hirakatas,
                   :combine_decomposed_hiraganas_and_katakanas, :to_fullwidth, :to_halfwidth,
                   :remove_ivs_svs, :charset
 
@@ -161,6 +164,8 @@ module Yosina
                    replace_ideographic_annotations: false, replace_radicals: false,
                    replace_spaces: false, replace_hyphens: false,
                    replace_mathematical_alphanumerics: false, replace_roman_numerals: false,
+                   replace_archaic_hirakatas: false, replace_small_hirakatas: false,
+                   convert_historical_hirakatas: nil,
                    combine_decomposed_hiraganas_and_katakanas: false,
                    to_fullwidth: false, to_halfwidth: false, remove_ivs_svs: false,
                    charset: 'unijis_2004')
@@ -176,6 +181,9 @@ module Yosina
       @replace_hyphens = replace_hyphens
       @replace_mathematical_alphanumerics = replace_mathematical_alphanumerics
       @replace_roman_numerals = replace_roman_numerals
+      @replace_archaic_hirakatas = replace_archaic_hirakatas
+      @replace_small_hirakatas = replace_small_hirakatas
+      @convert_historical_hirakatas = convert_historical_hirakatas
       @combine_decomposed_hiraganas_and_katakanas = combine_decomposed_hiraganas_and_katakanas
       @to_fullwidth = to_fullwidth
       @to_halfwidth = to_halfwidth
@@ -208,6 +216,9 @@ module Yosina
       ctx = apply_replace_hyphens(ctx)
       ctx = apply_replace_mathematical_alphanumerics(ctx)
       ctx = apply_replace_roman_numerals(ctx)
+      ctx = apply_replace_archaic_hirakatas(ctx)
+      ctx = apply_replace_small_hirakatas(ctx)
+      ctx = apply_convert_historical_hirakatas(ctx)
       ctx = apply_combine_decomposed_hiraganas_and_katakanas(ctx)
       ctx = apply_to_fullwidth(ctx)
       ctx = apply_hira_kata(ctx)
@@ -241,7 +252,7 @@ module Yosina
 
     def apply_hira_kata(ctx)
       if @hira_kata
-        ctx.insert_middle([:hira_kata, { mode: @hira_kata }])
+        ctx.insert_tail([:hira_kata, { mode: @hira_kata }])
       else
         ctx
       end
@@ -329,6 +340,35 @@ module Yosina
       end
     end
 
+    def apply_replace_archaic_hirakatas(ctx)
+      if @replace_archaic_hirakatas
+        ctx.insert_middle([:archaic_hirakatas, {}])
+      else
+        ctx
+      end
+    end
+
+    def apply_replace_small_hirakatas(ctx)
+      if @replace_small_hirakatas
+        ctx.insert_middle([:small_hirakatas, {}])
+      else
+        ctx
+      end
+    end
+
+    def apply_convert_historical_hirakatas(ctx)
+      if @convert_historical_hirakatas
+        mode = @convert_historical_hirakatas
+        ctx.insert_middle([:historical_hirakatas, {
+                            hiraganas: mode,
+                            katakanas: mode,
+                            voiced_katakanas: mode == 'decompose' ? 'decompose' : 'skip'
+                          }])
+      else
+        ctx
+      end
+    end
+
     def apply_combine_decomposed_hiraganas_and_katakanas(ctx)
       if @combine_decomposed_hiraganas_and_katakanas
         ctx.insert_head([:hira_kata_composition, { compose_non_combining_marks: true }])
@@ -373,3 +413,4 @@ module Yosina
     recipe.build_transliterator_configs
   end
 end
+# rubocop:enable Metrics/ClassLength

data/lib/yosina/transliterators/archaic_hirakatas.rb

@@ -0,0 +1,350 @@
+# frozen_string_literal: true
+
+module Yosina
+  module Transliterators
+    # Replaces archaic kana (hentaigana) with their modern equivalents.
+    module ArchaicHirakatas
+      # Generated mapping data from archaic_hirakatas.json
+      ARCHAIC_HIRAKATAS_MAPPINGS = {
+        "\u{1b000}" => "\u{30a8}",
+        "\u{1b001}" => "\u{3048}",
+        "\u{1b002}" => "\u{3042}",
+        "\u{1b003}" => "\u{3042}",
+        "\u{1b004}" => "\u{3042}",
+        "\u{1b005}" => "\u{3042}",
+        "\u{1b006}" => "\u{3044}",
+        "\u{1b007}" => "\u{3044}",
+        "\u{1b008}" => "\u{3044}",
+        "\u{1b009}" => "\u{3044}",
+        "\u{1b00a}" => "\u{3046}",
+        "\u{1b00b}" => "\u{3046}",
+        "\u{1b00c}" => "\u{3046}",
+        "\u{1b00d}" => "\u{3046}",
+        "\u{1b00e}" => "\u{3046}",
+        "\u{1b00f}" => "\u{3048}",
+        "\u{1b010}" => "\u{3048}",
+        "\u{1b011}" => "\u{3048}",
+        "\u{1b012}" => "\u{3048}",
+        "\u{1b013}" => "\u{3048}",
+        "\u{1b014}" => "\u{304a}",
+        "\u{1b015}" => "\u{304a}",
+        "\u{1b016}" => "\u{304a}",
+        "\u{1b017}" => "\u{304b}",
+        "\u{1b018}" => "\u{304b}",
+        "\u{1b019}" => "\u{304b}",
+        "\u{1b01a}" => "\u{304b}",
+        "\u{1b01b}" => "\u{304b}",
+        "\u{1b01c}" => "\u{304b}",
+        "\u{1b01d}" => "\u{304b}",
+        "\u{1b01e}" => "\u{304b}",
+        "\u{1b01f}" => "\u{304b}",
+        "\u{1b020}" => "\u{304b}",
+        "\u{1b021}" => "\u{304b}",
+        "\u{1b022}" => "\u{304b}",
+        "\u{1b023}" => "\u{304d}",
+        "\u{1b024}" => "\u{304d}",
+        "\u{1b025}" => "\u{304d}",
+        "\u{1b026}" => "\u{304d}",
+        "\u{1b027}" => "\u{304d}",
+        "\u{1b028}" => "\u{304d}",
+        "\u{1b029}" => "\u{304d}",
+        "\u{1b02a}" => "\u{304d}",
+        "\u{1b02b}" => "\u{304f}",
+        "\u{1b02c}" => "\u{304f}",
+        "\u{1b02d}" => "\u{304f}",
+        "\u{1b02e}" => "\u{304f}",
+        "\u{1b02f}" => "\u{304f}",
+        "\u{1b030}" => "\u{304f}",
+        "\u{1b031}" => "\u{304f}",
+        "\u{1b032}" => "\u{3051}",
+        "\u{1b033}" => "\u{3051}",
+        "\u{1b034}" => "\u{3051}",
+        "\u{1b035}" => "\u{3051}",
+        "\u{1b036}" => "\u{3051}",
+        "\u{1b037}" => "\u{3051}",
+        "\u{1b038}" => "\u{3053}",
+        "\u{1b039}" => "\u{3053}",
+        "\u{1b03a}" => "\u{3053}",
+        "\u{1b03b}" => "\u{3053}",
+        "\u{1b03c}" => "\u{3055}",
+        "\u{1b03d}" => "\u{3055}",
+        "\u{1b03e}" => "\u{3055}",
+        "\u{1b03f}" => "\u{3055}",
+        "\u{1b040}" => "\u{3055}",
+        "\u{1b041}" => "\u{3055}",
+        "\u{1b042}" => "\u{3055}",
+        "\u{1b043}" => "\u{3055}",
+        "\u{1b044}" => "\u{3057}",
+        "\u{1b045}" => "\u{3057}",
+        "\u{1b046}" => "\u{3057}",
+        "\u{1b047}" => "\u{3057}",
+        "\u{1b048}" => "\u{3057}",
+        "\u{1b049}" => "\u{3057}",
+        "\u{1b04a}" => "\u{3059}",
+        "\u{1b04b}" => "\u{3059}",
+        "\u{1b04c}" => "\u{3059}",
+        "\u{1b04d}" => "\u{3059}",
+        "\u{1b04e}" => "\u{3059}",
+        "\u{1b04f}" => "\u{3059}",
+        "\u{1b050}" => "\u{3059}",
+        "\u{1b051}" => "\u{3059}",
+        "\u{1b052}" => "\u{305b}",
+        "\u{1b053}" => "\u{305b}",
+        "\u{1b054}" => "\u{305b}",
+        "\u{1b055}" => "\u{305b}",
+        "\u{1b056}" => "\u{305b}",
+        "\u{1b057}" => "\u{305d}",
+        "\u{1b058}" => "\u{305d}",
+        "\u{1b059}" => "\u{305d}",
+        "\u{1b05a}" => "\u{305d}",
+        "\u{1b05b}" => "\u{305d}",
+        "\u{1b05c}" => "\u{305d}",
+        "\u{1b05d}" => "\u{305d}",
+        "\u{1b05e}" => "\u{305f}",
+        "\u{1b05f}" => "\u{305f}",
+        "\u{1b060}" => "\u{305f}",
+        "\u{1b061}" => "\u{305f}",
+        "\u{1b062}" => "\u{3061}",
+        "\u{1b063}" => "\u{3061}",
+        "\u{1b064}" => "\u{3061}",
+        "\u{1b065}" => "\u{3061}",
+        "\u{1b066}" => "\u{3061}",
+        "\u{1b067}" => "\u{3061}",
+        "\u{1b068}" => "\u{3061}",
+        "\u{1b069}" => "\u{3064}",
+        "\u{1b06a}" => "\u{3064}",
+        "\u{1b06b}" => "\u{3064}",
+        "\u{1b06c}" => "\u{3064}",
+        "\u{1b06d}" => "\u{3064}",
+        "\u{1b06e}" => "\u{3066}",
+        "\u{1b06f}" => "\u{3066}",
+        "\u{1b070}" => "\u{3066}",
+        "\u{1b071}" => "\u{3066}",
+        "\u{1b072}" => "\u{3066}",
+        "\u{1b073}" => "\u{3066}",
+        "\u{1b074}" => "\u{3066}",
+        "\u{1b075}" => "\u{3066}",
+        "\u{1b076}" => "\u{3066}",
+        "\u{1b077}" => "\u{3068}",
+        "\u{1b078}" => "\u{3068}",
+        "\u{1b079}" => "\u{3068}",
+        "\u{1b07a}" => "\u{3068}",
+        "\u{1b07b}" => "\u{3068}",
+        "\u{1b07c}" => "\u{3068}",
+        "\u{1b07d}" => "\u{3068}",
+        "\u{1b07e}" => "\u{306a}",
+        "\u{1b07f}" => "\u{306a}",
+        "\u{1b080}" => "\u{306a}",
+        "\u{1b081}" => "\u{306a}",
+        "\u{1b082}" => "\u{306a}",
+        "\u{1b083}" => "\u{306a}",
+        "\u{1b084}" => "\u{306a}",
+        "\u{1b085}" => "\u{306a}",
+        "\u{1b086}" => "\u{306a}",
+        "\u{1b087}" => "\u{306b}",
+        "\u{1b088}" => "\u{306b}",
+        "\u{1b089}" => "\u{306b}",
+        "\u{1b08a}" => "\u{306b}",
+        "\u{1b08b}" => "\u{306b}",
+        "\u{1b08c}" => "\u{306b}",
+        "\u{1b08d}" => "\u{306b}",
+        "\u{1b08e}" => "\u{306b}",
+        "\u{1b08f}" => "\u{306c}",
+        "\u{1b090}" => "\u{306c}",
+        "\u{1b091}" => "\u{306c}",
+        "\u{1b092}" => "\u{306d}",
+        "\u{1b093}" => "\u{306d}",
+        "\u{1b094}" => "\u{306d}",
+        "\u{1b095}" => "\u{306d}",
+        "\u{1b096}" => "\u{306d}",
+        "\u{1b097}" => "\u{306d}",
+        "\u{1b098}" => "\u{306d}",
+        "\u{1b099}" => "\u{306e}",
+        "\u{1b09a}" => "\u{306e}",
+        "\u{1b09b}" => "\u{306e}",
+        "\u{1b09c}" => "\u{306e}",
+        "\u{1b09d}" => "\u{306e}",
+        "\u{1b09e}" => "\u{306f}",
+        "\u{1b09f}" => "\u{306f}",
+        "\u{1b0a0}" => "\u{306f}",
+        "\u{1b0a1}" => "\u{306f}",
+        "\u{1b0a2}" => "\u{306f}",
+        "\u{1b0a3}" => "\u{306f}",
+        "\u{1b0a4}" => "\u{306f}",
+        "\u{1b0a5}" => "\u{306f}",
+        "\u{1b0a6}" => "\u{306f}",
+        "\u{1b0a7}" => "\u{306f}",
+        "\u{1b0a8}" => "\u{306f}",
+        "\u{1b0a9}" => "\u{3072}",
+        "\u{1b0aa}" => "\u{3072}",
+        "\u{1b0ab}" => "\u{3072}",
+        "\u{1b0ac}" => "\u{3072}",
+        "\u{1b0ad}" => "\u{3072}",
+        "\u{1b0ae}" => "\u{3072}",
+        "\u{1b0af}" => "\u{3072}",
+        "\u{1b0b0}" => "\u{3075}",
+        "\u{1b0b1}" => "\u{3075}",
+        "\u{1b0b2}" => "\u{3075}",
+        "\u{1b0b3}" => "\u{3078}",
+        "\u{1b0b4}" => "\u{3078}",
+        "\u{1b0b5}" => "\u{3078}",
+        "\u{1b0b6}" => "\u{3078}",
+        "\u{1b0b7}" => "\u{3078}",
+        "\u{1b0b8}" => "\u{3078}",
+        "\u{1b0b9}" => "\u{3078}",
+        "\u{1b0ba}" => "\u{307b}",
+        "\u{1b0bb}" => "\u{307b}",
+        "\u{1b0bc}" => "\u{307b}",
+        "\u{1b0bd}" => "\u{307b}",
+        "\u{1b0be}" => "\u{307b}",
+        "\u{1b0bf}" => "\u{307b}",
+        "\u{1b0c0}" => "\u{307b}",
+        "\u{1b0c1}" => "\u{307b}",
+        "\u{1b0c2}" => "\u{307e}",
+        "\u{1b0c3}" => "\u{307e}",
+        "\u{1b0c4}" => "\u{307e}",
+        "\u{1b0c5}" => "\u{307e}",
+        "\u{1b0c6}" => "\u{307e}",
+        "\u{1b0c7}" => "\u{307e}",
+        "\u{1b0c8}" => "\u{307e}",
+        "\u{1b0c9}" => "\u{307f}",
+        "\u{1b0ca}" => "\u{307f}",
+        "\u{1b0cb}" => "\u{307f}",
+        "\u{1b0cc}" => "\u{307f}",
+        "\u{1b0cd}" => "\u{307f}",
+        "\u{1b0ce}" => "\u{307f}",
+        "\u{1b0cf}" => "\u{307f}",
+        "\u{1b0d0}" => "\u{3080}",
+        "\u{1b0d1}" => "\u{3080}",
+        "\u{1b0d2}" => "\u{3080}",
+        "\u{1b0d3}" => "\u{3080}",
+        "\u{1b0d4}" => "\u{3081}",
+        "\u{1b0d5}" => "\u{3081}",
+        "\u{1b0d6}" => "\u{3081}",
+        "\u{1b0d7}" => "\u{3082}",
+        "\u{1b0d8}" => "\u{3082}",
+        "\u{1b0d9}" => "\u{3082}",
+        "\u{1b0da}" => "\u{3082}",
+        "\u{1b0db}" => "\u{3082}",
+        "\u{1b0dc}" => "\u{3082}",
+        "\u{1b0dd}" => "\u{3084}",
+        "\u{1b0de}" => "\u{3084}",
+        "\u{1b0df}" => "\u{3084}",
+        "\u{1b0e0}" => "\u{3084}",
+        "\u{1b0e1}" => "\u{3084}",
+        "\u{1b0e2}" => "\u{3084}",
+        "\u{1b0e3}" => "\u{3086}",
+        "\u{1b0e4}" => "\u{3086}",
+        "\u{1b0e5}" => "\u{3086}",
+        "\u{1b0e6}" => "\u{3086}",
+        "\u{1b0e7}" => "\u{3088}",
+        "\u{1b0e8}" => "\u{3088}",
+        "\u{1b0e9}" => "\u{3088}",
+        "\u{1b0ea}" => "\u{3088}",
+        "\u{1b0eb}" => "\u{3088}",
+        "\u{1b0ec}" => "\u{3088}",
+        "\u{1b0ed}" => "\u{3089}",
+        "\u{1b0ee}" => "\u{3089}",
+        "\u{1b0ef}" => "\u{3089}",
+        "\u{1b0f0}" => "\u{3089}",
+        "\u{1b0f1}" => "\u{308a}",
+        "\u{1b0f2}" => "\u{308a}",
+        "\u{1b0f3}" => "\u{308a}",
+        "\u{1b0f4}" => "\u{308a}",
+        "\u{1b0f5}" => "\u{308a}",
+        "\u{1b0f6}" => "\u{308a}",
+        "\u{1b0f7}" => "\u{308a}",
+        "\u{1b0f8}" => "\u{308b}",
+        "\u{1b0f9}" => "\u{308b}",
+        "\u{1b0fa}" => "\u{308b}",
+        "\u{1b0fb}" => "\u{308b}",
+        "\u{1b0fc}" => "\u{308b}",
+        "\u{1b0fd}" => "\u{308b}",
+        "\u{1b0fe}" => "\u{308c}",
+        "\u{1b0ff}" => "\u{308c}",
+        "\u{1b100}" => "\u{308c}",
+        "\u{1b101}" => "\u{308c}",
+        "\u{1b102}" => "\u{308d}",
+        "\u{1b103}" => "\u{308d}",
+        "\u{1b104}" => "\u{308d}",
+        "\u{1b105}" => "\u{308d}",
+        "\u{1b106}" => "\u{308d}",
+        "\u{1b107}" => "\u{308d}",
+        "\u{1b108}" => "\u{308f}",
+        "\u{1b109}" => "\u{308f}",
+        "\u{1b10a}" => "\u{308f}",
+        "\u{1b10b}" => "\u{308f}",
+        "\u{1b10c}" => "\u{308f}",
+        "\u{1b10d}" => "\u{3090}",
+        "\u{1b10e}" => "\u{3090}",
+        "\u{1b10f}" => "\u{3090}",
+        "\u{1b110}" => "\u{3090}",
+        "\u{1b111}" => "\u{3090}",
+        "\u{1b112}" => "\u{3091}",
+        "\u{1b113}" => "\u{3091}",
+        "\u{1b114}" => "\u{3091}",
+        "\u{1b115}" => "\u{3091}",
+        "\u{1b116}" => "\u{3092}",
+        "\u{1b117}" => "\u{3092}",
+        "\u{1b118}" => "\u{3092}",
+        "\u{1b119}" => "\u{3092}",
+        "\u{1b11a}" => "\u{3092}",
+        "\u{1b11b}" => "\u{3092}",
+        "\u{1b11c}" => "\u{3092}",
+        "\u{1b11d}" => "\u{3093}",
+        "\u{1b11e}" => "\u{3093}",
+        "\u{1b11f}" => "\u{3046}",
+        "\u{1b120}" => "\u{30a4}",
+        "\u{1b121}" => "\u{30a8}",
+        "\u{1b122}" => "\u{30a6}"
+      }.freeze
+
+      # Transliterator for archaic_hirakatas
+      class Transliterator < Yosina::BaseTransliterator
+        # Initialize the transliterator with options
+        #
+        # @param _options [Hash] Configuration options (currently unused)
+        def initialize(_options = {})
+          # Options currently unused for archaic_hirakatas transliterator
+          super()
+        end
+
+        # Replaces archaic kana (hentaigana) with their modern equivalents.
+        #
+        # @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 = ARCHAIC_HIRAKATAS_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 archaic_hirakatas transliterator
+      #
+      # @param options [Hash] Configuration options
+      # @return [Transliterator] A new archaic_hirakatas transliterator instance
+      def self.call(options = {})
+        Transliterator.new(options)
+      end
+    end
+  end
+end

data/lib/yosina/transliterators/hira_kata_table.rb

@@ -79,7 +79,11 @@ module Yosina
         ['ょ', 'ョ', 'ｮ'],
         ['ゎ', 'ヮ', nil],
         ['ゕ', 'ヵ', nil],
-        ['ゖ', 'ヶ', nil]
+        ['ゖ', 'ヶ', nil],
+        ["\u{1B132}", "\u{1B155}", nil],
+        ["\u{1B150}", "\u{1B164}", nil],
+        ["\u{1B151}", "\u{1B165}", nil],
+        ["\u{1B152}", "\u{1B166}", nil]
       ].freeze
 
       # Generate voiced character mappings

data/lib/yosina/transliterators/historical_hirakatas.rb

@@ -0,0 +1,153 @@
+# frozen_string_literal: true
+
+module Yosina
+  module Transliterators
+    # Convert historical hiragana/katakana characters to their modern equivalents
+    module HistoricalHirakatas
+      # Historical hiragana mappings: source => { simple:, decompose: }
+      HISTORICAL_HIRAGANA_MAPPINGS = {
+        "\u{3090}" => { simple: "\u{3044}", decompose: "\u{3046}\u{3043}" }, # ゐ → い / うぃ
+        "\u{3091}" => { simple: "\u{3048}", decompose: "\u{3046}\u{3047}" }  # ゑ → え / うぇ
+      }.freeze
+
+      # Historical katakana mappings: source => { simple:, decompose: }
+      HISTORICAL_KATAKANA_MAPPINGS = {
+        "\u{30F0}" => { simple: "\u{30A4}", decompose: "\u{30A6}\u{30A3}" }, # ヰ → イ / ウィ
+        "\u{30F1}" => { simple: "\u{30A8}", decompose: "\u{30A6}\u{30A7}" }  # ヱ → エ / ウェ
+      }.freeze
+
+      # Voiced historical katakana mappings: source => small vowel suffix
+      VOICED_HISTORICAL_KANA_MAPPINGS = {
+        "\u{30F7}" => "\u{30A1}", # ヷ → ァ
+        "\u{30F8}" => "\u{30A3}", # ヸ → ィ
+        "\u{30F9}" => "\u{30A7}", # ヹ → ェ
+        "\u{30FA}" => "\u{30A9}"  # ヺ → ォ
+      }.freeze
+
+      VOICED_HISTORICAL_KANA_DECOMPOSED_MAPPINGS = {
+        "\u{30EF}" => "\u{30A1}", # ヷ → ァ
+        "\u{30F0}" => "\u{30A3}", # ヸ → ィ
+        "\u{30F1}" => "\u{30A7}", # ヹ → ェ
+        "\u{30F2}" => "\u{30A9}"  # ヺ → ォ
+      }.freeze
+
+      COMBINING_DAKUTEN = "\u{3099}"
+      VU = "\u{30F4}"
+      U = "\u{30A6}"
+
+      # Transliterator for historical hiragana/katakana conversion
+      class Transliterator < Yosina::BaseTransliterator
+        # Initialize the transliterator with options
+        #
+        # @param options [Hash] Configuration options
+        # @option options [String] :hiraganas "simple" (default), "decompose", or "skip"
+        # @option options [String] :katakanas "simple" (default), "decompose", or "skip"
+        # @option options [String] :voiced_katakanas "decompose" or "skip" (default)
+        def initialize(options = {})
+          super()
+          @hiraganas = (options[:hiraganas] || :simple).to_sym
+          @katakanas = (options[:katakanas] || :simple).to_sym
+          @voiced_katakanas = (options[:voiced_katakanas] || :skip).to_sym
+        end
+
+        # Convert historical hiragana/katakana characters to modern equivalents
+        #
+        # @param input_chars [Enumerable<Char>] The characters to transliterate
+        # @return [Enumerable<Char>] The transliterated characters
+        def call(input_chars)
+          Chars.enum do |y|
+            offset = 0
+            pending = nil
+            input_chars.each do |char|
+              if char.sentinel?
+                offset = emit_char(y, pending, offset) if pending
+                pending = nil
+                y << char
+                break
+              end
+
+              if pending.nil?
+                pending = char
+                next
+              end
+
+              if char.c == COMBINING_DAKUTEN
+                # Check if pending char could be a decomposed voiced base
+                decomposed = VOICED_HISTORICAL_KANA_DECOMPOSED_MAPPINGS[pending.c]
+                if @voiced_katakanas == :skip || decomposed.nil?
+                  y << pending.with_offset(offset)
+                  offset += pending.c.length
+                  pending = char
+                  next
+                end
+                y << Char.new(c: U, offset: offset, source: pending)
+                offset += U.length
+                y << char.with_offset(offset)
+                offset += char.c.length
+                y << Char.new(c: decomposed, offset: offset, source: pending)
+                offset += decomposed.length
+                pending = nil
+                next
+              end
+
+              offset = emit_char(y, pending, offset)
+              pending = char
+            end
+            # Flush any remaining pending char
+            emit_char(y, pending, offset) if pending
+          end
+        end
+
+        private
+
+        # Emit a single char through the normal mapping logic
+        #
+        # @param y [Enumerator::Yielder] The yielder
+        # @param char [Char] The character to emit
+        # @param offset [Integer] The current offset
+        # @return [Integer] The new offset after emitting
+        # rubocop:disable Naming/MethodParameterName
+        def emit_char(y, char, offset)
+          # Historical hiragana
+          hira_mapping = HISTORICAL_HIRAGANA_MAPPINGS[char.c]
+          if hira_mapping && @hiraganas != :skip
+            replacement = hira_mapping[@hiraganas]
+            y << Char.new(c: replacement, offset: offset, source: char)
+            return offset + replacement.length
+          end
+
+          # Historical katakana
+          kata_mapping = HISTORICAL_KATAKANA_MAPPINGS[char.c]
+          if kata_mapping && @katakanas != :skip
+            replacement = kata_mapping[@katakanas]
+            y << Char.new(c: replacement, offset: offset, source: char)
+            return offset + replacement.length
+          end
+
+          # Voiced historical katakana
+          if @voiced_katakanas == :decompose
+            decomposed = VOICED_HISTORICAL_KANA_MAPPINGS[char.c]
+            if decomposed
+              y << Char.new(c: VU, offset: offset, source: char)
+              offset += VU.length
+              y << Char.new(c: decomposed, offset: offset, source: char)
+              return offset + decomposed.length
+            end
+          end
+
+          y << char.with_offset(offset)
+          offset + char.c.length
+        end
+      end
+      # rubocop:enable Naming/MethodParameterName
+
+      # Factory method to create a historical hirakatas transliterator
+      #
+      # @param options [Hash] Configuration options
+      # @return [Transliterator] A new historical hirakatas transliterator instance
+      def self.call(options = {})
+        Transliterator.new(options)
+      end
+    end
+  end
+end

data/lib/yosina/transliterators/small_hirakatas.rb

@@ -0,0 +1,117 @@
+# frozen_string_literal: true
+
+module Yosina
+  module Transliterators
+    # Replaces small hiragana/katakana with their ordinary-sized equivalents.
+    module SmallHirakatas
+      # Generated mapping data from small_hirakatas.json
+      SMALL_HIRAKATAS_MAPPINGS = {
+        "\u{3041}" => "\u{3042}",
+        "\u{3043}" => "\u{3044}",
+        "\u{3045}" => "\u{3046}",
+        "\u{3047}" => "\u{3048}",
+        "\u{3049}" => "\u{304a}",
+        "\u{3063}" => "\u{3064}",
+        "\u{3083}" => "\u{3084}",
+        "\u{3085}" => "\u{3086}",
+        "\u{3087}" => "\u{3088}",
+        "\u{308e}" => "\u{308f}",
+        "\u{3095}" => "\u{304b}",
+        "\u{3096}" => "\u{3051}",
+        "\u{30a1}" => "\u{30a2}",
+        "\u{30a3}" => "\u{30a4}",
+        "\u{30a5}" => "\u{30a6}",
+        "\u{30a7}" => "\u{30a8}",
+        "\u{30a9}" => "\u{30aa}",
+        "\u{30c3}" => "\u{30c4}",
+        "\u{30e3}" => "\u{30e4}",
+        "\u{30e5}" => "\u{30e6}",
+        "\u{30e7}" => "\u{30e8}",
+        "\u{30ee}" => "\u{30ef}",
+        "\u{30f5}" => "\u{30ab}",
+        "\u{30f6}" => "\u{30b1}",
+        "\u{31f0}" => "\u{30af}",
+        "\u{31f1}" => "\u{30b7}",
+        "\u{31f2}" => "\u{30b9}",
+        "\u{31f3}" => "\u{30c8}",
+        "\u{31f4}" => "\u{30cc}",
+        "\u{31f5}" => "\u{30cf}",
+        "\u{31f6}" => "\u{30d2}",
+        "\u{31f7}" => "\u{30d5}",
+        "\u{31f8}" => "\u{30d8}",
+        "\u{31f9}" => "\u{30db}",
+        "\u{31fa}" => "\u{30e0}",
+        "\u{31fb}" => "\u{30e9}",
+        "\u{31fc}" => "\u{30ea}",
+        "\u{31fd}" => "\u{30eb}",
+        "\u{31fe}" => "\u{30ec}",
+        "\u{31ff}" => "\u{30ed}",
+        "\u{ff67}" => "\u{ff71}",
+        "\u{ff68}" => "\u{ff72}",
+        "\u{ff69}" => "\u{ff73}",
+        "\u{ff6a}" => "\u{ff74}",
+        "\u{ff6b}" => "\u{ff75}",
+        "\u{ff6c}" => "\u{ff94}",
+        "\u{ff6d}" => "\u{ff95}",
+        "\u{ff6e}" => "\u{ff96}",
+        "\u{ff6f}" => "\u{ff82}",
+        "\u{1b132}" => "\u{3053}",
+        "\u{1b150}" => "\u{3090}",
+        "\u{1b151}" => "\u{3091}",
+        "\u{1b152}" => "\u{3092}",
+        "\u{1b155}" => "\u{30b3}",
+        "\u{1b164}" => "\u{30f0}",
+        "\u{1b165}" => "\u{30f1}",
+        "\u{1b166}" => "\u{30f2}",
+        "\u{1b167}" => "\u{30f3}"
+      }.freeze
+
+      # Transliterator for small_hirakatas
+      class Transliterator < Yosina::BaseTransliterator
+        # Initialize the transliterator with options
+        #
+        # @param _options [Hash] Configuration options (currently unused)
+        def initialize(_options = {})
+          # Options currently unused for small_hirakatas transliterator
+          super()
+        end
+
+        # Replaces small hiragana/katakana with their ordinary-sized equivalents.
+        #
+        # @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 = SMALL_HIRAKATAS_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 small_hirakatas transliterator
+      #
+      # @param options [Hash] Configuration options
+      # @return [Transliterator] A new small_hirakatas transliterator instance
+      def self.call(options = {})
+        Transliterator.new(options)
+      end
+    end
+  end
+end

data/lib/yosina/transliterators.rb

@@ -15,6 +15,9 @@ require_relative 'transliterators/jisx0201_and_alike'
 require_relative 'transliterators/circled_or_squared'
 require_relative 'transliterators/combined'
 require_relative 'transliterators/japanese_iteration_marks'
+require_relative 'transliterators/archaic_hirakatas'
+require_relative 'transliterators/small_hirakatas'
+require_relative 'transliterators/historical_hirakatas'
 
 module Yosina
   # Registry for transliterator factories
@@ -34,7 +37,10 @@ module Yosina
       jisx0201_and_alike: Transliterators::Jisx0201AndAlike,
       combined: Transliterators::Combined,
       circled_or_squared: CircledOrSquared,
-      japanese_iteration_marks: Transliterators::JapaneseIterationMarks
+      japanese_iteration_marks: Transliterators::JapaneseIterationMarks,
+      archaic_hirakatas: Transliterators::ArchaicHirakatas,
+      small_hirakatas: Transliterators::SmallHirakatas,
+      historical_hirakatas: Transliterators::HistoricalHirakatas
     }.freeze
 
     # Get a transliterator factory by name

data/lib/yosina/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Yosina
-  VERSION = '0.2.0'
+  VERSION = '1.1.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: yosina
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 1.1.0
 platform: ruby
 authors:
 - Moriyoshi Koizumi
-autorequire:
+autorequire: 
 bindir: exe
 cert_chain: []
-date: 2025-09-23 00:00:00.000000000 Z
+date: 2026-03-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: minitest
@@ -109,6 +109,7 @@ files:
 - lib/yosina/recipes.rb
 - lib/yosina/transliterator.rb
 - lib/yosina/transliterators.rb
+- lib/yosina/transliterators/archaic_hirakatas.rb
 - lib/yosina/transliterators/circled_or_squared.rb
 - lib/yosina/transliterators/circled_or_squared_data.rb
 - lib/yosina/transliterators/combined.rb
@@ -116,6 +117,7 @@ files:
 - lib/yosina/transliterators/hira_kata.rb
 - lib/yosina/transliterators/hira_kata_composition.rb
 - lib/yosina/transliterators/hira_kata_table.rb
+- lib/yosina/transliterators/historical_hirakatas.rb
 - lib/yosina/transliterators/hyphens.rb
 - lib/yosina/transliterators/hyphens_data.rb
 - lib/yosina/transliterators/ideographic_annotations.rb
@@ -129,6 +131,7 @@ files:
 - lib/yosina/transliterators/radicals.rb
 - lib/yosina/transliterators/roman_numerals.rb
 - lib/yosina/transliterators/roman_numerals_data.rb
+- lib/yosina/transliterators/small_hirakatas.rb
 - lib/yosina/transliterators/spaces.rb
 - lib/yosina/version.rb
 - yosina.gemspec
@@ -140,7 +143,7 @@ metadata:
   homepage_uri: https://github.com/yosina-lib/yosina
   source_code_uri: https://github.com/yosina-lib/yosina
   changelog_uri: https://github.com/yosina-lib/yosina/releases
-post_install_message:
+post_install_message: 
 rdoc_options: []
 require_paths:
 - lib
@@ -155,8 +158,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.11
-signing_key:
+rubygems_version: 3.0.3.1
+signing_key: 
 specification_version: 4
 summary: Japanese text transliteration library
 test_files: []