Class: Yosina::Transliterators::JapaneseIterationMarks::Transliterator

Inherits:

BaseTransliterator

• Object
• BaseTransliterator
• Yosina::Transliterators::JapaneseIterationMarks::Transliterator

Includes:

CharType

Defined in:

lib/yosina/transliterators/japanese_iteration_marks.rb

Overview

Transliterator for Japanese iteration marks

Instance Method Summary

• #call(input_chars) ⇒ Enumerable<Char>

  Replace iteration marks with appropriate repeated characters.

• #initialize(_options = {}) ⇒ Transliterator constructor

  Initialize the transliterator with options.

Methods included from CharType

#hatsuon?, #hiragana?, #iteration_mark?, #kanji?, #katakana?, #semi_voiced?, #sokuon?, #voiced?

Constructor Details

#initialize(_options = {}) ⇒ Transliterator

Initialize the transliterator with options

Parameters:

• options (Hash) —

  Configuration options (currently unused but kept for consistency)

# File 'lib/yosina/transliterators/japanese_iteration_marks.rb', line 120

def initialize(_options = {})
  super()
end

Instance Method Details

#call(input_chars) ⇒ Enumerable<Char>

Replace iteration marks with appropriate repeated characters

Parameters:

• input_chars (Enumerable<Char>) —

  The characters to transliterate

Returns:

• (Enumerable<Char>) —

  The transliterated characters

# File 'lib/yosina/transliterators/japanese_iteration_marks.rb', line 128

def call(input_chars)
  offset = 0
  prev_char_info = nil
  prev_was_iteration_mark = false

  Chars.enum do |y|
    input_chars.each do |char|
      # Skip empty/sentinel characters
      if char.c.empty?
        y << char
        next
      end

      current_char = char.c

      if iteration_mark?(current_char)
        # Check if previous character was also an iteration mark
        if prev_was_iteration_mark
          # Don't replace consecutive iteration marks
          y << char.with_offset(offset)
          offset += char.c.length
          prev_was_iteration_mark = true
          next
        end

        # We have an iteration mark, check if we can replace it
        replacement = nil
        if prev_char_info
          case current_char
          when HIRAGANA_ITERATION_MARK, VERTICAL_HIRAGANA_ITERATION_MARK
            # Repeat previous hiragana if valid
            if prev_char_info[:type] == :hiragana
              replacement = prev_char_info[:char]
            elsif prev_char_info[:type] == :hiragana_voiced
              # Voiced character with unvoiced mark: unvoice it
              replacement = HIRAGANA_UNVOICING[prev_char_info[:char]]
            end
          when HIRAGANA_VOICED_ITERATION_MARK, VERTICAL_HIRAGANA_VOICED_ITERATION_MARK
            # Repeat previous hiragana with voicing if possible
            if prev_char_info[:type] == :hiragana
              replacement = HIRAGANA_VOICING[prev_char_info[:char]]
            elsif prev_char_info[:type] == :hiragana_voiced
              # Voiced character with voiced mark: keep it voiced
              replacement = prev_char_info[:char]
            end
          when KATAKANA_ITERATION_MARK, VERTICAL_KATAKANA_ITERATION_MARK
            # Repeat previous katakana if valid
            if prev_char_info[:type] == :katakana
              replacement = prev_char_info[:char]
            elsif prev_char_info[:type] == :katakana_voiced
              # Voiced character with unvoiced mark: unvoice it
              replacement = KATAKANA_UNVOICING[prev_char_info[:char]]
            end
          when KATAKANA_VOICED_ITERATION_MARK, VERTICAL_KATAKANA_VOICED_ITERATION_MARK
            # Repeat previous katakana with voicing if possible
            if prev_char_info[:type] == :katakana
              replacement = KATAKANA_VOICING[prev_char_info[:char]]
            elsif prev_char_info[:type] == :katakana_voiced
              # Voiced character with voiced mark: keep it voiced
              replacement = prev_char_info[:char]
            end
          when KANJI_ITERATION_MARK
            # Repeat previous kanji
            replacement = prev_char_info[:char] if prev_char_info[:type] == :kanji
          end
        end

        if replacement
          # Create a new character with the replacement
          y << Char.new(c: replacement, offset: offset, source: char)
          offset += replacement.length
          # Don't update prev_char_info - keep the original one
          # This ensures consecutive iteration marks work correctly
        else
          # Couldn't replace the iteration mark
          y << char.with_offset(offset)
          offset += char.c.length
        end
        prev_was_iteration_mark = true
        next
      else
        # Not an iteration mark
        y << char.with_offset(offset)
        offset += char.c.length

        # Update previous character info
        char_code = current_char.ord
        char_type = get_char_type(current_char, char_code)

        # Only update prev_char_info if it's a repeatable character
        prev_char_info = ({ char: current_char, type: char_type } if char_type && char_type != :other)

        prev_was_iteration_mark = false
      end
    end
  end
end