Module: Kabosu

Defined in:

lib/kabosu.rb,
lib/kabosu/railtie.rb,
lib/kabosu/version.rb,
lib/kabosu/pos_matcher.rb,
lib/kabosu/dict_manager.rb,
lib/kabosu/morpheme_list.rb

Defined Under Namespace

Classes: ConfigError, DictManager, Dictionary, DictionaryError, Error, LookupError, Morpheme, MorphemeList, PosMatcher, Railtie, SentenceRange, SentenceSplitError, TokenizationError, Tokenizer

Constant Summary

MODE_A =

:a

MODE_B =

:b

MODE_C =

:c

VERSION =

"0.7.0"

Class Method Summary

• .split_sentences(text, limit: nil, with_checker: false, ranges: false, dictionary: nil) ⇒ Object
• .tokenize(text, tokenizer:) ⇒ Object

  Tokenize text using an explicitly provided tokenizer.

Class Method Details

.split_sentences(text, limit: nil, with_checker: false, ranges: false, dictionary: nil) ⇒ Object

# File 'lib/kabosu.rb', line 202

def self.split_sentences(text, limit: nil, with_checker: false, ranges: false, dictionary: nil)
  unless text.is_a?(String)
    raise ArgumentError, "text must be a String"
  end
  unless limit.nil? || limit.is_a?(Integer)
    raise ArgumentError, "limit must be an Integer or nil"
  end
  if limit && limit < 1
    raise ArgumentError, "limit must be greater than 0"
  end
  unless with_checker == true || with_checker == false
    raise ArgumentError, "with_checker must be true or false"
  end
  unless ranges == true || ranges == false
    raise ArgumentError, "ranges must be true or false"
  end
  unless dictionary.nil? || dictionary.is_a?(String)
    raise ArgumentError, "dictionary must be a String path or nil"
  end

  dict_path = nil
  if with_checker
    dict_path = dictionary || Dictionary.path
  end

  if ranges
    _split_sentences_with_ranges(text, limit, dict_path).map do |(start, finish, sentence)|
      SentenceRange.new(start, finish, sentence)
    end
  else
    _split_sentences(text, limit, dict_path)
  end
rescue RuntimeError => e
  raise SentenceSplitError.new(e.message), cause: e
end

.tokenize(text, tokenizer:) ⇒ Object

Tokenize text using an explicitly provided tokenizer.

dict = Kabosu::Dictionary.new(system_dict: Kabosu::Dictionary.path)
tok = dict.create(mode: :a)
Kabosu.tokenize("東京都に住んでいる", tokenizer: tok)

# File 'lib/kabosu.rb', line 246

def self.tokenize(text, tokenizer:)
  unless text.is_a?(String)
    raise ArgumentError, "text must be a String"
  end
  unless tokenizer.is_a?(Tokenizer)
    raise ArgumentError, "tokenizer must be a Kabosu::Tokenizer"
  end

  batch = tokenizer.__send__(:_tokenize, text)
  cost = batch.respond_to?(:internal_cost) ? batch.internal_cost : nil
  MorphemeList.new(batch, internal_cost: cost)
rescue RuntimeError => e
  raise TokenizationError.new(e.message), cause: e
end