class Enparser::Cli

Public Instance Methods

lemmatize(file_name, skip_sources = '') click to toggle source
# File lib/enparser/cli.rb, line 21
def lemmatize(file_name, skip_sources = '')
  lemmatizer = Enparser::Extractor.new
  lemmatizer.load_skip_deafult if options['load-default']
  lemmatizer.load_skip_file(skip_sources) unless skip_sources.empty?
  lemmatizer.parse_files(file_name)
  sep = options[:separator]
  lemmatizer.each do |word, count, forms|
    line = String.new(word)
    unless options['strict-by-lemmas']
      line << "#{sep}#{count}"
      line << "#{sep}(#{forms})" unless forms.empty?
    end
    puts line
  end
end
segment(file_name, skip_patterns = '') click to toggle source
# File lib/enparser/cli.rb, line 10
def segment(file_name, skip_patterns = '')
  segmenter = Enparser::Segmenter.new
  segmenter.load_skip_patters(skip_patterns) unless skip_patterns.empty?
  segmenter.segment_file(file_name) { |s| puts s }
end