translate.rb

#!/usr/bin/env ruby
require 'optparse'
require 'json'

ONLY_TWO_ALPHAS = /^[[:alpha:]]{2}$/
INVALID_LANGULAGE_CODE_ERROR_TEXT = 'Error: Language code should be ISO 639 alpha-2'


def parse_args(argv)
  parser = OptionParser.new
  parser.on('-s', '--source SOURCE_LANGUAGE', ONLY_TWO_ALPHAS)
  parser.on('-t', '--target TARGET_LANGUAGE', ONLY_TWO_ALPHAS)
  parser.on('-p', '--phrase PHRASE', '-w', '--word')

  opts = parser.getopts(argv)
  rest = parser.parse(argv)
  text = [opts['phrase'], rest].compact.join(' ')

  [opts['source'] || '', opts['target'] || '', text.strip]
end

# en or ja
def detect_language(text)
  case text
  when /\p{Han}|\p{Hiragana}|\p{Katakana}/
    'ja'
  else
    'en'
  end
end

def determine_source_language(target_lang, text)
  detected_lang = detect_language(text)
  return detected_lang if target_lang.empty? || target_lang != detected_lang

  case target_lang
  when 'ja'
    'en'
  else
    'ja'
  end
end

def determine_target_language(source_lang)
  case source_lang
  when 'ja'
    'en'
  else
    'ja'
  end
end


def load_translation_page(source_lang, target_lang, source_text)
  `curl -s --user-agent "" \
    -d "sl=#{source_lang}" \
    -d "tl=#{target_lang}" \
    -d "ie=utf_8" \
    -d "oe=utf_8" \
    -d "client=t" \
    --data-urlencode "text=#{source_text.strip}" \
    http://translate.google.com.br/translate_a/t`
end


def extract_translated_text(json)
  target = json.match(/\[(\[\[[\s\S]*?\]\])/)[1]
  array = JSON.parse(target);
  a = array.map {|el| el[0] }
  a.join('')
end



source_lang, target_lang, source_text = parse_args(ARGV)

source_text = STDIN.read if source_text.strip.empty?
source_lang = determine_source_language(target_lang, source_text) if source_lang.empty?
target_lang = determine_target_language(source_lang) if target_lang.empty?

response = load_translation_page(source_lang, target_lang, source_text)
# response = File.read('/tmp/log-translate')
# File.write('/tmp/log-translate', response)

puts extract_translated_text(response)