4 # :title: Translator plugin for rbot
6 # Author:: Yaohan Chen <yaohan.chen@gmail.com>
7 # Copyright:: (C) 2007 Yaohan Chen
10 # This plugin allows using rbot to translate text on a few translation services
14 # * Configuration for whether to show translation engine
15 # * Optionally sync default translators with karma.rb ranking
20 # base class for implementing a translation service
22 # direction:: supported translation directions, a hash where each key is a source
23 # language name, and each value is Set of target language names. The
24 # methods in the Direction module are convenient for initializing this
27 INFO = 'Some translation service'
29 class UnsupportedDirectionError < ArgumentError
32 class NoTranslationError < RuntimeError
35 attr_reader :directions, :cache
37 def initialize(directions, cache={})
38 @directions = directions
43 # whether the translator supports this direction
44 def support?(from, to)
45 from != to && @directions[from].include?(to)
48 # this implements argument checking and caching. subclasses should define the
49 # do_translate method to implement actual translation
50 def translate(text, from, to)
51 raise UnsupportedDirectionError unless support?(from, to)
52 raise ArgumentError, _("Cannot translate empty string") if text.empty?
53 request = [text, from, to]
54 unless @cache.has_key? request
55 translation = do_translate(text, from, to)
56 raise NoTranslationError if translation.empty?
57 @cache[request] = translation
64 # given the set of supported languages, return a hash suitable for the directions
65 # attribute which includes any language to any other language
66 def self.all_to_all(languages)
67 directions = all_to_none(languages)
68 languages.each {|l| directions[l] = languages.to_set}
72 # a hash suitable for the directions attribute which includes any language from/to
73 # the given set of languages (center_languages)
74 def self.all_from_to(languages, center_languages)
75 directions = all_to_none(languages)
76 center_languages.each {|l| directions[l] = languages - [l]}
77 (languages - center_languages).each {|l| directions[l] = center_languages.to_set}
81 # get a hash from a list of pairs
82 def self.pairs(list_of_pairs)
83 languages = list_of_pairs.flatten.to_set
84 directions = all_to_none(languages)
85 list_of_pairs.each do |(from, to)|
86 directions[from] << to
91 # an empty hash with empty sets as default values
92 def self.all_to_none(languages)
94 # always return empty set when the key is non-existent, but put empty set in the
95 # hash only if the key is one of the languages
96 if languages.include? k
107 class NiftyTranslator < Translator
108 INFO = '@nifty Translation <http://nifty.amikai.com/amitext/indexUTF8.jsp>'
110 def initialize(cache={})
112 super(Translator::Direction.all_from_to(%w[ja en zh_CN ko], %w[ja]), cache)
113 @form = WWW::Mechanize.new.
114 get('http://nifty.amikai.com/amitext/indexUTF8.jsp').
115 forms.name('translateForm').first
118 def do_translate(text, from, to)
119 @form.radiobuttons.name('langpair').value = "#{from},#{to}".upcase
120 @form.fields.name('sourceText').value = text
122 @form.submit(@form.buttons.name('translate')).
123 forms.name('translateForm').fields.name('translatedText').value
128 class ExciteTranslator < Translator
129 INFO = 'Excite.jp Translation <http://www.excite.co.jp/world/>'
131 def initialize(cache={})
135 super(Translator::Direction.all_from_to(%w[ja en zh_CN zh_TW ko], %w[ja]), cache)
137 @forms = Hash.new do |h, k|
140 h[k] = open_form('english')
141 when 'zh_CN', 'zh_TW'
142 # this way we don't need to fetch the same page twice
143 h['zh_CN'] = h['zh_TW'] = open_form('chinese')
145 h[k] = open_form('korean')
151 WWW::Mechanize.new.get("http://www.excite.co.jp/world/#{name}").
152 forms.name('world').first
155 def do_translate(text, from, to)
156 non_ja_language = from != 'ja' ? from : to
157 form = @forms[non_ja_language]
159 if non_ja_language =~ /zh_(CN|TW)/
160 form.fields.name('wb_lp').value = "#{from}#{to}".sub(/_(?:CN|TW)/, '').upcase
161 form.fields.name('big5').value = ($1 == 'TW' ? 'yes' : 'no')
163 # the en<->ja page is in Shift_JIS while other pages are UTF-8
164 text = Iconv.iconv('Shift_JIS', 'UTF-8', text) if non_ja_language == 'en'
165 form.fields.name('wb_lp').value = "#{from}#{to}".upcase
167 form.fields.name('before').value = text
168 result = form.submit.forms.name('world').fields.name('after').value
169 # the en<->ja page is in Shift_JIS while other pages are UTF-8
170 if non_ja_language == 'en'
171 Iconv.iconv('UTF-8', 'Shift_JIS', result)
180 class GoogleTranslator < Translator
181 INFO = 'Google Translate <http://www.google.com/translate_t>'
183 def initialize(cache={})
186 language_pairs = @lang_list.options.map do |o|
187 # these options have values like "en|zh-CN"; map to things like ['en', 'zh_CN'].
188 o.value.split('|').map {|l| l.sub('-', '_')}
190 super(Translator::Direction.pairs(language_pairs), cache)
194 agent = WWW::Mechanize.new
195 # without faking the user agent, Google Translate will serve non-UTF-8 text
196 agent.user_agent_alias = 'Linux Konqueror'
197 @form = agent.get('http://www.google.com/translate_t').
198 forms.action('/translate_t').first
199 @lang_list = @form.fields.name('langpair')
202 def do_translate(text, from, to)
205 @lang_list.value = "#{from}|#{to}".sub('_', '-')
206 @form.fields.name('text').value = text
207 @form.submit.parser.search('div#result_box').inner_html
212 class BabelfishTranslator < Translator
213 INFO = 'AltaVista Babel Fish Translation <http://babelfish.altavista.com/babelfish/>'
215 def initialize(cache)
218 @form = WWW::Mechanize.new.get('http://babelfish.altavista.com/babelfish/').
219 forms.name('frmTrText').first
220 @lang_list = @form.fields.name('lp')
221 language_pairs = @lang_list.options.map {|o| o.value.split('_')}.
222 reject {|p| p.empty?}
223 super(Translator::Direction.pairs(language_pairs), cache)
226 def do_translate(text, from, to)
227 if @form.fields.name('trtext').empty?
228 @form.add_field!('trtext', text)
230 @form.fields.name('trtext').value = text
232 @lang_list.value = "#{from}_#{to}"
233 @form.submit.parser.search("td.s/div[@style]").inner_html
237 class WorldlingoTranslator < Translator
238 INFO = 'WorldLingo Free Online Translator <http://www.worldlingo.com/en/products_services/worldlingo_translator.html>'
240 LANGUAGES = %w[en fr de it pt es ru nl el sv ar ja ko zh_CN zh_TW]
241 def initialize(cache)
243 super(Translator::Direction.all_to_all(LANGUAGES), cache)
246 def translate(text, from, to)
247 response = Irc::Utils.bot.httputil.get_response(URI.escape(
248 "http://www.worldlingo.com/SEfpX0LV2xIxsIIELJ,2E5nOlz5RArCY,/texttranslate?wl_srcenc=utf-8&wl_trgenc=utf-8&wl_text=#{text}&wl_srclang=#{from.upcase}&wl_trglang=#{to.upcase}"))
249 # WorldLingo seems to respond an XML when error occurs
250 case response['Content-Type']
254 raise Translator::NoTranslationError
259 class TranslatorPlugin < Plugin
260 Config.register Config::IntegerValue.new('translator.timeout',
261 :default => 30, :validate => Proc.new{|v| v > 0},
262 :desc => _("Number of seconds to wait for the translation service before timeout"))
265 'nifty' => NiftyTranslator,
266 'excite' => ExciteTranslator,
267 'google_translate' => GoogleTranslator,
268 'babelfish' => BabelfishTranslator,
269 'worldlingo' => WorldlingoTranslator,
276 TRANSLATORS.each_pair do |name, c|
278 @translators[name] = c.new(@registry.sub_registry(name))
279 map "#{name} :from :to *phrase",
280 :action => :cmd_translate, :thread => true
282 warning _("Translator %{name} cannot be used: %{reason}") %
283 {:name => name, :reason => $!}
287 Config.register Config::ArrayValue.new('translator.default_list',
288 :default => TRANSLATORS.keys,
289 :validate => Proc.new {|l| l.all? {|t| TRANSLATORS.has_key?(t)}},
290 :desc => _("List of translators to try in order when translator name not specified"),
291 :on_change => Proc.new {|bot, v| update_default})
295 def help(plugin, topic=nil)
296 if @translators.has_key?(plugin)
297 translator = @translators[plugin]
298 _('%{translator} <from> <to> <phrase> => Look up phrase using %{info}, supported from -> to languages: %{directions}') % {
299 :translator => plugin,
300 :info => translator.class::INFO,
301 :directions => translator.directions.map do |source, targets|
302 _('%{source} -> %{targets}') %
303 {:source => source, :targets => targets.to_a.join(', ')}
307 _('Command: <translator> <from> <to> <phrase>, where <translator> is one of: %{translators}. If "translator" is used in place of the translator name, the first translator in translator.default_list which supports the specified direction will be picked automatically. Use "help <translator>" to look up supported from and to languages') %
308 {:translators => @translators.keys.join(', ')}
313 @default_translators = bot.config['translator.default_list'] & @translators.keys
316 def cmd_translator(m, params)
317 from, to = params[:from], params[:to]
318 translator = @default_translators.find {|t| @translators[t].support?(from, to)}
320 cmd_translate m, params.merge({:translator => translator, :show_provider => true})
322 m.reply _('None of the default translators (translator.default_list) supports translating from %{source} to %{target}') % {:source => from, :target => to}
326 def cmd_translate(m, params)
327 # get the first word of the command
328 tname = params[:translator] || m.message[/\A(\w+)\s/, 1]
329 translator = @translators[tname]
330 from, to, phrase = params[:from], params[:to], params[:phrase].to_s
333 translation = Timeout.timeout(@bot.config['translator.timeout']) do
334 translator.translate(phrase, from, to)
336 m.reply(if params[:show_provider]
337 _('%{translation} (provided by %{translator})') %
338 {:translation => translation, :translator => tname}
343 rescue Translator::UnsupportedDirectionError
344 m.reply _("%{translator} doesn't support translating from %{source} to %{target}") %
345 {:translator => tname, :source => from, :target => to}
346 rescue Translator::NoTranslationError
347 m.reply _('%{translator} failed to provide a translation') %
348 {:translator => tname}
349 rescue Timeout::Error
350 m.reply _('The translator timed out')
353 m.reply _('No translator called %{name}') % {:name => tname}
358 plugin = TranslatorPlugin.new
359 plugin.map 'translator :from :to *phrase',
360 :action => :cmd_translator, :thread => true