X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Ftranslator.rb;h=0e09cb4855e717dd81473ead737443331570accd;hb=16336b4a240a4265d1f2df1e30d7b68d3a924287;hp=76392ccc85c5e32ebaa2c606e0100e5ecf5e78ff;hpb=b640b14d732d457fec50b89738206a911ec9de7a;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git diff --git a/data/rbot/plugins/translator.rb b/data/rbot/plugins/translator.rb index 76392ccc..0e09cb48 100644 --- a/data/rbot/plugins/translator.rb +++ b/data/rbot/plugins/translator.rb @@ -8,6 +8,11 @@ # License:: GPLv2 # # This plugin allows using rbot to translate text on a few translation services +# +# TODO +# +# * Configuration for whether to show translation engine +# * Optionally sync default translators with karma.rb ranking require 'set' require 'timeout' @@ -19,10 +24,14 @@ require 'timeout' # methods in the Direction module are convenient for initializing this # attribute class Translator + INFO = 'Some translation service' class UnsupportedDirectionError < ArgumentError end + class NoTranslationError < RuntimeError + end + attr_reader :directions, :cache def initialize(directions, cache={}) @@ -30,24 +39,32 @@ class Translator @cache = cache end - + # whether the translator supports this direction def support?(from, to) from != to && @directions[from].include?(to) end - # this implements checking of languages and caching. subclasses should define the + # this implements argument checking and caching. subclasses should define the # do_translate method to implement actual translation def translate(text, from, to) raise UnsupportedDirectionError unless support?(from, to) - @cache[[text, from, to]] ||= do_translate(text, from, to) + raise ArgumentError, _("Cannot translate empty string") if text.empty? + request = [text, from, to] + unless @cache.has_key? request + translation = do_translate(text, from, to) + raise NoTranslationError if translation.empty? + @cache[request] = translation + else + @cache[request] + end end module Direction # given the set of supported languages, return a hash suitable for the directions # attribute which includes any language to any other language def self.all_to_all(languages) - directions = all_to_all(languages) + directions = all_to_none(languages) languages.each {|l| directions[l] = languages.to_set} directions end @@ -88,25 +105,30 @@ end class NiftyTranslator < Translator + INFO = '@nifty Translation ' + def initialize(cache={}) require 'mechanize' super(Translator::Direction.all_from_to(%w[ja en zh_CN ko], %w[ja]), cache) @form = WWW::Mechanize.new. get('http://nifty.amikai.com/amitext/indexUTF8.jsp'). - forms.name('translateForm').first + forms_with(:name => 'translateForm').last end def do_translate(text, from, to) - @form.radiobuttons.name('langpair').value = "#{from},#{to}".upcase - @form.fields.name('sourceText').value = text + @radio = @form.radiobuttons_with(:name => 'langpair').first + @radio.value = "#{from},#{to}".upcase + @radio.check + @form.fields_with(:name => 'sourceText').last.value = text - @form.submit(@form.buttons.name('translate')). - forms.name('translateForm').fields.name('translatedText').value + @form.submit(@form.buttons_with(:name => 'translate').last). + forms_with(:name => 'translateForm').last.fields_with(:name => 'translatedText').last.value end end class ExciteTranslator < Translator + INFO = 'Excite.jp Translation ' def initialize(cache={}) require 'mechanize' @@ -129,7 +151,7 @@ class ExciteTranslator < Translator def open_form(name) WWW::Mechanize.new.get("http://www.excite.co.jp/world/#{name}"). - forms.name('world').first + forms_with(:name => 'world').first end def do_translate(text, from, to) @@ -137,15 +159,15 @@ class ExciteTranslator < Translator form = @forms[non_ja_language] if non_ja_language =~ /zh_(CN|TW)/ - form.fields.name('wb_lp').value = "#{from}#{to}".sub(/_(?:CN|TW)/, '').upcase - form.fields.name('big5').value = ($1 == 'TW' ? 'yes' : 'no') + form_with_fields(:name => 'wb_lp').first.value = "#{from}#{to}".sub(/_(?:CN|TW)/, '').upcase + form_with_fields(:name => 'big5').first.value = ($1 == 'TW' ? 'yes' : 'no') else # the en<->ja page is in Shift_JIS while other pages are UTF-8 text = Iconv.iconv('Shift_JIS', 'UTF-8', text) if non_ja_language == 'en' - form.fields.name('wb_lp').value = "#{from}#{to}".upcase + form.fields_with(:name => 'wb_lp').first.value = "#{from}#{to}".upcase end - form.fields.name('before').value = text - result = form.submit.forms.name('world').fields.name('after').value + form.fields_with(:name => 'before').first.value = text + result = form.submit.forms_with(:name => 'world').first.fields_with(:name => 'after').first.value # the en<->ja page is in Shift_JIS while other pages are UTF-8 if non_ja_language == 'en' Iconv.iconv('UTF-8', 'Shift_JIS', result) @@ -158,14 +180,18 @@ end class GoogleTranslator < Translator + INFO = 'Google Translate ' + def initialize(cache={}) require 'mechanize' load_form! - language_pairs = @lang_list.options.map do |o| - # these options have values like "en|zh-CN"; map to things like ['en', 'zh_CN']. - o.value.split('|').map {|l| l.sub('-', '_')} - end - super(Translator::Direction.pairs(language_pairs), cache) + + # we can probably safely assume that google translate is able to translate from + # any language in the source lang drop down list to any language in the target one + # so we create the language pairs based on that assumption + sl = @source_list.options.map { |o| o.value.sub('-', '_') } + tl = @target_list.options.map { |o| o.value.sub('-', '_') } + super(Translator::Direction.all_from_to(tl, sl), cache) end def load_form! @@ -173,113 +199,168 @@ class GoogleTranslator < Translator # without faking the user agent, Google Translate will serve non-UTF-8 text agent.user_agent_alias = 'Linux Konqueror' @form = agent.get('http://www.google.com/translate_t'). - forms.action('/translate_t').first - @lang_list = @form.fields.name('langpair') + forms_with(:action => '/translate_t').first + @source_list = @form.fields_with(:name => 'sl').last + @target_list = @form.fields_with(:name => 'tl').last end def do_translate(text, from, to) load_form! - @lang_list.value = "#{from}|#{to}".sub('_', '-') - @form.fields.name('text').value = text + @source_list.value = from.sub('_', '-') + @target_list.value = to.sub('_', '-') + @form.fields_with(:name => 'text').last.value = text @form.submit.parser.search('div#result_box').inner_html end end class BabelfishTranslator < Translator + INFO = 'AltaVista Babel Fish Translation ' + def initialize(cache) require 'mechanize' - + @form = WWW::Mechanize.new.get('http://babelfish.altavista.com/babelfish/'). - forms.name('frmTrText').first - @lang_list = @form.fields.name('lp') + forms_with(:name => 'frmTrText').first + @lang_list = @form.fields_with(:name => 'lp').first language_pairs = @lang_list.options.map {|o| o.value.split('_')}. reject {|p| p.empty?} super(Translator::Direction.pairs(language_pairs), cache) end def do_translate(text, from, to) - if @form.fields.name('trtext').empty? + if @form.fields_with(:name => 'trtext').empty? @form.add_field!('trtext', text) else - @form.fields.name('trtext').value = text + @form.fields_with(:name => 'trtext').first.value = text end @lang_list.value = "#{from}_#{to}" - @form.submit.parser.search("td.s/div[@style]").inner_html + @form.submit.parser.search("div[@id='result']/div[@style]").inner_html + end +end + +class WorldlingoTranslator < Translator + INFO = 'WorldLingo Free Online Translator ' + + LANGUAGES = %w[en fr de it pt es ru nl el sv ar ja ko zh_CN zh_TW] + def initialize(cache) + require 'uri' + super(Translator::Direction.all_to_all(LANGUAGES), cache) + end + + def translate(text, from, to) + response = Irc::Utils.bot.httputil.get_response(URI.escape( + "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}")) + # WorldLingo seems to respond an XML when error occurs + case response['Content-Type'] + when %r'text/plain' + response.body + else + raise Translator::NoTranslationError + end end end class TranslatorPlugin < Plugin - BotConfig.register BotConfigIntegerValue.new('translate.timeout', + Config.register Config::IntegerValue.new('translator.timeout', :default => 30, :validate => Proc.new{|v| v > 0}, :desc => _("Number of seconds to wait for the translation service before timeout")) + TRANSLATORS = { + 'nifty' => NiftyTranslator, + 'excite' => ExciteTranslator, + 'google_translate' => GoogleTranslator, + 'babelfish' => BabelfishTranslator, + 'worldlingo' => WorldlingoTranslator, + } + def initialize super - translator_classes = { - 'nifty' => NiftyTranslator, - 'excite' => ExciteTranslator, - 'google_translate' => GoogleTranslator, - 'babelfish' => BabelfishTranslator - } @translators = {} - - translator_classes.each_pair do |name, c| + TRANSLATORS.each_pair do |name, c| begin @translators[name] = c.new(@registry.sub_registry(name)) - map "#{name} :from :to *phrase", :action => :cmd_translate - rescue + map "#{name} :from :to *phrase", + :action => :cmd_translate, :thread => true + rescue Exception warning _("Translator %{name} cannot be used: %{reason}") % {:name => name, :reason => $!} end end + + Config.register Config::ArrayValue.new('translator.default_list', + :default => TRANSLATORS.keys, + :validate => Proc.new {|l| l.all? {|t| TRANSLATORS.has_key?(t)}}, + :desc => _("List of translators to try in order when translator name not specified"), + :on_change => Proc.new {|bot, v| update_default}) + update_default end def help(plugin, topic=nil) - if @translators.has_key?(topic) - _('Supported directions of translation for %{translator}: %{directions}') % { - :translator => topic, - :directions => @translators[topic].directions.map do |source, targets| + if @translators.has_key?(plugin) + translator = @translators[plugin] + _('%{translator} => Look up phrase using %{info}, supported from -> to languages: %{directions}') % { + :translator => plugin, + :info => translator.class::INFO, + :directions => translator.directions.map do |source, targets| _('%{source} -> %{targets}') % {:source => source, :targets => targets.to_a.join(', ')} end.join(' | ') } else - _('Command: , where is one of: %{translators}. Use help to look up supported from and to languages') % + _('Command: , where 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 " to look up supported from and to languages') % {:translators => @translators.keys.join(', ')} end end + def update_default + @default_translators = bot.config['translator.default_list'] & @translators.keys + end + + def cmd_translator(m, params) + from, to = params[:from], params[:to] + translator = @default_translators.find {|t| @translators[t].support?(from, to)} + if translator + cmd_translate m, params.merge({:translator => translator, :show_provider => true}) + else + m.reply _('None of the default translators (translator.default_list) supports translating from %{source} to %{target}') % {:source => from, :target => to} + end + end + def cmd_translate(m, params) # get the first word of the command - tname = m.message[/\A(\w+)\s/, 1] + tname = params[:translator] || m.message[/\A(\w+)\s/, 1] translator = @translators[tname] from, to, phrase = params[:from], params[:to], params[:phrase].to_s if translator begin - if translator.support?(from, to) - translation = Timeout.timeout(@bot.config['translate.timeout']) do - translator.translate(phrase, from, to) - end - if translation.empty? - m.reply _('No translation returned') - else - m.reply translation - end - else - m.reply _("%{translator} doesn't support translating from %{source} to %{target}") % - {:translator => tname, :source => from, :target => to} + translation = Timeout.timeout(@bot.config['translator.timeout']) do + translator.translate(phrase, from, to) end + m.reply(if params[:show_provider] + _('%{translation} (provided by %{translator})') % + {:translation => translation, :translator => tname} + else + translation + end) + + rescue Translator::UnsupportedDirectionError + m.reply _("%{translator} doesn't support translating from %{source} to %{target}") % + {:translator => tname, :source => from, :target => to} + rescue Translator::NoTranslationError + m.reply _('%{translator} failed to provide a translation') % + {:translator => tname} rescue Timeout::Error m.reply _('The translator timed out') end else - m.reply _('No translator called %{name}') % {:name => translator} + m.reply _('No translator called %{name}') % {:name => tname} end end end plugin = TranslatorPlugin.new - +plugin.map 'translator :from :to *phrase', + :action => :cmd_translator, :thread => true