X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Ftranslator.rb;h=933969a1a7f1aa62ffb388fefcf41f5a96443309;hb=12a508e1a75bc1de501f36ca9e4767f2349fe12e;hp=65348d0f4d7332837af3ad59549ac6b2f31e2a01;hpb=3b0979270bd28f1d7bcae81c7065cdc8e5867d96;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/translator.rb b/data/rbot/plugins/translator.rb
index 65348d0f..933969a1 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'
@@ -34,7 +39,15 @@ class Translator
     @cache = cache
   end
 
- 
+  # Many translators use Mechanize, which changed namespace around version 1.0
+  # To support both pre-1.0 and post-1.0 namespaces, we use these auxiliary
+  # method. The translator still needs to require 'mechanize' on initialization
+  # if it needs it.
+  def mechanize
+    return Mechanize if defined? Mechanize
+    return WWW::Mechanize
+  end
+
   # whether the translator supports this direction
   def support?(from, to)
     from != to && @directions[from].include?(to)
@@ -105,17 +118,19 @@ class NiftyTranslator < Translator
   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
   end
 
   def do_translate(text, from, to)
-    @form.radiobuttons.name('langpair').value = "#{from},#{to}".upcase
-    @form.fields.name('sourceText').value = text
-
-    @form.submit(@form.buttons.name('translate')).
-          forms.name('translateForm').fields.name('translatedText').value
+    @form ||= mechanize.new.
+              get('http://nifty.amikai.com/amitext/indexUTF8.jsp').
+              forms_with(:name => 'translateForm').last
+    @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_with(:name => 'translate').last).
+          forms_with(:name => 'translateForm').last.fields_with(:name => 'translatedText').last.value
   end
 end
 
@@ -143,8 +158,8 @@ class ExciteTranslator < Translator
   end
 
   def open_form(name)
-    WWW::Mechanize.new.get("http://www.excite.co.jp/world/#{name}").
-                   forms.name('world').first
+    mechanize.new.get("http://www.excite.co.jp/world/#{name}").
+                   forms_with(:name => 'world').first
   end
 
   def do_translate(text, from, to)
@@ -152,15 +167,15 @@ 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)
@@ -175,31 +190,30 @@ end
 class GoogleTranslator < Translator
   INFO = 'Google Translate <http://www.google.com/translate_t>'
 
+  LANGUAGES =
+    %w[af sq am ar hy az eu be bn bh bg my ca chr zh zh_CN zh_TW hr
+    cs da dv en eo et tl fi fr gl ka de el gn gu iw hi hu is id iu
+    ga it ja kn kk km ko lv lt mk ms ml mt mr mn ne no or ps fa pl
+    pt_PT pa ro ru sa sr sd si sk sl es sw sv tg ta tl te th bo tr
+    uk ur uz ug vi cy yi auto]
   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)
-  end
-
-  def load_form!
-    agent = WWW::Mechanize.new
-    # 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')
+    require "uri"
+    require "json"
+    super(Translator::Direction.all_to_all(LANGUAGES), cache)
   end
 
   def do_translate(text, from, to)
-    load_form!
+    langpair = [from == 'auto' ? '' : from, to].map { |e| e.tr('_', '-') }.join("|")
+    raw_json = Irc::Utils.bot.httputil.get_response(URI.escape(
+               "http://ajax.googleapis.com/ajax/services/language/translate?v=1.0&q=#{text}&langpair=#{langpair}")).body
+    response = JSON.parse(raw_json)
 
-    @lang_list.value = "#{from}|#{to}".sub('_', '-')
-    @form.fields.name('text').value = text
-    @form.submit.parser.search('div#result_box').inner_html
+    if response["responseStatus"] != 200
+      raise Translator::NoTranslationError, response["responseDetails"]
+    else
+      translation = response["responseData"]["translatedText"]
+      return Utils.decode_html_entities(translation)
+    end
   end
 end
 
@@ -209,23 +223,31 @@ class BabelfishTranslator < Translator
 
   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')
-    language_pairs = @lang_list.options.map {|o| o.value.split('_')}.
-                                            reject {|p| p.empty?}
+    (_, lang_list) = parse_page
+    language_pairs = lang_list.options.map {|o| o.value.split('_')}.
+                                           reject {|p| p.empty?}
     super(Translator::Direction.pairs(language_pairs), cache)
   end
 
+  def parse_page
+    form = mechanize.new.get('http://babelfish.altavista.com/babelfish/').
+           forms_with(:name => 'frmTrText').first
+    lang_list = form.fields_with(:name => 'lp').first
+    [form, lang_list]
+  end
+
   def do_translate(text, from, to)
-    if @form.fields.name('trtext').empty?
+    unless @form && @lang_list
+      @form, @lang_list = parse_page
+    end
+    
+    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
 
@@ -252,9 +274,12 @@ class WorldlingoTranslator < Translator
 end
 
 class TranslatorPlugin < Plugin
-  BotConfig.register BotConfigIntegerValue.new('translator.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"))
+  Config.register Config::StringValue.new('translator.destination',
+    :default => "en",
+    :desc => _("Default destination language to be used with translate command"))
 
   TRANSLATORS = {
     'nifty' => NiftyTranslator,
@@ -266,19 +291,17 @@ class TranslatorPlugin < Plugin
 
   def initialize
     super
-
+    @failed_translators = []
     @translators = {}
     TRANSLATORS.each_pair do |name, c|
-      begin
+      watch_for_fail(name) do
         @translators[name] = c.new(@registry.sub_registry(name))
-        map "#{name} :from :to *phrase", :action => :cmd_translate
-      rescue
-        warning _("Translator %{name} cannot be used: %{reason}") %
-               {:name => name, :reason => $!}
+        map "#{name} :from :to *phrase",
+          :action => :cmd_translate, :thread => true
       end
     end
 
-    BotConfig.register BotConfigArrayValue.new('translator.default_list',
+    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"),
@@ -286,33 +309,95 @@ class TranslatorPlugin < Plugin
     update_default
   end
 
+  def watch_for_fail(name, &block)
+    begin
+      yield
+    rescue Exception
+      @failed_translators << { :name => name, :reason => $!.to_s }
+
+      warning _("Translator %{name} cannot be used: %{reason}") %
+             {:name => name, :reason => $!}
+      map "#{name} [*args]", :action => :failed_translator,
+                             :defaults => {:name => name, :reason => $!}
+    end
+  end
+
+  def failed_translator(m, params)
+    m.reply _("Translator %{name} cannot be used: %{reason}") %
+            {:name => params[:name], :reason => params[:reason]}
+  end
+
   def help(plugin, topic=nil)
-    if @translators.has_key?(topic)
-      translator = @translators[topic]
-      _('%{info}, supported directions of translation: %{directions}') % {
-        :info => translator.class::INFO,
-        :directions => translator.directions.map do |source, targets|
-                         _('%{source} -> %{targets}') %
-                         {:source => source, :targets => targets.to_a.join(', ')}
-                       end.join(' | ')
-      }
+    case (topic.intern rescue nil)
+    when :failed
+      unless @failed_translators.empty?
+        failed_list = @failed_translators.map { |t| _("%{bold}%{translator}%{bold}: %{reason}") % {
+          :translator => t[:name],
+          :reason => t[:reason],
+          :bold => Bold
+        }}
+
+        _("Failed translators: %{list}") % { :list => failed_list.join(", ") }
+      else
+        _("None of the translators failed")
+      end
     else
-      _('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 <translator>" to look up supported from and to languages') %
-        {:translators => @translators.keys.join(', ')}
+      if @translators.has_key?(plugin)
+        translator = @translators[plugin]
+        _('%{translator} <from> <to> <phrase> => 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
+        help_str = _('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') %
+                     {:translators => @translators.keys.join(', ')}
+
+        help_str << "\n" + _("%{bold}Note%{bold}: %{failed_amt} translators failed, see %{reverse}%{prefix}help translate failed%{reverse} for details") % {
+          :failed_amt => @failed_translators.size,
+          :bold => Bold,
+          :reverse => Reverse,
+          :prefix => @bot.config['core.address_prefix'].first
+        }
+
+        help_str
+      end
     end
   end
 
+  def languages
+    @languages ||= @translators.map { |t| t.last.directions.keys }.flatten.uniq
+  end
+
   def update_default
-    @default_translators = bot.config['translator.default_list'] & @translators.keys 
+    @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)}
+    params[:to] = @bot.config['translator.destination'] if params[:to].nil?
+    params[:from] ||= 'auto'
+    translator = @default_translators.find {|t| @translators[t].support?(params[:from], params[:to])}
+
     if translator
-      cmd_translate m, params.merge({:translator => translator})
+      cmd_translate m, params.merge({:translator => translator, :show_provider => false})
     else
-      m.reply _('None of the default translators (translator.default_list) supports translating from %{source} to %{target}') % {:source => from, :target => to}
+      # When translate command is used without source language, "auto" as source
+      # language is assumed. It means that google translator is used and we let google
+      # figure out what the source language is.
+      #
+      # Problem is that the google translator will fail if the system that the bot is
+      # running on does not have the json gem installed.
+      if params[:from] == 'auto'
+        m.reply _("Unable to auto-detect source language due to broken google translator, see %{reverse}%{prefix}help translate failed%{reverse} for details") % {
+          :reverse => Reverse,
+          :prefix => @bot.config['core.address_prefix'].first
+        }
+      else
+        m.reply _('None of the default translators (translator.default_list) supports translating from %{source} to %{target}') % {:source => params[:from], :target => params[:to]}
+      end
     end
   end
 
@@ -322,25 +407,57 @@ class TranslatorPlugin < Plugin
     translator = @translators[tname]
     from, to, phrase = params[:from], params[:to], params[:phrase].to_s
     if translator
-      begin
-        translation = Timeout.timeout(@bot.config['translator.timeout']) do
-          translator.translate(phrase, from, to)
+      watch_for_fail(tname) do
+        begin
+          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.gsub("_", " ")}
+                  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
-        m.reply translation
-      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 => tname}
     end
   end
+
+  # URL translation has nothing to do with Translators so let's make it
+  # separate, and Google exclusive for now
+  def cmd_translate_url(m, params)
+    params[:to] = @bot.config['translator.destination'] if params[:to].nil?
+    params[:from] ||= 'auto'
+
+    translate_url = "http://translate.google.com/translate?sl=%{from}&tl=%{to}&u=%{url}" % {
+      :from => params[:from],
+      :to   => params[:to],
+      :url  => CGI.escape(params[:url].to_s)
+    }
+
+    m.reply(translate_url)
+  end
 end
 
 plugin = TranslatorPlugin.new
-plugin.map 'translator :from :to *phrase', :action => :cmd_translator
+req = Hash[*%w(from to).map { |e| [e.to_sym, /#{plugin.languages.join("|")}/] }.flatten]
+
+plugin.map 'translate [:from] [:to] :url',
+           :action => :cmd_translate_url, :requirements => req.merge(:url => %r{^https?://[^\s]*})
+plugin.map 'translator [:from] [:to] :url',
+           :action => :cmd_translate_url, :requirements => req.merge(:url => %r{^https?://[^\s]*})
+plugin.map 'translate [:from] [:to] *phrase',
+           :action => :cmd_translator, :thread => true, :requirements => req
+plugin.map 'translator [:from] [:to] *phrase',
+           :action => :cmd_translator, :thread => true, :requirements => req