-#-- vim:sw=2:et\r
-#++\r
-#\r
-# :title: Dictionary lookup plugin for rbot\r
-#\r
-# Author:: Giuseppe "Oblomov" Bilotta <giuseppe.bilotta@gmail.com>\r
-# Copyright:: (C) 2006-2007 Giuseppe Bilotta\r
-# License:: GPL v2\r
-#\r
-# Provides a link to the definition of a word in one of the supported\r
-# dictionaries. Currently available are\r
-# * the Oxford dictionary for (British) English\r
-# * the De Mauro/Paravia dictionary for Italian\r
-# * the Chambers dictionary for English (accepts both US and UK)\r
-#\r
-# Other plugins can use this one to check if a given word is valid in italian\r
-# or english by using the is_italian?/is_british?/is_english? methods\r
-#\r
-# TODO: cache results and reuse them if get_cached returns a cache copy\r
-\r
-DEMAURO_LEMMA = /<anchor>(.*?)(?: - (.*?))<go href="lemma.php\?ID=(\d+)"\/><\/anchor>/\r
-\r
-class DictPlugin < Plugin\r
- BotConfig.register BotConfigIntegerValue.new('dict.hits',\r
- :default => 3,\r
- :desc => "Number of hits to return from a dictionary lookup")\r
- BotConfig.register BotConfigIntegerValue.new('dict.first_par',\r
- :default => 0,\r
- :desc => "When set to n > 0, the bot will return the first paragraph from the first n dictionary hits")\r
-\r
- def initialize\r
- super\r
- @dmurl = "http://www.demauroparavia.it/"\r
- @dmwapurl = "http://wap.demauroparavia.it/index.php?lemma=%s"\r
- @dmwaplemma = "http://wap.demauroparavia.it/lemma.php?ID=%s"\r
- @oxurl = "http://www.askoxford.com/concise_oed/%s"\r
- @chambersurl = "http://www.chambersharrap.co.uk/chambers/features/chref/chref.py/main?query=%s&title=21st"\r
- end\r
-\r
-\r
- def help(plugin, topic="")\r
- case topic\r
- when "demauro"\r
- return "demauro <word> => provides a link to the definition of <word> from the De Mauro/Paravia dictionary"\r
- when "oxford"\r
- return "oxford <word> => provides a link to the definition of <word> (it can also be an expression) from the Concise Oxford dictionary"\r
- when "chambers"\r
- return "chambers <word> => provides a link to the definition of <word> (it can also be an expression) from the Chambers 21st Century Dictionary"\r
- end\r
- return "<dictionary> <word>: check for <word> on <dictionary> where <dictionary> can be one of: demauro, oxford, chambers"\r
- end\r
-\r
- def demauro(m, params)\r
- justcheck = params[:justcheck]\r
-\r
- word = params[:word].downcase\r
- url = @dmwapurl % URI.escape(word)\r
- xml = nil\r
- info = @bot.httputil.get_response(url) rescue nil\r
- xml = info.body if info\r
- if xml.nil?\r
- info = info ? " (#{info.code} - #{info.message})" : ""\r
- return false if justcheck\r
- m.reply "An error occurred while looking for #{word}#{info}"\r
- return\r
- end\r
- if xml=~ /Non ho trovato occorrenze per/\r
- return false if justcheck\r
- m.reply "Nothing found for #{word}"\r
- return\r
- end\r
- entries = xml.scan(DEMAURO_LEMMA)\r
- text = word\r
- urls = []\r
- if !entries.assoc(word) and !entries.assoc(word.upcase)\r
- return false if justcheck\r
- text += " not found. Similar words"\r
- end\r
- return true if justcheck\r
- text += ": "\r
- n = 0\r
- hits = @bot.config['dict.hits']\r
- text += entries[0...hits].map { |ar|\r
- n += 1\r
- urls << @dmwaplemma % ar[2]\r
- "#{n}. #{Bold}#{ar[0]}#{Bold} - #{ar[1].gsub(/<\/?em>/,'')}: #{@dmurl}#{ar[2]}"\r
- }.join(" | ")\r
- m.reply text\r
-\r
- first_pars = @bot.config['dict.first_par']\r
-\r
- return unless first_pars > 0\r
-\r
- Utils.get_first_pars urls, first_pars, :message => m,\r
- :strip => /^\S+\s+-\s+/\r
-\r
- end\r
-\r
- def is_italian?(word)\r
- return demauro(nil, :word => word, :justcheck => true)\r
- end\r
-\r
-\r
- def oxford(m, params)\r
- justcheck = params[:justcheck]\r
-\r
- word = params[:word].join\r
- [word, word + "_1"].each { |check|\r
- url = @oxurl % URI.escape(check)\r
- h = @bot.httputil.head(url, :max_redir => 5)\r
- if h\r
- m.reply("#{word} found: #{url}") unless justcheck\r
- return true\r
- end\r
- }\r
- return false if justcheck\r
- m.reply "#{word} not found"\r
- end\r
-\r
- def is_british?(word)\r
- return oxford(nil, :word => word, :justcheck => true)\r
- end\r
-\r
-\r
- def chambers(m, params)\r
- justcheck = params[:justcheck]\r
-\r
- word = params[:word].to_s.downcase\r
- url = @chambersurl % URI.escape(word)\r
- xml = nil\r
- info = @bot.httputil.get_response(url) rescue nil\r
- xml = info.body if info\r
- case xml\r
- when nil\r
- info = info ? " (#{info.code} - #{info.message})" : ""\r
- return false if justcheck\r
- m.reply "An error occurred while looking for #{word}#{info}"\r
- return\r
- when /Sorry, no entries for <b>.*?<\/b> were found./\r
- return false if justcheck\r
- m.reply "Nothing found for #{word}"\r
- return\r
- when /No exact matches for <b>.*?<\/b>, but the following may be helpful./\r
- return false if justcheck\r
- m.reply "Nothing found for #{word}, but see #{url} for possible suggestions"\r
- else\r
- return true if justcheck\r
- m.reply "#{word}: #{url}"\r
- end\r
- end\r
-\r
- def is_english?(word)\r
- return chambers(nil, :word => word, :justcheck => true)\r
- end\r
-\r
-end\r
-\r
-plugin = DictPlugin.new\r
-plugin.map 'demauro :word', :action => 'demauro'\r
-plugin.map 'oxford *word', :action => 'oxford'\r
-plugin.map 'chambers *word', :action => 'chambers'\r
-\r
+#-- vim:sw=2:et
+#++
+#
+# :title: Dictionary lookup plugin for rbot
+#
+# Author:: Giuseppe "Oblomov" Bilotta <giuseppe.bilotta@gmail.com>
+# Copyright:: (C) 2006-2007 Giuseppe Bilotta
+# License:: GPL v2
+#
+# Provides a link to the definition of a word in one of the supported
+# dictionaries. Currently available are
+# * the Oxford dictionary for (British) English
+# * the De Mauro/Paravia dictionary for Italian
+# * the Chambers dictionary for English (accepts both US and UK)
+# * the Littré dictionary for French
+#
+# Other plugins can use this one to check if a given word is valid in italian
+# or english or french by using the is_italian?, is_british?, is_english?,
+# is_french? methods
+#
+# TODO: cache results and reuse them if get_cached returns a cache copy
+
+DEMAURO_LEMMA = /<anchor>(.*?)(?: - (.*?))<go href="lemma.php\?ID=(\d+)"\/><\/anchor>/
+CHAMBERS_LEMMA = /<p><span class="hwd">(.*?)<\/span> <span class="psa">(.*?)<\/span>(.*?)<\/p>/
+
+class DictPlugin < Plugin
+ Config.register Config::IntegerValue.new('dict.hits',
+ :default => 3,
+ :desc => "Number of hits to return from a dictionary lookup")
+ Config.register Config::IntegerValue.new('dict.first_par',
+ :default => 0,
+ :desc => "When set to n > 0, the bot will return the first paragraph from the first n dictionary hits")
+
+ def demauro_filter(s)
+ # check if it's a page we can handle
+ loc = Utils.check_location(s, @dmurlrx)
+ # the location might be not good, but we might still be able to handle the
+ # page
+ if !loc and s[:text] !~ /<!-- Il dizionario della lingua italiana Paravia: /
+ debug "not our business"
+ return
+ end
+ # we want to grab the content from the WAP page, since it's in a much
+ # cleaner HTML, so first try to get the word ID
+ if s[:text] !~ %r{<li><a href="(\d+)" title="vai al lemma precedente" accesskey="p">lemma precedente</a></li>}
+ return
+ end
+ id = $1.to_i + 1
+ title = s[:text].ircify_html_title
+ content = @bot.filter(:htmlinfo, URI.parse(@dmwaplemma % id))[:content]
+ return {:title => title, :content => content.sub(/^\S+\s+-\s+/,'')}
+ end
+
+ def initialize
+ super
+ @dmurl = "http://old.demauroparavia.it/"
+ @dmurlrx = %r{http://(?:www|old\.)?demauroparavia\.it/(\d+)}
+ @dmwapurl = "http://wap.demauroparavia.it/index.php?lemma=%s"
+ @dmwaplemma = "http://wap.demauroparavia.it/lemma.php?ID=%s"
+ @oxurl = "http://www.askoxford.com/concise_oed/%s"
+ @chambersurl = "http://www.chambersharrap.co.uk/chambers/features/chref/chref.py/main?query=%s&title=21st"
+ @littreurl = "http://francois.gannaz.free.fr/Littre/xmlittre.php?requete=%s"
+
+ @bot.register_filter(:demauro, :htmlinfo) { |s| demauro_filter(s) }
+ end
+
+
+ def help(plugin, topic="")
+ case topic
+ when "demauro"
+ return "demauro <word> => provides a link to the definition of <word> from the De Mauro/Paravia dictionary"
+ when "oxford"
+ return "oxford <word> => provides a link to the definition of <word> (it can also be an expression) from the Concise Oxford dictionary"
+ when "chambers"
+ return "chambers <word> => provides a link to the definition of <word> (it can also be an expression) from the Chambers 21st Century Dictionary"
+ when "littre"
+ return "littre <word> => provides a link to the definition of <word> (it can also be an expression) from the Littré online dictionary"
+ end
+ return "<dictionary> <word>: check for <word> on <dictionary> where <dictionary> can be one of: demauro, oxford, chambers, littre"
+ end
+
+ def demauro(m, params)
+ justcheck = params[:justcheck]
+
+ word = params[:word].downcase
+ url = @dmwapurl % CGI.escape(word)
+ xml = nil
+ info = @bot.httputil.get_response(url) rescue nil
+ xml = info.body if info
+ if xml.nil?
+ info = info ? " (#{info.code} - #{info.message})" : ""
+ return false if justcheck
+ m.reply "An error occurred while looking for #{word}#{info}"
+ return
+ end
+ if xml=~ /Non ho trovato occorrenze per/
+ return false if justcheck
+ m.reply "Nothing found for #{word}"
+ return
+ end
+ entries = xml.scan(DEMAURO_LEMMA)
+ text = word
+ urls = []
+ if not entries.transpose.first.grep(/\b#{word}\b/)
+ return false if justcheck
+ text += " not found. Similar words"
+ end
+ return true if justcheck
+ text += ": "
+ n = 0
+ hits = @bot.config['dict.hits']
+ text += entries[0...hits].map { |ar|
+ n += 1
+ urls << @dmwaplemma % ar[2]
+ key = ar[1].ircify_html
+ "#{n}. #{Bold}#{ar[0]}#{Bold} - #{key}: #{@dmurl}#{ar[2]}"
+ }.join(" | ")
+ m.reply text
+
+ first_pars = @bot.config['dict.first_par']
+
+ return unless first_pars > 0
+
+ Utils.get_first_pars urls, first_pars, :message => m,
+ :strip => /^.+?\s+-\s+/
+
+ end
+
+ def is_italian?(word)
+ return demauro(nil, :word => word, :justcheck => true)
+ end
+
+
+ def oxford(m, params)
+ justcheck = params[:justcheck]
+
+ word = params[:word].join
+ [word, word + "_1"].each { |check|
+ url = @oxurl % CGI.escape(check)
+ if params[:british]
+ url << "?view=uk"
+ end
+ h = @bot.httputil.get(url, :max_redir => 5)
+ if h
+ defs = h.split("<span class=\"definition\">")
+ defs = defs[1..-1].map {|d| d.split("</span>")[0]}
+ if defs.size == 0
+ return false if justcheck
+ m.reply "#{word} not found"
+ return false
+ end
+ m.reply("#{word}: #{url}") unless justcheck
+ defn = defs[0]
+ m.reply("#{Bold}%s#{Bold}: %s" % [word, defn.ircify_html(:nbsp => :space)], :overlong => :truncate)
+ return true
+ end
+ }
+ end
+
+ def is_british?(word)
+ return oxford(nil, :word => word, :justcheck => true, :british => true)
+ end
+
+
+ def chambers(m, params)
+ justcheck = params[:justcheck]
+
+ word = params[:word].to_s.downcase
+ url = @chambersurl % CGI.escape(word)
+ xml = nil
+ info = @bot.httputil.get_response(url) rescue nil
+ xml = info.body if info
+ case xml
+ when nil
+ info = info ? " (#{info.code} - #{info.message})" : ""
+ return false if justcheck
+ m.reply "An error occurred while looking for #{word}#{info}"
+ return
+ when /Sorry, no entries for <b>.*?<\/b> were found./
+ return false if justcheck
+ m.reply "Nothing found for #{word}"
+ return
+ when /No exact matches for <b>.*?<\/b>, but the following may be helpful./
+ return false if justcheck
+ m.reply "Nothing found for #{word}, but see #{url} for possible suggestions"
+ return
+ end
+ # Else, we have a hit
+ return true if justcheck
+ m.reply "#{word}: #{url}"
+ entries = xml.scan(CHAMBERS_LEMMA)
+ hits = @bot.config['dict.hits']
+ entries[0...hits].map { |ar|
+ m.reply(("#{Bold}%s#{Bold} #{Underline}%s#{Underline}%s" % ar).ircify_html, :overlong => :truncate)
+ }
+ end
+
+ def is_english?(word)
+ return chambers(nil, :word => word, :justcheck => true)
+ end
+
+ def littre(m, params)
+ justcheck = params[:justcheck]
+
+ word = params[:word].to_s.downcase
+ url = @littreurl % CGI.escape(word)
+ xml = nil
+ info = @bot.httputil.get_response(url) rescue nil
+ xml = info.body if info
+ head ||= xml.match(/<div class="entree">(.*?)<\/div>/)[1] rescue nil
+ case xml
+ when nil
+ info = info ? " (#{info.code} - #{info.message})" : ""
+ return false if justcheck
+ m.reply "An error occurred while looking for #{word}#{info}"
+ return
+ when /Erreur : le mot <STRONG>.*?<\/STRONG> n'a pas./
+ return false if justcheck
+ if head
+ m.reply "Nothing found for #{word}, I'll assume you meant #{head}"
+ else
+ m.reply "Nothing found for #{word}"
+ return
+ end
+ end
+ return true if justcheck
+ entete = xml.match(/<div class="entete">(.*?)<\/div>/m)[1] rescue nil
+ m.reply "#{head}: #{url} : #{entete.ircify_html rescue nil}"
+ entries = xml.scan(/<span class="variante">(.*?)<\!--variante-->/m)
+ hits = @bot.config['dict.hits']
+ n = 0
+ entries[0...hits].map { |ar|
+ n += 1
+ m.reply(("#{Bold}#{n}#{Bold} %s" % ar).ircify_html, :overlong => :truncate)
+ }
+ end
+
+ def is_french?(word)
+ return littre(nil, :word => word, :justcheck => true)
+ end
+
+end
+
+plugin = DictPlugin.new
+plugin.map 'demauro :word', :action => 'demauro', :thread => true
+plugin.map 'oxford *word', :action => 'oxford', :thread => true
+plugin.map 'chambers *word', :action => 'chambers', :thread => true
+plugin.map 'littre *word', :action => 'littre', :thread => true
+
projects / user / henk / code / ruby / rbot.git / blobdiff