#-- vim:sw=2:et #++ # # :title: Google and Wikipedia search plugin for rbot # # Author:: Tom Gilbert (giblet) # Author:: Giuseppe "Oblomov" Bilotta # # Copyright:: (C) 2002-2005 Tom Gilbert # Copyright:: (C) 2006 Tom Gilbert, Giuseppe Bilotta # Copyright:: (C) 2006-2007 Giuseppe Bilotta # TODO:: use lr=lang_ or whatever is most appropriate to let google know # it shouldn't use the bot's location to find the preferred language require 'uri' Net::HTTP.version_1_2 GOOGLE_WAP_LINK = /(.*?)<\/a>/im GOOGLE_CALC_RESULT = %r{

 (.+)
} class SearchPlugin < Plugin BotConfig.register BotConfigIntegerValue.new('google.hits', :default => 3, :desc => "Number of hits to return from Google searches") BotConfig.register BotConfigIntegerValue.new('google.first_par', :default => 0, :desc => "When set to n > 0, the bot will return the first paragraph from the first n search hits") BotConfig.register BotConfigIntegerValue.new('wikipedia.hits', :default => 3, :desc => "Number of hits to return from Wikipedia searches") BotConfig.register BotConfigIntegerValue.new('wikipedia.first_par', :default => 1, :desc => "When set to n > 0, the bot will return the first paragraph from the first n wikipedia search hits") def help(plugin, topic="") case topic when "search", "google" "#{topic} => search google for " when "gcalc" "gcalc => use the google calculator to find the answer to " when "wp" "wp [] => search for on Wikipedia. You can select a national to only search the national Wikipedia" else "search (or: google ) => search google for | wp => search for on Wikipedia" end end def google(m, params) what = params[:words].to_s searchfor = URI.escape what # This method is also called by other methods to restrict searching to some sites if params[:site] site = "site:#{params[:site]}+" else site = "" end # It is also possible to choose a filter to remove constant parts from the titles # e.g.: "Wikipedia, the free encyclopedia" when doing Wikipedia searches filter = params[:filter] || "" url = "http://www.google.com/wml/search?q=#{site}#{searchfor}" hits = params[:hits] || @bot.config['google.hits'] begin wml = @bot.httputil.get(url) raise unless wml rescue => e m.reply "error googling for #{what}" return end results = wml.scan(GOOGLE_WAP_LINK) if results.length == 0 m.reply "no results found for #{what}" return end urls = Array.new results = results[0...hits].map { |res| n = res[0] t = Utils.decode_html_entities res[2].gsub(filter, '').strip u = URI.unescape res[1] urls.push(u) "#{n}. #{Bold}#{t}#{Bold}: #{u}" }.join(" | ") m.reply "Results for #{what}: #{results}", :split_at => /\s+\|\s+/ first_pars = params[:firstpar] || @bot.config['google.first_par'] return unless first_pars > 0 Utils.get_first_pars urls, first_pars, :message => m end def gcalc(m, params) what = params[:words].to_s searchfor = URI.escape(what).sub('+','%2B') debug "Getting gcalc thing: #{searchfor.inspect}" url = "http://www.google.com/search?q=#{searchfor}" begin html = @bot.httputil.get(url) rescue => e m.reply "error googlecalcing #{what}" return end debug "#{html.size} bytes of html recieved" results = html.scan(GOOGLE_CALC_RESULT) debug "results: #{results.inspect}" if results.length != 1 m.reply "couldn't calculate #{what}" return end result = results[0][0].ircify_html debug "replying with: #{result.inspect}" m.reply "#{result}" end def wikipedia(m, params) lang = params[:lang] site = "#{lang.nil? ? '' : lang + '.'}wikipedia.org" debug "Looking up things on #{site}" params[:site] = site params[:filter] = / - Wikipedia.*$/ params[:hits] = @bot.config['wikipedia.hits'] params[:firstpar] = @bot.config['wikipedia.first_par'] return google(m, params) end end plugin = SearchPlugin.new plugin.map "search *words", :action => 'google' plugin.map "google *words", :action => 'google' plugin.map "gcalc *words", :action => 'gcalc' plugin.map "wp :lang *words", :action => 'wikipedia', :requirements => { :lang => /^\w\w\w?$/ } plugin.map "wp *words", :action => 'wikipedia'