X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fsearch.rb;h=b2e950ecd544b81247e2f380dab041dfbaf125e8;hb=a51ae05ee0252576a10a77c9f92c0ce096be73aa;hp=aea6dc4cc4a5f9061261eeb2b88471118b1dc22f;hpb=9c98ff2e3e6dbf13eb650bff91b49b37594ca4b9;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/search.rb b/data/rbot/plugins/search.rb
index aea6dc4c..b2e950ec 100644
--- a/data/rbot/plugins/search.rb
+++ b/data/rbot/plugins/search.rb
@@ -12,25 +12,26 @@
 
 # TODO:: use lr=lang_<code> or whatever is most appropriate to let google know
 #        it shouldn't use the bot's location to find the preferred language
+# TODO:: support localized uncyclopedias -- not easy because they have different names
+#        for most languages
 
-require 'uri'
-
-Net::HTTP.version_1_2
-
+GOOGLE_SEARCH = "http://www.google.com/search?oe=UTF-8&q="
+GOOGLE_WAP_SEARCH = "http://www.google.com/wml/search?hl=en&q="
 GOOGLE_WAP_LINK = /<a accesskey="(\d)" href=".*?u=(.*?)">(.*?)<\/a>/im
-GOOGLE_CALC_RESULT = %r{<p><table><tr><td><img src=/images/calc_img\.gif></td><td>&nbsp;</td><td nowrap><font size=\+1><b>(.+)</b></td></tr><tr><td>}
+GOOGLE_CALC_RESULT = %r{<img src=/images/calc_img\.gif(?: width=40 height=30 alt="")?></td><td>&nbsp;</td><td nowrap>(?:<h2 class=r>)?<font size=\+1><b>(.+)</b>(?:</h2>)?</td>}
+GOOGLE_DEF_RESULT = %r{<p> (Web definitions for .*?)<br/>(.*?)<br/>(.*?)\s-\s+<a href}
 
 class SearchPlugin < Plugin
-  BotConfig.register BotConfigIntegerValue.new('google.hits',
+  Config.register Config::IntegerValue.new('google.hits',
     :default => 3,
     :desc => "Number of hits to return from Google searches")
-  BotConfig.register BotConfigIntegerValue.new('google.first_par',
+  Config.register Config::IntegerValue.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',
+  Config.register Config::IntegerValue.new('wikipedia.hits',
     :default => 3,
     :desc => "Number of hits to return from Wikipedia searches")
-  BotConfig.register BotConfigIntegerValue.new('wikipedia.first_par',
+  Config.register Config::IntegerValue.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")
 
@@ -40,16 +41,20 @@ class SearchPlugin < Plugin
       "#{topic} <string> => search google for <string>"
     when "gcalc"
       "gcalc <equation> => use the google calculator to find the answer to <equation>"
+    when "gdef"
+      "gdef <term(s)> => use the google define mechanism to find a definition of <term(s)>"
     when "wp"
       "wp [<code>] <string> => search for <string> on Wikipedia. You can select a national <code> to only search the national Wikipedia"
+    when "unpedia"
+      "unpedia <string> => search for <string> on Uncyclopedia"
     else
-      "search <string> (or: google <string>) => search google for <string> | wp <string> => search for <string> on Wikipedia"
+      "search <string> (or: google <string>) => search google for <string> | wp <string> => search for <string> on Wikipedia | unpedia <string> => search for <string> on Uncyclopedia"
     end
   end
 
   def google(m, params)
     what = params[:words].to_s
-    searchfor = URI.escape what
+    searchfor = CGI.escape what
     # This method is also called by other methods to restrict searching to some sites
     if params[:site]
       site = "site:#{params[:site]}+"
@@ -60,10 +65,14 @@ class SearchPlugin < Plugin
     # e.g.: "Wikipedia, the free encyclopedia" when doing Wikipedia searches
     filter = params[:filter] || ""
 
-    url = "http://www.google.com/wml/search?q=#{site}#{searchfor}"
+    url = GOOGLE_WAP_SEARCH + site + searchfor
 
     hits = params[:hits] || @bot.config['google.hits']
 
+    first_pars = params[:firstpar] || @bot.config['google.first_par']
+
+    single = (hits == 1 and first_pars == 1)
+
     begin
       wml = @bot.httputil.get(url)
       raise unless wml
@@ -76,18 +85,23 @@ class SearchPlugin < Plugin
       m.reply "no results found for #{what}"
       return
     end
+    single ||= (results.length==1)
     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}"
+      single ? u : "#{n}. #{Bold}#{t}#{Bold}: #{u}"
     }.join(" | ")
 
-    m.reply "Results for #{what}: #{results}", :split_at => /\s+\|\s+/
+    # If we return a single, full result, change the output to a more compact representation
+    if single
+      m.reply "Result for %s: %s -- %s" % [what, results, Utils.get_first_pars(urls, first_pars)], :overlong => :truncate
+      return
+    end
 
-    first_pars = params[:firstpar] || @bot.config['google.first_par']
+    m.reply "Results for #{what}: #{results}", :split_at => /\s+\|\s+/
 
     return unless first_pars > 0
 
@@ -97,10 +111,10 @@ class SearchPlugin < Plugin
 
   def gcalc(m, params)
     what = params[:words].to_s
-    searchfor = URI.escape(what).sub('+','%2B')
+    searchfor = CGI.escape(what)
     
     debug "Getting gcalc thing: #{searchfor.inspect}"
-    url = "http://www.google.com/search?q=#{searchfor}"
+    url = GOOGLE_SEARCH + searchfor
 
     begin
       html = @bot.httputil.get(url)
@@ -124,6 +138,35 @@ class SearchPlugin < Plugin
     m.reply "#{result}"
   end
 
+  def gdef(m, params)
+    what = params[:words].to_s
+    searchfor = CGI.escape("define " + what)
+    
+    debug "Getting gdef thing: #{searchfor.inspect}"
+    url = GOOGLE_WAP_SEARCH + searchfor
+
+    begin
+      html = @bot.httputil.get(url)
+    rescue => e
+      m.reply "error googledefining #{what}"
+      return
+    end
+
+    debug html
+    results = html.scan(GOOGLE_DEF_RESULT)
+    debug "results: #{results.inspect}"
+    
+    if results.length != 1
+      m.reply "couldn't find a definition for #{what} on Google"
+      return
+    end
+    
+    head = results[0][0].ircify_html
+    text = results[0][1].ircify_html
+    link = results[0][2]
+    m.reply "#{head} -- #{link}\n#{text}"
+  end
+
   def wikipedia(m, params)
     lang = params[:lang]
     site = "#{lang.nil? ? '' : lang + '.'}wikipedia.org"
@@ -134,6 +177,16 @@ class SearchPlugin < Plugin
     params[:firstpar] = @bot.config['wikipedia.first_par']
     return google(m, params)
   end
+
+  def unpedia(m, params)
+    site = "uncyclopedia.org"
+    debug "Looking up things on #{site}"
+    params[:site] = site
+    params[:filter] = / - Uncyclopedia.*$/
+    params[:hits] = @bot.config['wikipedia.hits']
+    params[:firstpar] = @bot.config['wikipedia.first_par']
+    return google(m, params)
+  end
 end
 
 plugin = SearchPlugin.new
@@ -141,6 +194,8 @@ plugin = SearchPlugin.new
 plugin.map "search *words", :action => 'google'
 plugin.map "google *words", :action => 'google'
 plugin.map "gcalc *words", :action => 'gcalc'
+plugin.map "gdef *words", :action => 'gdef'
 plugin.map "wp :lang *words", :action => 'wikipedia', :requirements => { :lang => /^\w\w\w?$/ }
 plugin.map "wp *words", :action => 'wikipedia'
+plugin.map "unpedia *words", :action => 'unpedia'