Initial implementation of proper caching based on last-modified and etag HTTP headers

[user/henk/code/ruby/rbot.git] / data / rbot / plugins / search.rb

diff --git a/data/rbot/plugins/search.rb b/data/rbot/plugins/search.rb
index 8118b2638971231c443535872c1d26178d1ce0ba..a498d47fa29da06997864f9365af28ca780671ed 100644 (file)
--- a/data/rbot/plugins/search.rb
+++ b/data/rbot/plugins/search.rb
@@ -6,18 +6,36 @@ GOOGLE_WAP_LINK = /<a accesskey="(\d)" href=".*?u=(.*?)">(.*?)<\/a>/im
 
 class SearchPlugin < Plugin
   def help(plugin, topic="")
+    case topic
+    when "search"
+    "search <string> => search google for <string>"
+    when "google"
     "google <string> => search google for <string>"
+    when "wp"
+      "wp [<code>] <string> => search for <string> on Wikipedia. You can select a national <code> to only search the national Wikipedia"
+    else
+    "search <string> (or: google <string>) => search google for <string> | wp <string> => search for <string> 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=#{searchfor}"
+    url = "http://www.google.com/wml/search?q=#{site}#{searchfor}"
 
 
     begin
-      wml = @bot.httputil.get(url)
+      wml = @bot.httputil.get_cached(url)
     rescue => e
       m.reply "error googling for #{what}"
       return
@@ -28,15 +46,28 @@ class SearchPlugin < Plugin
       return
     end
     results = results[0...3].map { |res|
-      "#{res[0]}. #{Bold}#{Utils.decode_html_entities res[2].strip}#{Bold}: #{URI.unescape res[1].strip}"
+      n = res[0]
+      t = Utils.decode_html_entities res[2].gsub(filter, '').strip
+      u = URI.unescape res[1]
+      "#{n}. #{Bold}#{t}#{Bold}: #{u}"
     }.join(" | ")
 
     m.reply "Results for #{what}: #{results}"
   end
+
+  def wikipedia(m, params)
+    lang = params[:lang]
+    site = "#{lang.nil? ? '' : lang + '.'}wikipedia.org"
+    params[:site] = site
+    params[:filter] = / - Wikipedia.*$/
+    return google(m, params)
+  end
 end
 
 plugin = SearchPlugin.new
 
 plugin.map "search *words", :action => 'google'
 plugin.map "google *words", :action => 'google'
+plugin.map "wp :lang *words", :action => 'wikipedia', :requirements => { :lang => /^\w\w\w?$/ }
+plugin.map "wp *words", :action => 'wikipedia'