Plugin header boilerplating.

[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 c44100b1abaa106ba5367f683223b599200906ff..3b7d682684f7a6f3be3dfcd9a3e6f55858c4ded0 100644 (file)
--- a/data/rbot/plugins/search.rb
+++ b/data/rbot/plugins/search.rb
@@ -1,3 +1,18 @@
+#-- vim:sw=2:et
+#++
+#
+# :title: Google and Wikipedia search plugin for rbot
+#
+# Author:: Tom Gilbert (giblet) <tom@linuxbrit.co.uk>
+# Author:: Giuseppe "Oblomov" Bilotta <giuseppe.bilotta@gmail.com>
+#
+# Copyright:: (C) 2002-2005 Tom Gilbert
+# Copyright:: (C) 2006 Tom Gilbert, Giuseppe Bilotta
+# Copyright:: (C) 2006-2007 Giuseppe Bilotta
+
+# 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
+
 require 'uri'
 
 Net::HTTP.version_1_2
@@ -5,33 +20,49 @@ Net::HTTP.version_1_2
 GOOGLE_WAP_LINK = /<a accesskey="(\d)" href=".*?u=(.*?)">(.*?)<\/a>/im
 
 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"
-    "search <string> => search google for <string>"
-    when "google"
-    "google <string> => search google for <string>"
+    when "search", "google"
+      "#{topic} <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"
+      "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=#{site}#{searchfor}"
 
+    hits = params[:hits] || @bot.config['google.hits']
 
     begin
-      wml = @bot.httputil.get(url)
+      wml = @bot.httputil.get_cached(url)
     rescue => e
       m.reply "error googling for #{what}"
       return
@@ -41,17 +72,33 @@ class SearchPlugin < Plugin
       m.reply "no results found for #{what}"
       return
     end
-    results = results[0...3].map { |res|
-      "#{res[0]}. #{Bold}#{Utils.decode_html_entities res[2].strip}#{Bold}: #{URI.unescape res[1].strip}"
+    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}"
+    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, :http_util => @bot.httputil, :message => m
+
   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