search: improve ddg output with missing sources

[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 3a12d30c949e783be05740329e569d5b0f812d78..a4c4a89c4aec0388fa697d80561fd40b51b79cd5 100644 (file)
--- a/data/rbot/plugins/search.rb
+++ b/data/rbot/plugins/search.rb
@@ -25,6 +25,11 @@ GOOGLE_TIME_RESULT = %r{alt="Clock"></td><td valign=[^>]+>(.+?)<(br|/td)>}
 
 DDG_API_SEARCH = "http://api.duckduckgo.com/?format=xml&no_html=1&skip_disambig=1&no_redirect=0&q="
 
+WOLFRAM_API_SEARCH = "http://api.wolframalpha.com/v2/query?input=%{terms}&appid=%{key}&format=plaintext"
+           "&scantimeout=3.0&podtimeout=4.0&formattimeout=8.0&parsetimeout=5.0"
+           "&excludepodid=SeriesRepresentations:*"
+WOLFRAM_API_KEY = "4EU37Y-TX9WJG3JH3"
+
 class SearchPlugin < Plugin
   Config.register Config::IntegerValue.new('duckduckgo.hits',
     :default => 3, :validate => Proc.new{|v| v > 0},
@@ -58,12 +63,14 @@ class SearchPlugin < Plugin
       "gdef <term(s)> => use the google define mechanism to find a definition of <term(s)>"
     when "gtime"
       "gtime <location> => use the google clock to find the current time at <location>"
+    when "wa"
+      "wa <string> => searches WolframAlpha for <string>"
     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> | ddg <string> to search DuckDuckGo | wp <string> => search for <string> on Wikipedia | unpedia <string> => search for <string> on Uncyclopedia"
+      "search <string> (or: google <string>) => search google for <string> | ddg <string> to search DuckDuckGo | wp <string> => search for <string> on Wikipedia | wa <string> => search for <string> on WolframAlpha | unpedia <string> => search for <string> on Uncyclopedia"
     end
   end
 
@@ -91,15 +98,28 @@ class SearchPlugin < Plugin
     answer = xml.elements['//Answer/text()'].to_s
     # abstract is returned for definitions etc
     abstract = xml.elements['//AbstractText/text()'].to_s
+    abfrom = ""
     unless abstract.empty?
-      absrc = xml.elements['//AbstractSource/text()']
-      aburl = xml.elements['//AbstractURL/text()']
+      absrc = xml.elements['//AbstractSource/text()'].to_s
+      aburl = xml.elements['//AbstractURL/text()'].to_s
+      unless absrc.empty? and aburl.empty?
+        abfrom = " --"
+        abfrom << " " << absrc unless absrc.empty?
+        abfrom << " " << aburl unless aburl.empty?
+      end
     end
+
     # but also definition (yes, you can have both, see e.g. printf)
     definition = xml.elements['//Definition/text()'].to_s
+    deffrom = ""
     unless definition.empty?
       defsrc = xml.elements['//Definition/@source/text()'].to_s
       defurl = xml.elements['//Definition/@url/text()'].to_s
+      unless defsrc.empty? and defurl.empty?
+        deffrom = " --"
+        deffrom << " " << defsrc unless defsrc.empty?
+        deffrom << " " << defurl unless defurl.empty?
+      end
     end
 
     if heading.empty? and answer.empty? and abstract.empty? and definition.empty?
@@ -115,15 +135,15 @@ class SearchPlugin < Plugin
 
     # otherwise, return the abstract, followed by as many hits as found
     unless heading.empty? or abstract.empty?
-      m.reply "%{bold}%{heading}:%{bold} %{abstract} -- %{absrc} %{aburl}" % {
+      m.reply "%{bold}%{heading}:%{bold} %{abstract}%{abfrom}" % {
         :bold => Bold, :heading => heading,
-        :abstract => abstract, :absrc => absrc, :aburl => aburl
+        :abstract => abstract, :abfrom => abfrom
       }
     end
     unless heading.empty? or definition.empty?
-      m.reply "%{bold}%{heading}:%{bold} %{abstract} -- %{absrc} %{aburl}" % {
+      m.reply "%{bold}%{heading}:%{bold} %{abstract}%{abfrom}" % {
         :bold => Bold, :heading => heading,
-        :abstract => definition, :absrc => defsrc, :aburl => defurl
+        :abstract => definition, :abfrom => deffrom
       }
     end
     # return zeroclick search results
@@ -386,6 +406,42 @@ class SearchPlugin < Plugin
     m.reply "#{head} -- #{text}"
   end
 
+  def wolfram(m, params)
+    what = params[:words].to_s
+    terms = CGI.escape what
+    url = WOLFRAM_API_SEARCH % {
+      :terms => terms, :key => WOLFRAM_API_KEY
+    }
+
+    begin
+      feed = @bot.httputil.get(url)
+      raise unless feed
+    rescue => e
+      m.reply "error asking WolframAlfa about #{what}"
+      return
+    end
+    debug feed
+
+    xml = REXML::Document.new feed
+    if xml.elements['/queryresult'].attributes['error'] == "true"
+      m.reply xml.elements['/queryresult/error/text()'].to_s
+      return
+    end
+    unless xml.elements['/queryresult'].attributes['success'] == "true"
+      m.reply "no data available"
+      return
+    end
+    answer = []
+    xml.elements.each("//pod/subpod/plaintext") { |element|
+      answer << element.text
+    }
+    # strip spaces and line breaks
+    answer[1].gsub!(/\n/, Bold + ' :: ' + Bold )
+    answer[1].gsub!(/\t/, ' ')
+    answer[1].gsub!(/\s+/, ' ')
+    m.reply answer[1].to_s
+  end
+
   def wikipedia(m, params)
     lang = params[:lang]
     site = "#{lang.nil? ? '' : lang + '.'}wikipedia.org"
@@ -436,7 +492,7 @@ end
 
 plugin = SearchPlugin.new
 
-plugin.map "ddg *words", :action => 'duckduckgo'
+plugin.map "ddg *words", :action => 'duckduckgo', :threaded => true
 plugin.map "search *words", :action => 'google', :threaded => true
 plugin.map "google *words", :action => 'google', :threaded => true
 plugin.map "lucky *words", :action => 'lucky', :threaded => true
@@ -444,6 +500,7 @@ plugin.map "gcount *words", :action => 'gcount', :threaded => true
 plugin.map "gcalc *words", :action => 'gcalc', :threaded => true
 plugin.map "gdef *words", :action => 'gdef', :threaded => true
 plugin.map "gtime *words", :action => 'gtime', :threaded => true
+plugin.map "wa *words", :action => 'wolfram', :threaded => true
 plugin.map "wp :lang *words", :action => 'wikipedia', :requirements => { :lang => /^\w\w\w?$/ }, :threaded => true
 plugin.map "wp *words", :action => 'wikipedia', :threaded => true
 plugin.map "unpedia *words", :action => 'unpedia', :threaded => true