X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fsearch.rb;h=f54c0df6bbadf558a8627783177315c22d0d674b;hb=95fab091ab9e2b42a6e8cfda72b231ef06971e39;hp=8b3b6848d11800084a22ac93a9b64d3c13802207;hpb=d372ec5babdbf5e6a9c49b21d16bcae74d0ff01c;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/search.rb b/data/rbot/plugins/search.rb
index 8b3b6848..f54c0df6 100644
--- a/data/rbot/plugins/search.rb
+++ b/data/rbot/plugins/search.rb
@@ -17,11 +17,10 @@
 
 GOOGLE_SEARCH = "http://www.google.com/search?oe=UTF-8&q="
 GOOGLE_WAP_SEARCH = "http://www.google.com/m/search?hl=en&q="
-# GOOGLE_WAP_LINK = /<a accesskey="(\d)" href=".*?u=(.*?)">(.*?)<\/a>/im
-GOOGLE_WAP_LINK = /<a href="(?:.*?u=(.*?)|(http:\/\/.*?))">(.*?)<\/a>/im
-GOOGLE_CALC_RESULT = %r{<img src=/images/calc_img\.gif(?: width=40 height=30 alt="")?>.*?<h2 class=r[^>]*><b>(.+?)</b>}
+GOOGLE_WAP_LINK = /result">(?:<div[^>]*>)?<a href="([^"]+)"[^>]*>(.*?)<\/a>/im
+GOOGLE_CALC_RESULT = %r{<img src=/images/calc_img\.gif(?: width=40 height=30 alt="")?>.*?<h[1-6] class=r[^>]*><b>(.+?)</b>}
 GOOGLE_COUNT_RESULT = %r{<font size=-1>Results <b>1<\/b> - <b>10<\/b> of about <b>(.*)<\/b> for}
-GOOGLE_DEF_RESULT = %r{<a href="([^"]*)"[^>]*>(Web definitions for .*?)<br/>(.*?)<br/>(.*?)\s-\s+<a href}
+GOOGLE_DEF_RESULT = %r{onebox_result">\s*(.*?)\s*<br/>\s*(.*?)<table}
 GOOGLE_TIME_RESULT = %r{alt="Clock"></td><td valign=[^>]+>(.+?)<(br|/td)>}
 
 class SearchPlugin < Plugin
@@ -90,7 +89,7 @@ class SearchPlugin < Plugin
       m.reply "error googling for #{what}"
       return
     end
-    results = wml.match('<p align="center">').pre_match.scan(GOOGLE_WAP_LINK)
+    results = wml.scan(GOOGLE_WAP_LINK)
 
     if results.length == 0
       m.reply "no results found for #{what}"
@@ -98,27 +97,46 @@ class SearchPlugin < Plugin
     end
 
     single ||= (results.length==1)
+    pretty = []
 
-    urls = Array.new
-    n = 0
-    results = results[0...hits].map { |res|
-      n += 1
-      t = res[2].ircify_html(:img => "[%{src} %{alt} %{dimensions}]").strip
-      u = URI.unescape(res[0] || res[1])
-      urls.push(u)
-      "%{n}%{b}%{t}%{b}%{sep}%{u}" % {
-        :n => (single ? "" : "#{n}. "),
-        :sep => (single ? " -- " : ": "),
-        :b => Bold, :t => t, :u => u
-      }
-    }
+    begin
+      urls = Array.new
+
+      debug results
+      results.each do |res|
+        t = res[1].ircify_html(:img => "[%{src} %{alt} %{dimensions}]").strip
+        u = res[0]
+        if u.sub!(%r{^http://www.google.com/aclk\?},'')
+          u = CGI::parse(u)['adurl'].first
+          debug "skipping ad for #{u}"
+          next
+        elsif u.sub!(%r{^http://www.google.com/gwt/x\?},'')
+          u = CGI::parse(u)['u'].first
+        elsif u.sub!(%r{^/url\?},'')
+          u = CGI::parse(u)['q'].first
+        end
+        urls.push(u)
+        pretty.push("%{n}%{b}%{t}%{b}%{sep}%{u}" % {
+          :n => (single ? "" : "#{urls.length}. "),
+          :sep => (single ? " -- " : ": "),
+          :b => Bold, :t => t, :u => u
+        })
+        break if urls.length == hits
+      end
+    rescue => e
+      m.reply "failed to understand what google found for #{what}"
+      error e
+      debug wml
+      debug results
+      return
+    end
 
     if params[:lucky]
-      m.reply results.first
+      m.reply pretty.first
       return
     end
 
-    result_string = results.join(" | ")
+    result_string = pretty.join(" | ")
 
     # If we return a single, full result, change the output to a more compact representation
     if single
@@ -172,7 +190,7 @@ class SearchPlugin < Plugin
     searchfor = CGI.escape(what)
 
     debug "Getting gcalc thing: #{searchfor.inspect}"
-    url = GOOGLE_SEARCH + searchfor
+    url = GOOGLE_WAP_SEARCH + searchfor
 
     begin
       html = @bot.httputil.get(url)
@@ -182,18 +200,19 @@ class SearchPlugin < Plugin
     end
 
     debug "#{html.size} bytes of html recieved"
+    debug html
 
-    results = html.scan(GOOGLE_CALC_RESULT)
-    debug "results: #{results.inspect}"
+    candidates = html.match(/font-weight:bold">(.*?)<\/(?:span|div)>/)
+    debug "candidates: #{candidates.inspect}"
 
-    if results.length != 1
+    if candidates.nil?
       m.reply "couldn't calculate #{what}"
       return
     end
+    result = candidates[1]
 
-    result = results[0][0].ircify_html
     debug "replying with: #{result.inspect}"
-    m.reply "#{result}"
+    m.reply result.ircify_html
   end
 
   def gcount(m, params)
@@ -249,21 +268,9 @@ class SearchPlugin < Plugin
       return
     end
 
-    gdef_link = "http://www.google.com" + CGI.unescapeHTML(results[0][0]) # could be used to extract all defs
-    head = results[0][1].ircify_html
-    text = results[0][2].ircify_html
-    link = results[0][3]
-    m.reply "#{head} -- #{link}\n#{text}"
-
-    ### gdef_link could be used for something like
-    # html_defs = @bot.httputil.get(gdef_link)
-    # related_index = html_defs.index(/Related phrases:/, 0)
-    # defs_index = html_defs.index(/Definitions of <b>/, related_index)
-
-    # related = html_defs[related_index..defs_index]
-    # defs = html_defs[defs_index..-1]
-
-    # m.reply defs.gsub('  <br/>','<li>').ircify_html
+    head = results[0][0].ircify_html
+    text = results[0][1].ircify_html
+    m.reply "#{head} -- #{text}"
   end
 
   def wikipedia(m, params)