seen: Ignore garbage phrases as a users last words.

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

diff --git a/data/rbot/plugins/url.rb b/data/rbot/plugins/url.rb
index ad8951212362b6c26bcad57403c2b73412cfaf38..8327dd945d169c0d1c3282e547fef01d08f1c32e 100644 (file)
--- a/data/rbot/plugins/url.rb
+++ b/data/rbot/plugins/url.rb
@@ -15,6 +15,12 @@ class UrlPlugin < Plugin
   Config.register Config::IntegerValue.new('url.display_link_info',
     :default => 0,
     :desc => "Get the title of links pasted to the channel and display it (also tells if the link is broken or the site is down). Do it for at most this many links per line (set to 0 to disable)")
+  Config.register Config::BooleanValue.new('url.auto_shorten',
+    :default => false,
+    :desc => "Automatically spit out shortened URLs when they're seen. Check shortenurls for config options")
+  Config.register Config::IntegerValue.new('url.auto_shorten_min_length',
+    :default => 48,
+    :desc => "Minimum length of URL to auto-shorten.  Only has an effect when url.auto_shorten is true.")
   Config.register Config::BooleanValue.new('url.titles_only',
     :default => false,
     :desc => "Only show info for links that have <title> tags (in other words, don't display info for jpegs, mpegs, etc.)")
@@ -88,6 +94,7 @@ class UrlPlugin < Plugin
       debug "+ getting info for #{url.request_uri}"
       info = @bot.filter(:htmlinfo, url)
       debug info
+      logopts[:htmlinfo] = info
       resp = info[:headers]
 
       logopts[:title] = title = info[:title]
@@ -152,6 +159,11 @@ class UrlPlugin < Plugin
     urls.each do |urlstr|
       debug "working on #{urlstr}"
       next unless urlstr =~ /^https?:\/\/./
+      if @bot.config['url.auto_shorten'] == true and
+         urlstr.length >= @bot.config['url.auto_shorten_min_length']
+        m.reply(bot.plugins['shortenurls'].shorten(nil, {:url=>urlstr, :called=>true}))
+        next
+      end
       title = nil
       debug "Getting title for #{urlstr}..."
       reply = nil
@@ -169,9 +181,17 @@ class UrlPlugin < Plugin
         # with the last character stripped. this might generate invalid URIs
         # (e.g. because "some.url" gets chopped to some.url%2, so catch that too
         if e.message =~ /\(404 - Not Found\)/i or e.kind_of?(URI::InvalidURIError)
-          # chop off last character, and retry if we still have enough string to
-          # look like a minimal URL
-          retry if urlstr.chop! and urlstr =~ /^https?:\/\/./
+          # chop off last non-word character from the unescaped version of
+          # the URL, and retry if we still have enough string to look like a
+          # minimal URL
+          unescaped = URI.unescape(urlstr)
+          debug "Unescaped: #{unescaped}"
+          if unescaped.sub!(/\W$/,'') and unescaped =~ /^https?:\/\/./
+            urlstr.replace URI.escape(unescaped, OUR_UNSAFE)
+            retry
+          else
+            debug "Not retrying #{unescaped}"
+          end
         end
         reply = "Error #{e.message}"
       end