Updated sender to try and get things into UTF8 where poss.

4 files changed, 146 insertions, 19 deletions

diff --git a/bin/mauvesend b/bin/mauvesend
index edb8e33..a61af49 100644
--- a/bin/mauvesend
+++ b/bin/mauvesend
@@ -318,8 +318,6 @@ begin
   error "Cannot send update -- not all libraries are available" if update.nil?
   error "No alerts specified" unless !update.alert.empty? || update.replace
 
-  update.transmission_id = rand(2**63)
-
   Mauve::Sender.new(ARGV).send(update, verbose)
 
 rescue ArgumentError => ae
diff --git a/debian/control b/debian/control
index 599ec3d..798f058 100644
--- a/debian/control
+++ b/debian/control
@@ -9,8 +9,9 @@ Standards-Version: 3.9.1
 Package: mauvealert-client
 Architecture: all
 Depends: ruby1.8,
-  mauvealert-common (>= 3.4.0),
+  mauvealert-common (>= 3.8.0),
   ${misc:Depends}
+Recommends: liblocale-ruby1.8 | ruby-locale
 Description: Mauve network alert system -- client
  Mauve is a network alert system for system and network administrators.  You
  can use it to quickly set up ad-hoc monitoring for a variety of services, and
@@ -22,7 +23,7 @@ Description: Mauve network alert system -- client
 Package: mauvealert-server
 Architecture: all
 Pre-Depends: libjs-jquery
-Depends: mauvealert-common (>= 3.4.0),
+Depends: mauvealert-common (>= 3.8.0),
  adduser,
  libdm-core-ruby1.8 (>= 1.2.0),
  libdm-migrations-ruby1.8 (>= 1.2.0),
diff --git a/lib/mauve/sender.rb b/lib/mauve/sender.rb
index c84dcb0..054313e 100644
--- a/lib/mauve/sender.rb
+++ b/lib/mauve/sender.rb
@@ -1,8 +1,21 @@
 # encoding: UTF-8
 require 'ipaddr'
 require 'socket'
+begin
+  require 'locale'
+rescue LoadError
+  # Do nothing -- these are bonus libraries :)
+end
+
+begin
+  require 'iconv'
+rescue LoadError
+  # Do nothing -- these are bonus libraries :)
+end
+
 require 'mauve/mauve_resolv'
 require 'mauve/mauve_time'
+require 'mauve/proto'
 
 module Mauve
   #
@@ -129,32 +142,68 @@ module Mauve
 
     end
 
-    # Send an update.
-    #
-    # @param [Mauve::Proto] update The update to send
-    # @param [Integer] vebose The verbosity -- higher is more.
+    # Sanitise all fields in an update, such that when we send, they are
+    # normal.
+    # 
     #
-    # @return [Integer] the number of packets sent.
-    def send(update, verbose=0)
-
+    def sanitize(update)
       #
       # Must have a source, so default to hostname if user doesn't care 
       update.source ||= Socket.gethostname
-      
+
+      #
+      # Check the locale charset.  This is to maximise the amout of information
+      # mauve receives, rather than provide proper sanitised data for the server.
+      #
+      from_charset = (Locale.current.charset || Locale.charset) if defined?(Locale)
+      from_charset ||= "UTF-8"
+
       #
-      # Make sure all alerts default to "-r now"
+      # 
       #
+      update.each_field do |field, value|
+        #
+        # Make sure all string fields are UTF8 -- to ensure the maximal amount of information is sent.
+        #
+        update.__send__("#{field.name}=", Iconv.conv("UTF-8//IGNORE", from_charset, value)) if value.is_a?(String) and defined?(Iconv)
+      end
+
       update.alert.each do |alert|
-        next if alert.raise_time || alert.clear_time
-        alert.raise_time = Time.now.to_i
+        #
+        # Make sure all alerts default to "-r now"
+        #
+        alert.raise_time = Time.now.to_i unless (alert.raise_time > 0 or alert.clear_time > 0)
+
+        alert.each_field do |field, value|
+          #
+          # Make sure all string fields are UTF8 -- to ensure the maximal amount of information is sent.
+          #
+          alert.__send__("#{field.name}=", Iconv.conv("UTF-8//IGNORE", from_charset, value)) if value.is_a?(String) and defined?(Iconv)
+        end
       end
-     
+
+      #
+      # Make sure we set the transmission time and ID.
+      #
+      update.transmission_time = Time.now.to_i if update.transmission_time.nil? or update.transmission_time == 0
+      update.transmission_id   = rand(2**63)   if update.transmission_id.nil? or update.transmission_id == 0
+
+      update
+    end
+
+    # Send an update.
+    #
+    # @param [Mauve::Proto] update The update to send
+    # @param [Integer] vebose The verbosity -- higher is more.
+    #
+    # @return [Integer] the number of packets sent.
+    def send(update, verbose=0)
       #
-      # Make sure we set the transmission time
+      # Clean up the update, and set any missing fields.
       #
-      update.transmission_time = Time.now.to_i 
+      update = sanitise(update)
 
-      data = update.serialize_to_string
+      data = sanitize(update).serialize_to_string
 
       if verbose == 1
         summary = "#{update.transmission_id} from #{update.source}"
diff --git a/test/tc_mauve_sender.rb b/test/tc_mauve_sender.rb
new file mode 100644
index 0000000..7db06c1
--- /dev/null
+++ b/test/tc_mauve_sender.rb
@@ -0,0 +1,79 @@
+$:.unshift "../lib"
+
+require 'th_mauve_resolv'
+require 'test/unit'
+require 'pp'
+require 'timecop'
+require 'mauve/sender'
+require 'locale'
+require 'iconv'
+
+
+class TcMauveSender < Test::Unit::TestCase 
+  include Mauve
+
+  def setup
+    Timecop.freeze(Time.local(2011,8,1,0,0,0,0))
+  end
+
+  def teardown
+    Timecop.return
+  end
+
+  def test_sanitise
+    Locale.clear
+    Locale.current = "en_GB.ISO-8859-1"
+    
+    #
+    # Set up a couple of crazy sources.
+    #
+    utf8_source = "Å ðîßtáñt plàñët"
+    iso88591_source = Iconv.conv(Locale.current.charset, "UTF-8", utf8_source)
+
+    #
+    # Make sure our two sources are distinct
+    #
+    assert(iso88591_source != utf8_source)
+
+    sender = Sender.new("test-1.example.com")
+    update = Mauve::Proto::AlertUpdate.new
+    update.source = iso88591_source
+    update.replace = false
+
+    alert = Mauve::Proto::Alert.new
+    update.alert << alert
+    
+    alert_cleared = Mauve::Proto::Alert.new
+    update.alert << alert_cleared
+    alert_cleared.clear_time = Time.now.to_i
+    
+    #
+    # Make sure the update has the correct source
+    #
+    assert_equal(iso88591_source, update.source)
+
+    #
+    # Sanitize
+    #
+    update = sender.sanitize(update)
+
+    #
+    # Now make sure the sanitization has changed it back to UTF-8
+    #
+    assert_equal(utf8_source, update.source)
+
+    #
+    # Now make sure the transmission time + id have been set
+    #
+    assert_equal(Time.now.to_i, update.transmission_time)
+    assert_kind_of(Integer, update.transmission_id)
+
+    #
+    # Make sure that the alert has its raise time set by default
+    #
+    assert_equal(Time.now.to_i, alert.raise_time)
+    assert_equal(0, alert_cleared.raise_time)
+  end
+
+
+end