Raise error on non utf-8 string in 1.8

lib/mongo/errors.rb

@@ -27,6 +27,9 @@ class ConfigurationError < MongoRubyError; end
   # Raised when invalid arguments are sent to Mongo Ruby methods.
   class MongoArgumentError < MongoRubyError; end
 
+  # Raised when given a string is not valid utf-8 (Ruby 1.8 only).
+  class InvalidStringEncoding < MongoRubyError; end
+
   # Raised on failures in connection to the database server.
   class ConnectionError < MongoRubyError; end
 

lib/mongo/util/bson.rb

@@ -55,7 +55,12 @@ def self.to_utf8(str)
     end
   else
     def self.to_utf8(str)
-      str                       # TODO Ruby 1.8 punt for now
+      begin
+      str.unpack("U*")
+      rescue => ex
+        raise InvalidStringEncoding, "String not valid utf-8: #{str}"
+      end
+      str
     end
   end
 

test/test_bson.rb

@@ -1,6 +1,7 @@
 $LOAD_PATH[0,0] = File.join(File.dirname(__FILE__), '..', 'lib')
 require 'mongo'
 require 'mongo/util/ordered_hash'
+require 'iconv'
 require 'test/unit'
 
 class BSONTest < Test::Unit::TestCase
@@ -20,6 +21,20 @@ def test_string
     assert_equal doc, @b.deserialize
   end
 
+  def test_valid_utf8_string
+    doc = {'doc' => "aéあ"}
+    @b.serialize(doc)
+    assert_equal doc, @b.deserialize
+  end
+
+  def test_invalid_string
+    string = Iconv.conv('iso-8859-1', 'utf-8', 'aé').first
+    doc = {'doc' => string} 
+    assert_raise InvalidStringEncoding do
+      @b.serialize(doc)
+    end
+  end
+
   def test_code
     doc = {'$where' => Code.new('this.a.b < this.b')}
     @b.serialize(doc)