Merge branch 'aqua_13_1_acu70514_binary_string_encoding'

.gitignore

@@ -1,5 +1,6 @@
 pkg
 *.rbc
 .redcar
+.idea
 doc/
 .yardoc/

Gemfile.lock

@@ -4,7 +4,7 @@ PATH
     cassandra (0.12.1)
       json
       rake
-      simple_uuid (>= 0.1.0)
+      simple_uuid (>= 0.2.0)
       thrift_client (>= 0.7.0)
 
 GEM
@@ -19,16 +19,16 @@ GEM
       rubyforge (>= 2.0.4)
     eventmachine (0.12.10)
     gemcutter (0.7.0)
-    json (1.5.3)
+    json (1.7.5)
     json_pure (1.5.3)
     rake (0.9.2)
     rdoc (3.8)
     rubyforge (2.0.4)
       json_pure (>= 1.1.7)
-    simple_uuid (0.2.0)
-    thrift (0.7.0)
-    thrift_client (0.7.0)
-      thrift (~> 0.7.0)
+    simple_uuid (0.3.0)
+    thrift (0.8.0)
+    thrift_client (0.8.2)
+      thrift (~> 0.8.0)
 
 PLATFORMS
   ruby

lib/cassandra/columns.rb

@@ -90,8 +90,8 @@ def _standard_insert_mutation(column_family, column_name, value, timestamp, ttl
       CassandraThrift::Mutation.new(
         :column_or_supercolumn => CassandraThrift::ColumnOrSuperColumn.new(
           :column => CassandraThrift::Column.new(
-            :name      => column_name_class(column_family).new(column_name).to_s,
-            :value     => value,
+            :name      => Cassandra::Helpers.convert_to_string_with_encoding(column_name_class(column_family).new(column_name)),
+            :value     => Cassandra::Helpers.convert_to_string_with_encoding(value),
             :timestamp => timestamp,
             :ttl       => ttl
           )

lib/cassandra/helpers.rb

@@ -1,5 +1,22 @@
 class Cassandra
   module Helpers
+    # TODO Aqua remove this helper and its callers when thrift 0.9.1
+    # is released with bugfixes about BINARY strings
+    def self.convert_to_string_with_encoding(object)
+      case object
+      when SimpleUUID::UUID
+        # Due to a bug in Thrift 0.9.0, we must pretend like
+        # UUID byte-vectors (BINARY encoded Strings) are
+        # UTF-8 encoded, otherwise thrift tries to transcode
+        # to UTF-8 which is not kosher.
+        object.to_s.force_encoding(Encoding::UTF_8)
+      else
+        # All other strings can be passed with natural encoding
+        # (UTF-8, or any encoding that can be transcoded to it).
+        object.to_s
+      end
+    end
+
     def extract_and_validate_params(column_family, keys, args, options)
       options = options.dup
       column_family = column_family.to_s
@@ -34,7 +51,7 @@ def s_map(el, klass)
       when Array then el.map { |i| s_map(i, klass) }
       when NilClass then nil
       else
-        klass.new(el).to_s
+        Cassandra::Helpers.convert_to_string_with_encoding(klass.new(el))
       end
     end
   end