fix: Set charset of encoded structured events (#55)

Signed-off-by: Daniel Azuma <[email protected]>

Author: dazuma

lib/cloud_events/http_binding.rb

@@ -183,7 +183,10 @@ def encode_structured_content event, format, **format_args
       handlers = @structured_formatters[format] || []
       handlers.reverse_each do |handler|
         content = handler.encode event, **format_args
-        return [{ "Content-Type" => "application/cloudevents+#{format}" }, content] if content
+        if content
+          content_type = "application/cloudevents+#{format}; charset=#{charset_of content}"
+          return [{ "Content-Type" => content_type }, content]
+        end
       end
       raise HttpContentError, "Unknown cloudevents format: #{format.inspect}"
     end
@@ -206,7 +209,10 @@ def encode_batched_content events, format, **format_args
       handlers = @batched_formatters[format] || []
       handlers.reverse_each do |handler|
         content = handler.encode_batch events, **format_args
-        return [{ "Content-Type" => "application/cloudevents-batch+#{format}" }, content] if content
+        if content
+          content_type = "application/cloudevents-batch+#{format}; charset=#{charset_of content}"
+          return [{ "Content-Type" => content_type }, content]
+        end
       end
       raise HttpContentError, "Unknown cloudevents format: #{format.inspect}"
     end
@@ -254,7 +260,7 @@ def encode_binary_content event
     # @return [String] Resulting decoded string in UTF-8.
     #
     def percent_decode str
-      str = str.gsub(/"((?:[^"\\]|\\.)*)"/) { Regexp.last_match(1).gsub(/\\(.)/, '\1') }
+      str = str.gsub(/"((?:[^"\\]|\\.)*)"/) { ::Regexp.last_match(1).gsub(/\\(.)/, '\1') }
       decoded_str = str.gsub(/%[0-9a-fA-F]{2}/) { |m| [m[1..-1].to_i(16)].pack "C" }
       decoded_str.force_encoding ::Encoding::UTF_8
     end
@@ -293,18 +299,28 @@ def read_with_charset io, charset
         begin
           str.force_encoding charset
         rescue ::ArgumentError
-          # Do nothing for now if the charset is unrecognized
+          # Use binary for now if the charset is unrecognized
+          str.force_encoding ::Encoding::ASCII_8BIT
         end
       end
       str
     end
 
     def string_content_type str
-      if str.encoding == ::Encoding.ASCII_8BIT
+      if str.encoding == ::Encoding::ASCII_8BIT
         "application/octet-stream"
       else
         "text/plain; charset=#{str.encoding.name.downcase}"
       end
     end
+
+    def charset_of str
+      encoding = str.encoding
+      if encoding == ::Encoding::ASCII_8BIT
+        "binary"
+      else
+        encoding.name.downcase
+      end
+    end
   end
 end

test/test_http_binding.rb

@@ -85,7 +85,7 @@
     assert_equal my_subject, event.subject
     assert_equal my_time, event.time
     headers, body = http_binding.encode_batched_content [event], "json", sort: true
-    assert_equal({ "Content-Type" => "application/cloudevents-batch+json" }, headers)
+    assert_equal({ "Content-Type" => "application/cloudevents-batch+json; charset=utf-8" }, headers)
     assert_equal my_json_batch_encoded, body
   end
 
@@ -144,7 +144,7 @@
     assert_equal my_subject, event.subject
     assert_equal my_time, event.time
     headers, body = http_binding.encode_structured_content event, "json", sort: true
-    assert_equal({ "Content-Type" => "application/cloudevents+json" }, headers)
+    assert_equal({ "Content-Type" => "application/cloudevents+json; charset=utf-8" }, headers)
     assert_equal my_json_struct_encoded, body
   end
 
@@ -171,7 +171,7 @@
     assert_equal my_subject, event.subject
     assert_equal my_time, event.time
     headers, body = http_binding.encode_structured_content event, "json", sort: true
-    assert_equal({ "Content-Type" => "application/cloudevents+json" }, headers)
+    assert_equal({ "Content-Type" => "application/cloudevents+json; charset=utf-8" }, headers)
     assert_equal my_json_struct_encoded, body
   end