Fail when extra fields are in the payload

lib/avro-patches/logical_types/schema_validator.rb

@@ -1,27 +1,29 @@
 module AvroPatches
   module LogicalTypes
     module SchemaValidatorPatch
-      def validate!(expected_schema, logical_datum, options = { recursive: true, encoded: false })
+      def validate!(expected_schema, logical_datum, options = { recursive: true, encoded: false, fail_on_extra_fields: false})
         options ||= {}
         options[:recursive] = true unless options.key?(:recursive)
 
         result = Avro::SchemaValidator::Result.new
         if options[:recursive]
-          validate_recursive(expected_schema, logical_datum, Avro::SchemaValidator::ROOT_IDENTIFIER, result, options[:encoded])
+          validate_recursive(expected_schema, logical_datum,
+                             Avro::SchemaValidator::ROOT_IDENTIFIER, result, options)
         else
-          validate_simple(expected_schema, logical_datum, Avro::SchemaValidator::ROOT_IDENTIFIER, result, options[:encoded])
+          validate_simple(expected_schema, logical_datum,
+                          Avro::SchemaValidator::ROOT_IDENTIFIER, result, options)
         end
         fail Avro::SchemaValidator::ValidationError, result if result.failure?
         result
       end
 
       private
 
-      def validate_recursive(expected_schema, logical_datum, path, result, encoded = false)
-        datum = resolve_datum(expected_schema, logical_datum, encoded)
+      def validate_recursive(expected_schema, logical_datum, path, result, options={})
+        datum = resolve_datum(expected_schema, logical_datum, options[:encoded])
 
         # The entire method is overridden so that encoded: true can be passed here
-        validate_simple(expected_schema, datum, path, result, true)
+        validate_simple(expected_schema, datum, path, result, encoded: true)
 
         case expected_schema.type_sym
         when :array
@@ -36,13 +38,20 @@ def validate_recursive(expected_schema, logical_datum, path, result, encoded = f
             deeper_path = deeper_path_for_hash(field.name, path)
             validate_recursive(field.type, datum[field.name], deeper_path, result)
           end
+          if options[:fail_on_extra_fields]
+            datum_fields = datum.keys.map(&:to_s)
+            schema_fields = expected_schema.fields.map(&:name)
+            (datum_fields - schema_fields).each do |extra_field|
+              result.add_error(path, "extra field '#{extra_field}' - not in schema")
+            end
+          end
         end
       rescue Avro::SchemaValidator::TypeMismatchError
         result.add_error(path, "expected type #{expected_schema.type_sym}, got #{actual_value_message(datum)}")
       end
 
-      def validate_simple(expected_schema, logical_datum, path, result, encoded = false)
-        datum = resolve_datum(expected_schema, logical_datum, encoded)
+      def validate_simple(expected_schema, logical_datum, path, result, options = {})
+        datum = resolve_datum(expected_schema, logical_datum, options[:encoded])
         super(expected_schema, datum, path, result)
       end
 

test/schema_validator/test_schema_validator.rb

@@ -17,8 +17,8 @@
 require 'test_help'
 
 class TestSchema < Test::Unit::TestCase
-  def validate!(schema, value)
-    Avro::SchemaValidator.validate!(schema, value)
+  def validate!(schema, value, options=nil)
+    Avro::SchemaValidator.validate!(schema, value, options)
   end
 
   def validate_simple!(schema, value)
@@ -469,4 +469,24 @@ def test_validate_multiple_errors
       exception.to_s
     )
   end
+
+  def test_validate_extra_fields
+    schema = hash_to_schema(
+               type: 'record',
+               name: 'fruits',
+               fields: [
+                 {
+                   name: 'veggies',
+                   type: 'string'
+                 }
+                     ]
+    )
+    exception = assert_raise(Avro::SchemaValidator::ValidationError) do
+      validate!(schema, {'veggies' => 'tomato', 'bread' => 'rye'}, fail_on_extra_fields: true)
+    end
+    assert_equal(1, exception.result.errors.size)
+    assert_equal("at . extra field 'bread' - not in schema",
+                 exception.to_s)
+  end
+
 end