fluent-plugin-kafka-enchanced 0.5.31 → 0.5.32

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 126ac61c90d73c137a2617876ee242a394659b14
-  data.tar.gz: fc9180ce1429f3aa2af0a4aea3eb12e9254b58cd
+  metadata.gz: 33fd86ef94e5d7e9340ed79329dff51ea2f32a4a
+  data.tar.gz: 1caa6add35ca52ee497068a8a68f77e71f8aff76
 SHA512:
-  metadata.gz: 6d2c68b75b7a6b19bed5f1b4dc7fde749ab7355632433c9a982c2daa43e2b554aa40c29ccfef903e3071ace576680e06424b2303ae325201d807ae3b55f96a49
-  data.tar.gz: f9fe60aa46e4be239a965d58063355823c12c046353226723b9edc93ab58f366b7a16800d61f1a28244dcc4b1cd7351c61a56d9da5a2fcff35b314ee3327d5c6
+  metadata.gz: 775560dc0091e06d95bbf45c9658c66834e774bf0c019da2f30d0e9ac25b644883568e09dba118ddb5a58baae262f41bb7af52ba6345e4b7d512df254913548d
+  data.tar.gz: 7eba27507c8ee929ea05dba4c845b995717429585d5b3b785b8cd4511353899688f484b3103371b26a94604b4d342b89d1c2fcd2731ddf90dc56466426b89b3d

data/fluent-plugin-kafka.gemspec

@@ -12,7 +12,7 @@ Gem::Specification.new do |gem|
   gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
   gem.name          = "fluent-plugin-kafka-enchanced"
   gem.require_paths = ["lib"]
-  gem.version       = '0.5.31'
+  gem.version       = '0.5.32'
   gem.required_ruby_version = ">= 2.1.0"
 
   gem.add_dependency "fluentd", [">= 0.10.58", "< 2"]

data/lib/fluent/plugin/out_kafka.rb

@@ -180,7 +180,7 @@ DESC
           schema_id = registry.register("#{schema_name}-value", schema)
 
           stored_schema = {
-              'schema' => schema,
+              'schema_json' => schema_json,
               'schema_id' => schema_id,
               'field_types' => field_types
           }

data/lib/fluent/plugin/out_kafka_buffered.rb

@@ -202,23 +202,52 @@ DESC
       require "avro/builder"
       init_redis
       Proc.new do |tag, time, record|
-        record = record.select{|key, value| !key.nil? && !key.empty?}.map{|k, v| [k.tr('[]-', '_').delete('$'), ((!v.is_a?(Fixnum) && !v.is_a?(Float)) ? v.to_s.force_encoding("UTF-8") : v)]}.to_h
+        record = record.select{|key, value| !key.nil? && !key.empty?}.map do |k, v|
+          [
+              k.tr('[]-', '_').delete('$'),
+              (v.is_a?(Fixnum) || v.is_a?(Float) || v.nil? ? v : v.to_s.force_encoding("UTF-8"))
+          ]
+        end.to_h
         timestamp = Time.new
         record['enchilada_time_with_format'] = timestamp.strftime("%Y-%m-%dT%H:%M:%S.%LZ")
+        @topic_name = schema_name = "#{tag.to_s.tr('.$:', '_')}_#{Digest::MD5.new.hexdigest(record.keys.to_s)[0..5]}"
+
+        avro = AvroTurf::Messaging.new(registry_url: @schema_registry)
+
+        unless (stored_schema = get_schema_from_redis_by_name(schema_name))
+          fields = record.map do |key, value|
+            {
+                'name' => key,
+                'type' => ['null', (value.is_a?(Fixnum) ? 'int' : (value.is_a?(Float) ? 'float' : 'string'))]
+            }
+          end
+          field_types = fields.map{|field| [field['name'], (field['type'] - ['null']).first]}.to_h
+          fields << {"name" => "enchilada_timestamp", "type" => "long"}
+          schema_json = {
+              "type": "record",
+              "name": schema_name,
+              "fields": fields
+          }.to_json
+          registry = avro.instance_variable_get('@registry')
+          schema = Avro::Schema.parse(schema_json)
+          schema_id = registry.register("#{schema_name}-value", schema)
+
+          stored_schema = {
+              'schema_json' => schema_json,
+              'schema_id' => schema_id,
+              'field_types' => field_types
+          }
+
+          set_schema_to_redis(schema_name, stored_schema)
+
+        end
 
-        fields = record.map{|key, value| {'name' => key, 'type' => (value.is_a?(Fixnum) ? 'int' : (value.is_a?(Float) ? 'float' : 'string'))}}
         record['enchilada_timestamp'] = timestamp.strftime('%s%3N').to_i
-        fields << {"name" => "enchilada_timestamp", "type" => "long"}
-        @topic_name = schema_name = "#{tag.to_s.tr('.$:', '_')}_#{Digest::MD5.new.hexdigest(fields.to_s)[0..5]}"
-        schema_json = {
-            "type": "record",
-            "name": schema_name,
-            "fields": fields
-        }.to_json
-        schema = Avro::Schema.parse(schema_json)
+        record = record.map do |key, val|
+          [key, (stored_schema['field_types'][key] != 'string' || val.nil? ? val : val.to_s)]
+        end.to_h
 
-        avro = AvroTurf::Messaging.new(registry_url: @schema_registry)
-        avro.encode(record, schema: schema, subject: "#{schema_name}-value")
+        avro.encode(record, stored_schema['schema_id'], schema: Avro::Schema.parse(stored_schema['schema_json']))
       end
     elsif @output_data_type =~ /^attr:(.*)$/
       @custom_attributes = $1.split(',').map(&:strip).reject(&:empty?)
@@ -318,7 +347,7 @@ DESC
       parsed_schema = JSON.parse($redis.get(schema_name))
       {
           'schema_id' => parsed_schema['schema_id'],
-          'schema' => Avro::Schema.parse(parsed_schema['schema']),
+          'schema_json' => Avro::Schema.parse(parsed_schema['schema_json']),
           'field_types' => parsed_schema['field_types']
       }
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: fluent-plugin-kafka-enchanced
 version: !ruby/object:Gem::Version
-  version: 0.5.31
+  version: 0.5.32
 platform: ruby
 authors:
 - Paschenko Konstantin
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-03-29 00:00:00.000000000 Z
+date: 2017-03-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: fluentd