使用Avro Java API,我可以创建一个简单的记录模式,如:
Schema schemaWithTimestamp = SchemaBuilder
.record("MyRecord").namespace("org.demo")
.fields()
.name("timestamp").type().longType().noDefault()
.endRecord();
如何使用逻辑类型标记架构字段,具体如下: https://avro.apache.org/docs/1.8.1/api/java/org/apache/avro/LogicalTypes.TimestampMillis.html
答案 0 :(得分:11)
感谢DontPanic:
Schema timestampMilliType = LogicalTypes.timestampMillis().addToSchema(Schema.create(Schema.Type.LONG));
Schema schemaWithTimestamp = SchemaBuilder
.record("MyRecord").namespace("org.demo")
.fields()
.name("timestamp_with_logical_type").type(timestampMilliType).noDefault()
.name("timestamp_no_logical_type").type().longType().noDefault()
.endRecord();
System.out.println(schemaWithTimestamp.toString(true));
这导致:
{
"type" : "record",
"name" : "MyRecord",
"namespace" : "org.demo",
"fields" : [ {
"name" : "timestamp_with_logical_type",
"type" : {
"type" : "long",
"logicalType" : "timestamp-millis"
}
}, {
"name" : "timestamp_no_logical_type",
"type" : "long"
} ]
}
答案 1 :(得分:2)
我认为您可以手动创建架构:
live == false
你的架构:
List<Schema.Field> fields = new ArrayList<>();
Schema timeStampField = Schema.create(Schema.Type.LONG);
fields.add(new Schema.Field("timestamp", LogicalTypes.timestampMillis().addToSchema(timeStampField), null, null));
Schema resultSchema = Schema.createRecord("MyRecord", null, "org.demo", false, fields);
System.out.println(resultSchema);
resultSchema with timestampMillis:
{"type":"record","name":"MyRecord","namespace":"org.demo","fields":[{"name":"timestamp","type":"long"}]}
答案 2 :(得分:0)
感谢第一个解决方案,现在提供了可为null的逻辑类型,例如:
{
"name":"maturityDate",
"type":["null", {
"type":"long","logicalType":"timestamp-millis"
}]
},
我认为:
Schema timestampMilliType = LogicalTypes.timestampMillis().addToSchema(Schema.create(Schema.Type.LONG));
Schema clientIdentifier = SchemaBuilder.record("ClientIdentifier")
.namespace("com.baeldung.avro")
.fields()
.requiredString("hostName")
.requiredString("ipAddress")
.name("maturityDate")
.type()
.unionOf()
.nullType()
.and()
.type(timestampMilliType)
.endUnion()
.noDefault()
.endRecord();