卡夫卡流中的自定义时间戳提取器

Question

我正在尝试使用 kafka 流来处理来自 kafka 主题的一些数据。数据来自 kafka 0.11.0 写入的 kafka 主题，其中没有嵌入时间戳。在互联网上阅读一些内容后，我开始明白我可以通过在自定义 class 中扩展 TimestampExtractor class 并将其传递到 StreamsConfig 来解决这个问题。

我是这样做的--

class MyEventTimestampExtractor extends TimestampExtractor {
  override def extract(record: ConsumerRecord[AnyRef, AnyRef], prev: Long) = {
    record.value() match {
        case w: String => 1000L
        case _ => throw new RuntimeException(s"Called for $record")
    }
  }
}

我基于 this code on github

但是，当我执行 sbt run

时出现此错误

[error] /home/someuser/app/blahblah/src/main/scala/main.scala:34: class MyEventTimestampExtractor needs to be abstract, since method extract in trait TimestampExtractor of type (x: org.apache.kafka.clients.consumer.ConsumerRecord[Object,Object], x: Long)Long is not defined
[error] (Note that Long does not match Long)
[error] class MyEventTimestampExtractor extends TimestampExtractor {
[error]       ^
[error] /home/someuser/app/blahblah/src/main/scala/main.scala:35: method extract overrides nothing.
[error] Note: the super classes of class MyEventTimestampExtractor contain the following, non final members named extract:
[error] def extract(x: org.apache.kafka.clients.consumer.ConsumerRecord[Object,Object],x: Long): Long
[error]   override def extract(record: ConsumerRecord[AnyRef, AnyRef], prev: Long): Long = {
[error]                ^
[error] two errors found
[error] (compile:compileIncremental) Compilation failed

我的build.sbt文件是这个--

name := "kafka streams experiment"
version := "1.0"
scalaVersion := "2.12.4"

libraryDependencies ++= Seq(
  "org.apache.kafka" % "kafka-streams" % "1.0.0"
)

我不太明白这个错误。特别是 Note that Long does not match Long 周围的部分。我做错了什么？谢谢！

Answer 1

尝试使用（观察 prev 函数参数的类型：

  override def extract(record: ConsumerRecord[AnyRef, AnyRef], prev: java.lang.Long) = {

Answer 2

您需要 java.Long，因为此 API 是在 Java 中定义的，而您使用的是 Scala Long

卡夫卡流中的自定义时间戳提取器

custom timestamp extractor in kafka streams

apache-kafka

apache-kafka-streams

confluent-platform