2

我正在寻找与Flink-SQL 中的 Spark-SQL 中存在的函数from_unixtime(bigint unixtime)的等效项。

我的目标是转换这种格式:1439799094

变成这种格式:2015-05-18 05:43:37

4

2 回答 2

1

只需使用 UDF!

https://ci.apache.org/projects/flink/flink-docs-stable/dev/table/udfs.html

示例使用

test.csv

creation_date|key
1535816823|1
1536392928|2
1536272308|3

EpochTimeConverter.scala

import java.time.format.DateTimeFormatter
import java.time.{Instant, LocalDateTime, ZoneId}

import org.apache.flink.table.functions.ScalarFunction

class EpochTimeConverter extends ScalarFunction {
  def eval(epochTime: Int): String = {
    // For performance, you may cache `DateTimeFormatter` in real life
    val timePattern = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss")
    timePattern.format(LocalDateTime.ofInstant(Instant.ofEpochSecond(epochTime), ZoneId.systemDefault()))
  }
}

UdfExample.scala

import org.apache.flink.api.scala.{ExecutionEnvironment, _}
import org.apache.flink.table.api.scala._
import org.apache.flink.table.api.{TableEnvironment, Types}
import org.apache.flink.table.sources.CsvTableSource
import org.apache.flink.types.Row

object UdfExample {
  def main(args: Array[String]): Unit = {
    val env = ExecutionEnvironment.getExecutionEnvironment
    val tableEnv = TableEnvironment.getTableEnvironment(env)
    val csvData = CsvTableSource
      .builder()
      .path("test.csv")
      .ignoreFirstLine()
      .fieldDelimiter("|")
      .field("creation_date", Types.INT)
      .field("key", Types.INT)
      .build()

    tableEnv.registerTableSource("temp_table", csvData)

    println("Without udf:")
    tableEnv.sqlQuery("SELECT creation_date, key FROM temp_table").toDataSet[Row].print()

    tableEnv.registerFunction("from_unixtime", new EpochTimeConverter())

    println()
    println("With udf:")
    tableEnv.sqlQuery("select from_unixtime(creation_date),key from temp_table").toDataSet[Row].print()
  }
}

如果你运行UdfExample.scala,它将产生类似的输出,如下所示:

Without udf:
1535816823,1
1536272308,3
1536392928,2

With udf:
2018-09-01 18:47:03,1
2018-09-07 01:18:28,3
2018-09-08 10:48:48,2
于 2019-01-29T22:10:10.420 回答
-1

我想你正在寻找DATE_FORMAT(timestamp, '%Y-%m-%d %H:%i:%s').

有关更多信息,请参阅有关内置函数日期格式说明符的文档。

于 2018-09-07T05:28:47.373 回答