自定义一个函数实现查询字符串长度。首先创建测试的DataFrame:
val spark = SparkSession.builder().master("local").appName("UDF").getOrCreate()
val nameList: List[String] = List[String]("zhangsan", "lisi", "wangwu", "zhaoliu", "tianqi")
import spark.implicits._
val nameDF: DataFrame = nameList.toDF("name")
nameDF.createOrReplaceTempView("students")
注册函数,包含函数名和参数类型:
spark.udf.register("STRLEN",(name:String)=>{
name.length
})
// spark.udf.register("STRLEN",(name:String,i:Int)=>{
// name.length+i
// })
使用自定义函数:
spark.sql("select name ,STRLEN(name) as length from students order by length desc").show(100)
// spark.sql("select name ,STRLEN(name,10) as length from students order by length desc").show(100)