在开始正式数据处理之前，我觉得有必要去学习理解下UDF。

UDF

UDF全称User-Defined Functions，用户自定义函数，是Spark SQL的一项功能，用于定义新的基于列的函数，这些函数扩展了Spark SQL的DSL用于转换数据集的词汇表。

我在databricks上找到一个比较简单理解的入门栗子：

Register the function as a UDF

val squared = (s: Int) => {
  s * s
}
spark.udf.register("square", squared)

Call the UDF in Spark SQL

spark.range(1, 20).registerTempTable("test")
%sql select id, square(id) as id_squared from test

我理解就是先定义一个函数squared，返回输入数字的平方，然后register，并绑定square方法名为square，然后就在Spark SQL中直接使用square方法。

实例一：温度转化

import org.apache.spark.sql.SparkSession
import org.apache.spark.SparkConf

object ScalaUDFExample {
  def main(args: Array[String]) {
    val conf       = new SparkConf().setAppName("Scala UDF Example")
    val spark      = SparkSession.builder().enableHiveSupport().config(conf).getOrCreate() 

    val ds = spark.read.json("temperatures.json")
    ds.createOrReplaceTempView("citytemps")

    // Register the UDF with our SparkSession     spark.udf.register("CTOF", (degreesCelcius: Double) => ((degreesCelcius * 9.0 / 5.0) + 32.0))

    spark.sql("SELECT city, CTOF(avgLow) AS avgLowF, CTOF(avgHigh) AS avgHighF FROM citytemps").show()
  }
}

我们将定义一个 UDF 来将以下 JSON 数据中的温度从摄氏度（degrees Celsius）转换为华氏度（degrees Fahrenheit）：

{"city":"St. John's","avgHigh":8.7,"avgLow":0.6}
{"city":"Charlottetown","avgHigh":9.7,"avgLow":0.9}
{"city":"Halifax","avgHigh":11.0,"avgLow":1.6}
{"city":"Fredericton","avgHigh":11.2,"avgLow":-0.5}
{"city":"Quebec","avgHigh":9.0,"avgLow":-1.0}
{"city":"Montreal","avgHigh":11.1,"avgLow":1.4}
...

实例二：时间转化

case class Purchase(customer_id: Int, purchase_id: Int, date: String, time: String, tz: String, amount:Double)

val x = sc.parallelize(Array(
  Purchase(123, 234, "2007-12-12", "20:50", "UTC", 500.99),
  Purchase(123, 247, "2007-12-12", "15:30", "PST", 300.22),
  Purchase(189, 254, "2007-12-13", "00:50", "EST", 122.19),
  Purchase(187, 299, "2007-12-12", "07:30", "UTC", 524.37)
))

val df = sqlContext.createDataFrame(x)
df.registerTempTable("df")

自定义函数

def makeDT(date: String, time: String, tz: String) = s"$date $time $tz"
sqlContext.udf.register("makeDt", makeDT(_:String,_:String,_:String))

// Now we can use our function directly in SparkSQL. sqlContext.sql("SELECT amount, makeDt(date, time, tz) from df").take(2)
// but not outside df.select($"customer_id", makeDt($"date", $"time", $"tz"), $"amount").take(2) // fails

如果想要在SQL外面使用，必须通过spark.sql.function.udf来创建UDF

import org.apache.spark.sql.functions.udf
val makeDt = udf(makeDT(_:String,_:String,_:String))
// now this works df.select($"customer_id", makeDt($"date", $"time", $"tz"), $"amount").take(2)

实践操作

写一个UDF来将一些Int数字分类

val formatDistribution = (view: Int) => {
  if (view < 10) {
    "<10"
  } else if (view <= 100) {
    "10~100"
  } else if (view <= 1000) {
    "100~1K"
  } else if (view <= 10000) {
    "1K~10K"
  } else if (view <= 100000) {
    "10K~100K"
  } else {
    ">100K"
  }
}

注册：

session.udf.register("formatDistribution", UDF.formatDistribution)

SQL：

session.sql("select user_id, formatDistribution(variance_digg_count) as variance from video")

写到这里，再回顾UDF，我感觉这就像是去为了方便做一个分类转化等操作，和Python里面的函数一样，只不过这里的UDF一般特指Spark SQL里面使用的函数。然后发现这里和SQL中的自定义函数挺像的:

CREATE FUNCTION [函数所有者.]<函数名称> 
(   
    -- 添加函数所需的参数，可以没有参数     [<@param1> <参数类型>]
    [,<@param1> <参数类型>]…
)
RETURNS TABLE 
AS
RETURN 
(
    -- 查询返回的SQL语句     SELECT查询语句
)

/* * 创建内联表值函数，查询交易总额大于1W的开户人个人信息 */
create function getCustInfo()
returns @CustInfo table  --返回table类型
(
    --账户ID
    CustID int,
    --帐户名称
    CustName varchar(20) not null,
    --身份证号
    IDCard varchar(18),
    --电话
    TelePhone varchar(13) not null,
    --地址
    Address varchar(50) default('地址不详')
)
as
begin
    --为table表赋值
    insert into @CustInfo
    select CustID,CustName,IDCard,TelePhone,Address from AccountInfo 
    where CustID in (select CustID from CardInfo 
    where CardID in (select CardID from TransInfo group by CardID,transID,TransType,TransMoney,TransDate having sum(TransMoney)>10000))
    return
end
go
-- 调用内联表值函数 select * from getCustInfo()
go

好像有异曲同工之妙~

欢迎关注：Spark实战–学习UDF