DataFrame和SparkSql取值误区

来源：互联网发布：化工设计软件编辑：程序博客网时间：2024/06/14 02:42

1、DataFrame返回的不是对象。

2、DataFrame查出来的数据返回的是一个dataframe数据集。

3、DataFrame只有遇见Action的算子才能执行

4、SparkSql查出来的数据返回的是一个dataframe数据集。

原始数据

scala>    val parquetDF = sqlContext.read.parquet("hdfs://hadoop14:9000/yuhui/parquet/part-r-00004.gz.parquet")df: org.apache.spark.sql.DataFrame = [timestamp: string, appkey: string, app_version: string, channel: string, lang: string, os_type: string, os_version: string, display: string, device_type: string, mac: string, network: string, nettype: string, suuid: string, register_days: int, country: string, area: string, province: string, city: string, event: string, use_interval_cat: string, use_duration_cat: string, use_interval: bigint, use_duration: bigint, os_upgrade_from: string, app_upgrade_from: string, page_name: string, event_name: string, error_type: string]

代码

package DataFrameimport org.apache.spark.sql.SQLContextimport org.apache.spark.{SparkConf, SparkContext}/**  * Created by yuhui on 2016/6/14.  */object DataFrameTest {  def main(args: Array[String]) {        DataFrameInto()  }  def DataFrameInto() {    val conf = new SparkConf()    val sc = new SparkContext(conf)    val sqlContext = new SQLContext(sc)    val df = sqlContext.read.parquet("hdfs://hadoop14:9000/yuhui/parquet")   //df.map(line => printinfo(line.getString(0)))   //df.foreach(line => printinfo(line.getString(0)+" , "+line.getString(14)+" , "+line.getString(15)))   //df.select("timestamp","country","area").foreach(line=>printinfo(line.toString))    df.registerTempTable("infotable")    sqlContext.sql("SELECT timestamp , country , area from infotable").foreach(line=>printinfo(line.toString))  }  def printinfo(msg: String) {println("printinfo函数-->" + msg) }}

代码解析

1、df.map(line => printinfo(line.getString(0)))

这段代码不行执行printinfo()函数，因为只有map算子，没有Action算子。

2、df.foreach(line => printinfo(line.getString(0)+" , "+line.getString(14)+" , "+line.getString(15)))

通过Spark的Action算子接收数据进行操作，执行结果如下：

3、df.select("timestamp","country","area").foreach(line=>printinfo(line.toString))

通过DataFrame的API进行操作，再通过Spark的Action算子打印出来，执行结果如下：

4、sqlContext.sql("SELECT timestamp , country , area from infotable").foreach(line=>printinfo(line.toString))

执行结果如下：

0 0