Spark编程实现SQL查询的实例

1、Oracle中的SQL

select count(1)
from a_V_PWYZL_CUSTACCT_PSMIS t
   where not exists (select 1
            from tb_show_multi_question q
           WHERE q.dqmp_rule_code = ‘仅比对系统有‘
             and q.dqmp_role_id = ‘105754659‘
             and q.DQMP_target_id = t.dqmp_mrid)
     AND NOT EXISTS (select /*+ index(s) */
           1
            from a_V_PWYZL_CUSTACCT_GIS s
           where s.dqmp_cpk = t.dqmp_cpk)
     and t.is_repeat = ‘0‘;

2、Hive/Shark版

select count(1) from
  (
  select t.*,q.dqmp_question_id(列索引q.0) as f1,s.gis_mrid(列索引s.0)as f2
  from (select * from a_V_PWYZL_CUSTACCT_PSMIS t where t.is_repeat(列索引t.19) = ‘0‘) t
       left outer join (select * from tb_show_multi_question q where q.dqmp_rule_code(列索引q.26)= ‘仅比对系统有‘ and q.dqmp_role_id(列索引q.31)= ‘105754659‘ ) q on q.DQMP_target_id（列索引q.13）= t.dqmp_mrid（列索引t.32）
       left outer join a_V_PWYZL_CUSTACCT_GIS s on s.dqmp_cpk(列索引s.31)= t.dqmp_cpk（列索列t.31）
  ) tv
where tv.f1 is NULL and tv.f2 is NULL;

3、Spark编程实现

 

package cn.com.gzkit.spark

import org.apache.spark.SparkContext

import org.apache.spark.SparkContext._

import org.apache.spark.rdd.PairRDDFunctions;

object test2 {

  def main(args: Array[String]) {

      var master="spark://kit-b1:7077";

      var sc=new SparkContext(master, "HdfsTest",System.getenv("SPARK_HOME"),SparkContext.jarOfClass(this.getClass));

      var file1 = sc.textFile("hdfs://kit-b1/demodata/utf8_a_v_pwyzl_custacct_psmis.txt").map(_.split(‘|‘));

      var file2 = sc.textFile("hdfs://kit-b1/demodata/utf8_a_v_pwyzl_custacct_gis.txt").map(_.split(‘|‘)).map(m=>(m(31),m(0)));

      var file3 = sc.textFile("hdfs://kit-b1/demodata/utf8_tb_show_multi_question.txt").map(_.split(‘|‘)).map(m=>(m(13),m(31),m(26),m(0)));

      file1.cache();

      file2.cache();

      file3.cache();

      //file1.count();

      //file2.count();

      //file3.count();

      System.out.println("-----------begin-----------------");

      var t=file1.filter(_(19)=="0");

      var t1=t.map(t=>(t(32),t));

      var q=file3.filter(_._2=="105754659").filter(_._3=="仅比对系统有");

      var q1=q.map(q=>(q._1,q));

      var tq=t1.leftOuterJoin(q1);

      System.out.println("-----------end-----------------");

      System.out.println("-----------begin2-----------------");

      var t2=tq.map(m=>(m._2._1(31),m._2));

      var s=file2;

      var ts=t2.leftOuterJoin(s);

      var rs=ts.filter(_._2._2==None).filter(_._2._1._2==None);

      System.out.println("sava file");

      rs.saveAsTextFile("hdfs://kit-b1/demodata/test/02");

      System.out.println("sava file end");

      var v2=rs.count();

      System.out.println("v2="+v2);

      System.out.println("-----------end2-----------------");

      sc.stop();

  }

}

Spark编程实现SQL查询的实例,布布扣,bubuko.com

Spark编程实现SQL查询的实例

原文：http://www.cnblogs.com/lam99v/p/3793031.html