Spark 操作JSON格式数据

scala> val json=spark.read.json("/opt/data/emp.json")
json: org.apache.spark.sql.DataFrame = [comm: string, deptno: bigint ... 6 more fields]

scala> json.show
+----+------+-----+------+----------+---------+----+----+
|comm|deptno|empno| ename|  hiredate|      job| mgr| sal|
+----+------+-----+------+----------+---------+----+----+
|    |    20| 7369| SMITH|1980/12/17|    CLERK|7902| 800|
| 300|    30| 7499| ALLEN| 1981/2/20| SALESMAN|7698|1600|
| 500|    30| 7521|  WARD| 1981/2/22| SALESMAN|7698|1250|
|    |    20| 7566| JONES|  1981/4/2|  MANAGER|7839|2975|
|1400|    30| 7654|MARTIN| 1981/9/28| SALESMAN|7698|1250|
|    |    30| 7698| BLAKE|  1981/5/1|  MANAGER|7839|2850|
|    |    10| 7782| CLARK|  1981/6/9|  MANAGER|7839|2450|
|    |    20| 7788| SCOTT| 1987/4/19|  ANALYST|7566|3000|
|    |    10| 7839|  KING|1981/11/17|PRESIDENT|    |5000|
|   0|    30| 7844|TURNER|  1981/9/8| SALESMAN|7698|1500|
|    |    20| 7876| ADAMS| 1987/5/23|    CLERK|7788|1100|
|    |    30| 7900| JAMES| 1981/12/3|    CLERK|7698| 950|
|    |    20| 7902|  FORD| 1981/12/3|  ANALYST|7566|3000|
|    |    10| 7934|MILLER| 1982/1/23|    CLERK|7782|1300|
+----+------+-----+------+----------+---------+----+----+

scala> json.printSchema
root
 |-- comm: string (nullable = true)
 |-- deptno: long (nullable = true)
 |-- empno: long (nullable = true)
 |-- ename: string (nullable = true)
 |-- hiredate: string (nullable = true)
 |-- job: string (nullable = true)
 |-- mgr: string (nullable = true)
 |-- sal: long (nullable = true)

scala> json.createOrReplaceTempView("emp")

scala> spark.sql("select deptno,count(1) from emp group by deptno").show
+------+--------+
|deptno|count(1)|
+------+--------+
|    10|       3|
|    30|       6|
|    20|       5|
+------+--------+

scala>