准备数据

必须是一行,不能是多行

[{       "name":"zhangsan",      "age":"18"},{"name":"lisi","age":"25"}]

SparkShell读取Linux本地json文件

root/soft/person.json就是本机路径

var df = spark.read.format(“json”).load(“file:///root/soft/person.json”)

scala> var df =   spark.read.format("json").load("file:///root/soft/person.json")
df: org.apache.spark.sql.DataFrame = [_corrupt_record: string]
scala> df.show
+--------------------+
|     _corrupt_record|
+--------------------+
|                   [|
|                   {|
|        "name":"z...|
|          "age":"18"|
|                  },|
|                   {|
|        "name":"l...|
|          "age":"25"|
|                   }|
|                   ]|
+--------------------+
scala> var df =   spark.read.format("json").load("file:///root/soft/person.json")
df: org.apache.spark.sql.DataFrame = [age: string, name: string]
scala> df.show
+---+--------+
|age|    name|
+---+--------+
| 18|zhangsan|
| 25|    lisi|
+---+--------+

SparkShell读取HDFS上的json文件

文件在hdfs的data目录下面

scala> var df2 =   spark.read.json("/data/person.json")
df2: org.apache.spark.sql.DataFrame = [age: string, name: string]
scala> df2.show
+---+--------+
|age|    name|
+---+--------+
| 18|zhangsan|
| 25|    lisi|
+---+--------+
scala>