Spark DataFrame简介(二)

Spark DataFrame基础操作

 

创建SparkSession和SparkContext

val spark = SparkSession.builder.master("local").getOrCreate()
val sc = spark.sparkContext

从数组创建DataFrame

spark.range(1000).toDF("number").show()

指定Schema创建DataFrame

val data = Seq(
  Row("A", 10, 112233),
  Row("B", 20, 223311),
  Row("C", 30, 331122))

val schema = StructType(List(
  StructField("name", StringType),
  StructField("age", IntegerType),
  StructField("phone", IntegerType)))

spark.createDataFrame(sc.makeRDD(data), schema).show()

从JSON文件加载DataFrame

/* data.json
   {"name":"A","age":10,"phone":112233}
   {"name":"B", "age":20,"phone":223311}
   {"name":"C", "age":30,"phone":331122}
 */
spark.read.format("json").load("/Users/tobe/temp2/data.json").show()

从CSV文件加载DataFrame

/* data.csv
   name,age,phone
   A,10,112233
   B,20,223311
   C,30,331122
 */
spark.read.option("header", true).csv("/Users/tobe/temp2/data.csv").show()

读取MySQL数据库加载DataFrame

/* data.csv
   name,age,phone
   A,10,112233
   B,20,223311
   C,30,331122
 */
spark.read.option("header", true).csv("/Users/tobe/temp2/data.csv").show()

RDD转DataFrame

/* data.csv
   name,age,phone
   A,10,112233
   B,20,223311
   C,30,331122
 */
spark.read.option("header", true).csv("/Users/tobe/temp2/data.csv").show()

创建Timestamp数据

Spark的TimestampType类型与Java的java.sql.Timestamp对应，

/* data.csv
   name,age,phone
   A,10,112233
   B,20,223311
   C,30,331122
 */
spark.read.option("header", true).csv("/Users/tobe/temp2/data.csv").show()

创建DateType数据

Spark的DateType类型与Java的java.sql.Date对应，

/* data.csv
   name,age,phone
   A,10,112233
   B,20,223311
   C,30,331122
 */
spark.read.option("header", true).csv("/Users/tobe/temp2/data.csv").show()