direct-spark-sql

a hyper-optimized single-node(local) version of spark sql engine, which's fundamental data structure is scala Iterator rather than RDD.

Usage

see DirectExecSuite.scala

val spark = DirectSparkSession.builder().getOrCreate()
  spark
      .createDataFrame(Seq(("a", 2, 0), ("bbb", 2, 1), ("c", 3, 0), ("ddd", 4, 1), ("e", 5, 1)))
      .toDF("name", "age", "genda")
      .createOrReplaceTempView("people")
    spark
      .createDataFrame(List(("a", 1, 0), ("b", 2, 1), ("c", 3, 0)))
      .toDF("name", "age", "genda")
      .createOrReplaceTempView("people2")
      
 val resultTable = spark.sqlDirectly("""
                                             |select
                                             |* from people t1
                                             |join people2 t2
                                             |on t1.name = t2.name
                                             |""".stripMargin)

Contributing

It's under active developing now , may be released in one or two months.

Please run follow command before submitting a PR:

 mvn clean install -Pcheck-style