GaParser.scala 2.88 KB
package mobvista.dmp.datasource.ga

import java.net.URI
import java.util.regex.Pattern

import mobvista.dmp.common.CommonSparkJob
import org.apache.commons.cli.Options
import org.apache.hadoop.fs.{FileSystem, Path}
import org.apache.spark.sql.{SaveMode, SparkSession}

/**
 * @package: mobvista.dmp.datasource.ga
 * @author: wangjf
 * @date: 2018/11/4
 * @time: 上午9:50
 * @email: jinfeng.wang@mobvista.com
 * @phone: 152-1062-7698
 */
class GaParser extends CommonSparkJob with Serializable {

  override protected def run(args: Array[String]): Int = {

    val commandLine = commParser.parse(options, args)
    if (!checkMustOption(commandLine)) {
      printUsage(options)
      return -1
    } else {
      printOptions(commandLine)
    }

    val input = commandLine.getOptionValue("input")
    val date_str_midline = commandLine.getOptionValue("date_str_midline")
    val coalesce = commandLine.getOptionValue("coalesce")
    val output = commandLine.getOptionValue("output")

    val spark = SparkSession.builder()
      .appName("GaParser")
      .config("spark.rdd.compress", "true")
      .config("spark.io.compression.codec", "snappy")
      .config("spark.sql.orc.filterPushdown", "true")
      .config("spark.sql.warehouse.dir", "s3://mob-emr-test/spark-warehouse")
      .config("spark.serializer", "org.apache.spark.serializer.KryoSerializer")
      .enableHiveSupport()
      .getOrCreate()
    val sc = spark.sparkContext

    FileSystem.get(new URI(s"s3://mob-emr-test"), sc.hadoopConfiguration).delete(new Path(output), true)
    try {

      val ga_filter = "s3://mob-emr-test/ga-filter-lists/games.csv"
      val filter_rdd = if (FileSystem.get(new URI(s"s3://mob-emr-test"), sc.hadoopConfiguration).exists(new Path(ga_filter))) {
        sc.textFile(ga_filter).map(_.split(",")).filter(r => {
          Pattern.compile("^\\d+$").matcher(r(0)).matches()
        }).map(r => {
          r(0).toInt
        })
      } else {
        sc.emptyRDD[Int]
      }
      val filter_gameId = filter_rdd.collect().toSet[Int]

      println("filter_gameId -->> " + filter_gameId)

      sc.broadcast(filter_gameId)
      val rdd = sc.textFile(path = input).coalesce(coalesce.toInt)
      val df = rdd.mapPartitions(Constant.parserGa(_, date_str_midline, filter_gameId))

      import spark.implicits._
      df.filter(_ != null).toDF
        .write.mode(SaveMode.Overwrite)
        .option("orc.compress", "zlib")
        .orc(output)

    } finally {
      sc.stop()
      spark.stop()
    }
    0
  }

  override protected def buildOptions(): Options = {
    val options = new Options
    options.addOption("input", true, "[must] input")
    options.addOption("date_str_midline", true, "[must] date_str_midline")
    options.addOption("coalesce", true, "[must] coalesce")
    options.addOption("output", true, "[must] output")
    options
  }
}

object GaParser {
  def main(args: Array[String]): Unit = {
    new GaParser().run(args)
  }
}