2021-10-20 17:37:42 +02:00
|
|
|
package eu.dnetlib.dhp.datacite
|
2021-09-28 08:58:59 +02:00
|
|
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.{Oaf, Result}
|
|
|
|
import org.apache.spark.SparkConf
|
|
|
|
import org.apache.spark.sql.functions.max
|
2021-10-20 17:37:42 +02:00
|
|
|
import org.apache.spark.sql.{Encoder, Encoders, SparkSession}
|
2021-09-28 08:58:59 +02:00
|
|
|
import org.slf4j.{Logger, LoggerFactory}
|
|
|
|
|
|
|
|
import java.text.SimpleDateFormat
|
2021-10-20 17:37:42 +02:00
|
|
|
import java.util.Locale
|
2021-09-28 08:58:59 +02:00
|
|
|
import scala.io.Source
|
|
|
|
|
|
|
|
object SparkDownloadUpdateDatacite {
|
|
|
|
val log: Logger = LoggerFactory.getLogger(getClass)
|
|
|
|
|
|
|
|
def main(args: Array[String]): Unit = {
|
|
|
|
|
|
|
|
val conf = new SparkConf
|
2022-01-11 16:57:48 +01:00
|
|
|
val parser = new ArgumentApplicationParser(
|
|
|
|
Source
|
|
|
|
.fromInputStream(
|
|
|
|
getClass.getResourceAsStream("/eu/dnetlib/dhp/datacite/generate_dataset_params.json")
|
|
|
|
)
|
|
|
|
.mkString
|
|
|
|
)
|
2021-09-28 08:58:59 +02:00
|
|
|
parser.parseArgument(args)
|
|
|
|
val master = parser.get("master")
|
|
|
|
val sourcePath = parser.get("sourcePath")
|
|
|
|
val workingPath = parser.get("workingPath")
|
|
|
|
|
|
|
|
val hdfsuri = parser.get("namenode")
|
|
|
|
log.info(s"namenode is $hdfsuri")
|
|
|
|
|
2022-01-11 16:57:48 +01:00
|
|
|
val spark: SparkSession = SparkSession
|
|
|
|
.builder()
|
|
|
|
.config(conf)
|
2021-09-28 08:58:59 +02:00
|
|
|
.appName(getClass.getSimpleName)
|
|
|
|
.master(master)
|
|
|
|
.getOrCreate()
|
|
|
|
|
|
|
|
implicit val oafEncoder: Encoder[Oaf] = Encoders.kryo[Oaf]
|
|
|
|
implicit val resEncoder: Encoder[Result] = Encoders.kryo[Result]
|
|
|
|
|
|
|
|
import spark.implicits._
|
|
|
|
|
2022-01-11 16:57:48 +01:00
|
|
|
val maxDate: String = spark.read
|
|
|
|
.load(workingPath)
|
|
|
|
.as[Oaf]
|
|
|
|
.filter(s => s.isInstanceOf[Result])
|
|
|
|
.map(r => r.asInstanceOf[Result].getDateofcollection)
|
|
|
|
.select(max("value"))
|
|
|
|
.first()
|
|
|
|
.getString(0)
|
2021-09-28 08:58:59 +02:00
|
|
|
val ISO8601FORMAT = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ssZ", Locale.US)
|
2021-10-20 17:37:42 +02:00
|
|
|
val string_to_date = ISO8601FORMAT.parse(maxDate)
|
2021-09-28 08:58:59 +02:00
|
|
|
val ts = string_to_date.getTime
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|