2021-10-15 15:00:15 +02:00
|
|
|
package eu.dnetlib.dhp.sx.bio.ebi
|
2021-06-24 17:20:00 +02:00
|
|
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser
|
2021-12-06 11:26:36 +01:00
|
|
|
import eu.dnetlib.dhp.collection.CollectionUtils
|
2021-06-24 17:20:00 +02:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Oaf
|
2021-10-15 15:00:15 +02:00
|
|
|
import eu.dnetlib.dhp.sx.bio.BioDBToOAF
|
|
|
|
import eu.dnetlib.dhp.sx.bio.BioDBToOAF.EBILinkItem
|
2021-06-24 17:20:00 +02:00
|
|
|
import org.apache.commons.io.IOUtils
|
|
|
|
import org.apache.spark.SparkConf
|
2021-10-12 08:11:53 +02:00
|
|
|
import org.apache.spark.sql._
|
2021-06-24 17:20:00 +02:00
|
|
|
import org.slf4j.{Logger, LoggerFactory}
|
2021-10-12 08:11:53 +02:00
|
|
|
|
2021-06-24 17:20:00 +02:00
|
|
|
object SparkEBILinksToOaf {
|
|
|
|
|
|
|
|
def main(args: Array[String]): Unit = {
|
2021-06-30 17:27:55 +02:00
|
|
|
val log: Logger = LoggerFactory.getLogger(getClass)
|
2021-06-24 17:20:00 +02:00
|
|
|
val conf: SparkConf = new SparkConf()
|
2022-01-11 16:57:48 +01:00
|
|
|
val parser = new ArgumentApplicationParser(
|
|
|
|
IOUtils.toString(
|
|
|
|
getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/ebi/ebi_to_df_params.json")
|
|
|
|
)
|
|
|
|
)
|
2021-06-24 17:20:00 +02:00
|
|
|
parser.parseArgument(args)
|
|
|
|
val spark: SparkSession =
|
|
|
|
SparkSession
|
|
|
|
.builder()
|
|
|
|
.config(conf)
|
|
|
|
.appName(SparkEBILinksToOaf.getClass.getSimpleName)
|
2022-01-11 16:57:48 +01:00
|
|
|
.master(parser.get("master"))
|
|
|
|
.getOrCreate()
|
2021-10-12 08:11:53 +02:00
|
|
|
|
|
|
|
import spark.implicits._
|
2021-06-24 17:20:00 +02:00
|
|
|
val sourcePath = parser.get("sourcePath")
|
|
|
|
log.info(s"sourcePath -> $sourcePath")
|
|
|
|
val targetPath = parser.get("targetPath")
|
|
|
|
log.info(s"targetPath -> $targetPath")
|
2021-10-12 08:11:53 +02:00
|
|
|
implicit val PMEncoder: Encoder[Oaf] = Encoders.kryo(classOf[Oaf])
|
2021-06-24 17:20:00 +02:00
|
|
|
|
2022-01-11 16:57:48 +01:00
|
|
|
val ebLinks: Dataset[EBILinkItem] = spark.read
|
|
|
|
.load(sourcePath)
|
|
|
|
.as[EBILinkItem]
|
|
|
|
.filter(l => l.links != null && l.links.startsWith("{"))
|
2021-06-24 17:20:00 +02:00
|
|
|
|
2022-01-11 16:57:48 +01:00
|
|
|
CollectionUtils.saveDataset(
|
|
|
|
ebLinks
|
|
|
|
.flatMap(j => BioDBToOAF.parse_ebi_links(j.links))
|
|
|
|
.filter(p => BioDBToOAF.EBITargetLinksFilter(p))
|
|
|
|
.flatMap(p => BioDBToOAF.convertEBILinksToOaf(p)),
|
|
|
|
targetPath
|
|
|
|
)
|
2021-06-24 17:20:00 +02:00
|
|
|
}
|
|
|
|
}
|