diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/SparkSplitOafTODLIEntities.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/SparkSplitOafTODLIEntities.scala index f359f73f9..9bdf26acb 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/SparkSplitOafTODLIEntities.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/SparkSplitOafTODLIEntities.scala @@ -3,7 +3,7 @@ package eu.dnetlib.dhp.sx.graph import eu.dnetlib.dhp.application.ArgumentApplicationParser import eu.dnetlib.dhp.schema.oaf.{Oaf, Relation, Result} import eu.dnetlib.dhp.schema.scholexplorer.{DLIDataset, DLIPublication, DLIUnknown} -import eu.dnetlib.dhp.sx.ebi.EBIAggregator +import eu.dnetlib.dhp.sx.graph.ebi.EBIAggregator import org.apache.commons.io.IOUtils import org.apache.commons.lang3.StringUtils import org.apache.spark.sql.{Dataset, Encoder, Encoders, SaveMode, SparkSession} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/BioDBToOAF.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/BioDBToOAF.scala similarity index 99% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/BioDBToOAF.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/BioDBToOAF.scala index f435c56dd..a19c6fb12 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/BioDBToOAF.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/BioDBToOAF.scala @@ -1,4 +1,4 @@ -package eu.dnetlib.dhp.sx.bio +package eu.dnetlib.dhp.sx.graph.bio import eu.dnetlib.dhp.schema.common.ModelConstants import eu.dnetlib.dhp.schema.oaf.utils.{GraphCleaningFunctions, OafMapperUtils} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/SparkTransformBioDatabaseToOAF.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/SparkTransformBioDatabaseToOAF.scala similarity index 91% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/SparkTransformBioDatabaseToOAF.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/SparkTransformBioDatabaseToOAF.scala index 1bca02a73..d66cc84ec 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/SparkTransformBioDatabaseToOAF.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/SparkTransformBioDatabaseToOAF.scala @@ -1,8 +1,8 @@ -package eu.dnetlib.dhp.sx.bio +package eu.dnetlib.dhp.sx.graph.bio import eu.dnetlib.dhp.application.ArgumentApplicationParser import eu.dnetlib.dhp.schema.oaf.{Oaf, Result} -import eu.dnetlib.dhp.sx.bio.BioDBToOAF.ScholixResolved +import BioDBToOAF.ScholixResolved import org.apache.commons.io.IOUtils import org.apache.spark.SparkConf import org.apache.spark.sql.{Encoder, Encoders, SaveMode, SparkSession} @@ -13,7 +13,7 @@ object SparkTransformBioDatabaseToOAF { def main(args: Array[String]): Unit = { val conf: SparkConf = new SparkConf() val log: Logger = LoggerFactory.getLogger(getClass) - val parser = new ArgumentApplicationParser(IOUtils.toString(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/bio_to_oaf_params.json"))) + val parser = new ArgumentApplicationParser(IOUtils.toString(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/bio/bio_to_oaf_params.json"))) parser.parseArgument(args) val database: String = parser.get("database") log.info("database: {}", database) diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMArticle.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMArticle.java similarity index 97% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMArticle.java rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMArticle.java index 881528425..211cbcffb 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMArticle.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMArticle.java @@ -1,5 +1,5 @@ -package eu.dnetlib.dhp.sx.bio.pubmed; +package eu.dnetlib.dhp.sx.graph.bio.pubmed; import java.io.Serializable; import java.util.ArrayList; diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMAuthor.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMAuthor.java similarity index 92% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMAuthor.java rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMAuthor.java index cef92d003..ba69998c5 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMAuthor.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMAuthor.java @@ -1,5 +1,5 @@ -package eu.dnetlib.dhp.sx.bio.pubmed; +package eu.dnetlib.dhp.sx.graph.bio.pubmed; import java.io.Serializable; diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMGrant.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMGrant.java similarity index 93% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMGrant.java rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMGrant.java index ce9420cc1..0c3fd4601 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMGrant.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMGrant.java @@ -1,5 +1,5 @@ -package eu.dnetlib.dhp.sx.bio.pubmed; +package eu.dnetlib.dhp.sx.graph.bio.pubmed; public class PMGrant { diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMJournal.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMJournal.java similarity index 94% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMJournal.java rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMJournal.java index 863a23bd5..d251354d4 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMJournal.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMJournal.java @@ -1,5 +1,5 @@ -package eu.dnetlib.dhp.sx.bio.pubmed; +package eu.dnetlib.dhp.sx.graph.bio.pubmed; import java.io.Serializable; diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMParser.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMParser.scala similarity index 99% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMParser.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMParser.scala index 80cb0667c..8744bdfb4 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMParser.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMParser.scala @@ -1,4 +1,4 @@ -package eu.dnetlib.dhp.sx.bio.pubmed +package eu.dnetlib.dhp.sx.graph.bio.pubmed import scala.xml.MetaData import scala.xml.pull.{EvElemEnd, EvElemStart, EvText, XMLEventReader} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMSubject.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMSubject.java similarity index 94% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMSubject.java rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMSubject.java index 862d39a94..354b2cbe5 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PMSubject.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PMSubject.java @@ -1,5 +1,5 @@ -package eu.dnetlib.dhp.sx.bio.pubmed; +package eu.dnetlib.dhp.sx.graph.bio.pubmed; public class PMSubject { private String value; diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PubMedToOaf.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PubMedToOaf.scala similarity index 99% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PubMedToOaf.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PubMedToOaf.scala index 2a1392353..ae4a72062 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/bio/pubmed/PubMedToOaf.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/PubMedToOaf.scala @@ -1,4 +1,4 @@ -package eu.dnetlib.dhp.sx.bio.pubmed +package eu.dnetlib.dhp.sx.graph.bio.pubmed import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup import eu.dnetlib.dhp.schema.common.ModelConstants diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/EBIAggregator.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/EBIAggregator.scala similarity index 99% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/EBIAggregator.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/EBIAggregator.scala index ee2dbadfd..4e3fda800 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/EBIAggregator.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/EBIAggregator.scala @@ -1,4 +1,5 @@ -package eu.dnetlib.dhp.sx.ebi +package eu.dnetlib.dhp.sx.graph.ebi + import eu.dnetlib.dhp.oa.merge.AuthorMerger import eu.dnetlib.dhp.schema.oaf.{Publication, Relation, Dataset => OafDataset} import eu.dnetlib.dhp.schema.scholexplorer.{DLIDataset, DLIPublication, DLIUnknown} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkAddLinkUpdates.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkAddLinkUpdates.scala similarity index 98% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkAddLinkUpdates.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkAddLinkUpdates.scala index 294cd97ad..a8dd93688 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkAddLinkUpdates.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkAddLinkUpdates.scala @@ -1,10 +1,11 @@ -package eu.dnetlib.dhp.sx.ebi +package eu.dnetlib.dhp.sx.graph.ebi + import eu.dnetlib.dhp.application.ArgumentApplicationParser import eu.dnetlib.dhp.schema.common.ModelConstants import eu.dnetlib.dhp.schema.oaf.{Author, Instance, Journal, KeyValue, Oaf, Publication, Relation, Dataset => OafDataset} import eu.dnetlib.dhp.schema.scholexplorer.OafUtils.createQualifier import eu.dnetlib.dhp.schema.scholexplorer.{DLIDataset, DLIPublication, OafUtils, ProvenaceInfo} -import eu.dnetlib.dhp.sx.bio.pubmed.{PMArticle, PMAuthor, PMJournal} +import eu.dnetlib.dhp.sx.graph.bio.pubmed.{PMArticle, PMAuthor, PMJournal} import eu.dnetlib.dhp.utils.DHPUtils import eu.dnetlib.scholexplorer.relation.RelationMapper import org.apache.commons.io.IOUtils @@ -165,7 +166,7 @@ case class EBILinks(relation:String, pubdate:String, tpid:String, tpidType:Strin def main(args: Array[String]): Unit = { val conf: SparkConf = new SparkConf() - val parser = new ArgumentApplicationParser(IOUtils.toString(SparkEBILinksToOaf.getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/ebi/ebi_to_df_params.json"))) + val parser = new ArgumentApplicationParser(IOUtils.toString(SparkEBILinksToOaf.getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/ebi/ebi_to_df_params.json"))) parser.parseArgument(args) val spark: SparkSession = SparkSession diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkCreateBaselineDataFrame.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkCreateBaselineDataFrame.scala similarity index 92% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkCreateBaselineDataFrame.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkCreateBaselineDataFrame.scala index 27dd4adba..b9cccf4e1 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkCreateBaselineDataFrame.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkCreateBaselineDataFrame.scala @@ -1,9 +1,9 @@ -package eu.dnetlib.dhp.sx.ebi +package eu.dnetlib.dhp.sx.graph.ebi import eu.dnetlib.dhp.application.ArgumentApplicationParser import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup import eu.dnetlib.dhp.schema.oaf.Result -import eu.dnetlib.dhp.sx.bio.pubmed.{PMArticle, PMAuthor, PMJournal, PMParser, PubMedToOaf} +import eu.dnetlib.dhp.sx.graph.bio.pubmed.{PMArticle, PMAuthor, PMJournal, PMParser, PubMedToOaf} import eu.dnetlib.dhp.utils.ISLookupClientFactory import org.apache.commons.io.IOUtils import org.apache.spark.SparkConf @@ -41,7 +41,7 @@ object SparkCreateBaselineDataFrame { def main(args: Array[String]): Unit = { val conf: SparkConf = new SparkConf() val log: Logger = LoggerFactory.getLogger(getClass) - val parser = new ArgumentApplicationParser(IOUtils.toString(SparkEBILinksToOaf.getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/ebi/baseline_to_oaf_params.json"))) + val parser = new ArgumentApplicationParser(IOUtils.toString(SparkEBILinksToOaf.getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/ebi/baseline_to_oaf_params.json"))) parser.parseArgument(args) val isLookupUrl: String = parser.get("isLookupUrl") log.info("isLookupUrl: {}", isLookupUrl) @@ -81,6 +81,6 @@ object SparkCreateBaselineDataFrame { exported_dataset .map(a => PubMedToOaf.convert(a, vocabularies)).as[Result] .filter(p => p!= null) - .write.mode(SaveMode.Overwrite).save(s"$workingPath/baseline_oaf") + .write.mode(SaveMode.Overwrite).save(s"$workingPath/oaf/baseline_oaf") } } diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkEBILinksToOaf.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkEBILinksToOaf.scala similarity index 89% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkEBILinksToOaf.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkEBILinksToOaf.scala index 250b285ff..b4314631a 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/ebi/SparkEBILinksToOaf.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/ebi/SparkEBILinksToOaf.scala @@ -1,9 +1,9 @@ -package eu.dnetlib.dhp.sx.ebi +package eu.dnetlib.dhp.sx.graph.ebi import eu.dnetlib.dhp.application.ArgumentApplicationParser import eu.dnetlib.dhp.schema.oaf.Oaf -import eu.dnetlib.dhp.sx.bio.BioDBToOAF -import eu.dnetlib.dhp.sx.bio.BioDBToOAF.EBILinkItem +import eu.dnetlib.dhp.sx.graph.bio.BioDBToOAF +import eu.dnetlib.dhp.sx.graph.bio.BioDBToOAF.EBILinkItem import org.apache.commons.io.IOUtils import org.apache.spark.SparkConf import org.apache.spark.sql.{Dataset, Encoder, Encoders, SaveMode, SparkSession} @@ -13,7 +13,7 @@ object SparkEBILinksToOaf { def main(args: Array[String]): Unit = { val log: Logger = LoggerFactory.getLogger(SparkEBILinksToOaf.getClass) val conf: SparkConf = new SparkConf() - val parser = new ArgumentApplicationParser(IOUtils.toString(SparkEBILinksToOaf.getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/ebi/ebi_to_df_params.json"))) + val parser = new ArgumentApplicationParser(IOUtils.toString(SparkEBILinksToOaf.getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/ebi/ebi_to_df_params.json"))) parser.parseArgument(args) val spark: SparkSession = SparkSession diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/sx/pangaea/PangaeaUtils.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/pangaea/PangaeaUtils.scala similarity index 98% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/sx/pangaea/PangaeaUtils.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/pangaea/PangaeaUtils.scala index bbd9124bd..193512474 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/sx/pangaea/PangaeaUtils.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/pangaea/PangaeaUtils.scala @@ -1,4 +1,4 @@ -package eu.dnetlib.sx.pangaea +package eu.dnetlib.dhp.sx.graph.pangaea import org.apache.spark.sql.expressions.Aggregator import org.apache.spark.sql.{Encoder, Encoders} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/sx/pangaea/SparkGeneratePanagaeaDataset.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/pangaea/SparkGeneratePanagaeaDataset.scala similarity index 94% rename from dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/sx/pangaea/SparkGeneratePanagaeaDataset.scala rename to dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/pangaea/SparkGeneratePanagaeaDataset.scala index 24ed08968..bf726cf59 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/sx/pangaea/SparkGeneratePanagaeaDataset.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/pangaea/SparkGeneratePanagaeaDataset.scala @@ -1,7 +1,7 @@ -package eu.dnetlib.sx.pangaea +package eu.dnetlib.dhp.sx.graph.pangaea import eu.dnetlib.dhp.application.ArgumentApplicationParser -import eu.dnetlib.dhp.sx.ebi.SparkEBILinksToOaf +import eu.dnetlib.dhp.sx.graph.ebi.SparkEBILinksToOaf import org.apache.spark.rdd.RDD import org.apache.spark.{SparkConf, SparkContext} import org.apache.spark.sql.{Encoder, Encoders, SaveMode, SparkSession} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/bio_to_oaf_params.json b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/bio_to_oaf_params.json similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/bio_to_oaf_params.json rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/bio_to_oaf_params.json diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/oozie_app/config-default.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/oozie_app/config-default.xml similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/oozie_app/config-default.xml rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/oozie_app/config-default.xml diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/oozie_app/workflow.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/oozie_app/workflow.xml similarity index 94% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/oozie_app/workflow.xml rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/oozie_app/workflow.xml index 90e2ae34f..0df085ee1 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/bio/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/bio/oozie_app/workflow.xml @@ -41,7 +41,7 @@ yarn cluster Convert PDB to OAF Dataset - eu.dnetlib.dhp.sx.bio.SparkTransformBioDatabaseToOAF + eu.dnetlib.dhp.sx.graph.bio.SparkTransformBioDatabaseToOAF dhp-graph-mapper-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} @@ -68,7 +68,7 @@ yarn cluster Convert UNIPROT to OAF Dataset - eu.dnetlib.dhp.sx.bio.SparkTransformBioDatabaseToOAF + eu.dnetlib.dhp.sx.graph.bio.SparkTransformBioDatabaseToOAF dhp-graph-mapper-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} @@ -95,7 +95,7 @@ yarn cluster Convert EBI Links to OAF Dataset - eu.dnetlib.dhp.sx.ebi.SparkEBILinksToOaf + eu.dnetlib.dhp.sx.graph.ebi.SparkEBILinksToOaf dhp-graph-mapper-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} @@ -121,7 +121,7 @@ yarn cluster Convert Scholix to OAF Dataset - eu.dnetlib.dhp.sx.bio.SparkTransformBioDatabaseToOAF + eu.dnetlib.dhp.sx.graph.bio.SparkTransformBioDatabaseToOAF dhp-graph-mapper-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} @@ -148,7 +148,7 @@ yarn cluster Convert Crossref Links to OAF Dataset - eu.dnetlib.dhp.sx.bio.SparkTransformBioDatabaseToOAF + eu.dnetlib.dhp.sx.graph.bio.SparkTransformBioDatabaseToOAF dhp-graph-mapper-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/baseline_to_oaf_params.json b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/baseline_to_oaf_params.json similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/baseline_to_oaf_params.json rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/baseline_to_oaf_params.json diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/ebi_to_df_params.json b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/ebi_to_df_params.json similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/ebi_to_df_params.json rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/ebi_to_df_params.json diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/oozie_app/config-default.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/oozie_app/config-default.xml similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/oozie_app/config-default.xml rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/oozie_app/config-default.xml diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/oozie_app/workflow.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/oozie_app/workflow.xml similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/ebi/oozie_app/workflow.xml rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/ebi/oozie_app/workflow.xml diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/oozie_app/config-default.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/pubmed/oozie_app/config-default.xml similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/oozie_app/config-default.xml rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/pubmed/oozie_app/config-default.xml diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/oozie_app/workflow.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/pubmed/oozie_app/workflow.xml similarity index 94% rename from dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/oozie_app/workflow.xml rename to dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/pubmed/oozie_app/workflow.xml index 81a6167b2..63abba24e 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/graph/pubmed/oozie_app/workflow.xml @@ -21,7 +21,7 @@ yarn cluster Convert Baseline to Dataset - eu.dnetlib.dhp.sx.ebi.SparkCreateBaselineDataFrame + eu.dnetlib.dhp.sx.graph.ebi.SparkCreateBaselineDataFrame dhp-graph-mapper-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/pangaea_to_dataset.json b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/pangaea_to_dataset.json deleted file mode 100644 index 366f1426e..000000000 --- a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/sx/pubmed/pangaea_to_dataset.json +++ /dev/null @@ -1,4 +0,0 @@ -[ - {"paramName":"mt", "paramLongName":"master", "paramDescription": "should be local or yarn", "paramRequired": true}, - {"paramName":"w", "paramLongName":"workingPath", "paramDescription": "the path of the sequencial file to read", "paramRequired": true} -] \ No newline at end of file diff --git a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/SparkScholexplorerAggregationTest.scala b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/SparkScholexplorerAggregationTest.scala index 373683190..278af2a16 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/SparkScholexplorerAggregationTest.scala +++ b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/SparkScholexplorerAggregationTest.scala @@ -2,7 +2,7 @@ package eu.dnetlib.dhp.sx.graph import com.fasterxml.jackson.databind.{ObjectMapper, SerializationFeature} import eu.dnetlib.dhp.schema.scholexplorer.DLIPublication -import eu.dnetlib.dhp.sx.ebi.EBIAggregator +import eu.dnetlib.dhp.sx.graph.ebi.EBIAggregator import org.apache.spark.sql.{Dataset, Encoder, Encoders, SparkSession} import org.junit.jupiter.api.Assertions._ import org.junit.jupiter.api.Test diff --git a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/bio/pubmed/AbstractVocabularyTest.java b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/AbstractVocabularyTest.java similarity index 96% rename from dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/bio/pubmed/AbstractVocabularyTest.java rename to dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/AbstractVocabularyTest.java index cb3451a87..30fac5ac8 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/bio/pubmed/AbstractVocabularyTest.java +++ b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/AbstractVocabularyTest.java @@ -1,5 +1,5 @@ -package eu.dnetlib.dhp.sx.bio.pubmed; +package eu.dnetlib.dhp.sx.graph.bio.pubmed; import static org.mockito.Mockito.lenient; diff --git a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/bio/pubmed/BioScholixTest.scala b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/BioScholixTest.scala similarity index 92% rename from dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/bio/pubmed/BioScholixTest.scala rename to dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/BioScholixTest.scala index de13a08d6..894c0a795 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/bio/pubmed/BioScholixTest.scala +++ b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/graph/bio/pubmed/BioScholixTest.scala @@ -1,9 +1,9 @@ -package eu.dnetlib.dhp.sx.bio.pubmed +package eu.dnetlib.dhp.sx.graph.bio.pubmed import com.fasterxml.jackson.databind.{DeserializationFeature, ObjectMapper, SerializationFeature} import eu.dnetlib.dhp.schema.oaf.{Oaf, Relation, Result} -import eu.dnetlib.dhp.sx.bio.BioDBToOAF -import eu.dnetlib.dhp.sx.bio.BioDBToOAF.ScholixResolved +import eu.dnetlib.dhp.sx.graph.bio.BioDBToOAF.ScholixResolved +import eu.dnetlib.dhp.sx.graph.bio.BioDBToOAF import org.json4s.DefaultFormats import org.json4s.JsonAST.{JField, JObject, JString} import org.json4s.jackson.JsonMethods.parse @@ -72,7 +72,7 @@ class BioScholixTest extends AbstractVocabularyTest{ assertNotNull(vocabularies) assertTrue(vocabularies.vocabularyExists("dnet:publication_resource")) - val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/pdb_dump")).mkString + val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/bio/pdb_dump")).mkString records.lines.foreach(s => assertTrue(s.nonEmpty)) val result:List[Oaf]= records.lines.toList.flatMap(o => BioDBToOAF.pdbTOOaf(o)) @@ -94,7 +94,7 @@ class BioScholixTest extends AbstractVocabularyTest{ assertNotNull(vocabularies) assertTrue(vocabularies.vocabularyExists("dnet:publication_resource")) - val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/uniprot_dump")).mkString + val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/bio/uniprot_dump")).mkString records.lines.foreach(s => assertTrue(s.nonEmpty)) val result:List[Oaf]= records.lines.toList.flatMap(o => BioDBToOAF.uniprotToOAF(o)) @@ -133,7 +133,7 @@ class BioScholixTest extends AbstractVocabularyTest{ @Test def testCrossrefLinksToOAF():Unit = { - val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/crossref_links")).mkString + val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/bio/crossref_links")).mkString records.lines.foreach(s => assertTrue(s.nonEmpty)) @@ -148,7 +148,7 @@ class BioScholixTest extends AbstractVocabularyTest{ @Test def testEBILinksToOAF():Unit = { - val iterator = GzFileIterator(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/ebi_links.gz"), "UTF-8") + val iterator = GzFileIterator(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/bio/ebi_links.gz"), "UTF-8") val data = iterator.next() val res = BioDBToOAF.parse_ebi_links(BioDBToOAF.extractEBILinksFromDump(data).links).filter(BioDBToOAF.EBITargetLinksFilter).flatMap(BioDBToOAF.convertEBILinksToOaf) @@ -165,7 +165,7 @@ class BioScholixTest extends AbstractVocabularyTest{ @Test def scholixResolvedToOAF():Unit ={ - val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/bio/scholix_resolved")).mkString + val records:String =Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/bio/scholix_resolved")).mkString records.lines.foreach(s => assertTrue(s.nonEmpty)) implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats diff --git a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/pangaea/PangaeaTransformTest.scala b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/pangaea/PangaeaTransformTest.scala index 053e4d63e..b90827e81 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/pangaea/PangaeaTransformTest.scala +++ b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/sx/pangaea/PangaeaTransformTest.scala @@ -1,6 +1,6 @@ package eu.dnetlib.dhp.sx.pangaea -import eu.dnetlib.sx.pangaea.PangaeaUtils +import eu.dnetlib.dhp.sx.graph.pangaea.PangaeaUtils import org.junit.jupiter.api.Test import java.util.TimeZone diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/crossref_links b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/crossref_links similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/crossref_links rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/crossref_links diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/ebi_links.gz b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/ebi_links.gz similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/ebi_links.gz rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/ebi_links.gz diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/pdb_dump b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/pdb_dump similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/pdb_dump rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/pdb_dump diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/pubmed/pubmed.xml b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/pubmed/pubmed.xml similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/pubmed/pubmed.xml rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/pubmed/pubmed.xml diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/pubmed/pubmed_dump b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/pubmed/pubmed_dump similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/pubmed/pubmed_dump rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/pubmed/pubmed_dump diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/scholix_resolved b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/scholix_resolved similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/scholix_resolved rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/scholix_resolved diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/uniprot_dump b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/uniprot_dump similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/bio/uniprot_dump rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/bio/uniprot_dump diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/ebi/rel1.json b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/ebi/rel1.json similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/ebi/rel1.json rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/ebi/rel1.json diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/ebi/rel_multiple.json b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/ebi/rel_multiple.json similarity index 100% rename from dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/ebi/rel_multiple.json rename to dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/sx/graph/ebi/rel_multiple.json