|
|
|
@ -3,8 +3,8 @@ package eu.dnetlib.doiboost.mag
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser
|
|
|
|
|
import org.apache.commons.io.IOUtils
|
|
|
|
|
import org.apache.spark.SparkConf
|
|
|
|
|
import org.apache.spark.sql.{SaveMode, SparkSession}
|
|
|
|
|
import org.apache.spark.sql.types._
|
|
|
|
|
import org.apache.spark.sql.{SaveMode, SparkSession}
|
|
|
|
|
import org.slf4j.{Logger, LoggerFactory}
|
|
|
|
|
|
|
|
|
|
object SparkImportMagIntoDataset {
|
|
|
|
@ -24,13 +24,13 @@ object SparkImportMagIntoDataset {
|
|
|
|
|
"Affiliations" -> Tuple2("mag/Affiliations.txt", Seq("AffiliationId:long", "Rank:uint", "NormalizedName:string", "DisplayName:string", "GridId:string", "OfficialPage:string", "WikiPage:string", "PaperCount:long", "PaperFamilyCount:long", "CitationCount:long", "Iso3166Code:string", "Latitude:float?", "Longitude:float?", "CreatedDate:DateTime")),
|
|
|
|
|
"AuthorExtendedAttributes" -> Tuple2("mag/AuthorExtendedAttributes.txt", Seq("AuthorId:long", "AttributeType:int", "AttributeValue:string")),
|
|
|
|
|
"Authors" -> Tuple2("mag/Authors.txt", Seq("AuthorId:long", "Rank:uint", "NormalizedName:string", "DisplayName:string", "LastKnownAffiliationId:long?", "PaperCount:long", "PaperFamilyCount:long", "CitationCount:long", "CreatedDate:DateTime")),
|
|
|
|
|
"ConferenceInstances" -> Tuple2("mag/ConferenceInstances.txt", Seq("ConferenceInstanceId:long", "NormalizedName:string", "DisplayName:string", "ConferenceSeriesId:long", "Location:string", "OfficialUrl:string", "StartDate:DateTime?", "EndDate:DateTime?", "AbstractRegistrationDate:DateTime?", "SubmissionDeadlineDate:DateTime?", "NotificationDueDate:DateTime?", "FinalVersionDueDate:DateTime?", "PaperCount:long", "PaperFamilyCount:long" ,"CitationCount:long", "Latitude:float?", "Longitude:float?", "CreatedDate:DateTime")),
|
|
|
|
|
"ConferenceInstances" -> Tuple2("mag/ConferenceInstances.txt", Seq("ConferenceInstanceId:long", "NormalizedName:string", "DisplayName:string", "ConferenceSeriesId:long", "Location:string", "OfficialUrl:string", "StartDate:DateTime?", "EndDate:DateTime?", "AbstractRegistrationDate:DateTime?", "SubmissionDeadlineDate:DateTime?", "NotificationDueDate:DateTime?", "FinalVersionDueDate:DateTime?", "PaperCount:long", "PaperFamilyCount:long", "CitationCount:long", "Latitude:float?", "Longitude:float?", "CreatedDate:DateTime")),
|
|
|
|
|
"ConferenceSeries" -> Tuple2("mag/ConferenceSeries.txt", Seq("ConferenceSeriesId:long", "Rank:uint", "NormalizedName:string", "DisplayName:string", "PaperCount:long", "PaperFamilyCount:long", "CitationCount:long", "CreatedDate:DateTime")),
|
|
|
|
|
"EntityRelatedEntities" -> Tuple2("advanced/EntityRelatedEntities.txt", Seq("EntityId:long", "EntityType:string", "RelatedEntityId:long", "RelatedEntityType:string", "RelatedType:int", "Score:float")),
|
|
|
|
|
"FieldOfStudyChildren" -> Tuple2("advanced/FieldOfStudyChildren.txt", Seq("FieldOfStudyId:long", "ChildFieldOfStudyId:long")),
|
|
|
|
|
"FieldOfStudyExtendedAttributes" -> Tuple2("advanced/FieldOfStudyExtendedAttributes.txt", Seq("FieldOfStudyId:long", "AttributeType:int", "AttributeValue:string")),
|
|
|
|
|
"FieldsOfStudy" -> Tuple2("advanced/FieldsOfStudy.txt", Seq("FieldOfStudyId:long", "Rank:uint", "NormalizedName:string", "DisplayName:string", "MainType:string", "Level:int", "PaperCount:long", "PaperFamilyCount:long", "CitationCount:long", "CreatedDate:DateTime")),
|
|
|
|
|
"Journals" -> Tuple2("mag/Journals.txt", Seq("JournalId:long", "Rank:uint", "NormalizedName:string", "DisplayName:string", "Issn:string", "Publisher:string", "Webpage:string", "PaperCount:long", "PaperFamilyCount:long" ,"CitationCount:long", "CreatedDate:DateTime")),
|
|
|
|
|
"Journals" -> Tuple2("mag/Journals.txt", Seq("JournalId:long", "Rank:uint", "NormalizedName:string", "DisplayName:string", "Issn:string", "Publisher:string", "Webpage:string", "PaperCount:long", "PaperFamilyCount:long", "CitationCount:long", "CreatedDate:DateTime")),
|
|
|
|
|
"PaperAbstractsInvertedIndex" -> Tuple2("nlp/PaperAbstractsInvertedIndex.txt.*", Seq("PaperId:long", "IndexedAbstract:string")),
|
|
|
|
|
"PaperAuthorAffiliations" -> Tuple2("mag/PaperAuthorAffiliations.txt", Seq("PaperId:long", "AuthorId:long", "AffiliationId:long?", "AuthorSequenceNumber:uint", "OriginalAuthor:string", "OriginalAffiliation:string")),
|
|
|
|
|
"PaperCitationContexts" -> Tuple2("nlp/PaperCitationContexts.txt", Seq("PaperId:long", "PaperReferenceId:long", "CitationContext:string")),
|
|
|
|
@ -75,7 +75,6 @@ object SparkImportMagIntoDataset {
|
|
|
|
|
.master(parser.get("master")).getOrCreate()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stream.foreach { case (k, v) =>
|
|
|
|
|
val s: StructType = getSchema(k)
|
|
|
|
|
val df = spark.read
|