From a7763d249205f71a6ae274eb3e4b7c3f53619aaf Mon Sep 17 00:00:00 2001 From: Sandro La Bruzzo Date: Fri, 12 Nov 2021 09:56:45 +0100 Subject: [PATCH] removed alternate identifier in resolutionMap --- .../resolution/SparkResolveRelation.scala | 24 ++++++++++++------- 1 file changed, 16 insertions(+), 8 deletions(-) diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/resolution/SparkResolveRelation.scala b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/resolution/SparkResolveRelation.scala index 9e44e017c..cd517dd5e 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/resolution/SparkResolveRelation.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/resolution/SparkResolveRelation.scala @@ -96,6 +96,21 @@ object SparkResolveRelation { .text(s"$graphBasePath/relation") } + def extractInstanceCF(input: String): List[(String, String)] = { + implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats + lazy val json: json4s.JValue = parse(input) + val result: List[(String, String)] = for { + JObject(iObj) <- json \ "instance" + JField("collectedfrom", JObject(cf)) <- iObj + JField("instancetype", JObject(instancetype)) <- iObj + JField("value", JString(collectedFrom)) <- cf + JField("classname", JString(classname)) <- instancetype + } yield (classname, collectedFrom) + + result + + } + def extractPidsFromRecord(input: String): (String, List[(String, String)]) = { implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats @@ -108,14 +123,7 @@ object SparkResolveRelation { JField("classid", JString(pidType)) <- qualifier } yield (pidValue, pidType) - val alternateIds: List[(String, String)] = for { - JObject(pids) <- json \\ "alternateIdentifier" - JField("value", JString(pidValue)) <- pids - JField("qualifier", JObject(qualifier)) <- pids - JField("classid", JString(pidType)) <- qualifier - } yield (pidValue, pidType) - - (id, result ::: alternateIds) + (id, result) }