sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

[DOIBoost - Mapping] Changed the logic which brought in in the instance urls that should not be there: The urld of the doi in the json is reachable from the root (json/"URL") other urls where added from the links element. Now the mapping from the link element has been removed

This commit is contained in:
Miriam Baglioni 2021-11-03 16:43:36 +01:00
parent 683fe093cf
commit 96769b4481
1 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/Crossref2Oaf.scala
View File

@ -206,11 +206,16 @@ case object Crossref2Oaf {
else { else {
instance.setDateofacceptance(asField(createdDate.getValue)) instance.setDateofacceptance(asField(createdDate.getValue))
} }
val s: String = (json \ "URL").extract[String] val s: List[String] = List((json \ "URL").extract[String])
val links: List[String] = ((for {JString(url) <- json \ "link" \ "URL"} yield url) ::: List(s)).filter(p => p != null && p.toLowerCase().contains(doi.toLowerCase())).distinct // val links: List[String] = ((for {JString(url) <- json \ "link" \ "URL"} yield url) ::: List(s)).filter(p => p != null && p.toLowerCase().contains(doi.toLowerCase())).distinct
if (links.nonEmpty) { // if (links.nonEmpty) {
instance.setUrl(links.asJava) // instance.setUrl(links.asJava)
// }
if(s.nonEmpty)
{
instance.setUrl(s.asJava)
} }
result.setInstance(List(instance).asJava) result.setInstance(List(instance).asJava)
//IMPORTANT //IMPORTANT