Merge pull request 'Change the access right in DoiBoost' (#126) from doiboosi_accessright into beta

Reviewed-on: D-Net/dnet-hadoop#126

dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/DoiBoostMappingUtil.scala

@@ -1,12 +1,16 @@
 package eu.dnetlib.doiboost
 
+import java.time.LocalDate
+import java.time.format.DateTimeFormatter
+
 import eu.dnetlib.dhp.schema.action.AtomicAction
-import eu.dnetlib.dhp.schema.oaf.{AccessRight, DataInfo, Dataset, Field, Instance, KeyValue, Oaf, Organization, Publication, Qualifier, Relation, Result, StructuredProperty}
+import eu.dnetlib.dhp.schema.oaf.{AccessRight, DataInfo, Dataset, Field, Instance, KeyValue, Oaf, OpenAccessRoute, Organization, Publication, Qualifier, Relation, Result, StructuredProperty}
 import eu.dnetlib.dhp.utils.DHPUtils
 import org.apache.commons.lang3.StringUtils
 import com.fasterxml.jackson.databind.ObjectMapper
 import eu.dnetlib.dhp.schema.common.ModelConstants
 import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils
+import eu.dnetlib.doiboost.DoiBoostMappingUtil.{getClosedAccessQualifier, getEmbargoedAccessQualifier, getUnknownQualifier}
 import org.json4s
 import org.json4s.DefaultFormats
 import org.json4s.jackson.JsonMethods.parse
@@ -118,10 +122,74 @@ object DoiBoostMappingUtil {
   }
 
 
+  def decideAccessRight(lic : Field[String], date:String) : AccessRight = {
+    if(lic == null){
+      //Default value Unknown
+      return getUnknownQualifier()
+    }
+    val license : String = lic.getValue
+    //CC licenses
+    if(license.startsWith("cc") ||
+      license.startsWith("http://creativecommons.org/licenses") ||
+      license.startsWith("https://creativecommons.org/licenses") ||
+
+      //ACS Publications Author choice licenses (considered OPEN also by Unpaywall)
+      license.equals("http://pubs.acs.org/page/policy/authorchoice_ccby_termsofuse.html") ||
+      license.equals("http://pubs.acs.org/page/policy/authorchoice_termsofuse.html") ||
+      license.equals("http://pubs.acs.org/page/policy/authorchoice_ccbyncnd_termsofuse.html") ||
+
+      //APA (considered OPEN also by Unpaywall)
+      license.equals("http://www.apa.org/pubs/journals/resources/open-access.aspx")){
+
+      val oaq : AccessRight = getOpenAccessQualifier()
+      oaq.setOpenAccessRoute(OpenAccessRoute.hybrid)
+      return oaq
+    }
+
+    //OUP (BUT ONLY AFTER 12 MONTHS FROM THE PUBLICATION DATE, OTHERWISE THEY ARE EMBARGOED)
+    if(license.equals("https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model")){
+      val now = java.time.LocalDate.now
+
+      try{
+        val pub_date = LocalDate.parse(date, DateTimeFormatter.ofPattern("yyyy-MM-dd"))
+        if (((now.toEpochDay - pub_date.toEpochDay)/365.0) > 1){
+          val oaq : AccessRight = getOpenAccessQualifier()
+          oaq.setOpenAccessRoute(OpenAccessRoute.hybrid)
+          return oaq
+        }
+        else{
+          return getEmbargoedAccessQualifier()
+        }
+      }catch {
+        case e: Exception => {
+          try{
+          val pub_date = LocalDate.parse(date, DateTimeFormatter.ofPattern("yyyy-MM-dd'T'HH:mm:ss'Z'"))
+          if (((now.toEpochDay - pub_date.toEpochDay)/365.0) > 1){
+            val oaq : AccessRight = getOpenAccessQualifier()
+            oaq.setOpenAccessRoute(OpenAccessRoute.hybrid)
+            return oaq
+          }
+          else{
+            return getEmbargoedAccessQualifier()
+          }
+          }catch{
+            case ex: Exception => return getClosedAccessQualifier()
+          }
+        }
+
+      }
+
+    }
+
+    return getClosedAccessQualifier()
+
+  }
+
 
 
   def getOpenAccessQualifier():AccessRight = {
-    OafMapperUtils.accessRight("OPEN","Open Access", ModelConstants.DNET_ACCESS_MODES, ModelConstants.DNET_ACCESS_MODES)
+
+    OafMapperUtils.accessRight(ModelConstants.ACCESS_RIGHT_OPEN,"Open Access", ModelConstants.DNET_ACCESS_MODES, ModelConstants.DNET_ACCESS_MODES)
   }
 
   def getRestrictedQualifier():AccessRight = {
@@ -129,6 +197,20 @@ object DoiBoostMappingUtil {
   }
 
 
+  def getUnknownQualifier():AccessRight = {
+    OafMapperUtils.accessRight(ModelConstants.UNKNOWN, ModelConstants.NOT_AVAILABLE,ModelConstants.DNET_ACCESS_MODES, ModelConstants.DNET_ACCESS_MODES)
+  }
+
+
+  def getEmbargoedAccessQualifier():AccessRight = {
+    OafMapperUtils.accessRight("EMBARGO","Embargo",ModelConstants.DNET_ACCESS_MODES, ModelConstants.DNET_ACCESS_MODES)
+  }
+
+  def getClosedAccessQualifier():AccessRight = {
+    OafMapperUtils.accessRight("CLOSED","Closed Access", ModelConstants.DNET_ACCESS_MODES, ModelConstants.DNET_ACCESS_MODES)
+  }
+
+
   def extractInstance(r:Result):Option[Instance] = {
     r.getInstance().asScala.find(i => i.getInstancetype != null && i.getInstancetype.getClassid.nonEmpty)
   }
@@ -150,10 +232,11 @@ object DoiBoostMappingUtil {
       if (item != null) {
         hb.setValue(item.officialname)
         hb.setKey(generateDSId(item.id))
-        if (item.openAccess)
+        if (item.openAccess) {
           i.setAccessright(getOpenAccessQualifier())
-        val ar = getOpenAccessQualifier()
-        publication.setBestaccessright(OafMapperUtils.qualifier(ar.getClassid, ar.getClassname, ar.getSchemeid, ar.getSchemename))
+          i.getAccessright.setOpenAccessRoute(OpenAccessRoute.gold)
+        }
+
       }
       else {
         hb = ModelConstants.UNKNOWN_REPOSITORY
@@ -161,17 +244,8 @@ object DoiBoostMappingUtil {
       i.setHostedby(hb)
     })
 
-    val ar = publication.getInstance().asScala.filter(i => i.getInstancetype != null && i.getAccessright!= null && i.getAccessright.getClassid!= null).map(f=> f.getAccessright.getClassid)
-    if (ar.nonEmpty) {
-      if(ar.contains(ModelConstants.ACCESS_RIGHT_OPEN)){
-        val ar = getOpenAccessQualifier()
-        publication.setBestaccessright(OafMapperUtils.qualifier(ar.getClassid, ar.getClassname, ar.getSchemeid, ar.getSchemename))
-      }
-      else {
-        val ar = getRestrictedQualifier()
-        publication.setBestaccessright(OafMapperUtils.qualifier(ar.getClassid, ar.getClassname, ar.getSchemeid, ar.getSchemename))
-      }
-    }
+    publication.setBestaccessright(OafMapperUtils.createBestAccessRights(publication.getInstance()))
+
     publication
   }
 

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

@@ -4,7 +4,7 @@ import eu.dnetlib.dhp.schema.common.ModelConstants
 import eu.dnetlib.dhp.schema.oaf._
 import eu.dnetlib.dhp.schema.oaf.utils.{IdentifierFactory, OafMapperUtils}
 import eu.dnetlib.dhp.utils.DHPUtils
-import eu.dnetlib.doiboost.DoiBoostMappingUtil._
+import eu.dnetlib.doiboost.DoiBoostMappingUtil.{decideAccessRight, _}
 import org.apache.commons.lang.StringUtils
 import org.json4s
 import org.json4s.DefaultFormats
@@ -168,12 +168,22 @@ case object Crossref2Oaf {
     // Mapping instance
     val instance = new Instance()
     val license = for {
-      JString(lic) <- json \ "license" \ "URL"
-    } yield asField(lic)
-    val l = license.filter(d => StringUtils.isNotBlank(d.getValue))
-    if (l.nonEmpty)
-      instance.setLicense(l.head)
-
+      JObject(license) <- json \ "license"
+      JField("URL", JString(lic)) <- license
+      JField("content-version", JString(content_version)) <- license
+    } yield (asField(lic), content_version)
+    val l = license.filter(d => StringUtils.isNotBlank(d._1.getValue))
+    if (l.nonEmpty){
+      if (l exists (d => d._2.equals("vor"))){
+        for(d <- l){
+          if (d._2.equals("vor")){
+            instance.setLicense(d._1)
+          }
+        }
+      }
+      else{
+        instance.setLicense(l.head._1)}
+    }
 
     // Ticket #6281 added pid to Instance
     instance.setPid(result.getPid)
@@ -185,7 +195,7 @@ case object Crossref2Oaf {
         OafMapperUtils.qualifier("0001", "peerReviewed", ModelConstants.DNET_REVIEW_LEVELS, ModelConstants.DNET_REVIEW_LEVELS))
     }
 
-    instance.setAccessright(getRestrictedQualifier())
+    instance.setAccessright(decideAccessRight(instance.getLicense, result.getDateofacceptance.getValue))
     instance.setInstancetype(OafMapperUtils.qualifier(cobjCategory.substring(0, 4), cobjCategory.substring(5), ModelConstants.DNET_PUBLICATION_RESOURCE, ModelConstants.DNET_PUBLICATION_RESOURCE))
     result.setResourcetype(OafMapperUtils.qualifier(cobjCategory.substring(0, 4),   cobjCategory.substring(5), ModelConstants.DNET_PUBLICATION_RESOURCE, ModelConstants.DNET_PUBLICATION_RESOURCE))
 

dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/uw/UnpayWallToOAF.scala

@@ -11,6 +11,7 @@ import org.slf4j.{Logger, LoggerFactory}
 
 import scala.collection.JavaConverters._
 import eu.dnetlib.doiboost.DoiBoostMappingUtil._
+import eu.dnetlib.doiboost.uw.UnpayWallToOAF.get_unpaywall_color
 
 
 
@@ -23,6 +24,21 @@ case class OALocation(evidence:Option[String], host_type:Option[String], is_best
 object UnpayWallToOAF {
   val logger: Logger = LoggerFactory.getLogger(getClass)
 
+
+  def get_unpaywall_color(input:String):Option[OpenAccessRoute] = {
+    if(input == null || input.equalsIgnoreCase("close"))
+      return None
+    if(input.equalsIgnoreCase("green"))
+      return Some(OpenAccessRoute.green)
+    if(input.equalsIgnoreCase("bronze"))
+      return Some(OpenAccessRoute.bronze)
+    if(input.equalsIgnoreCase("hybrid"))
+      return Some(OpenAccessRoute.hybrid)
+    else
+      return Some(OpenAccessRoute.gold)
+
+  }
+
   def get_color(is_oa:Boolean, location: OALocation, journal_is_oa:Boolean):Option[OpenAccessRoute] = {
     if (is_oa) {
       if (location.host_type.isDefined) {
@@ -65,7 +81,7 @@ object UnpayWallToOAF {
 
     val oaLocation:OALocation = (json \ "best_oa_location").extractOrElse[OALocation](null)
 
-    val colour = get_color(is_oa, oaLocation, journal_is_oa)
+    val colour = get_unpaywall_color((json \ "oa_status").extractOrElse[String](null))
 
     pub.setCollectedfrom(List(createUnpayWallCollectedFrom()).asJava)
     pub.setDataInfo(generateDataInfo())

dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/process/oozie_app/workflow.xml

@@ -99,7 +99,7 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
-                --conf spark.sql.shuffle.partitions=3840
+                --conf spark.sql.shuffle.partitions=7680
                 --conf spark.extraListeners=${spark2ExtraListeners}
                 --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                 --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
@@ -124,7 +124,7 @@
                 --executor-memory=${sparkExecutorIntersectionMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
-                --conf spark.sql.shuffle.partitions=3840
+                --conf spark.sql.shuffle.partitions=7680
                 --conf spark.extraListeners=${spark2ExtraListeners}
                 --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                 --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}

dhp-workflows/dhp-doiboost/src/test/java/eu/dnetlib/doiboost/crossref/CrossrefMappingTest.scala

@@ -492,6 +492,124 @@ class CrossrefMappingTest {
 
   }
 
+  @Test
+  def testLicenseVorClosed() :Unit = {
+    val json = Source.fromInputStream(getClass.getResourceAsStream("publication_license_vor.json")).mkString
+
+
+    assertNotNull(json)
+    assertFalse(json.isEmpty);
+
+    val resultList: List[Oaf] = Crossref2Oaf.convert(json)
+
+    assertTrue(resultList.nonEmpty)
+
+
+    val item : Result = resultList.filter(p => p.isInstanceOf[Result]).head.asInstanceOf[Result]
+
+    mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
+    println(mapper.writeValueAsString(item))
+
+    assertTrue(item.getInstance().asScala exists (i => i.getLicense.getValue.equals("https://www.springer.com/vor")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getClassid.equals("CLOSED")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getOpenAccessRoute == null))
+
+
 
 
+  }
+
+  @Test
+  def testLicenseOpen() :Unit = {
+    val json = Source.fromInputStream(getClass.getResourceAsStream("publication_license_open.json")).mkString
+
+
+    assertNotNull(json)
+    assertFalse(json.isEmpty);
+
+    val resultList: List[Oaf] = Crossref2Oaf.convert(json)
+
+    assertTrue(resultList.nonEmpty)
+
+
+    val item : Result = resultList.filter(p => p.isInstanceOf[Result]).head.asInstanceOf[Result]
+
+    assertTrue(item.getInstance().asScala exists (i => i.getLicense.getValue.equals("http://pubs.acs.org/page/policy/authorchoice_ccby_termsofuse.html")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getClassid.equals("OPEN")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getOpenAccessRoute == OpenAccessRoute.hybrid))
+    mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
+    println(mapper.writeValueAsString(item))
+
+  }
+
+  @Test
+  def testLicenseEmbargoOpen() :Unit = {
+    val json = Source.fromInputStream(getClass.getResourceAsStream("publication_license_embargo_open.json")).mkString
+
+
+    assertNotNull(json)
+    assertFalse(json.isEmpty);
+
+    val resultList: List[Oaf] = Crossref2Oaf.convert(json)
+
+    assertTrue(resultList.nonEmpty)
+
+
+    val item : Result = resultList.filter(p => p.isInstanceOf[Result]).head.asInstanceOf[Result]
+
+    assertTrue(item.getInstance().asScala exists (i => i.getLicense.getValue.equals("https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getClassid.equals("OPEN")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getOpenAccessRoute == OpenAccessRoute.hybrid))
+    mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
+    println(mapper.writeValueAsString(item))
+
+  }
+
+  @Test
+  def testLicenseEmbargo() :Unit = {
+    val json = Source.fromInputStream(getClass.getResourceAsStream("publication_license_embargo.json")).mkString
+
+
+    assertNotNull(json)
+    assertFalse(json.isEmpty);
+
+    val resultList: List[Oaf] = Crossref2Oaf.convert(json)
+
+    assertTrue(resultList.nonEmpty)
+
+
+    val item : Result = resultList.filter(p => p.isInstanceOf[Result]).head.asInstanceOf[Result]
+
+    assertTrue(item.getInstance().asScala exists (i => i.getLicense.getValue.equals("https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getClassid.equals("EMBARGO")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getOpenAccessRoute == null))
+    mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
+    println(mapper.writeValueAsString(item))
+
+  }
+
+
+  @Test
+  def testLicenseEmbargoDateTime() :Unit = {
+    val json = Source.fromInputStream(getClass.getResourceAsStream("publication_license_embargo_datetime.json")).mkString
+
+
+    assertNotNull(json)
+    assertFalse(json.isEmpty);
+
+    val resultList: List[Oaf] = Crossref2Oaf.convert(json)
+
+    assertTrue(resultList.nonEmpty)
+
+
+    val item : Result = resultList.filter(p => p.isInstanceOf[Result]).head.asInstanceOf[Result]
+
+    assertTrue(item.getInstance().asScala exists (i => i.getLicense.getValue.equals("https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getClassid.equals("EMBARGO")))
+    assertTrue(item.getInstance().asScala exists (i => i.getAccessright.getOpenAccessRoute == null))
+    mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
+    println(mapper.writeValueAsString(item))
+
+  }
+
 }