diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java index a8d09e4a7f..c157be51a0 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java @@ -30,6 +30,7 @@ import eu.dnetlib.dhp.schema.common.ModelSupport; import eu.dnetlib.dhp.schema.oaf.*; import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory; import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils; +import eu.dnetlib.dhp.schema.oaf.utils.PidType; public abstract class AbstractMdRecordToOafMapper { @@ -68,6 +69,17 @@ public abstract class AbstractMdRecordToOafMapper { nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3); } + // lowercase pidTypes as keys, normal casing for the values + protected static final Map pidTypeWithAuthority = new HashMap<>(); + + static { + IdentifierFactory.PID_AUTHORITY + .keySet() + .stream() + .forEach(entry -> pidTypeWithAuthority.put(entry.toString().toLowerCase(), entry.toString())); + + } + protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId, final boolean forceOriginalId) { this.vocs = vocs; diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java index 5781988e62..a25bcd47e8 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java @@ -22,9 +22,12 @@ import com.google.common.collect.Lists; import eu.dnetlib.dhp.common.PacePerson; import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup; +import eu.dnetlib.dhp.schema.common.ModelSupport; +import eu.dnetlib.dhp.schema.common.RelationInverse; import eu.dnetlib.dhp.schema.oaf.*; import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions; import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory; +import eu.dnetlib.dhp.schema.oaf.utils.PidType; public class OdfToOafMapper extends AbstractMdRecordToOafMapper { @@ -393,40 +396,53 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper { final List res = new ArrayList<>(); for (final Object o : doc - .selectNodes("//*[local-name()='relatedIdentifier' and ./@relatedIdentifierType='OPENAIRE']")) { + .selectNodes("//*[local-name()='relatedIdentifier']")) { - final String originalId = ((Node) o).getText(); + final String originalId = ((Node) o).getText().trim(); if (StringUtils.isNotBlank(originalId)) { - final String otherId = createOpenaireId(50, originalId, false); - final String type = ((Node) o).valueOf("@relationType"); - - if (type.equalsIgnoreCase(IS_SUPPLEMENT_TO)) { - res - .add( - getRelation( - docId, otherId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENT_TO, entity)); - res - .add( - getRelation( - otherId, docId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENTED_BY, entity)); - } else if (type.equalsIgnoreCase(IS_PART_OF)) { - res - .add( - getRelation( - docId, otherId, RESULT_RESULT, PART, IS_PART_OF, entity)); - res - .add( - getRelation( - otherId, docId, RESULT_RESULT, PART, HAS_PART, entity)); - } else { - // TODO catch more semantics + final String idType = ((Node) o).valueOf("@relatedIdentifierType"); + final String relType = ((Node) o).valueOf("@relationType"); + String otherId = guessRelatedIdentifier(idType, originalId); + if (StringUtils.isNotBlank(otherId)) { + res.addAll(getRelations(relType, docId, otherId, entity)); } + } } return res; } + protected String guessRelatedIdentifier(final String idType, final String value) { + if (StringUtils.isBlank(idType) || StringUtils.isBlank(value)) + return null; + if (idType.equalsIgnoreCase("OPENAIRE")) + return createOpenaireId(50, value, false); + if (pidTypeWithAuthority.containsKey(idType.toLowerCase())) { + return IdentifierFactory.idFromPid("50", pidTypeWithAuthority.get(idType.toLowerCase()), value, true); + } + return null; + + } + + protected List getRelations(final String reltype, final String entityId, final String otherId, + final OafEntity entity) { + final List res = new ArrayList<>(); + RelationInverse rel = ModelSupport.findRelation(reltype); + if (rel != null) { + res + .add( + getRelation( + entityId, otherId, rel.getRelType(), rel.getSubReltype(), rel.getRelClass(), entity)); + res + .add( + getRelation( + otherId, entityId, rel.getRelType(), rel.getSubReltype(), rel.getInverseRelClass(), entity)); + + } + return res; + } + @Override protected Qualifier prepareResourceType(final Document doc, final DataInfo info) { return prepareQualifier( diff --git a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java index 64b68e6af1..7552d1789c 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java +++ b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java @@ -579,8 +579,10 @@ class MappersTest { final List list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml); - assertEquals(1, list.size()); + assertEquals(3, list.size()); assertTrue(list.get(0) instanceof Software); + assertTrue(list.get(1) instanceof Relation); + assertTrue(list.get(2) instanceof Relation); final Software s = (Software) list.get(0); @@ -590,6 +592,22 @@ class MappersTest { assertTrue(s.getAuthor().size() > 0); assertTrue(s.getSubject().size() > 0); assertTrue(s.getInstance().size() > 0); + + final Relation r1 = (Relation) list.get(1); + final Relation r2 = (Relation) list.get(2); + + assertEquals(s.getId(), r1.getSource()); + assertEquals("50|doi_________::b453e7b4b2130ace57ff0c3db470a982", r1.getTarget()); + assertEquals(ModelConstants.RESULT_RESULT, r1.getRelType()); + assertEquals(ModelConstants.RELATIONSHIP, r1.getSubRelType()); + assertEquals(ModelConstants.IS_REFERENCED_BY, r1.getRelClass()); + + assertEquals(s.getId(), r2.getTarget()); + assertEquals("50|doi_________::b453e7b4b2130ace57ff0c3db470a982", r2.getSource()); + assertEquals(ModelConstants.RESULT_RESULT, r2.getRelType()); + assertEquals(ModelConstants.RELATIONSHIP, r2.getSubRelType()); + assertEquals(ModelConstants.REFERENCES, r2.getRelClass()); + } @Test @@ -912,13 +930,13 @@ class MappersTest { } @Test - void testROHub() throws IOException, DocumentException { + void testROHub() throws IOException { final String xml = IOUtils.toString(Objects.requireNonNull(getClass().getResourceAsStream("rohub.xml"))); final List list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml); System.out.println("***************"); System.out.println(new ObjectMapper().writeValueAsString(list)); System.out.println("***************"); -// final Dataset p = (Dataset) list.get(0); +// final OtherResearchProduct p = (OtherResearchProduct) list.get(0); // assertValidId(p.getId()); // assertValidId(p.getCollectedfrom().get(0).getKey()); // System.out.println(p.getTitle().get(0).getValue()); @@ -926,13 +944,36 @@ class MappersTest { } @Test - void testROHub2() throws IOException, DocumentException { + void testROHub2() throws IOException { final String xml = IOUtils .toString(Objects.requireNonNull(getClass().getResourceAsStream("rohub-modified.xml"))); final List list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml); System.out.println("***************"); System.out.println(new ObjectMapper().writeValueAsString(list)); System.out.println("***************"); + assertEquals(7, list.size()); + final OtherResearchProduct p = (OtherResearchProduct) list.get(0); + assertValidId(p.getId()); + assertValidId(p.getCollectedfrom().get(0).getKey()); + assertEquals("50|w3id________::afc7592914ae190a50570db90f55f9c2", p.getId()); + assertTrue(StringUtils.isNotBlank(p.getTitle().get(0).getValue())); + assertEquals("w3id", (p.getPid().get(0).getQualifier().getClassid())); + assertEquals("https://w3id.org/ro-id/0ab171a7-45c5-4194-82d4-850955504bca", (p.getPid().get(0).getValue())); + + assertEquals(1, list.stream().filter(o -> o instanceof OtherResearchProduct).count()); + assertEquals(6, list.stream().filter(o -> o instanceof Relation).count()); + + for (Oaf oaf : list) { + if (oaf instanceof Relation) { + String source = ((Relation) oaf).getSource(); + String target = ((Relation) oaf).getTarget(); + assertNotEquals(source, target); + assertTrue(source.equals(p.getId()) || target.equals(p.getId())); + assertNotNull(((Relation) oaf).getSubRelType()); + assertNotNull(((Relation) oaf).getRelClass()); + assertNotNull(((Relation) oaf).getRelType()); + } + } } @Test diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_dataset.xml b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_dataset.xml index 4f41ee6eac..4633d62c39 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_dataset.xml +++ b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_dataset.xml @@ -69,7 +69,6 @@ - 10.5281/zenodo.3234525 https://zenodo.org/communities/epfl 1.0.0 diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_software.xml b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_software.xml index 6a9170ce17..387b1ee866 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_software.xml +++ b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_software.xml @@ -20,7 +20,7 @@ bio.tools http://maplab.imppc.org/chainy/ - 10.1093/bioinformatics/btw839 + 10.1093/bioinformatics/btw839 https://bio.tools/ diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/rohub-modified.xml b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/rohub-modified.xml index 95d65ac8d5..ce846a2cf9 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/rohub-modified.xml +++ b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/rohub-modified.xml @@ -30,6 +30,9 @@ https://w3id.org/ro-id/0ab171a7-45c5-4194-82d4-850955504bca/resources/6d3427a8-352e-49f4-9796-f618c44dc16d + + fsh_____4119::afc7592914ae190a50570db90f55f9c3 + RO-crate diff --git a/pom.xml b/pom.xml index a1b26966e1..9b60b90789 100644 --- a/pom.xml +++ b/pom.xml @@ -807,7 +807,7 @@ 3.3.3 3.4.2 [2.12,3.0) - [3.14.0] + [3.15.0] [4.0.3] [6.0.5] [3.1.6]