add a default datainfo if missing

dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java

@@ -1,15 +1,46 @@
 package eu.dnetlib.dhp.oa.graph.raw;
 
-import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.*;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.createOpenaireId;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.dataInfo;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.field;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.journal;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.keyValue;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listFields;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.oaiIProvenance;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
+import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Date;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
 
-import eu.dnetlib.dhp.schema.oaf.*;
-import java.util.*;
 import org.apache.commons.lang3.StringUtils;
 import org.dom4j.Document;
 import org.dom4j.DocumentFactory;
 import org.dom4j.DocumentHelper;
 import org.dom4j.Node;
 
+import eu.dnetlib.dhp.schema.oaf.Author;
+import eu.dnetlib.dhp.schema.oaf.DataInfo;
+import eu.dnetlib.dhp.schema.oaf.Dataset;
+import eu.dnetlib.dhp.schema.oaf.Field;
+import eu.dnetlib.dhp.schema.oaf.GeoLocation;
+import eu.dnetlib.dhp.schema.oaf.Instance;
+import eu.dnetlib.dhp.schema.oaf.Journal;
+import eu.dnetlib.dhp.schema.oaf.KeyValue;
+import eu.dnetlib.dhp.schema.oaf.OAIProvenance;
+import eu.dnetlib.dhp.schema.oaf.Oaf;
+import eu.dnetlib.dhp.schema.oaf.OtherResearchProduct;
+import eu.dnetlib.dhp.schema.oaf.Publication;
+import eu.dnetlib.dhp.schema.oaf.Qualifier;
+import eu.dnetlib.dhp.schema.oaf.Relation;
+import eu.dnetlib.dhp.schema.oaf.Result;
+import eu.dnetlib.dhp.schema.oaf.Software;
+import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
+
 public abstract class AbstractMdRecordToOafMapper {
 
 	protected final Map<String, String> code2name;
@@ -18,17 +49,15 @@ public abstract class AbstractMdRecordToOafMapper {
 			qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
 
 	protected static final Qualifier PUBLICATION_RESULTTYPE_QUALIFIER =
-            qualifier(
+			qualifier("publication", "publication", "dnet:result_typologies", "dnet:result_typologies");
-                    "publication",
-                    "publication",
-                    "dnet:result_typologies",
-                    "dnet:result_typologies");
 	protected static final Qualifier DATASET_RESULTTYPE_QUALIFIER =
 			qualifier("dataset", "dataset", "dnet:result_typologies", "dnet:result_typologies");
 	protected static final Qualifier SOFTWARE_RESULTTYPE_QUALIFIER =
 			qualifier("software", "software", "dnet:result_typologies", "dnet:result_typologies");
 	protected static final Qualifier OTHER_RESULTTYPE_QUALIFIER =
 			qualifier("other", "other", "dnet:result_typologies", "dnet:result_typologies");
+	protected static final Qualifier REPOSITORY_QUALIFIER =
+			qualifier("sysimport:crosswalk:repository", "sysimport:crosswalk:repository", "dnet:provenanceActions", "dnet:provenanceActions");
 
 	protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
 		this.code2name = code2name;
@@ -47,22 +76,15 @@ public abstract class AbstractMdRecordToOafMapper {
 			DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
 
 			final Document doc =
-                    DocumentHelper.parseText(
+					DocumentHelper.parseText(xml.replaceAll("http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3"));
-                            xml.replaceAll(
-                                    "http://datacite.org/schema/kernel-4",
-                                    "http://datacite.org/schema/kernel-3"));
 
 			final String type = doc.valueOf("//dr:CobjCategory/@type");
 			final KeyValue collectedFrom =
-                    keyValue(
+					keyValue(doc.valueOf("//oaf:collectedFrom/@id"), doc.valueOf("//oaf:collectedFrom/@name"));
-                            doc.valueOf("//oaf:collectedFrom/@id"),
-                            doc.valueOf("//oaf:collectedFrom/@name"));
 			final KeyValue hostedBy =
 					StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
 							? collectedFrom
-                            : keyValue(
+							: keyValue(doc.valueOf("//oaf:hostedBy/@id"), doc.valueOf("//oaf:hostedBy/@name"));
-                                    doc.valueOf("//oaf:hostedBy/@id"),
-                                    doc.valueOf("//oaf:hostedBy/@name"));
 
 			final DataInfo info = prepareDataInfo(doc);
 			final long lastUpdateTimestamp = new Date().getTime();
@@ -192,14 +214,7 @@ public abstract class AbstractMdRecordToOafMapper {
 		r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
 		r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
 		r.setCollectedfrom(Arrays.asList(collectedFrom));
-        r.setPid(
+		r.setPid(prepareListStructProps(doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
-                prepareListStructProps(
-                        doc,
-                        "//oaf:identifier",
-                        "@identifierType",
-                        "dnet:pid_types",
-                        "dnet:pid_types",
-                        info));
 		r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
 		r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
 		r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
@@ -228,7 +243,10 @@ public abstract class AbstractMdRecordToOafMapper {
 	protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
 
 	protected abstract List<Instance> prepareInstances(
-            Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
+			Document doc,
+			DataInfo info,
+			KeyValue collectedfrom,
+			KeyValue hostedby);
 
 	protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
 
@@ -253,28 +271,34 @@ public abstract class AbstractMdRecordToOafMapper {
 	protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
 
 	protected abstract List<Field<String>> prepareOtherResearchProductTools(
-            Document doc, DataInfo info);
+			Document doc,
+			DataInfo info);
 
 	protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
-            Document doc, DataInfo info);
+			Document doc,
+			DataInfo info);
 
 	protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
-            Document doc, DataInfo info);
+			Document doc,
+			DataInfo info);
 
 	protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
 
 	protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
 
 	protected abstract List<StructuredProperty> prepareSoftwareLicenses(
-            Document doc, DataInfo info);
+			Document doc,
+			DataInfo info);
 
 	protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
-            Document doc, DataInfo info);
+			Document doc,
+			DataInfo info);
 
 	protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
 
 	protected abstract Field<String> prepareDatasetMetadataVersionNumber(
-            Document doc, DataInfo info);
+			Document doc,
+			DataInfo info);
 
 	protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
 
@@ -298,27 +322,16 @@ public abstract class AbstractMdRecordToOafMapper {
 			final String sp = n.valueOf("@sp");
 			final String vol = n.valueOf("@vol");
 			final String edition = n.valueOf("@edition");
-            if (StringUtils.isNotBlank(name)) {
+			if (StringUtils.isNotBlank(name)) { return journal(name, issnPrinted, issnOnline, issnLinking, ep, iss, sp, vol, edition, null, null, info); }
-                return journal(
-                        name,
-                        issnPrinted,
-                        issnOnline,
-                        issnLinking,
-                        ep,
-                        iss,
-                        sp,
-                        vol,
-                        edition,
-                        null,
-                        null,
-                        info);
-            }
 		}
 		return null;
 	}
 
 	protected Qualifier prepareQualifier(
-            final Node node, final String xpath, final String schemeId, final String schemeName) {
+			final Node node,
+			final String xpath,
+			final String schemeId,
+			final String schemeName) {
 		final String classId = node.valueOf(xpath);
 		final String className = code2name.get(classId);
 		return qualifier(classId, className, schemeId, schemeName);
@@ -336,15 +349,16 @@ public abstract class AbstractMdRecordToOafMapper {
 			final Node n = (Node) o;
 			final String classId = n.valueOf(xpathClassId);
 			final String className = code2name.get(classId);
-            res.add(
+			res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info));
-                    structuredProperty(
-                            n.getText(), classId, className, schemeId, schemeName, info));
 		}
 		return res;
 	}
 
 	protected List<StructuredProperty> prepareListStructProps(
-            final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
+			final Node node,
+			final String xpath,
+			final Qualifier qualifier,
+			final DataInfo info) {
 		final List<StructuredProperty> res = new ArrayList<>();
 		for (final Object o : node.selectNodes(xpath)) {
 			final Node n = (Node) o;
@@ -354,52 +368,37 @@ public abstract class AbstractMdRecordToOafMapper {
 	}
 
 	protected List<StructuredProperty> prepareListStructProps(
-            final Node node, final String xpath, final DataInfo info) {
+			final Node node,
+			final String xpath,
+			final DataInfo info) {
 		final List<StructuredProperty> res = new ArrayList<>();
 		for (final Object o : node.selectNodes(xpath)) {
 			final Node n = (Node) o;
-            res.add(
+			res.add(structuredProperty(n.getText(), n.valueOf("@classid"), n.valueOf("@classname"), n.valueOf("@schemeid"), n.valueOf("@schemename"), info));
-                    structuredProperty(
-                            n.getText(),
-                            n.valueOf("@classid"),
-                            n.valueOf("@classname"),
-                            n.valueOf("@schemeid"),
-                            n.valueOf("@schemename"),
-                            info));
 		}
 		return res;
 	}
 
 	protected OAIProvenance prepareOAIprovenance(final Document doc) {
 		final Node n =
-                doc.selectSingleNode(
+				doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
-                        "//*[local-name()='provenance']/*[local-name()='originDescription']");
 
-        if (n == null) {
+		if (n == null) { return null; }
-            return null;
-        }
 
 		final String identifier = n.valueOf("./*[local-name()='identifier']");
-        final String baseURL = n.valueOf("./*[local-name()='baseURL']");
+		final String baseURL = n.valueOf("./*[local-name()='baseURL']");;
-        ;
+		final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");;
-        final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
-        ;
 		final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
-        final String datestamp = n.valueOf("./*[local-name()='datestamp']");
+		final String datestamp = n.valueOf("./*[local-name()='datestamp']");;
-        ;
+		final String harvestDate = n.valueOf("@harvestDate");;
-        final String harvestDate = n.valueOf("@harvestDate");
-        ;
 
-        return oaiIProvenance(
+		return oaiIProvenance(identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
-                identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
 	}
 
 	protected DataInfo prepareDataInfo(final Document doc) {
 		final Node n = doc.selectSingleNode("//oaf:datainfo");
 
-        if (n == null) {
+		if (n == null) { return dataInfo(false, null, false, false, REPOSITORY_QUALIFIER, "0.9"); }
-            return null;
-        }
 
 		final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
 		final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
@@ -412,13 +411,7 @@ public abstract class AbstractMdRecordToOafMapper {
 		final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
 		final String trust = n.valueOf("./oaf:trust");
 
-        return dataInfo(
+		return dataInfo(deletedbyinference, inferenceprovenance, inferred, false, qualifier(paClassId, paClassName, paSchemeId, paSchemeName), trust);
-                deletedbyinference,
-                inferenceprovenance,
-                inferred,
-                false,
-                qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
-                trust);
 	}
 
 	protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
@@ -426,7 +419,9 @@ public abstract class AbstractMdRecordToOafMapper {
 	}
 
 	protected List<Field<String>> prepareListFields(
-            final Node node, final String xpath, final DataInfo info) {
+			final Node node,
+			final String xpath,
+			final DataInfo info) {
 		return listFields(info, prepareListString(node, xpath));
 	}