core utilities in dhp-common moved in external module dhp-schemas

This commit is contained in:
Claudio Atzori 2021-04-27 15:44:01 +02:00
parent ac77a245a3
commit 5afa7d3e0c
49 changed files with 46 additions and 1751 deletions

View File

@ -10,8 +10,8 @@ import org.apache.commons.lang3.StringUtils;
import com.google.common.collect.Maps;
import eu.dnetlib.dhp.schema.oaf.OafMapperUtils;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
public class Vocabulary implements Serializable {

View File

@ -7,8 +7,8 @@ import java.util.stream.Collectors;
import org.apache.commons.lang3.StringUtils;
import eu.dnetlib.dhp.schema.oaf.OafMapperUtils;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;

View File

@ -1,399 +0,0 @@
package eu.dnetlib.dhp.schema.oaf;
import java.util.*;
import java.util.function.Function;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import org.apache.commons.lang3.StringUtils;
import com.clearspring.analytics.util.Lists;
import com.google.common.collect.Sets;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.utils.PidBlacklistProvider;
public class CleaningFunctions {
public static final String DOI_PREFIX_REGEX = "(^10\\.|\\/10.)";
public static final String DOI_PREFIX = "10.";
public static final String ORCID_PREFIX_REGEX = "^http(s?):\\/\\/orcid\\.org\\/";
public static final String CLEANING_REGEX = "(?:\\n|\\r|\\t)";
public static final Set<String> PID_BLACKLIST = new HashSet<>();
public static final String INVALID_AUTHOR_REGEX = ".*deactivated.*";
public static final String TITLE_FILTER_REGEX = "[.*test.*\\W\\d]";
public static final int TITLE_FILTER_RESIDUAL_LENGTH = 10;
static {
PID_BLACKLIST.add("none");
PID_BLACKLIST.add("na");
}
public static <T extends Oaf> T fixVocabularyNames(T value) {
if (value instanceof Datasource) {
// nothing to clean here
} else if (value instanceof Project) {
// nothing to clean here
} else if (value instanceof Organization) {
Organization o = (Organization) value;
if (Objects.nonNull(o.getCountry())) {
fixVocabName(o.getCountry(), ModelConstants.DNET_COUNTRY_TYPE);
}
} else if (value instanceof Relation) {
// nothing to clean here
} else if (value instanceof Result) {
Result r = (Result) value;
fixVocabName(r.getLanguage(), ModelConstants.DNET_LANGUAGES);
fixVocabName(r.getResourcetype(), ModelConstants.DNET_DATA_CITE_RESOURCE);
fixVocabName(r.getBestaccessright(), ModelConstants.DNET_ACCESS_MODES);
if (Objects.nonNull(r.getSubject())) {
r.getSubject().forEach(s -> fixVocabName(s.getQualifier(), ModelConstants.DNET_SUBJECT_TYPOLOGIES));
}
if (Objects.nonNull(r.getInstance())) {
for (Instance i : r.getInstance()) {
fixVocabName(i.getAccessright(), ModelConstants.DNET_ACCESS_MODES);
fixVocabName(i.getRefereed(), ModelConstants.DNET_REVIEW_LEVELS);
}
}
if (Objects.nonNull(r.getAuthor())) {
r.getAuthor().stream().filter(Objects::nonNull).forEach(a -> {
if (Objects.nonNull(a.getPid())) {
a.getPid().stream().filter(Objects::nonNull).forEach(p -> {
fixVocabName(p.getQualifier(), ModelConstants.DNET_PID_TYPES);
});
}
});
}
if (value instanceof Publication) {
} else if (value instanceof eu.dnetlib.dhp.schema.oaf.Dataset) {
} else if (value instanceof OtherResearchProduct) {
} else if (value instanceof Software) {
}
}
return value;
}
public static <T extends Oaf> boolean filter(T value) {
if (value instanceof Datasource) {
// nothing to evaluate here
} else if (value instanceof Project) {
// nothing to evaluate here
} else if (value instanceof Organization) {
// nothing to evaluate here
} else if (value instanceof Relation) {
// nothing to clean here
} else if (value instanceof Result) {
Result r = (Result) value;
if (Objects.nonNull(r.getTitle()) && r.getTitle().isEmpty()) {
return false;
}
if (value instanceof Publication) {
} else if (value instanceof eu.dnetlib.dhp.schema.oaf.Dataset) {
} else if (value instanceof OtherResearchProduct) {
} else if (value instanceof Software) {
}
}
return true;
}
public static <T extends Oaf> T cleanup(T value) {
if (value instanceof Datasource) {
// nothing to clean here
} else if (value instanceof Project) {
// nothing to clean here
} else if (value instanceof Organization) {
Organization o = (Organization) value;
if (Objects.isNull(o.getCountry()) || StringUtils.isBlank(o.getCountry().getClassid())) {
o.setCountry(ModelConstants.UNKNOWN_COUNTRY);
}
} else if (value instanceof Relation) {
// nothing to clean here
} else if (value instanceof Result) {
Result r = (Result) value;
if (Objects.nonNull(r.getPublisher()) && StringUtils.isBlank(r.getPublisher().getValue())) {
r.setPublisher(null);
}
if (Objects.isNull(r.getLanguage()) || StringUtils.isBlank(r.getLanguage().getClassid())) {
r
.setLanguage(
qualifier("und", "Undetermined", ModelConstants.DNET_LANGUAGES));
}
if (Objects.nonNull(r.getSubject())) {
r
.setSubject(
r
.getSubject()
.stream()
.filter(Objects::nonNull)
.filter(sp -> StringUtils.isNotBlank(sp.getValue()))
.filter(sp -> Objects.nonNull(sp.getQualifier()))
.filter(sp -> StringUtils.isNotBlank(sp.getQualifier().getClassid()))
.map(CleaningFunctions::cleanValue)
.collect(Collectors.toList()));
}
if (Objects.nonNull(r.getTitle())) {
r
.setTitle(
r
.getTitle()
.stream()
.filter(Objects::nonNull)
.filter(sp -> StringUtils.isNotBlank(sp.getValue()))
.filter(
sp -> sp
.getValue()
.toLowerCase()
.replaceAll(TITLE_FILTER_REGEX, "")
.length() > TITLE_FILTER_RESIDUAL_LENGTH)
.map(CleaningFunctions::cleanValue)
.collect(Collectors.toList()));
}
if (Objects.nonNull(r.getDescription())) {
r
.setDescription(
r
.getDescription()
.stream()
.filter(Objects::nonNull)
.filter(sp -> StringUtils.isNotBlank(sp.getValue()))
.map(CleaningFunctions::cleanValue)
.collect(Collectors.toList()));
}
if (Objects.nonNull(r.getPid())) {
r.setPid(processPidCleaning(r.getPid()));
}
if (Objects.isNull(r.getResourcetype()) || StringUtils.isBlank(r.getResourcetype().getClassid())) {
r
.setResourcetype(
qualifier(ModelConstants.UNKNOWN, "Unknown", ModelConstants.DNET_DATA_CITE_RESOURCE));
}
if (Objects.nonNull(r.getInstance())) {
for (Instance i : r.getInstance()) {
Optional
.ofNullable(i.getPid())
.ifPresent(pid -> {
final Set<StructuredProperty> pids = pid
.stream()
.filter(Objects::nonNull)
.filter(p -> StringUtils.isNotBlank(p.getValue()))
.collect(Collectors.toCollection(HashSet::new));
Optional
.ofNullable(i.getAlternateIdentifier())
.ifPresent(altId -> {
final Set<StructuredProperty> altIds = altId
.stream()
.filter(Objects::nonNull)
.filter(p -> StringUtils.isNotBlank(p.getValue()))
.collect(Collectors.toCollection(HashSet::new));
i.setAlternateIdentifier(Lists.newArrayList(Sets.difference(altIds, pids)));
});
});
if (Objects.isNull(i.getAccessright()) || StringUtils.isBlank(i.getAccessright().getClassid())) {
i
.setAccessright(
accessRight(
ModelConstants.UNKNOWN, ModelConstants.NOT_AVAILABLE,
ModelConstants.DNET_ACCESS_MODES));
}
if (Objects.isNull(i.getHostedby()) || StringUtils.isBlank(i.getHostedby().getKey())) {
i.setHostedby(ModelConstants.UNKNOWN_REPOSITORY);
}
if (Objects.isNull(i.getRefereed())) {
i.setRefereed(qualifier("0000", "Unknown", ModelConstants.DNET_REVIEW_LEVELS));
}
}
}
if (Objects.isNull(r.getBestaccessright()) || StringUtils.isBlank(r.getBestaccessright().getClassid())) {
Qualifier bestaccessrights = OafMapperUtils.createBestAccessRights(r.getInstance());
if (Objects.isNull(bestaccessrights)) {
r
.setBestaccessright(
qualifier(
ModelConstants.UNKNOWN, ModelConstants.NOT_AVAILABLE,
ModelConstants.DNET_ACCESS_MODES));
} else {
r.setBestaccessright(bestaccessrights);
}
}
if (Objects.nonNull(r.getAuthor())) {
final List<Author> authors = Lists.newArrayList();
for (Author a : r.getAuthor()) {
if (Objects.isNull(a.getPid())) {
a.setPid(Lists.newArrayList());
} else {
a
.setPid(
a
.getPid()
.stream()
.filter(Objects::nonNull)
.filter(p -> Objects.nonNull(p.getQualifier()))
.filter(p -> StringUtils.isNotBlank(p.getValue()))
.map(p -> {
p.setValue(p.getValue().trim().replaceAll(ORCID_PREFIX_REGEX, ""));
return p;
})
.filter(p -> StringUtils.isNotBlank(p.getValue()))
.collect(
Collectors
.toMap(
StructuredProperty::getValue, Function.identity(), (p1, p2) -> p1,
LinkedHashMap::new))
.values()
.stream()
.collect(Collectors.toList()));
}
if (StringUtils.isBlank(a.getFullname())) {
if (StringUtils.isNotBlank(a.getName()) && StringUtils.isNotBlank(a.getSurname())) {
a.setFullname(a.getSurname() + ", " + a.getName());
}
}
if (StringUtils.isNotBlank(a.getFullname()) && isValidAuthorName(a)) {
authors.add(a);
}
}
boolean nullRank = authors
.stream()
.anyMatch(a -> Objects.isNull(a.getRank()));
if (nullRank) {
int i = 1;
for (Author author : authors) {
author.setRank(i++);
}
}
r.setAuthor(authors);
}
if (value instanceof Publication) {
} else if (value instanceof eu.dnetlib.dhp.schema.oaf.Dataset) {
} else if (value instanceof OtherResearchProduct) {
} else if (value instanceof Software) {
}
}
return value;
}
private static boolean isValidAuthorName(Author a) {
return !Stream
.of(a.getFullname(), a.getName(), a.getSurname())
.filter(s -> s != null && !s.isEmpty())
.collect(Collectors.joining(""))
.toLowerCase()
.matches(INVALID_AUTHOR_REGEX);
}
private static List<StructuredProperty> processPidCleaning(List<StructuredProperty> pids) {
return pids
.stream()
.filter(Objects::nonNull)
.filter(sp -> StringUtils.isNotBlank(StringUtils.trim(sp.getValue())))
.filter(sp -> !PID_BLACKLIST.contains(sp.getValue().trim().toLowerCase()))
.filter(sp -> Objects.nonNull(sp.getQualifier()))
.filter(sp -> StringUtils.isNotBlank(sp.getQualifier().getClassid()))
.map(CleaningFunctions::normalizePidValue)
.filter(CleaningFunctions::pidFilter)
.collect(Collectors.toList());
}
protected static StructuredProperty cleanValue(StructuredProperty s) {
s.setValue(s.getValue().replaceAll(CLEANING_REGEX, " "));
return s;
}
protected static Field<String> cleanValue(Field<String> s) {
s.setValue(s.getValue().replaceAll(CLEANING_REGEX, " "));
return s;
}
// HELPERS
private static void fixVocabName(Qualifier q, String vocabularyName) {
if (Objects.nonNull(q) && StringUtils.isBlank(q.getSchemeid())) {
q.setSchemeid(vocabularyName);
q.setSchemename(vocabularyName);
}
}
private static AccessRight accessRight(String classid, String classname, String scheme) {
return OafMapperUtils
.accessRight(
classid, classname, scheme, scheme);
}
private static Qualifier qualifier(String classid, String classname, String scheme) {
return OafMapperUtils
.qualifier(
classid, classname, scheme, scheme);
}
/**
* Utility method that filter PID values on a per-type basis.
* @param s the PID whose value will be checked.
* @return false if the pid matches the filter criteria, true otherwise.
*/
public static boolean pidFilter(StructuredProperty s) {
final String pidValue = s.getValue();
if (Objects.isNull(s.getQualifier()) ||
StringUtils.isBlank(pidValue) ||
StringUtils.isBlank(pidValue.replaceAll("(?:\\n|\\r|\\t|\\s)", ""))) {
return false;
}
if (CleaningFunctions.PID_BLACKLIST.contains(pidValue)) {
return false;
}
if (PidBlacklistProvider.getBlacklist(s.getQualifier().getClassid()).contains(pidValue)) {
return false;
}
return true;
}
/**
* Utility method that normalises PID values on a per-type basis.
* @param pid the PID whose value will be normalised.
* @return the PID containing the normalised value.
*/
public static StructuredProperty normalizePidValue(StructuredProperty pid) {
String value = Optional
.ofNullable(pid.getValue())
.map(String::trim)
.orElseThrow(() -> new IllegalArgumentException("PID value cannot be empty"));
switch (pid.getQualifier().getClassid()) {
// TODO add cleaning for more PID types as needed
case "doi":
pid.setValue(value.toLowerCase().replaceFirst(DOI_PREFIX_REGEX, DOI_PREFIX));
break;
}
return pid;
}
}

View File

@ -1,22 +0,0 @@
package eu.dnetlib.dhp.schema.oaf;
public class ModelHardLimits {
public static final String LAYOUT = "index";
public static final String INTERPRETATION = "openaire";
public static final String SEPARATOR = "-";
public static final int MAX_EXTERNAL_ENTITIES = 50;
public static final int MAX_AUTHORS = 200;
public static final int MAX_AUTHOR_FULLNAME_LENGTH = 1000;
public static final int MAX_TITLE_LENGTH = 5000;
public static final int MAX_TITLES = 10;
public static final int MAX_ABSTRACT_LENGTH = 150000;
public static final int MAX_INSTANCES = 10;
public static String getCollectionName(String format) {
return format + SEPARATOR + LAYOUT + SEPARATOR + INTERPRETATION;
}
}

View File

@ -1,364 +0,0 @@
package eu.dnetlib.dhp.schema.oaf;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import java.util.*;
import java.util.concurrent.ConcurrentHashMap;
import java.util.function.Function;
import java.util.function.Predicate;
import java.util.stream.Collectors;
import org.apache.commons.lang3.StringUtils;
import eu.dnetlib.dhp.schema.common.AccessRightComparator;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.utils.DHPUtils;
public class OafMapperUtils {
public static Oaf merge(final Oaf left, final Oaf right) {
if (ModelSupport.isSubClass(left, OafEntity.class)) {
return mergeEntities((OafEntity) left, (OafEntity) right);
} else if (ModelSupport.isSubClass(left, Relation.class)) {
((Relation) left).mergeFrom((Relation) right);
} else {
throw new RuntimeException("invalid Oaf type:" + left.getClass().getCanonicalName());
}
return left;
}
public static OafEntity mergeEntities(OafEntity left, OafEntity right) {
if (ModelSupport.isSubClass(left, Result.class)) {
return mergeResults((Result) left, (Result) right);
} else if (ModelSupport.isSubClass(left, Datasource.class)) {
((Datasource) left).mergeFrom((Datasource) right);
} else if (ModelSupport.isSubClass(left, Organization.class)) {
((Organization) left).mergeFrom((Organization) right);
} else if (ModelSupport.isSubClass(left, Project.class)) {
((Project) left).mergeFrom((Project) right);
} else {
throw new RuntimeException("invalid OafEntity subtype:" + left.getClass().getCanonicalName());
}
return left;
}
public static Result mergeResults(Result left, Result right) {
if (new ResultTypeComparator().compare(left, right) < 0) {
left.mergeFrom(right);
return left;
} else {
right.mergeFrom(left);
return right;
}
}
public static KeyValue keyValue(final String k, final String v) {
final KeyValue kv = new KeyValue();
kv.setKey(k);
kv.setValue(v);
return kv;
}
public static List<KeyValue> listKeyValues(final String... s) {
if (s.length % 2 > 0) {
throw new RuntimeException("Invalid number of parameters (k,v,k,v,....)");
}
final List<KeyValue> list = new ArrayList<>();
for (int i = 0; i < s.length; i += 2) {
list.add(keyValue(s[i], s[i + 1]));
}
return list;
}
public static <T> Field<T> field(final T value, final DataInfo info) {
if (value == null || StringUtils.isBlank(value.toString())) {
return null;
}
final Field<T> field = new Field<>();
field.setValue(value);
field.setDataInfo(info);
return field;
}
public static List<Field<String>> listFields(final DataInfo info, final String... values) {
return Arrays
.stream(values)
.map(v -> field(v, info))
.filter(Objects::nonNull)
.filter(distinctByKey(f -> f.getValue()))
.collect(Collectors.toList());
}
public static List<Field<String>> listFields(final DataInfo info, final List<String> values) {
return values
.stream()
.map(v -> field(v, info))
.filter(Objects::nonNull)
.filter(distinctByKey(f -> f.getValue()))
.collect(Collectors.toList());
}
public static Qualifier unknown(final String schemeid, final String schemename) {
return qualifier("UNKNOWN", "Unknown", schemeid, schemename);
}
public static AccessRight accessRight(
final String classid,
final String classname,
final String schemeid,
final String schemename) {
return accessRight(classid, classname, schemeid, schemename, null);
}
public static AccessRight accessRight(
final String classid,
final String classname,
final String schemeid,
final String schemename,
final OpenAccessRoute openAccessRoute) {
final AccessRight accessRight = new AccessRight();
accessRight.setClassid(classid);
accessRight.setClassname(classname);
accessRight.setSchemeid(schemeid);
accessRight.setSchemename(schemename);
accessRight.setOpenAccessRoute(openAccessRoute);
return accessRight;
}
public static Qualifier qualifier(
final String classid,
final String classname,
final String schemeid,
final String schemename) {
final Qualifier q = new Qualifier();
q.setClassid(classid);
q.setClassname(classname);
q.setSchemeid(schemeid);
q.setSchemename(schemename);
return q;
}
public static Qualifier qualifier(final Qualifier qualifier) {
final Qualifier q = new Qualifier();
q.setClassid(qualifier.getClassid());
q.setClassname(qualifier.getClassname());
q.setSchemeid(qualifier.getSchemeid());
q.setSchemename(qualifier.getSchemename());
return q;
}
public static StructuredProperty structuredProperty(
final String value,
final String classid,
final String classname,
final String schemeid,
final String schemename,
final DataInfo dataInfo) {
return structuredProperty(value, qualifier(classid, classname, schemeid, schemename), dataInfo);
}
public static StructuredProperty structuredProperty(
final String value,
final Qualifier qualifier,
final DataInfo dataInfo) {
if (value == null) {
return null;
}
final StructuredProperty sp = new StructuredProperty();
sp.setValue(value);
sp.setQualifier(qualifier);
sp.setDataInfo(dataInfo);
return sp;
}
public static ExtraInfo extraInfo(
final String name,
final String value,
final String typology,
final String provenance,
final String trust) {
final ExtraInfo info = new ExtraInfo();
info.setName(name);
info.setValue(value);
info.setTypology(typology);
info.setProvenance(provenance);
info.setTrust(trust);
return info;
}
public static OAIProvenance oaiIProvenance(
final String identifier,
final String baseURL,
final String metadataNamespace,
final Boolean altered,
final String datestamp,
final String harvestDate) {
final OriginDescription desc = new OriginDescription();
desc.setIdentifier(identifier);
desc.setBaseURL(baseURL);
desc.setMetadataNamespace(metadataNamespace);
desc.setAltered(altered);
desc.setDatestamp(datestamp);
desc.setHarvestDate(harvestDate);
final OAIProvenance p = new OAIProvenance();
p.setOriginDescription(desc);
return p;
}
public static Journal journal(
final String name,
final String issnPrinted,
final String issnOnline,
final String issnLinking,
final DataInfo dataInfo) {
return journal(
name,
issnPrinted,
issnOnline,
issnLinking,
null,
null,
null,
null,
null,
null,
null,
dataInfo);
}
public static Journal journal(
final String name,
final String issnPrinted,
final String issnOnline,
final String issnLinking,
final String ep,
final String iss,
final String sp,
final String vol,
final String edition,
final String conferenceplace,
final String conferencedate,
final DataInfo dataInfo) {
if (StringUtils.isNotBlank(name)
|| StringUtils.isNotBlank(issnPrinted)
|| StringUtils.isNotBlank(issnOnline)
|| StringUtils.isNotBlank(issnLinking)) {
final Journal j = new Journal();
j.setName(name);
j.setIssnPrinted(issnPrinted);
j.setIssnOnline(issnOnline);
j.setIssnLinking(issnLinking);
j.setEp(ep);
j.setIss(iss);
j.setSp(sp);
j.setVol(vol);
j.setEdition(edition);
j.setConferenceplace(conferenceplace);
j.setConferencedate(conferencedate);
j.setDataInfo(dataInfo);
return j;
} else {
return null;
}
}
public static DataInfo dataInfo(
final Boolean deletedbyinference,
final String inferenceprovenance,
final Boolean inferred,
final Boolean invisible,
final Qualifier provenanceaction,
final String trust) {
final DataInfo d = new DataInfo();
d.setDeletedbyinference(deletedbyinference);
d.setInferenceprovenance(inferenceprovenance);
d.setInferred(inferred);
d.setInvisible(invisible);
d.setProvenanceaction(provenanceaction);
d.setTrust(trust);
return d;
}
public static String createOpenaireId(
final int prefix,
final String originalId,
final boolean to_md5) {
if (StringUtils.isBlank(originalId)) {
return null;
} else if (to_md5) {
final String nsPrefix = StringUtils.substringBefore(originalId, "::");
final String rest = StringUtils.substringAfter(originalId, "::");
return String.format("%s|%s::%s", prefix, nsPrefix, DHPUtils.md5(rest));
} else {
return String.format("%s|%s", prefix, originalId);
}
}
public static String createOpenaireId(
final String type,
final String originalId,
final boolean to_md5) {
switch (type) {
case "datasource":
return createOpenaireId(10, originalId, to_md5);
case "organization":
return createOpenaireId(20, originalId, to_md5);
case "person":
return createOpenaireId(30, originalId, to_md5);
case "project":
return createOpenaireId(40, originalId, to_md5);
default:
return createOpenaireId(50, originalId, to_md5);
}
}
public static String asString(final Object o) {
return o == null ? "" : o.toString();
}
public static <T> Predicate<T> distinctByKey(
final Function<? super T, ?> keyExtractor) {
final Map<Object, Boolean> seen = new ConcurrentHashMap<>();
return t -> seen.putIfAbsent(keyExtractor.apply(t), Boolean.TRUE) == null;
}
public static Qualifier createBestAccessRights(final List<Instance> instanceList) {
return getBestAccessRights(instanceList);
}
protected static Qualifier getBestAccessRights(final List<Instance> instanceList) {
if (instanceList != null) {
final Optional<AccessRight> min = instanceList
.stream()
.map(i -> i.getAccessright())
.min(new AccessRightComparator<>());
final Qualifier rights = min.isPresent() ? qualifier(min.get()) : new Qualifier();
if (StringUtils.isBlank(rights.getClassid())) {
rights.setClassid(UNKNOWN);
}
if (StringUtils.isBlank(rights.getClassname())
|| UNKNOWN.equalsIgnoreCase(rights.getClassname())) {
rights.setClassname(NOT_AVAILABLE);
}
if (StringUtils.isBlank(rights.getSchemeid())) {
rights.setSchemeid(DNET_ACCESS_MODES);
}
if (StringUtils.isBlank(rights.getSchemename())) {
rights.setSchemename(DNET_ACCESS_MODES);
}
return rights;
}
return null;
}
}

View File

@ -1,78 +0,0 @@
package eu.dnetlib.dhp.schema.oaf;
import static eu.dnetlib.dhp.schema.common.ModelConstants.CROSSREF_ID;
import java.util.Comparator;
import java.util.HashSet;
import java.util.Optional;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import com.google.common.collect.Sets;
import eu.dnetlib.dhp.schema.common.ModelConstants;
public class ResultTypeComparator implements Comparator<Result> {
@Override
public int compare(Result left, Result right) {
if (left == null && right == null)
return 0;
if (left == null)
return 1;
if (right == null)
return -1;
HashSet<String> lCf = getCollectedFromIds(left);
HashSet<String> rCf = getCollectedFromIds(right);
if (lCf.contains(CROSSREF_ID) && !rCf.contains(CROSSREF_ID)) {
return -1;
}
if (!lCf.contains(CROSSREF_ID) && rCf.contains(CROSSREF_ID)) {
return 1;
}
String lClass = left.getResulttype().getClassid();
String rClass = right.getResulttype().getClassid();
if (lClass.equals(rClass))
return 0;
if (lClass.equals(ModelConstants.PUBLICATION_RESULTTYPE_CLASSID))
return -1;
if (rClass.equals(ModelConstants.PUBLICATION_RESULTTYPE_CLASSID))
return 1;
if (lClass.equals(ModelConstants.DATASET_RESULTTYPE_CLASSID))
return -1;
if (rClass.equals(ModelConstants.DATASET_RESULTTYPE_CLASSID))
return 1;
if (lClass.equals(ModelConstants.SOFTWARE_RESULTTYPE_CLASSID))
return -1;
if (rClass.equals(ModelConstants.SOFTWARE_RESULTTYPE_CLASSID))
return 1;
if (lClass.equals(ModelConstants.ORP_RESULTTYPE_CLASSID))
return -1;
if (rClass.equals(ModelConstants.ORP_RESULTTYPE_CLASSID))
return 1;
// Else (but unlikely), lexicographical ordering will do.
return lClass.compareTo(rClass);
}
protected HashSet<String> getCollectedFromIds(Result left) {
return Optional
.ofNullable(left.getCollectedfrom())
.map(
cf -> cf
.stream()
.map(c -> c.getKey())
.collect(Collectors.toCollection(HashSet::new)))
.orElse(new HashSet<>());
}
}

View File

@ -1,212 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import static com.google.common.base.Preconditions.checkArgument;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import java.io.Serializable;
import java.util.*;
import java.util.function.Function;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import org.apache.commons.lang3.StringUtils;
import com.google.common.collect.HashBiMap;
import com.google.common.collect.Maps;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.utils.DHPUtils;
/**
* Factory class for OpenAIRE identifiers in the Graph
*/
public class IdentifierFactory implements Serializable {
public static final String ID_SEPARATOR = "::";
public static final String ID_PREFIX_SEPARATOR = "|";
public static final int ID_PREFIX_LEN = 12;
/**
* Declares the associations PID_TYPE -> [DATASOURCE ID, NAME] considered authoritative for that PID_TYPE
*/
public static final Map<PidType, HashBiMap<String, String>> PID_AUTHORITY = Maps.newHashMap();
static {
PID_AUTHORITY.put(PidType.doi, HashBiMap.create());
PID_AUTHORITY.get(PidType.doi).put(CROSSREF_ID, "Crossref");
PID_AUTHORITY.get(PidType.doi).put(DATACITE_ID, "Datacite");
PID_AUTHORITY.put(PidType.pmc, HashBiMap.create());
PID_AUTHORITY.get(PidType.pmc).put(EUROPE_PUBMED_CENTRAL_ID, "Europe PubMed Central");
PID_AUTHORITY.get(PidType.pmc).put(PUBMED_CENTRAL_ID, "PubMed Central");
PID_AUTHORITY.put(PidType.pmid, HashBiMap.create());
PID_AUTHORITY.get(PidType.pmid).put(EUROPE_PUBMED_CENTRAL_ID, "Europe PubMed Central");
PID_AUTHORITY.get(PidType.pmid).put(PUBMED_CENTRAL_ID, "PubMed Central");
PID_AUTHORITY.put(PidType.arXiv, HashBiMap.create());
PID_AUTHORITY.get(PidType.arXiv).put(ARXIV_ID, "arXiv.org e-Print Archive");
}
public static List<StructuredProperty> getPids(List<StructuredProperty> pid, KeyValue collectedFrom) {
return pidFromInstance(pid, collectedFrom, true).distinct().collect(Collectors.toList());
}
public static <T extends Result> String createDOIBoostIdentifier(T entity) {
if (entity == null)
return null;
StructuredProperty pid = null;
if (entity.getPid() != null) {
pid = entity
.getPid()
.stream()
.filter(Objects::nonNull)
.filter(s -> s.getQualifier() != null && "doi".equalsIgnoreCase(s.getQualifier().getClassid()))
.filter(CleaningFunctions::pidFilter)
.findAny()
.orElse(null);
} else {
if (entity.getInstance() != null) {
pid = entity
.getInstance()
.stream()
.filter(i -> i.getPid() != null)
.flatMap(i -> i.getPid().stream())
.filter(CleaningFunctions::pidFilter)
.findAny()
.orElse(null);
}
}
if (pid != null)
return idFromPid(entity, pid, true);
return null;
}
/**
* Creates an identifier from the most relevant PID (if available) provided by a known PID authority in the given
* entity T. Returns entity.id when none of the PIDs meet the selection criteria is available.
*
* @param entity the entity providing PIDs and a default ID.
* @param <T> the specific entity type. Currently Organization and Result subclasses are supported.
* @param md5 indicates whether should hash the PID value or not.
* @return an identifier from the most relevant PID, entity.id otherwise
*/
public static <T extends OafEntity> String createIdentifier(T entity, boolean md5) {
checkArgument(StringUtils.isNoneBlank(entity.getId()), "missing entity identifier");
final Map<String, Set<StructuredProperty>> pids = extractPids(entity);
return pids
.values()
.stream()
.flatMap(s -> s.stream())
.min(new PidComparator<>(entity))
.map(
min -> Optional
.ofNullable(pids.get(min.getQualifier().getClassid()))
.map(
p -> p
.stream()
.sorted(new PidValueComparator())
.findFirst()
.map(s -> idFromPid(entity, s, md5))
.orElseGet(entity::getId))
.orElseGet(entity::getId))
.orElseGet(entity::getId);
}
private static <T extends OafEntity> Map<String, Set<StructuredProperty>> extractPids(T entity) {
if (entity instanceof Result) {
return Optional
.ofNullable(((Result) entity).getInstance())
.map(
instance -> mapPids(instance))
.orElse(new HashMap<>());
} else {
return entity
.getPid()
.stream()
.map(CleaningFunctions::normalizePidValue)
.filter(CleaningFunctions::pidFilter)
.collect(
Collectors
.groupingBy(
p -> p.getQualifier().getClassid(),
Collectors.mapping(p -> p, Collectors.toCollection(HashSet::new))));
}
}
private static Map<String, Set<StructuredProperty>> mapPids(List<Instance> instance) {
return instance
.stream()
.map(i -> pidFromInstance(i.getPid(), i.getCollectedfrom(), false))
.flatMap(Function.identity())
.collect(
Collectors
.groupingBy(
p -> p.getQualifier().getClassid(),
Collectors.mapping(p -> p, Collectors.toCollection(HashSet::new))));
}
private static Stream<StructuredProperty> pidFromInstance(List<StructuredProperty> pid, KeyValue collectedFrom,
boolean mapHandles) {
return Optional
.ofNullable(pid)
.map(
pp -> pp
.stream()
// filter away PIDs provided by a DS that is not considered an authority for the
// given PID Type
.filter(p -> {
return shouldFilterPid(collectedFrom, p, mapHandles);
})
.map(CleaningFunctions::normalizePidValue)
.filter(CleaningFunctions::pidFilter))
.orElse(Stream.empty());
}
private static boolean shouldFilterPid(KeyValue collectedFrom, StructuredProperty p, boolean mapHandles) {
final PidType pType = PidType.tryValueOf(p.getQualifier().getClassid());
return (mapHandles && pType.equals(PidType.handle)) || Optional.ofNullable(collectedFrom).isPresent() &&
Optional
.ofNullable(PID_AUTHORITY.get(pType))
.map(authorities -> {
return authorities.containsKey(collectedFrom.getKey())
|| authorities.containsValue(collectedFrom.getValue());
})
.orElse(false);
}
/**
* @see {@link IdentifierFactory#createIdentifier(OafEntity, boolean)}
*/
public static <T extends OafEntity> String createIdentifier(T entity) {
return createIdentifier(entity, true);
}
private static <T extends OafEntity> String idFromPid(T entity, StructuredProperty s, boolean md5) {
return new StringBuilder()
.append(ModelSupport.getIdPrefix(entity.getClass()))
.append(ID_PREFIX_SEPARATOR)
.append(createPrefix(s.getQualifier().getClassid()))
.append(ID_SEPARATOR)
.append(md5 ? DHPUtils.md5(s.getValue()) : s.getValue())
.toString();
}
// create the prefix (length = 12)
private static String createPrefix(String pidType) {
StringBuilder prefix = new StringBuilder(StringUtils.left(pidType, ID_PREFIX_LEN));
while (prefix.length() < ID_PREFIX_LEN) {
prefix.append("_");
}
return prefix.substring(0, ID_PREFIX_LEN);
}
}

View File

@ -1,38 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.util.Comparator;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class OrganizationPidComparator implements Comparator<StructuredProperty> {
@Override
public int compare(StructuredProperty left, StructuredProperty right) {
PidType lClass = PidType.tryValueOf(left.getQualifier().getClassid());
PidType rClass = PidType.tryValueOf(right.getQualifier().getClassid());
if (lClass.equals(PidType.openorgs))
return -1;
if (rClass.equals(PidType.openorgs))
return 1;
if (lClass.equals(PidType.GRID))
return -1;
if (rClass.equals(PidType.GRID))
return 1;
if (lClass.equals(PidType.mag_id))
return -1;
if (rClass.equals(PidType.mag_id))
return 1;
if (lClass.equals(PidType.urn))
return -1;
if (rClass.equals(PidType.urn))
return 1;
return 0;
}
}

View File

@ -1,8 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.util.HashMap;
import java.util.HashSet;
public class PidBlacklist extends HashMap<String, HashSet<String>> {
}

View File

@ -1,37 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.io.IOException;
import java.util.HashSet;
import java.util.Optional;
import java.util.Set;
import org.apache.commons.io.IOUtils;
import com.fasterxml.jackson.databind.ObjectMapper;
public class PidBlacklistProvider {
private static final PidBlacklist blacklist;
static {
try {
String json = IOUtils.toString(IdentifierFactory.class.getResourceAsStream("pid_blacklist.json"));
blacklist = new ObjectMapper().readValue(json, PidBlacklist.class);
} catch (IOException e) {
throw new RuntimeException(e);
}
}
public static PidBlacklist getBlacklist() {
return blacklist;
}
public static Set<String> getBlacklist(String pidType) {
return Optional
.ofNullable(getBlacklist().get(pidType))
.orElse(new HashSet<>());
}
}

View File

@ -1,48 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.util.Comparator;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.OafEntity;
import eu.dnetlib.dhp.schema.oaf.Organization;
import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class PidComparator<T extends OafEntity> implements Comparator<StructuredProperty> {
private T entity;
public PidComparator(T entity) {
this.entity = entity;
}
@Override
public int compare(StructuredProperty left, StructuredProperty right) {
if (left == null && right == null)
return 0;
if (left == null)
return 1;
if (right == null)
return -1;
if (ModelSupport.isSubClass(entity, Result.class)) {
return compareResultPids(left, right);
}
if (ModelSupport.isSubClass(entity, Organization.class)) {
return compareOrganizationtPids(left, right);
}
// Else (but unlikely), lexicographical ordering will do.
return left.getQualifier().getClassid().compareTo(right.getQualifier().getClassid());
}
private int compareResultPids(StructuredProperty left, StructuredProperty right) {
return new ResultPidComparator().compare(left, right);
}
private int compareOrganizationtPids(StructuredProperty left, StructuredProperty right) {
return new OrganizationPidComparator().compare(left, right);
}
}

View File

@ -1,29 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import org.apache.commons.lang3.EnumUtils;
public enum PidType {
// Result
doi, pmid, pmc, handle, arXiv, nct, pdb,
// Organization
openorgs, corda, corda_h2020, GRID, mag_id, urn,
// Used by dedup
undefined, original;
public static boolean isValid(String type) {
return EnumUtils.isValidEnum(PidType.class, type);
}
public static PidType tryValueOf(String s) {
try {
return PidType.valueOf(s);
} catch (Exception e) {
return PidType.original;
}
}
}

View File

@ -1,33 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.util.Comparator;
import java.util.Optional;
import eu.dnetlib.dhp.schema.oaf.*;
public class PidValueComparator implements Comparator<StructuredProperty> {
@Override
public int compare(StructuredProperty left, StructuredProperty right) {
if (left == null && right == null)
return 0;
if (left == null)
return 1;
if (right == null)
return -1;
StructuredProperty l = CleaningFunctions.normalizePidValue(left);
StructuredProperty r = CleaningFunctions.normalizePidValue(right);
return Optional
.ofNullable(l.getValue())
.map(
lv -> Optional
.ofNullable(r.getValue())
.map(rv -> lv.compareTo(rv))
.orElse(-1))
.orElse(1);
}
}

View File

@ -1,53 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.util.Comparator;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class ResultPidComparator implements Comparator<StructuredProperty> {
@Override
public int compare(StructuredProperty left, StructuredProperty right) {
PidType lClass = PidType.tryValueOf(left.getQualifier().getClassid());
PidType rClass = PidType.tryValueOf(right.getQualifier().getClassid());
if (lClass.equals(PidType.doi))
return -1;
if (rClass.equals(PidType.doi))
return 1;
if (lClass.equals(PidType.pmid))
return -1;
if (rClass.equals(PidType.pmid))
return 1;
if (lClass.equals(PidType.pmc))
return -1;
if (rClass.equals(PidType.pmc))
return 1;
if (lClass.equals(PidType.handle))
return -1;
if (rClass.equals(PidType.handle))
return 1;
if (lClass.equals(PidType.arXiv))
return -1;
if (rClass.equals(PidType.arXiv))
return 1;
if (lClass.equals(PidType.nct))
return -1;
if (rClass.equals(PidType.nct))
return 1;
if (lClass.equals(PidType.pdb))
return -1;
if (rClass.equals(PidType.pdb))
return 1;
return 0;
}
}

View File

@ -1,69 +0,0 @@
package eu.dnetlib.dhp.schema.oaf;
import static org.junit.jupiter.api.Assertions.*;
import java.io.IOException;
import java.util.HashSet;
import java.util.List;
import java.util.stream.Collectors;
import org.apache.commons.io.IOUtils;
import org.jetbrains.annotations.NotNull;
import org.junit.jupiter.api.Assertions;
import org.junit.jupiter.api.Test;
import com.fasterxml.jackson.databind.DeserializationFeature;
import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import it.unimi.dsi.fastutil.Hash;
public class OafMapperUtilsTest {
private static ObjectMapper OBJECT_MAPPER = new ObjectMapper()
.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);
@Test
public void testMergePubs() throws IOException {
Publication p1 = read("publication_1.json", Publication.class);
Publication p2 = read("publication_2.json", Publication.class);
Dataset d1 = read("dataset_1.json", Dataset.class);
Dataset d2 = read("dataset_2.json", Dataset.class);
assertEquals(p1.getCollectedfrom().size(), 1);
assertEquals(p1.getCollectedfrom().get(0).getKey(), ModelConstants.CROSSREF_ID);
assertEquals(d2.getCollectedfrom().size(), 1);
assertFalse(cfId(d2.getCollectedfrom()).contains(ModelConstants.CROSSREF_ID));
assertTrue(
OafMapperUtils
.mergeResults(p1, d2)
.getResulttype()
.getClassid()
.equals(ModelConstants.PUBLICATION_RESULTTYPE_CLASSID));
assertEquals(p2.getCollectedfrom().size(), 1);
assertFalse(cfId(p2.getCollectedfrom()).contains(ModelConstants.CROSSREF_ID));
assertEquals(d1.getCollectedfrom().size(), 1);
assertTrue(cfId(d1.getCollectedfrom()).contains(ModelConstants.CROSSREF_ID));
assertTrue(
OafMapperUtils
.mergeResults(p2, d1)
.getResulttype()
.getClassid()
.equals(ModelConstants.DATASET_RESULTTYPE_CLASSID));
}
@NotNull
protected HashSet<String> cfId(List<KeyValue> collectedfrom) {
return collectedfrom.stream().map(c -> c.getKey()).collect(Collectors.toCollection(HashSet::new));
}
protected <T extends Result> T read(String filename, Class<T> clazz) throws IOException {
final String json = IOUtils.toString(getClass().getResourceAsStream(filename));
return OBJECT_MAPPER.readValue(json, clazz);
}
}

View File

@ -1,21 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import java.util.Set;
import org.junit.jupiter.api.Assertions;
import org.junit.jupiter.api.Test;
public class BlackListProviderTest {
@Test
public void blackListTest() {
Assertions.assertNotNull(PidBlacklistProvider.getBlacklist());
Assertions.assertNotNull(PidBlacklistProvider.getBlacklist().get("doi"));
Assertions.assertTrue(PidBlacklistProvider.getBlacklist().get("doi").size() > 0);
final Set<String> xxx = PidBlacklistProvider.getBlacklist("xxx");
Assertions.assertNotNull(xxx);
Assertions.assertEquals(0, xxx.size());
}
}

View File

@ -1,75 +0,0 @@
package eu.dnetlib.dhp.schema.oaf.utils;
import static org.junit.jupiter.api.Assertions.*;
import java.io.IOException;
import org.apache.commons.io.IOUtils;
import org.junit.jupiter.api.Test;
import com.fasterxml.jackson.databind.DeserializationFeature;
import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.utils.DHPUtils;
public class IdentifierFactoryTest {
private static ObjectMapper OBJECT_MAPPER = new ObjectMapper()
.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);
@Test
public void testCreateIdentifierForPublication() throws IOException {
verifyIdentifier(
"publication_doi1.json", "50|doi_________::79dbc7a2a56dc1532659f9038843256e", true);
verifyIdentifier(
"publication_doi2.json", "50|doi_________::79dbc7a2a56dc1532659f9038843256e", true);
verifyIdentifier(
"publication_doi3.json", "50|pmc_________::94e4cb08c93f8733b48e2445d04002ac", true);
verifyIdentifier(
"publication_doi4.json", "50|od______2852::38861c44e6052a8d49f59a4c39ba5e66", true);
verifyIdentifier(
"publication_pmc1.json", "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f", true);
verifyIdentifier(
"publication_pmc2.json", "50|pmc_________::94e4cb08c93f8733b48e2445d04002ac", true);
final String defaultID = "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f";
verifyIdentifier("publication_3.json", defaultID, true);
verifyIdentifier("publication_4.json", defaultID, true);
verifyIdentifier("publication_5.json", defaultID, true);
}
@Test
public void testCreateIdentifierForPublicationNoHash() throws IOException {
verifyIdentifier("publication_doi1.json", "50|doi_________::10.1016/j.cmet.2010.03.013", false);
verifyIdentifier("publication_doi2.json", "50|doi_________::10.1016/j.cmet.2010.03.013", false);
verifyIdentifier("publication_pmc1.json", "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f", false);
verifyIdentifier(
"publication_urn1.json", "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f", false);
final String defaultID = "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f";
verifyIdentifier("publication_3.json", defaultID, false);
verifyIdentifier("publication_4.json", defaultID, false);
verifyIdentifier("publication_5.json", defaultID, false);
}
protected void verifyIdentifier(String filename, String expectedID, boolean md5) throws IOException {
final String json = IOUtils.toString(getClass().getResourceAsStream(filename));
final Publication pub = OBJECT_MAPPER.readValue(json, Publication.class);
String id = IdentifierFactory.createIdentifier(pub, md5);
assertNotNull(id);
assertEquals(expectedID, id);
}
}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1g", "resuttype" : { "classid" : "dataset" }, "pid":[{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2011.03.013"},{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}], "collectedfrom" : [ { "key" : "10|openaire____::081b82f96300b6a6e3d282bad31cb6e2", "value" : "Crossref"} ]}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1g", "resuttype" : { "classid" : "dataset" }, "pid":[{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2011.03.013"},{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}], "collectedfrom" : [ { "key" : "10|openaire____::081b82f96300b6a6e3d282bad31cb6e3", "value" : "Repository B"} ]}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f", "resuttype" : { "classid" : "publication" }, "pid":[{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2011.03.013"},{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}], "collectedfrom" : [ { "key" : "10|openaire____::081b82f96300b6a6e3d282bad31cb6e2", "value" : "Crossref"} ]}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f", "resuttype" : { "classid" : "publication" }, "pid":[{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2011.03.013"},{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}], "collectedfrom" : [ { "key" : "10|openaire____::081b82f96300b6a6e3d282bad31cb6e3", "value" : "Repository A"} ]}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f","pid":[{"qualifier":{"classid":"scp-number"},"value":"79953761260"}]}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f","pid":[]}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f"}

View File

@ -1,33 +0,0 @@
{
"id": "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f",
"instance": [
{
"collectedfrom": {
"key": "10|openaire____::081b82f96300b6a6e3d282bad31cb6e2",
"value": "Crossref"
},
"pid": [
{
"qualifier": {"classid": "doi"},
"value": "10.1016/j.cmet.2010.03.013"
}
]
},
{
"pid": [
{
"qualifier": {"classid": "urn"},
"value": "urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"
},
{
"qualifier": {"classid": "scp-number"},
"value": "79953761260"
},
{
"qualifier": {"classid": "pmc"},
"value": "21459329"
}
]
}
]
}

View File

@ -1,37 +0,0 @@
{
"id": "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f",
"instance": [
{
"collectedfrom": {
"key": "10|openaire____::081b82f96300b6a6e3d282bad31cb6e2",
"value": "Crossref"
},
"pid": [
{
"qualifier": {"classid": "doi"},
"value": "10.1016/j.cmet.2010.03.013"
}
]
},
{
"collectedfrom": {
"key": "10|opendoar____::8b6dd7db9af49e67306feb59a8bdc52c",
"value": "Europe PubMed Central"
},
"pid": [
{
"qualifier": {"classid": "urn"},
"value": "urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"
},
{
"qualifier": {"classid": "scp-number"},
"value": "79953761260"
},
{
"qualifier": {"classid": "pmc"},
"value": "21459329"
}
]
}
]
}

View File

@ -1,37 +0,0 @@
{
"id": "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f",
"instance": [
{
"collectedfrom": {
"key": "10|openaire____::1234",
"value": "Zenodo"
},
"pid": [
{
"qualifier": {"classid": "doi"},
"value": "10.1016/j.cmet.2010.03.013"
}
]
},
{
"collectedfrom": {
"key": "10|opendoar____::8b6dd7db9af49e67306feb59a8bdc52c",
"value": "Europe PubMed Central"
},
"pid": [
{
"qualifier": {"classid": "urn"},
"value": "urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"
},
{
"qualifier": {"classid": "scp-number"},
"value": "79953761260"
},
{
"qualifier": {"classid": "pmc"},
"value": "21459329"
}
]
}
]
}

View File

@ -1,37 +0,0 @@
{
"id": "50|od______2852::38861c44e6052a8d49f59a4c39ba5e66",
"instance": [
{
"collectedfrom": {
"key": "10|openaire____::1234",
"value": "Zenodo"
},
"pid": [
{
"qualifier": {"classid": "doi"},
"value": "10.1016/j.cmet.2010.03.013"
},
{
"qualifier": {"classid": "handle"},
"value": "11012/83840"
}
]
},
{
"collectedfrom": {
"key": "10|opendoar____::2852",
"value": "Digital library of Brno University of Technology"
},
"pid": [
{
"qualifier": {"classid": "pmc"},
"value": "21459329"
},
{
"qualifier": {"classid": "handle"},
"value": "11012/83840"
}
]
}
]
}

View File

@ -1 +0,0 @@
{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f","pid":[{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}]}

View File

@ -1,21 +0,0 @@
{
"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f",
"instance": [
{
"collectedfrom": {
"key": "10|opendoar____::8b6dd7db9af49e67306feb59a8bdc52c",
"value": "Europe PubMed Central"
},
"pid": [
{
"qualifier": {"classid": "doi"},
"value": "10.1016/j.cmet.2010.03.013"
},
{
"qualifier":{"classid":"pmc"},
"value":"21459329"
}
]
}
]
}

View File

@ -1,23 +0,0 @@
{
"id": "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f",
"pid": [
{
"qualifier": {
"classid": "urn"
},
"value": "urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"
},
{
"qualifier": {
"classid": "scp-number"
},
"value": "79953761260"
},
{
"qualifier": {
"classid": "pmcid"
},
"value": "21459329"
}
]
}

View File

@ -4,10 +4,10 @@ import com.fasterxml.jackson.databind.ObjectMapper
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup
import eu.dnetlib.dhp.schema.action.AtomicAction
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf.{Author, DataInfo, Instance, KeyValue, Oaf, OafMapperUtils, OtherResearchProduct, Publication, Qualifier, Relation, Result, Software, StructuredProperty, Dataset => OafDataset}
import eu.dnetlib.dhp.schema.oaf.{Author, DataInfo, Instance, KeyValue, Oaf, OtherResearchProduct, Publication, Qualifier, Relation, Result, Software, StructuredProperty, Dataset => OafDataset}
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory
import eu.dnetlib.dhp.schema.oaf.{AccessRight, Author, DataInfo, Instance, KeyValue, Oaf, OafMapperUtils, OtherResearchProduct, Publication, Qualifier, Relation, Result, Software, StructuredProperty, Dataset => OafDataset}
import eu.dnetlib.dhp.schema.oaf.utils.{IdentifierFactory, OafMapperUtils}
import eu.dnetlib.dhp.schema.oaf.{AccessRight, Author, DataInfo, Instance, KeyValue, Oaf, OtherResearchProduct, Publication, Qualifier, Relation, Result, Software, StructuredProperty, Dataset => OafDataset}
import eu.dnetlib.dhp.utils.DHPUtils
import org.apache.commons.lang3.StringUtils
import org.json4s.DefaultFormats

View File

@ -1,26 +1,6 @@
package eu.dnetlib.dhp.collection.plugin.rest;
import eu.dnetlib.dhp.collection.CollectorException;
import eu.dnetlib.dhp.collection.HttpClientParams;
import eu.dnetlib.dhp.collection.JsonUtils;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.http.HttpHeaders;
import org.apache.http.entity.ContentType;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.w3c.dom.Node;
import org.w3c.dom.NodeList;
import org.xml.sax.InputSource;
import javax.xml.transform.OutputKeys;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerConfigurationException;
import javax.xml.transform.TransformerFactory;
import javax.xml.transform.dom.DOMSource;
import javax.xml.transform.stream.StreamResult;
import javax.xml.xpath.*;
import java.io.InputStream;
import java.io.StringWriter;
import java.io.UnsupportedEncodingException;
@ -32,6 +12,28 @@ import java.util.Iterator;
import java.util.Queue;
import java.util.concurrent.PriorityBlockingQueue;
import javax.xml.transform.OutputKeys;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerConfigurationException;
import javax.xml.transform.TransformerFactory;
import javax.xml.transform.dom.DOMSource;
import javax.xml.transform.stream.StreamResult;
import javax.xml.xpath.*;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.http.HttpHeaders;
import org.apache.http.entity.ContentType;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.w3c.dom.Node;
import org.w3c.dom.NodeList;
import org.xml.sax.InputSource;
import eu.dnetlib.dhp.collection.CollectorException;
import eu.dnetlib.dhp.collection.HttpClientParams;
import eu.dnetlib.dhp.collection.JsonUtils;
/**
* log.info(...) equal to log.trace(...) in the application-logs
* <p>

View File

@ -32,6 +32,7 @@ import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.common.HdfsSupport;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import scala.Tuple2;
/**

View File

@ -2,15 +2,11 @@
package eu.dnetlib.dhp.oa.dedup;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.Optional;
import org.apache.commons.io.IOUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Encoders;
import org.apache.spark.sql.SaveMode;
import org.apache.spark.sql.SparkSession;
@ -18,20 +14,12 @@ import org.dom4j.DocumentException;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.sun.media.sound.ModelChannelMixer;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.KeyValue;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.utils.DHPUtils;
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
import eu.dnetlib.pace.config.DedupConfig;
import net.sf.saxon.ma.trie.Tuple2;
public class SparkCopyOpenorgsMergeRels extends AbstractSparkAction {
private static final Logger log = LoggerFactory.getLogger(SparkCopyOpenorgsMergeRels.class);

View File

@ -15,6 +15,7 @@ import eu.dnetlib.dhp.schema.common.EntityType;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.dhp.schema.oaf.utils.PidComparator;
import eu.dnetlib.dhp.schema.oaf.utils.PidType;

View File

@ -23,6 +23,7 @@ import com.google.common.collect.Lists;
import eu.dnetlib.dhp.oa.dedup.model.Identifier;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.pace.util.MapDocumentUtil;
import scala.Tuple2;

View File

@ -3,7 +3,6 @@ package eu.dnetlib.dhp.oa.dedup;
import static java.nio.file.Files.createTempDirectory;
import static org.apache.spark.sql.functions.count;
import static org.junit.jupiter.api.Assertions.*;
import static org.mockito.Mockito.lenient;
@ -23,21 +22,11 @@ import java.util.Properties;
import org.apache.commons.io.FileUtils;
import org.apache.commons.io.IOUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaPairRDD;
import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.JavaSparkContext;
import org.apache.spark.api.java.function.FilterFunction;
import org.apache.spark.api.java.function.ForeachFunction;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.api.java.function.PairFunction;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Encoders;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.SparkSession;
import org.apache.spark.util.CollectionsUtils;
import org.junit.jupiter.api.*;
import org.junit.jupiter.api.extension.ExtendWith;
import org.junit.platform.commons.util.StringUtils;
import org.mockito.Mock;
import org.mockito.Mockito;
import org.mockito.junit.jupiter.MockitoExtension;
@ -46,14 +35,10 @@ import com.fasterxml.jackson.databind.DeserializationFeature;
import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.schema.oaf.OafMapperUtils;
import eu.dnetlib.dhp.schema.oaf.Organization;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.utils.DHPUtils;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
import eu.dnetlib.pace.util.MapDocumentUtil;
import scala.Tuple2;
@ExtendWith(MockitoExtension.class)
@TestMethodOrder(MethodOrderer.OrderAnnotation.class)

View File

@ -18,6 +18,7 @@ import com.google.gson.*;
import eu.dnetlib.dhp.common.PacePerson;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.dhp.utils.DHPUtils;
import eu.dnetlib.doiboost.orcidnodoi.util.DumpToActionsUtility;
import eu.dnetlib.doiboost.orcidnodoi.util.Pair;

View File

@ -2,12 +2,11 @@
package eu.dnetlib.dhp.oa.graph.clean;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
import static eu.dnetlib.dhp.schema.oaf.CleaningFunctions.*;
import static eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions.*;
import java.util.Optional;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.function.FilterFunction;
import org.apache.spark.api.java.function.MapFunction;

View File

@ -2,7 +2,7 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.*;
import static eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils.*;
import java.util.*;
import java.util.stream.Collectors;
@ -19,6 +19,7 @@ import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
public abstract class AbstractMdRecordToOafMapper {

View File

@ -2,15 +2,7 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.asString;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.createOpenaireId;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.dataInfo;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.field;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.journal;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.listFields;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.listKeyValues;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.qualifier;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.structuredProperty;
import static eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils.*;
import java.io.Closeable;
import java.io.IOException;

View File

@ -2,7 +2,7 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.*;
import static eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils.*;
import java.util.ArrayList;
import java.util.HashSet;
@ -20,8 +20,9 @@ import com.google.common.collect.Lists;
import eu.dnetlib.dhp.common.PacePerson;
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.CleaningFunctions;
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
import eu.dnetlib.dhp.schema.oaf.utils.ModelHardLimits;
public class OafToOafMapper extends AbstractMdRecordToOafMapper {

View File

@ -2,7 +2,7 @@
package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import static eu.dnetlib.dhp.schema.oaf.OafMapperUtils.*;
import static eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils.*;
import java.util.*;
import java.util.stream.Collectors;
@ -13,8 +13,8 @@ import org.dom4j.Node;
import eu.dnetlib.dhp.common.PacePerson;
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
public class OdfToOafMapper extends AbstractMdRecordToOafMapper {

View File

@ -21,6 +21,7 @@ import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;

View File

@ -19,6 +19,7 @@ import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.oa.graph.clean.CleaningFunctionTest;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;

View File

@ -30,6 +30,7 @@ import com.google.common.collect.Lists;
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
@ExtendWith(MockitoExtension.class)
public class MigrateDbEntitiesApplicationTest {

View File

@ -28,6 +28,7 @@ import eu.dnetlib.dhp.oa.provision.model.RelatedEntity;
import eu.dnetlib.dhp.oa.provision.model.RelatedEntityWrapper;
import eu.dnetlib.dhp.schema.common.EntityType;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.ModelHardLimits;
import scala.Tuple2;
/**

View File

@ -30,6 +30,7 @@ import eu.dnetlib.dhp.oa.provision.model.RelatedEntityWrapper;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.ModelHardLimits;
import scala.Tuple2;
import scala.collection.JavaConverters;
import scala.collection.Seq;