From 513407a1d15e7ee27fa854d57e892c08e6bdaa13 Mon Sep 17 00:00:00 2001
From: "Stefan E. Funk" <funk@sub.uni-goettingen.de>
Date: Mon, 19 Jul 2021 15:41:01 +0200
Subject: [PATCH] Add "%2F" instead of "_" to JSON file names
---
.../textgrid/middleware/OAIPMHUtilities.java | 3 +
.../middleware/RecordDelivererDATACITE.java | 223 ++++++++++++++----
.../webapp/WEB-INF/oaipmh.dariah.properties | 4 +-
pom.xml | 2 +-
4 files changed, 182 insertions(+), 50 deletions(-)
diff --git a/oaipmh-core/src/main/java/info/textgrid/middleware/OAIPMHUtilities.java b/oaipmh-core/src/main/java/info/textgrid/middleware/OAIPMHUtilities.java
index 952dab03..e3f822b0 100644
--- a/oaipmh-core/src/main/java/info/textgrid/middleware/OAIPMHUtilities.java
+++ b/oaipmh-core/src/main/java/info/textgrid/middleware/OAIPMHUtilities.java
@@ -236,6 +236,9 @@ public class OAIPMHUtilities {
public static String datestampAsString(String originalDateTimeString) throws ParseException {
SimpleDateFormat tgItemTime = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss.S");
+
+ System.out.println("DATE? " + originalDateTimeString);
+
Date date = tgItemTime.parse(originalDateTimeString);
SimpleDateFormat outFormatter = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss'Z'");
outFormatter.setTimeZone(TimeZone.getTimeZone("UTC"));
diff --git a/oaipmh-core/src/main/java/info/textgrid/middleware/RecordDelivererDATACITE.java b/oaipmh-core/src/main/java/info/textgrid/middleware/RecordDelivererDATACITE.java
index b1b100e0..d8fc2afe 100644
--- a/oaipmh-core/src/main/java/info/textgrid/middleware/RecordDelivererDATACITE.java
+++ b/oaipmh-core/src/main/java/info/textgrid/middleware/RecordDelivererDATACITE.java
@@ -11,6 +11,7 @@ import org.apache.commons.logging.LogFactory;
import org.elasticsearch.common.Strings;
import org.json.JSONObject;
import org.springframework.stereotype.Component;
+import info.textgrid.middleware.common.TextGridMimetypes;
import info.textgrid.middleware.oaipmh.ContributorType;
import info.textgrid.middleware.oaipmh.DateType;
import info.textgrid.middleware.oaipmh.GetRecordType;
@@ -46,7 +47,6 @@ import info.textgrid.middleware.oaipmh.Resource.Subjects;
import info.textgrid.middleware.oaipmh.Resource.Subjects.Subject;
import info.textgrid.middleware.oaipmh.Resource.Titles;
import info.textgrid.middleware.oaipmh.Resource.Titles.Title;
-import info.textgrid.middleware.oaipmh.TitleType;
/**
* @author Maximilian Brodhun, SUB Göttingen
@@ -118,6 +118,11 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
Strings.EMPTY_ARRAY)
.getSource());
+ if (this.jsonObj == null || this.jsonObj.isEmpty()) {
+ throw new IOException(
+ "No data could be retrieved from ElasticSearch for ID: " + idInElasticSearchIndex);
+ }
+
System.out.println("jsonObject: " + this.jsonObj);
// Set response header.
@@ -204,14 +209,14 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
this.resource.setRelatedIdentifiers(this.addRelatedIdentifiers());
this.resource.setIdentifier(this.addIdentifier());
this.resource.setCreators(this.addCreators());
- this.addGeoLocation();
+ // FIXME Check if we can have empty elements here! Was: Check for null in this.addGeoLocation()
+ // method!)
+ this.resource.setGeoLocations(this.addGeoLocation());
this.resource.setResourceType(addResourceType());
this.resource.setVersion(this.addVersion());
-
- if (addSubjects().getSubject().get(0).getValue() != null) {
- this.resource.setSubjects(this.addSubjects()); // TODO what if subjects == null?
- }
-
+ // FIXME Check if we can have empty elements here! Was: Check for null in this.addSubjects()
+ // method!)
+ this.resource.setSubjects(this.addSubjects());
this.resource.setSizes(this.addSize());
return this.resource;
@@ -241,19 +246,43 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
Subjects subjects = new Subjects();
- for (String subjectField : this.oarSubjectFields) {
- Subject subject = new Subject();
- subject.setValue(OAIPMHUtilities.fieldLoader(this.jsonObj, subjectField + ".value"));
- subject.setSchemeURI(OAIPMHUtilities.fieldLoader(this.jsonObj, subjectField + ".id.type"));
- subject
- .setSubjectScheme(OAIPMHUtilities.fieldLoader(this.jsonObj, subjectField + ".id.value"));
- subjects.getSubject().add(subject);
+ // **
+ // TextGrid
+ // **
+
+ if (this.textgrid) {
+ for (String subjectField : this.oarSubjectFields) {
+ Subject subject = new Subject();
+ subject.setValue(OAIPMHUtilities.fieldLoader(this.jsonObj, subjectField + ".value"));
+ subject.setSchemeURI(OAIPMHUtilities.fieldLoader(this.jsonObj, subjectField + ".id.type"));
+ subject.setSubjectScheme(
+ OAIPMHUtilities.fieldLoader(this.jsonObj, subjectField + ".id.value"));
+ subjects.getSubject().add(subject);
+ }
}
- if (subjects.getSubject().get(0) != null) {
- return subjects;
- } else {
- return null;
+
+ // **
+ // DARIAH
+ // **
+
+ else if (this.dariah) {
+ for (String subjectField : this.oarSubjectFields) {
+ List<String> sList = OAIPMHUtilities.fieldLoader(this.jsonObj, new String[] {subjectField});
+ for (String s : sList) {
+ Subject subject = new Subject();
+
+ log.debug("subject: " + s);
+
+ subject.setValue(s);
+ // TODO Extract subject schema and schema URI somehow?
+ // subject.setSchemeURI();
+ // subject.setSubjectScheme();
+ subjects.getSubject().add(subject);
+ }
+ }
}
+
+ return subjects;
}
/**
@@ -273,12 +302,73 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
/**
* @return
*/
- private static ResourceType addResourceType() {
+ private ResourceType addResourceType() {
ResourceType resourceType = new ResourceType();
- resourceType.setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.DATASET);
- // FIXME What about collection? Could editions be seen as collection datasets?
+ // **
+ // TextGrid
+ // **
+
+ if (this.textgrid) {
+ for (String format : this.oarFormatFields) {
+ String resourceValue = OAIPMHUtilities.fieldLoader(this.jsonObj, format);
+ // TODO Set resource value here?
+ resourceType.setValue(resourceValue);
+ // Set collection for all aggregation types, take also into account: images and text.
+ if (TextGridMimetypes.AGGREGATION_SET.contains(format)) {
+ resourceType
+ .setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.COLLECTION);
+ } else if (TextGridMimetypes.IMAGE_SET.contains(format)) {
+ resourceType
+ .setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.IMAGE);
+ } else if (TextGridMimetypes.ORIGINAL_SET.contains(format)) {
+ resourceType
+ .setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.TEXT);
+ }
+ // Use dataset for everything else.
+ else {
+ resourceType.setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.DATASET);
+ }
+ // We take only the first of all the resource types!
+ if (resourceType.getValue() != null && !resourceType.getValue().isEmpty()) {
+ break;
+ }
+ }
+ }
+
+ // **
+ // DARIAH
+ // **
+
+ else if (this.dariah) {
+
+ System.out.println(" ## oarFormatFields: " + this.oarFormatFields.length);
+
+ for (String format : this.oarFormatFields) {
+ // Only the first value is taken here! We do not need to create a list!
+ // TODO Please see addLanguage(), maybe do it as done there to use the other fielsLoader()
+ // method!
+ String resourceValue = OAIPMHUtilities.fieldLoader(this.jsonObj, format);
+
+ System.out.println(" ## resourceType: " + format + " --> " + resourceValue);
+
+ resourceType.setValue(resourceValue);
+ // Set collection for DARIAH collection type here, data object for everything else.
+ if (TextGridMimetypes.DARIAH_COLLECTION.equals(format)) {
+ resourceType
+ .setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.COLLECTION);
+ }
+ // Use dataset for everything else.
+ else {
+ resourceType.setResourceTypeGeneral(info.textgrid.middleware.oaipmh.ResourceType.DATASET);
+ }
+ // We take only the first of all the resource types!
+ if (resourceType.getValue() != null && !resourceType.getValue().isEmpty()) {
+ break;
+ }
+ }
+ }
return resourceType;
}
@@ -286,7 +376,7 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
/**
*
*/
- private void addGeoLocation() {
+ private GeoLocations addGeoLocation() {
GeoLocations geoLocations = new GeoLocations();
@@ -296,11 +386,12 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
OAIPMHUtilities.fieldLoader(this.jsonObj, geoLocationField + ".value"));
geoLocations.getGeoLocation().add(geoLocation);
}
- if (geoLocations.getGeoLocation().get(0).getGeoLocationPlace() != null) {
- this.resource.setGeoLocations(geoLocations);
- }
- // return geoLocations;
+ // if (geoLocations.getGeoLocation().get(0).getGeoLocationPlace() != null) {
+ // this.resource.setGeoLocations(geoLocations);
+ // }
+
+ return geoLocations;
}
/**
@@ -313,7 +404,7 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
for (String relatedIdentifierField : this.oarRelatedIdentifierFields) {
- System.out.println("relatedIdentifierField: " + relatedIdentifierField);
+ log.debug("relatedIdentifierField: " + relatedIdentifierField);
// **
// TextGrid
@@ -350,7 +441,7 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
OAIPMHUtilities.fieldLoader(this.jsonObj, new String[] {relatedIdentifierField});
for (String i : relatedIdentiferList) {
- System.out.println("i: " + i);
+ log.debug("i: " + i);
RelatedIdentifier relatedID = new RelatedIdentifier();
relatedID.setValue(i);
@@ -360,13 +451,15 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
relatedID.setRelatedIdentifierType(RelatedIdentifierType.DOI);
} else if (i.startsWith("http")) {
relatedID.setRelatedIdentifierType(RelatedIdentifierType.URL);
- } else {
- // TODO Check other prefixes and add other values, too?
- // TODO No OTHER value existing! What do we do here, if none of the above apply?
}
- // TODO We have to check relation type, too? Firstly we take part of a collection
- // here(that's what we put in while publishing).
- relatedID.setRelationType(RelationType.IS_PART_OF);
+ // TODO Check other prefixes and add other values, too?
+ else {
+ // TODO No type OTHER existing! Value is mandatory, so we chose URL, what else can we
+ // do?
+ relatedID.setRelatedIdentifierType(RelatedIdentifierType.URL);
+ }
+ // Relation type is REFERENCES for the time being (coming from dc:relation at the moment).
+ relatedID.setRelationType(RelationType.REFERENCES);
relatedIdentifiers.getRelatedIdentifier().add(relatedID);
}
}
@@ -446,16 +539,48 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
RightsList rightsList = new RightsList();
- for (String rightsField : this.oarRightsFields) {
- Rights rights = new Rights();
- rights.setRightsURI(OAIPMHUtilities.fieldLoader(this.jsonObj, rightsField + ".licenseUri"));
- rights.setValue(OAIPMHUtilities.fieldLoader(this.jsonObj, rightsField + ".value"));
- rightsList.getRights().add(rights);
+ // **
+ // Textgrid
+ // **
+
+ if (this.textgrid) {
+ for (String rightsField : this.oarRightsFields) {
+ Rights rights = new Rights();
+ rights.setRightsURI(OAIPMHUtilities.fieldLoader(this.jsonObj, rightsField + ".licenseUri"));
+ rights.setValue(OAIPMHUtilities.fieldLoader(this.jsonObj, rightsField + ".value"));
+ rightsList.getRights().add(rights);
+ }
+ if (rightsList.getRights().get(0).getValue() != null) {
+ Rights openAccesRight = new Rights();
+ openAccesRight.setRightsURI("info:eu-repo/semantics/openAccess");
+ rightsList.getRights().add(openAccesRight);
+ }
}
- if (rightsList.getRights().get(0).getValue() != null) {
- Rights openAccesRight = new Rights();
- openAccesRight.setRightsURI("info:eu-repo/semantics/openAccess");
- rightsList.getRights().add(openAccesRight);
+
+ // **
+ // DARIAH
+ // **
+
+ else if (this.dariah) {
+ for (String rightsField : this.oarRightsFields) {
+ Rights rights = new Rights();
+
+ List<String> rList = OAIPMHUtilities.fieldLoader(this.jsonObj, new String[] {rightsField});
+
+ for (String r : rList) {
+
+ log.debug("right: " + r);
+
+ // Set rights URI assuming every "http://" or "https://" really IS an URI.
+ if (r.startsWith("http")) {
+ rights.setRightsURI(r);
+ }
+ // Set value in every case, even if doubled with URI.
+ rights.setValue(r);
+ }
+
+ rightsList.getRights().add(rights);
+ }
}
return rightsList;
@@ -528,10 +653,13 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
String language = null;
- // TODO Use only ONE language here? What if we have more then one in DARIAH DC data?
-
+ // TODO Only ONE language is permitted in Datacite schema? What if we have more then one in
+ // DARIAH DC data? Take only first one for the moment!
for (String languageField : this.oarLanguageFields) {
- language = OAIPMHUtilities.fieldLoader(this.jsonObj, languageField);
+ List<String> langs = OAIPMHUtilities.fieldLoader(this.jsonObj, new String[] {languageField});
+ if (!langs.isEmpty()) {
+ language = langs.get(0);
+ }
}
return language;
@@ -735,7 +863,8 @@ public class RecordDelivererDATACITE extends RecordDelivererAbstract {
for (String title : titleList) {
Title tileInOpenAireRecord = new Title();
tileInOpenAireRecord.setValue(title);
- tileInOpenAireRecord.setTitleType(TitleType.OTHER);
+ // Title type OTHER is not permitted, leave type out for the moment!
+ // tileInOpenAireRecord.setTitleType(TitleType.OTHER);
titles.getTitle().add(tileInOpenAireRecord);
}
}
diff --git a/oaipmh-webapp/src/main/webapp/WEB-INF/oaipmh.dariah.properties b/oaipmh-webapp/src/main/webapp/WEB-INF/oaipmh.dariah.properties
index 0240a8d9..f9d99173 100644
--- a/oaipmh-webapp/src/main/webapp/WEB-INF/oaipmh.dariah.properties
+++ b/oaipmh-webapp/src/main/webapp/WEB-INF/oaipmh.dariah.properties
@@ -43,7 +43,7 @@ metadataFormatListInstance = MetadataFormatListDelivererDH
## Field Settings ##
######################
-fields = administrativeMetadata.dcterms:modified, administrativeMetadata.dcterms:created, administrativeMetadata.dcterms:identifier, administrativeMetadata.dcterms:extent, administrativeMetadata.dcterms:relation, descriptiveMetadata.dc:contributor, descriptiveMetadata.dc:coverage, descriptiveMetadata.dc:creator, descriptiveMetadata.dc:date, descriptiveMetadata.dc:description, descriptiveMetadata.dc:format, descriptiveMetadata.dc:identifier, descriptiveMetadata.dc:language, descriptiveMetadata.dc:publisher, descriptiveMetadata.dc:relation, descriptiveMetadata.dc:rights, descriptiveMetadata.dc:source, descriptiveMetadata.dc:subject, descriptiveMetadata.dc:title, descriptiveMetadata.dc:type, descriptiveMetadata.dc:relation
+fields = administrativeMetadata.dcterms:modified, administrativeMetadata.dcterms:created, administrativeMetadata.dcterms:identifier, administrativeMetadata.dcterms:extent, administrativeMetadata.dcterms:relation, administrativeMetadata.dcterms:format, descriptiveMetadata.dc:contributor, descriptiveMetadata.dc:coverage, descriptiveMetadata.dc:creator, descriptiveMetadata.dc:date, descriptiveMetadata.dc:description, descriptiveMetadata.dc:format, descriptiveMetadata.dc:identifier, descriptiveMetadata.dc:language, descriptiveMetadata.dc:publisher, descriptiveMetadata.dc:relation, descriptiveMetadata.dc:rights, descriptiveMetadata.dc:source, descriptiveMetadata.dc:subject, descriptiveMetadata.dc:title, descriptiveMetadata.dc:type, descriptiveMetadata.dc:relation
workFields = descriptiveMetadata.dc:title
######################
@@ -66,7 +66,7 @@ oar.geoLocationFields = descriptiveMetadata.dc:coverage
oar.handle = administrativeMetadata.dcterms:identifier
oar.versionFields =
oar.subjectFields = descriptiveMetadata.dc:subject
-oar.sizeField = administrativeMetadata.dcterms:extent
+oar.sizeField = administrativeMetadata.dcterms:extent
##########################
## Dublin Core Fields ##
diff --git a/pom.xml b/pom.xml
index 27a0f462..49ebb55d 100644
--- a/pom.xml
+++ b/pom.xml
@@ -8,7 +8,7 @@
<packaging>pom</packaging>
<name>DARIAHDE :: OAI-PMH DataProvider</name>
<properties>
- <common.version>3.7.0-ES6-SNAPSHOT</common.version>
+ <common.version>3.9.0-ES6-SNAPSHOT</common.version>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<cxf.version>3.2.8</cxf.version>
<spring.version>4.0.2.RELEASE</spring.version>
--
GitLab