MaDgIK
/
uoa-validator-engine2
13
1
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'master' of code-repo.d4science.org:MaDgIK/validator-engine 

# Conflicts:
#	src/test/java/eu/dnetlib/validator2/engine/Test_FAIR.java
This commit is contained in:
pispis 2023-07-27 14:52:58 +03:00
parent d04630465d a4c7c33907
commit 6e490f9a63
12 changed files with 168 additions and 158 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
src/main/java/eu/dnetlib/validator2/engine/RuleEngine.java
View File

@ -2,18 +2,16 @@ package eu.dnetlib.validator2.engine;
public class RuleEngine {
public static <T, R extends Rule<T>> void applyAndReport(R rule, T t, Reporter<T, R> reporter) {
public static <T, R extends Rule<T>> void applyAndReport(R rule, T t, Reporter<T, R> reporter) throws IllegalArgumentException
{
if (reporter == null) throw new IllegalArgumentException("Reporter cannot be null");
if (rule == null) throw new IllegalArgumentException("Rule cannot be null");
try {
if(rule.test(t)) {
if ( rule.test(t) )
reporter.reportSuccessFor(rule, t);
}
else {
else
reporter.reportFailureFor(rule, t);
}
} catch (Throwable throwable) {
reporter.reportErrorFor(rule, t, throwable);
}

2
src/main/java/eu/dnetlib/validator2/engine/builtins/XPathExpressionHelper.java
View File

@ -42,8 +42,6 @@ public class XPathExpressionHelper {
if ( logger.isTraceEnabled() )
logger.trace("Compiling {}", s);
XPathExpression expr = XPATH.compile(s);
if ( logger.isTraceEnabled() )
logger.trace("Compiled {} = {}", s, expr);
return expr;
} catch (XPathExpressionException e) {
logger.error("Compilation failure", e);

7
src/main/java/eu/dnetlib/validator2/validation/guideline/ElementSpecCompiler.java
View File

@ -69,7 +69,7 @@ class ElementSpecCompiler {
// TODO: Should we escape the arg?
private static String xpathForNodeName(String name) {
return "local-name()='" + name + "'";
return "name()='" + name + "'";
}
private static String xpathForNodeName(String name, String valuePrefix) {
@ -436,12 +436,13 @@ class ElementSpecCompiler {
@Override
public boolean test(Document doc) throws RuleEvaluationException {
GuidelineEvaluation guidelineEvaluation = runtimeInfo.get();
String thisId = getContext().getIdProperty().getValue();
XMLContext context = getContext();
String thisId = context.getIdProperty().getValue();
if (isApplicable(this, guidelineEvaluation).test(doc)) {
try {
logger.debug("Applying {}", thisId);
NodeList nodes = getContext().getXPathExpressionProperty().evaluate(doc);
NodeList nodes = context.getXPathExpressionProperty().evaluate(doc);
boolean result = predicate.test(nodes);
if (result) {
logger.debug("Setting node list of this rule {}", thisId);

22
src/main/java/eu/dnetlib/validator2/validation/guideline/GuidelineEvaluation.java
View File

@ -47,27 +47,27 @@ class GuidelineEvaluation {
logger.debug("Evaluating " + rules);
for (SyntheticRule<Document> rule: rules) {
for ( SyntheticRule<Document> rule: rules ) {
String id = rule.getContext().getIdProperty().getValue();
RuleEngine.applyAndReport(rule, doc, reporter);
Status status = diagnostics.getLastReportedStatus();
if (status == Status.ERROR) {
if ( status == Status.ERROR ) {
// fail fast in case of errors (no reason to proceed, since results may be rubbish)
return StandardResult.forError(diagnostics.getLastReportedError().getMessage());
}
if (status == Status.SUCCESS && getRequirementLevelOf(id) == RequirementLevel.NOT_APPLICABLE) {
if ( status == Status.SUCCESS && getRequirementLevelOf(id) == RequirementLevel.NOT_APPLICABLE ) {
// Report the non-applicability of a rule as a warning
// The check for both status and non-applicable requirement level is redundant
// (non-applicable rules always succeed), yet it is more clear hopefully.
// (non-applicable rules always succeed), yet it is more clear, hopefully.
logger.warn("Non-applicable: " + rule);
warnings.add(synthesizeFailureMessage(rule));
}
else if (status == Status.FAILURE) {
if (getRequirementLevelOf(id) == RequirementLevel.MANDATORY) {
else if ( status == Status.FAILURE ) {
if ( getRequirementLevelOf(id) == RequirementLevel.MANDATORY ) {
// A mandatory rule has failed, yet we don't know whether we should report is as such.
// Let's check the parent of the rule
@ -78,8 +78,7 @@ class GuidelineEvaluation {
logger.error("Fail fast for root failure: " + rule);
errors.add(synthesizeFailureMessage(rule));
return StandardResult.forFailure(warnings, errors);
}
else {
} else {
// The current rule has a parent/ancestor which:
// (a) is non-mandatory or
// (b) it was successful.
@ -87,8 +86,7 @@ class GuidelineEvaluation {
logger.warn("Mandatory failure: " + rule);
warnings.add(synthesizeFailureMessage(rule));
}
}
else {
} else {
// This is a warning: a non-mandatory rule has failed.
// Note that MA rules are treated as non-mandatory.
// We let the evaluation loop proceed.
@ -99,7 +97,7 @@ class GuidelineEvaluation {
}
int returnedWeight = weight;
if (rules.size() == warnings.size()) {
if ( rules.size() == warnings.size() ) {
// TODO: I think this is the desired behavior, but we need to get confirmation for this.
// All rules have failed, but with warnings (thus being either optional or recommended).
// This indicates a success with 0 weight.
@ -109,12 +107,10 @@ class GuidelineEvaluation {
}
private String synthesizeFailureMessage(Rule<Document> rule) {
return subjectId + ": rule " + Helper.stringify(rule) + " has failed";
}
private String synthesizeNotApplicableMessage(Rule<Document> rule) {
return subjectId + ": rule " + Helper.stringify(rule) + " is not applicable";
}

2
src/main/java/eu/dnetlib/validator2/validation/guideline/openaire/AbstractOpenAireProfile.java
View File

@ -63,7 +63,7 @@ public abstract class AbstractOpenAireProfile implements XMLApplicationProfile {
return XMLCardinalityRule
.builder()
.setId(ElementSpec.APPLICABILITY_RULE_ID)
.setXPathExpression("//*[local-name()='" + elementName + "']")
.setXPathExpression("//*[name()='" + elementName + "']")
.setIsInclusive(true).setRange(1, Long.MAX_VALUE)
.build();
}

84
src/main/java/eu/dnetlib/validator2/validation/guideline/openaire/FAIR_Data_GuidelinesProfile.java
View File

@ -386,6 +386,7 @@ class F2_01M_SPEC extends AbstractGuideline<Document> {
// TODO: iterate over results and build one Guideline.Result
try {
XMLApplicationProfile.ValidationResult res_F = profile.validate(id, t);
Map<String, Guideline.Result> results = res_F.results();
// Get actual score and not (%) to incorporate to FAIR score
final int MaxScoreF2_01M_SPEC = (int) ((res_F.score()*getWeight())/100);
@ -397,19 +398,23 @@ class F2_01M_SPEC extends AbstractGuideline<Document> {
List<String> warnings2 = new ArrayList<>();
List<String> errors2 = new ArrayList<>();
int score = 0;
for (Map.Entry entry : res_F.results().entrySet()) {
if (res_F.results().get(entry.getKey()).warnings().toString().length() > 2) {
warnings2.add(res_F.results().get(entry.getKey()).warnings().toString());
for (Map.Entry<String, Guideline.Result> entry : results.entrySet()) {
String key = entry.getKey();
String warningsStr = results.get(key).warnings().toString();
String errorsStr = results.get(key).errors().toString();
if ( warningsStr.length() > 2 ) {
warnings2.add(warningsStr);
}
if (res_F.results().get(entry.getKey()).errors().toString().length() > 2) {
errors2.add(res_F.results().get(entry.getKey()).errors().toString());
if ( errorsStr.length() > 2 ) {
errors2.add(errorsStr);
}
if (entry.getValue().toString().contains("SUCCESS")) {
score += 2;
}
logger.debug(String.valueOf(res_F.results().get(entry.getKey()).warnings().getClass()));
logger.debug(String.valueOf(warnings2.getClass()));
if ( logger.isTraceEnabled() ) {
logger.trace(String.valueOf(results.get(key).warnings().getClass()));
logger.trace(String.valueOf(warnings2.getClass()));
}
}
@ -470,7 +475,7 @@ class F3_01M_SPEC extends AbstractGuideline<Document> {
try {
// System.out.println("\nMetadata includes the identifier for the data");
XMLApplicationProfile.ValidationResult res_F = profile.validate(id, t);
Map<String, Result> results = res_F.results();
Map<String, Guideline.Result> results = res_F.results();
// int MaxScoreF3_01M_SPEC = (int) ((res_F.score()*getWeight())/100);
int MaxScoreF3_01M_SPEC;
@ -483,12 +488,15 @@ class F3_01M_SPEC extends AbstractGuideline<Document> {
ArrayList<String> warnings2 = new ArrayList<>();
ArrayList<String> errors2 = new ArrayList<>();
for (Map.Entry entry : res_F.results().entrySet()) {
if (res_F.results().get(entry.getKey()).warnings().toString().length() > 2) {
warnings2.add(res_F.results().get(entry.getKey()).warnings().toString());
for (Map.Entry<String, Guideline.Result> entry : results.entrySet()) {
String key = entry.getKey();
String warningsStr = results.get(key).warnings().toString();
String errorsStr = results.get(key).errors().toString();
if ( warningsStr.length() > 2 ) {
warnings2.add(warningsStr);
}
if (res_F.results().get(entry.getKey()).errors().toString().length() > 2) {
errors2.add(res_F.results().get(entry.getKey()).errors().toString());
if ( errorsStr.length() > 2 ) {
errors2.add(errorsStr);
}
}
@ -544,9 +552,9 @@ class I2_01M_SPEC extends AbstractGuideline<Document> {
// TODO: iterate over results and build one Guideline.Result
try {
logger.debug("\nMetadata uses FAIR-compliant vocabularies");
logger.debug("Metadata uses FAIR-compliant vocabularies");
XMLApplicationProfile.ValidationResult res_F = profile.validate(id, t);
Map<String, Result> results = res_F.results();
Map<String, Guideline.Result> results = res_F.results();
// int MaxScoreI2_01M_SPEC = (int) ((res_F.score()*getWeight())/100);
int MaxScoreI2_01M_SPEC;
@ -559,12 +567,15 @@ class I2_01M_SPEC extends AbstractGuideline<Document> {
ArrayList<String> warnings2 = new ArrayList<>();
ArrayList<String> errors2 = new ArrayList<>();
for (Map.Entry entry : res_F.results().entrySet()) {
if (res_F.results().get(entry.getKey()).warnings().toString().length() > 2) {
warnings2.add(res_F.results().get(entry.getKey()).warnings().toString());
for (Map.Entry<String, Guideline.Result> entry : results.entrySet()) {
String key = entry.getKey();
String warningsStr = results.get(key).warnings().toString();
String errorsStr = results.get(key).errors().toString();
if ( warningsStr.length() > 2 ) {
warnings2.add(warningsStr);
}
if (res_F.results().get(entry.getKey()).errors().toString().length() > 2) {
errors2.add(res_F.results().get(entry.getKey()).errors().toString());
if ( errorsStr.length() > 2 ) {
errors2.add(errorsStr);
}
}
@ -624,18 +635,21 @@ class R1_01M_SPEC extends AbstractGuideline<Document> {
try {
// System.out.println("\nPlurality of accurate and relevant attributes are provided to allow reuse");
XMLApplicationProfile.ValidationResult res_F = profile.validate(id, t);
Map<String, Result> results = res_F.results();
Map<String, Guideline.Result> results = res_F.results();
int MaxScoreR1_01M_SPEC = (int) ((res_F.score()*getWeight())/100);
ArrayList<String> warnings2 = new ArrayList<>();
ArrayList<String> errors2 = new ArrayList<>();
int score = 0;
for (Map.Entry entry : res_F.results().entrySet()) {
if (res_F.results().get(entry.getKey()).warnings().toString().length() > 2) {
warnings2.add(res_F.results().get(entry.getKey()).warnings().toString());
for (Map.Entry<String, Guideline.Result> entry : results.entrySet()) {
String key = entry.getKey();
String warningsStr = results.get(key).warnings().toString();
String errorsStr = results.get(key).errors().toString();
if ( warningsStr.length() > 2 ) {
warnings2.add(warningsStr);
}
if (res_F.results().get(entry.getKey()).errors().toString().length() > 2) {
errors2.add(res_F.results().get(entry.getKey()).errors().toString());
if ( errorsStr.length() > 2 ) {
errors2.add(errorsStr);
}
if (entry.getValue().toString().contains("SUCCESS")) {
score += 3;
@ -698,19 +712,21 @@ class R1_2_01M_SPEC extends AbstractGuideline<Document> {
try {
// System.out.println("\nMetadata includes provenance information according to a cross-community language");
XMLApplicationProfile.ValidationResult res_F = profile.validate(id, t);
Map<String, Result> results = res_F.results();
Map<String, Guideline.Result> results = res_F.results();
int MaxScoreR1_2_01M_SPEC = (int) ((res_F.score()*getWeight())/100);
ArrayList<String> warnings2 = new ArrayList<>();
ArrayList<String> errors2 = new ArrayList<>();
int score = 0;
for (Map.Entry entry : res_F.results().entrySet()) {
if (res_F.results().get(entry.getKey()).warnings().toString().length() > 2) {
warnings2.add(res_F.results().get(entry.getKey()).warnings().toString());
for (Map.Entry<String, Guideline.Result> entry : results.entrySet()) {
String key = entry.getKey();
String warningsStr = results.get(key).warnings().toString();
String errorsStr = results.get(key).errors().toString();
if ( warningsStr.length() > 2 ) {
warnings2.add(warningsStr);
}
if (res_F.results().get(entry.getKey()).errors().toString().length() > 2) {
errors2.add(res_F.results().get(entry.getKey()).errors().toString());
if ( errorsStr.length() > 2 ) {
errors2.add(errorsStr);
}
if (entry.getValue().toString().contains("SUCCESS")) {
score += 3;

50
src/main/java/eu/dnetlib/validator2/validation/guideline/openaire/LiteratureGuidelinesV3Profile.java
View File

@ -77,69 +77,69 @@ public final class LiteratureGuidelinesV3Profile extends AbstractOpenAireProfile
};
private static final ElementSpec TITLE_SPEC = Builders
.forMandatoryElement("title", ONE_TO_N)
.forMandatoryElement("dc:title", ONE_TO_N)
.build();
private static final ElementSpec CREATOR_SPEC = Builders
.forMandatoryElement("creator", ONE_TO_N)
.forMandatoryElement("dc:creator", ONE_TO_N)
.build();
private static final ElementSpec PROJECT_IDENTIFIER_SPEC = Builders
.forMandatoryIfApplicableElement("relation", ONE, elementIsPresent("relation"))
.forMandatoryIfApplicableElement("dc:relation", ONE, elementIsPresent("dc:relation"))
.allowedValues(new RegexValuePredicate(COMPILED_PROJECT_IDENTIFIER_REGEX))
.build();
private static final ElementSpec ACCESS_LEVEL_SPEC = Builders
.forMandatoryElement("rights", ONE).allowedValues(repoAccessTerms)
.forMandatoryElement("dc:rights", ONE).allowedValues(repoAccessTerms)
.build();
private static final ElementSpec LICENSE_CONDITION_SPEC = Builders
.forRecommendedRepeatableElement("rights")
.forRecommendedRepeatableElement("dc:rights")
.allowedValues(new RegexValuePredicate(COMPILED_LICENSE_CONDITION_REG_EX))
.build();
private static final ElementSpec EMBARGO_END_DATE_SPEC = Builders
.forMandatoryIfApplicableElement("date", ONE, applicabilityRuleForEmbargoEndDate())
.forMandatoryIfApplicableElement("dc:date", ONE, applicabilityRuleForEmbargoEndDate())
.allowedValues(new EmbargoedEndDateValuePredicate())
.build();
private static final ElementSpec ALT_IDENTIFIER_SPEC = Builders
.forRecommendedRepeatableElement("relation")
.forRecommendedRepeatableElement("dc:relation")
.allowedValues(new RegexValuePredicate(COMPILED_ALT_IDENTIFIER_REG_EX))
.build();
private static final ElementSpec PUBLICATION_REF_SPEC = Builders
.forRecommendedRepeatableElement("relation")
.forRecommendedRepeatableElement("dc:relation")
.allowedValues(new RegexValuePredicate(COMPILED_PUBLICATION_REFERENCE_REG_EX))
.build();
private static final ElementSpec DATASET_REF_SPEC = Builders
.forRecommendedRepeatableElement("relation")
.forRecommendedRepeatableElement("dc:relation")
.allowedValues(new RegexValuePredicate(COMPILED_DATASET_REFERENCE_REG_EX))
.build();
//TODO value is either a keyword (free text) or a classification (info:eu-repo/classification)
// v3 guideliness recommends ddc classification (Dewey Decimal Classification)
private static final ElementSpec SUBJECT_SPEC = Builders
.forMandatoryIfApplicableElement("subject", ONE_TO_N, elementIsPresent("subject"))
.forMandatoryIfApplicableElement("dc:subject", ONE_TO_N, elementIsPresent("dc:subject"))
.build();
private static final ElementSpec DESCRIPTION_SPEC = Builders
.forMandatoryIfApplicableElement("description", ONE_TO_N, elementIsPresent("description"))
.forMandatoryIfApplicableElement("dc:description", ONE_TO_N, elementIsPresent("dc:description"))
.build();
private static final ElementSpec PUBLISHER_SPEC = Builders
.forMandatoryIfApplicableElement("publisher", ONE_TO_N, elementIsPresent("publisher"))
.forMandatoryIfApplicableElement("dc:publisher", ONE_TO_N, elementIsPresent("dc:publisher"))
.build();
private static final ElementSpec CONTRIBUTOR_SPEC = Builders
.forRecommendedRepeatableElement("contributor")
.forRecommendedRepeatableElement("dc:contributor")
.build();
//TODO
// Search element -> dc:date AND NOT embargoedDate
private static final ElementSpec PUBLICATION_DATE_SPEC = Builders
.forMandatoryElement("date", ONE)
.forMandatoryElement("dc:date", ONE)
.allowedValues(new RegexValuePredicate(COMPILED_PUBLICATION_DATE_REG_EX))
.build();
@ -147,7 +147,7 @@ public final class LiteratureGuidelinesV3Profile extends AbstractOpenAireProfile
// Search element -> dc:type AND values IN publicationTypes set
//TODO: Check it is first occurrence
private static final ElementSpec PUBLICATION_TYPE_M_SPEC = Builders
.forMandatoryElement("type", ONE)
.forMandatoryElement("dc:type", ONE)
.atPosition(ElementPosition.FIRST)
.allowedValues(publicationTypes)
.build();
@ -156,47 +156,47 @@ public final class LiteratureGuidelinesV3Profile extends AbstractOpenAireProfile
// Search element -> dc:type AND values NOT IN publicationTypes
//TODO: check it is second occurrence
private static final ElementSpec PUBLICATION_TYPE_O_SPEC = Builders
.forOptionalElement("type")
.forOptionalElement("dc:type")
.atPosition(ElementPosition.SECOND)
.allowedValues(new Predicates.SetOfCaseInsensitiveAllowedValues(publicationTypesAndVersions).negate())
.build();
private static final ElementSpec PUBLICATION_VERSION_SPEC = Builders
.forRecommendedElement("type")
.forRecommendedElement("dc:type")
.allowedValues(publicationVersions)
.build();
private static final ElementSpec FORMAT_SPEC = Builders
.forRecommendedRepeatableElement("format")
.forRecommendedRepeatableElement("dc:format")
.allowedValues(new MediaTypesValuePredicate())
.build();
private static final ElementSpec RESOURCE_IDENTIFIER_SPEC = Builders
.forMandatoryElement("identifier", ONE_TO_N)
.forMandatoryElement("dc:identifier", ONE_TO_N)
.build();
private static final ElementSpec SOURCE_SPEC = Builders
.forRecommendedRepeatableElement("source")
.forRecommendedRepeatableElement("dc:source")
.build();
//TODO values from ISO 639-1 or 639-2 or 639-3 (recommended)
private static final ElementSpec LANGUAGE_SPEC = Builders
.forRecommendedRepeatableElement("language")
.forRecommendedRepeatableElement("dc:language")
.allowedValues(new ISO639ValuePredicate())
.build();
//TODO: Should exclude other dc:relation elements e.g. !containsAllowedValuesOF -> Project Identifier (MA), Alternative Identifier (R), Publication Reference (R), Dataset Reference (R)
private static final ElementSpec RELATION_SPEC = Builders
.forOptionalRepeatableElement("relation")
.forOptionalRepeatableElement("dc:relation")
.allowedValues(relationSpecAllowedValuesPredicate())
.build();
private static final ElementSpec COVERAGE_SPEC = Builders
.forRecommendedRepeatableElement("coverage")
.forRecommendedRepeatableElement("dc:coverage")
.build();
private static final ElementSpec AUDIENCE_SPEC = Builders
.forRecommendedRepeatableElement("audience")
.forRecommendedRepeatableElement("dc:audience")
.allowedValues(audiences)
.build();
@ -204,7 +204,7 @@ public final class LiteratureGuidelinesV3Profile extends AbstractOpenAireProfile
return XMLCardinalityRule.builder().
setId(ElementSpec.APPLICABILITY_RULE_ID).
// first predicate count(...) makes sure there is only one Access Level set and then the second predicate verifies its value.
setXPathExpression("//*[count(//*[local-name()='rights' and starts-with(normalize-space(text()), 'info:eu-repo/semantics/')])=1][local-name()='rights' and normalize-space(text())='info:eu-repo/semantics/embargoedAccess']").
setXPathExpression("//*[count(//*[name()='dc:rights' and starts-with(normalize-space(text()), 'info:eu-repo/semantics/')])=1][name()='dc:rights' and normalize-space(text())='info:eu-repo/semantics/embargoedAccess']").
setRange(1,1).
setIsInclusive(true).
build();

132
src/main/java/eu/dnetlib/validator2/validation/guideline/openaire/LiteratureGuidelinesV4Profile.java
View File

@ -137,61 +137,61 @@ public final class LiteratureGuidelinesV4Profile extends AbstractOpenAireProfile
};
private static final ElementSpec TITLE_SPEC = Builders.
forMandatoryElement("title", ONE_TO_N).
forMandatoryElement("datacite:title", ONE_TO_N).
withOptionalAttribute("xml:lang", new RegexValuePredicate(COMPILED_BCP47_LANG_TAGS_REG_EX)).
withOptionalAttribute("titleType", TITLE_TYPES).
build();
private static final ElementSpec CREATOR_SPEC = Builders.
forMandatoryElement("creator", ONE_TO_N).
forMandatoryElement("datacite:creator", ONE_TO_N).
withSubElement(Builders.
forMandatoryElement("creatorName", ONE).
forMandatoryElement("datacite:creatorName", ONE).
withRecommendedAttribute("nameType", NAME_TYPES)).
withSubElement(Builders.
forRecommendedElement("givenName")).
forRecommendedElement("datacite:givenName")).
withSubElement(Builders.
forRecommendedElement("familyName")).
forRecommendedElement("datacite:familyName")).
withSubElement(Builders.
forRecommendedRepeatableElement("nameIdentifier").
forRecommendedRepeatableElement("datacite:nameIdentifier").
withMandatoryAttribute("nameIdentifierScheme").
withRecommendedAttribute("schemeURI")).
withSubElement(Builders.
forRecommendedRepeatableElement("affiliation")).
forRecommendedRepeatableElement("datacite:affiliation")).
build();
private static final ElementSpec CONTRIBUTOR_SPEC = Builders.
forMandatoryIfApplicableElement("contributor", ONE_TO_N, elementIsPresent("contributor")).
forMandatoryIfApplicableElement("datacite:contributor", ONE_TO_N, elementIsPresent("datacite:contributor")).
withMandatoryAttribute("contributorType", CONTRIBUTOR_TYPES).
withSubElement(Builders.
forMandatoryElement("contributorName", ONE).
forMandatoryElement("datacite:contributorName", ONE).
withRecommendedAttribute("nameType", NAME_TYPES)).
withSubElement(Builders.
forOptionalElement("familyName")).
forOptionalElement("datacite:familyName")).
withSubElement(Builders.
forOptionalElement("givenName")).
forOptionalElement("datacite:givenName")).
withSubElement(Builders.
forRecommendedRepeatableElement("nameIdentifier").
forRecommendedRepeatableElement("datacite:nameIdentifier").
withMandatoryAttribute("nameIdentifierScheme").
withRecommendedAttribute("schemeURI")).
withSubElement(Builders.
forRecommendedRepeatableElement("affiliation")).
forRecommendedRepeatableElement("datacite:affiliation")).
build();
//This property has some issues/annotations in documentation
private static final ElementSpec FUNDING_REFERENCE_SPEC = Builders.
forMandatoryIfApplicableElement("fundingReference", ONE_TO_N, elementIsPresent("fundingReference")).
forMandatoryIfApplicableElement("oaire:fundingReference", ONE_TO_N, elementIsPresent("oaire:fundingReference")).
withSubElement(Builders.
forMandatoryElement("funderName", ONE)).
forMandatoryElement("oaire:funderName", ONE)).
withSubElement(Builders.
forRecommendedElement("funderIdentifier").
forRecommendedElement("oaire:funderIdentifier").
withRecommendedAttribute("funderIdentifierType", FUNDER_IDENTIFIER_TYPES)).
withSubElement(Builders.
forOptionalElement("fundingStream")).
forOptionalElement("oaire:fundingStream")).
withSubElement(Builders.
forMandatoryIfApplicableElement("awardNumber", ONE, elementIsPresent("awardNumber")).
forMandatoryIfApplicableElement("oaire:awardNumber", ONE, elementIsPresent("oaire:awardNumber")).
withRecommendedAttribute("awardURI")).
withSubElement(Builders.
forRecommendedElement("awardTitle")).
forRecommendedElement("oaire:awardTitle")).
build();
//TODO: Allowed values are referred as "suggested" in the documentation, but then a controlled list is given.
@ -199,12 +199,12 @@ public final class LiteratureGuidelinesV4Profile extends AbstractOpenAireProfile
// https://bitbucket.org/saikos/openaire-validator/issues/40
// https://bitbucket.org/saikos/openaire-validator/issues/32/
private static final ElementSpec ALTERNATE_IDENTIFIER_SPEC = Builders.
forRecommendedRepeatableElement("alternateIdentifier").
forRecommendedRepeatableElement("datacite:alternateIdentifier").
withMandatoryAttribute("alternateIdentifierType", IDENTIFIER_TYPES).
build();
private static final ElementSpec RELATED_IDENTIFIER_SPEC = Builders.
forRecommendedRepeatableElement("relatedIdentifier").
forRecommendedRepeatableElement("datacite:relatedIdentifier").
withMandatoryAttribute("relatedIdentifierType", IDENTIFIER_TYPES).
withMandatoryAttribute("relationType", RELATION_TYPES).
//TODO: For following 3 attributes. Need a way to target relationType attribute of current element
@ -223,13 +223,13 @@ public final class LiteratureGuidelinesV4Profile extends AbstractOpenAireProfile
//TODO: Implement proper applicability rule
//LEONIDAS: The withMandatoryAttribute fails whe another date element eg. for Publication Date exists
private static final ElementSpec EMBARGO_PERIOD_DATE_SPEC = Builders.
forMandatoryIfApplicableElement("date", ONE_TO_N, applicabilityRuleForEmbargoPeriodDate()).
forMandatoryIfApplicableElement("datacite:date", ONE_TO_N, applicabilityRuleForEmbargoPeriodDate()).
withMandatoryAttribute("dateType", EMBARGO_DATE_TYPES).
allowedValues(new RegexValuePredicate(COMPILED_YYYY_MM_DD_RANGE_REGEX).or(new RegexValuePredicate(COMPILED_YEAR_YYYY_REG_EX))).
build();
// private static final ElementSpec EMBARGO_PERIOD_DATE_SPEC = Builders.
// forMandatoryIfApplicableElement("date", TWO, applicabilityRuleForEmbargoPeriodDate()).
// forMandatoryIfApplicableElement("datacite:date", TWO, applicabilityRuleForEmbargoPeriodDate()).
// withMandatoryAttribute("dateType", EMBARGO_DATE_TYPES).
// build();
@ -237,12 +237,12 @@ public final class LiteratureGuidelinesV4Profile extends AbstractOpenAireProfile
There are no "strict" allowed values. Recommendations are IETF BCP 47 and ISO 639-x
*/
private static final ElementSpec LANGUAGE_SPEC = Builders.
forMandatoryIfApplicableElement("language", ONE_TO_N, elementIsPresent("language")).
forMandatoryIfApplicableElement("dc:language", ONE_TO_N, elementIsPresent("dc:language")).
allowedValues(new RegexValuePredicate(COMPILED_BCP47_LANG_TAGS_REG_EX).or(new ISO639ValuePredicate())).
build();
private static final ElementSpec PUBLISHER_SPEC = Builders.
forMandatoryIfApplicableElement("publisher", ONE_TO_N, elementIsPresent("publisher")).
forMandatoryIfApplicableElement("dc:publisher", ONE_TO_N, elementIsPresent("dc:publisher")).
build();
/*
@ -251,98 +251,98 @@ public final class LiteratureGuidelinesV4Profile extends AbstractOpenAireProfile
"Recommended" best practice for encoding the date value is ISO 8601 [W3CDTF] (YYYY-MM-DD) (YYYY mandatory)
*/
private static final ElementSpec PUBLICATION_DATE_SPEC = Builders.
forMandatoryElement("date", ONE).
forMandatoryElement("datacite:date", ONE).
withMandatoryAttribute("dateType", PUBLICATION_DATE_TYPE).
allowedValues(new RegexValuePredicate(COMPILED_PUBLICATION_DATE_REG_EX).or(new RegexValuePredicate(COMPILED_YYYY_MM_DD_RANGE_REGEX).or(new RegexValuePredicate(COMPILED_YEAR_YYYY_REG_EX)))).
build();
private static final ElementSpec RESOURCE_TYPE_SPEC = Builders.
forMandatoryElement("resourceType", ONE).
forMandatoryElement("oaire:resourceType", ONE).
withMandatoryAttribute("resourceTypeGeneral", RESOURCE_GENERAL_TYPES).
withMandatoryAttribute("uri", RESOURCE_CONCEPT_URIS).
build();
private static final ElementSpec DESCRIPTION_SPEC = Builders.
forMandatoryIfApplicableElement("description", ONE_TO_N, elementIsPresent("description")).
forMandatoryIfApplicableElement("dc:description", ONE_TO_N, elementIsPresent("dc:description")).
withOptionalAttribute("xml:lang", new RegexValuePredicate(COMPILED_BCP47_LANG_TAGS_REG_EX)).
build();
private static final ElementSpec FORMAT_SPEC = Builders.
forRecommendedRepeatableElement("format").
forRecommendedRepeatableElement("dc:format").
allowedValues(new MediaTypesValuePredicate()).
build();
private static final ElementSpec RESOURCE_IDENTIFIER_SPEC = Builders.
forMandatoryElement("identifier", ONE).
forMandatoryElement("datacite:identifier", ONE).
withMandatoryAttribute("identifierType", RESOURCE_IDENTIFIER_TYPES).
build();
private static final ElementSpec ACCESS_RIGHTS_SPEC = Builders.
forMandatoryElement("rights", ONE).
forMandatoryElement("datacite:rights", ONE).
withMandatoryAttribute("uri", ACCESS_RIGHTS_URIS).
allowedValues(ACCESS_RIGHTS_TYPES).
build();
private static final ElementSpec SOURCE_SPEC = Builders.
forRecommendedRepeatableElement("source").
forRecommendedRepeatableElement("dc:source").
build();
//TODO: Should we check URI attribute values are valid?
private static final ElementSpec SUBJECT_SPEC = Builders.
forMandatoryIfApplicableElement("subject", ONE_TO_N, elementIsPresent("subject")).
forMandatoryIfApplicableElement("datacite:subject", ONE_TO_N, elementIsPresent("datacite:subject")).
withOptionalAttribute("subjectScheme").
withOptionalAttribute("schemeURI").
withOptionalAttribute("valueURI").
build();
private static final ElementSpec LICENSE_CONDITION_SPEC = Builders.
forRecommendedElement("licenseCondition").
withMandatoryIfApplicableAttribute("uri", elementIsPresent("licenseCondition")).
withMandatoryIfApplicableAttribute("startDate", elementIsPresent("licenseCondition")).
forRecommendedElement("oaire:licenseCondition").
withMandatoryIfApplicableAttribute("uri", elementIsPresent("oaire:licenseCondition")).
withMandatoryIfApplicableAttribute("startDate", elementIsPresent("oaire:licenseCondition")).
build();
private static final ElementSpec COVERAGE_SPEC = Builders.
forRecommendedRepeatableElement("coverage").
forRecommendedRepeatableElement("dc:coverage").
build();
private static final ElementSpec SIZE_SPEC = Builders.
forOptionalRepeatableElement("size").
forOptionalRepeatableElement("datacite:size").
build();
private static final ElementSpec GEO_LOCATION_SPEC = Builders.
forOptionalRepeatableElement("geoLocation").
forOptionalRepeatableElement("datacite:geoLocation").
withSubElement(Builders.
forOptionalElement("geoLocationPoint").
forOptionalElement("datacite:geoLocationPoint").
withSubElement(Builders.
forMandatoryElement("pointLongitude", ONE)).
forMandatoryElement("datacite:pointLongitude", ONE)).
withSubElement(Builders.
forMandatoryElement("pointLatitude", ONE))).
forMandatoryElement("datacite:pointLatitude", ONE))).
withSubElement(Builders.
forOptionalElement("geoLocationBox").
forOptionalElement("datacite:geoLocationBox").
withSubElement(Builders.
forMandatoryElement("westBoundLongitude", ONE)).
forMandatoryElement("datacite:westBoundLongitude", ONE)).
withSubElement(Builders.
forMandatoryElement("eastBoundLongitude", ONE)).
forMandatoryElement("datacite:eastBoundLongitude", ONE)).
withSubElement(Builders.
forMandatoryElement("southBoundLatitude", ONE)).
forMandatoryElement("datacite:southBoundLatitude", ONE)).
withSubElement(Builders.
forMandatoryElement("northBoundLatitude", ONE))).
forMandatoryElement("datacite:northBoundLatitude", ONE))).
withSubElement(Builders.
forOptionalElement("geoLocationPlace")).
forOptionalElement("datacite:geoLocationPlace")).
withSubElement(Builders.
forOptionalRepeatableElement("geoLocationPolygon").
forOptionalRepeatableElement("datacite:geoLocationPolygon").
withSubElement(Builders.
forMandatoryElement("polygonPoint", FOUR_TO_N).
forMandatoryElement("datacite:polygonPoint", FOUR_TO_N).
withSubElement(Builders.
forMandatoryElement("pointLongitude", ONE)).
forMandatoryElement("datacite:pointLongitude", ONE)).
withSubElement(Builders.
forMandatoryElement("pointLatitude", ONE))).
forMandatoryElement("datacite:pointLatitude", ONE))).
withSubElement(Builders.
forOptionalElement("inPolygonPoint").
forOptionalElement("datacite:inPolygonPoint").
withSubElement(Builders.
forMandatoryElement("pointLongitude", ONE)).
forMandatoryElement("datacite:pointLongitude", ONE)).
withSubElement(Builders.
forMandatoryElement("pointLatitude", ONE)))).
forMandatoryElement("datacite:pointLatitude", ONE)))).
build();
/*
@ -354,57 +354,57 @@ public final class LiteratureGuidelinesV4Profile extends AbstractOpenAireProfile
TODO: Should we cross-check attribute and element value are relevant?
*/
private static final ElementSpec RESOURCE_VERSION_SPEC = Builders.
forRecommendedElement("version").
forRecommendedElement("oaire:version").
withMandatoryIfApplicableAttribute("uri", applicabilityRuleForURIAttributeOfResourceVersion(), RESOURCE_VERSION_URIS).
build();
//TODO: Has annotation/issue: accessRightsURI attribute values also appears on ACCESS_RIGHTS_SPEC. Should check it's the same?
private static final ElementSpec FILE_LOCATION_SPEC = Builders.
forMandatoryIfApplicableElement("file", ONE_TO_N, elementIsPresent("file")).
forMandatoryIfApplicableElement("oaire:file", ONE_TO_N, elementIsPresent("oaire:file")).
withRecommendedAttribute("accessRightsURI", ACCESS_RIGHTS_URIS).
withRecommendedAttribute("mimeType", new MediaTypesValuePredicate()).
withRecommendedAttribute("objectType", FILE_OBJECT_TYPES).
build();
private static final ElementSpec CITATION_TITLE_SPEC = Builders.
forRecommendedElement("citationTitle").
forRecommendedElement("oaire:citationTitle").
build();
private static final ElementSpec CITATION_VOLUME_SPEC = Builders.
forRecommendedElement("citationVolume").
forRecommendedElement("oaire:citationVolume").
build();
private static final ElementSpec CITATION_ISSUE_SPEC = Builders.
forRecommendedElement("citationIssue").
forRecommendedElement("oaire:citationIssue").
build();
private static final ElementSpec CITATION_START_PAGE_SPEC = Builders.
forRecommendedElement("citationStartPage").
forRecommendedElement("oaire:citationStartPage").
build();
private static final ElementSpec CITATION_END_PAGE_SPEC = Builders.
forRecommendedElement("citationEndPage").
forRecommendedElement("oaire:citationEndPage").
build();
private static final ElementSpec CITATION_EDITION_SPEC = Builders.
forRecommendedElement("citationEdition").
forRecommendedElement("oaire:citationEdition").
build();
private static final ElementSpec CITATION_CONFERENCE_PLACE_SPEC = Builders.
forRecommendedElement("citationConferencePlace").
forRecommendedElement("oaire:citationConferencePlace").
build();
//TODO: Implement regex/allowedValuesPredicate
// Date has recommended best practice ISO 8601 [W3CDTF], and two [single date] [start date - end date] formats
private static final ElementSpec CITATION_CONFERENCE_DATE_SPEC = Builders.
forRecommendedElement("citationConferenceDate").
forRecommendedElement("oaire:citationConferenceDate").
allowedValues(new RegexValuePredicate(COMPILED_YYYY_MM_DD_REGEX).or(new RegexValuePredicate(COMPILED_YYYY_MM_DD_RANGE_REGEX))).
build();
//TODO: A non-exhaustive list is provided for values, derived from the Common Education Data Standards vocabulary
// Should we add it?
private static final ElementSpec AUDIENCE_SPEC = Builders.
forOptionalRepeatableElement("audience").
forOptionalRepeatableElement("dcterms:audience").
allowedValues(AUDIENCE_VOCABULARY).
build();

6
src/test/groovy/eu/dnetlib/validator2/engine/XMLHelper.groovy
View File

@ -1,6 +1,6 @@
package eu.dnetlib.validator2.engine
import eu.dnetlib.validator2.engine.Helper
import groovy.xml.DOMBuilder
import groovy.xml.FactorySupport
import org.w3c.dom.Document
@ -31,8 +31,8 @@ class XMLHelper {
int bufferSize,
Predicate<ZipEntry> filter,
BiConsumer < String, Document > documentConsumer) {
if (documentConsumer) {
if (documentConsumer)
{
DocumentBuilderFactory factory = FactorySupport.createDocumentBuilderFactory()
factory.setNamespaceAware(true)
factory.setValidating(false)

3
src/test/java/eu/dnetlib/validator2/engine/Test.java
View File

@ -21,7 +21,8 @@ public class Test {
// "src/test/resources/openaireguidelinesV3/dia.library.tuc.gr/Record_21811.xml",
// "src/test/resources/openaireguidelinesV3/cris.vtt.fi/01.xml",
// "src/test/resources/openaireguidelinesV3/cris.vtt.fi/02.xml",
"src/test/resources/openaireguidelinesV3/cris.vtt.fi/03.xml"
// "src/test/resources/openaireguidelinesV3/cris.vtt.fi/03.xml"
"src/test/resources/openaireguidelinesV3/ngi.brage.unit.no/ListRecords_prefix_oai_dc_set_openaire.xml"
};
public static void main(String[] args) {

4
src/test/java/eu/dnetlib/validator2/engine/Test_FAIR.java
View File

@ -24,8 +24,8 @@ public class Test_FAIR {
"src/test/resources/openaireguidelinesV3/cris.vtt.fi/01_data.xml",
"src/test/resources/openaireguidelinesV3/cris.vtt.fi/02_data.xml",
"src/test/resources/openaireguidelinesV3/cris.vtt.fi/04_data.xml",
// "src/test/resources/openaireguidelinesV3/cris.vtt.fi/03_data.xml",
// "src/test/resources/openaireguidelinesV4/01_gv4.xml"
"src/test/resources/openaireguidelinesV3/cris.vtt.fi/03_data.xml",
//"src/test/resources/openaireguidelinesV4/01_gv4.xml" // This file does not exist!
};
public static void main(String[] args) {

4
src/test/java/eu/dnetlib/validator2/engine/Test_v4.java
View File

@ -20,8 +20,8 @@ public class Test_v4 {
private static final String[] FILES = new String[] {
"src/test/resources/openaireguidelinesV4/v4_literature_all_invalid_guidelines_record.xml",
// "src/test/resources/openaireguidelinesV4/v4_literature_all_guidelines_record.xml",
//"src/test/resources/openaireguidelinesV4/v4_literature_all_invalid_guidelines_record.xml",
"src/test/resources/openaireguidelinesV4/v4_literature_all_guidelines_record.xml",
// "src/test/resources/openaireguidelinesV4/oai_mediarep_org_doc_2534.xml",
// "src/test/resources/openaireguidelinesV4/01_gv4.xml"
};