private Graph GenerateDefinitionsGraph(JObject metadataJson)
{
var definitionsGraph = new Graph();
var metadataExtractor = new MetdataExtractor();
ProgressLog.Info("Extracting column property definitions");
metadataExtractor.GenerateColumnDefinitions(metadataJson, definitionsGraph);
return(definitionsGraph);
}
public void ItAssertsRdfType()
{
var ex = new MetdataExtractor();
var obj = new JObject(new JProperty("url", "http://datadock.io/test/repo/data"));
var g = new Graph();
ex.Run(obj, g, new Uri("http://datadock.io/test/repo/publisher"), 100, null);
var expectSubject = g.CreateUriNode(new Uri("http://datadock.io/test/repo/data"));
GraphShouldContainRdfTypeStatement(g, expectSubject);
}
public void ItAssertsVoidTriples()
{
var ex = new MetdataExtractor();
var obj = new JObject(new JProperty("url", "http://datadock.io/test/repo/data"));
var g = new Graph();
ex.Run(obj, g, new Uri("http://datadock.io/test/repo/publisher"), 100, null);
var expectSubject = g.CreateUriNode(new Uri("http://datadock.io/test/repo/data"));
var voidTriples = g.CreateUriNode(new Uri("http://rdfs.org/ns/void#triples"));
var expectTripleCount = g.CreateLiteralNode("100", new Uri("http://www.w3.org/2001/XMLSchema#integer"));
var statements = g.GetTriplesWithSubjectPredicate(expectSubject, voidTriples);
statements.Count().Should().Be(1);
statements.Should().Contain(x => x.Object.Equals(expectTripleCount));
}
public void ItCanProcessDcTitleStringField()
{
var ex = new MetdataExtractor();
var obj = new JObject(new JProperty("url", "http://datadock.io/test/repo/data"),
new JProperty("dc:title", "This is a title"));
var g = new Graph();
ex.Run(obj, g, new Uri("http://datadock.io/test/repo/publisher"), 100, null);
var expectSubject = g.CreateUriNode(new Uri("http://datadock.io/test/repo/data"));
var expectPredicate = g.CreateUriNode(new Uri("http://purl.org/dc/terms/title"));
var titleStatement = g.GetTriplesWithSubjectPredicate(expectSubject, expectPredicate).FirstOrDefault();
titleStatement.Should().NotBeNull();
titleStatement.Object.Should().BeAssignableTo <ILiteralNode>().Which.Value.Should().Be("This is a title");
}
public void ItCanProcessDcLicenseUrlField()
{
var ex = new MetdataExtractor();
var obj = new JObject(new JProperty("url", "http://datadock.io/test/repo/data"),
new JProperty("dc:license", "http://creativecommons.org/ns#cc-0"));
var g = new Graph();
ex.Run(obj, g, new Uri("http://datadock.io/test/repo/publisher"), 100, null);
var expectSubject = g.CreateUriNode(new Uri("http://datadock.io/test/repo/data"));
var expectPredicate = g.CreateUriNode(new Uri("http://purl.org/dc/terms/license"));
var licenseStatement = g.GetTriplesWithSubjectPredicate(expectSubject, expectPredicate).FirstOrDefault();
licenseStatement.Should().NotBeNull();
licenseStatement.Object.Should()
.BeAssignableTo <IUriNode>()
.Which.Uri.Should()
.Be(new Uri("http://creativecommons.org/ns#cc-0"));
}
public void ItAssertsModifiedDate()
{
var ex = new MetdataExtractor();
var obj = new JObject(new JProperty("url", "http://datadock.io/test/repo/data"),
new JProperty("dc:license", "http://creativecommons.org/ns#cc-0"));
var g = new Graph();
ex.Run(obj, g, new Uri("http://datadock.io/test/repo/publisher"), 100, new DateTime(2017, 01, 02, 03, 04, 05));
var expectSubject = g.CreateUriNode(new Uri("http://datadock.io/test/repo/data"));
var expectPredicate = g.CreateUriNode(new Uri("http://purl.org/dc/terms/modified"));
var statements = g.GetTriplesWithSubjectPredicate(expectSubject, expectPredicate).ToList();
statements.Count.Should().Be(1);
var val = statements[0].Object as ILiteralNode;
val.Should().NotBeNull();
val.Value.Should().Be("2017-01-02");
val.DataType.ToString().Should().Be("http://www.w3.org/2001/XMLSchema#date");
}
private Graph GenerateMetadataGraph(Uri datasetUri, Uri publisherIri, JObject metadataJson, IEnumerable <Uri> downloadUris, IGraph dataGraph)
{
var metadataGraph = new Graph();
var metadataExtractor = new MetdataExtractor();
ProgressLog.Info("Extracting dataset metadata");
metadataExtractor.Run(metadataJson, metadataGraph, publisherIri, dataGraph.Triples.Count, DateTime.UtcNow);
var dsNode = metadataGraph.CreateUriNode(datasetUri);
var ddNode = metadataGraph.CreateUriNode(new Uri("http://rdfs.org/ns/void#dataDump"));
var exampleResource = metadataGraph.CreateUriNode(new Uri("http://rdfs.org/ns/void#exampleResource"));
foreach (var downloadUri in downloadUris)
{
metadataGraph.Assert(dsNode, ddNode, metadataGraph.CreateUriNode(downloadUri));
}
foreach (var distinctSubject in dataGraph.Triples.Select(t => t.Subject).OfType <IUriNode>().Distinct().Take(10))
{
metadataGraph.Assert(dsNode, exampleResource, distinctSubject);
}
return(metadataGraph);
}
public void ItCanProcessMultipleDcatKeywords()
{
var ex = new MetdataExtractor();
var obj = new JObject(new JProperty("url", "http://datadock.io/test/repo/data"),
new JProperty("dcat:keyword", new JArray(new JValue("one"), new JValue("two"), new JValue("three"))));
var g = new Graph();
ex.Run(obj, g, new Uri("http://datadock.io/test/repo/publisher"), 100, null);
var expectSubject = g.CreateUriNode(new Uri("http://datadock.io/test/repo/data"));
var expectPredicate = g.CreateUriNode(new Uri("http://www.w3.org/ns/dcat#keyword"));
var statements = g.GetTriplesWithSubjectPredicate(expectSubject, expectPredicate).ToList();
statements.Count.Should().Be(3);
foreach (var statement in statements)
{
statement.Object.Should()
.BeAssignableTo <ILiteralNode>()
.Which.Value.Should()
.BeOneOf("one", "two", "three");
}
}
