Tightening up the tests a lot

lib/catalog.js

@@ -31,21 +31,20 @@ const RELATED = [ 'publications', 'websites', 'metadata', 'data' ];
 const RELATED_TYPES = {
 	'publications': 'ScholarlyArticle',
 	'websites': 'WebSite',
-	'metadata': 'DataSet',  // schema.org hasn't got a dedicated type
+	'metadata': 'CreativeWork',  // schema.org hasn't got a dedicated type
 	'data': 'DataSet'
 };
 
+const ROOT_PATH = 'data/';
+
 const FUNDING_FIELDS = [ 'foaf:fundedBy_foaf:Agent', 'foaf:fundedBy_vivo:Grant' ];
 
 
 const SUBJECT_FIELDS = [ 'dc:subject_anzsrc:for', 'dc:subject_anzsrc:seo' ];
 
-// these will do for now
-
 const SUBJECT_IRI_PREFIX = {
     'dc:subject_anzsrc:for': '_:FOR/',
-    'dc:subject_anzsrc:seo': '_:SEO/',
-    'finalKeywords': '_:keywords/'
+    'dc:subject_anzsrc:seo': '_:SEO/'
 };
 
 	// dataset -> about -> subjects
@@ -164,11 +163,12 @@ async function datapub2catalog(options) {
 			const eitems = items.filter((i) => i);
 			if( eitems ) {
 				dataset[property] = items.map(link_id);
-				graph.push(eitems);
+				graph.push(..._.flatten(eitems));
 			}
 		}
 	});
 
+	// todo: trim context here?
 	return {
 		'@context': context,
 		'@graph': graph.filter((e) => e)
@@ -199,7 +199,7 @@ function make_dataset(id, datapub, organisation, dates, files) {
 	const ds = {
 		'@id': id,
 		'@type': 'Dataset',
-		'path': './',
+		'path': ROOT_PATH,
 		'name': datapub['title'],
 		'description': datapub['description'],
 		'dateCreated': dates['dateCreated'],
@@ -207,6 +207,10 @@ function make_dataset(id, datapub, organisation, dates, files) {
 		'publisher': link_id(organisation),
 		'hasPart': files.map(link_id),
 	};
+	// see https://schema.org/keywords
+	if( datapub['finalKeywords'] ) {
+		ds['keywords'] = datapub['finalKeywords'].join(', ');
+	}
 	return ds;
 }
 
@@ -287,9 +291,9 @@ function make_files(datapub) {
 				'name': dl['name'],
 				'path': dl['name'],
 				'@type': 'File',
+				'description': dl['notes'],
 				'contentSize': dl['contentSize'],
-				'encodingFormat': dl['encodingFormat'],
-				'fileFormat': dl['fileFormat']
+				'encodingFormat': dl['mimeType']
 			}
 		} else {
 			// todo: URLs and physical locations
@@ -386,17 +390,23 @@ function make_funders(datapub) {
 
 }
 
+// note - keywords are added on the dataset item
+
 function make_subjects(datapub) {
 	return SUBJECT_FIELDS.map((field) => {
 		if( datapub[field] ) {
-			return datapub[field].map((subj) => {
-				const id = SUBJECT_IRI_PREFIX[field] + subj;
+			return datapub[field].map((cat) => {
+				const id = SUBJECT_IRI_PREFIX[field] + cat['notation'];
 				return {
 					'@id': id,
+					'@type': 'StructuredValue',
+					'url': id,
 					'identifier': id,
-					'name': subj 
+					'name': cat['name'] 
 				};
 			})
+		} else {
+			return [];
 		}
 	});
 }

test/catalog.spec.js

@@ -77,21 +77,33 @@ describe("Convert a ReDBox 2.0 DataPub to CATALOG.json", () => {
 			'owner': OWNER,
 			'approver': APPROVER
 		});
-		const roots = cj['@graph'].filter((item) => {item['path'] === './'});
+		const roots = cj['@graph'].filter((item) => {return item['path'] === 'data/'});
 		cjds = roots[0];
-		console.log("cjds = " + JSON.stringify(cjds));
-		console.log("cj = " + JSON.stringify(cj));
+		await fs.writeJson('./test_data/CATALOG.json', cj, { 'spaces': 4 });
+	});
+
+	it("has a root dataset", () => {
+  		expect(cjds).to.be.a('object');
+	});
+
+	it("has string @id values for every graph item", () => {
+		cj['@graph'].map((i) => {
+			console.log("graph items @id: " + JSON.stringify(i['@id']));
+			expect(i).to.have.property('@id');
+			expect(i['@id']).to.be.a('string');
+		})
 	});
 
 
-	it("has a dataset item with correct metadata in the catalog", () => {
+	it("has a dataset item with correct metadata", () => {
 		assert(cj, "Got an object");
 		assert(cj['@graph'], "Catalog has a @graph");
 		const dataset = get_id(cj, DATASET_ID);
 		assert(dataset, "Graph has an item with id " + DATASET_ID);
 		expect(dataset['name']).to.equal(dp['title']);
 		expect(dataset['description']).to.equal(dp['description']);
-		expect(dataset['publisher']['@id']).to.equal(ORG['id']);
+		expect(dataset['publisher']['@id']).to.equal(ORG['id'])
+		expect(dataset['keywords']).to.equal(dp['finalKeywords'].join(', '));
 	});
 
 	it("has a create action with the owner", () => {
@@ -127,33 +139,40 @@ describe("Convert a ReDBox 2.0 DataPub to CATALOG.json", () => {
 // };
 
 	it("has subjects", () => {
-		// finalKeywords, FORs and SEOs
-		const fields = [ 'finalKeywords', 'dc:subject_anzsrc:for', 'dc:subject_anzsrc:seo'];
+		// FORs and SEOs
+		const fields = [ 'dc:subject_anzsrc:for', 'dc:subject_anzsrc:seo'];
+		const SUBJECT_IRI_PREFIX = {
+    	'dc:subject_anzsrc:for': '_:FOR/',
+    	'dc:subject_anzsrc:seo': '_:SEO/'
+		};
 		const about = cjds['about'].map((i) => i['@id']);
 		_.forEach(fields, (field) => {
-			const subjects = cj.filter((i) => { i['name'] === dp[field]});
-			expect(subjects.length).to.equal(1);
+			const pref = SUBJECT_IRI_PREFIX[field];
+			const expectsubs = dp[field].map((f) => { return f['name'] });
+			const subjects = cj['@graph'].filter((i) => { return i['@id'].startsWith(pref) });
+			const gotsubs = subjects.map((f) => { return f['name']} );
+			expect(gotsubs.sort()).to.eql(expectsubs.sort());			
 		})
 	});
 
 	it("has spatial coverage", () => {
-
+		assert(false);
 	});
 
 	it("has temporal coverage", () => {
-
+		assert(false);
 	});
 
 	it("has funders", () => {
-
+		assert(false);
 	});
 
 	it("has a licence", () => {
-
+		assert(false);
 	});
 
 	it("has related works", () => {
-
+		assert(false);
 	});
 
 

test_data/datapub.json

@@ -114,7 +114,30 @@
                 ]
             }
         ],
-        "dc:subject_anzsrc:seo" : [],
+        "dc:subject_anzsrc:seo" : [ 
+            {
+                "name" : "81 - DEFENCE",
+                "label" : "DEFENCE",
+                "notation" : "81"
+            }, 
+            {
+                "name" : "8201 - FORESTRY",
+                "label" : "FORESTRY",
+                "notation" : "8201",
+                "geneaology" : [ 
+                    "82"
+                ]
+            }, 
+            {
+                "name" : "830101 - Aquaculture Crustaceans (excl. Rock Lobster and Prawns)",
+                "label" : "Aquaculture Crustaceans (excl. Rock Lobster and Prawns)",
+                "notation" : "830101",
+                "geneaology" : [ 
+                    "83", 
+                    "8301"
+                ]
+            }
+        ],
         "startDate" : "2018-11-18",
         "endDate" : "2018-12-02",
         "timePeriod" : "The Anthropocene",