Now makes history actions (create and publish) and adds people items for

the agents of these who weren't already listed as creators

lib/catalog.js

@@ -18,28 +18,51 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 const CONTEXT = './defaults/context.json';
 
-fs = require('fs-extra');
+const fs = require('fs-extra');
+const _ = require('lodash');
 
 
-/* Convert metadata from a redbox data publication and
-   its data record to a JSON-LD datacrate catalogue */
+/* datapub2catalog(options)
 
+Convert metadata from a redbox data publication and
+its data record to a JSON-LD datacrate catalogue 
+
+options = {
+	id: redbox oid,
+	datapub: redbox data publication record,
+	org: {
+	  @id: organisation identifier,
+	  name: organisation name
+	},
+	owner: email of the data pub's record,
+	approver: email of the librarian who approved publication
+}
+
+The owner and approver are used to build two Actions against
+the object - creation and publication.
+
+
+TODO; add contentSize, encodingFormat and fileFormat from
+siegfried to the dataLocations array
+
+Remember to keep the connection to the data record
+
+*/
 
-// TODO; add contentSize, encodingFormat and fileFormat
-// from siegfried to the dataLocations array
-// org is the organisation as an array with keys @id
-// and name:
-// { 
-//   "@id": "https://www.uts.edu.au/"
-//   "name": "University of Technology Sydney",
-// }
 
-async function datapub2catalog(id, datapub, org) {
+
+async function datapub2catalog(options) {
+
+	const id = options['id'];
+	const datapub = options['datapub'];
+	const org = options['organisation'];
+	const owner = options['owner'];
+	const approver = options['approver'];
+
 	const context = await fs.readJson(CONTEXT);
 
 	const catalog = {
-		'@context': context,
-		'@graph': []
+		'@context': context
 	};
 
 	const organisation = { 
@@ -51,25 +74,36 @@ async function datapub2catalog(id, datapub, org) {
 
 	// assuming that all creators are affiliated to this one organisation
 
-	const people = creators(datapub, organisation);
-	const files = attachments(datapub);
+	var people = make_creators(datapub, organisation);
+
+	const files = make_files(datapub);
+	const dataset = make_dataset(id, datapub, organisation, files);
+	const [ history, more_people ] = make_history(dataset, people, owner, approver);
+
+	if( more_people.length > 0 ) {
+		people = people.concat(more_people);
+	}
 
-	const graph = [
-		dataset(id, datapub, organisation, files),
+	const graph = _.flatten([
+		dataset,
 		organisation, 
-		funding(datapub),
-		about(datapub),
-		spatial(datapub),
-		temporal(datapub),
+		make_funding(datapub),
+		make_about(datapub),
+		make_spatial(datapub),
+		make_temporal(datapub),
 		people,
-		licence(datapub),
-		citation(datapub),
+		make_licence(datapub),
+		make_citation(datapub),
 		files,
-		related(datapub)
-	];
+		make_related(datapub),
+		history
+	]);
+
 
-	catalog['@graph'] = graph.filter((e) => e);
-	return catalog;
+	return {
+		'@context': context,
+		'@graph': graph.filter((e) => e)
+	};
 }
 
 
@@ -77,8 +111,9 @@ function link_id(item) {
 	return { '@id': item['@id'] };
 }
 
+// TODO: isBasedOn the data record
 
-function dataset(id, datapub, organisation, files) {
+function make_dataset(id, datapub, organisation, files) {
 		return {
 			'@id': id,
 			'type': 'Dataset',
@@ -86,11 +121,58 @@ function dataset(id, datapub, organisation, files) {
 			'description': datapub['description'],
 			'Publisher': link_id(organisation),
 			'hasPart': files.map(link_id),
-
-
 		}
 }
 
+function find_by_email(people, email) {
+	const match = people.filter((x) => x['email'] == email);
+	if( match.length ) {
+		return match[0];
+	} else {
+		return undefined;
+	}
+}	
+
+function make_history(dataset, people, owner, approver) {
+	var owner_item = find_by_email(people, owner);
+	var approver_item = find_by_email(people, approver);
+	const new_people = []
+	if( !owner_item ) {
+		owner_item = { 
+			'@id': owner,
+			'email': owner
+		};
+		new_people.push(owner_item)
+	}
+	if( !approver_item ) {
+		approver_item = { 
+			'@id': approver,
+			'email': approver
+		};
+		new_people.push(approver_item)
+	}
+	// TODO: descriptions for these and maybe a link to the 
+	// data record for the CreateAction
+  const history = [
+		{
+			'@id': dataset['@id'] + '_h0',
+			'@type': 'CreateAction',
+			'name': 'Create',
+			'object': link_id(dataset),
+			'agent': link_id(owner_item),
+			'startTime': 'yyyy-mm-dd'
+		},
+		{
+			'@id': dataset['@id'] + '_h1',
+			'@type': 'UpdateAction',
+			'name': 'Publish',
+			'object': link_id(dataset),
+			'agent': link_id(approver_item),
+			'startTime': 'yyyy-mm-dd'
+		}
+	];
+	return [ history, new_people ];
+}
 
 
 // files(datapub)
@@ -99,7 +181,7 @@ function dataset(id, datapub, organisation, files) {
 // The dataLocations are expected to have contentSize, 
 // encodingFormat and fileFormat already set
 
-function attachments(datapub) {
+function make_files(datapub) {
 	if( !datapub['dataLocations'] ) {
 		return [];
 	}
@@ -128,7 +210,7 @@ function attachments(datapub) {
 
 // TODO: affiliations
 
-function creators(datapub, organisation) {
+function make_creators(datapub, organisation) {
 	if( !datapub['creators'] ) {
 		throw Error("A DataCrate has to have at least one creator");
 	}	
@@ -149,37 +231,38 @@ function creators(datapub, organisation) {
 			return undefined;
 		}
 	})
-
-	return creators.filter((x) => x)
-
+	console.log("make_creators " + typeof(creators));
+	const nc = creators.filter((x) => x)
+	console.log("make_creators " + typeof(nc));
+	return nc;
 }
 
 
-function funding(datapub) {
+function make_funding(datapub) {
 	return undefined;
 }
 
-function about(datapub) {
+function make_about(datapub) {
 	return undefined;
 }
 
-function spatial(datapub) {
+function make_spatial(datapub) {
 	return undefined;
 }
 
-function temporal(datapub) {
+function make_temporal(datapub) {
 	return undefined;
 }
 
-function licence(datapub) {
+function make_licence(datapub) {
 	return undefined;
 }
 
-function citation(datapub) {
+function make_citation(datapub) {
 	return undefined;
 }
 
-function related(datapub) {
+function make_related(datapub) {
 	return undefined;
 }
 

package-lock.json

@@ -405,6 +405,11 @@
         "verror": "1.10.0"
       }
     },
+    "lodash": {
+      "version": "4.17.11",
+      "resolved": "https://registry.npmjs.org/lodash/-/lodash-4.17.11.tgz",
+      "integrity": "sha512-cQKh8igo5QUhZ7lg38DYWAxMvjSAKG0A8wGSVimP07SIUEK2UO+arSRKbRZWtelMtN5V0Hkwh5ryOto/SshYIg=="
+    },
     "mime-db": {
       "version": "1.36.0",
       "resolved": "https://registry.npmjs.org/mime-db/-/mime-db-1.36.0.tgz",

package.json

@@ -15,7 +15,8 @@
   "repository": "https://code.research.uts.edu.au/eresearch/datacrate",
   "dependencies": {
     "fs-extra": "^7.0.0",
-    "jsonld": "^1.1.0"
+    "jsonld": "^1.1.0",
+    "lodash": "^4.17.11"
   },
   "devDependencies": {
     "chai": "^4.1.2",

test/catalog.spec.js

@@ -34,14 +34,23 @@ const ORG = {
 	'name': 'University of Technology Sydney'
 };
 
+const OWNER = 'Michael.Lynch@uts.edu.au';
+const APPROVER = 'data.librarian@uts.edu.au';
+
 const DATASET_ID = 'https://data.research.uts.edu.au/public/XYZ';
 
 describe("ReDBox data publication to DataCrate catalog.json", () => {
 
 	it("Can run at all ", async () => {
 		const dp = await fs.readJson('./test_data/datapub.json');
-		const cj = await catalog.datapub2catalog(DATASET_ID, dp, ORG);
-		console.log(JSON.stringify(cj, null, 2));
+		const cj = await catalog.datapub2catalog({
+			'@id': DATASET_ID,
+			'datapub': dp,
+			'organisation': ORG,
+			'owner': OWNER,
+			'approver': APPROVER
+		}); 
+		console.log(JSON.stringify(cj['@graph'], null, 2));
 		assert(cj, "Got something back")
 	});