Randomised DataCrates are now working. Fixed a bug in catalog.js - it wasn't

linking creators to the top-level dataset item.

lib/catalog.js

@@ -146,7 +146,7 @@ async function datapub2catalog(options) {
 	var people = make_creators(datapub, organisation);
 
 	const files = make_files(datapub);
-	const dataset = make_dataset(id, datapub, organisation, dates, files);
+	const dataset = make_dataset(id, datapub, organisation, dates, files, people);
 	const [ history, more_people ] = make_history(
 		dataset, people, owner, approver, dates
 	);
@@ -227,7 +227,7 @@ function link_id(item) {
 
 // TODO: isBasedOn the data record
 
-function make_dataset(id, datapub, organisation, dates, files) {
+function make_dataset(id, datapub, organisation, dates, files, creators) {
 	const ds = {
 		'@id': id,
 		'@type': 'Dataset',
@@ -238,6 +238,7 @@ function make_dataset(id, datapub, organisation, dates, files) {
 		'datePublished': dates['datePublished'],
 		'publisher': link_id(organisation),
 		'hasPart': files.map(link_id),
+		'creator': creators.map(link_id)
 	};
 	// see https://schema.org/keywords
 	if( datapub['finalKeywords'] ) {
@@ -344,6 +345,9 @@ function make_files(datapub) {
 // returns the array of Person items from the creators
 // field of the data publication
 
+// Doesn't add the creators to the datapub - the calling
+// code is responsible for that.
+
 // TODO: affiliations
 
 function make_creators(datapub, organisation) {
@@ -383,7 +387,11 @@ function make_creators(datapub, organisation) {
 function make_related_works(datapub) {
 	return RELATED.map((thing) => {
 		const field = 'related_' + thing;
-		return datapub[field].filter((f) => f['related_url']).map((item) => {
+		var values = [];
+		if( field in datapub ) {
+			values = datapub[field];
+		}
+		return values.filter((f) => f['related_url']).map((item) => {
 			const ld_item = {
 				'@id': item['related_url'],
 				'@type': RELATED_TYPES[thing],
@@ -529,7 +537,7 @@ function make_temporal(datapub) {
 function make_license(datapub, prefix) {
 	const license = [];
 	if( datapub['license_other_url'] || datapub['license_notes'] ) {
-		if( datapub['license_other_url'] );
+		if( datapub['license_other_url'] ) {
 			licenses.push({
 				'@id': datapub['license_other_url'],
 				'@type': 'CreativeWork',
@@ -550,7 +558,7 @@ function make_license(datapub, prefix) {
 			'@type': 'CreativeWork',
 			'name': datapub['license_identifier'],
 			'url': datapub['license_identifier']
-		}];
+		});
 	}
 	return [];
 }

lib/random.js

 // Module to generate a bunch of CATALOG.json files which 
 // have arbitrary but realistic data
 
+const ORGANISATION = {
+	'id': 'https://examples.edu',
+	'name': 'Examples University'
+};
+
+const EMAIL_DOMAIN = 'examples.edu';
+const OWNER = 'owner@examples.edu';
+const APPROVER = 'approver@examples.edu';
+
+const NAME_MIN = 3;
+const NAME_MAX = 10;
+const KEYWORD_MIN = 3;
+const KEYWORD_MAX = 12;
+const WORD_MIN = 2;
+const WORD_MAX = 14;
+const SENTENCE_MIN = 3;
+const SENTENCE_MAX = 30;
+const PARA_MIN = 1;
+const PARA_MAX = 10;
+
+const N_KEYWORD_MIN = 2;
+const N_KEYWORD_MAX = 10;
+
+const N_PEOPLE_MIN = 1;
+const N_PEOPLE_MAX = 5;
+
+const HONORIFICS = [ 'Dr', 'A/Prof', 'Prof', 'Dr', 'Dr', 'Dr', 'Mr', 'Ms' ];
+
+const datacrate = require('./catalog.js');
+const _ = require('lodash');
+const fs = require('fs-extra');
+const randdict = require('random-word');
+const path = require('path');
+const uuidv4 = require('uuid/v4');
+
+
+function randrange(min, max) {
+	return Math.floor(Math.random() * (max - min)) + min;
+}
+
+function randoms(n, fn) {
+	return Array(n).fill(0).map(fn);
+}
+
+function randname() {
+	return _.upperFirst(randdict());
+}
+
+function randperson() {
+	const honorific = _.sample(HONORIFICS);
+	const surname = randname();
+	const givenname = randname();
+	const name = givenname + ' ' + surname;
+	const email = givenname + '.' + surname + '@' + EMAIL_DOMAIN;
+	const id = uuidv4();
+	return {
+		'dc:identifier': id,
+		'text_full_name': name,
+		'full_name_honorific': honorific + ' ' + name,
+		'email': email
+	}
+}
+
+function randkeyword() {
+	return randdict();
+}
+
+function randsentence() {
+	const nwords = randrange(SENTENCE_MIN, SENTENCE_MAX);
+	const s = randoms(nwords, randdict).join(' ') + '.';
+	return _.upperFirst(s);
+}
+
+function randtext() {
+	const nsentences = randrange(PARA_MIN, PARA_MAX);
+	console.log(nsentences);
+	return randoms(nsentences, randsentence).join(' ') + '\n';
+}
+
+function randdatapub(keywords, people) {
+	const k = _.sampleSize(keywords, randrange(N_KEYWORD_MIN, N_KEYWORD_MAX));
+	const title = _.startCase(_.camelCase(randsentence()));
+	const desc = randtext();
+	const creators = _.clone(_.sampleSize(people, randrange(N_PEOPLE_MIN, N_PEOPLE_MAX)));
+	const contributors = _.clone(creators);
+	const collabs = contributors.splice(1);
+	contributors[0].role = 'Chief Investigator';
+	return {
+		keywords: k,
+		contributor_ci: contributors[0],
+		contributors: collabs,
+		creators: creators,
+		title: title,
+		description: desc
+	}
+}
+
+
+function randdatapubs(n) {
+	const keywords = randoms(Math.floor(n / 2), randkeyword);
+	const people = randoms(n * 2, randperson);
+	return randoms(n, () => randdatapub(keywords, people))
+}
+
+async function makedatacrate(dest, datapub) {
+	const id = uuidv4();
+	await fs.ensureDir(path.join(dest, id));
+	const catalog = await datacrate.datapub2catalog({
+		id: id,
+		datapub: datapub,
+		organisation: ORGANISATION,
+		owner: OWNER,
+		approver: APPROVER
+	});
+	const catfile = path.join(dest, id, 'CATALOG.json');
+	await fs.writeFile(catfile, JSON.stringify(catalog, null, 2));
+	console.log('Wrote ' + catfile);
+	const datapubfile = path.join(dest, id, 'datapub.json');
+	await fs.writeFile(datapubfile, JSON.stringify(datapub, null, 2));
+	console.log('Wrote ' + datapubfile);
+}
+
+
+async function makedatacrates(dest, n) {
+	const datapubs = randdatapubs(n);
+	Promise.all(
+		datapubs.map(async p => {
+			await makedatacrate(dest, p);
+		})
+	);
+	console.log("Done");
+}
+
+makedatacrates('./output/', 5)
+
+
+

package-lock.json

 {
   "name": "datacrate",
-  "version": "1.0.3",
+  "version": "1.0.4",
   "lockfileVersion": 1,
   "requires": true,
   "dependencies": {
@@ -1235,6 +1235,15 @@
       "resolved": "https://registry.npmjs.org/qs/-/qs-6.5.2.tgz",
       "integrity": "sha512-N5ZAX4/LxJmF+7wN74pUD6qAh9/wnvdQcjq9TZjevvXzSUo7bfmw91saqMjzGS2xq91/odN2dW/WOl7qQHNDGA=="
     },
+    "random-word": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/random-word/-/random-word-2.0.0.tgz",
+      "integrity": "sha1-Je0Y1bpamUCerOtZ2r5lt20v9uk=",
+      "requires": {
+        "unique-random-array": "^1.0.0",
+        "word-list": "^2.0.0"
+      }
+    },
     "rdf-canonize": {
       "version": "0.2.4",
       "resolved": "https://registry.npmjs.org/rdf-canonize/-/rdf-canonize-0.2.4.tgz",
@@ -1527,6 +1536,19 @@
       "integrity": "sha512-0fr/mIH1dlO+x7TlcMy+bIDqKPsw/70tVyeHW787goQjhmqaZe10uwLujubK9q9Lg6Fiho1KUKDYz0Z7k7g5/g==",
       "dev": true
     },
+    "unique-random": {
+      "version": "1.0.0",
+      "resolved": "https://registry.npmjs.org/unique-random/-/unique-random-1.0.0.tgz",
+      "integrity": "sha1-zj4iTIJCzTOg53sNcYDXfmti0MQ="
+    },
+    "unique-random-array": {
+      "version": "1.0.1",
+      "resolved": "https://registry.npmjs.org/unique-random-array/-/unique-random-array-1.0.1.tgz",
+      "integrity": "sha512-z9J/SV8CUIhIRROcHe9YUoAT6XthUJt0oUyLGgobiXJprDP9O9dsErNevvSaAv5BkhwFEVPn6nIEOKeNE6Ck1Q==",
+      "requires": {
+        "unique-random": "^1.0.0"
+      }
+    },
     "universalify": {
       "version": "0.1.2",
       "resolved": "https://registry.npmjs.org/universalify/-/universalify-0.1.2.tgz",
@@ -1616,6 +1638,11 @@
       "resolved": "https://registry.npmjs.org/which-module/-/which-module-2.0.0.tgz",
       "integrity": "sha1-2e8H3Od7mQK4o6j6SzHD4/fm6Ho="
     },
+    "word-list": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/word-list/-/word-list-2.0.0.tgz",
+      "integrity": "sha1-VN36Sq4fqnF3LPtukgjE2PmZNj4="
+    },
     "wordwrap": {
       "version": "1.0.0",
       "resolved": "https://registry.npmjs.org/wordwrap/-/wordwrap-1.0.0.tgz",

package.json

@@ -14,10 +14,12 @@
   "license": "GPL-3.0",
   "repository": "https://code.research.uts.edu.au/eresearch/datacrate",
   "dependencies": {
+    "calcyte": "^1.0.2",
     "fs-extra": "^7.0.0",
     "jsonld": "^1.1.0",
     "lodash": "^4.17.11",
-    "calcyte": "^1.0.2"
+    "random-word": "^2.0.0",
+    "uuid": "^3.3.2"
   },
   "devDependencies": {
     "chai": "^4.1.2",