Merge pull request #153 from guardian/pf/rename-iptccat-to-ap

Rename cat codes labelled 'iptccat' as 'apCat' instead

ingestion-lambda/src/categoryCodes.test.ts

@@ -9,16 +9,16 @@ describe('processFingerpostAPCategoryCodes', () => {
 		expect(processFingerpostAPCategoryCodes(['service:news'])).toEqual([]);
 	});
 
-	it('should return simple category codes as they were received', () => {
+	it('should return simple codes labelled "iptccat" as simple "apCat" codes', () => {
 		expect(
 			processFingerpostAPCategoryCodes(['iptccat:a', 'iptccat:b']),
-		).toEqual(['iptccat:a', 'iptccat:b']);
+		).toEqual(['apCat:a', 'apCat:b']);
 	});
 
 	it('should expand category codes with multiple subcodes', () => {
 		expect(processFingerpostAPCategoryCodes(['iptccat:c+d'])).toEqual([
-			'iptccat:c',
-			'iptccat:d',
+			'apCat:c',
+			'apCat:d',
 		]);
 	});
 
@@ -31,7 +31,7 @@ describe('processFingerpostAPCategoryCodes', () => {
 	it('should remove empty strings', () => {
 		expect(
 			processFingerpostAPCategoryCodes(['iptccat:a', '', 'iptccat:c']),
-		).toEqual(['iptccat:a', 'iptccat:c']);
+		).toEqual(['apCat:a', 'apCat:c']);
 	});
 
 	it('should remove trailing and leading whitespace', () => {
@@ -42,7 +42,7 @@ describe('processFingerpostAPCategoryCodes', () => {
 				' service:news ',
 				'qCode:value ',
 			]),
-		).toEqual(['iptccat:a', 'iptccat:c', 'qCode:value']);
+		).toEqual(['apCat:a', 'apCat:c', 'qCode:value']);
 	});
 
 	it('should deduplicate category codes after stripping whitespace', () => {
@@ -52,6 +52,6 @@ describe('processFingerpostAPCategoryCodes', () => {
 				' iptccat:a',
 				'iptccat:c',
 			]),
-		).toEqual(['iptccat:a', 'iptccat:c']);
+		).toEqual(['apCat:a', 'apCat:c']);
 	});
 });

ingestion-lambda/src/categoryCodes.ts

@@ -14,21 +14,26 @@ function partition<T>(
 	return [first, second];
 }
 
+/**
+ * We receive AP codes from Fingerpost in the format `prefix:code1+code2+code3:code4+code5`.
+ * At the time of writing these are AP category codes, but mislabelled as `iptccat` codes.
+ * This function transforms the prefix, and splits the codes into individual category codes.
+ */
 function flattenCategoryCodes(categoryCodes: string): string[] {
 	const [prefix, ...codes] = categoryCodes.split(':');
 	return codes
 		.flatMap((_) => _.split('+'))
-		.flatMap((code) => `${prefix}:${code}`);
+		.map((code) => `${prefix?.trim() === 'iptccat' ? 'apCat' : prefix}:${code}`);
 }
 
 export function processFingerpostAPCategoryCodes(original: string[]): string[] {
-	const remainingNotServiceCodes = original.filter((_) => !_.includes('service:'));
-	const [iptccatCodes, rest] = partition(remainingNotServiceCodes, (code) =>
+	const notServiceCodes = original.filter((_) => !_.includes('service:')); // we aren't interested in keeping the service codes here
+	const [categoryCodes, rest] = partition(notServiceCodes, (code) =>
 		code.includes('iptccat:'),
 	);
-	const transformedIptccatCodes = iptccatCodes.flatMap(flattenCategoryCodes);
+	const transformedCategoryCodes = categoryCodes.flatMap(flattenCategoryCodes);
 
-	const allCategoryCodes = [...transformedIptccatCodes, ...rest]
+	const allCategoryCodes = [...transformedCategoryCodes, ...rest]
 		.map((_) => _.trim())
 		.filter((_) => _.length > 0);
 	const deduped = [...new Set(allCategoryCodes)];