Improve moderation controls and term management

src/lib/server/community-moderation.test.ts

@@ -3,6 +3,7 @@ import {
 	assessCommunityModeration,
 	findLocalModerationMatches,
 	normalizeModerationTerm,
+	parseModerationTermBatch,
 	type CommunityModerationPolicy
 } from './community-moderation';
 
@@ -34,6 +35,27 @@ describe('community moderation policy', () => {
 		expect(normalizeModerationTerm('  Rogue\u00a0Agent  ')).toBe('rogue agent');
 	});
 
+	it('parses pasted terms with automatic word and phrase matching', () => {
+		expect(parseModerationTermBatch('weapon\nkill, machine gun; assault rifle')).toEqual({
+			terms: [
+				{ value: 'weapon', normalizedValue: 'weapon', matchMode: 'word' },
+				{ value: 'kill', normalizedValue: 'kill', matchMode: 'word' },
+				{ value: 'machine gun', normalizedValue: 'machine gun', matchMode: 'phrase' },
+				{ value: 'assault rifle', normalizedValue: 'assault rifle', matchMode: 'phrase' }
+			],
+			duplicateCount: 0
+		});
+	});
+
+	it('deduplicates pasted terms after compatibility and case normalization', () => {
+		const parsed = parseModerationTermBatch('Weapon\nｗｅａｐｏｎ\nMACHINE   GUN\nmachine gun');
+		expect(parsed.terms).toEqual([
+			{ value: 'Weapon', normalizedValue: 'weapon', matchMode: 'word' },
+			{ value: 'MACHINE GUN', normalizedValue: 'machine gun', matchMode: 'phrase' }
+		]);
+		expect(parsed.duplicateCount).toBe(2);
+	});
+
 	it('allows an exact local exception to suppress a matching banned term', () => {
 		const matches = findLocalModerationMatches('Rogue Agent tactics', 'comment', policy({
 			detectPii: false,

src/lib/server/community-moderation.ts

@@ -107,6 +107,37 @@ export function normalizeModerationTerm(value: string) {
 	return value.normalize('NFKC').trim().replace(/\s+/g, ' ').toLocaleLowerCase();
 }
 
+export type ParsedModerationTermBatch = {
+	terms: Array<{
+		value: string;
+		normalizedValue: string;
+		matchMode: 'word' | 'phrase';
+	}>;
+	duplicateCount: number;
+};
+
+export function parseModerationTermBatch(value: string): ParsedModerationTermBatch {
+	const entries = value.normalize('NFKC')
+		.split(/[\r\n,;]+/u)
+		.map((entry) => entry.trim().replace(/\s+/g, ' '))
+		.filter(Boolean);
+	const terms: ParsedModerationTermBatch['terms'] = [];
+	const seen = new Set<string>();
+	let duplicateCount = 0;
+	for (const entry of entries) {
+		const normalizedValue = normalizeModerationTerm(entry);
+		const matchMode = /\s/u.test(normalizedValue) ? 'phrase' : 'word';
+		const key = `${matchMode}:${normalizedValue}`;
+		if (seen.has(key)) {
+			duplicateCount += 1;
+			continue;
+		}
+		seen.add(key);
+		terms.push({ value: entry, normalizedValue, matchMode });
+	}
+	return { terms, duplicateCount };
+}
+
 export function hashModeratedContent(value: string) {
 	return createHash('sha256').update(value.normalize('NFC')).digest('hex');
 }