feat: use levenshtein distance for summary matches name validation rule

packages/moderation/src/locales/en-US/index.json

@@ -435,7 +435,7 @@
     "defaultMessage": "Visit links settings"
   },
   "project.text-validation.summary-matches-title": {
-    "defaultMessage": "A project summary cannot be the same as it's title."
+    "defaultMessage": "A project summary cannot be the same as or too similar to its title."
   },
   "project.text-validation.summary-too-short": {
     "defaultMessage": "Your summary is too short. Add a sentence or two which describes your project."

packages/moderation/src/validation-rules/rules/summary.ts

@@ -66,7 +66,7 @@ const messages = defineMessages({
 	},
 	matchesName: {
 		id: 'project.text-validation.summary-matches-title',
-		defaultMessage: "A project summary cannot be the same as it's title.",
+		defaultMessage: 'A project summary cannot be the same as or too similar to its title.',
 	},
 	tooShort: {
 		id: 'project.text-validation.summary-too-short',
@@ -89,6 +89,7 @@ const messages = defineMessages({
 })
 
 export const MIN_SUMMARY_CHARS = 25
+export const MAX_SUMMARY_NAME_SIMILARITY = 0.8
 
 export interface ProjectSummaryValidationInput {
 	summary: string | null | undefined
@@ -105,11 +106,42 @@ function findProjectSummaryLinkOrIp(summary: string): string | null {
 	return summaryLinkify.match(summary)?.[0].raw ?? null
 }
 
-export function projectSummaryMatchesName(summary: string, name: string) {
-	const normalizedSummary = normalizeProjectFieldText(summary).replace(/\s+/g, '')
-	const normalizedName = normalizeProjectFieldText(name).replace(/\s+/g, '')
+function getLevenshteinDistance(left: string[], right: string[]) {
+	if (left.length > right.length) return getLevenshteinDistance(right, left)
 
-	return normalizedSummary.length > 0 && normalizedSummary === normalizedName
+	let previousRow = Array.from({ length: left.length + 1 }, (_, index) => index)
+
+	for (let rightIndex = 0; rightIndex < right.length; rightIndex++) {
+		const currentRow = [rightIndex + 1]
+
+		for (let leftIndex = 0; leftIndex < left.length; leftIndex++) {
+			currentRow.push(
+				Math.min(
+					currentRow[leftIndex] + 1,
+					previousRow[leftIndex + 1] + 1,
+					previousRow[leftIndex] + (left[leftIndex] === right[rightIndex] ? 0 : 1),
+				),
+			)
+		}
+
+		previousRow = currentRow
+	}
+
+	return previousRow[left.length]
+}
+
+function normalizeProjectFieldTextForSimilarity(value: string) {
+	return Array.from(normalizeProjectFieldText(value).toLocaleLowerCase('en-US').replace(/\s+/g, ''))
+}
+
+export function getProjectSummaryNameSimilarity(summary: string, name: string) {
+	const normalizedSummary = normalizeProjectFieldTextForSimilarity(summary)
+	const normalizedName = normalizeProjectFieldTextForSimilarity(name)
+	const longestLength = Math.max(normalizedSummary.length, normalizedName.length)
+
+	if (longestLength === 0) return 0
+
+	return 1 - getLevenshteinDistance(normalizedSummary, normalizedName) / longestLength
 }
 
 export function hasProjectSummaryFormatting(summary: string) {
@@ -172,7 +204,7 @@ export const projectSummaryValidationRules = {
 				!summary ||
 				findProjectSummaryLinkOrIp(summary) !== null ||
 				!name ||
-				!projectSummaryMatchesName(summary, name),
+				getProjectSummaryNameSimilarity(summary, name) < MAX_SUMMARY_NAME_SIMILARITY,
 		}),
 		presentation: {
 			message: messages.matchesName,

packages/moderation/src/validation-rules/tests.ts

@@ -21,8 +21,8 @@ import { validateProjectDisclosures } from './rules/disclosures.ts'
 import { validateProjectGalleryDescription, validateProjectGalleryName } from './rules/gallery.ts'
 import { projectNameValidationRules, validateProjectNameField } from './rules/name.ts'
 import {
+	getProjectSummaryNameSimilarity,
 	hasProjectSummaryFormatting,
-	projectSummaryMatchesName,
 	validateProjectSummary,
 } from './rules/summary.ts'
 import { projectVersionValidationRules } from './rules/versions.ts'
@@ -166,7 +166,7 @@ test('requires known environments for mods and modpacks', () => {
 })
 
 test('validates summary content from one rule set', () => {
-	assert.equal(projectSummaryMatchesName('  Caf\u00e9  ', 'Cafe\u0301'), true)
+	assert.equal(getProjectSummaryNameSimilarity('  Caf\u00e9  ', 'Cafe\u0301'), 1)
 	assert.deepEqual(
 		validateProjectSummary({ summary: '# Short summary', name: 'Project title' }).map(
 			({ code, severity }) => ({ code, severity }),
@@ -185,6 +185,26 @@ test('validates summary content from one rule set', () => {
 	)
 })
 
+test('rejects project summaries that are too similar to the project name', () => {
+	for (const summary of ['My Project', 'myproject', 'My Projec', 'My Projects']) {
+		assert.equal(
+			validateProjectSummary({ summary, name: 'My Project' }).some(
+				({ code }) => code === 'project-summary-matches-title',
+			),
+			true,
+			summary,
+		)
+	}
+
+	assert.equal(
+		validateProjectSummary({
+			summary: 'A detailed summary of what My Project provides',
+			name: 'My Project',
+		}).some(({ code }) => code === 'project-summary-matches-title'),
+		false,
+	)
+})
+
 test('warns when a project summary is mostly non-English', () => {
 	assert.deepEqual(
 		validateProjectSummary({