fix(n8n): preserve root docs pages

Author: auscaster

nodes/Sourcey/sourcey/retrieval.ts

@@ -337,8 +337,12 @@ export function relativeOutputPath(url: string, siteUrl: string): string {
 	const sitePath = new URL(normalizeSiteUrl(siteUrl)).pathname.replace(/\/+$/, '');
 	let parsedPath = new URL(url).pathname;
 
+	if (sitePath && parsedPath === sitePath) {
+		return ROOT_OUTPUT_PATH;
+	}
 	if (sitePath && parsedPath.startsWith(`${sitePath}/`)) {
 		parsedPath = parsedPath.slice(sitePath.length + 1);
+		if (!parsedPath) return ROOT_OUTPUT_PATH;
 	} else {
 		parsedPath = parsedPath.replace(/^\/+/, '');
 	}
@@ -661,10 +665,10 @@ function normalizeOutputPath(path: string): string {
 	}
 
 	value = value.replace(/^\/+/, '').replace(/\/+$/, '');
-	if (!value) return '';
+	if (!value) return ROOT_OUTPUT_PATH;
 	if (value.endsWith('.html')) value = value.slice(0, -5);
 	else if (value.endsWith('.htm')) value = value.slice(0, -4);
-	if (value === 'index') return '';
+	if (value === ROOT_OUTPUT_PATH) return ROOT_OUTPUT_PATH;
 	if (value.endsWith('/index')) value = value.slice(0, -'/index'.length);
 	return value;
 }
@@ -723,6 +727,7 @@ const HEAD_RE = /<head\b.*?<\/head>/gis;
 const SCRIPT_STYLE_RE = /<(?:script|style)\b.*?>.*?<\/(?:script|style)>/gis;
 const TAG_RE = /<[^>]+>/g;
 const WHITESPACE_RE = /\s+/g;
+const ROOT_OUTPUT_PATH = 'index';
 const STOPWORDS = new Set([
 	'a',
 	'an',

test/sourcey-retrieval.test.mjs

@@ -94,6 +94,64 @@ Use Sourcey search-index.json and llms-full.txt to answer docs questions with ci
 		);
 	});
 
+	it('keeps root and index docs pages addressable', async () => {
+		const pages = parseLlmsFullText(`### Home
+
+Path: \`/index.html\`
+
+Root page.
+`);
+
+		assert.equal(pages.index.title, 'Home');
+
+		const httpGet = fixtureHttpGet({
+			'/reference/search-index.json': JSON.stringify([
+				{
+					title: 'Home',
+					content: 'Landing docs page',
+					url: '/reference/',
+					tab: 'Guides',
+					category: 'Pages',
+				},
+			]),
+			'/reference/llms-full.txt': `### Home
+
+Path: \`/reference/index.html\`
+
+Full landing page from llms-full.
+`,
+		});
+
+		const results = await searchSourceyDocs({
+			httpGet,
+			siteUrl: SITE_URL,
+			query: 'landing docs',
+			topK: 1,
+		});
+
+		assert.equal(results[0].path, 'index');
+		assert.equal(results[0].url, 'https://docs.example.com/reference/');
+
+		const context = await retrieveSourceyContext({
+			httpGet,
+			siteUrl: SITE_URL,
+			query: 'landing docs',
+			topK: 1,
+			maxContextChars: 1000,
+		});
+
+		assert.match(context.context, /Full landing page from llms-full/);
+
+		const fetched = await fetchSourceyPage({
+			httpGet,
+			siteUrl: SITE_URL,
+			pathOrUrl: SITE_URL,
+		});
+
+		assert.equal(fetched.path, 'index');
+		assert.equal(fetched.pageContent, 'Full landing page from llms-full.');
+	});
+
 	it('fetches a page from llms-full or falls back to HTML', async () => {
 		const fromLlms = await fetchSourceyPage({
 			httpGet: fixtureHttpGet({