Janady13/thatstupidcomputer / tests/site/search.test.ts

1import test from 'node:test'; 2import assert from 'node:assert/strict'; 3import { ROUTES, getPage } from '../../packages/site/content.ts'; 4import { buildGraph, metadataForPage, renderHead, robotsTxt, safeJson, sitemapXml } from '../../packages/site/search.ts'; 5import { configFromProjection, loadSiteConfig, validateProjection } from '../../packages/site/truth.ts'; 6import { projectionFixture, TEST_CRAWLER_POLICY } from './fixture.ts'; 7 8test('all indexable pages self-canonicalize and have correct stable graph references', () => { 9 const config = configFromProjection(validateProjection(projectionFixture()), 'production', TEST_CRAWLER_POLICY); 10 for (const page of ROUTES) { 11 const canonical = `${config.canonicalOrigin}${page.path}`; 12 const meta = metadataForPage(page, config); 13 assert.equal(meta.canonical, canonical); 14 assert.match(meta.robots, /^index/); 15 const graph = buildGraph(page, config)!['@graph']; 16 const internal = new Set(graph.map((node) => node['@id'])); 17 const external = new Set([config.publisher?.id, config.provider?.id, config.brand?.id]); 18 function checkRefs(value: unknown): void { 19 if (!value || typeof value !== 'object') return; 20 const record = value as Record<string, unknown>; 21 if (typeof record['@id'] === 'string') assert.ok(internal.has(record['@id']) || external.has(record['@id']), `${page.path}: ${record['@id']}`); 22 Object.values(record).forEach(checkRefs); 23 } 24 checkRefs(graph); 25 assert.equal(internal.size, graph.length); 26 assert.ok(graph.every((node) => !['Person', 'Organization', 'Brand'].includes(node['@type']))); 27 const service = graph.find((node) => node['@type'] === 'Service'); 28 if (page.service) { assert.equal(service?.description, page.intro); assert.deepEqual(service?.provider, { '@id': config.provider!.id }); } 29 const head = renderHead(page, config); 30 assert.equal((head.match(/<title>/g) ?? []).length, 1); 31 assert.equal((head.match(/rel="canonical"/g) ?? []).length, 1); 32 assert.doesNotMatch(head, /sameAs|AggregateRating|LocalBusiness|FAQPage|SearchAction/); 33 } 34}); 35test('preview remains noindex and unknown pages never masquerade as the homepage', () => { 36 const config = loadSiteConfig({ deployment: 'preview' }); 37 assert.match(metadataForPage(getPage('/'), config).robots, /^noindex/); 38 assert.equal(metadataForPage(undefined, config).canonical, undefined); 39 assert.equal(buildGraph(undefined, config), undefined); 40 assert.match(renderHead(undefined, config), /Page not found/); 41 assert.doesNotMatch(renderHead(undefined, config), /rel="canonical"/); 42 for (const path of ['/about//', '/services///', '/%61bout/']) { 43 const unknown = getPage(path); 44 assert.equal(unknown, undefined); 45 assert.doesNotMatch(renderHead(unknown, config), /rel="canonical"|site-structured-data/); 46 assert.match(renderHead(unknown, config), /Page not found/); 47 } 48 assert.doesNotMatch(sitemapXml(config), /<loc>/); 49 assert.doesNotMatch(robotsTxt(config), /Sitemap:/); 50}); 51test('production sitemap includes only the canonical route inventory and no invented freshness', () => { 52 const config = configFromProjection(validateProjection(projectionFixture()), 'production', TEST_CRAWLER_POLICY); 53 const xml = sitemapXml(config); 54 assert.equal((xml.match(/<loc>/g) ?? []).length, ROUTES.length); 55 for (const page of ROUTES) assert.ok(xml.includes(`<loc>${config.canonicalOrigin}${page.path}</loc>`)); 56 assert.doesNotMatch(xml, /lastmod|changefreq|priority|private|api/); 57 const robots = robotsTxt(config); 58 assert.match(robots, /User-agent: OAI-SearchBot\nAllow: \/\nDisallow: \/api\//); 59 assert.match(robots, /User-agent: GPTBot\nDisallow: \//); 60 assert.match(robots, /User-agent: Google-Extended\nDisallow: \//); 61}); 62test('metadata and JSON-LD embedding escape HTML boundaries', () => { 63 const config = loadSiteConfig({ deployment: 'preview' }); 64 const page = { ...ROUTES[0]!, title: 'A & B </title><script>alert(1)</script>', description: '" < > &' }; 65 const head = renderHead(page, config); 66 assert.doesNotMatch(head, /<script>alert/); 67 assert.match(head, /&amp;/); 68 const input = { text: '</script><script>alert(1)</script>\u2028&' }; 69 const output = safeJson(input); 70 assert.doesNotMatch(output, /<|>|&/); 71 assert.deepEqual(JSON.parse(output), input); 72}); 73test('configured model-use decisions remain independent and preserve utility exclusions when allowed', () => { 74 const config = configFromProjection(validateProjection(projectionFixture()), 'production', { training: 'deny', googleExtended: 'allow', source: 'deployment-config' }); 75 let robots = robotsTxt(config); 76 assert.match(robots, /User-agent: GPTBot\nDisallow: \//); 77 assert.match(robots, /User-agent: ClaudeBot\nDisallow: \//); 78 assert.match(robots, /User-agent: Google-Extended\nAllow: \/\nDisallow: \/api\//); 79 config.crawlerPolicy = { training: 'allow', googleExtended: 'deny', source: 'deployment-config' }; 80 robots = robotsTxt(config); 81 assert.match(robots, /User-agent: GPTBot\nAllow: \/\nDisallow: \/api\//); 82 assert.match(robots, /User-agent: ClaudeBot\nAllow: \/\nDisallow: \/api\//); 83 assert.match(robots, /User-agent: Google-Extended\nDisallow: \//); 84 config.crawlerPolicy.source = 'preview-default'; 85 assert.throws(() => robotsTxt(config), /explicitly configured/); 86}); 87test('ownership metadata is emitted only for actually configured public tokens', () => { 88 const configured = loadSiteConfig({ deployment: 'preview', googleSiteVerification: 'google_safe-token-123', bingSiteVerification: 'ABCDEF1234567890' }); 89 const head = renderHead(getPage('/'), configured); 90 assert.match(head, /name="google-site-verification" content="google_safe-token-123"/); 91 assert.match(head, /name="msvalidate.01" content="ABCDEF1234567890"/); 92 const noVerification = { ...configured, verification: undefined }; 93 assert.doesNotMatch(renderHead(getPage('/'), noVerification), /google-site-verification|msvalidate.01/); 94}); 95 96test('OpenAI and Anthropic training choices can differ without changing discovery or Google grounding', () => { 97 const config = loadSiteConfig({ deployment: 'preview', aiTrainingPolicy: 'deny', googleExtendedPolicy: 'deny', openaiTrainingPolicy: 'allow', anthropicTrainingPolicy: 'deny' }); 98 const robots = robotsTxt(config); 99 assert.match(robots, /User-agent: GPTBot\nAllow: \/\nDisallow: \/api\//); 100 assert.match(robots, /User-agent: ClaudeBot\nDisallow: \//); 101 assert.match(robots, /User-agent: Google-Extended\nDisallow: \//); 102 for (const agent of ['OAI-SearchBot', 'Claude-SearchBot', 'PerplexityBot', 'ChatGPT-User', 'Claude-User', 'Perplexity-User']) { 103 assert.ok(robots.includes(`User-agent: ${agent}\nAllow: /\nDisallow: /api/`)); 104 } 105 const opposite = robotsTxt(loadSiteConfig({ deployment: 'preview', aiTrainingPolicy: 'allow', googleExtendedPolicy: 'allow', openaiTrainingPolicy: 'deny', anthropicTrainingPolicy: 'allow' })); 106 assert.match(opposite, /User-agent: GPTBot\nDisallow: \//); 107 assert.match(opposite, /User-agent: ClaudeBot\nAllow: \/\nDisallow: \/api\//); 108}); 109 110test('invalid provider overrides fail closed even when the legacy policy is valid', () => { 111 assert.throws(() => loadSiteConfig({ deployment: 'preview', aiTrainingPolicy: 'allow', openaiTrainingPolicy: 'sometimes' }), /OPENAI_TRAINING_POLICY/); 112 assert.throws(() => loadSiteConfig({ deployment: 'preview', aiTrainingPolicy: 'allow', anthropicTrainingPolicy: 'sometimes' }), /ANTHROPIC_TRAINING_POLICY/); 113}); 114