|
| 1 | +// Generates sitemap.xml and llms.txt from the routes on disk + src/site-meta.js. |
| 2 | +// |
| 3 | +// Routes are discovered from app/**/page.{jsx,tsx,js,ts} (the same files the app |
| 4 | +// router globs at runtime), so the sitemap is always complete. llms.txt pairs |
| 5 | +// each route with its curated description from site-meta; a route with no |
| 6 | +// description is reported (not silently dropped). |
| 7 | +// |
| 8 | +// Used by vite-discoverability.mjs at build/dev start, and runnable directly: |
| 9 | +// node scripts/discoverability.mjs # writes into public/ |
| 10 | +import { globSync, mkdirSync, readFileSync, writeFileSync } from "node:fs"; |
| 11 | +import path from "node:path"; |
| 12 | +import { fileURLToPath } from "node:url"; |
| 13 | +import { |
| 14 | + SITE_URL, |
| 15 | + SITE_NAME, |
| 16 | + TAGLINE, |
| 17 | + SUMMARY, |
| 18 | + SECTION_ORDER, |
| 19 | + PAGES, |
| 20 | + FOOTER, |
| 21 | + FULL_DOC, |
| 22 | + SITEMAP_EXTRA, |
| 23 | + SITEMAP_EXCLUDE, |
| 24 | +} from "../src/site-meta.js"; |
| 25 | + |
| 26 | +const APP_DIR = fileURLToPath(new URL("../app", import.meta.url)); |
| 27 | +const PUBLIC_DIR = fileURLToPath(new URL("../public", import.meta.url)); |
| 28 | +const REPO_ROOT = fileURLToPath(new URL("../../", import.meta.url)); |
| 29 | + |
| 30 | +// All routes, derived from page files: app/docs/glob/page.jsx -> /docs/glob, |
| 31 | +// app/page.jsx -> /. |
| 32 | +export function discoverRoutes(appDir = APP_DIR) { |
| 33 | + const files = globSync("**/page.*", { cwd: appDir }).filter((f) => |
| 34 | + /(^|\/)page\.(jsx|tsx|js|ts)$/.test(f), |
| 35 | + ); |
| 36 | + const routes = files.map((f) => { |
| 37 | + const trimmed = f.replace(/(^|\/)page\.(jsx|tsx|js|ts)$/, ""); |
| 38 | + return trimmed === "" ? "/" : "/" + trimmed; |
| 39 | + }); |
| 40 | + return [...new Set(routes)].filter((r) => !SITEMAP_EXCLUDE.has(r)).sort(); |
| 41 | +} |
| 42 | + |
| 43 | +function priorityFor(route) { |
| 44 | + if (route === "/") return "1.0"; |
| 45 | + const depth = route.split("/").filter(Boolean).length; |
| 46 | + return depth <= 1 ? "0.9" : depth === 2 ? "0.7" : "0.6"; |
| 47 | +} |
| 48 | + |
| 49 | +export function renderSitemap(routes) { |
| 50 | + const entries = [ |
| 51 | + { path: "/", priority: "1.0", weekly: true }, |
| 52 | + ...routes |
| 53 | + .filter((r) => r !== "/") |
| 54 | + .map((r) => ({ path: r, priority: priorityFor(r), weekly: /^\/(docs|api)$/.test(r) })), |
| 55 | + ...SITEMAP_EXTRA, |
| 56 | + ]; |
| 57 | + const urls = entries.map((e) => { |
| 58 | + const loc = `${SITE_URL}${e.path === "/" ? "/" : e.path}`; |
| 59 | + const cf = e.weekly ? " <changefreq>weekly</changefreq>\n" : ""; |
| 60 | + return ` <url>\n <loc>${loc}</loc>\n${cf} <priority>${e.priority || "0.6"}</priority>\n </url>`; |
| 61 | + }); |
| 62 | + return `<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n${urls.join("\n")}\n</urlset>\n`; |
| 63 | +} |
| 64 | + |
| 65 | +export function renderLlmsTxt(routes) { |
| 66 | + const present = new Set(routes); |
| 67 | + // A discovered route with no metadata, and metadata for a route that's gone. |
| 68 | + const missing = routes.filter((r) => r !== "/" && !PAGES[r]); |
| 69 | + const stale = Object.keys(PAGES).filter((r) => !present.has(r)); |
| 70 | + |
| 71 | + let out = `# ${SITE_NAME}\n\n> ${TAGLINE}\n\n${SUMMARY}\n`; |
| 72 | + for (const section of SECTION_ORDER) { |
| 73 | + // PAGES insertion order = curated reading order; include only live routes. |
| 74 | + const routesInSection = Object.keys(PAGES).filter( |
| 75 | + (r) => PAGES[r].section === section && present.has(r), |
| 76 | + ); |
| 77 | + if (routesInSection.length === 0) continue; |
| 78 | + out += `\n## ${section}\n\n`; |
| 79 | + for (const r of routesInSection) { |
| 80 | + const { title, description } = PAGES[r]; |
| 81 | + out += `- [${title}](${SITE_URL}${r}): ${description}\n`; |
| 82 | + } |
| 83 | + } |
| 84 | + for (const f of FOOTER) out += `\n## ${f.section}\n\n${f.body}\n`; |
| 85 | + return { content: out, missing, stale }; |
| 86 | +} |
| 87 | + |
| 88 | +// Inlines the canonical repo docs (README + docs/API.md) as one markdown file. |
| 89 | +export function renderLlmsFull(repoRoot = REPO_ROOT) { |
| 90 | + const header = `# ${SITE_NAME} — full documentation\n\n> ${FULL_DOC.tagline}\n\n_${FULL_DOC.note}_\n\n---\n`; |
| 91 | + const parts = FULL_DOC.sources.map((rel) => |
| 92 | + readFileSync(path.join(repoRoot, rel), "utf8").trim(), |
| 93 | + ); |
| 94 | + return `${header}\n${parts.join("\n\n---\n\n")}\n`; |
| 95 | +} |
| 96 | + |
| 97 | +// Builds all three files; returns their contents plus any warnings. |
| 98 | +export function buildDiscoverability(appDir = APP_DIR) { |
| 99 | + const routes = discoverRoutes(appDir); |
| 100 | + const sitemap = renderSitemap(routes); |
| 101 | + const { content: llms, missing, stale } = renderLlmsTxt(routes); |
| 102 | + return { |
| 103 | + routes, |
| 104 | + files: { |
| 105 | + "sitemap.xml": sitemap, |
| 106 | + "llms.txt": llms, |
| 107 | + "llms-full.txt": renderLlmsFull(), |
| 108 | + }, |
| 109 | + missing, |
| 110 | + stale, |
| 111 | + }; |
| 112 | +} |
| 113 | + |
| 114 | +// Run directly: write into public/ and report. |
| 115 | +if (process.argv[1] && path.resolve(process.argv[1]) === fileURLToPath(import.meta.url)) { |
| 116 | + const { routes, files, missing, stale } = buildDiscoverability(); |
| 117 | + mkdirSync(PUBLIC_DIR, { recursive: true }); |
| 118 | + for (const [name, content] of Object.entries(files)) { |
| 119 | + writeFileSync(path.join(PUBLIC_DIR, name), content); |
| 120 | + } |
| 121 | + if (missing.length) console.warn(`⚠ no llms.txt metadata (add to src/site-meta.js): ${missing.join(", ")}`); |
| 122 | + if (stale.length) console.warn(`⚠ site-meta.js entries for missing routes: ${stale.join(", ")}`); |
| 123 | + console.log(`Wrote ${Object.keys(files).join(", ")} to public/ (${routes.length} routes).`); |
| 124 | +} |
0 commit comments