|
| 1 | +/** |
| 2 | + * Post-build crawl: every internal page href in docs/.vitepress/dist must |
| 3 | + * resolve to a rendered HTML file (or index.html). External URLs are ignored. |
| 4 | + * |
| 5 | + * Run after docs:build (wired into docs:verify). |
| 6 | + */ |
| 7 | +import { existsSync } from 'node:fs' |
| 8 | +import { readdir, readFile } from 'node:fs/promises' |
| 9 | +import { dirname, relative, resolve } from 'node:path' |
| 10 | +import { fileURLToPath } from 'node:url' |
| 11 | + |
| 12 | +const __dirname = dirname(fileURLToPath(import.meta.url)) |
| 13 | +const distDir = resolve(__dirname, '../docs/.vitepress/dist') |
| 14 | +const SITE_HOSTS = new Set(['openaidoc.org', 'www.openaidoc.org', 'localhost']) |
| 15 | + |
| 16 | +if (!existsSync(distDir)) { |
| 17 | + console.error( |
| 18 | + `verify-internal-links: dist not found at ${distDir}. Run docs:build first.` |
| 19 | + ) |
| 20 | + process.exit(1) |
| 21 | +} |
| 22 | + |
| 23 | +const htmlFiles = await listFiles(distDir, '.html') |
| 24 | +/** @type {Map<string, Set<string>>} */ |
| 25 | +const deadPages = new Map() |
| 26 | +let internalRefs = 0 |
| 27 | + |
| 28 | +for (const file of htmlFiles) { |
| 29 | + const rel = toPosix(relative(distDir, file)) |
| 30 | + if (rel === '404.html') continue |
| 31 | + |
| 32 | + const html = await readFile(file, 'utf8') |
| 33 | + const hrefs = collectHrefs(html) |
| 34 | + const fromUrl = fileToUrl(file) |
| 35 | + |
| 36 | + for (const href of hrefs) { |
| 37 | + const resolved = resolveInternalPage(fromUrl, href) |
| 38 | + if (!resolved) continue |
| 39 | + internalRefs++ |
| 40 | + if (!pageExists(resolved.candidates)) { |
| 41 | + if (!deadPages.has(resolved.path)) deadPages.set(resolved.path, new Set()) |
| 42 | + deadPages.get(resolved.path).add(fromUrl) |
| 43 | + } |
| 44 | + } |
| 45 | +} |
| 46 | + |
| 47 | +if (deadPages.size > 0) { |
| 48 | + const ranked = [...deadPages.entries()] |
| 49 | + .map(([path, sources]) => ({ |
| 50 | + path, |
| 51 | + sourceCount: sources.size, |
| 52 | + sources: [...sources].slice(0, 5), |
| 53 | + })) |
| 54 | + .sort((a, b) => b.sourceCount - a.sourceCount) |
| 55 | + |
| 56 | + console.error( |
| 57 | + `Internal link verification failed: ${ranked.length} dead page path(s), ` + |
| 58 | + `${internalRefs} internal page href(s) checked across ${htmlFiles.length} HTML files.\n` |
| 59 | + ) |
| 60 | + for (const item of ranked.slice(0, 80)) { |
| 61 | + console.error( |
| 62 | + `- ${item.path} (${item.sourceCount} source page(s), e.g. ${item.sources.join(', ')})` |
| 63 | + ) |
| 64 | + } |
| 65 | + if (ranked.length > 80) { |
| 66 | + console.error(`… and ${ranked.length - 80} more`) |
| 67 | + } |
| 68 | + process.exit(1) |
| 69 | +} |
| 70 | + |
| 71 | +console.log( |
| 72 | + `Internal link verification passed: ${htmlFiles.length} HTML pages, ` + |
| 73 | + `${internalRefs} internal page href(s), 0 dead targets.` |
| 74 | +) |
| 75 | + |
| 76 | +function collectHrefs(html) { |
| 77 | + const out = new Set() |
| 78 | + const re = /\bhref=["']([^"']+)["']/gi |
| 79 | + let match |
| 80 | + while ((match = re.exec(html)) !== null) { |
| 81 | + out.add(match[1]) |
| 82 | + } |
| 83 | + return out |
| 84 | +} |
| 85 | + |
| 86 | +function fileToUrl(file) { |
| 87 | + let rel = toPosix(relative(distDir, file)) |
| 88 | + if (rel === 'index.html') return '/' |
| 89 | + if (rel.endsWith('/index.html')) { |
| 90 | + return `/${rel.slice(0, -'index.html'.length)}` |
| 91 | + } |
| 92 | + if (rel.endsWith('.html')) { |
| 93 | + return `/${rel.slice(0, -'.html'.length)}` |
| 94 | + } |
| 95 | + return `/${rel}` |
| 96 | +} |
| 97 | + |
| 98 | +/** |
| 99 | + * @returns {{ path: string, candidates: string[] } | null} |
| 100 | + */ |
| 101 | +function resolveInternalPage(fromUrl, href) { |
| 102 | + if ( |
| 103 | + !href || |
| 104 | + href.startsWith('#') || |
| 105 | + href.startsWith('mailto:') || |
| 106 | + href.startsWith('tel:') || |
| 107 | + href.startsWith('data:') || |
| 108 | + href.startsWith('javascript:') |
| 109 | + ) { |
| 110 | + return null |
| 111 | + } |
| 112 | + |
| 113 | + let pathWithQuery = href |
| 114 | + if (/^https?:\/\//i.test(href) || href.startsWith('//')) { |
| 115 | + try { |
| 116 | + const url = new URL(href.startsWith('//') ? `https:${href}` : href) |
| 117 | + if (!SITE_HOSTS.has(url.hostname)) return null |
| 118 | + pathWithQuery = `${url.pathname}${url.search}${url.hash}` |
| 119 | + } catch { |
| 120 | + return null |
| 121 | + } |
| 122 | + } |
| 123 | + |
| 124 | + const cut = pathWithQuery.search(/[?#]/) |
| 125 | + const pathOnly = cut === -1 ? pathWithQuery : pathWithQuery.slice(0, cut) |
| 126 | + if (!pathOnly) return null |
| 127 | + |
| 128 | + const lastSeg = pathOnly.split('/').pop() || '' |
| 129 | + // Static assets and leftover .md / api: refs are not page routes for this gate. |
| 130 | + if (lastSeg.includes('.') && !lastSeg.endsWith('.html')) { |
| 131 | + return null |
| 132 | + } |
| 133 | + |
| 134 | + let urlPath = pathOnly |
| 135 | + if (!urlPath.startsWith('/')) { |
| 136 | + const base = fromUrl.endsWith('/') |
| 137 | + ? fromUrl |
| 138 | + : `${fromUrl.split('/').slice(0, -1).join('/') || ''}/` |
| 139 | + urlPath = posixNormalize(`${base}${urlPath}`) |
| 140 | + } else { |
| 141 | + urlPath = posixNormalize(urlPath) |
| 142 | + } |
| 143 | + if (!urlPath.startsWith('/')) urlPath = `/${urlPath}` |
| 144 | + |
| 145 | + const candidates = [] |
| 146 | + if (urlPath.endsWith('/')) { |
| 147 | + candidates.push(resolve(distDir, `.${urlPath}`, 'index.html')) |
| 148 | + } else if (urlPath.endsWith('.html')) { |
| 149 | + candidates.push(resolve(distDir, `.${urlPath}`)) |
| 150 | + } else { |
| 151 | + candidates.push(resolve(distDir, `.${urlPath}.html`)) |
| 152 | + candidates.push(resolve(distDir, `.${urlPath}`, 'index.html')) |
| 153 | + } |
| 154 | + |
| 155 | + return { path: urlPath, candidates } |
| 156 | +} |
| 157 | + |
| 158 | +function pageExists(candidates) { |
| 159 | + return candidates.some((file) => existsSync(file)) |
| 160 | +} |
| 161 | + |
| 162 | +function posixNormalize(path) { |
| 163 | + const trailingSlash = path.endsWith('/') && path !== '/' |
| 164 | + const parts = [] |
| 165 | + for (const part of path.split('/')) { |
| 166 | + if (!part || part === '.') continue |
| 167 | + if (part === '..') { |
| 168 | + parts.pop() |
| 169 | + continue |
| 170 | + } |
| 171 | + parts.push(part) |
| 172 | + } |
| 173 | + const body = `/${parts.join('/')}` |
| 174 | + if (body === '/') return '/' |
| 175 | + return trailingSlash ? `${body}/` : body |
| 176 | +} |
| 177 | + |
| 178 | +function toPosix(value) { |
| 179 | + return value.replaceAll('\\', '/') |
| 180 | +} |
| 181 | + |
| 182 | +async function listFiles(directory, extension) { |
| 183 | + const entries = await readdir(directory, { withFileTypes: true }) |
| 184 | + const nested = await Promise.all( |
| 185 | + entries.map(async (entry) => { |
| 186 | + const target = resolve(directory, entry.name) |
| 187 | + if (entry.isDirectory()) return listFiles(target, extension) |
| 188 | + return entry.isFile() && entry.name.endsWith(extension) ? [target] : [] |
| 189 | + }) |
| 190 | + ) |
| 191 | + return nested.flat() |
| 192 | +} |
0 commit comments