blog/scripts/build-perf/content-lint.mjs

#!/usr/bin/env node
// 构建期内容 lint（T3，零依赖）：扫 dist 产物三检——站内断链 / 图片存在 / zh↔en 配对
// 用法：node scripts/build-perf/content-lint.mjs [dist目录=dist]
// 逃生门：CONTENT_LINT=skip 整体跳过；CONTENT_LINT=warn 只警告不炸构建（首版兼容期）
// 口径：断链/缺图 = FAIL（exit 1）；配对缺失 = WARN（en 缺页可能是有意的，人工裁决）
import { readdirSync, readFileSync, existsSync, statSync } from 'node:fs';
import { join, resolve as pathResolve } from 'node:path';

const DIST = pathResolve(process.argv[2] ?? 'dist');
const MODE = process.env.CONTENT_LINT ?? 'fail';
if (MODE === 'skip') {
  console.log('[content-lint] CONTENT_LINT=skip，整体跳过');
  process.exit(0);
}

const SKIP_PREFIX = ['/admin/', '/photos-originals/']; // admin=CMS 路由；photos-originals=部署期 rsync 叠加层（不在 dist，另有部署链保障）
const SKIP_FILE = new Set(['/404.html']); // Astro 404 页不参与配对

// 一次性预览路由（无 en 对应页）：完全不参与检查。用户选定方案后随路由一起删除。
const SKIP_ALL = new Set(['/heatmap-preview/index.html']);


function walk(dir, out = []) {
  for (const e of readdirSync(dir)) {
    const p = join(dir, e);
    if (statSync(p).isDirectory()) walk(p, out);
    else if (e.endsWith('.html')) out.push(p);
  }
  return out;
}

/** URL 路径 → 磁盘路径：先解百分号编码（中文/空格文件名在 HTML 里是 %XX 形式） */
function decodePath(p) {
  try {
    return decodeURI(p);
  } catch {
    return p;
  }
}

/** /x/ → dist/x/index.html；/a.svg → dist/a.svg；找不到即 null */
function targetExists(url) {
  if (SKIP_PREFIX.some((s) => url.startsWith(s))) return true;
  const p = decodePath(url.split('#')[0].split('?')[0]);
  if (!p || p === '/') return existsSync(join(DIST, 'index.html'));
  if (p.endsWith('/')) return existsSync(join(DIST, p, 'index.html'));
  return existsSync(join(DIST, p)) || existsSync(join(DIST, `${p}.html`)) || existsSync(join(DIST, p, 'index.html'));
}

const files = walk(DIST);
const fails = [];
const warns = [];

for (const f of files) {
  const rel = f.slice(DIST.length) || '/';
  if (SKIP_ALL.has(rel)) continue;
  const html = readFileSync(f, 'utf8');

  // 站内 <a href>：仅根相对路径；#锚点/mailto/tel/外部/data: 全跳过
  for (const m of html.matchAll(/<a\s[^>]*href="([^"]*)"/g)) {
    const href = m[1];
    if (!href.startsWith('/') || href.startsWith('//')) continue;
    if (!targetExists(href)) fails.push(`断链 ${rel} → ${href}`);
  }

  // <img src> 与 srcset：产物必须存在（构建期副本检查，防部署后 404 图）
  for (const m of html.matchAll(/<img\s[^>]*>/g)) {
    const tag = m[0];
    const src = /(?:\ssrc="([^"]*)")/.exec(tag)?.[1];
    const srcset = /(?:\ssrcset="([^"]*)")/.exec(tag)?.[1];
    for (const u of [src, ...(srcset ?? '').split(',').map((s) => s.trim().split(/\s+/)[0])]) {
      if (!u || !u.startsWith('/') || u.startsWith('//')) continue;
      if (!targetExists(u)) fails.push(`缺图 ${rel} → ${u}`);
    }
  }

  // zh↔en 配对：dist/en/x/index.html ↔ dist/x/index.html，双向 WARN
  if (!SKIP_FILE.has(rel) && !SKIP_PREFIX.some((s) => rel.startsWith(s))) {
    const other = rel.startsWith('/en/') ? rel.slice(3) : `/en${rel === '/index.html' ? '/index.html' : rel}`;
    if (!existsSync(join(DIST, other))) warns.push(`配对缺 ${rel} ↔ ${other}`);
  }
}

for (const w of warns) console.warn(`[content-lint] WARN ${w}`);
if (fails.length) {
  for (const x of fails) console.error(`[content-lint] FAIL ${x}`);
  if (MODE === 'warn') {
    console.warn(`[content-lint] WARN 模式：${fails.length} 个失败降级为警告`);
    process.exit(0);
  }
  console.error(`[content-lint] ${fails.length} 失败 / ${warns.length} 警告`);
  process.exit(1);
}
console.log(`[content-lint] 通过：${files.length} 页扫描，0 断链 0 缺图，${warns.length} 配对警告`);