test(qa): e2e 装配修正 + 集成/门禁测试与工具链重构
- e2e：修死选择器与蜜罐误命中、补断言、GA4/security-headers 用例诚实标注 skip 边界；新增 assert-helpers/fixtures/hydrated/primary-nav/touch-targets 等复用桩。 - 集成层：config/test/jest.integration.config.js + tests-integration/ 真库 一次性 SQLite 用例，teardown 守卫开发库指纹。 - 门禁工具链：jest.setup 归位到根、scripts/accessibility 全站 axe 节点计数、 check-brand-text-token/check-motion-constraints 机械守卫、stryker/lighthouse 配置收敛。
This commit is contained in:
@@ -0,0 +1,283 @@
|
||||
// 全站 axe 节点计数门禁（验收 §8-⑤ 的永久化版本）。
|
||||
//
|
||||
// 来源：本文件由 `docs/acceptance/2026-09-21-axe/axe-contrast-evidence.mjs` 平移而来（2026-09-23），
|
||||
// 行为口径不变：同一份路由清单在 **chromium + firefox** × **light + dark** 四个组合下逐页跑两遍 axe-core
|
||||
// 1) tag 通道 `runOnly:{type:'tag',values:['wcag2a','wcag2aa','wcag21a','wcag21aa']}`
|
||||
// 2) 规则级通道 `runOnly:{type:'rules',values:EXTRA_RULES}`
|
||||
// 第 2 遍是 `presentation-role-conflict` 的**唯一真实覆盖**（该规则属 `best-practice` tag，不在 tag 通道内），
|
||||
// 也是 Lighthouse 那三条死断言删除后 `autocomplete-valid` / `svg-img-alt` 的兜底。
|
||||
// 证据链：docs/acceptance/2026-09-21-gates/final-tree-results.md §⑤ 与「Lighthouse 的 3 条死断言」。
|
||||
//
|
||||
// 与一次性版本的关键差异：**门禁判定从「打印 PASSED」升级为「exit code」**，并新增分母自证断言，
|
||||
// 使「悄悄不再覆盖」无法以绿灯形式通过（CI 见 Jenkinsfile 的 ♿♿ 阶段）。
|
||||
//
|
||||
// 用法：
|
||||
// npm run check:axe:routes # 生成路由清单（默认 /tmp/axe-routes.xml）
|
||||
// npm run check:axe # 用该清单扫描并判定
|
||||
// 环境变量：BASE（默认 http://localhost:3100，须是 `node dist/standalone/server.js` 起的生产服务，
|
||||
// 非 dev server）、SITEMAP（路由清单 XML）、OUT（证据 JSON 输出目录，默认 docs/acceptance/2026-09-21-axe，
|
||||
// 历史证据仍留在这里）。
|
||||
//
|
||||
// 退出码：0 通过 / 1 门禁判红（violation 节点 > 0、主题或底色不匹配、规则级通道分母不闭合）/
|
||||
// 2 输入不可用（路由清单缺失或解析出 0 条）。
|
||||
|
||||
import { chromium, firefox } from 'playwright';
|
||||
import { mkdirSync, writeFileSync, readFileSync, existsSync } from 'fs';
|
||||
import { fileURLToPath } from 'url';
|
||||
import { dirname, isAbsolute, join, resolve } from 'path';
|
||||
|
||||
// 本脚本从证据目录 docs/acceptance/2026-09-21-axe/ 搬到 scripts/accessibility/，cwd 相对路径一律改为
|
||||
// 仓库根相对，这样无论从哪个目录调用都指向同一份 axe-core 与同一处证据目录（行为等价，只是不再依赖 cwd）。
|
||||
const REPO_ROOT = resolve(dirname(fileURLToPath(import.meta.url)), '..', '..');
|
||||
const repoPath = (p) => (isAbsolute(p) ? p : resolve(REPO_ROOT, p));
|
||||
|
||||
const BASE = process.env.BASE || 'http://localhost:3100';
|
||||
const OUT = repoPath(process.env.OUT || 'docs/acceptance/2026-09-21-axe');
|
||||
const SITEMAP = repoPath(process.env.SITEMAP || '/tmp/axe-routes.xml');
|
||||
const AXE_SCRIPT = repoPath('node_modules/axe-core/axe.min.js');
|
||||
const TAGS = ['wcag2a', 'wcag2aa', 'wcag21a', 'wcag21aa'];
|
||||
// Lighthouse 12.6.1 不包装这三条 axe 规则，且 presentation-role-conflict 属 best-practice tag（不在 TAGS 内）
|
||||
// ⇒ 只有这里的规则级 runOnly 才真正测到它们；证据链见 docs/acceptance/2026-09-21-gates/final-tree-results.md
|
||||
//
|
||||
// 两条数组刻意分开：
|
||||
// CANONICAL_EXTRA_RULES = 门禁**要求被判定**的三条规则（绝对下限，写死在此）
|
||||
// EXTRA_RULES = 实际传给 axe 的 runOnly 列表（默认等于下限）
|
||||
// 只删 EXTRA_RULES 会立即判红（分母不再等于 路由 × 组合 × 3 条下限），删 CANONICAL 则需要改动带注释的门禁契约本身。
|
||||
const CANONICAL_EXTRA_RULES = ['autocomplete-valid', 'presentation-role-conflict', 'svg-img-alt'];
|
||||
const EXTRA_RULES = [...CANONICAL_EXTRA_RULES];
|
||||
const ENGINES = [['chromium', chromium], ['firefox', firefox]];
|
||||
const THEMES = ['light', 'dark'];
|
||||
const COMBOS = ENGINES.length * THEMES.length; // 4 个「引擎 × 主题」组合
|
||||
// 底色必须等于本站令牌值：未加载样式时所有元素都落在默认黑白底上，对比度会「意外通过」。
|
||||
const EXPECTED_BG = { light: 'rgb(255, 255, 255)', dark: 'rgb(10, 14, 20)' };
|
||||
// 逐页状态码期望：清单里绝大多数路由必须 2xx/3xx，`/_not-found` 是本站 404 页、**期望** 404。
|
||||
// 为什么必须逐页判：404/500 页几乎没有可比对节点，`violationNodes` 仍为 0，而 `s.pages` 计数照加；
|
||||
// 聚合保险丝 `okResponses > 0` 只在**全部**响应都非 2xx 时才炸 ⇒ 单个路由退化成 404 会被当「干净」计过。
|
||||
// 证据：docs/acceptance/2026-09-21-axe/axe-evidence.json 实测 136 行 = 132×200 + 4×404（全部来自 /_not-found）。
|
||||
const EXPECTED_404 = ['/_not-found'];
|
||||
const statusOffenders = [];
|
||||
|
||||
function fatal(message, code = 2) {
|
||||
console.error(`FATAL: ${message}`);
|
||||
process.exit(code);
|
||||
}
|
||||
|
||||
// 扫描中途抛错（浏览器启动失败、服务不可达、axe 注入失败）会以非 0 退出，这里只是把原因说清楚。
|
||||
for (const ev of ['unhandledRejection', 'uncaughtException']) {
|
||||
process.on(ev, (err) => fatal(`扫描中断（${ev}）：${err instanceof Error ? err.message : String(err)}`, 1));
|
||||
}
|
||||
|
||||
if (!existsSync(AXE_SCRIPT)) {
|
||||
fatal(`缺少 ${AXE_SCRIPT}（axe-core 未安装？先 npm ci）`, 2);
|
||||
}
|
||||
if (!existsSync(SITEMAP)) {
|
||||
fatal(`路由清单 ${SITEMAP} 不存在。先跑 npm run check:axe:routes 生成清单`, 2);
|
||||
}
|
||||
|
||||
const urls = [...new Set(
|
||||
readFileSync(SITEMAP, 'utf8')
|
||||
.split('<url>')
|
||||
.slice(1)
|
||||
.map((b) => (b.match(/<loc>([^<]+)<\/loc>/) || [])[1])
|
||||
.map((loc) => {
|
||||
try {
|
||||
return new URL(loc).pathname;
|
||||
} catch {
|
||||
return null;
|
||||
}
|
||||
})
|
||||
.filter(Boolean)
|
||||
)];
|
||||
|
||||
// 空清单时下面的循环零次执行、计数器全为 0，「PASSED = true」是假绿，必须拒绝出证据。
|
||||
if (urls.length === 0) {
|
||||
fatal(`从 ${SITEMAP} 解析出 0 条路由，扫描等于没测，退出 2`, 2);
|
||||
}
|
||||
|
||||
async function settle(page) {
|
||||
await page.waitForTimeout(1200);
|
||||
await page.evaluate(async () => {
|
||||
let last = -1;
|
||||
for (let i = 0; i < 12; i += 1) {
|
||||
window.scrollTo(0, document.body.scrollHeight);
|
||||
await new Promise((r) => setTimeout(r, 180));
|
||||
if (Math.round(window.scrollY) === last) break;
|
||||
last = Math.round(window.scrollY);
|
||||
}
|
||||
window.scrollTo(0, 0);
|
||||
});
|
||||
await page.waitForTimeout(400);
|
||||
}
|
||||
|
||||
const rows = [];
|
||||
for (const [engineName, launcher] of ENGINES) {
|
||||
const browser = await launcher.launch();
|
||||
for (const theme of THEMES) {
|
||||
const ctx = await browser.newContext({ colorScheme: theme });
|
||||
await ctx.addInitScript((t) => {
|
||||
localStorage.setItem('novalon-theme', t);
|
||||
localStorage.setItem('novalon-cookie-preferences',
|
||||
JSON.stringify({ necessary: true, analytics: false, functional: false, timestamp: '2026-09-21T00:00:00.000Z' }));
|
||||
}, theme);
|
||||
const page = await ctx.newPage();
|
||||
for (const url of urls) {
|
||||
const res = await page.goto(`${BASE}${url}`, { waitUntil: 'domcontentloaded', timeout: 45000 });
|
||||
const statusCode = res ? res.status() : 0;
|
||||
await settle(page);
|
||||
const observedTheme = await page.evaluate(() =>
|
||||
document.documentElement.getAttribute('data-theme'));
|
||||
await page.addScriptTag({ path: AXE_SCRIPT });
|
||||
const tagRun = await page.evaluate((tags) => window.axe.run(document, {
|
||||
runOnly: { type: 'tag', values: tags },
|
||||
resultTypes: ['violations'],
|
||||
}), TAGS);
|
||||
const violations = tagRun.violations.map((v) => ({
|
||||
id: v.id, impact: v.impact, nodes: v.nodes.length,
|
||||
sample: v.nodes.slice(0, 3).map((n) => (n.target || []).join(' ')),
|
||||
}));
|
||||
// 规则级复跑：axe 遇到未知 rule id 会直接抛错；extraRulesChecked 断言三条规则确实参与了本页运行。
|
||||
const ruleRes = await page.evaluate((rules) => window.axe.run(document, {
|
||||
runOnly: { type: 'rules', values: rules },
|
||||
}), EXTRA_RULES);
|
||||
const ruleViolations = ruleRes.violations.map((v) => ({
|
||||
id: v.id, impact: v.impact, nodes: v.nodes.length,
|
||||
sample: v.nodes.slice(0, 3).map((n) => (n.target || []).join(' ')),
|
||||
}));
|
||||
const ruleIds = new Set([
|
||||
...ruleRes.violations, ...ruleRes.passes, ...ruleRes.incomplete, ...ruleRes.inapplicable,
|
||||
].map((r) => r.id));
|
||||
const covered = CANONICAL_EXTRA_RULES.filter((id) => ruleIds.has(id));
|
||||
const pageBg = await page.evaluate(() => getComputedStyle(document.body).backgroundColor);
|
||||
rows.push({
|
||||
engine: engineName, theme, url, observedTheme, statusCode, pageBg,
|
||||
bgMatchesToken: pageBg === EXPECTED_BG[theme],
|
||||
violationRules: violations.length,
|
||||
violationNodes: violations.reduce((a, v) => a + v.nodes, 0),
|
||||
contrastNodes: violations.filter((v) => v.id === 'color-contrast').reduce((a, v) => a + v.nodes, 0),
|
||||
extraRuleNodes: ruleViolations.reduce((a, v) => a + v.nodes, 0),
|
||||
extraRulesChecked: ruleIds.size,
|
||||
extraRulesCovered: covered,
|
||||
extraRulesMissing: CANONICAL_EXTRA_RULES.filter((id) => !ruleIds.has(id)),
|
||||
violations,
|
||||
extraRuleViolations: ruleViolations,
|
||||
});
|
||||
console.log(`${engineName}/${theme} ${url} status=${statusCode} theme=${observedTheme} bg=${pageBg} rules=${violations.length} nodes=${rows.at(-1).violationNodes} contrast=${rows.at(-1).contrastNodes} extra=${rows.at(-1).extraRuleNodes}/${covered.length}/${rows.at(-1).extraRulesChecked}`);
|
||||
}
|
||||
await ctx.close();
|
||||
}
|
||||
await browser.close();
|
||||
}
|
||||
|
||||
const summary = {};
|
||||
for (const r of rows) {
|
||||
const k = `${r.engine}/${r.theme}`;
|
||||
summary[k] = summary[k] || {
|
||||
pages: 0, themeMismatch: 0, bgMismatch: 0, contrastNodes: 0, violationNodes: 0,
|
||||
extraRuleNodes: 0, extraRulePagesUnderCovered: 0, extraRuleCoveragePairs: 0,
|
||||
non2xxResponses: 0, okResponses: 0, unexpectedStatus: 0,
|
||||
};
|
||||
summary[k].pages += 1;
|
||||
summary[k].contrastNodes += r.contrastNodes;
|
||||
summary[k].violationNodes += r.violationNodes;
|
||||
summary[k].extraRuleNodes += r.extraRuleNodes;
|
||||
summary[k].extraRuleCoveragePairs += r.extraRulesCovered.length;
|
||||
// 两条都要满足：三条下限规则逐条判定过（缺一条即覆盖腐化），且通道本身没跑出预期之外的形状。
|
||||
if (r.extraRulesMissing.length > 0 || r.extraRulesChecked !== EXTRA_RULES.length) {
|
||||
summary[k].extraRulePagesUnderCovered += 1;
|
||||
}
|
||||
if (r.observedTheme !== r.theme) summary[k].themeMismatch += 1;
|
||||
if (!r.bgMatchesToken) summary[k].bgMismatch += 1;
|
||||
if (r.statusCode >= 200 && r.statusCode < 400) summary[k].okResponses += 1;
|
||||
else summary[k].non2xxResponses += 1;
|
||||
// 逐页状态码判据（独立于上面两条聚合计数）
|
||||
const expects404 = EXPECTED_404.includes(r.url);
|
||||
const statusMatches = expects404
|
||||
? r.statusCode === 404
|
||||
: r.statusCode >= 200 && r.statusCode < 400;
|
||||
if (!statusMatches) {
|
||||
summary[k].unexpectedStatus += 1;
|
||||
statusOffenders.push(
|
||||
`${k} ${r.url} status=${r.statusCode} expected=${expects404 ? '404' : '2xx/3xx'}`,
|
||||
);
|
||||
}
|
||||
}
|
||||
|
||||
const expectedCombos = ENGINES.flatMap(([e]) => THEMES.map((t) => `${e}/${t}`));
|
||||
const totalRuleChecks = rows.reduce((a, r) => a + r.extraRulesChecked, 0);
|
||||
const totalCanonicalCoverage = rows.reduce((a, r) => a + r.extraRulesCovered.length, 0);
|
||||
const expectedCanonicalCoverage = urls.length * COMBOS * CANONICAL_EXTRA_RULES.length;
|
||||
const missingRuleUnion = [...new Set(rows.flatMap((r) => r.extraRulesMissing))];
|
||||
|
||||
// ---- 门禁断言（exit code 由此决定，不再只是打印） ----
|
||||
const failures = [];
|
||||
const push = (cond, msg) => { if (!cond) failures.push(msg); };
|
||||
|
||||
// 分母自证：缺了它，「少扫了几个组合」与「扫了且全绿」在证据上无法区分。
|
||||
push(rows.length === urls.length * COMBOS,
|
||||
`分母不闭合：rows=${rows.length}，应为 ${urls.length} 路由 × ${COMBOS} 组合 = ${urls.length * COMBOS}`);
|
||||
for (const combo of expectedCombos) {
|
||||
const s = summary[combo];
|
||||
if (!s) { push(false, `${combo} 组合完全缺失（浏览器未启动或中途中断）`); continue; }
|
||||
push(s.pages === urls.length, `${combo} 只扫了 ${s.pages} 页，清单有 ${urls.length} 页`);
|
||||
push(s.violationNodes === 0, `${combo} violationNodes=${s.violationNodes}（阈值 0）`);
|
||||
push(s.contrastNodes === 0, `${combo} contrastNodes=${s.contrastNodes}（阈值 0）`);
|
||||
push(s.extraRuleNodes === 0, `${combo} extraRuleNodes=${s.extraRuleNodes}（三条规则级判据，阈值 0）`);
|
||||
push(s.extraRulePagesUnderCovered === 0,
|
||||
`${combo} 有 ${s.extraRulePagesUnderCovered} 页的规则级通道未覆盖满 ${CANONICAL_EXTRA_RULES.length} 条下限规则（缺：${missingRuleUnion.join(', ') || '形状异常'}）`);
|
||||
push(s.extraRuleCoveragePairs === urls.length * CANONICAL_EXTRA_RULES.length,
|
||||
`${combo} 规则级覆盖 ${s.extraRuleCoveragePairs} ≠ ${urls.length} 路由 × ${CANONICAL_EXTRA_RULES.length} 规则`);
|
||||
push(s.themeMismatch === 0, `${combo} themeMismatch=${s.themeMismatch}（以为在量深色其实还是浅色）`);
|
||||
push(s.bgMismatch === 0, `${combo} bgMismatch=${s.bgMismatch}（底色不等于令牌值 ⇒ 样式未加载 ⇒ 对比度「意外达标」）`);
|
||||
// 保险丝：整组全非 2xx ⇒ 服务或清单不对，扫的是 404 页而不是本站页面。
|
||||
push(s.okResponses > 0, `${combo} 所有响应都非 2xx（BASE=${BASE} 不是可用的生产服务，或清单已过期）`);
|
||||
push(s.unexpectedStatus === 0,
|
||||
`${combo} 有 ${s.unexpectedStatus} 页状态码不符期望（404/500 页会被 axe 判为「无违规」且仍计入 pages 分母；期望见 EXPECTED_404）`);
|
||||
}
|
||||
// offenders 明细直接进判据文本（而非只打 stdout）：本仓曾因 line reporter 不带 stdout
|
||||
// 而在并发失败时「查不到规则名」，同理这里若不落进 failures 就查不到是哪一条路由 404。
|
||||
if (statusOffenders.length > 0) {
|
||||
push(false, `逐页状态码断言失败 ${statusOffenders.length} 处：\n ${statusOffenders.join('\n ')}`);
|
||||
}
|
||||
push(totalRuleChecks === urls.length * COMBOS * EXTRA_RULES.length,
|
||||
`规则级通道总量 ${totalRuleChecks} ≠ ${urls.length}×${COMBOS}×${EXTRA_RULES.length} = ${urls.length * COMBOS * EXTRA_RULES.length}`);
|
||||
// 绝对下限的分母闭合：三条规则 × 每一页 × 每一个组合，一个都不能少（0 违规 ≠ 没测到）。
|
||||
push(totalCanonicalCoverage === expectedCanonicalCoverage,
|
||||
`规则级覆盖 ${totalCanonicalCoverage} ≠ ${expectedCanonicalCoverage}` +
|
||||
`（= ${urls.length} 路由 × ${COMBOS} 组合 × ${CANONICAL_EXTRA_RULES.length} 条下限规则 ${CANONICAL_EXTRA_RULES.join(', ')}）`);
|
||||
|
||||
const artifact = {
|
||||
generatedAt: new Date().toISOString(),
|
||||
base: BASE,
|
||||
axeCore: JSON.parse(readFileSync(repoPath('node_modules/axe-core/package.json'), 'utf8')).version,
|
||||
routeSource: `${SITEMAP} → ${urls.length} 条路由（sitemap ∪ 预渲染产物 ∪ 站内链接 BFS）`,
|
||||
routeCount: urls.length,
|
||||
routes: urls,
|
||||
combinations: expectedCombos,
|
||||
tags: TAGS,
|
||||
extraRules: EXTRA_RULES,
|
||||
requiredRuleCoverage: CANONICAL_EXTRA_RULES,
|
||||
thresholds: {
|
||||
colorContrastNodes: 0, violationNodes: 0, themeMismatch: 0, bgMismatch: 0,
|
||||
extraRuleNodes: 0, extraRulePagesUnderCovered: 0,
|
||||
rows: urls.length * COMBOS,
|
||||
totalExtraRuleChecks: urls.length * COMBOS * EXTRA_RULES.length,
|
||||
requiredRuleCoverageTotal: expectedCanonicalCoverage,
|
||||
},
|
||||
totalExtraRuleChecks: totalRuleChecks,
|
||||
expectedExtraRuleChecks: urls.length * COMBOS * EXTRA_RULES.length,
|
||||
totalRequiredRuleCoverage: totalCanonicalCoverage,
|
||||
expectedRequiredRuleCoverage: expectedCanonicalCoverage,
|
||||
passed: failures.length === 0,
|
||||
failures,
|
||||
summary,
|
||||
rows,
|
||||
};
|
||||
mkdirSync(OUT, { recursive: true });
|
||||
writeFileSync(`${OUT}/axe-evidence.json`, JSON.stringify(artifact, null, 2));
|
||||
console.log('\nSUMMARY', JSON.stringify(summary, null, 2));
|
||||
console.log('PASSED =', artifact.passed);
|
||||
if (!artifact.passed) {
|
||||
console.error(`\nFAIL: axe 节点计数门禁 ${failures.length} 条判据未通过（证据已写入 ${OUT}/axe-evidence.json）`);
|
||||
for (const f of failures) console.error(` ✗ ${f}`);
|
||||
process.exit(1);
|
||||
}
|
||||
@@ -0,0 +1,165 @@
|
||||
// 全站页面清单生成器（`check:axe` 的分母来源）。
|
||||
//
|
||||
// 来源：本文件由 `docs/acceptance/2026-09-21-axe/crawl-sitemap.mjs` 平移而来（2026-09-23），口径不变：
|
||||
// 从 BASE 起 BFS 跟随同源链接，并与 sitemap / 预渲染产物取并集 —— 因为 `src/app/sitemap.ts` 只列
|
||||
// 「希望被收录」的页面，404 与根错误兜底页天然不在其中（见 docs/lessons-learned.md §5.24）。
|
||||
//
|
||||
// 用法（零参数即可跑通，Jenkins 的 ♿♿ 阶段就是这么调的）：
|
||||
// BASE=http://localhost:3100 npm run check:axe:routes
|
||||
// 环境变量：
|
||||
// BASE 生产服务地址（默认 http://localhost:3100；须是 `node dist/standalone/server.js`，不是 dev server）
|
||||
// SEED 清单种子：既可以是本地 sitemap.xml 路径，也可以是 http(s) URL（默认取 `${BASE}/sitemap.xml`）
|
||||
// MANIFEST 预渲染产物目录（默认 dist/standalone/dist/server/app），存在才参与并集
|
||||
// OUT 输出的路由清单 XML（默认 /tmp/axe-routes.xml，即 axe-node-count.mjs 的默认 SITEMAP）
|
||||
// MAX_PAGES BFS 上限（默认 200）
|
||||
// 退出码：0 成功 / 2 清单为空或 BASE 完全不可达（宁可不产出，也不要让门禁扫到 0 页还判绿）。
|
||||
|
||||
import { readFile, writeFile, readdir } from 'node:fs/promises';
|
||||
import { existsSync } from 'node:fs';
|
||||
import { dirname, isAbsolute, relative, resolve } from 'node:path';
|
||||
import { fileURLToPath } from 'node:url';
|
||||
|
||||
// 搬到 scripts/accessibility/ 后不再依赖 cwd：相对路径按仓库根解析。
|
||||
const REPO_ROOT = resolve(dirname(fileURLToPath(import.meta.url)), '..', '..');
|
||||
const repoPath = (p) => (isAbsolute(p) ? p : resolve(REPO_ROOT, p));
|
||||
|
||||
const BASE = process.env.BASE ?? 'http://localhost:3100';
|
||||
const SEED = process.env.SEED ?? `${BASE}/sitemap.xml`;
|
||||
const OUT = repoPath(process.env.OUT ?? '/tmp/axe-routes.xml');
|
||||
const MANIFEST = repoPath(process.env.MANIFEST ?? 'dist/standalone/dist/server/app');
|
||||
const MAX_PAGES = Number(process.env.MAX_PAGES ?? 200);
|
||||
const CONCURRENCY = 5;
|
||||
|
||||
const origin = new URL(BASE).origin;
|
||||
const found = new Set();
|
||||
const visited = new Set();
|
||||
|
||||
function normalize(raw) {
|
||||
let u;
|
||||
try {
|
||||
u = new URL(raw, BASE);
|
||||
} catch {
|
||||
return null;
|
||||
}
|
||||
if (u.origin !== origin) return null;
|
||||
if (u.search || u.hash) return null;
|
||||
const p = decodeURIComponent(u.pathname);
|
||||
if (/\.(svg|png|jpe?g|webp|avif|ico|css|js|mjs|txt|xml|json|woff2?|mp4|pdf|zip)$/i.test(p)) return null;
|
||||
if (p.startsWith('/api/') || p.startsWith('/_next/') || p.startsWith('/admin')) return null;
|
||||
if (p === '/_global-error') return null; // 框架内置 500 文档，见下方 MANIFEST 段的按内容判定
|
||||
if (p.includes('/uploads/')) return null;
|
||||
return p === '/' ? '/' : p.replace(/\/+$/, '') || '/';
|
||||
}
|
||||
|
||||
async function readSeed() {
|
||||
if (!SEED) return '';
|
||||
if (/^https?:\/\//i.test(SEED)) {
|
||||
try {
|
||||
const res = await fetch(SEED);
|
||||
if (!res.ok) {
|
||||
console.warn(`WARN: 种子 sitemap ${SEED} 返回 ${res.status}，仅用预渲染产物 + BFS 作为清单来源`);
|
||||
return '';
|
||||
}
|
||||
return await res.text();
|
||||
} catch (err) {
|
||||
console.warn(`WARN: 种子 sitemap ${SEED} 不可达（${err instanceof Error ? err.message : err}），仅用预渲染产物 + BFS`);
|
||||
return '';
|
||||
}
|
||||
}
|
||||
return readFile(repoPath(SEED), 'utf8'); // 本地文件路径：读不到就抛错，不静默降级
|
||||
}
|
||||
|
||||
for (const m of (await readSeed()).matchAll(/<loc>([^<]+)<\/loc>/g)) {
|
||||
const n = normalize(m[1]);
|
||||
if (n) found.add(n);
|
||||
}
|
||||
|
||||
// 预渲染产物清单：sitemap 只收静态页 + 已发布 slug，站内无入口的孤儿页不在其中。
|
||||
// 构建产物里的 .html 才是「实际会渲染的公开页面」全集，故与 sitemap 取并集。
|
||||
async function walk(dir) {
|
||||
const out = [];
|
||||
for (const ent of await readdir(dir, { withFileTypes: true })) {
|
||||
const p = `${dir}/${ent.name}`;
|
||||
if (ent.isDirectory()) out.push(...(await walk(p)));
|
||||
else if (ent.name.endsWith('.html')) out.push(p);
|
||||
}
|
||||
return out;
|
||||
}
|
||||
let manifestCount = 0;
|
||||
let skippedInternal = 0;
|
||||
if (existsSync(MANIFEST)) {
|
||||
for (const file of await walk(MANIFEST)) {
|
||||
// Next.js 为根错误路径预渲染的是**框架内置 500 文档**（`<html id="__next_error__">`，零应用标记、
|
||||
// 无 lang、深色落到浏览器默认画布），它不是本站编写的路由，扫它等于测框架模板。
|
||||
// 本站确实提供了 src/app/global-error.tsx（佐证：dist/.../chunks/ssr/src_app_global-error_tsx_*.js 内含本站文案），
|
||||
// 所以按**内容**判定而不是按名字硬编码：一旦框架把本站组件预渲染到此路径，条件不再命中，路由自动回到清单。
|
||||
if (/<html[^>]*id="__next_error__"/.test(await readFile(file, 'utf8'))) {
|
||||
skippedInternal++;
|
||||
continue;
|
||||
}
|
||||
const rel = relative(MANIFEST, file).replace(/\.html$/, '');
|
||||
const path = rel === 'index' ? '/' : `/${rel.replace(/\/index$/, '')}`;
|
||||
const n = normalize(path);
|
||||
if (n && !found.has(n)) manifestCount++;
|
||||
if (n) found.add(n);
|
||||
}
|
||||
} else {
|
||||
console.warn(`WARN: 预渲染产物目录 ${MANIFEST} 不存在，清单将缺少孤儿页（先 npm run build）`);
|
||||
}
|
||||
|
||||
let reachable = 0;
|
||||
let frontier = [...found];
|
||||
while (frontier.length && visited.size < MAX_PAGES) {
|
||||
const batch = frontier.splice(0, CONCURRENCY * 4);
|
||||
const results = await Promise.all(
|
||||
batch.map(async (path) => {
|
||||
if (visited.has(path)) return [];
|
||||
visited.add(path);
|
||||
try {
|
||||
const res = await fetch(BASE + path);
|
||||
if (!res.ok) return [];
|
||||
reachable += 1;
|
||||
const html = await res.text();
|
||||
const links = new Set();
|
||||
for (const m of html.matchAll(/<a\b[^>]*href="([^"]+)"/g)) {
|
||||
const n = normalize(m[1]);
|
||||
if (n && !visited.has(n)) links.add(n);
|
||||
}
|
||||
return [...links];
|
||||
} catch {
|
||||
return [];
|
||||
}
|
||||
})
|
||||
);
|
||||
for (const links of results) {
|
||||
for (const l of links) {
|
||||
if (!found.has(l)) {
|
||||
found.add(l);
|
||||
frontier.push(l);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
const paths = [...found].sort();
|
||||
|
||||
// 两条「宁红不假绿」的保险丝：清单为空 ⇒ 门禁等于没测；一条都不可达 ⇒ BASE 起错了/没起。
|
||||
// 两种情况下都**不写 OUT**，避免用空清单覆盖上一份可用清单。
|
||||
if (paths.length === 0) {
|
||||
console.error(`FATAL: 三源（${SEED} ∪ ${MANIFEST} ∪ BFS）并集为 0 条路由，不写出 ${OUT}`);
|
||||
process.exit(2);
|
||||
}
|
||||
if (visited.size > 0 && reachable === 0) {
|
||||
console.error(`FATAL: BASE=${BASE} 上 ${visited.size} 条候选路由全部不可达，清单不可信，不写出 ${OUT}`);
|
||||
process.exit(2);
|
||||
}
|
||||
|
||||
const xml =
|
||||
'<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n' +
|
||||
paths.map((p) => ` <url><loc>${BASE}${p}</loc></url>`).join('\n') +
|
||||
'\n</urlset>\n';
|
||||
await writeFile(OUT, xml);
|
||||
console.log(
|
||||
`crawled visited=${visited.size} reachable=${reachable} urls=${paths.length} manifestExtra=${manifestCount} skippedInternalDocs=${skippedInternal} -> ${OUT}`
|
||||
);
|
||||
console.log(paths.join('\n'));
|
||||
Reference in New Issue
Block a user