Auto-sanitize + gate translation quote artifacts

scripts/__tests__/frontmatter-check.test.mjs

@@ -0,0 +1,19 @@
+import { test } from 'node:test';
+import assert from 'node:assert/strict';
+import { frontmatterError } from '../check-mdx-parse.mjs';
+
+test('frontmatterError flags nested-quote frontmatter', () => {
+  const content = '---\ntitle: "ok"\ndescription: "带有、"显示所有方案"链接。"\n---\n\nBody\n';
+  const err = frontmatterError(content);
+  assert.ok(err, 'expected an error object');
+  assert.equal(typeof err.message, 'string');
+});
+
+test('frontmatterError returns null for valid frontmatter', () => {
+  const content = '---\ntitle: "ok"\ndescription: "a clean one"\n---\n\nBody\n';
+  assert.equal(frontmatterError(content), null);
+});
+
+test('frontmatterError returns null when there is no frontmatter', () => {
+  assert.equal(frontmatterError('Just body text, no frontmatter.\n'), null);
+});

scripts/__tests__/nested-quote.test.mjs

@@ -0,0 +1,114 @@
+import { test } from 'node:test';
+import assert from 'node:assert/strict';
+import {
+  fixFrontmatterQuotes,
+  fixTagAttrQuotes,
+  fixNestedQuotesInBody,
+  fixNestedQuotes,
+  parsesClean,
+} from '../lint-mdx.mjs';
+
+test('fixFrontmatterQuotes converts inner double-quotes to single quotes', () => {
+  const fm = 'title: "ok"\ndescription: "构建带有、"显示所有方案"链接的付费墙。"';
+  const { result, changed } = fixFrontmatterQuotes(fm);
+  assert.equal(changed, true);
+  assert.equal(
+    result,
+    'title: "ok"\ndescription: "构建带有、\'显示所有方案\'链接的付费墙。"',
+  );
+});
+
+test('fixFrontmatterQuotes leaves clean frontmatter untouched', () => {
+  const fm = 'title: "ok"\ndescription: "a clean description"';
+  const { result, changed } = fixFrontmatterQuotes(fm);
+  assert.equal(changed, false);
+  assert.equal(result, fm);
+});
+
+test('fixFrontmatterQuotes preserves escaped quotes', () => {
+  const fm = 'description: "she said \\"hi\\""';
+  const { result, changed } = fixFrontmatterQuotes(fm);
+  assert.equal(changed, false);
+  assert.equal(result, fm);
+});
+
+test('fixTagAttrQuotes fixes inner quotes in the last attribute', () => {
+  const tag = '<ZoomImage id="a.webp" width="500px" alt="配置了"打开 URL"操作的链接" />';
+  const { result, changed } = fixTagAttrQuotes(tag);
+  assert.equal(changed, true);
+  assert.equal(
+    result,
+    '<ZoomImage id="a.webp" width="500px" alt="配置了\'打开 URL\'操作的链接" />',
+  );
+});
+
+test('fixTagAttrQuotes leaves a clean tag untouched', () => {
+  const tag = '<ZoomImage id="a.webp" width="500px" alt="clean alt text" />';
+  const { result, changed } = fixTagAttrQuotes(tag);
+  assert.equal(changed, false);
+  assert.equal(result, tag);
+});
+
+test('fixTagAttrQuotes does not flag a valueless/boolean attribute', () => {
+  // `default` is a boolean attribute — the closing quote of `label` is
+  // followed by ` default>`, which must be recognized as a boundary.
+  const tag = '<TabItem value="builder" label="Paywall Builder" default>';
+  const { result, changed } = fixTagAttrQuotes(tag);
+  assert.equal(changed, false);
+  assert.equal(result, tag);
+});
+
+test('fixTagAttrQuotes leaves a clean tag with trailing boolean attr + close', () => {
+  const tag = '<Tag a="x" disabled />';
+  const { result, changed } = fixTagAttrQuotes(tag);
+  assert.equal(changed, false);
+  assert.equal(result, tag);
+});
+
+test('fixNestedQuotesInBody skips fenced code blocks', () => {
+  // Use a realistic CJK artifact: a stray inner quote followed by non-ASCII
+  // text (what the translator actually produces). ASCII continuations are
+  // intentionally read as adjacent attributes, not inner quotes.
+  const body = [
+    '```jsx',
+    '<ZoomImage alt="甲"乙"丙" />',
+    '```',
+    '<ZoomImage alt="甲"乙"丙" />',
+  ].join('\n');
+  const { result, changed } = fixNestedQuotesInBody(body);
+  assert.equal(changed, true);
+  const out = result.split('\n');
+  // Code-fence line (index 1) preserved verbatim; only the real tag (index 3) fixed.
+  assert.equal(out[1], '<ZoomImage alt="甲"乙"丙" />');
+  assert.equal(out[3], '<ZoomImage alt="甲\'乙\'丙" />');
+});
+
+const BROKEN_FILE = [
+  '---',
+  'title: "ok"',
+  'description: "带有、"显示所有方案"链接。"',
+  '---',
+  '',
+  'Intro.',
+  '<ZoomImage id="a.webp" alt="配置了"打开 URL"操作" />',
+].join('\n');
+
+test('fixNestedQuotes repairs frontmatter and body together', () => {
+  const { result, changed } = fixNestedQuotes(BROKEN_FILE);
+  assert.equal(changed, true);
+  assert.ok(result.includes('description: "带有、\'显示所有方案\'链接。"'));
+  assert.ok(result.includes('alt="配置了\'打开 URL\'操作"'));
+});
+
+test('parsesClean: false for broken file, true after fix', async () => {
+  assert.equal(await parsesClean(BROKEN_FILE), false);
+  const { result } = fixNestedQuotes(BROKEN_FILE);
+  assert.equal(await parsesClean(result), true);
+});
+
+test('fixNestedQuotes leaves a fully valid file unchanged', () => {
+  const ok = '---\ntitle: "ok"\n---\n\n<ZoomImage id="a.webp" alt="clean" />\n';
+  const { result, changed } = fixNestedQuotes(ok);
+  assert.equal(changed, false);
+  assert.equal(result, ok);
+});

scripts/check-mdx-parse.mjs

@@ -16,9 +16,11 @@
 
 import fs from 'node:fs/promises';
 import path from 'node:path';
+import { fileURLToPath } from 'node:url';
 import { compile } from '@mdx-js/mdx';
 import remarkDirective from 'remark-directive';
 import { remarkAside } from '../src/plugins/remark-aside.mjs';
+import yaml from 'js-yaml';
 
 const ROOT = process.cwd();
 const SCAN_DIRS = ['src/content/docs', 'src/locales', 'src/components/reusable'];
@@ -42,9 +44,39 @@ async function* walk(dir) {
   }
 }
 
+// Returns a parse-error descriptor for the file's YAML frontmatter, or null if
+// the frontmatter is valid / absent. @mdx-js/mdx does not validate frontmatter,
+// so this closes the gap that let bad-YAML translations reach the slow build.
+export function frontmatterError(content) {
+  const m = content.match(/^?---\r?\n([\s\S]*?)\r?\n---/);
+  if (!m) return null;
+  try {
+    yaml.load(m[1]);
+    return null;
+  } catch (err) {
+    return {
+      message: err.message.split('\n')[0],
+      // js-yaml mark line is 0-based and relative to the frontmatter body;
+      // +2 maps it to the file (1 for the opening `---`, 1 for 1-based lines).
+      line: err.mark ? err.mark.line + 2 : null,
+      column: err.mark ? err.mark.column + 1 : null,
+    };
+  }
+}
+
 async function checkFile(file) {
+  const content = await fs.readFile(file, 'utf-8');
+  const fmErr = frontmatterError(content);
+  if (fmErr) {
+    return {
+      file: path.relative(ROOT, file),
+      message: `frontmatter: ${fmErr.message}`,
+      line: fmErr.line,
+      column: fmErr.column,
+    };
+  }
   try {
-    await compile(await fs.readFile(file, 'utf-8'), {
+    await compile(content, {
       jsx: true,
       remarkPlugins: [remarkDirective, remarkAside],
     });
@@ -59,29 +91,37 @@ async function checkFile(file) {
   }
 }
 
-const files = [];
-for (const dir of SCAN_DIRS) {
-  for await (const f of walk(path.join(ROOT, dir))) files.push(f);
-}
+async function main() {
+  const files = [];
+  for (const dir of SCAN_DIRS) {
+    for await (const f of walk(path.join(ROOT, dir))) files.push(f);
+  }
 
-const issues = [];
-for (let i = 0; i < files.length; i += CONCURRENCY) {
-  const chunk = files.slice(i, i + CONCURRENCY);
-  const results = await Promise.all(chunk.map(checkFile));
-  for (const r of results) if (r) issues.push(r);
-}
+  const issues = [];
+  for (let i = 0; i < files.length; i += CONCURRENCY) {
+    const chunk = files.slice(i, i + CONCURRENCY);
+    const results = await Promise.all(chunk.map(checkFile));
+    for (const r of results) if (r) issues.push(r);
+  }
 
-issues.sort((a, b) => a.file.localeCompare(b.file));
+  issues.sort((a, b) => a.file.localeCompare(b.file));
 
-if (issues.length === 0) {
-  console.log(`check-mdx-parse: ${files.length} file(s) parsed cleanly`);
-  process.exit(0);
+  if (issues.length === 0) {
+    console.log(`check-mdx-parse: ${files.length} file(s) parsed cleanly`);
+    process.exit(0);
+  }
+
+  console.error(`check-mdx-parse: ${files.length} scanned, ${issues.length} parse error(s):\n`);
+  for (const i of issues) {
+    const loc = i.line ? `${i.file}:${i.line}${i.column ? ':' + i.column : ''}` : i.file;
+    console.error(`  ${loc}`);
+    console.error(`    ${i.message}\n`);
+  }
+  process.exit(1);
 }
 
-console.error(`check-mdx-parse: ${files.length} scanned, ${issues.length} parse error(s):\n`);
-for (const i of issues) {
-  const loc = i.line ? `${i.file}:${i.line}${i.column ? ':' + i.column : ''}` : i.file;
-  console.error(`  ${loc}`);
-  console.error(`    ${i.message}\n`);
+const isMain = process.argv[1]
+  && path.resolve(process.argv[1]) === fileURLToPath(import.meta.url);
+if (isMain) {
+  main();
 }
-process.exit(1);