fix(mindspace): canonicalize public page links without MindSpace segment

2026-07-10 11:39:21 +08:00
parent aede1e6fcb
commit adc13a167e
2 changed files with 72 additions and 1 deletions
@@ -444,6 +444,8 @@ const PUBLIC_HTML_LINK_PATTERN =
  /https?:\/\/[^\s<>"')\]]+\/MindSpace\/([0-9a-f-]{36}|[a-z0-9._-]+)\/([^\s<>"')\]]+\.html)/gi;
 const PUBLIC_HTML_MARKDOWN_LINK_PATTERN =
  /\[([^\]\n]*)\]\((https?:\/\/[^\s<>"')\]]+\/MindSpace\/([0-9a-f-]{36}|[a-z0-9._-]+)\/([^\s<>"')\]]+\.html))\)/gi;
+const MARKDOWN_LINK_PATTERN = /\[([^\]\n]*)\]\((https?:\/\/[^\s<>"')\]]+)\)/g;
+const PLAIN_HTML_URL_PATTERN = /https?:\/\/[^\s<>"')\]]+\.html/gi;
 const USER_IDENTITY_BLOCK_PATTERN = /^\[用户身份\][\s\S]*?(?:\n{2,}|$)/;
 const IMAGE_URL_LINES_PATTERN = /\n*\[图片\d+]: [^\n]+/g;
 const TKMIND_VISION_NOTE_PATTERN = /\n*【TKMind 图片分析结果[\s\S]*$/;
@@ -515,6 +517,36 @@ function sanitizeOwnPublicHtmlUrl(publicUrl, owner, rawRelativePath, currentUser
  };
 }

+function isCurrentUserPublicOwner(owner, currentUser) {
+  const normalizedOwner = String(owner ?? '').trim().toLowerCase();
+  const normalizedUserId = String(currentUser?.id ?? '').trim().toLowerCase();
+  const normalizedUsername = String(currentUser?.username ?? '').trim().toLowerCase();
+  return Boolean(normalizedOwner && (
+    normalizedOwner === normalizedUserId ||
+    normalizedOwner === normalizedUsername
+  ));
+}
+
+function sanitizeMissingMindSpacePublicHtmlUrl(url, currentUser) {
+  let parsed;
+  try {
+    parsed = new URL(url);
+  } catch {
+    return null;
+  }
+  const parts = parsed.pathname
+    .split('/')
+    .filter(Boolean)
+    .map((part) => decodePathSegment(part));
+  if (parts[0]?.toLowerCase() === 'mindspace') return null;
+  const owner = parts[0];
+  if (!isCurrentUserPublicOwner(owner, currentUser)) return null;
+  const rawRelativePath = parts.slice(1).join('/');
+  if (!rawRelativePath.toLowerCase().endsWith('.html')) return null;
+  const result = sanitizeOwnPublicHtmlUrl(url, owner, rawRelativePath, currentUser);
+  return result.ok ? result.url : result.notice;
+}
+
 export function sanitizePublicHtmlLinksInText(text, currentUser) {
  let next = String(text ?? '').replace(
    PUBLIC_HTML_MARKDOWN_LINK_PATTERN,
@@ -524,10 +556,18 @@ export function sanitizePublicHtmlLinksInText(text, currentUser) {
      return label ? `[${label}](${result.url})` : result.url;
    },
  );
-  return next.replace(PUBLIC_HTML_LINK_PATTERN, (match, owner, rawRelativePath) => {
+  next = next.replace(PUBLIC_HTML_LINK_PATTERN, (match, owner, rawRelativePath) => {
    const result = sanitizeOwnPublicHtmlUrl(match, owner, rawRelativePath, currentUser);
    return result.ok ? result.url : result.notice;
  });
+  next = next.replace(MARKDOWN_LINK_PATTERN, (match, label, url) => {
+    const sanitizedUrl = sanitizeMissingMindSpacePublicHtmlUrl(url, currentUser);
+    if (!sanitizedUrl) return match;
+    return label ? `[${label}](${sanitizedUrl})` : sanitizedUrl;
+  });
+  return next.replace(PLAIN_HTML_URL_PATTERN, (match) => (
+    sanitizeMissingMindSpacePublicHtmlUrl(match, currentUser) ?? match
+  ));
 }

 export function sanitizeUserVisibleMessageText(text, currentUser) {
@@ -216,6 +216,37 @@ test('sanitizePublicHtmlLinksInText canonicalizes wrong MindSpace public hosts',
  }
 });

+test('sanitizePublicHtmlLinksInText canonicalizes public html links missing MindSpace segment', () => {
+  const owner = `test-user-${Date.now()}-missing-mindspace`;
+  const previousBase = process.env.H5_PUBLIC_BASE_URL;
+  process.env.H5_PUBLIC_BASE_URL = 'https://m.tkmind.cn';
+  const htmlPath = path.join(process.cwd(), 'MindSpace', owner, 'public', 'kuanting-plan.html');
+  try {
+    fs.mkdirSync(path.dirname(htmlPath), { recursive: true });
+    fs.writeFileSync(htmlPath, '<!doctype html><title>宽庭</title>');
+    const text =
+      `[宽庭方案](https://mindspace.tkmind.com/${owner}/public/kuanting-plan.html)`;
+    const next = sanitizePublicHtmlLinksInText(text, { id: owner, username: 'john' });
+    assert.doesNotMatch(next, /mindspace\.tkmind\.com/);
+    assert.match(
+      next,
+      new RegExp(`https://m\\.tkmind\\.cn/MindSpace/${owner}/public/kuanting-plan\\.html`),
+    );
+    assert.doesNotMatch(next, /页面生成未完成/);
+  } finally {
+    if (previousBase == null) delete process.env.H5_PUBLIC_BASE_URL;
+    else process.env.H5_PUBLIC_BASE_URL = previousBase;
+    fs.rmSync(path.join(process.cwd(), 'MindSpace', owner), { recursive: true, force: true });
+  }
+});
+
+test('sanitizePublicHtmlLinksInText leaves other users missing-MindSpace links unchanged', () => {
+  const owner = `test-user-${Date.now()}-other-missing-mindspace`;
+  const text = `[外部页面](https://mindspace.tkmind.com/${owner}/public/kuanting-plan.html)`;
+  const next = sanitizePublicHtmlLinksInText(text, { id: `current-${owner}`, username: 'john' });
+  assert.equal(next, text);
+});
+
 test('sanitizeSessionConversationPublicHtmlLinks updates text items inside conversation messages', () => {
  const owner = `test-user-${Date.now()}-conversation`;
  const publicRoot = path.join(process.cwd(), 'MindSpace', owner, 'public');