memind/mindspace-page-sync.mjs

import fs from 'node:fs/promises';
import path from 'node:path';

const PUBLIC_HTML_SKIP_DIRS = new Set([
  '.tmp-images',
  'assets',
  'images',
  'shared',
  'node_modules',
]);

function normalizePublicHtmlPath(relativePath) {
  const parts = String(relativePath ?? '')
    .replace(/^\/+/, '')
    .split('/')
    .filter((part) => part && part !== '.' && part !== '..');
  if (parts.length === 0) return '';
  if (parts[0].toLowerCase() === 'public') return parts.join('/');
  if (parts.length === 1 && parts[0].toLowerCase().endsWith('.html')) return `public/${parts[0]}`;
  return parts.join('/');
}

function titleFromRelativePath(relativePath) {
  const basename = path.basename(String(relativePath ?? ''), '.html');
  return basename.replace(/[-_]+/g, ' ').trim() || '生成的页面';
}

function extractHtmlTitle(content) {
  const match = String(content).match(/<title[^>]*>([^<]+)<\/title>/i);
  return match?.[1]?.trim() ?? '';
}

function extractHtmlSummary(content) {
  const match = String(content).match(/<meta[^>]+name=["']description["'][^>]+content=["']([^"']+)["']/i);
  return match?.[1]?.trim().slice(0, 1000) ?? '';
}

async function findPageByWorkspaceRelativePath(pool, userId, relativePath) {
  const [rows] = await pool.query(
    `SELECT p.id, p.updated_at
     FROM h5_page_records p
     JOIN h5_page_versions pv ON pv.id = p.current_version_id
     WHERE p.user_id = ?
       AND p.status <> 'deleted'
       AND JSON_UNQUOTE(JSON_EXTRACT(pv.source_snapshot_json, '$.relative_path')) = ?
     ORDER BY p.updated_at DESC, p.id DESC
     LIMIT 1`,
    [userId, relativePath],
  );
  const row = rows[0];
  if (!row) return null;
  return { id: row.id, updatedAt: Number(row.updated_at ?? 0) };
}

function parseJsonColumn(value, fallback = {}) {
  if (value == null || value === '') return { ...fallback };
  if (typeof value === 'object' && !Buffer.isBuffer(value)) return value;
  try {
    return JSON.parse(String(value));
  } catch {
    return { ...fallback };
  }
}

async function loadIndexedWorkspacePages(pool, userId) {
  const [rows] = await pool.query(
    `SELECT p.id, p.updated_at, p.source_asset_id, pv.source_snapshot_json
     FROM h5_page_records p
     JOIN h5_page_versions pv ON pv.id = p.current_version_id
     WHERE p.user_id = ? AND p.status <> 'deleted'`,
    [userId],
  );
  const byPath = new Map();
  for (const row of rows) {
    const snapshot = parseJsonColumn(row.source_snapshot_json);
    const relativePath = normalizePublicHtmlPath(snapshot.relative_path);
    if (relativePath) {
      byPath.set(relativePath, {
        id: row.id,
        updatedAt: Number(row.updated_at ?? 0),
      });
    }
    if (row.source_asset_id && relativePath) {
      byPath.set(`asset:${row.source_asset_id}`, { id: row.id, updatedAt: Number(row.updated_at ?? 0) });
    }
  }
  return byPath;
}

async function listPublishPublicHtmlFiles(publishDir) {
  const publicDir = path.join(publishDir, 'public');
  const files = [];

  async function walk(absDir, relWithinPublic) {
    let entries;
    try {
      entries = await fs.readdir(absDir, { withFileTypes: true });
    } catch {
      return;
    }
    for (const entry of entries) {
      if (entry.name.startsWith('.') || PUBLIC_HTML_SKIP_DIRS.has(entry.name)) continue;
      const absPath = path.join(absDir, entry.name);
      const relPath = relWithinPublic ? `${relWithinPublic}/${entry.name}` : entry.name;
      if (entry.isDirectory()) {
        await walk(absPath, relPath);
        continue;
      }
      if (!entry.isFile() || !entry.name.toLowerCase().endsWith('.html')) continue;
      const stat = await fs.stat(absPath);
      files.push({
        relativePath: normalizePublicHtmlPath(`public/${relPath}`),
        absolutePath: absPath,
        mtimeMs: stat.mtimeMs,
      });
    }
  }

  await walk(publicDir, '');
  return files;
}

async function upsertWorkspaceHtmlPage({
  pool,
  pageService,
  userId,
  indexedPages,
  relativePath,
  content,
  assetId = null,
  sourceMtimeMs = null,
}) {
  const title = extractHtmlTitle(content) || titleFromRelativePath(relativePath);
  const summary = extractHtmlSummary(content);
  let existing =
    indexedPages.get(relativePath) ??
    (assetId ? indexedPages.get(`asset:${assetId}`) : null);

  if (!existing && pool) {
    existing = await findPageByWorkspaceRelativePath(pool, userId, relativePath);
    if (existing) {
      indexedPages.set(relativePath, existing);
      if (assetId) indexedPages.set(`asset:${assetId}`, existing);
    }
  }

  const pageInput = {
    title,
    summary,
    content,
    contentFormat: 'html',
    pageType: 'html',
    templateId: 'editorial',
    categoryCode: 'draft',
  };
  const snapshot = {
    ...(assetId ? { source_asset_id: assetId } : {}),
    source_category: 'public',
    content_mode: 'static_html',
    relative_path: relativePath,
    auto_synced: true,
  };

  if (existing) {
    if (sourceMtimeMs != null && existing.updatedAt >= sourceMtimeMs) {
      return 'skipped';
    }
    await pageService.updatePage(userId, existing.id, {
      ...pageInput,
      changeNote: '同步工作区页面更新',
    });
    indexedPages.set(relativePath, { id: existing.id, updatedAt: Date.now() });
    return 'updated';
  }

  const page = await pageService.createFromChat(userId, pageInput, {
    ...(assetId ? { assetId } : {}),
    snapshot,
    createdAt:
      sourceMtimeMs != null && Number.isFinite(Number(sourceMtimeMs))
        ? Math.floor(Number(sourceMtimeMs))
        : undefined,
  });
  indexedPages.set(relativePath, { id: page.id, updatedAt: Date.now() });
  if (assetId) indexedPages.set(`asset:${assetId}`, { id: page.id, updatedAt: Date.now() });
  return 'created';
}

export async function syncGeneratedPagesFromPublicAssets({
  pool,
  pageService,
  assetService,
  userId,
  publishDir = null,
  syncWorkspaceAssets = null,
} = {}) {
  if (!pool || !pageService || !userId) {
    return { created: 0, skipped: 0, updated: 0 };
  }

  if (typeof syncWorkspaceAssets === 'function') {
    await syncWorkspaceAssets(userId, { categoryCode: 'public' }).catch(() => {});
  }

  const indexedPages = await loadIndexedWorkspacePages(pool, userId);
  let created = 0;
  let updated = 0;
  let skipped = 0;

  if (publishDir) {
    const htmlFiles = await listPublishPublicHtmlFiles(publishDir);
    for (const file of htmlFiles) {
      try {
        const content = await fs.readFile(file.absolutePath, 'utf8');
        if (!content.trim()) {
          skipped += 1;
          continue;
        }
        const result = await upsertWorkspaceHtmlPage({
          pool,
          pageService,
          userId,
          indexedPages,
          relativePath: file.relativePath,
          content,
          sourceMtimeMs: file.mtimeMs,
        });
        if (result === 'created') created += 1;
        else if (result === 'updated') updated += 1;
        else skipped += 1;
      } catch (error) {
        console.warn(
          `[MindSpace] page sync failed for ${file.relativePath}:`,
          error?.message ?? error,
        );
        skipped += 1;
      }
    }
  }

  if (!assetService) {
    return { created, updated, skipped };
  }

  const [rows] = await pool.query(
    `SELECT a.id, a.display_name, a.original_filename, a.updated_at
     FROM h5_assets a
     JOIN h5_space_categories c ON c.id = a.category_id AND c.user_id = a.user_id
     WHERE a.user_id = ?
       AND c.category_code = 'public'
       AND a.mime_type = 'text/html'
       AND a.status = 'ready'
     ORDER BY a.updated_at DESC
     LIMIT 200`,
    [userId],
  );

  for (const asset of rows) {
    const relativePath = normalizePublicHtmlPath(
      String(asset.original_filename ?? '').includes('/')
        ? asset.original_filename
        : `public/${asset.original_filename}`,
    );
    if (indexedPages.has(relativePath)) {
      skipped += 1;
      continue;
    }
    try {
      const { path: assetPath } = await assetService.readAsset(userId, asset.id);
      const content = await fs.readFile(assetPath, 'utf8');
      const result = await upsertWorkspaceHtmlPage({
        pool,
        pageService,
        userId,
        indexedPages,
        relativePath,
        content,
        assetId: asset.id,
        sourceMtimeMs: Number(asset.updated_at ?? 0),
      });
      if (result === 'created') created += 1;
      else if (result === 'updated') updated += 1;
      else skipped += 1;
    } catch (error) {
      console.warn(
        `[MindSpace] asset page sync failed for ${asset.original_filename}:`,
        error?.message ?? error,
      );
      skipped += 1;
    }
  }

  return { created, updated, skipped };
}