memind/page-data-workspace-bind.mjs

import fs from 'node:fs/promises';
import path from 'node:path';
import { createPageService, normalizeWorkspaceRelativePath } from './mindspace-pages.mjs';
import { createPublicationService, publicationInternals } from './mindspace-publications.mjs';
import { slugFromPageTitle } from './mindspace-chat-plaza.mjs';
import { writePageAccessPolicy } from './page-data-policy-store.mjs';
import { normalizePageAccessPolicy } from './page-access-policy.mjs';
import { upsertPageDataPolicyIndex } from './page-data-policy-index.mjs';
import { syncPageDataPolicyAccessMode } from './page-data-publish-sync.mjs';
import { createMindSpacePublicUrl } from './mindspace-canonical-url.mjs';
import { resolvePageDataDeliveryBaseUrl } from './user-publish.mjs';
import { createUserDataSpaceService } from './user-data-space-service.mjs';
import {
  assertPolicyMatchesHtmlDatasets,
  buildPageDataPolicyDatasetsFromRegistry,
  detectPageDataDatasetUsageFromHtml,
} from './page-data-html-detect.mjs';

export const DEFAULT_PAGE_DATA_ADMIN_PASSWORD = '88888888';

export function resolvePageDataBindAccess(accessMode = 'public') {
  return publicationInternals.normalizeAccessMode(accessMode);
}

/** password 模式：用户未传口令时使用 DEFAULT_PAGE_DATA_ADMIN_PASSWORD；<8 位抛错 */
export function resolvePageDataBindPassword(accessMode, password = null) {
  const mode = resolvePageDataBindAccess(accessMode);
  if (mode !== 'password') return null;
  const explicit = String(password ?? '').trim();
  if (explicit) {
    return publicationInternals.normalizePassword(explicit, true);
  }
  return DEFAULT_PAGE_DATA_ADMIN_PASSWORD;
}

function titleFromPublicHtml(content, relativePath) {
  const match = String(content ?? '').match(/<title[^>]*>([^<]+)<\/title>/i);
  const fromTitle = match?.[1]?.replace(/\s+/g, ' ').trim();
  if (fromTitle) return fromTitle;
  const basename = path.basename(String(relativePath ?? ''), '.html');
  return basename || 'MindSpace 页面';
}

function normalizePublicHtmlRelativePath(relativePath) {
  const normalized = normalizeWorkspaceRelativePath(relativePath);
  if (!normalized || !normalized.startsWith('public/') || !normalized.toLowerCase().endsWith('.html')) {
    throw Object.assign(new Error('仅支持 workspace 公开 HTML 页面（public/*.html）'), {
      code: 'invalid_public_html_path',
    });
  }
  return normalized;
}

async function ensureWorkspaceHtmlPage({ userId, relativePath, content, title, mindSpacePages }) {
  const normalized = normalizePublicHtmlRelativePath(relativePath);
  let page = await mindSpacePages.findPageByRelativePath(userId, normalized);
  const pageInput = {
    title: title ?? page?.title ?? titleFromPublicHtml(content, normalized),
    content,
    contentFormat: 'html',
    pageType: 'html',
    categoryCode: 'draft',
  };
  if (page) {
    return mindSpacePages.updatePage(
      userId,
      page.id,
      {
        ...pageInput,
        expectedVersion: page.versionNo,
      },
      {
        snapshot: {
          content_mode: 'static_html',
          relative_path: normalized,
        },
      },
    );
  }
  return mindSpacePages.createFromChat(userId, pageInput, {
    snapshot: {
      content_mode: 'static_html',
      relative_path: normalized,
    },
  });
}

async function ensureWorkspaceHtmlPublished({
  userId,
  page,
  mindSpacePublications,
  accessMode,
  password,
  urlSlug,
}) {
  const normalizedAccessMode = resolvePageDataBindAccess(accessMode);
  const resolvedPassword = resolvePageDataBindPassword(normalizedAccessMode, password);
  const existing = await mindSpacePublications.getCurrent(userId, page.id);
  if (existing?.id) {
    const accessChanged = existing.accessMode !== normalizedAccessMode;
    const needsPassword =
      normalizedAccessMode === 'password' && Boolean(resolvedPassword);
    if (accessChanged || needsPassword) {
      const updated = await mindSpacePublications.updatePublicationAccess(userId, existing.id, {
        accessMode: normalizedAccessMode,
        password: resolvedPassword,
      });
      existing.accessMode = updated.accessMode;
      existing.id = updated.id;
      existing.publicUrl = updated.publicUrl;
    }
    if (page.pageType === 'html' && mindSpacePublications.refreshOnlinePublicationHtml) {
      return (
        (await mindSpacePublications.refreshOnlinePublicationHtml(userId, page.id)) ?? existing
      );
    }
    return existing;
  }
  const preferredSlug = urlSlug || slugFromPageTitle(page.title, page.id);
  return mindSpacePublications.publish(userId, page.id, {
    pageVersionId: page.currentVersionId,
    accessMode: normalizedAccessMode,
    password: resolvedPassword,
    urlSlug: preferredSlug,
    autoAcknowledgeFindings: true,
  });
}

function buildWorkspacePublicUrl(userId, relativePath) {
  const publicBaseUrl = resolvePageDataDeliveryBaseUrl();
  return createMindSpacePublicUrl({
    publicBaseUrl,
    ownerKey: userId,
    relativePath,
  });
}

function assertRegisteredDatasetTables(userDataSpace, registryDatasets, usage) {
  for (const dataset of registryDatasets) {
    const requiredUsage = usage.get(dataset.name) ?? {};
    if (requiredUsage.insert && !dataset.actions.includes('insert')) {
      throw Object.assign(new Error(`dataset「${dataset.name}」未开放写入`), {
        code: 'dataset_action_not_registered',
        datasetName: dataset.name,
        action: 'insert',
      });
    }
    if (requiredUsage.read && !dataset.actions.includes('read')) {
      throw Object.assign(new Error(`dataset「${dataset.name}」未开放读取`), {
        code: 'dataset_action_not_registered',
        datasetName: dataset.name,
        action: 'read',
      });
    }
    const actualColumns = userDataSpace.listTableColumns(dataset.table);
    if (!actualColumns.length) {
      throw Object.assign(new Error(`dataset 对应表不存在：${dataset.table}`), {
        code: 'table_not_found',
        datasetName: dataset.name,
      });
    }
    const actualNames = new Set(actualColumns.map((column) => column.name));
    const configuredColumns = Object.values(dataset.columns ?? {}).flat();
    const missingColumns = [...new Set(configuredColumns)].filter((column) => !actualNames.has(column));
    if (missingColumns.length) {
      throw Object.assign(
        new Error(`dataset「${dataset.name}」注册字段不存在：${missingColumns.join(', ')}`),
        { code: 'dataset_schema_mismatch', datasetName: dataset.name, missingColumns },
      );
    }
  }
}

async function assertRegisteredDatasetTablesAsync(userDataSpace, registryDatasets, usage) {
  for (const dataset of registryDatasets) {
    const adapter = {
      listTableColumns: () => userDataSpace.listTableColumns(dataset.table),
    };
    const columns = await adapter.listTableColumns();
    assertRegisteredDatasetTables({ listTableColumns: () => columns }, [dataset], usage);
  }
}

/**
 * Page Data policy must be derived from a registered SQLite dataset, never
 * accepted solely from an Agent-supplied policy. This runs before page or
 * publication creation so an incomplete data setup cannot leave a live page.
 */
export function resolveRegisteredPageDataPolicy({
  html,
  userId,
  accessMode,
  pageDataPolicy = null,
  userDataSpace,
} = {}) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  if (!usage.size) return null;
  if (!userDataSpace) throw new Error('缺少 Page Data 数据空间');

  if (pageDataPolicy?.datasets) {
    assertPolicyMatchesHtmlDatasets(html, pageDataPolicy.datasets);
  }

  const registryDatasets = userDataSpace.listDatasets();
  const datasets = buildPageDataPolicyDatasetsFromRegistry({
    html,
    registryDatasets,
    usage,
  });
  assertRegisteredDatasetTables(
    userDataSpace,
    registryDatasets.filter((dataset) => datasets[dataset.name]),
    usage,
  );

  return {
    ...pageDataPolicy,
    ownerUserId: String(pageDataPolicy?.ownerUserId ?? userId).trim(),
    accessMode: pageDataPolicy?.accessMode ?? accessMode,
    // The registry is authoritative for actions and allowed columns. Agent
    // input can describe the page, but cannot manufacture a dataset policy.
    datasets,
  };
}

export async function resolveRegisteredPageDataPolicyAsync(options = {}) {
  const { html, userId, accessMode, pageDataPolicy = null, userDataSpace } = options;
  const usage = detectPageDataDatasetUsageFromHtml(html);
  if (!usage.size) return null;
  if (!userDataSpace) throw new Error('缺少 Page Data 数据空间');
  if (pageDataPolicy?.datasets) assertPolicyMatchesHtmlDatasets(html, pageDataPolicy.datasets);
  const registryDatasets = await userDataSpace.listDatasets();
  const datasets = buildPageDataPolicyDatasetsFromRegistry({ html, registryDatasets, usage });
  await assertRegisteredDatasetTablesAsync(
    userDataSpace,
    registryDatasets.filter((dataset) => datasets[dataset.name]),
    usage,
  );
  return {
    ...pageDataPolicy,
    ownerUserId: String(pageDataPolicy?.ownerUserId ?? userId).trim(),
    accessMode: pageDataPolicy?.accessMode ?? accessMode,
    datasets,
  };
}

export async function bindWorkspaceHtmlForPageData({
  pool,
  h5Root,
  storageRoot,
  userId,
  workspaceRoot,
  relativePath,
  title = null,
  accessMode = 'public',
  password = null,
  pageDataPolicy = null,
  urlSlug = null,
  publicPageLimit,
}) {
  if (!pool) throw new Error('数据库未配置，无法绑定页面');
  if (!userId) throw new Error('缺少 userId');
  if (!workspaceRoot) throw new Error('缺少 workspaceRoot');
  const normalized = normalizePublicHtmlRelativePath(relativePath);
  const absoluteHtmlPath = path.join(workspaceRoot, ...normalized.split('/'));
  const content = await fs.readFile(absoluteHtmlPath, 'utf8');
  const normalizedAccessMode = resolvePageDataBindAccess(accessMode);
  const resolvedPassword = resolvePageDataBindPassword(normalizedAccessMode, password);

  const userDataSpace = createUserDataSpaceService({ workspaceRoot, userId, query: pool });
  const resolvedPageDataPolicy = await resolveRegisteredPageDataPolicyAsync({
    html: content,
    userId,
    accessMode: normalizedAccessMode,
    pageDataPolicy,
    userDataSpace,
  });

  const mindSpacePages = createPageService(pool, { h5Root, storageRoot });
  const mindSpacePublications = createPublicationService(pool, {
    h5Root,
    storageRoot,
    publicPageLimit,
  });

  const page = await ensureWorkspaceHtmlPage({
    userId,
    relativePath: normalized,
    content,
    title,
    mindSpacePages,
  });

  const publication = await ensureWorkspaceHtmlPublished({
    userId,
    page,
    mindSpacePublications,
    accessMode: normalizedAccessMode,
    password: resolvedPassword,
    urlSlug,
  });

  let policy = null;

  if (resolvedPageDataPolicy) {
    const ownerUserId = String(resolvedPageDataPolicy.ownerUserId ?? userId).trim();
    policy = normalizePageAccessPolicy(
      {
        pageId: page.id,
        ownerUserId,
        accessMode: resolvedPageDataPolicy.accessMode ?? normalizedAccessMode,
        datasets: resolvedPageDataPolicy.datasets,
        defaultVisitorRole: resolvedPageDataPolicy.defaultVisitorRole,
        visitors: resolvedPageDataPolicy.visitors,
        roles: resolvedPageDataPolicy.roles,
      },
      { fallbackPageId: page.id, fallbackOwnerUserId: ownerUserId },
    );
    policy = writePageAccessPolicy(workspaceRoot, policy);
    await upsertPageDataPolicyIndex(pool, policy).catch(() => null);
    policy =
      syncPageDataPolicyAccessMode(workspaceRoot, page.id, publication.accessMode, ownerUserId) ??
      policy;
  }

  const workspaceUrl = buildWorkspacePublicUrl(userId, normalized);
  return {
    pageId: page.id,
    pageTitle: page.title,
    relativePath: normalized,
    publicationId: publication.id,
    publicationAccessMode: publication.accessMode,
    publicationPasswordApplied: normalizedAccessMode === 'password',
    publicationUrl: workspaceUrl,
    workspaceUrl,
    deliveryUrl: workspaceUrl,
    deliveryHint:
      '向用户交付静态页时优先使用 deliveryUrl（MindSpace /MindSpace/<用户ID>/public/... 路径），publicationUrl 仅作补充',
    policy,
  };
}