memind/page-data-html-detect.mjs

/**
 * 从 MindSpace 公开 HTML 中解析 Page Data API 引用的 dataset 与读写意图。
 */

const PAGE_DATA_CLIENT_SCRIPT_PATTERN = /\/assets\/page-data-client\.js/i;
const FORBIDDEN_LEGACY_PAGE_DATA_API_PATTERN = /\/api\/page-data\b/i;

export function htmlUsesForbiddenLegacyPageDataApi(html) {
  return FORBIDDEN_LEGACY_PAGE_DATA_API_PATTERN.test(String(html ?? ''));
}

export function htmlUsesPageDataApi(html) {
  const content = String(html ?? '');
  const usage = detectPageDataDatasetUsageFromHtml(content);
  return (
    usage.size > 0 ||
    PAGE_DATA_CLIENT_SCRIPT_PATTERN.test(content) ||
    /\bMindSpacePageData\b/.test(content) ||
    htmlUsesForbiddenLegacyPageDataApi(content)
  );
}

export function detectPageDataDatasetUsageFromHtml(html) {
  const text = String(html ?? '');
  const datasets = new Map();
  const constants = new Map();

  for (const match of text.matchAll(/(?:const|let|var)\s+([A-Za-z_$][\w$]*)\s*=\s*['"]([^'"]+)['"]/g)) {
    constants.set(match[1], match[2]);
  }

  function remember(name, patch) {
    const key = String(name ?? '').trim();
    if (!key) return;
    datasets.set(key, { ...(datasets.get(key) ?? {}), ...patch });
  }

  for (const match of text.matchAll(/\.insertRow\(\s*['"]([^'"]+)['"]/g)) {
    remember(match[1], { insert: true });
  }
  for (const match of text.matchAll(/\.listRows\(\s*['"]([^'"]+)['"]/g)) {
    remember(match[1], { read: true });
  }
  for (const match of text.matchAll(/\.updateRow\(\s*['"]([^'"]+)['"]/g)) {
    remember(match[1], { update: true });
  }
  for (const match of text.matchAll(/\.deleteRow\(\s*['"]([^'"]+)['"]/g)) {
    // The browser client's deleteRow endpoint intentionally falls back to a
    // soft delete unless a policy explicitly grants hard_delete. A page using
    // this API must therefore only require the safe, default capability.
    remember(match[1], { softDelete: true });
  }
  // `${...}` inside an HTML template literal is a row-id interpolation, not a
  // JavaScript identifier passed as the dataset argument. Without this guard,
  // snippets such as `onclick="deleteRow(${row.id})"` are detected as a `$`
  // dataset and can make the whole workspace binding pass fail.
  for (const match of text.matchAll(/\.insertRow\(\s*(?!\$\{)([A-Za-z_$][\w$]*)/g)) {
    remember(constants.get(match[1]) ?? match[1], { insert: true });
  }
  for (const match of text.matchAll(/\.listRows\(\s*(?!\$\{)([A-Za-z_$][\w$]*)/g)) {
    remember(constants.get(match[1]) ?? match[1], { read: true });
  }
  for (const match of text.matchAll(/\.updateRow\(\s*(?!\$\{)([A-Za-z_$][\w$]*)/g)) {
    remember(constants.get(match[1]) ?? match[1], { update: true });
  }
  for (const match of text.matchAll(/\.deleteRow\(\s*(?!\$\{)([A-Za-z_$][\w$]*)/g)) {
    remember(constants.get(match[1]) ?? match[1], { softDelete: true });
  }

  return datasets;
}

export function inferPageDataBindAccessMode(relativePath, html) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  const hasRead = [...usage.values()].some((item) => item.read);
  const hasInsert = [...usage.values()].some((item) => item.insert);
  // A page that exchanges a password for a Page Data token is intentionally
  // protected even when it both reads and writes its dataset (for example, a
  // personal tracker). Do not silently re-bind it as public merely because it
  // is not named "-admin.html".
  const usesServerAuthentication = /\.\s*authenticate\s*\(/.test(String(html ?? ''));
  if (
    /-admin\.html$/i.test(String(relativePath ?? '')) ||
    (hasRead && !hasInsert) ||
    usesServerAuthentication
  ) {
    return 'password';
  }
  return 'public';
}

export function assertPolicyMatchesHtmlDatasets(html, policyDatasets) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  const htmlNames = [...usage.keys()].sort();
  const policyNames = Object.keys(policyDatasets ?? {}).sort();
  if (!htmlNames.length) return usage;
  if (htmlNames.join(',') !== policyNames.join(',')) {
    throw Object.assign(
      new Error(
        `Page Data 策略 dataset（${policyNames.join(', ') || '无'}）与 HTML 中 insertRow/listRows/updateRow/deleteRow 引用的 dataset（${htmlNames.join(', ')}）不一致。请使用与页面脚本相同的 dataset 名称。`,
      ),
      {
        code: 'dataset_policy_html_mismatch',
        htmlDatasets: htmlNames,
        policyDatasets: policyNames,
      },
    );
  }
  return usage;
}

export function buildPageDataPolicyDatasetsFromRegistry({ html, registryDatasets, usage }) {
  const detected = usage ?? detectPageDataDatasetUsageFromHtml(html);
  if (!detected.size) return null;

  const registryMap = new Map(registryDatasets.map((dataset) => [dataset.name, dataset]));
  const datasets = {};

  for (const [name, perms] of detected) {
    const registered = registryMap.get(name);
    if (!registered) {
      throw Object.assign(
        new Error(`HTML 引用的 dataset「${name}」尚未注册，请先执行 private_data_register_dataset`),
        { code: 'dataset_not_registered', datasetName: name },
      );
    }

    const entry = {
      insert: false,
      read: false,
      update: false,
      softDelete: false,
      hardDelete: false,
      columns: {},
    };

    if (perms.insert && !perms.read) {
      entry.insert = true;
      entry.columns.insert = registered.columns?.insert ?? [];
    } else if (perms.read && !perms.insert) {
      entry.read = true;
      entry.columns.read = registered.columns?.read ?? [];
    } else {
      if (perms.insert) {
        entry.insert = true;
        entry.columns.insert = registered.columns?.insert ?? [];
      }
      if (perms.read) {
        entry.read = true;
        entry.columns.read = registered.columns?.read ?? [];
      }
    }
    if (perms.softDelete) {
      entry.softDelete = true;
      entry.columns.soft_delete = registered.columns?.soft_delete ?? ['id'];
    }
    if (perms.update) {
      entry.update = true;
      entry.columns.update = registered.columns?.update ?? [];
    }

    datasets[name] = entry;
  }

  return datasets;
}