memind/page-data-delivery-assess.mjs

import fs from 'node:fs/promises';
import {
  detectPageDataDatasetUsageFromHtml,
  htmlUsesForbiddenLegacyPageDataApi,
  inferPageDataBindAccessMode,
} from './page-data-html-detect.mjs';
import { listPageAccessPolicies, readPageAccessPolicy } from './page-data-policy-store.mjs';
import { createUserDataSpaceService } from './user-data-space-service.mjs';

const INJECTION_PAGE_ID_PATTERN =
  /window\.__MINDSPACE_PAGE_DATA__\s*=\s*\{[^}]*(?:"pageId"|pageId)\s*:\s*"([^"]+)"/i;
const INJECTION_META_PAGE_ID_PATTERN =
  /<meta[^>]+name=["']mindspace-page-data-page-id["'][^>]+content=["']([^"']+)["']/i;

export function normalizePageDataApiBase(apiBase) {
  const raw = String(apiBase ?? '').trim().replace(/\/$/, '');
  if (!raw) return '/api';
  if (/\/api$/i.test(raw)) return raw;
  return `${raw}/api`;
}

export function extractInjectedPageIdFromHtml(html) {
  const content = String(html ?? '');
  const fromScript = content.match(INJECTION_PAGE_ID_PATTERN)?.[1]?.trim();
  if (fromScript) return fromScript;
  return content.match(INJECTION_META_PAGE_ID_PATTERN)?.[1]?.trim() ?? null;
}

export function buildSmokeInsertRow(policy, datasetName) {
  const columns = policy?.datasets?.[datasetName]?.columns?.insert ?? [];
  const row = {};
  for (const column of columns) {
    if (/rating|score|星级|评分/i.test(column)) {
      row[column] = 3;
    } else if (/phone|手机|电话/i.test(column)) {
      row[column] = '13800000000';
    } else {
      row[column] = 'smoke-test';
    }
  }
  return row;
}

export function assessPageDataAuthenticationContract({ policy, html } = {}) {
  const accessMode = String(policy?.accessMode ?? '').trim();
  const content = String(html ?? '');
  const usesServerAuthentication = /\.\s*authenticate\s*\(/.test(content);
  const passwordValue = String.raw`\b(?:pwd|pass(?:word)?|passwordInput|adminPassword)\w*(?:\.value)?`;
  const passwordComparison = String.raw`(?:===?|!==?)`;
  const passwordOperand = String.raw`(?:[A-Za-z_$][\w$]*|['"\x60][^'"\x60]*['"\x60])`;
  const usesClientSidePasswordCheck = new RegExp(
    `(?:${passwordValue}\\s*${passwordComparison}\\s*${passwordOperand}|${passwordOperand}\\s*${passwordComparison}\\s*${passwordValue})`,
    'i',
  ).test(content);
  const reasons = [];
  if (accessMode === 'password' && !usesServerAuthentication) {
    reasons.push('password_page_missing_server_authentication');
  }
  if (accessMode === 'public' && usesServerAuthentication) {
    reasons.push('public_page_cannot_use_password_authentication');
  }
  if (accessMode === 'public' && usesClientSidePasswordCheck) {
    reasons.push('public_page_uses_client_side_password_gate');
  }
  if (accessMode === 'password' && usesClientSidePasswordCheck) {
    reasons.push('password_page_uses_client_side_password_gate');
  }
  return reasons;
}

export async function queryOnlinePublication(pool, pageId) {
  if (!pool || !pageId) return null;
  const [rows] = await pool.query(
    `SELECT id, page_id, status, access_mode
     FROM h5_publish_records
     WHERE page_id = ? AND status = 'online'
     ORDER BY published_at DESC
     LIMIT 1`,
    [pageId],
  );
  return rows[0] ?? null;
}

function findWorkspacePolicyForHtml({ publishDir, relativePath, html }) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  const htmlDatasetNames = [...usage.keys()].sort().join(',');
  const expectedAccessMode = inferPageDataBindAccessMode(relativePath, html);
  for (const policy of listPageAccessPolicies(publishDir)) {
    const policyNames = Object.keys(policy?.datasets ?? {}).sort().join(',');
    if (policyNames !== htmlDatasetNames) continue;
    if (String(policy.accessMode ?? '').trim() !== expectedAccessMode) continue;
    let matches = true;
    for (const [datasetName, perms] of usage) {
      const policyDataset = policy.datasets?.[datasetName];
      if (!policyDataset) {
        matches = false;
        break;
      }
      if (perms.insert && !policyDataset.insert) matches = false;
      if (perms.read && !policyDataset.read) matches = false;
    }
    if (matches) return policy;
  }
  return null;
}

function assessWorkspacePageDataPolicyReadiness({ publishDir, relativePath, html }) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  if (!usage.size) {
    return { ready: true, reasons: [] };
  }

  const reasons = [];
  const policy = findWorkspacePolicyForHtml({ publishDir, relativePath, html });
  if (!policy) {
    reasons.push('missing_workspace_policy');
  }

  return { ready: reasons.length === 0, reasons, policy };
}

export async function assessWorkspacePageDataReadiness({ userId, publishDir, relativePath, html }) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  const policyAssessment = assessWorkspacePageDataPolicyReadiness({ publishDir, relativePath, html });
  if (!usage.size) return policyAssessment;

  const reasons = [...policyAssessment.reasons];
  const dataSpace = createUserDataSpaceService({ workspaceRoot: publishDir, userId });
  for (const datasetName of usage.keys()) {
    if (!(await dataSpace.getDataset(datasetName))) {
      reasons.push(`dataset_not_registered:${datasetName}`);
    }
  }
  return { ...policyAssessment, ready: reasons.length === 0, reasons };
}

export async function assessPageDataHtmlBinding({
  pool,
  userId,
  publishDir,
  relativePath,
  html,
  findPageByRelativePath,
}) {
  const usage = detectPageDataDatasetUsageFromHtml(html);
  const reasons = [];

  if (htmlUsesForbiddenLegacyPageDataApi(html)) {
    reasons.push('forbidden_legacy_page_data_api');
  }

  if (!usage.size) {
    return {
      bound: reasons.length === 0,
      reasons: [...new Set(reasons)],
      pageId: null,
      policy: null,
    };
  }

  const workspace = await assessWorkspacePageDataReadiness({ userId, publishDir, relativePath, html });
  reasons.push(...workspace.reasons);

  let pageId = null;
  if (typeof findPageByRelativePath === 'function') {
    const page = await findPageByRelativePath(userId, relativePath).catch(() => null);
    pageId = page?.id ? String(page.id).trim() : null;
    if (!pageId) {
      reasons.push('missing_page_record');
    }
  } else {
    reasons.push('missing_page_lookup');
  }

  if (pageId) {
    const policy = readPageAccessPolicy(publishDir, pageId);
    if (!policy) {
      reasons.push('missing_policy_for_page');
    } else {
      const htmlDatasetNames = [...usage.keys()].sort().join(',');
      const policyNames = Object.keys(policy.datasets ?? {}).sort().join(',');
      if (htmlDatasetNames !== policyNames) {
        reasons.push('policy_dataset_mismatch');
      }
      const expectedAccessMode = inferPageDataBindAccessMode(relativePath, html);
      if (String(policy.accessMode ?? '').trim() !== expectedAccessMode) {
        reasons.push('policy_access_mode_mismatch');
      }
      reasons.push(...assessPageDataAuthenticationContract({ policy, html }));
      for (const [datasetName, perms] of usage) {
        const policyDataset = policy.datasets?.[datasetName];
        if (!policyDataset) continue;
        if (perms.insert && !policyDataset.insert) reasons.push('missing_insert_permission');
        if (perms.read && !policyDataset.read) reasons.push('missing_read_permission');
      }
    }

    if (pool && pageId) {
      const publication = await queryOnlinePublication(pool, pageId);
      const expectedAccessMode = inferPageDataBindAccessMode(relativePath, html);
      if (publication && String(publication.access_mode ?? '').trim() !== expectedAccessMode) {
        reasons.push('publication_access_mode_mismatch');
      }
    }
  }

  return {
    bound: reasons.length === 0,
    reasons: [...new Set(reasons)],
    pageId,
    policy: pageId ? readPageAccessPolicy(publishDir, pageId) : workspace.policy ?? null,
  };
}

export async function fetchWorkspaceHtmlInjection({ url, fetchImpl = fetch }) {
  const response = await fetchImpl(String(url));
  if (!response.ok) {
    return {
      ok: false,
      reason: `http_${response.status}`,
      pageId: null,
      html: '',
    };
  }
  const html = await response.text();
  const pageId = extractInjectedPageIdFromHtml(html);
  if (!pageId) {
    return { ok: false, reason: 'missing_page_data_injection', pageId: null, html };
  }
  return { ok: true, reason: null, pageId, html };
}

export async function smokeTestPageDataInsert({
  apiBase,
  pageId,
  dataset,
  policy,
  fetchImpl = fetch,
}) {
  const base = normalizePageDataApiBase(apiBase);
  const row = buildSmokeInsertRow(policy, dataset);
  if (!pageId || !dataset || !Object.keys(row).length) {
    return { ok: false, reason: 'smoke_payload_unavailable', status: 0, body: null };
  }

  const response = await fetchImpl(
    `${base}/public/pages/${encodeURIComponent(pageId)}/data/${encodeURIComponent(dataset)}/rows`,
    {
      method: 'POST',
      headers: { 'content-type': 'application/json', accept: 'application/json' },
      body: JSON.stringify(row),
    },
  );
  const body = await response.json().catch(() => ({}));
  if (!response.ok) {
    return {
      ok: false,
      reason: body?.error?.code ?? 'insert_failed',
      status: response.status,
      body,
    };
  }
  return { ok: true, reason: null, status: response.status, body };
}

export async function verifyPageDataDeliveryArtifacts({
  artifacts = [],
  publishDir,
  apiBase,
  pool,
  userId,
  findPageByRelativePath,
  fetchImpl = fetch,
}) {
  const failures = [];
  for (const artifact of artifacts) {
    const relativePath = artifact.relativePath;
    const html = await fs.readFile(artifact.localPath, 'utf8').catch(() => '');
    const usage = detectPageDataDatasetUsageFromHtml(html);
    const insertDataset = [...usage.entries()].find(([, perms]) => perms.insert)?.[0] ?? null;

    const assessment = await assessPageDataHtmlBinding({
      pool,
      userId,
      publishDir,
      relativePath,
      html,
      findPageByRelativePath,
    });
    if (!assessment.bound) {
      failures.push({
        relativePath,
        stage: 'binding',
        reason: assessment.reasons.join(',') || 'not_bound',
        bindingReasons: assessment.reasons,
      });
      continue;
    }

    if (artifact.isAdmin || !insertDataset) continue;

    const injection = await fetchWorkspaceHtmlInjection({ url: artifact.url, fetchImpl });
    if (!injection.ok) {
      failures.push({ relativePath, stage: 'injection', reason: injection.reason });
      continue;
    }

    const pageId = injection.pageId;
    const policy =
      readPageAccessPolicy(publishDir, pageId) ?? assessment.policy;

    const smoke = await smokeTestPageDataInsert({
      apiBase,
      pageId,
      dataset: insertDataset,
      policy,
      fetchImpl,
    });
    if (smoke.ok) {
      continue;
    }

    failures.push({
      relativePath,
      stage: 'insert_smoke',
      reason: smoke.reason,
      status: smoke.status,
      bindingReasons: assessment.reasons,
    });
  }
  return failures;
}