memind/mindspace-h5-html-finish-guard.mjs

import crypto from 'node:crypto';
import fs from 'node:fs';
import path from 'node:path';
import { extractStaticPageLinks } from './mindspace-chat-save.mjs';
import {
  createPublicHtmlLinkExists,
  shouldRetryHtmlGenerationReply,
} from './wechat-mp.mjs';
import {
  isStubPublicHtmlContent,
  materializeMissingPublicHtmlWrites,
} from './mindspace-public-finish-sync.mjs';
import { scanWorkspaceFilesForProhibitedBrowserStorage } from './mindspace-browser-storage-policy.mjs';

const PUBLIC_HTML_PATH_PATTERN_GLOBAL = /(?:^|[^a-z0-9_./-])(public\/[a-z0-9][a-z0-9._/-]{0,255}\.html)\b/gi;
const PUBLIC_ASSET_ATTR_PATTERN = /(?:href|src)=["']([^"'#?\s]+)["']/gi;
const PUBLIC_ASSET_EXT_PATTERN = /\.(svg|png|jpe?g|webp|gif)$/i;

const repairAttemptsBySession = new Map();

function envFlag(value, fallback = false) {
  const raw = String(value ?? '').trim().toLowerCase();
  if (!raw) return fallback;
  return ['1', 'true', 'yes', 'on'].includes(raw);
}

function messageText(message) {
  const content = message?.content;
  if (typeof content === 'string') return content.trim();
  if (!Array.isArray(content)) return '';
  return content
    .filter((item) => item?.type === 'text')
    .map((item) => String(item.text ?? '').trim())
    .filter(Boolean)
    .join('\n')
    .trim();
}

function normalizePublicRelativePath(value, { publishDir } = {}) {
  const clean = String(value ?? '')
    .split('?')[0]
    .split('#')[0]
    .replace(/^\.\//, '')
    .trim()
    .replace(/\\/g, '/');
  if (!clean || clean.includes('://') || clean.startsWith('data:')) return null;
  if (!clean.startsWith('public/')) {
    if (clean.startsWith('assets/')) return `public/${clean}`;
    return null;
  }
  const root = path.resolve(String(publishDir ?? ''));
  const target = path.resolve(root, clean);
  if (target !== root && !target.startsWith(`${root}${path.sep}`)) return null;
  return clean;
}

function publicFileExists(publishDir, relativePath) {
  const normalized = normalizePublicRelativePath(relativePath, { publishDir });
  if (!normalized) return false;
  const target = path.resolve(String(publishDir ?? ''), normalized);
  return fs.existsSync(target) && fs.statSync(target).isFile();
}

function publicHtmlIsMissingOrStub(publishDir, relativePath) {
  const normalized = normalizePublicRelativePath(relativePath, { publishDir });
  if (!normalized || !normalized.toLowerCase().endsWith('.html')) return false;
  const target = path.resolve(String(publishDir ?? ''), normalized);
  if (!fs.existsSync(target) || !fs.statSync(target).isFile()) return true;
  try {
    return isStubPublicHtmlContent(fs.readFileSync(target, 'utf8'));
  } catch {
    return true;
  }
}

export function isH5HtmlFinishGuardEnabled(env = process.env) {
  return envFlag(env.MEMIND_H5_HTML_FINISH_GUARD, false);
}

export function resetH5HtmlFinishGuardAttempts(sessionId = null) {
  if (sessionId) repairAttemptsBySession.delete(String(sessionId));
  else repairAttemptsBySession.clear();
}

export function collectReferencedPublicHtmlPaths(messages, currentUser) {
  const paths = new Set();
  for (const message of Array.isArray(messages) ? messages : []) {
    if (message?.role !== 'assistant') continue;
    const text = messageText(message);
    if (!text) continue;
    for (const link of extractStaticPageLinks(text, {
      userId: currentUser?.id,
      username: currentUser?.username ?? null,
    })) {
      if (link?.relativePath) paths.add(link.relativePath);
    }
    for (const match of text.matchAll(PUBLIC_HTML_PATH_PATTERN_GLOBAL)) {
      if (match[1]) paths.add(match[1]);
    }
  }
  return [...paths].sort();
}

export function collectMissingPublicHtmlPaths(messages, currentUser, publishDir) {
  return collectReferencedPublicHtmlPaths(messages, currentUser).filter((relativePath) =>
    publicHtmlIsMissingOrStub(publishDir, relativePath),
  );
}

export function collectTouchedPublicCodePaths(messages, publishDir, syncResult = null) {
  const paths = new Set(
    (syncResult?.publicHtmlRelativePaths ?? [])
      .map((item) => normalizePublicRelativePath(item, { publishDir }))
      .filter(Boolean),
  );
  for (const message of Array.isArray(messages) ? messages : []) {
    for (const item of message?.content ?? []) {
      if (item?.type !== 'toolRequest') continue;
      const toolCall = item.toolCall?.value;
      const name = String(toolCall?.name ?? '').trim().split('__').at(-1);
      if (!['write_file', 'edit_file', 'write', 'edit'].includes(name)) continue;
      const args = toolCall?.arguments ?? {};
      const relativePath = normalizePublicRelativePath(args.path ?? args.file_path, { publishDir });
      if (relativePath) paths.add(relativePath);
    }
  }
  return [...paths];
}

export function collectMissingPublicAssetReferences(publishDir) {
  const root = path.resolve(String(publishDir ?? ''));
  const publicDir = path.join(root, 'public');
  if (!fs.existsSync(publicDir) || !fs.statSync(publicDir).isDirectory()) return [];

  const missing = new Map();
  for (const file of fs.readdirSync(publicDir)) {
    if (!file.toLowerCase().endsWith('.html')) continue;
    const htmlRelativePath = `public/${file}`;
    const content = fs.readFileSync(path.join(publicDir, file), 'utf8');
    for (const match of content.matchAll(PUBLIC_ASSET_ATTR_PATTERN)) {
      const relativePath = normalizePublicRelativePath(match[1], { publishDir: root });
      if (!relativePath || !PUBLIC_ASSET_EXT_PATTERN.test(relativePath)) continue;
      if (publicFileExists(root, relativePath)) continue;
      missing.set(relativePath, { relativePath, htmlRelativePath });
    }
  }
  return [...missing.values()].sort((a, b) => a.relativePath.localeCompare(b.relativePath));
}

export function evaluateH5HtmlFinishGuard({
  messages,
  currentUser,
  publishDir,
  syncResult = null,
} = {}) {
  materializeMissingPublicHtmlWrites({ messages, publishDir });
  const missingHtml = collectMissingPublicHtmlPaths(messages, currentUser, publishDir);
  const missingAssets = collectMissingPublicAssetReferences(publishDir);
  const browserStorageViolations = scanWorkspaceFilesForProhibitedBrowserStorage({
    publishDir,
    relativePaths: collectTouchedPublicCodePaths(messages, publishDir, syncResult),
  });

  const lastAssistant = [...(Array.isArray(messages) ? messages : [])]
    .reverse()
    .find((message) => message?.role === 'assistant');
  const reply = {
    text: messageText(lastAssistant),
    messages: Array.isArray(messages) ? messages : [],
  };
  const intent = { agentText: reply.text };
  const linkExists = createPublicHtmlLinkExists(publishDir);
  const confirmedArtifacts = (syncResult?.publicHtmlRelativePaths ?? [])
    .filter((relativePath) => publicFileExists(publishDir, relativePath))
    .map((relativePath) => ({ relativePath, localPath: path.join(publishDir, relativePath) }));

  const htmlGenerationNeedsRetry = shouldRetryHtmlGenerationReply({
    reply,
    intent,
    confirmedArtifacts,
    hasValidLinkInReply: false,
  });

  const needsRepair =
    missingHtml.length > 0 ||
    missingAssets.length > 0 ||
    browserStorageViolations.length > 0 ||
    htmlGenerationNeedsRetry;

  return {
    needsRepair,
    missingHtml,
    missingAssets,
    browserStorageViolations,
    htmlGenerationNeedsRetry,
    linkExists,
  };
}

export function buildH5HtmlRepairPrompt({
  missingHtml = [],
  missingAssets = [],
  browserStorageViolations = [],
} = {}) {
  const lines = [
    '【系统补盘请求】检测到 public 页面交付不完整。请立即使用 write_file 写入缺失文件（完整内容），并在完成后确认磁盘存在。',
    '要求：先 load_skill static-page-publish，再逐个 write_file；不要用空回复或仅 edit_file 改 HTML 代替资源落盘。',
  ];
  if (missingHtml.length) {
    lines.push('', '缺失或仍为 stub 的 HTML：');
    for (const relativePath of missingHtml) lines.push(`- ${relativePath}`);
  }
  if (missingAssets.length) {
    lines.push('', '页面已引用但磁盘缺失的资源：');
    for (const item of missingAssets) {
      lines.push(`- ${item.relativePath}（来自 ${item.htmlRelativePath}）`);
    }
  }
  if (browserStorageViolations.length) {
    lines.push(
      '',
      '以下页面违反数据存储硬约束：',
      ...browserStorageViolations.map((item) => `- ${item.relativePath}：禁止 ${item.apis.join(', ')}`),
      '',
      '必须删除浏览器存储代码，load_skill → page-data-collect，使用 private_data_* 在当前用户专属 PostgreSQL schema 建表和注册 dataset，HTML 只能通过 Page Data API 读写。禁止 SQLite 或任何本地 fallback。',
    );
  }
  lines.push('', '写完后请 list_dir public/assets 或对应目录，确认文件真实存在。');
  return lines.join('\n');
}

export async function maybeRepairH5HtmlAfterFinish({
  sessionId,
  userId,
  currentUser,
  messages,
  publishDir,
  syncResult = null,
  tkmindProxy = null,
  maxAttempts = 1,
  env = process.env,
  logger = console,
} = {}) {
  const evaluation = evaluateH5HtmlFinishGuard({
    messages,
    currentUser,
    publishDir,
    syncResult,
  });
  if (!evaluation.needsRepair) {
    resetH5HtmlFinishGuardAttempts(sessionId);
    return { repaired: false, skipped: 'ok', ...evaluation };
  }
  if (!isH5HtmlFinishGuardEnabled(env)) {
    logger.warn?.(
      `[MindSpace][h5-html-guard] incomplete public delivery for session ${sessionId}: `
      + `html=${evaluation.missingHtml.join(',') || 'none'} `
      + `assets=${evaluation.missingAssets.map((item) => item.relativePath).join(',') || 'none'}`,
    );
    return { repaired: false, skipped: 'disabled', ...evaluation };
  }
  if (!tkmindProxy?.submitSessionReplyForUser) {
    return { repaired: false, skipped: 'no_proxy', ...evaluation };
  }

  const key = String(sessionId ?? '');
  const attempts = repairAttemptsBySession.get(key) ?? 0;
  if (!key || attempts >= maxAttempts) {
    logger.warn?.(
      `[MindSpace][h5-html-guard] repair limit reached for session ${sessionId}`,
      evaluation,
    );
    return { repaired: false, skipped: 'limit', attempts, ...evaluation };
  }
  repairAttemptsBySession.set(key, attempts + 1);

  const prompt = buildH5HtmlRepairPrompt(evaluation);
  // goosed validates request_id as a UUID, so repair metadata must not be
  // encoded by prefixing the id. The message metadata below carries the type.
  const requestId = crypto.randomUUID();
  const userMessage = {
    role: 'user',
    content: [{ type: 'text', text: prompt }],
    metadata: {
      displayText: '请补全缺失的 public 页面与资源文件',
      userVisible: false,
      agentVisible: true,
      memindRun: { htmlFinishRepair: true },
    },
  };

  try {
    await tkmindProxy.submitSessionReplyForUser(
      userId,
      sessionId,
      requestId,
      userMessage,
    );
    logger.info?.(
      `[MindSpace][h5-html-guard] triggered repair for session ${sessionId} `
      + `(attempt ${attempts + 1}/${maxAttempts})`,
    );
    return { repaired: true, triggered: true, attempts: attempts + 1, ...evaluation };
  } catch (err) {
    logger.warn?.(
      `[MindSpace][h5-html-guard] repair failed for session ${sessionId}: `
      + `${err instanceof Error ? err.message : err}`,
    );
    return {
      repaired: false,
      skipped: 'error',
      error: err instanceof Error ? err.message : String(err),
      attempts: attempts + 1,
      ...evaluation,
    };
  }
}