memind/mindspace-public-finish-sync.mjs

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

import { extractStaticPageLinks, materializePrivateAssetsInPublicHtmlFiles } from './mindspace-chat-save.mjs';
import { DOWNLOADABLE_FILE_PATTERN } from './mindspace-html-download-links.mjs';
import { scheduleWorkspaceHtmlThumbnailSidecars } from './mindspace-workspace-thumbnails.mjs';

/**
 * Public HTML finish-sync invariants (regression guard — do not simplify away).
 * edit_file patches must materialize onto MindSpace/<userId>/public/*.html.
 * See docs/regression-guards/mindspace-publish-and-chat-finish.md
 */

const PUBLIC_HTML_PATH_PATTERN = /(?:^|[^a-z0-9_./-])(public\/[a-z0-9][a-z0-9._/-]{0,255}\.html)\b/i;
const PUBLIC_HTML_PATH_PATTERN_GLOBAL = /(?:^|[^a-z0-9_./-])(public\/[a-z0-9][a-z0-9._/-]{0,255}\.html)\b/gi;
const PUBLIC_DOWNLOAD_ATTR_PATTERN = /(?:href|src)=["']([^"'#?\s]+)["']/gi;
const DEFAULT_DOWNLOAD_SYNC_MIN_SIZE = 1024;

function normalizePublicDownloadHref(href) {
  const clean = String(href ?? '')
    .split('?')[0]
    .split('#')[0]
    .replace(/^\.\//, '')
    .trim();
  if (
    !clean ||
    clean.includes('://') ||
    clean.startsWith('data:') ||
    !DOWNLOADABLE_FILE_PATTERN.test(clean)
  ) {
    return null;
  }
  if (clean.startsWith('../oa/')) return clean.slice(3);
  if (clean.startsWith('oa/')) return clean;
  if (!clean.includes('/')) return `public/${clean}`;
  return clean;
}

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

  const refs = new Map();
  for (const file of fs.readdirSync(publicDir)) {
    if (!file.toLowerCase().endsWith('.html')) continue;
    const content = fs.readFileSync(path.join(publicDir, file), 'utf8');
    const htmlRelativePath = `public/${file}`;
    for (const match of content.matchAll(PUBLIC_DOWNLOAD_ATTR_PATTERN)) {
      const relativePath = normalizePublicDownloadHref(match[1]);
      if (relativePath && !refs.has(relativePath)) {
        refs.set(relativePath, { relativePath, htmlRelativePath });
      }
    }
  }
  return [...refs.values()];
}

function listOaDownloadCandidates(publishDir, extension, { minSize = DEFAULT_DOWNLOAD_SYNC_MIN_SIZE } = {}) {
  const oaDir = path.join(path.resolve(String(publishDir ?? '')), 'oa');
  if (!fs.existsSync(oaDir) || !fs.statSync(oaDir).isDirectory()) return [];

  return fs
    .readdirSync(oaDir)
    .filter((name) => name.toLowerCase().endsWith(extension))
    .map((name) => {
      const absolutePath = path.join(oaDir, name);
      const stat = fs.statSync(absolutePath);
      return { name, absolutePath, size: stat.size, mtimeMs: stat.mtimeMs };
    })
    .filter((item) => item.size >= minSize)
    .sort((a, b) => b.mtimeMs - a.mtimeMs || b.size - a.size);
}

function resolvePublicDownloadSyncSource(
  root,
  relativePath,
  htmlRelativePath,
  { minSize = DEFAULT_DOWNLOAD_SYNC_MIN_SIZE } = {},
) {
  const extension = path.extname(relativePath).toLowerCase();
  const basename = path.basename(relativePath);
  const htmlStemFilename = path.posix.basename(htmlRelativePath).replace(/\.html$/i, extension);
  const exactCandidates = [
    path.posix.join('oa', basename),
    path.posix.join('oa', htmlStemFilename),
  ];

  for (const candidate of exactCandidates) {
    const absolutePath = path.resolve(root, candidate);
    if (absolutePath === path.resolve(root, relativePath)) continue;
    try {
      const stat = fs.statSync(absolutePath);
      if (stat.isFile() && stat.size >= minSize) {
        return {
          name: path.basename(candidate),
          absolutePath,
          size: stat.size,
          strategy: 'exact_oa_match',
        };
      }
    } catch {
      // Ignore missing exact candidates and continue.
    }
  }

  const fallbackCandidates = listOaDownloadCandidates(root, extension, { minSize });
  if (fallbackCandidates.length === 1) {
    return { ...fallbackCandidates[0], strategy: 'single_oa_fallback' };
  }

  return null;
}

export function syncPublicDocxDownloads({
  publishDir,
  minCompleteSize = DEFAULT_DOWNLOAD_SYNC_MIN_SIZE,
} = {}) {
  const root = path.resolve(String(publishDir ?? ''));
  if (!root) return { synced: [], skipped: [], missing: [], source: null, sourceByRelativePath: {} };

  const synced = [];
  const skipped = [];
  const missing = [];
  const sourceByRelativePath = {};
  for (const reference of extractPublicDownloadReferencesFromHtml(root)) {
    const { relativePath, htmlRelativePath } = reference;
    const destination = path.resolve(root, relativePath);
    if (destination !== root && !destination.startsWith(`${root}${path.sep}`)) {
      skipped.push(relativePath);
      continue;
    }
    const source = resolvePublicDownloadSyncSource(root, relativePath, htmlRelativePath, {
      minSize: minCompleteSize,
    });
    const destinationExists = fs.existsSync(destination) && fs.statSync(destination).isFile();
    if (destinationExists && !source) {
      skipped.push(relativePath);
      continue;
    }
    if (!source) {
      missing.push(relativePath);
      continue;
    }
    if (destinationExists) {
      try {
        if (fs.statSync(destination).size >= source.size) {
          skipped.push(relativePath);
          continue;
        }
      } catch {
        // Fall through and rewrite from source.
      }
    }
    try {
      fs.mkdirSync(path.dirname(destination), { recursive: true });
      fs.copyFileSync(source.absolutePath, destination);
      synced.push(relativePath);
      sourceByRelativePath[relativePath] = source.name;
    } catch {
      skipped.push(relativePath);
    }
  }
  const sourceNames = [...new Set(Object.values(sourceByRelativePath))];
  return {
    synced,
    skipped,
    missing,
    source: sourceNames.length === 1 ? sourceNames[0] : null,
    sourceByRelativePath,
  };
}

function messageText(message) {
  const textParts = Array.isArray(message?.content)
    ? message.content
        .filter((item) => item?.type === 'text' && typeof item.text === 'string')
        .map((item) => item.text.trim())
        .filter(Boolean)
    : [];
  const displayText =
    typeof message?.metadata?.displayText === 'string' ? message.metadata.displayText.trim() : '';
  return [...textParts, displayText].filter(Boolean).join('\n');
}

function messageId(message) {
  const id =
    message?.id ??
    message?.message_id ??
    message?.metadata?.messageId ??
    message?.metadata?.message_id ??
    null;
  const normalized = String(id ?? '').trim();
  return normalized || null;
}

export function normalizePublicHtmlRelativePath(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 publicHtmlFileExists(publishDir, relativePath) {
  if (!publishDir) return false;
  const root = path.resolve(String(publishDir ?? ''));
  const normalized = normalizePublicHtmlRelativePath(relativePath);
  if (!root || !normalized.toLowerCase().endsWith('.html')) return false;
  const target = path.resolve(root, normalized);
  if (target !== root && !target.startsWith(`${root}${path.sep}`)) return false;
  return fs.existsSync(target) && fs.statSync(target).isFile();
}

export function collectOwnPublicHtmlRelativePaths({
  messages,
  currentUser,
  publishDir,
  materialized = [],
  skipped = [],
} = {}) {
  const paths = new Set();
  const add = (relativePath) => {
    const normalized = normalizePublicHtmlRelativePath(relativePath);
    if (normalized && publicHtmlFileExists(publishDir, normalized)) paths.add(normalized);
  };
  for (const relativePath of [...materialized, ...skipped]) add(relativePath);
  for (const message of Array.isArray(messages) ? messages : []) {
    const text = messageText(message);
    if (!text) continue;
    for (const link of extractStaticPageLinks(text, {
      userId: currentUser?.id,
      username: currentUser?.username ?? null,
    })) {
      add(link.relativePath);
    }
    for (const match of text.matchAll(PUBLIC_HTML_PATH_PATTERN_GLOBAL)) {
      add(match[1]);
    }
  }
  return [...paths].sort();
}

export function collectOwnPublicHtmlArtifactRefs({
  messages,
  currentUser,
  publishDir,
  materialized = [],
  skipped = [],
} = {}) {
  const refs = new Map();
  const add = (relativePath, sourceMessageId = null) => {
    const normalized = normalizePublicHtmlRelativePath(relativePath);
    if (!normalized || !publicHtmlFileExists(publishDir, normalized)) return;
    const existing = refs.get(normalized);
    refs.set(normalized, {
      relativePath: normalized,
      messageId: existing?.messageId ?? sourceMessageId ?? null,
    });
  };
  for (const relativePath of [...materialized, ...skipped]) add(relativePath);
  for (const artifact of extractPublicHtmlWriteArtifacts(messages, { publishDir })) {
    add(artifact.relativePath, artifact.messageId ?? null);
  }
  for (const message of Array.isArray(messages) ? messages : []) {
    const sourceMessageId = messageId(message);
    const text = messageText(message);
    if (!text) continue;
    for (const link of extractStaticPageLinks(text, {
      userId: currentUser?.id,
      username: currentUser?.username ?? null,
    })) {
      add(link.relativePath, sourceMessageId);
    }
    for (const match of text.matchAll(PUBLIC_HTML_PATH_PATTERN_GLOBAL)) {
      add(match[1], sourceMessageId);
    }
  }
  return [...refs.values()].sort((a, b) => a.relativePath.localeCompare(b.relativePath));
}

function isWriteLikeHtmlTool(name, args) {
  const action = String(args?.action ?? '').toLowerCase();
  const normalizedName = String(name ?? '').trim();
  const writeLikeDeveloper =
    (normalizedName === 'developer' && (action === 'write' || action === 'edit')) ||
    normalizedName === 'write' ||
    normalizedName.endsWith('__write');
  const writeLikeSandbox =
    (normalizedName === 'write_file' ||
      normalizedName === 'edit_file' ||
      normalizedName.endsWith('__write_file') ||
      normalizedName.endsWith('__edit_file')) &&
    typeof args?.path === 'string';
  return (writeLikeDeveloper && typeof args?.path === 'string') || writeLikeSandbox;
}

function isEditLikeHtmlTool(name, args) {
  const normalizedName = String(name ?? '').trim();
  return (
    (normalizedName === 'edit_file' ||
      normalizedName.endsWith('__edit_file') ||
      (normalizedName === 'developer' && String(args?.action ?? '').toLowerCase() === 'edit')) &&
    typeof args?.path === 'string'
  );
}

function readPublicHtmlBaseline(publishDir, relativePath) {
  const root = path.resolve(String(publishDir ?? ''));
  if (!root) return '';
  const destination = path.resolve(root, relativePath);
  if (destination !== root && !destination.startsWith(`${root}${path.sep}`)) return '';
  if (!fs.existsSync(destination) || !fs.statSync(destination).isFile()) return '';
  try {
    return fs.readFileSync(destination, 'utf8');
  } catch {
    return '';
  }
}

function applyPublicHtmlEdit(existing, args) {
  if (typeof args?.new_str !== 'string') return null;
  const oldStr = typeof args.old_str === 'string' ? args.old_str : '';
  if (oldStr && !existing.includes(oldStr)) return null;
  return oldStr ? existing.replace(oldStr, args.new_str) : args.new_str;
}

export function extractPublicHtmlWriteArtifacts(messages = [], { publishDir = null } = {}) {
  const artifacts = new Map();
  for (const message of messages) {
    const sourceMessageId = messageId(message);
    for (const item of message?.content ?? []) {
      if (item?.type !== 'toolRequest') continue;
      const toolCall = item.toolCall?.value;
      const args = toolCall?.arguments ?? {};
      if (!isWriteLikeHtmlTool(toolCall?.name, args)) continue;
      const candidate = String(args.path ?? '').trim();
      if (!candidate.toLowerCase().endsWith('.html')) continue;
      const relativePath = normalizePublicHtmlRelativePath(candidate);
      if (!relativePath) continue;
      let content = null;
      if (typeof args.content === 'string') {
        content = args.content;
      } else if (isEditLikeHtmlTool(toolCall?.name, args)) {
        const baseline =
          artifacts.get(relativePath)?.content ??
          readPublicHtmlBaseline(publishDir, relativePath);
        content = applyPublicHtmlEdit(baseline, args);
      } else if (typeof args.new_str === 'string' && !args.old_str) {
        content = args.new_str;
      }
      if (content == null) continue;
      artifacts.set(relativePath, { relativePath, content, messageId: sourceMessageId });
    }
  }
  return [...artifacts.values()];
}

const PUBLIC_HTML_STUB_MARKERS = ['临时补出', '服务号兜底', '服务号自动补出简版页面'];

export function isStubPublicHtmlContent(content) {
  const value = String(content ?? '');
  return PUBLIC_HTML_STUB_MARKERS.some((marker) => value.includes(marker));
}

function shouldReplaceExistingPublicHtml(destination, nextContent) {
  if (!fs.existsSync(destination) || !fs.statSync(destination).isFile()) return true;
  try {
    const existing = fs.readFileSync(destination, 'utf8');
    if (isStubPublicHtmlContent(existing)) return true;
    return typeof nextContent === 'string' && nextContent.length > 0 && existing !== nextContent;
  } catch {
    return Boolean(nextContent);
  }
}

export function materializeMissingPublicHtmlWrites({ messages, publishDir }) {
  const root = path.resolve(String(publishDir ?? ''));
  if (!root) return { materialized: [], skipped: [] };

  const materialized = [];
  const skipped = [];
  for (const artifact of extractPublicHtmlWriteArtifacts(messages, { publishDir: root })) {
    const destination = path.resolve(root, artifact.relativePath);
    if (destination !== root && !destination.startsWith(`${root}${path.sep}`)) {
      skipped.push(artifact.relativePath);
      continue;
    }
    if (fs.existsSync(destination) && fs.statSync(destination).isFile()) {
      if (!shouldReplaceExistingPublicHtml(destination, artifact.content)) {
        skipped.push(artifact.relativePath);
        continue;
      }
    }
    try {
      fs.mkdirSync(path.dirname(destination), { recursive: true });
      fs.writeFileSync(destination, artifact.content, 'utf8');
      materialized.push(artifact.relativePath);
      scheduleWorkspaceHtmlThumbnailSidecars(root, artifact.relativePath);
    } catch {
      skipped.push(artifact.relativePath);
    }
  }
  return { materialized, skipped };
}

export function materializePublicHtmlWritesFromSessionEvent(
  event,
  { publishDir, recentCount = 20 } = {},
) {
  if (!event || !publishDir) {
    return { materialized: [], skipped: [] };
  }
  if (event.type === 'Message' && event.message) {
    return materializeMissingPublicHtmlWrites({
      messages: [event.message],
      publishDir,
    });
  }
  if (event.type === 'UpdateConversation' && Array.isArray(event.conversation)) {
    return materializeMissingPublicHtmlWrites({
      messages: event.conversation.slice(-Math.max(1, recentCount)),
      publishDir,
    });
  }
  return { materialized: [], skipped: [] };
}

/** Read the public HTML targets from a stream event without writing them. */
export function collectPublicHtmlWritePathsFromSessionEvent(event, { publishDir, recentCount = 20 } = {}) {
  if (!event || !publishDir) return [];
  const messages = event.type === 'Message' && event.message
    ? [event.message]
    : event.type === 'UpdateConversation' && Array.isArray(event.conversation)
      ? event.conversation.slice(-Math.max(1, recentCount))
      : [];
  return [...new Set(extractPublicHtmlWriteArtifacts(messages, { publishDir })
    .map((artifact) => artifact.relativePath)
    .filter(Boolean))];
}

function messageHasPublicHtmlToolRequest(message, { publishDir = null } = {}) {
  if (extractPublicHtmlWriteArtifacts([message], { publishDir }).length > 0) {
    return true;
  }
  for (const item of message?.content ?? []) {
    if (item?.type !== 'toolRequest') continue;
    const toolCall = item.toolCall?.value;
    const args = toolCall?.arguments ?? {};
    if (!isWriteLikeHtmlTool(toolCall?.name, args)) continue;
    const candidate = String(args.path ?? '').trim();
    if (!candidate.toLowerCase().endsWith('.html')) continue;
    if (normalizePublicHtmlRelativePath(candidate)) return true;
  }
  return false;
}

export function hasRecentOwnPublicHtmlReference(
  messages,
  currentUser,
  { recentCount = 80, publishDir = null } = {},
) {
  if (!Array.isArray(messages) || messages.length === 0 || !currentUser?.id) return false;
  const recentMessages = messages.slice(-Math.max(1, recentCount));
  for (const message of recentMessages) {
    if (messageHasPublicHtmlToolRequest(message, { publishDir })) {
      return true;
    }
    const text = messageText(message);
    if (!text) continue;
    if (
      extractStaticPageLinks(text, {
        userId: currentUser.id,
        username: currentUser.username ?? null,
      }).length > 0
    ) {
      return true;
    }
    if (PUBLIC_HTML_PATH_PATTERN.test(text)) {
      return true;
    }
  }
  return false;
}

export async function syncPublicHtmlAfterFinish({
  messages,
  currentUser,
  publishDir,
  syncWorkspaceAssets,
  registerPublicHtmlArtifacts,
  sessionId = null,
  pool = null,
  storageRoot = null,
  h5Root = null,
} = {}) {
  if (!hasRecentOwnPublicHtmlReference(messages, currentUser, { publishDir })) {
    return { materialized: [], skipped: [], synced: false };
  }

  const { materialized, skipped } = materializeMissingPublicHtmlWrites({ messages, publishDir });
  const docxSync = syncPublicDocxDownloads({ publishDir });
  const publicHtmlArtifactRefs = collectOwnPublicHtmlArtifactRefs({
    messages,
    currentUser,
    publishDir,
    materialized,
    skipped,
  });
  const publicHtmlRelativePaths = publicHtmlArtifactRefs.map((ref) => ref.relativePath);
  const sourceMessageId = publicHtmlArtifactRefs.find((ref) => ref.messageId)?.messageId ?? null;
  let assetMaterialization = { materialized: [], skipped: [] };
  if (pool && storageRoot && h5Root && currentUser?.id && publicHtmlRelativePaths.length > 0) {
    assetMaterialization = await materializePrivateAssetsInPublicHtmlFiles({
      pool,
      storageRoot,
      h5Root,
      userId: currentUser.id,
      publishDir,
      relativePaths: publicHtmlRelativePaths,
    }).catch(() => ({ materialized: [], skipped: publicHtmlRelativePaths }));
  }
  let synced = false;
  if (typeof syncWorkspaceAssets === 'function' && currentUser?.id) {
    await syncWorkspaceAssets(currentUser.id, {
      categoryCode: 'public',
      sourceSessionId: sessionId,
      sourceMessageId,
      onlyRelativePaths: publicHtmlRelativePaths,
    });
    synced = true;
  }
  if (typeof registerPublicHtmlArtifacts === 'function' && currentUser?.id && sessionId) {
    await registerPublicHtmlArtifacts(currentUser.id, {
      sessionId,
      relativePaths: publicHtmlRelativePaths,
      artifactRefs: publicHtmlArtifactRefs,
    });
  }
  return {
    materialized,
    skipped,
    synced,
    docxSync,
    publicHtmlRelativePaths,
    publicHtmlArtifactRefs,
    assetMaterialization,
  };
}