memind/mindspace-wechat-page-draft.mjs

import fs from 'node:fs';
import path from 'node:path';
import { fetch as undiciFetch } from 'undici';
import sharp from 'sharp';
import {
  buildPublicUrl,
  PUBLISH_ROOT_DIR,
  PUBLIC_ZONE_DIR,
  resolvePublicBaseUrl,
} from './user-publish.mjs';
import { extractMindspaceCoverPath } from './wechat/verify/generated-thumbnail.mjs';
import { extractPageTitle } from './wechat/verify/share-preview-repair.mjs';
import { readWorkspacePublishHtml } from './mindspace-workspace-path.mjs';
import { resolvePageWorkspaceRelativePath } from './mindspace-workspace-relative-path.mjs';
import { fetchWechatMpAccessToken } from './mindspace-wechat-mp-config.mjs';
import {
  addWechatDraftArticle,
  buildDailyNewsWechatDraftArticleForPush,
  convertNewsPageHtmlToWechatArticle,
  isDailyNewsFormat,
  uploadWechatArticleContentImage,
  uploadWechatPermanentThumb,
} from './wechat-news-morning-draft.mjs';
import { finalizeWechatDraftArticleForPush } from './wechat-draft-article-layout.mjs';
import {
  verifyWechatDraftCoverResolution,
  verifyWechatDraftPublicationContent,
  WECHAT_DRAFT_PUBLICATION_STANDARD_VERSION,
} from './wechat-draft-publication-standard.mjs';

export { WECHAT_DRAFT_PUBLICATION_STANDARD_VERSION };

const RUNS_TABLE = 'h5_user_wechat_page_draft_runs';
const DEFAULT_TOKEN_URL = 'https://api.weixin.qq.com/cgi-bin/stable_token';

function escapeHtml(value) {
  return String(value ?? '')
    .replace(/&/g, '&')
    .replace(/</g, '&lt;')
    .replace(/>/g, '&gt;')
    .replace(/"/g, '&quot;');
}

function extractMetaDescription(html) {
  const match = String(html ?? '').match(
    /<meta[^>]+name=["']description["'][^>]+content=["']([^"']*)["']/i,
  );
  return match?.[1]?.trim() ?? '';
}

function stripHtml(value) {
  return String(value ?? '')
    .replace(/<br\s*\/?>/gi, '\n')
    .replace(/<[^>]+>/g, '')
    .replace(/\s+/g, ' ')
    .trim();
}

function inlineHtmlToWechatText(value) {
  return String(value ?? '')
    .replace(/<br\s*\/?>/gi, '\n')
    .replace(/<em[^>]*>([\s\S]*?)<\/em>/gi, '<span style="color:#d98f4a;font-style:normal;">$1</span>')
    .replace(/<strong[^>]*>([\s\S]*?)<\/strong>/gi, '<strong>$1</strong>')
    .replace(/<b[^>]*>([\s\S]*?)<\/b>/gi, '<strong>$1</strong>')
    .replace(/<[^>]+>/g, '')
    .replace(/\n{3,}/g, '\n\n')
    .trim();
}

function stripLegacyReadOriginalFooter(content) {
  return String(content ?? '')
    .replace(/<p[^>]*>\s*阅读原文[：:][\s\S]*?<\/p>/gi, '')
    .trim();
}

export function isNewsHotspotFormat(html) {
  const source = String(html ?? '');
  if (isDailyNewsFormat(source)) return false;
  if (/news-hotspot|今日新闻热点/u.test(source)) return true;
  if (!/class="card"/.test(source)) return false;
  return /<h2[^>]*>/.test(source)
    && (/<span class="tag"/.test(source) || /<div class="k"/.test(source));
}

function normalizeImageRef(value) {
  const ref = String(value ?? '').trim();
  if (!ref || /^data:/i.test(ref)) return '';
  return ref;
}

function resolveImageFilePath(imageRef, publishDir, htmlRelativePath = '') {
  const ref = normalizeImageRef(imageRef);
  if (!ref) return null;
  if (/^https?:\/\//i.test(ref)) return ref;
  const htmlDir = htmlRelativePath
    ? path.dirname(path.join(publishDir, htmlRelativePath))
    : publishDir;
  const publicDir = path.join(publishDir, PUBLIC_ZONE_DIR);
  const candidates = [
    path.resolve(htmlDir, ref.replace(/^\.\//, '')),
    path.resolve(publicDir, ref.replace(/^\.\//, '')),
    path.resolve(publishDir, ref.replace(/^\.\//, '')),
    path.resolve(publishDir, ref.replace(/^\/+/, '')),
  ];
  for (const candidate of candidates) {
    if (/^https?:\/\//i.test(candidate)) return candidate;
    if (fs.existsSync(candidate)) return candidate;
  }
  return null;
}

function normalizeWechatImageUrl(imageUrl) {
  return String(imageUrl ?? '').trim().replace(/^http:\/\//i, 'https://');
}

function renderWechatImage(imageUrl, alt = '') {
  const safeUrl = normalizeWechatImageUrl(imageUrl);
  if (!safeUrl) return '';
  return [
    '<section style="margin:0 0 16px;text-align:center;">',
    `<img src="${safeUrl}" alt="${escapeHtml(alt)}" style="width:100%;max-width:100%;display:block;margin:0 auto;border-radius:10px;" />`,
    '</section>',
  ].join('');
}

function resolveSidecarThumbnailPaths(publishDir, htmlRelativePath) {
  if (!publishDir || !htmlRelativePath) return [];
  const htmlPath = path.join(publishDir, htmlRelativePath);
  const pngPath = htmlPath.replace(/\.html$/i, '.thumbnail.png');
  const svgPath = htmlPath.replace(/\.html$/i, '.thumbnail.svg');
  if (fs.existsSync(pngPath)) return [pngPath];
  if (fs.existsSync(svgPath)) return [svgPath];
  return [];
}

function collectPageImageCandidatesFromHtml(html, { publishDir = '', htmlRelativePath = '' } = {}) {
  const source = String(html ?? '');
  const refs = [];
  const coverPath = extractMindspaceCoverPath(source);
  if (coverPath) refs.push(coverPath);
  for (const match of source.matchAll(/background-image:\s*url\(["']?([^"')]+)["']?\)/gi)) {
    refs.push(match[1]);
  }
  for (const match of source.matchAll(/<img[^>]+src=["']([^"']+)["']/gi)) {
    refs.push(match[1]);
  }
  const seen = new Set();
  const resolved = [];
  for (const ref of refs) {
    const normalized = normalizeImageRef(ref);
    if (!normalized || seen.has(normalized)) continue;
    seen.add(normalized);
    resolved.push({
      ref: normalized,
      absPath: resolveImageFilePath(normalized, publishDir, htmlRelativePath),
    });
  }
  return resolved;
}

export function collectPageImageCandidates(html, { publishDir = '', htmlRelativePath = '' } = {}) {
  const resolved = collectPageImageCandidatesFromHtml(html, { publishDir, htmlRelativePath });
  if (
    htmlRelativePath
    && publishDir
    && !resolveHeroImagePath(html, { publishDir, htmlRelativePath })
  ) {
    const seen = new Set(resolved.map((item) => item.ref));
    for (const sidecarPath of resolveSidecarThumbnailPaths(publishDir, htmlRelativePath)) {
      const ref = path.basename(sidecarPath);
      if (seen.has(ref)) continue;
      seen.add(ref);
      resolved.push({ ref, absPath: sidecarPath });
    }
  }
  return resolved;
}

function countBodyImages(html) {
  return collectPageImageCandidates(html).length;
}

function extractBodyHtml(source) {
  const match = String(source ?? '').match(/<body[^>]*>([\s\S]*?)<\/body>/i);
  return match ? match[1] : String(source ?? '');
}

function stripMindspaceDeliveryChrome(source) {
  let body = extractBodyHtml(source);
  body = body.replace(/<style id="mindspace-public-share-style">[\s\S]*?<\/script>\s*/i, '');
  body = body.replace(/<script[\s\S]*?<\/script>/gi, '');
  body = body.replace(/<style[\s\S]*?<\/style>/gi, '');
  body = body.replace(/<link[^>]*>/gi, '');
  body = body.replace(/<div[^>]*data-mindspace-wechat-dialog[\s\S]*?<\/div>\s*/gi, '');
  body = body.replace(/<div[^>]*data-mindspace-public-share[\s\S]*?<\/div>\s*/gi, '');
  body = body.replace(/<small[^>]*aria-live="polite"[^>]*>[\s\S]*?<\/small>/gi, '');
  return body.trim();
}

function isMindspaceBoilerplateText(text) {
  const normalized = stripHtml(text);
  if (!normalized) return true;
  if (/^(TKMind · 智趣|AI 配图服务暂时不可用)$/u.test(normalized)) return true;
  if (/^(取消|确认推送|知道了|关闭|保存长图|公开分享|公众号|发布 Plaza)$/u.test(normalized)) return true;
  return false;
}

function resolveMindspaceMainContentHtml(body) {
  const main = body.match(/<main[^>]*>([\s\S]*?)<\/main>/i)?.[1];
  if (main) return main;
  const article = body.match(/<article[^>]*>([\s\S]*?)<\/article>/i)?.[1];
  if (article) return article;
  const page = body.match(/<div class="page"[^>]*>([\s\S]*?)<\/div>/i)?.[1];
  if (page) return page;
  const content = body.match(/<div class="content"[^>]*>([\s\S]*?)<\/div>/i)?.[1];
  if (content) return content;
  return body;
}

function appendParagraphsFromHtml(html, blocks, seen) {
  for (const match of String(html ?? '').matchAll(/<p[^>]*>([\s\S]*?)<\/p>/gi)) {
    const text = stripHtml(match[1]);
    if (!text || isMindspaceBoilerplateText(text) || seen.has(text)) continue;
    seen.add(text);
    blocks.push(
      `<p style="margin:0 0 16px;font-size:16px;line-height:1.9;color:#333;text-align:justify;">${escapeHtml(text)}</p>`,
    );
  }
}

function appendListsFromHtml(html, blocks, seen) {
  for (const match of String(html ?? '').matchAll(/<li[^>]*>([\s\S]*?)<\/li>/gi)) {
    const text = stripHtml(match[1]);
    if (!text || isMindspaceBoilerplateText(text) || seen.has(text)) continue;
    seen.add(text);
    blocks.push(
      `<p style="margin:0 0 10px;font-size:15px;line-height:1.8;color:#444;padding-left:12px;">• ${escapeHtml(text)}</p>`,
    );
  }
}

function appendBlockquotesFromHtml(html, blocks, seen) {
  for (const match of String(html ?? '').matchAll(/<blockquote[^>]*>([\s\S]*?)<\/blockquote>/gi)) {
    const text = stripHtml(match[1]);
    if (!text || isMindspaceBoilerplateText(text) || seen.has(text)) continue;
    seen.add(text);
    blocks.push(
      `<blockquote style="margin:0 0 16px;padding:12px 16px;border-left:3px solid #d98f4a;background:#faf7f1;color:#2a1a12;font-size:15px;line-height:1.8;">${escapeHtml(text)}</blockquote>`,
    );
  }
}

function appendFallbackMindspaceArticleSections(sections, source, { skipTexts = new Set() } = {}) {
  let body = stripMindspaceDeliveryChrome(source);
  body = body
    .replace(/<header class="hero"[^>]*>[\s\S]*?<\/header>/gi, '')
    .replace(/<div class="hero"[^>]*>[\s\S]*?<\/div>/gi, '')
    .replace(/<div class="title-block"[^>]*>[\s\S]*?<\/div>/gi, '')
    .replace(/<div class="divider"[^>]*>\s*<\/div>/gi, '')
    .replace(/<div class="tags"[^>]*>[\s\S]*?<\/div>/gi, '')
    .replace(/<footer[^>]*>[\s\S]*?<\/footer>/gi, '');

  const contentHtml = resolveMindspaceMainContentHtml(body);
  const seen = new Set([...skipTexts].map((text) => stripHtml(text)).filter(Boolean));
  const blocks = [];

  for (const match of contentHtml.matchAll(/<h2[^>]*>([\s\S]*?)<\/h2>([\s\S]*?)(?=<h2|<h3|$)/gi)) {
    const heading = stripHtml(match[1]);
    if (!heading || seen.has(heading)) continue;
    seen.add(heading);
    blocks.push(
      `<p style="margin:0 0 10px;font-size:17px;font-weight:700;color:#2a1a12;">${escapeHtml(heading)}</p>`,
    );
    appendParagraphsFromHtml(match[2], blocks, seen);
    appendListsFromHtml(match[2], blocks, seen);
  }

  for (const match of contentHtml.matchAll(/<h3[^>]*>([\s\S]*?)<\/h3>([\s\S]*?)(?=<h[23]|$)/gi)) {
    const heading = stripHtml(match[1]);
    if (!heading || seen.has(heading)) continue;
    seen.add(heading);
    blocks.push(
      `<p style="margin:0 0 8px;font-size:15px;font-weight:700;color:#444;">${escapeHtml(heading)}</p>`,
    );
    appendParagraphsFromHtml(match[2], blocks, seen);
    appendListsFromHtml(match[2], blocks, seen);
  }

  appendParagraphsFromHtml(contentHtml, blocks, seen);
  appendListsFromHtml(contentHtml, blocks, seen);
  appendBlockquotesFromHtml(contentHtml, blocks, seen);

  if (blocks.length === 0) return;
  sections.push('<section style="margin:0 0 20px;">');
  sections.push(...blocks);
  sections.push('</section>');
}

export function convertRichMindSpacePageHtmlToWechatArticle(
  html,
  { pageTitle = '', pageSummary = '', imageUrlMap = new Map() } = {},
) {
  const source = String(html ?? '');
  const title = (extractPageTitle(source) || pageTitle || 'TKMind 页面').slice(0, 64);
  const digest = (extractMetaDescription(source) || pageSummary || title).slice(0, 120);
  const sections = [];

  const heroTitle = stripHtml(
    source.match(/<header class="hero"[^>]*>[\s\S]*?<h1[^>]*>([\s\S]*?)<\/h1>/i)?.[1]
    ?? source.match(/<article[^>]*>[\s\S]*?<h1[^>]*>([\s\S]*?)<\/h1>/i)?.[1]
    ?? '',
  );
  const heroLead = inlineHtmlToWechatText(
    source.match(/<header class="hero"[^>]*>[\s\S]*?<p class="lead"[^>]*>([\s\S]*?)<\/p>/i)?.[1]
    ?? source.match(/<p class="pitch-sub"[^>]*>([\s\S]*?)<\/p>/i)?.[1]
    ?? '',
  );
  const titleBlockTitle = stripHtml(
    source.match(/<div class="title-block"[^>]*>[\s\S]*?<h1[^>]*class="title"[^>]*>([\s\S]*?)<\/h1>/i)?.[1]
    ?? source.match(/<h1[^>]*class="title"[^>]*>([\s\S]*?)<\/h1>/i)?.[1]
    ?? '',
  );
  const titleBlockSubtitle = stripHtml(
    source.match(/<div class="title-block"[^>]*>[\s\S]*?<p[^>]*class="subtitle"[^>]*>([\s\S]*?)<\/p>/i)?.[1] ?? '',
  );
  const displayTitle = heroTitle || titleBlockTitle || title;
  const displayLead = heroLead || titleBlockSubtitle;
  const coverRef = extractMindspaceCoverPath(source);
  const sidecarThumbRef = [...imageUrlMap.keys()].find((ref) => /\.thumbnail\.(png|svg)$/i.test(ref)) ?? '';
  const heroImageUrl = normalizeWechatImageUrl(
    imageUrlMap.get(coverRef)
    ?? (sidecarThumbRef ? imageUrlMap.get(sidecarThumbRef) : undefined)
    ?? imageUrlMap.values().next().value
    ?? '',
  );

  if (heroImageUrl) {
    sections.push(renderWechatImage(heroImageUrl, displayTitle));
  }

  sections.push([
    '<section style="text-align:center;padding:20px 14px 16px;background:linear-gradient(135deg,#2a1a12,#4a3020);color:#fff;border-radius:12px;margin:0 0 16px;">',
    `<p style="margin:0 0 8px;font-size:22px;font-weight:700;line-height:1.35;">${escapeHtml(displayTitle)}</p>`,
    displayLead
      ? `<p style="margin:0;font-size:14px;opacity:.88;line-height:1.8;">${displayLead.replace(/\n/g, '<br/>')}</p>`
      : '',
    '</section>',
  ].filter(Boolean).join(''));

  const summary = stripHtml(pageSummary);
  if (summary && summary !== stripHtml(displayLead) && summary !== displayTitle) {
    sections.push(
      `<p style="margin:0 0 16px;font-size:14px;color:#666;line-height:1.8;">${escapeHtml(summary)}</p>`,
    );
  }

  const proseBlocks = [...source.matchAll(/<div class="prose"[^>]*>([\s\S]*?)<\/div>/gi)];
  if (proseBlocks.length > 0) {
    sections.push('<section style="margin:0 0 20px;">');
    for (const [, block] of proseBlocks) {
      for (const [, paragraph] of block.matchAll(/<p[^>]*>([\s\S]*?)<\/p>/gi)) {
        const text = stripHtml(paragraph);
        if (!text) continue;
        sections.push(
          `<p style="margin:0 0 16px;font-size:16px;line-height:2;color:#5c3a1e;text-align:justify;">${escapeHtml(text)}</p>`,
        );
      }
    }
    sections.push('</section>');
  }

  const tagBlocks = [...source.matchAll(/<div class="tags"[^>]*>([\s\S]*?)<\/div>/gi)];
  if (tagBlocks.length > 0) {
    const tags = tagBlocks.flatMap(([, block]) =>
      [...block.matchAll(/<span class="tag"[^>]*>([\s\S]*?)<\/span>/gi)]
        .map(([, tag]) => stripHtml(tag))
        .filter(Boolean),
    );
    if (tags.length > 0) {
      sections.push(
        `<p style="margin:0 0 20px;font-size:13px;line-height:1.8;color:#8b6b4a;text-align:center;">${tags.map((tag) => escapeHtml(tag)).join(' · ')}</p>`,
      );
    }
  }

  const stanzaBlocks = [...source.matchAll(/<div class="stanza"[^>]*>([\s\S]*?)<\/div>/gi)];
  if (stanzaBlocks.length > 0) {
    sections.push('<section style="margin:0 0 20px;">');
    sections.push('<p style="margin:0 0 12px;font-size:13px;font-weight:600;color:#888;letter-spacing:.2em;">正文</p>');
    for (const [, block] of stanzaBlocks) {
      const num = stripHtml(block.match(/<span class="num"[^>]*>([\s\S]*?)<\/span>/i)?.[1] ?? '');
      const poem = inlineHtmlToWechatText(block.match(/<p[^>]*>([\s\S]*?)<\/p>/i)?.[1] ?? '');
      if (!poem) continue;
      sections.push([
        '<div style="margin:0 0 18px;padding:14px 12px;background:#faf7f1;border-radius:10px;">',
        num ? `<p style="margin:0 0 8px;font-size:11px;color:#d98f4a;letter-spacing:.24em;">${escapeHtml(num)}</p>` : '',
        `<p style="margin:0;font-size:16px;line-height:2;color:#2a1a12;white-space:pre-wrap;">${poem.replace(/\n/g, '<br/>')}</p>`,
        '</div>',
      ].filter(Boolean).join(''));
    }
    sections.push('</section>');
  }

  const finalText = inlineHtmlToWechatText(
    source.match(/<div class="final"[^>]*>([\s\S]*?)<\/div>/i)?.[1] ?? '',
  );
  if (finalText) {
    sections.push(
      `<p style="margin:0 0 20px;font-size:15px;line-height:1.9;color:#666;text-align:center;">${finalText.replace(/\n/g, '<br/>')}</p>`,
    );
  }

  const cardBlocks = [...source.matchAll(/<div class="card"[^>]*>([\s\S]*?)<\/div>/gi)];
  const imageryCards = cardBlocks.filter(([, block]) =>
    /<span class="glyph"/.test(block) || /<h3[^>]*>/.test(block),
  );
  if (imageryCards.length > 0) {
    sections.push('<section style="margin:0 0 20px;">');
    sections.push('<p style="margin:0 0 12px;font-size:13px;font-weight:600;color:#888;letter-spacing:.2em;">意象</p>');
    for (const [, block] of imageryCards) {
      const glyph = stripHtml(block.match(/<span class="glyph"[^>]*>([\s\S]*?)<\/span>/i)?.[1] ?? '');
      const cardTitle = stripHtml(block.match(/<h3[^>]*>([\s\S]*?)<\/h3>/i)?.[1] ?? '');
      const cardBody = stripHtml(block.match(/<p[^>]*>([\s\S]*?)<\/p>/i)?.[1] ?? '');
      sections.push([
        '<div style="margin:0 0 14px;padding:16px;background:#fff;border:1px solid #eee;border-radius:10px;">',
        glyph ? `<p style="margin:0 0 8px;font-size:24px;color:#d98f4a;">${escapeHtml(glyph)}</p>` : '',
        cardTitle ? `<p style="margin:0 0 8px;font-size:15px;font-weight:700;color:#2a1a12;">${escapeHtml(cardTitle)}</p>` : '',
        cardBody ? `<p style="margin:0;font-size:14px;line-height:1.8;color:#666;">${escapeHtml(cardBody)}</p>` : '',
        '</div>',
      ].filter(Boolean).join(''));
    }
    sections.push('</section>');
  }

  const noteBlock = source.match(/<section class="note"[^>]*>([\s\S]*?)<\/section>/i)?.[1] ?? '';
  if (noteBlock) {
    sections.push('<section style="margin:0 0 8px;">');
    sections.push('<p style="margin:0 0 12px;font-size:13px;font-weight:600;color:#888;letter-spacing:.2em;">创作手记</p>');
    const noteTitle = stripHtml(noteBlock.match(/<h2[^>]*>([\s\S]*?)<\/h2>/i)?.[1] ?? '');
    if (noteTitle) {
      sections.push(`<p style="margin:0 0 12px;font-size:16px;font-weight:700;color:#2a1a12;">${escapeHtml(noteTitle)}</p>`);
    }
    for (const [, paragraph] of noteBlock.matchAll(/<p[^>]*>([\s\S]*?)<\/p>/gi)) {
      const text = stripHtml(paragraph);
      if (!text) continue;
      sections.push(`<p style="margin:0 0 12px;font-size:14px;line-height:1.9;color:#555;">${escapeHtml(text)}</p>`);
    }
    const quote = stripHtml(noteBlock.match(/<div class="quote"[^>]*>([\s\S]*?)<\/div>/i)?.[1] ?? '');
    if (quote) {
      sections.push(
        `<blockquote style="margin:0;padding:12px 16px;border-left:3px solid #d98f4a;background:#faf7f1;color:#2a1a12;font-size:15px;line-height:1.8;">${escapeHtml(quote)}</blockquote>`,
      );
    }
    sections.push('</section>');
  }

  const specializedContentExtracted =
    proseBlocks.length > 0
    || stanzaBlocks.length > 0
    || Boolean(finalText)
    || imageryCards.length > 0
    || Boolean(noteBlock);
  if (!specializedContentExtracted) {
    appendFallbackMindspaceArticleSections(sections, source, {
      skipTexts: new Set([displayTitle, displayLead, summary].filter(Boolean)),
    });
  }

  for (const match of source.matchAll(/<img[^>]+src=["']([^"']+)["']/gi)) {
    const ref = normalizeImageRef(match[1]);
    const uploaded = normalizeWechatImageUrl(imageUrlMap.get(ref));
    if (uploaded && uploaded !== heroImageUrl) {
      sections.push(renderWechatImage(uploaded));
    }
  }

  return {
    title: displayTitle.slice(0, 64),
    digest,
    content: sections.join('\n'),
    imageCount: imageUrlMap.size,
  };
}

export function resolveWechatDraftProfile(html) {
  const source = String(html ?? '');
  if (isDailyNewsFormat(source)) return 'daily-news';
  if (isNewsHotspotFormat(source)) return 'news-hotspot';
  if (/name=["']mindspace-cover["']/i.test(source)) return 'mindspace-page';
  if (/data-mindspace-page-tag=["']platform-brand["']/i.test(source)) return 'mindspace-page';
  if (/class="stanza"/.test(source) || /class="hero"/.test(source)) return 'mindspace-page';
  return 'generic';
}

export function validatePageForWechatDraft(
  html,
  { pageTitle = '', pageSummary = '', publicUrl = '' } = {},
) {
  const source = String(html ?? '').trim();
  const profile = resolveWechatDraftProfile(source);
  const warnings = [];
  const blockers = [];

  if (!source) {
    blockers.push('页面 HTML 为空，无法推送草稿。');
  }
  if (!stripHtml(pageTitle || extractPageTitle(source))) {
    blockers.push('缺少页面标题。');
  }
  if (!publicUrl) {
    warnings.push('页面尚未公开或缺少阅读原文链接，草稿中将省略「阅读原文」。');
  }
  if (countBodyImages(source) === 0 && profile === 'mindspace-page') {
    warnings.push('页面未检测到可上传图片，草稿将以文字为主。');
  }
  if (profile === 'mindspace-page') {
    warnings.push('MindSpace 页面结构转换后视觉可能与原页略有差异。');
  }

  return {
    ok: blockers.length === 0,
    profile,
    warnings,
    blockers,
  };
}

async function uploadPageImagesForWechat(
  accessToken,
  html,
  { publishDir = '', htmlRelativePath = '', wechatFetch = undiciFetch } = {},
) {
  const imageUrlMap = new Map();
  const candidates = collectPageImageCandidates(html, { publishDir, htmlRelativePath });
  for (const { ref, absPath } of candidates) {
    if (!absPath || imageUrlMap.has(ref)) continue;
    if (/^https?:\/\//i.test(absPath)) {
      imageUrlMap.set(ref, absPath);
      continue;
    }
    let buffer = fs.readFileSync(absPath);
    if (/\.(webp|svg)$/i.test(absPath)) {
      buffer = await sharp(buffer).png().toBuffer();
    }
    const uploaded = normalizeWechatImageUrl(
      await uploadWechatArticleContentImage(
        accessToken,
        buffer,
        { wechatFetch, filename: `${path.basename(absPath).replace(/\.(webp|svg)$/i, '')}.png` },
      ),
    );
    imageUrlMap.set(ref, uploaded);
  }
  return imageUrlMap;
}

export async function buildMindSpacePageWechatDraftArticleForPush({
  html,
  publicUrl = '',
  pageTitle = '',
  pageSummary = '',
  accessToken = null,
  publishDir = '',
  htmlRelativePath = '',
  wechatFetch = undiciFetch,
  memindLibRoot = process.cwd(),
} = {}) {
  const imageUrlMap = accessToken
    ? await uploadPageImagesForWechat(accessToken, html, { publishDir, htmlRelativePath, wechatFetch })
    : new Map();
  const profile = resolveWechatDraftProfile(html);
  let article;
  if (profile === 'news-hotspot') {
    article = convertNewsPageHtmlToWechatArticle(html, { publicUrl });
    article.content = stripLegacyReadOriginalFooter(article.content);
  } else if (profile === 'mindspace-page') {
    article = convertRichMindSpacePageHtmlToWechatArticle(html, {
      pageTitle,
      pageSummary,
      imageUrlMap,
    });
  } else {
    article = convertGenericPageHtmlToWechatArticle(html, {
      pageTitle,
      pageSummary,
    });
    article.content = stripLegacyReadOriginalFooter(article.content);
  }
  const finalized = await finalizeWechatDraftArticleForPush({
    content: article.content,
    publicUrl,
    accessToken,
    wechatFetch,
    memindLibRoot,
  });
  return {
    ...article,
    content: finalized.content,
    contentSourceUrl: publicUrl || undefined,
    uploadedImageCount: imageUrlMap.size,
  };
}

export async function buildWechatDraftPublicationBundleForPush({
  html,
  publicUrl = '',
  pageTitle = '',
  pageSummary = '',
  accessToken = null,
  publishDir = '',
  htmlRelativePath = '',
  h5Root = process.cwd(),
  userId = '',
  workspaceRelativePath = htmlRelativePath,
  wechatFetch = undiciFetch,
  memindLibRoot = process.cwd(),
} = {}) {
  const profile = resolveWechatDraftProfile(html);
  const article = isDailyNewsFormat(html)
    ? await buildDailyNewsWechatDraftArticleForPush({
        html,
        publicUrl,
        accessToken,
        wechatFetch,
        memindLibRoot,
        publishDir,
      })
    : await buildMindSpacePageWechatDraftArticleForPush({
        html,
        publicUrl,
        pageTitle,
        pageSummary,
        accessToken,
        publishDir,
        htmlRelativePath,
        wechatFetch,
        memindLibRoot,
      });
  const thumbResolved = resolveWechatDraftThumbPath(html, {
    h5Root,
    userId,
    workspaceRelativePath,
    publishDir,
  });
  const contentCheck = verifyWechatDraftPublicationContent(article.content, { publicUrl });
  const coverCheck = verifyWechatDraftCoverResolution(thumbResolved);
  if (!contentCheck.ok || !coverCheck.ok) {
    throw Object.assign(
      new Error(`草稿不符合推送标准 ${WECHAT_DRAFT_PUBLICATION_STANDARD_VERSION}`),
      {
        code: 'wechat_draft_standard_violation',
        contentIssues: contentCheck.issues,
        coverIssues: coverCheck.issues,
      },
    );
  }
  return {
    article,
    thumbResolved,
    profile,
    standardVersion: WECHAT_DRAFT_PUBLICATION_STANDARD_VERSION,
  };
}

export function convertGenericPageHtmlToWechatArticle(
  html,
  { publicUrl = '', pageTitle = '', pageSummary = '' } = {},
) {
  const title = (extractPageTitle(html) || pageTitle || 'TKMind 页面').slice(0, 64);
  const digest = (extractMetaDescription(html) || pageSummary || title).slice(0, 120);
  const sections = [];
  const summary = stripHtml(pageSummary);
  if (summary) {
    sections.push(
      `<p style="margin:0 0 16px;font-size:14px;color:#666;line-height:1.8;">${escapeHtml(summary)}</p>`,
    );
  }
  appendFallbackMindspaceArticleSections(sections, html, {
    skipTexts: new Set([title, summary].filter(Boolean)),
  });
  if (sections.length === (summary ? 1 : 0)) {
    sections.push(
      `<p style="margin:0;font-size:15px;line-height:1.8;color:#333;">${escapeHtml(title)}</p>`,
    );
  }
  return {
    title,
    digest,
    content: sections.join('\n').slice(0, 20000),
    contentSourceUrl: publicUrl || undefined,
  };
}

export function convertPageHtmlToWechatArticle(
  html,
  { publicUrl = '', pageTitle = '', pageSummary = '' } = {},
) {
  if (isDailyNewsFormat(html)) {
    return convertNewsPageHtmlToWechatArticle(html, { publicUrl });
  }
  if (isNewsHotspotFormat(html)) {
    return convertNewsPageHtmlToWechatArticle(html, { publicUrl });
  }
  if (resolveWechatDraftProfile(html) === 'mindspace-page') {
    return convertRichMindSpacePageHtmlToWechatArticle(html, { pageTitle, pageSummary });
  }
  return convertGenericPageHtmlToWechatArticle(html, { publicUrl, pageTitle, pageSummary });
}

async function ensureRunsTable(pool) {
  await pool.query(`
    CREATE TABLE IF NOT EXISTS ${RUNS_TABLE} (
      id CHAR(36) PRIMARY KEY,
      user_id CHAR(36) NOT NULL,
      page_id CHAR(36) NOT NULL,
      status VARCHAR(16) NOT NULL,
      page_title VARCHAR(255) NULL,
      page_url VARCHAR(512) NULL,
      draft_media_id VARCHAR(128) NULL,
      error_message TEXT NULL,
      triggered_by CHAR(36) NULL,
      created_at BIGINT NOT NULL,
      KEY idx_wechat_page_draft_user (user_id, created_at),
      KEY idx_wechat_page_draft_page (page_id, created_at)
    ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci
  `);
}

function cryptoRandomId() {
  return globalThis.crypto?.randomUUID?.() ?? `${Date.now()}-${Math.random().toString(16).slice(2)}`;
}

function resolveThumbPath(h5Root, userId, workspaceRelativePath) {
  if (!h5Root || !userId || !workspaceRelativePath) return null;
  const parts = String(workspaceRelativePath)
    .replace(/^\/+/, '')
    .split('/')
    .filter(Boolean);
  const baseParts = parts[0]?.toLowerCase() === PUBLIC_ZONE_DIR
    ? parts
    : [PUBLIC_ZONE_DIR, ...parts];
  const htmlPath = path.join(
    h5Root,
    PUBLISH_ROOT_DIR,
    String(userId),
    ...baseParts,
  );
  const pngPath = htmlPath.replace(/\.html$/i, '.thumbnail.png');
  if (fs.existsSync(pngPath)) return pngPath;
  const svgPath = htmlPath.replace(/\.html$/i, '.thumbnail.svg');
  if (fs.existsSync(svgPath)) return svgPath;
  return null;
}

function isWechatDraftDecorativeImageRef(ref = '') {
  return /mp-follow-qrcode|tkmind-icon|\/brand\//i.test(String(ref ?? ''));
}

export function resolveHeroImagePath(html, { publishDir = '', htmlRelativePath = '' } = {}) {
  const coverRef = extractMindspaceCoverPath(html);
  if (coverRef && !/\.svg$/i.test(coverRef)) {
    const coverPath = resolveImageFilePath(coverRef, publishDir, htmlRelativePath);
    if (coverPath && !/\.svg$/i.test(coverPath)) {
      return coverPath;
    }
  }
  const candidates = collectPageImageCandidatesFromHtml(html, { publishDir, htmlRelativePath });
  for (const { ref, absPath } of candidates) {
    if (!absPath || /\.svg$/i.test(absPath) || isWechatDraftDecorativeImageRef(ref)) {
      continue;
    }
    return absPath;
  }
  return null;
}

function resolveRasterThumbSidecarPath(h5Root, userId, workspaceRelativePath) {
  const sidecarPath = resolveThumbPath(h5Root, userId, workspaceRelativePath);
  if (sidecarPath && /\.(png|svg)$/i.test(sidecarPath)) {
    return sidecarPath;
  }
  return null;
}

export function resolveWechatDraftThumbPath(
  html,
  { h5Root, userId, workspaceRelativePath, publishDir } = {},
) {
  const heroPath = resolveHeroImagePath(html, {
    publishDir,
    htmlRelativePath: workspaceRelativePath,
  });
  if (heroPath) {
    return { source: 'hero', path: heroPath };
  }
  const sidecarPath = resolveRasterThumbSidecarPath(h5Root, userId, workspaceRelativePath);
  if (sidecarPath) {
    return { source: 'thumbnail-sidecar', path: sidecarPath };
  }
  return { source: 'placeholder', path: null };
}

async function loadThumbBuffer(thumbPath, { wechatFetch = undiciFetch } = {}) {
  if (!thumbPath) {
    return sharp(
      Buffer.from(
        '<svg xmlns="http://www.w3.org/2000/svg" width="900" height="900"><rect width="900" height="900" fill="#1a1a2e"/><text x="50%" y="50%" fill="#fff" font-size="42" text-anchor="middle" dominant-baseline="middle">TKMind</text></svg>',
      ),
    )
      .png()
      .toBuffer();
  }

  let raw;
  if (/^https?:\/\//i.test(thumbPath)) {
    const response = await wechatFetch(thumbPath);
    if (!response.ok) {
      throw new Error(`封面图片下载失败: ${thumbPath}`);
    }
    raw = Buffer.from(await response.arrayBuffer());
  } else {
    raw = fs.readFileSync(thumbPath);
  }

  return sharp(raw).resize(900, 900, { fit: 'cover' }).png().toBuffer();
}

export function resolvePagePublicUrl(
  page,
  {
    userId = '',
    workspaceRelativePath = '',
    publication = null,
    publicBaseUrl = '',
  } = {},
) {
  const resolved = String(
    publication?.publicUrl
    ?? publication?.public_url
    ?? page?.publication?.publicUrl
    ?? page?.publicationUrl
    ?? page?.workspacePublicUrl
    ?? '',
  ).trim();
  if (resolved) return resolved;
  const relativePath = String(
    workspaceRelativePath
    ?? page?.workspaceRelativePath
    ?? resolvePageWorkspaceRelativePath(page)
    ?? '',
  ).trim();
  if (relativePath && userId && publicBaseUrl) {
    return buildPublicUrl(publicBaseUrl, userId, relativePath);
  }
  return '';
}

export function createMindSpaceWechatPageDraftService(
  pool,
  {
    getMindSpacePages = () => null,
    getMindSpacePublications = () => null,
    getWechatMpConfig = () => null,
    h5Root = process.cwd(),
    memindLibRoot = h5Root,
    env = process.env,
    tokenUrl = env.H5_WECHAT_MP_TOKEN_URL?.trim() || DEFAULT_TOKEN_URL,
    wechatFetch = undiciFetch,
  } = {},
) {
  let ensurePromise = null;
  const accessTokenCache = new Map();

  async function ensureReady() {
    if (!ensurePromise) {
      ensurePromise = ensureRunsTable(pool);
    }
    await ensurePromise;
  }

  async function getAccessToken(userId) {
    const configService = getWechatMpConfig?.();
    if (!configService) throw new Error('公众号配置服务未启用');
    const credentials = await configService.getCredentials(userId);
    if (!credentials?.appId || !credentials?.appSecret) {
      throw Object.assign(new Error('请先在 M 配置中填写公众号 AppID 和 AppSecret'), {
        code: 'wechat_mp_not_configured',
      });
    }
    const cacheKey = `${userId}:${credentials.appId}`;
    const cached = accessTokenCache.get(cacheKey);
    if (cached && cached.expiresAt > Date.now() + 60_000) {
      return { accessToken: cached.token, credentials };
    }
    const token = await fetchWechatMpAccessToken(credentials, { tokenUrl, wechatFetch });
    accessTokenCache.set(cacheKey, {
      token: token.accessToken,
      expiresAt: Date.now() + token.expiresIn * 1000,
    });
    return { accessToken: token.accessToken, credentials };
  }

  async function recordRun(payload) {
    await ensureReady();
    const id = cryptoRandomId();
    const now = Date.now();
    await pool.query(
      `INSERT INTO ${RUNS_TABLE}
        (id, user_id, page_id, status, page_title, page_url, draft_media_id, error_message, triggered_by, created_at)
       VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
      [
        id,
        payload.userId,
        payload.pageId,
        payload.status,
        payload.pageTitle ?? null,
        payload.pageUrl ?? null,
        payload.draftMediaId ?? null,
        payload.errorMessage ?? null,
        payload.triggeredBy ?? null,
        now,
      ],
    );
    return {
      id,
      status: payload.status,
      pageId: payload.pageId,
      pageTitle: payload.pageTitle ?? null,
      pageUrl: payload.pageUrl ?? null,
      draftMediaId: payload.draftMediaId ?? null,
      errorMessage: payload.errorMessage ?? null,
      triggeredBy: payload.triggeredBy ?? null,
      createdAt: now,
    };
  }

  async function resolvePageHtml(userId, pageId) {
    const pages = getMindSpacePages?.();
    if (!pages?.renderPreview || !pages?.getPage) {
      throw new Error('MindSpace 页面服务未启用');
    }
    const page = await pages.getPage(userId, pageId);
    const { html } = await pages.renderPreview(userId, pageId);
    const workspaceRelativePath = resolvePageWorkspaceRelativePath(page)
      ?? page.workspaceRelativePath
      ?? '';
    const publications = getMindSpacePublications?.();
    const publication = publications?.getCurrent
      ? await publications.getCurrent(userId, pageId).catch(() => null)
      : null;
    const publicBaseUrl = resolvePublicBaseUrl(env);
    const publicUrl = resolvePagePublicUrl(page, {
      userId,
      workspaceRelativePath,
      publication,
      publicBaseUrl,
    });
    let sourceHtml = html;
    if (workspaceRelativePath) {
      const workspaceHtml = await readWorkspacePublishHtml(h5Root, userId, workspaceRelativePath);
      if (workspaceHtml?.trim()) {
        sourceHtml = workspaceHtml;
      }
    }
    return {
      page,
      html: sourceHtml,
      publicUrl,
      workspaceRelativePath,
    };
  }

  return {
    async pushPageDraft(userId, pageId, { triggeredBy = null } = {}) {
      let preview = null;
      try {
        preview = await resolvePageHtml(userId, pageId);
        const { accessToken, credentials } = await getAccessToken(userId);
        const publishDir = path.join(h5Root, PUBLISH_ROOT_DIR, String(userId));
        const bundle = await buildWechatDraftPublicationBundleForPush({
          html: preview.html,
          publicUrl: preview.publicUrl,
          pageTitle: preview.page.title,
          pageSummary: preview.page.summary,
          accessToken,
          publishDir,
          htmlRelativePath: preview.workspaceRelativePath ?? '',
          h5Root,
          userId,
          workspaceRelativePath: preview.workspaceRelativePath,
          wechatFetch,
          memindLibRoot,
        });
        const { article, thumbResolved } = bundle;
        const thumbBuffer = await loadThumbBuffer(thumbResolved.path, { wechatFetch });
        const thumbMediaId = await uploadWechatPermanentThumb(accessToken, thumbBuffer, { wechatFetch });
        const draft = await addWechatDraftArticle(
          accessToken,
          {
            title: article.title,
            author: credentials.author || 'TKMind',
            digest: article.digest,
            content: article.content,
            content_source_url: article.contentSourceUrl,
            thumb_media_id: thumbMediaId,
            need_open_comment: 0,
            only_fans_can_comment: 0,
          },
          { wechatFetch },
        );
        const run = await recordRun({
          userId,
          pageId,
          status: 'success',
          pageTitle: preview.page.title,
          pageUrl: preview.publicUrl,
          draftMediaId: draft.draftMediaId,
          triggeredBy,
        });
        return {
          ok: true,
          draftMediaId: draft.draftMediaId,
          pageTitle: preview.page.title,
          pageUrl: preview.publicUrl,
          standardVersion: bundle.standardVersion,
          coverSource: thumbResolved.source,
          run,
        };
      } catch (error) {
        const message = error instanceof Error ? error.message : String(error);
        const run = await recordRun({
          userId,
          pageId,
          status: 'failed',
          pageTitle: preview?.page?.title ?? null,
          pageUrl: preview?.publicUrl ?? null,
          errorMessage: message,
          triggeredBy,
        });
        throw Object.assign(new Error(message), {
          code: error?.code ?? 'wechat_draft_push_failed',
          run,
        });
      }
    },

    async listRuns(userId, { pageId = null, limit = 20 } = {}) {
      await ensureReady();
      const safeLimit = Math.min(Math.max(Number(limit) || 20, 1), 100);
      const params = [userId];
      let sql = `
        SELECT id, page_id AS pageId, status, page_title AS pageTitle, page_url AS pageUrl,
               draft_media_id AS draftMediaId, error_message AS errorMessage,
               triggered_by AS triggeredBy, created_at AS createdAt
          FROM ${RUNS_TABLE}
         WHERE user_id = ?`;
      if (pageId) {
        sql += ' AND page_id = ?';
        params.push(pageId);
      }
      sql += ' ORDER BY created_at DESC LIMIT ?';
      params.push(safeLimit);
      const [rows] = await pool.query(sql, params);
      return rows.map((row) => ({
        ...row,
        createdAt: Number(row.createdAt ?? 0),
      }));
    },
  };
}

export const mindspaceWechatPageDraftInternals = {
  convertGenericPageHtmlToWechatArticle,
  convertPageHtmlToWechatArticle,
  convertRichMindSpacePageHtmlToWechatArticle,
  collectPageImageCandidates,
  isNewsHotspotFormat,
  resolveWechatDraftThumbPath,
  resolveHeroImagePath,
};