memind/wechat/verify/page-artifact.mjs

import fs from 'node:fs';
import path from 'node:path';
import { verifyHtmlImageSourcesAllowed } from '../../chat-image-materialize.mjs';

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

export function resolveArtifactLocalPath(artifact, publishDir = '') {
  const localPath = String(artifact?.localPath ?? '').trim();
  if (localPath) return localPath;
  const relativePath = String(artifact?.relativePath ?? '').trim();
  const root = String(publishDir ?? '').trim();
  if (!relativePath || !root) return '';
  const resolved = path.resolve(root, relativePath);
  const normalizedRoot = path.resolve(root);
  if (
    resolved !== normalizedRoot &&
    !resolved.startsWith(`${normalizedRoot}${path.sep}`)
  ) {
    return '';
  }
  return resolved;
}

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

export function artifactFileExists(artifact, { publishDir = '' } = {}) {
  const localPath = resolveArtifactLocalPath(artifact, publishDir);
  if (localPath) {
    try {
      return fs.existsSync(localPath) && fs.statSync(localPath).isFile();
    } catch {
      return false;
    }
  }
  if (typeof artifact?.exists === 'boolean') {
    return artifact.exists;
  }
  return false;
}

export function isStubPublicHtmlArtifact(artifact, { publishDir = '' } = {}) {
  if (typeof artifact?.isStub === 'boolean') {
    return artifact.isStub;
  }
  const localPath = resolveArtifactLocalPath(artifact, publishDir);
  if (!localPath) return false;
  try {
    return isStubPublicHtmlContent(fs.readFileSync(localPath, 'utf8'));
  } catch {
    return false;
  }
}

/** Real HTML artifacts only — never fall back to stub placeholders. */
export function filterSendableHtmlArtifacts(artifacts = [], options = {}) {
  return artifacts.filter(
    (artifact) =>
      artifactFileExists(artifact, options) &&
      !isStubPublicHtmlArtifact(artifact, options),
  );
}

export function selectSendableHtmlArtifacts({
  verifiedArtifacts = [],
  confirmedArtifacts = [],
  publishDir = '',
} = {}) {
  const options = { publishDir };
  const candidates = verifiedArtifacts.length > 0 ? verifiedArtifacts : confirmedArtifacts;
  return filterSendableHtmlArtifacts(candidates, options);
}

export function verifyPageArtifactContent(artifact, { minBytes = 512, publishDir = '' } = {}) {
  const localPath = resolveArtifactLocalPath(artifact, publishDir);
  const resolvedArtifact = localPath ? { ...artifact, localPath } : artifact;
  if (!artifactFileExists(resolvedArtifact, { publishDir })) {
    return { ok: false, reason: 'missing_file' };
  }
  if (isStubPublicHtmlArtifact(resolvedArtifact, { publishDir })) {
    return { ok: false, reason: 'stub_placeholder' };
  }
  if (!localPath) {
    if (Number(artifact?.sizeBytes ?? 0) < minBytes) {
      return {
        ok: false,
        reason: 'too_small',
      };
    }
    if (artifact?.isHtmlDocument !== true) {
      return {
        ok: false,
        reason: 'not_html_document',
      };
    }
    return { ok: true, reason: null };
  }
  const size = fs.statSync(localPath).size;
  if (size < minBytes) {
    return { ok: false, reason: 'too_small' };
  }
  const content = fs.readFileSync(localPath, 'utf8');
  if (!/<(?:html|body|main|article)\b/i.test(content)) {
    return { ok: false, reason: 'not_html_document' };
  }
  return { ok: true, reason: null };
}

export function verifyPageArtifactImageSources(
  artifact,
  {
    publishDir = '',
    allowedImageEmbedKeys = null,
  } = {},
) {
  const allowed = allowedImageEmbedKeys instanceof Set
    ? allowedImageEmbedKeys
    : Array.isArray(allowedImageEmbedKeys)
      ? new Set(allowedImageEmbedKeys)
      : null;
  if (!allowed || allowed.size === 0) {
    return { ok: true, reason: null, offendingKeys: [] };
  }
  const localPath = resolveArtifactLocalPath(artifact, publishDir);
  if (!localPath) {
    return { ok: false, reason: 'missing_file', offendingKeys: [] };
  }
  let content = '';
  try {
    content = fs.readFileSync(localPath, 'utf8');
  } catch {
    return { ok: false, reason: 'missing_file', offendingKeys: [] };
  }
  const verification = verifyHtmlImageSourcesAllowed(content, allowed);
  return {
    ok: verification.ok,
    reason: verification.reason,
    offendingKeys: verification.offendingKeys ?? [],
  };
}