diff --git a/docs/news-morning-templates.md b/docs/news-morning-templates.md
index 6202071..7e839e1 100644
--- a/docs/news-morning-templates.md
+++ b/docs/news-morning-templates.md
@@ -39,7 +39,7 @@ DB 值优先于环境变量。非法值一律回落到 `news001`。
- 每张卡片必须带 `data-event-key="{事件指纹}"`，指纹 = 核心主体 + 核心动作，
中英文报道共用同一指纹，全页不得重复。
-### 2. 运行时去重闸门
+### 2. 运行时去重
prompt 层的「禁止重复」无法验证，所以在落盘后再做一次程序化清理：
[`wechat-news-morning-dedup.mjs`](../wechat-news-morning-dedup.mjs)，由 worker 在生成成功后调用。
@@ -102,9 +102,15 @@ news002 的卡片图会先上传到微信正文素材接口，再由 inline 渲
静态页与微信草稿均使用 **左上角 72～92px 小图**（`.card` / `.lead-card` 左侧缩略图），不再用全宽大图。
能拿到真实配图的卡片必须显示并上传；没有真实图就只保留文字卡，不放默认小图。
远程配图会先下载到当日 MindSpace `public/`，改写成本站相对 URL，避免海外图源在国内打不开。
-推送前不再删除无图故事卡（微信草稿保留纯文字卡）；有图卡必须成功上传微信 CDN。
-生成闸门会拦截：配图上传失败、头条不是 3～4 条、事件卡超过 30 条，或页面仍混有 news001 DOM，
-才会由生成/推送闸门阻断，不允许进入草稿箱。
+推送前不再删除无图故事卡（微信草稿保留纯文字卡）；有图卡未能上传微信 CDN 时只去掉该图，卡片保留文字。
+
+配图 / 结构 / 写满检查**只引导修复、不阻断生成和推送**：
+检查不过时依次做 SearXNG 补位、补图、头条规整；仍不满足就删掉完全空的栏目，
+以现有内容出稿并照常推草稿，同时记 `news002 audit incomplete, publishing best-effort page` 告警。
+
+强制重生成（`forceGenerateOnce`）在执行前就清掉标记，只跑一次、失败不重试；
+生成失败时把归档的原稿放回原位，当日页面不会因重生成失败而 404。
+降级页自动重生成每天最多一次。
真实图片来源按优先级取 SearXNG 结果里的 `img_src`/`thumbnail`，以及文章来源的
`og:image:secure_url`、`og:image`、`twitter:image`；不使用与事件无法核验关系的泛图片搜索结果。
diff --git a/news-content-safety-rules.mjs b/news-content-safety-rules.mjs
index e44c710..69f9e53 100644
--- a/news-content-safety-rules.mjs
+++ b/news-content-safety-rules.mjs
@@ -48,6 +48,7 @@ function rx(source, flags = 'iu') {
const TERM_RULES = Object.freeze([
// 成人低俗
{ id: 'adult-explicit', category: 'adult', level: 'block', re: rx('色情(?:网站|视频|图片|直播|片)|黄色(?:网站|视频|小说)|黄网|黄片|成人(?:网站|视频|直播|电影)|AV女优|裸聊|约炮|援交|一夜情|性爱视频|艳照|露点(?:照|视频|写真)|福利姬|情色|擦边(?:视频|直播|写真|女主播)|porn|onlyfans|nsfw|xvideos') },
+ { id: 'adult-vulgar', category: 'adult', level: 'block', re: rx('鸡巴|几把|肉棒|屌大|操逼|草逼|骚逼|口交|肛交|乳交|群交|乱伦|啪啪啪|自慰视频|无码(?:视频|片|高清)|熟女|海角(?:网|社区)|黑料(?:网|社|不打烊)|(?:51|91)(?:吃瓜|黑料)|吃瓜(?:网|黑料)') },
{ id: 'adult-sexual-crime', category: 'adult', level: 'review', re: rx('性侵|强奸|猥亵|性骚扰|卖淫|嫖娼|淫秽|裸照|不雅视频|偷拍') },
// 自杀自残
@@ -131,7 +132,7 @@ const SELF_MEDIA_DOMAINS = Object.freeze([
'sohu.com/a/',
]);
-const UNSAFE_DOMAIN_RE = /(?:porn|xxx|sex|casino|bet365|gamble|18av|hentai|jav)/i;
+const UNSAFE_DOMAIN_RE = /(?:porn|xxx|sex|casino|bet365|gamble|18av|hentai|jav|haijiao|heiliao|51cg|chigua)/i;
const POLITICAL_RE = rx(POLITICAL_TERMS);
const POLITICAL_SECTION_IDS = new Set(['policy', 'geopolitics', 'politics']);
const POLITICAL_TAG_RE = /时政|政策|政务|外交/u;
diff --git a/news-content-safety-rules.test.mjs b/news-content-safety-rules.test.mjs
index 1750568..f885b68 100644
--- a/news-content-safety-rules.test.mjs
+++ b/news-content-safety-rules.test.mjs
@@ -17,6 +17,12 @@ test('adult explicit content is blocked, sexual-crime news goes to review', () =
assert.equal(levelOf('91每日大赛-今日吃瓜爆料资讯平台，热门黑料绯闻每日更新'), 'block');
assert.equal(levelOf('男子强奸案一审宣判，被判有期徒刑十年'), 'review');
assert.equal(levelOf('央行宣布下调存款准备金率 0.5 个百分点'), 'clean');
+ assert.equal(levelOf('AI短剧王鹤棣的3厘米鸡巴第1集 打玻尿酸坏死只剩3cm... | 海角网'), 'block');
+ assert.equal(levelOf('熟女 - 热门标签，每日更新最新内容'), 'block');
+ assert.equal(levelOf('51吃瓜 今日最新'), 'block');
+ assert.equal(assessNewsSourceUrl('https://www.haijiao.com/post/123').level, 'block');
+ assert.equal(assessNewsSourceUrl('https://heiliao.com/archives/1').level, 'block');
+ assert.equal(levelOf('海角七号重映，票房破亿'), 'clean');
});
test('common news phrases are not false positives', () => {
diff --git a/wechat-news-morning-draft-worker.mjs b/wechat-news-morning-draft-worker.mjs
index 0413e2f..4a2dca0 100644
--- a/wechat-news-morning-draft-worker.mjs
+++ b/wechat-news-morning-draft-worker.mjs
@@ -30,6 +30,7 @@ import {
enrichNews002HtmlWithSourceImages,
localizeNews002CardImages,
promoteNews002LeadCards,
+ pruneEmptyNews002Sections,
} from './wechat-news-morning-images.mjs';
import {
isWechatNewsMorningDraftWorkerEnabled,
@@ -119,9 +120,7 @@ async function applyNews002ImageGate(
{ searchGroups = [], safetyExemptSectionIds = [] } = {},
) {
if (config?.templateId !== 'news002') return null;
- if (!page?.localPath || !fs.existsSync(page.localPath)) {
- throw new Error('news002 配图闸门无法读取生成页面');
- }
+ if (!page?.localPath || !fs.existsSync(page.localPath)) return null;
let source = fs.readFileSync(page.localPath, 'utf8');
const normalizedBefore = promoteNews002LeadCards(source);
if (normalizedBefore.promotedCount > 0 || normalizedBefore.demotedCount > 0) {
@@ -269,6 +268,20 @@ async function applyNews002ImageGate(
}
}
+ if (!audit.ok) {
+ const pruned = pruneEmptyNews002Sections(localized.html, { requireImage: false });
+ if (pruned.removedSections > 0) {
+ localized = { ...localized, html: pruned.html };
+ fs.writeFileSync(page.localPath, localized.html);
+ audit = auditNews002ImageCoverage(localized.html, {
+ requireTemplateStructure: true,
+ requireSectionFill: true,
+ exemptSectionIds: safetyExemptSectionIds,
+ env,
+ });
+ }
+ }
+
logger.log?.('[NewsMorningDraft] news002 image audit', {
slug: page.slug,
attempted: enriched.attemptedCount,
@@ -278,12 +291,15 @@ async function applyNews002ImageGate(
...audit,
});
if (!audit.ok) {
- throw new Error(
- `news002 配图/结构/写满闸门未通过：${audit.violations.join(', ')}`
- + `（卡片 ${audit.cardCount}，真实配图 ${audit.imageCount}，头条配图 ${audit.leadImageCount}/${audit.leadCount}）`,
- );
+ logger.warn?.('[NewsMorningDraft] news002 audit incomplete, publishing best-effort page', {
+ slug: page.slug,
+ violations: audit.violations,
+ cardCount: audit.cardCount,
+ imageCount: audit.imageCount,
+ leadImages: `${audit.leadImageCount}/${audit.leadCount}`,
+ });
}
- return audit;
+ return { ...audit, degraded: !audit.ok };
}
function archiveTodayPageForForce(config, h5Root, now, logger) {
@@ -292,15 +308,47 @@ function archiveTodayPageForForce(config, h5Root, now, logger) {
const stamp = new Date(now).toISOString().replace(/[:.]/g, '-');
const archived = path.join(path.dirname(page.localPath), `_archived-${page.slug}-${stamp}.html`);
fs.renameSync(page.localPath, archived);
+ let archivedThumb = null;
if (page.thumbPath && fs.existsSync(page.thumbPath)) {
- const archivedThumb = archived.replace(/\.html$/i, '.thumbnail.png');
+ archivedThumb = archived.replace(/\.html$/i, '.thumbnail.png');
fs.renameSync(page.thumbPath, archivedThumb);
}
logger.log?.('[NewsMorningDraft] archived today page before force generate', {
from: page.relativePath,
to: path.basename(archived),
});
- return archived;
+ return {
+ originalPath: page.localPath,
+ archivedPath: archived,
+ originalThumbPath: archivedThumb ? page.thumbPath : null,
+ archivedThumbPath: archivedThumb,
+ };
+}
+
+/** 强制重生成失败时把原稿放回原位；新稿已落盘则保留新稿。 */
+function restoreArchivedTodayPage(archived, logger) {
+ if (!archived?.archivedPath || !fs.existsSync(archived.archivedPath)) return false;
+ if (fs.existsSync(archived.originalPath)) return false;
+ try {
+ fs.renameSync(archived.archivedPath, archived.originalPath);
+ if (
+ archived.archivedThumbPath
+ && fs.existsSync(archived.archivedThumbPath)
+ && !fs.existsSync(archived.originalThumbPath)
+ ) {
+ fs.renameSync(archived.archivedThumbPath, archived.originalThumbPath);
+ }
+ logger.log?.('[NewsMorningDraft] restored archived today page after failed generate', {
+ to: path.basename(archived.originalPath),
+ });
+ return true;
+ } catch (error) {
+ logger.warn?.(
+ '[NewsMorningDraft] restore archived today page failed:',
+ error instanceof Error ? error.message : error,
+ );
+ return false;
+ }
}
export function startWechatNewsMorningDraftWorker({
@@ -338,13 +386,15 @@ export function startWechatNewsMorningDraftWorker({
let stopped = false;
let running = false;
let generationInFlightDateKey = null;
+ let degradedRegenerateDateKey = null;
const ensureTodayPage = async (config, dateKey, { now = Date.now(), force = false } = {}) => {
if (generationInFlightDateKey === dateKey) return null;
generationInFlightDateKey = dateKey;
+ let archived = null;
try {
if (force) {
- archiveTodayPageForForce(config, h5Root, now, logger);
+ archived = archiveTodayPageForForce(config, h5Root, now, logger);
}
let searchBrief = '';
let searchGroups = [];
@@ -387,7 +437,14 @@ export function startWechatNewsMorningDraftWorker({
stage: 'post-generate',
});
const safetyExemptSectionIds = safety?.exemptSectionIds ?? [];
- await applyNews002ImageGate(page, config, logger, env, { searchGroups, safetyExemptSectionIds });
+ try {
+ await applyNews002ImageGate(page, config, logger, env, { searchGroups, safetyExemptSectionIds });
+ } catch (imageError) {
+ logger.warn?.(
+ '[NewsMorningDraft] news002 image repair failed, keeping generated page:',
+ imageError instanceof Error ? imageError.message : imageError,
+ );
+ }
await applyNewsMorningContentSafety(page, {
reviewer: contentReviewer,
reviewCache,
@@ -411,6 +468,7 @@ export function startWechatNewsMorningDraftWorker({
errorMessage: message,
}).catch(() => {});
logger.warn?.('[NewsMorningDraft] auto-generate failed:', message);
+ restoreArchivedTodayPage(archived, logger);
throw error;
} finally {
if (generationInFlightDateKey === dateKey) {
@@ -462,6 +520,7 @@ export function startWechatNewsMorningDraftWorker({
todayPageDegraded
&& !config.forceGenerateOnce
&& generationInFlightDateKey !== dateKey
+ && degradedRegenerateDateKey !== dateKey
&& (
isWithinNewsMorningGenerateLeadWindow(config, now)
|| isLocalScheduleDue(
@@ -476,6 +535,7 @@ export function startWechatNewsMorningDraftWorker({
userId: config.sourceUserId,
dateKey,
});
+ degradedRegenerateDateKey = dateKey;
try {
todayPage = await ensureTodayPage(config, dateKey, { now, force: true });
if (
@@ -507,11 +567,19 @@ export function startWechatNewsMorningDraftWorker({
if (config.forceGenerateOnce && generationInFlightDateKey !== dateKey) {
try {
- todayPage = await ensureTodayPage(config, dateKey, { now, force: true });
await wechatNewsMorningDraftService.updateConfig(
{ forceGenerateOnce: false },
{ updatedBy: 'auto-generate' },
);
+ } catch (error) {
+ logger.warn?.(
+ '[NewsMorningDraft] clear forceGenerateOnce failed, skip force generate:',
+ error instanceof Error ? error.message : error,
+ );
+ return;
+ }
+ try {
+ todayPage = await ensureTodayPage(config, dateKey, { now, force: true });
if (config.autoPushEnabled && todayPage) {
const result = await wechatNewsMorningDraftService.pushDraft({
triggeredBy: 'manual-generate',
@@ -522,8 +590,11 @@ export function startWechatNewsMorningDraftWorker({
pageSlug: result.preview?.page?.slug ?? todayPage.slug,
});
}
- } catch {
- // keep the flag so the next scan retries
+ } catch (error) {
+ logger.warn?.(
+ '[NewsMorningDraft] force generate failed:',
+ error instanceof Error ? error.message : error,
+ );
}
return;
}
diff --git a/wechat-news-morning-draft-worker.test.mjs b/wechat-news-morning-draft-worker.test.mjs
index 4d59080..cc919dc 100644
--- a/wechat-news-morning-draft-worker.test.mjs
+++ b/wechat-news-morning-draft-worker.test.mjs
@@ -244,6 +244,140 @@ test('runOnce honors forceGenerateOnce even when today page exists', async () =>
}
});
+test('failed forceGenerateOnce clears the flag, restores the original page and does not retry', async () => {
+ const tmpRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'news-draft-force-fail-'));
+ const userId = 'user-news';
+ const publicDir = path.join(tmpRoot, PUBLISH_ROOT_DIR, userId, PUBLIC_ZONE_DIR);
+ fs.mkdirSync(publicDir, { recursive: true });
+ const pagePath = path.join(publicDir, 'daily-news-0911.html');
+ fs.writeFileSync(pagePath, HEALTHY_TODAY_HTML);
+
+ const service = createWechatNewsMorningDraftService(createPool(), {
+ mpConfig: { enabled: true, appId: 'app', appSecret: 'secret' },
+ h5Root: tmpRoot,
+ env: {
+ H5_WECHAT_NEWS_MORNING_DRAFT_ENABLED: '1',
+ H5_WECHAT_NEWS_MORNING_DRAFT_USER_ID: userId,
+ },
+ });
+ let config = {
+ enabled: true,
+ sourceUserId: userId,
+ autoPushEnabled: true,
+ autoGenerateEnabled: true,
+ forceGenerateOnce: true,
+ timezone: 'Asia/Shanghai',
+ pushHour: 6,
+ pushMinute: 0,
+ pageSlugPattern: 'daily-news-*',
+ };
+ service.getConfig = async () => config;
+ service.updateConfig = async (payload) => {
+ config = { ...config, ...payload };
+ return config;
+ };
+ const pushCalls = [];
+ service.pushDraft = async (options = {}) => {
+ pushCalls.push(options);
+ return { ok: true, draftMediaId: 'draft-x' };
+ };
+ service.hasSuccessfulAutoPushToday = async () => true;
+
+ let executeCount = 0;
+ const fixedNow = Date.UTC(2026, 8, 10, 22, 30, 0);
+ const originalNow = Date.now;
+ Date.now = () => fixedNow;
+ try {
+ const worker = startWechatNewsMorningDraftWorker({
+ wechatNewsMorningDraftService: service,
+ mpConfig: { enabled: true },
+ userAuth: { canUseChat: async () => ({ ok: true }) },
+ tkmindProxy: { id: 'proxy' },
+ h5Root: tmpRoot,
+ env: { H5_WECHAT_NEWS_MORNING_DRAFT_WORKER_ENABLED: '1' },
+ executeTask: async () => {
+ executeCount += 1;
+ throw new Error('agent timeout');
+ },
+ prefetchNewsSearch: async () => ({ brief: '', groups: [] }),
+ logger: { log() {}, warn() {} },
+ intervalMs: 30_000,
+ runOnStart: false,
+ setIntervalFn: () => ({ unref() {} }),
+ });
+
+ await worker.runOnce();
+ await worker.runOnce();
+ assert.equal(executeCount, 1);
+ assert.equal(config.forceGenerateOnce, false);
+ assert.equal(pushCalls.length, 0);
+ assert.equal(fs.readFileSync(pagePath, 'utf8'), HEALTHY_TODAY_HTML);
+ assert.equal(
+ fs.readdirSync(publicDir).some((name) => name.startsWith('_archived-daily-news-0911-')),
+ false,
+ );
+ } finally {
+ Date.now = originalNow;
+ fs.rmSync(tmpRoot, { recursive: true, force: true });
+ }
+});
+
+test('news002 page that misses structure checks is still published with empty sections pruned', async () => {
+ const tmpRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'news-draft-news002-degraded-'));
+ const userId = 'user-news';
+ const publicDir = path.join(tmpRoot, PUBLISH_ROOT_DIR, userId, PUBLIC_ZONE_DIR);
+ fs.mkdirSync(publicDir, { recursive: true });
+
+ const service = createWechatNewsMorningDraftService(createPool(), {
+ mpConfig: { enabled: true, appId: 'app', appSecret: 'secret' },
+ h5Root: tmpRoot,
+ env: {
+ H5_WECHAT_NEWS_MORNING_DRAFT_ENABLED: '1',
+ H5_WECHAT_NEWS_MORNING_DRAFT_USER_ID: userId,
+ H5_WECHAT_NEWS_MORNING_TEMPLATE_ID: 'news002',
+ },
+ });
+ const warnings = [];
+ const html = '每日新闻早报
'
+ + '国内
'
+ + '
国产大飞机交付第100架
'
+ + '
'
+ + '
来源声音
'
+ + '';
+ const worker = startWechatNewsMorningDraftWorker({
+ wechatNewsMorningDraftService: service,
+ mpConfig: { enabled: true },
+ userAuth: { canUseChat: async () => ({ ok: true }) },
+ tkmindProxy: { id: 'proxy' },
+ h5Root: tmpRoot,
+ env: {
+ H5_WECHAT_NEWS_MORNING_DRAFT_WORKER_ENABLED: '1',
+ MEMIND_NEWS_MORNING_SEARXNG_FALLBACK: '0',
+ },
+ executeTask: async () => {
+ fs.writeFileSync(path.join(publicDir, 'daily-news-0911.html'), html);
+ },
+ prefetchNewsSearch: async () => ({ brief: '', groups: [] }),
+ contentReviewer: async () => ({ ok: true, violations: [] }),
+ logger: { log() {}, warn: (...args) => warnings.push(args) },
+ intervalMs: 30_000,
+ runOnStart: false,
+ setIntervalFn: () => ({ unref() {} }),
+ });
+
+ try {
+ const result = await worker.generateToday({ now: Date.UTC(2026, 8, 10, 22, 0, 0), force: true });
+ assert.equal(result.skipped, false);
+ const saved = fs.readFileSync(path.join(publicDir, 'daily-news-0911.html'), 'utf8');
+ assert.match(saved, /国产大飞机交付第100架/u);
+ assert.doesNotMatch(saved, /id="voices"/);
+ assert.ok(warnings.some(([message]) => /publishing best-effort page/.test(String(message))));
+ } finally {
+ fs.rmSync(tmpRoot, { recursive: true, force: true });
+ }
+});
+
test('generateToday skips existing page unless force is set', async () => {
const tmpRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'news-draft-force-'));
const userId = 'user-news';
diff --git a/wechat-news-morning-draft.mjs b/wechat-news-morning-draft.mjs
index 3c9b8a7..46ca71c 100644
--- a/wechat-news-morning-draft.mjs
+++ b/wechat-news-morning-draft.mjs
@@ -31,7 +31,6 @@ import {
import {
auditNews002ImageCoverage,
extractNews002CardImageUrls,
- isWeakNewsMorningImageUrl,
} from './wechat-news-morning-images.mjs';
import {
NEWS001_FRESH_CONTENT_SPEC,
@@ -854,38 +853,51 @@ export async function buildDailyNewsWechatDraftArticleForPush({
wechatFetch,
memindLibRoot,
});
- if (/class="[^"]*\bmasthead\b/i.test(String(html ?? ''))) {
- const audit = auditNews002ImageCoverage(html, {
+ let draftHtml = String(html ?? '');
+ const pushWarnings = [];
+ if (/class="[^"]*\bmasthead\b/i.test(draftHtml)) {
+ const audit = auditNews002ImageCoverage(draftHtml, {
requireTemplateStructure: true,
requireAllCoreSections: false,
requireLeadCount: false,
});
- const pushBlocking = audit.violations.filter((item) => (
+ pushWarnings.push(...audit.violations.filter((item) => (
item === 'story-image-missing'
|| item === 'lead-image-missing'
|| item === 'too-many-story-cards'
|| item === 'legacy-news001-dom'
- ));
- if (pushBlocking.length > 0) {
- throw new Error(`news002 页面未通过推送闸门：${pushBlocking.join(', ')}`);
- }
- const missingUploads = extractNews002CardImageUrls(html).filter((ref) => (
- !isWeakNewsMorningImageUrl(ref)
- )).filter((ref) => (
+ )));
+ const missingUploads = extractNews002CardImageUrls(draftHtml).filter((ref) => (
!imageUrlMap.has(ref)
&& !imageUrlMap.has(ref.replace(/^\.\//, ''))
&& !imageUrlMap.has(ref.replace(/^\//, ''))
));
if (missingUploads.length > 0) {
- throw new Error(`news002 有 ${missingUploads.length} 张正文配图未成功上传微信，已阻止草稿推送`);
+ draftHtml = stripNews002ImagesByRef(draftHtml, missingUploads);
+ pushWarnings.push(`image-upload-missing:${missingUploads.length}`);
}
}
- return buildDailyNewsWechatDraftArticle({
- html,
- publicUrl,
- qrcodeImageUrl,
- imageUrlMap,
- });
+ return {
+ ...buildDailyNewsWechatDraftArticle({
+ html: draftHtml,
+ publicUrl,
+ qrcodeImageUrl,
+ imageUrlMap,
+ }),
+ pushWarnings,
+ };
+}
+
+/** 微信正文只认已上传的 mmbiz 图片：未上传成功的配图整块去掉，卡片保留文字。 */
+export function stripNews002ImagesByRef(html, refs = []) {
+ const missing = new Set(refs);
+ if (missing.size === 0) return String(html ?? '');
+ return String(html ?? '')
+ .replace(/]*class="[^"]*\bmedia\b[^"]*"[^>]*>[\s\S]*?<\/div>/gi, (block) => {
+ const src = block.match(/]*\bsrc=["']([^"']+)["']/i)?.[1];
+ return src && missing.has(src) ? '' : block;
+ })
+ .replace(/]*\bsrc=["']([^"']+)["'][^>]*>/gi, (tag, src) => (missing.has(src) ? '' : tag));
}
export function convertNewsPageHtmlToWechatArticle(html, { publicUrl = '', qrcodeImageUrl = '' } = {}) {
@@ -1588,6 +1600,12 @@ export function createWechatNewsMorningDraftService(
publishDir: path.dirname(page.localPath),
})
: preview.article;
+ if (article.pushWarnings?.length) {
+ logger.warn?.('[NewsMorningDraft] draft pushed with repairs', {
+ pageSlug: page.slug,
+ warnings: article.pushWarnings,
+ });
+ }
if (articleTitleOverride) {
article.title = String(articleTitleOverride).trim().slice(0, 64);
}
diff --git a/wechat-news-morning-draft.test.mjs b/wechat-news-morning-draft.test.mjs
index e538949..105b1a5 100644
--- a/wechat-news-morning-draft.test.mjs
+++ b/wechat-news-morning-draft.test.mjs
@@ -10,10 +10,21 @@ import {
createWechatNewsMorningDraftService,
findLatestNewsMorningPage,
NEWS_MORNING_TEMPLATE_VERSION,
+ stripNews002ImagesByRef,
wechatNewsMorningDraftInternals,
} from './wechat-news-morning-draft.mjs';
import { PUBLIC_ZONE_DIR, PUBLISH_ROOT_DIR } from './user-publish.mjs';
+test('stripNews002ImagesByRef drops only images that failed to upload and keeps card text', () => {
+ const html = ''
+ + '';
+ const out = stripNews002ImagesByRef(html, ['./a.jpg']);
+ assert.doesNotMatch(out, /a\.jpg/);
+ assert.match(out, /b\.jpg/);
+ assert.match(out, /甲<\/h3>/u);
+ assert.equal(stripNews002ImagesByRef(html, []), html);
+});
+
const SAMPLE_HTML = `
diff --git a/wechat-news-morning-images.mjs b/wechat-news-morning-images.mjs
index e87f7c8..20baa70 100644
--- a/wechat-news-morning-images.mjs
+++ b/wechat-news-morning-images.mjs
@@ -420,7 +420,7 @@ export function promoteNews002LeadCards(html, { min = 3, max = 4 } = {}) {
const NEWS002_KEEP_EMPTY_SECTION_IDS = new Set(['weather', 'history', 'brief']);
-function pruneEmptyNews002Sections(html) {
+export function pruneEmptyNews002Sections(html, { requireImage = true } = {}) {
let output = String(html ?? '');
const re = /]*class=["'][^"']*\bsection\b[^"']*["'][^>]*id=["']([^"']+)["'][^>]*>/gi;
const matches = [...output.matchAll(re)];
@@ -435,7 +435,8 @@ function pruneEmptyNews002Sections(html) {
match.index + match[0].length,
);
const slice = output.slice(start, end);
- const cards = extractNews002StoryCards(slice).filter((card) => cardHasDisplayableImage(card.raw));
+ const cards = extractNews002StoryCards(slice)
+ .filter((card) => (requireImage ? cardHasDisplayableImage(card.raw) : Boolean(card.title)));
const hasDeep = /]*class=["'][^"']*\bdeep\b/i.test(slice);
if (cards.length === 0 && !hasDeep) {
output = output.slice(0, start) + output.slice(end);
