fix(news): relax SearXNG fallback when prefetch images are missing

Allow no-image candidates and create missing section shells so news002
image gate can pass after source-page enrichment on thin agent output.

Co-authored-by: Cursor <cursoragent@cursor.com>

wechat-news-morning-searxng-fallback.mjs

@@ -223,6 +223,32 @@ function pickSearchCandidates(groupMap, poolIds, seenTitles, { requireImage = tr
   return picked;
 }
 
+/** 先取带图候选，不足时再放宽为无图（后续 enrichment 会从来源页补 og:image）。 */
+function pickSearchCandidatesWithFallback(groupMap, poolIds, seenTitles, needCount) {
+  if (needCount <= 0) return [];
+  const workingSeen = [...seenTitles];
+  const withImage = pickSearchCandidates(groupMap, poolIds, workingSeen, { requireImage: true });
+  let merged = withImage;
+  if (merged.length < needCount) {
+    const withoutImage = pickSearchCandidates(groupMap, poolIds, workingSeen, { requireImage: false });
+    const usedUrls = new Set(merged.map((item) => item.url));
+    for (const item of withoutImage) {
+      if (merged.length >= needCount) break;
+      if (usedUrls.has(item.url)) continue;
+      merged.push(item);
+      usedUrls.add(item.url);
+    }
+  }
+  const picked = merged.slice(0, needCount);
+  for (const item of picked) {
+    seenTitles.push({
+      urlKey: normalizeNewsMorningUrl(item.url),
+      title: normalizeNewsMorningTitle(item.title),
+    });
+  }
+  return picked;
+}
+
 function buildLeadCardHtml(item, eventKey) {
   const title = escapeHtml(item.title);
   const snippet = escapeHtml(truncateText(item.snippet || item.title, 120));
@@ -340,17 +366,15 @@ export function supplementNews002HtmlFromSearxngGroups(
     || countLeadCardsInHeadlines(output) < 3;
   let remainingBudget = Math.max(0, maxAdds - existingCards.length);
 
-  const leadPool = pickSearchCandidates(
-    groupMap,
-    SECTION_SEARCH_POOLS.headlines,
-    [...seenTitles],
-    { requireImage: true },
-  );
-
   if (needLeads && remainingBudget > 0) {
     const currentLeads = countLeadCardsInHeadlines(output);
     const needCount = Math.max(0, 3 - currentLeads);
-    const toAdd = leadPool.splice(0, Math.min(needCount, remainingBudget));
+    const toAdd = pickSearchCandidatesWithFallback(
+      groupMap,
+      SECTION_SEARCH_POOLS.headlines,
+      seenTitles,
+      Math.min(needCount, remainingBudget),
+    );
     if (toAdd.length) {
       const cardsHtml = toAdd.map((item) => (
         buildLeadCardHtml(item, buildNewsMorningEventKey(item.title, usedEventKeys))
@@ -394,9 +418,17 @@ export function supplementNews002HtmlFromSearxngGroups(
     if (!needCount && hasNews002Section(output, sectionId)) continue;
 
     const pool = SECTION_SEARCH_POOLS[sectionId] ?? [SEARXNG_GROUP_TO_NEWS002_SECTION[sectionId]].filter(Boolean);
-    const candidates = pickSearchCandidates(groupMap, pool, [...seenTitles], { requireImage: true });
-    const toAdd = candidates.slice(0, Math.min(needCount || meta.min, remainingBudget));
-    if (!toAdd.length) continue;
+    const targetCount = Math.min(needCount || meta.min, remainingBudget);
+    const toAdd = pickSearchCandidatesWithFallback(groupMap, pool, seenTitles, targetCount);
+    const sectionMissing = missingSections.includes(sectionId) || !hasNews002Section(output, sectionId);
+
+    if (!toAdd.length) {
+      if (sectionMissing) {
+        output = createAndInsertSection(output, sectionId, '');
+        stats.addedSections.push(sectionId);
+      }
+      continue;
+    }
 
     const cardsHtml = toAdd.map((item) => (
       buildStoryCardHtml(item, buildNewsMorningEventKey(item.title, usedEventKeys))

wechat-news-morning-searxng-fallback.test.mjs

@@ -75,6 +75,37 @@ test('buildNewsMorningEventKey deduplicates keys', () => {
   assert.equal(buildNewsMorningEventKey('Same Title', used), 'same-title-2');
 });
 
+test('supplementNews002HtmlFromSearxngGroups fills sections without prefetch images', () => {
+  const groupsNoImage = [
+    {
+      id: 'world',
+      title: '国际',
+      results: [
+        { title: '国际无图一', url: 'https://news.example/w-no1', snippet: '国际一' },
+        { title: '国际无图二', url: 'https://news.example/w-no2', snippet: '国际二' },
+      ],
+    },
+    {
+      id: 'tech',
+      title: '科技',
+      results: [
+        { title: '科技无图一', url: 'https://news.example/t-no1', snippet: '科技一' },
+        { title: '科技无图二', url: 'https://news.example/t-no2', snippet: '科技二' },
+      ],
+    },
+  ];
+  const result = supplementNews002HtmlFromSearxngGroups(
+    THIN_AGENT_HTML,
+    groupsNoImage,
+    { violations: ['missing-sections:world|tech'] },
+  );
+  assert.equal(result.changed, true);
+  assert.match(result.html, /id="world"/u);
+  assert.match(result.html, /id="tech"/u);
+  assert.match(result.html, /国际无图一/u);
+  assert.match(result.html, /科技无图一/u);
+});
+
 test('supplementNews002HtmlFromSearxngGroups fills missing core sections and leads', () => {
   const beforeAudit = auditNews002ImageCoverage(THIN_AGENT_HTML, { requireTemplateStructure: true });
   assert.equal(beforeAudit.ok, false);