fix: edit registry filter

scripts/harvest.js

@@ -125,6 +125,11 @@ function originalFilename(titleEn, sourceName) {
 
 // ── Registry ─────────────────────────────────────────────────────────────────
 
+function normalizeUrl(u) {
+  try { const p = new URL(u); p.search = ''; return p.toString(); }
+  catch (e) { return u; }
+}
+
 function buildDedupSet(config, sourceId) {
   const seen = new Set();
   const windowDays = config.registryWindowDays || 7;
@@ -132,20 +137,35 @@ function buildDedupSet(config, sourceId) {
   const source = config.sources.find(s => s.id === sourceId);
   if (!source) return seen;
 
-  const folder = path.join(vaultPath, source.vaultFolder);
-  if (!fs.existsSync(folder)) return seen;
+  // vaultFolder 含 "/" 时（如 "日经亚洲/世界"），从主目录遍历所有二级子目录的 registry.json，
+  // 实现同主目录下不同栏目的跨栏目去重；否则只扫描自身目录。
+  const parts = source.vaultFolder.split('/');
+  let dayRoots;
+  if (parts.length > 1) {
+    const primary = path.join(vaultPath, parts[0]);
+    if (!fs.existsSync(primary)) return seen;
+    dayRoots = fs.readdirSync(primary)
+      .filter(sub => fs.statSync(path.join(primary, sub)).isDirectory())
+      .map(sub => path.join(primary, sub));
+  } else {
+    const folder = path.join(vaultPath, source.vaultFolder);
+    if (!fs.existsSync(folder)) return seen;
+    dayRoots = [folder];
+  }
 
   const today = new Date();
   for (let i = 0; i < windowDays; i++) {
     const d = new Date(today); d.setDate(today.getDate() - i);
     const dayStr = `${d.getFullYear()}${String(d.getMonth()+1).padStart(2,'0')}${String(d.getDate()).padStart(2,'0')}`;
-    const regPath = path.join(folder, dayStr, 'registry.json');
-    if (fs.existsSync(regPath)) {
-      try {
-        const reg = JSON.parse(fs.readFileSync(regPath, 'utf8'));
-        (reg.articles || []).forEach(a => seen.add(a.url));
-      } catch (e) {
-        console.error(`  ⚠️  Failed to parse registry ${regPath}: ${e.message}`);
+    for (const root of dayRoots) {
+      const regPath = path.join(root, dayStr, 'registry.json');
+      if (fs.existsSync(regPath)) {
+        try {
+          const reg = JSON.parse(fs.readFileSync(regPath, 'utf8'));
+          (reg.articles || []).forEach(a => seen.add(normalizeUrl(a.url)));
+        } catch (e) {
+          console.error(`  ⚠️  Failed to parse registry ${regPath}: ${e.message}`);
+        }
       }
     }
   }
@@ -484,7 +504,6 @@ Examples:
   console.error(`   Collected ${links.length} candidate links`);
 
   // Normalize URLs (strip query params for dedup)
-  const normalizeUrl = (u) => { try { const p = new URL(u); p.search = ''; return p.toString(); } catch(e) { return u; } };
   const normalizedLinks = [...new Set(links.map(u => normalizeUrl(u)))];
   // Keep ALL new candidates rather than slicing to `count`: recipe sources with
   // validateArticle reject section/category pages only at fetch time, so capping