feedkeeper 0.28.0 → 0.28.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "feedkeeper",
-  "version": "0.28.0",
+  "version": "0.28.1",
   "description": "A lightweight, self-hosted RSS reader and remote MCP server for your news, blogs, and AI workflows.",
   "keywords": [
     "rss",

package/server/dist/feeds/contentNoise.js

@@ -121,8 +121,14 @@ function removeReaderNoise(root, title, stats, author) {
             if (ancestor.querySelector("a[href]") && PRICE.test(text))
                 card = ancestor;
         }
-        if (card)
+        if (card) {
+            const parent = card.parentElement;
             drop(card, "product-card");
+            // A long product name can put the enclosing header just beyond the card's size limit.
+            if (parent && PRODUCT_LABEL.test(flat(parent.textContent)) && !parent.querySelector("img, picture, iframe, video, a[href]")) {
+                drop(parent, "empty-product-label");
+            }
+        }
     }
     // Structured data is for machines. Frameworks render it as an element of any name (`<component type="application/ld+json">`),
     // and a stray block may hold nothing else: neither is text for a reader.
@@ -248,6 +254,27 @@ function removeReaderNoise(root, title, stats, author) {
         drop(block, "comment-rules");
         break;
     }
+    // Empty nested slots are structural evidence: the next story paragraph is outside this ad container.
+    // Do not infer an empty ad from a standalone heading followed by text (which may itself be sponsored).
+    for (const container of [...root.querySelectorAll("div, aside, section, p")]) {
+        if (!root.contains(container))
+            continue;
+        const text = flat(container.textContent);
+        const product = PRODUCT_LABEL.test(text);
+        if (!product && !AD_LABEL.test(text))
+            continue;
+        if (container.querySelector("img, picture, iframe, video, audio, a[href], object, embed"))
+            continue;
+        const emptySlot = container.querySelector("div:empty, ins:empty");
+        // Readability may remove the empty slot, leaving only its enclosing block and label paragraph.
+        const isolated = container.matches("div, aside, section") && container.querySelector("p") &&
+            (product || AD_SLOT_LABEL.test(text));
+        const next = container.nextElementSibling;
+        const nextHasMedia = next && (next.matches("img, picture, iframe, video, audio, a[href]") ||
+            next.querySelector("img, picture, iframe, video, audio, a[href]"));
+        if (emptySlot || (isolated && !nextHasMedia))
+            drop(container, product ? "empty-product-label" : "empty-ad-label");
+    }
     // An ad label whose ad is not there (a script fills it): sponsored content itself stays, the label above nothing goes.
     const labelled = leafBlocks(root);
     labelled.forEach((block, index) => {
@@ -362,13 +389,14 @@ export function cleanReaderHtml(html, title, stats = {}, author) {
         return html;
     const root = JSDOM.fragment(html);
     const images = resolveCustomImages(root, stats);
+    const duplicates = collapseAdjacentImages(root, stats);
     const variants = collapseImageVariants(root, stats);
     const videos = resolveVideoPlayers(root, stats);
     const noise = removeReaderNoise(root, title, stats, author);
     const typography = removeTypography(root, stats);
     videos.links.forEach(link => { if (root.contains(link))
         link.setAttribute("data-media", "video"); });
-    return images || variants || videos.changed || noise || typography ? serialize(root) : html;
+    return images || duplicates || variants || videos.changed || noise || typography ? serialize(root) : html;
 }
 const PLAY_LABEL = /^(video starten|video abspielen|abspielen|wiedergabe( starten)?|play( video)?|watch( video)?|start video)$/i;
 const PLAY_SOURCE = /(?:^|[/_.-])play[-_]?(?:icon|button|btn)?\.(?:svg|png|webp|gif)(?:[?#]|$)/i;
@@ -539,6 +567,46 @@ function collapseImageVariants(root, stats) {
     }
     return changed;
 }
+/** Responsive copies may be separate images whose CSS normally hides one. Only exact adjacent copies go;
+ * text, other media, different links, captions and responsive sources keep images distinct. */
+function collapseAdjacentImages(root, stats) {
+    const walker = root.ownerDocument.createTreeWalker(root, 1 | 4);
+    let previous = null;
+    const duplicates = [];
+    for (let node = walker.nextNode(); node; node = walker.nextNode()) {
+        if (node.nodeType === 3) {
+            if (flat(node.textContent))
+                previous = null;
+            continue;
+        }
+        const element = node;
+        if (element.matches("video, audio, iframe, object, embed, hr"))
+            previous = null;
+        if (element.tagName !== "IMG")
+            continue;
+        const src = flat(element.getAttribute("data-src") || element.getAttribute("data-lazy-src") ||
+            element.getAttribute("data-original") || element.getAttribute("src"));
+        if (!src || /^(data|blob):/i.test(src)) {
+            previous = null;
+            continue;
+        }
+        const picture = element.closest("picture");
+        const key = JSON.stringify([src, element.getAttribute("srcset"), element.getAttribute("data-srcset"),
+            element.getAttribute("alt"), element.getAttribute("title"), element.closest("a")?.getAttribute("href"),
+            [...(picture?.querySelectorAll("source") ?? [])].map(source => [source.getAttribute("srcset"), source.getAttribute("media"), source.getAttribute("type")])]);
+        if (previous === key)
+            duplicates.push(picture?.querySelectorAll("img").length === 1 ? picture : element);
+        else
+            previous = key;
+    }
+    for (const duplicate of duplicates) {
+        const parent = duplicate.parentNode;
+        duplicate.remove();
+        pruneEmpty(parent);
+        stats["duplicate-image"] = (stats["duplicate-image"] ?? 0) + 1;
+    }
+    return duplicates.length > 0;
+}
 const CREDIT_PATTERNS = [
     /^(©|\(c\)\s|copyright\b)/i,
     /^[([]?(Bild|Foto|Fotos|Bildquelle|Bildrechte|Bildnachweis|Titelbild|Beitragsbild|Photo|Image|Credits?|Grafik|Illustration)[)\]]?\s*[:/|]/i,
@@ -589,7 +657,9 @@ const COMMENT_PATTERNS = [
     /\b(let us know|tell us) .{0,40}in the comments\b/i,
     /\bNur angemeldete Benutzer können kommentieren\b/i,
 ];
-const AD_LABEL = /^(Anzeige|Werbung|Advertisement|Advertising|Advertorial|Ads?|Sponsored( content| post)?|Paid (content|post)|Gesponsert|Promoted)\s*:?$/i;
+const AD_LABEL = /^(Anzeige|Werbung|Advertisement|Advertising|Advertorial|Ads?|Sponsored( content| post)?|Paid (content|post)|Gesponsert|Promoted|広告|広告掲載|スポンサー(?:記事|コンテンツ)?|宣伝|PR)\s*[:：]?$/i;
+const AD_SLOT_LABEL = /^(Anzeige|Werbung|Advertisement|Advertising|Ads?|広告|広告掲載|宣伝)\s*[:：]?$/i;
+const PRODUCT_LABEL = /^(Produkthinweis|Produktempfehlung|Product (notice|recommendation)|商品紹介|商品案内|製品紹介)\s*[:：]?$/i;
 const COMMENT_RULES_HEADING = /^(Regeln für Kommentare|Kommentarregeln|Kommentar-Richtlinien|Community[- ]Richtlinien|(Comment(ing)?|Community|Discussion|House) (rules|guidelines|policy|policies)|Rules for comments|Commenting rules)\s*:?$/i;
 const TAG_LABEL = /^(Schlagw(ö|oe)rter|Schlagworte|Tags?|Keywords|Themen|Topics)\s*:?$/i;
 const VIDEO_CARD = /^\d{1,2}:\d{2}(:\d{2})?(?=\D)\s*\S/;

package/server/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@feedkeeper/server",
-  "version": "0.28.0",
+  "version": "0.28.1",
   "private": true,
   "type": "module"
 }