feedkeeper 0.26.2 → 0.26.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "feedkeeper",
-  "version": "0.26.2",
+  "version": "0.26.4",
   "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

@@ -76,12 +76,28 @@ function removeReaderNoise(root, title, stats) {
     const headline = normalized(title ?? "");
     const seen = new Set();
     let leading = 0;
+    let reachedText = false;
     for (const block of leafBlocks(root)) {
         const text = flat(block.textContent);
         if (!text)
             continue;
         const position = leading++;
         const heading = /^H[1-3]$/.test(block.tagName);
+        // Before the first real paragraph: a standalone date and a section chip repeat what the page head says.
+        if (!reachedText && text.length >= 80)
+            reachedText = true;
+        if (!reachedText && position < 12) {
+            const anchorsInBlock = block.querySelectorAll("a");
+            const chip = anchorsInBlock.length === 1 && flat(anchorsInBlock[0].textContent) === text && text.length <= 30 && text.split(" ").length <= 3;
+            if (chip) {
+                drop(block, "section-chip");
+                continue;
+            }
+            if (/^(\d{1,2}\.\s?\d{1,2}\.\s?\d{2,4}|[A-Z][a-z]{2,8}\.? \d{1,2}, \d{4})([,\s•·|-]+\d{1,2}[:.]\d{2}( Uhr)?)?$/.test(text)) {
+                drop(block, "dateline");
+                continue;
+            }
+        }
         if (headline && normalized(text) === headline && (position < 4 || (heading && position < 14))) {
             drop(block, "title-repeat");
             continue;
@@ -228,9 +244,37 @@ export function cleanReaderHtml(html, title, stats = {}) {
     if (!html)
         return html;
     const root = JSDOM.fragment(html);
+    const images = resolveCustomImages(root, stats);
     const noise = removeReaderNoise(root, title, stats);
     const typography = removeTypography(root, stats);
-    return noise || typography ? serialize(root) : html;
+    return images || noise || typography ? serialize(root) : html;
+}
+/**
+ * Some publishers wrap a picture in their own element and let a script fill the image: `<a-img src="/imgs/…">`
+ * around an `<img>` that only holds a grey placeholder. Without the script no reader would ever show the picture, so
+ * the wrapper's address takes the placeholder's place and the wrapper, which means nothing outside the page, goes.
+ */
+function resolveCustomImages(root, stats) {
+    let changed = false;
+    for (const wrapper of [...root.querySelectorAll("a-img")]) {
+        const source = (wrapper.getAttribute("src") ?? "").trim();
+        if (source && !/^data:/i.test(source)) {
+            let image = wrapper.querySelector("img");
+            if (!image) {
+                image = wrapper.ownerDocument.createElement("img");
+                image.setAttribute("alt", wrapper.getAttribute("alt") ?? "");
+                wrapper.append(image);
+            }
+            const current = (image.getAttribute("src") ?? "").trim();
+            if (!current || /^data:/i.test(current))
+                image.setAttribute("src", source);
+        }
+        wrapper.replaceWith(...wrapper.childNodes);
+        changed = true;
+    }
+    if (changed)
+        stats["custom-image"] = (stats["custom-image"] ?? 0) + 1;
+    return changed;
 }
 const CREDIT_PATTERNS = [
     /^(©|\(c\)\s|copyright\b)/i,

package/server/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@feedkeeper/server",
-  "version": "0.26.2",
+  "version": "0.26.4",
   "private": true,
   "type": "module"
 }