feedkeeper 0.35.5 → 0.35.6

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "feedkeeper",
-  "version": "0.35.5",
+  "version": "0.35.6",
   "description": "A lightweight, self-hosted RSS reader and remote MCP server for your news, blogs, and AI workflows.",
   "keywords": [
     "rss",

package/server/dist/feeds/poller.js

@@ -3,6 +3,7 @@ import { fetchFeed } from "./fetcher.js";
 import { discoverIconUrls, iconCheckDue } from "./icon.js";
 import { refreshFeedIcon } from "./feedIcon.js";
 import { decodeEntities, plainTitle } from "./text.js";
+import { unwrapRedirectLink } from "./redirectLinks.js";
 import { enqueueNewItemNotifications, startPushOutbox, stopPushOutbox } from "../pushOutbox.js";
 import { hasUserCapability } from "../auth/capabilities.js";
 import { findFeedById, listSubscriberIds, updateFeedAfterPoll, updateFeedIcon, upsertItems } from "./repository.js";
@@ -123,8 +124,8 @@ export function itemGuid(item) {
 /** The article address: its link, or its guid when the feed gives only that and the guid is a web address. */
 export function itemLink(item) {
     if (item.link)
-        return item.link;
-    return item.guid && /^https?:\/\//i.test(item.guid) ? item.guid : undefined;
+        return unwrapRedirectLink(item.link);
+    return item.guid && /^https?:\/\//i.test(item.guid) ? unwrapRedirectLink(item.guid) : undefined;
 }
 /** One attempt; the queue owns retries, admission, single-flight and the total deadline. */
 async function pollAttempt(feed, signal) {

package/server/dist/feeds/redirectLinks.js

@@ -0,0 +1,22 @@
+/**
+ * Some feeds wrap every article address in a redirect through the publisher's tracker. Google Alerts links through
+ * `https://www.google.com/url?...&url=<article>`, which is not the page to read and fails the full-text fetch. The article address
+ * inside the wrapper is what an item should point to.
+ */
+const GOOGLE_HOST = /^(?:www\.)?google\.[a-z]{2,3}(?:\.[a-z]{2})?$/i;
+export function unwrapRedirectLink(link) {
+    if (!link)
+        return link;
+    let url;
+    try {
+        url = new URL(link);
+    }
+    catch {
+        return link;
+    }
+    if (!GOOGLE_HOST.test(url.hostname) || url.pathname !== "/url")
+        return link;
+    const target = url.searchParams.get("url") ?? url.searchParams.get("q");
+    return target && /^https?:\/\//i.test(target) ? target : link;
+}
+//# sourceMappingURL=redirectLinks.js.map

package/server/dist/feeds/repository.js

@@ -3,7 +3,8 @@ import { articleText } from "./articleContent.js";
 import { LEAD_VERSION } from "./contentNoise.js";
 import { largerHeroImage } from "./heroImage.js";
 import { refreshGalleryClassification, refreshLead } from "./lead.js";
-import { decodeEntities } from "./text.js";
+import { decodeEntities, plainTitle } from "./text.js";
+import { unwrapRedirectLink } from "./redirectLinks.js";
 export function findFeedByUrl(url) {
     return db.prepare("SELECT * FROM feeds WHERE url = ?").get(url);
 }
@@ -570,19 +571,24 @@ export function moveSubscription(userId, fromFeedId, toFeedId) {
 export function updateFeedIcon(feedId, iconUrl) {
     db.prepare("UPDATE feeds SET icon_url = ?, icon_checked_at = strftime('%Y-%m-%dT%H:%M:%fZ', 'now') WHERE id = ?").run(iconUrl, feedId);
 }
-/** Fix item titles, snippets, feed titles and labels stored before entity decoding existed; cheap once nothing matches. */
+/**
+ * Fix item titles, snippets, links, feed titles and labels stored before entity decoding, title markup removal and redirect unwrapping
+ * existed; cheap once nothing matches.
+ */
 export function repairEncodedText() {
     const rows = db
-        .prepare("SELECT id, title, content_snippet FROM items WHERE title LIKE '%&%;%' OR content_snippet LIKE '%&%;%'")
+        .prepare(`SELECT id, title, content_snippet, link FROM items
+       WHERE title LIKE '%&%;%' OR title LIKE '%<%>%' OR content_snippet LIKE '%&%;%' OR link LIKE 'https://www.google.%/url?%'`)
         .all();
-    const update = db.prepare("UPDATE items SET title = ?, content_snippet = ? WHERE id = ?");
+    const update = db.prepare("UPDATE items SET title = ?, content_snippet = ?, link = ? WHERE id = ?");
     let fixed = 0;
     db.transaction(() => {
         for (const row of rows) {
-            const title = decodeEntities(row.title);
+            const title = row.title === null ? null : (plainTitle(row.title) ?? row.title);
             const snippet = decodeEntities(row.content_snippet);
-            if (title !== row.title || snippet !== row.content_snippet) {
-                update.run(title ?? null, snippet ?? null, row.id);
+            const link = unwrapRedirectLink(row.link);
+            if (title !== row.title || snippet !== row.content_snippet || link !== row.link) {
+                update.run(title, snippet ?? null, link ?? null, row.id);
                 refreshLead(row.id);
                 fixed++;
             }

package/server/dist/feeds/text.js

@@ -15,9 +15,14 @@ function decodeOnce(text) {
         return NAMED_ENTITIES[name] ?? match;
     });
 }
-/** A title as one line of text: entities decoded, line breaks and runs of spaces collapsed to one space, trimmed. */
+/**
+ * Inline formatting that some feeds put into a title, such as the `<b>` Google Alerts wraps around the search term. Only these tags are
+ * removed, so a title that really contains an angle bracket stays as it is.
+ */
+const TITLE_MARKUP = /<\/?(?:b|i|u|em|strong|mark|span|small|sub|sup|font|a|br)\b[^>]*>/gi;
+/** A title as one line of text: entities decoded, inline markup removed, line breaks and runs of spaces collapsed to one space, trimmed. */
 export function plainTitle(text) {
-    const cleaned = decodeEntities(text)?.replace(/\s+/g, " ").trim();
+    const cleaned = decodeEntities(text)?.replace(TITLE_MARKUP, tag => (/^<br/i.test(tag) ? " " : "")).replace(/\s+/g, " ").trim();
     return cleaned || undefined;
 }
 export function decodeEntities(text) {

package/server/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@feedkeeper/server",
-  "version": "0.35.5",
+  "version": "0.35.6",
   "private": true,
   "type": "module"
 }