From a51b5540682f16e2e895be408f4d6d313716c0e7 Mon Sep 17 00:00:00 2001
From: Ricardo <rick@rmendes.net>
Date: Thu, 5 Mar 2026 22:52:59 +0100
Subject: [PATCH] feat: delete stale items older than 30 days

Stripped dedup skeletons and unread items older than 30 days are now
hard-deleted on startup. Previously, stripped skeletons accumulated
forever and unread items had no expiry, causing unbounded collection
growth.

Confab-Link: http://localhost:8080/sessions/4d40ef89-a713-48c1-b4ed-0ffafca25677
---
 index.js             | 11 +++++++-
 lib/storage/items.js | 61 ++++++++++++++++++++++++++++++++++++++++++++
 package.json         |  2 +-
 3 files changed, 72 insertions(+), 2 deletions(-)

diff --git a/index.js b/index.js
index 72d8882..127df93 100644
--- a/index.js
+++ b/index.js
@@ -9,7 +9,11 @@ import { readerController } from "./lib/controllers/reader.js";
 import { handleMediaProxy } from "./lib/media/proxy.js";
 import { startScheduler, stopScheduler } from "./lib/polling/scheduler.js";
 import { ensureActivityPubChannel } from "./lib/storage/channels.js";
-import { cleanupAllReadItems, createIndexes } from "./lib/storage/items.js";
+import {
+  cleanupAllReadItems,
+  cleanupStaleItems,
+  createIndexes,
+} from "./lib/storage/items.js";
 import { webmentionReceiver } from "./lib/webmention/receiver.js";
 import { websubHandler } from "./lib/websub/handler.js";
 
@@ -210,6 +214,11 @@ export default class MicrosubEndpoint {
       cleanupAllReadItems(indiekit).catch((error) => {
         console.warn("[Microsub] Startup cleanup failed:", error.message);
       });
+
+      // Delete stale items (stripped skeletons + unread older than 30 days)
+      cleanupStaleItems(indiekit).catch((error) => {
+        console.warn("[Microsub] Stale cleanup failed:", error.message);
+      });
     } else {
       console.warn(
         "[Microsub] Database not available at init, scheduler not started",
diff --git a/lib/storage/items.js b/lib/storage/items.js
index 31369e1..27cbb8a 100644
--- a/lib/storage/items.js
+++ b/lib/storage/items.js
@@ -387,6 +387,10 @@ export async function countReadItems(application, channelId, userId) {
 // uid, readBy) so the poller doesn't re-ingest them as new unread entries.
 const MAX_FULL_READ_ITEMS = 200;
 
+// Maximum age (in days) for stripped skeletons and unread items.
+// After this period, both are hard-deleted to prevent unbounded growth.
+const MAX_ITEM_AGE_DAYS = 30;
+
 /**
  * Cleanup old read items by stripping content but preserving dedup skeletons.
  * This prevents the vicious cycle where deleted read items get re-ingested as
@@ -575,6 +579,63 @@ export async function cleanupAllReadItems(application) {
   return totalCleaned;
 }
 
+/**
+ * Delete stale items: stripped skeletons and unread items older than MAX_ITEM_AGE_DAYS.
+ * Stripped skeletons have served their dedup purpose; stale unread items are unlikely
+ * to be read. Both are hard-deleted to prevent unbounded collection growth.
+ * @param {object} application - Indiekit application
+ * @returns {Promise<number>} Total number of items deleted
+ */
+export async function cleanupStaleItems(application) {
+  const collection = getCollection(application);
+  const cutoff = new Date();
+  cutoff.setDate(cutoff.getDate() - MAX_ITEM_AGE_DAYS);
+
+  // Delete stripped skeletons older than cutoff
+  const strippedResult = await collection.deleteMany({
+    _stripped: true,
+    $or: [
+      { published: { $lt: cutoff } },
+      { published: { $exists: false }, createdAt: { $lt: cutoff.toISOString() } },
+    ],
+  });
+
+  // Delete unread items older than cutoff
+  const unreadResult = await collection.deleteMany({
+    readBy: { $in: [null, []] },
+    _stripped: { $ne: true },
+    $or: [
+      { published: { $lt: cutoff } },
+      { published: { $exists: false }, createdAt: { $lt: cutoff.toISOString() } },
+    ],
+  });
+
+  // Also catch items with no readBy field at all
+  const noReadByResult = await collection.deleteMany({
+    readBy: { $exists: false },
+    _stripped: { $ne: true },
+    $or: [
+      { published: { $lt: cutoff } },
+      { published: { $exists: false }, createdAt: { $lt: cutoff.toISOString() } },
+    ],
+  });
+
+  const total =
+    strippedResult.deletedCount +
+    unreadResult.deletedCount +
+    noReadByResult.deletedCount;
+
+  if (total > 0) {
+    console.info(
+      `[Microsub] Stale cleanup: deleted ${strippedResult.deletedCount} stripped skeletons, ` +
+        `${unreadResult.deletedCount + noReadByResult.deletedCount} stale unread items ` +
+        `(cutoff: ${MAX_ITEM_AGE_DAYS} days)`,
+    );
+  }
+
+  return total;
+}
+
 export async function markItemsRead(application, channelId, entryIds, userId) {
   const collection = getCollection(application);
   const channelObjectId =
diff --git a/package.json b/package.json
index 9c03ef4..6b56fc0 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@rmdes/indiekit-endpoint-microsub",
-  "version": "1.0.42",
+  "version": "1.0.43",
   "description": "Microsub endpoint for Indiekit. Enables subscribing to feeds and reading content using the Microsub protocol.",
   "keywords": [
     "indiekit",