fix: clean up stale and orphaned entries from subscriptions.json

Clear jsonStore before seeding from MongoDB on startup to prevent stale
disk cache entries from persisting. Add cleanup for orphaned resource
documents that have no corresponding subscription. Run cleanup
immediately on startup instead of waiting for the first 24-hour cycle.

Author: andrewshell

app.js

@@ -18,6 +18,11 @@ console.log(`${config.appName} ${config.appVersion}`);
 
 // Schedule cleanup tasks
 function scheduleCleanupTasks() {
+    // Run cleanup immediately on startup
+    removeExpiredSubscriptions()
+        .then(() => console.log('Startup subscription cleanup completed'))
+        .catch(err => console.error('Error in startup subscription cleanup:', err));
+
     // Run subscription cleanup every 24 hours
     setInterval(async() => {
         try {
@@ -70,6 +75,7 @@ app.use(require('./controllers'));
 
 // Start server
 async function seedJsonStore() {
+    jsonStore.clear();
     const db = mongodb.get('rsscloud');
     const resources = await db.collection('resources').find({}).toArray();
     const subscriptions = await db.collection('subscriptions').find({}).toArray();

services/remove-expired-subscriptions.js

@@ -134,14 +134,48 @@ async function removeExpiredSubscriptions() {
             console.log(`Created ${resourcesCreated} missing resource documents`);
         }
 
-        console.log(`Subscription cleanup completed: ${totalRemoved} expired/errored subscriptions removed, ${documentsProcessed} documents processed, ${documentsDeleted} empty documents deleted, ${urlsFixed} URLs fixed`);
+        // Find resources with no corresponding subscription and remove them
+        let orphanedResourcesRemoved = 0;
+        const orphanedResourcesCursor = db.collection('resources').aggregate([
+            {
+                $lookup: {
+                    from: 'subscriptions',
+                    localField: '_id',
+                    foreignField: '_id',
+                    as: 'subscription'
+                }
+            },
+            {
+                $match: {
+                    subscription: { $size: 0 }
+                }
+            }
+        ]);
+
+        while (await orphanedResourcesCursor.hasNext()) {
+            const doc = await orphanedResourcesCursor.next();
+            // Skip recently-checked resources (preserved by the subscription cleanup above)
+            if (doc.whenLastCheck && dayjs(doc.whenLastCheck).isAfter(dayjs().subtract(24, 'hours'))) {
+                continue;
+            }
+            await db.collection('resources').deleteOne({ _id: doc._id });
+            jsonStore.removeEntry(doc._id);
+            orphanedResourcesRemoved++;
+        }
+
+        if (orphanedResourcesRemoved > 0) {
+            console.log(`Removed ${orphanedResourcesRemoved} orphaned resource documents`);
+        }
+
+        console.log(`Subscription cleanup completed: ${totalRemoved} expired/errored subscriptions removed, ${documentsProcessed} documents processed, ${documentsDeleted} empty documents deleted, ${urlsFixed} URLs fixed, ${orphanedResourcesRemoved} orphaned resources removed`);
 
         return {
             subscriptionsRemoved: totalRemoved,
             documentsProcessed,
             documentsDeleted,
             resourcesCreated,
-            urlsFixed
+            urlsFixed,
+            orphanedResourcesRemoved
         };
 
     } catch (error) {

test/remove-expired-subscriptions.js

@@ -155,6 +155,32 @@ describe('RemoveExpiredSubscriptions', function() {
         expect(resDoc).to.not.be.null;
     });
 
+    it('should remove orphaned resource with no subscription document', async function() {
+        const feedPath = '/rss.xml',
+            resourceUrl = mock.serverUrl + feedPath,
+            dayjs = await getDayjs();
+
+        // Add resource but no subscription document (last checked 48 hours ago)
+        await mongodb.addResource(resourceUrl, {
+            lastHash: 'abc',
+            lastSize: 100,
+            ctChecks: 1,
+            ctUpdates: 0,
+            whenLastCheck: new Date(dayjs().utc().subtract(48, 'hours').format())
+        });
+        jsonStore.setResource(resourceUrl, { lastHash: 'abc', lastSize: 100 });
+
+        await removeExpiredSubscriptions();
+
+        // Resource document should be removed
+        const resDoc = await mongodb.findResource(resourceUrl);
+        expect(resDoc).to.be.null;
+
+        // JSON store entry should be fully removed
+        const storeData = jsonStore.getData();
+        expect(storeData).to.not.have.property(resourceUrl);
+    });
+
     it('should create resource for subscription without one', async function() {
         const feedPath = '/rss.xml',
             resourceUrl = mock.serverUrl + feedPath,