From fbcaab53d03a1d0111367f7cea539f27eddaf89d Mon Sep 17 00:00:00 2001
From: Bruno Bernardino <me@brunobernardino.com>
Date: Mon, 14 Oct 2024 15:13:55 +0100
Subject: [PATCH] Add Cron to delete old, read articles. Update Deno.

---
 .dvmrc                            |  2 +-
 Dockerfile                        |  2 +-
 crons/index.ts                    | 10 +++++---
 crons/news.ts                     | 41 +++++++++++++++++++++++++++++++
 crons/{cleanup.ts => sessions.ts} |  0
 5 files changed, 50 insertions(+), 5 deletions(-)
 rename crons/{cleanup.ts => sessions.ts} (100%)

diff --git a/.dvmrc b/.dvmrc
index 94ed04f..a82727c 100644
--- a/.dvmrc
+++ b/.dvmrc
@@ -1 +1 @@
-1.46.2
+1.46.3
diff --git a/Dockerfile b/Dockerfile
index e46b810..a00f264 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,4 +1,4 @@
-FROM denoland/deno:ubuntu-1.46.2
+FROM denoland/deno:ubuntu-1.46.3
 
 EXPOSE 8000
 
diff --git a/crons/index.ts b/crons/index.ts
index f380a6c..9368994 100644
--- a/crons/index.ts
+++ b/crons/index.ts
@@ -1,8 +1,8 @@
-import { Cron } from 'https://deno.land/x/croner@8.0.1/dist/croner.js';
+import { Cron } from 'https://deno.land/x/croner@8.1.2/dist/croner.js';
 
 import { isAppEnabled } from '/lib/config.ts';
-import { cleanupSessions } from './cleanup.ts';
-import { fetchNewArticles } from './news.ts';
+import { cleanupSessions } from './sessions.ts';
+import { cleanupOldArticles, fetchNewArticles } from './news.ts';
 
 export function startCrons() {
   new Cron(
@@ -14,6 +14,10 @@ export function startCrons() {
     },
     async () => {
       await cleanupSessions();
+
+      if (isAppEnabled('news')) {
+        await cleanupOldArticles();
+      }
     },
   );
 
diff --git a/crons/news.ts b/crons/news.ts
index 84f2fcc..3d2cfe0 100644
--- a/crons/news.ts
+++ b/crons/news.ts
@@ -25,3 +25,44 @@ export async function fetchNewArticles(forceFetch = false) {
     console.error(error);
   }
 }
+
+export async function cleanupOldArticles() {
+  const oneMonthAgo = new Date(new Date().setUTCMonth(new Date().getUTCMonth() - 1));
+
+  try {
+    console.info('Will cleanup old articles');
+
+    const feedIdsToSkip = new Set<string>();
+
+    const feeds = await db.query<Pick<NewsFeed, 'id' | 'feed_url'>>(
+      sql`SELECT "id", "feed_url" FROM "bewcloud_news_feeds" ORDER BY "last_crawled_at" ASC`,
+    );
+
+    for (const feed of feeds) {
+      const recentArticlesCount = (await db.query<{ count: number }>(
+        sql`SELECT COUNT("id") AS "count" FROM "bewcloud_news_feed_articles" WHERE "feed_id" = $1 AND "article_date" >= $2`,
+        [feed.id, oneMonthAgo],
+      ))[0].count;
+
+      // Don't delete old articles if the feed doesn't have recent articles (to skip feeds with less items in total)
+      if (recentArticlesCount <= 5) {
+        feedIdsToSkip.add(feed.id);
+      }
+    }
+
+    const result = await db.query<{ count: number }>(
+      sql`WITH "deleted" AS (
+        DELETE FROM "bewcloud_news_feed_articles" WHERE "is_read" = TRUE AND "article_date" <= $1 AND "feed_id" != ANY($2) RETURNING *
+      )
+        SELECT COUNT(*) FROM "deleted"`,
+      [
+        oneMonthAgo.toISOString().substring(0, 10),
+        [...feedIdsToSkip],
+      ],
+    );
+
+    console.info('Deleted', result[0].count, 'old articles');
+  } catch (error) {
+    console.error(error);
+  }
+}
diff --git a/crons/cleanup.ts b/crons/sessions.ts
similarity index 100%
rename from crons/cleanup.ts
rename to crons/sessions.ts