From ec6889bcdf0e96b7c950b2a8d29197671d03dc45 Mon Sep 17 00:00:00 2001
From: Eugene Shulga <ujeenator@gmail.com>
Date: Thu, 15 May 2025 13:08:06 +0200
Subject: [PATCH] implemented historical data for 7 days

---
 .gitignore         |  2 ++
 app.ts             | 66 ++++++++++++++++++++++++++++++++++++++++++----
 docker-compose.yml |  2 ++
 3 files changed, 65 insertions(+), 5 deletions(-)

diff --git a/.gitignore b/.gitignore
index a14702c..8b46890 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,5 @@
+data
+
 # dependencies (bun install)
 node_modules
 
diff --git a/app.ts b/app.ts
index a9c2645..9f06e94 100644
--- a/app.ts
+++ b/app.ts
@@ -2,9 +2,65 @@ import { Hono } from 'hono';
 import { serve } from 'bun';
 import Parser from 'rss-parser';
 import { Feed, type Item } from 'feed';
+import fs from 'fs';
+import path from 'path';
 
 const app = new Hono();
 const parser = new Parser();
+const DATA_DIR = path.resolve('.', 'data');
+
+// Ensure data directory exists
+if (!fs.existsSync(DATA_DIR)) fs.mkdirSync(DATA_DIR, { recursive: true });
+
+// Derive a safe filename from the feed URL
+function getStorePath(feedUrl: string): string {
+  const safeName = feedUrl.replace(/[^a-z0-9]/gi, '_').toLowerCase();
+  return path.join(DATA_DIR, `${safeName}.json`);
+}
+
+// Async feed: fetch remote, load+merge local, persist only on change, return merged data
+async function syncFeed(feedUrl: string): Promise<{ title: string; feedId: string; feedLink: string; items: Parser.Item[] }> {
+  // Fetch remote feed
+  const { title, feedUrl: feedId, link: feedLink, items: remoteItems } = await parser.parseURL(feedUrl);
+
+  // Determine storage path
+  const storePath = getStorePath(feedUrl);
+  let localItems: Parser.Item[] = [];
+  let localItemsJson = '';
+
+  // Load and parse local file once, reset if corrupted
+  if (fs.existsSync(storePath)) {
+    try {
+      localItemsJson = fs.readFileSync(storePath, 'utf-8');
+      localItems = JSON.parse(localItemsJson) as Parser.Item[];
+    } catch {
+      localItems = [];
+      localItemsJson = '';
+    }
+  }
+
+  const sevenDaysAgo = new Date();
+  sevenDaysAgo.setHours(0, 0, 0, 0);
+  sevenDaysAgo.setDate(sevenDaysAgo.getDate() - 7);
+
+  // Merge fresh items (dedupe by link)
+  const localLinks = new Set(localItems.map(i => i.link));
+  const mergedItems = [...localItems.filter(p => new Date(p.pubDate as string) >= sevenDaysAgo)];
+  for (const remoteItem of remoteItems) {
+    if (remoteItem.link && !localLinks.has(remoteItem.link)) mergedItems.push(remoteItem);
+  }
+
+  // Serialize merged
+  const mergedItemsJson = JSON.stringify(mergedItems, null, 2);
+
+  // Persist only if changed
+  if (mergedItemsJson !== localItemsJson) {
+    fs.writeFileSync(storePath, mergedItemsJson, 'utf-8');
+  }
+
+  return { title: title as string, feedId: feedId as string, feedLink: feedLink as string, items: mergedItems };
+}
+
 
 app.get('/group-by-day', async (c) => {
   const feedUrl = c.req.query('feedUrl');
@@ -12,13 +68,13 @@ app.get('/group-by-day', async (c) => {
     return c.text('Missing feedUrl query parameter', 400);
   }
   try {
-    const { title, feedUrl: feedId, link: feedLink, items } = await parser.parseURL(feedUrl);
+    const { title, feedId, feedLink, items } = await syncFeed(feedUrl);
 
-    const grouped = items.reduce((acc, { pubDate, content, contentSnippet, summary, link }) => {
+    const grouped = items.reduce((acc: Record<string, ({link: string, content: string})[]>, { pubDate, content, contentSnippet, summary, link }) => {
       const day = new Date(pubDate as string).toISOString().slice(0, 10);
       acc[day] = acc[day] || [];
       acc[day].push({
-        link,
+        link: link as string,
         content: content || contentSnippet || summary || ''
       });
       return acc;
@@ -38,8 +94,8 @@ app.get('/group-by-day', async (c) => {
         title: `${title} - ${dateStr}`,
         id: day,
         date: new Date(day),
-        link: grouped[day][0].link,
-        content: grouped[day].map((p: any) => p.content).join('\n\n---------------------\n\n')
+        link: grouped[day]?.at(0)?.link as string,
+        content: grouped[day]?.map((p: any) => p.content).join('\n\n---------------------\n\n') as string
       };
     }).filter(p => p.date < today);
 
diff --git a/docker-compose.yml b/docker-compose.yml
index 0d69f23..fd405dc 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -8,4 +8,6 @@ services:
       - PORT=${PORT:-3000}
     ports:
       - "${PORT:-3000}:${PORT:-3000}"
+    volumes:
+      - ./data:/app/data
     restart: unless-stopped
\ No newline at end of file