News.
Authored by tinyfiles.tv, plus wire headlines from AP and Reuters via free RSS. Two kinds of news, labeled. If it matters, click through to the original.
This page is a small, personal news page — not a wire service. Authored means we wrote it: opinion, site notes, explainers. Wire means the headline + link came from AP or Reuters. We do not rewrite wire headlines. We do not pretend our take is the wire. If a story matters to you, open the original and check it there.
Our personal news source is the Daily Token broadcast — 15 minutes, no yelling. This page is the written companion to that.
Wire — AP + Reuters
Free RSS via Google News site: feeds. Headlines link to the original.AP and Reuters no longer offer working official public RSS feeds — AP’s broke in 2024, Reuters closed theirs in 2020. So this section reads their stories through free Google News RSS feeds limited to site:apnews.com and site:reuters.com. No key, no paywall on the feed itself. The browser can’t fetch RSS directly (CORS), so the page tries a public JSON proxy first, then a CORS proxy, then direct — and always shows direct links if the fetch fails.
Copy any URL below into your RSS reader. They are free, no key. If one breaks, try the next — feed URLs change.
AP — via Google News (primary, free): https://news.google.com/rss/search?q=when:24h+site:apnews.com&hl=en-US&gl=US&ceid=US:en Reuters — via Google News (primary, free): https://news.google.com/rss/search?q=when:24h+site:reuters.com&ceid=US:en&hl=en-US&gl=US AP — backups (if Google News feed fails): https://flipboard.com/@associatedpress/top-stories-u7govdf1z.rss https://rsshub.app/apnews/topics/apf-topnews
Heads-up: the Google News URLs return AP / Reuters stories that Google News indexed. The story links go to the original AP / Reuters article. We don’t host or rewrite the wire copy.
Pictures — Click Between Them
JavaScript gallery. No feed, no algorithm. Just pictures + nicer things.The news above can be a lot. This part is the opposite. Eight pictures, eight nicer things. Click a thumbnail, or Prev / Next, or Shuffle. The big picture changes. The nice thing changes with it.
Sunrise
Good morning. You don\'t have to be perfect today to deserve a good one.
News Backend — Proposed Architecture
Right now this page is local-only: authored articles are hard-coded HTML, and the wire is fetched live in the browser through a public JSON proxy (rss2json) + a CORS proxy (allorigins) because browsers block direct RSS fetches (CORS). That works for a demo, but it depends on third-party proxies, rate limits, and feed URLs that change. This is the map to make it real: server fetches + caches the wire, authored articles live in the DB, and the front-end reads from our own API — never from Google News directly.
01 / Files to create
api/news/list.php GET /api/news/list?tab=all&q= api/news/article.php GET /api/news/article?id=3 api/news/rss.php GET /api/news/rss (our own RSS out, optional) admin/news_write.php POST /admin/news_write (authored only, login required) cron/fetch_wire.php CLI — fetch + cache AP + Reuters, every 15 min sql/news_tables.sql
Keep the tinyfiles PHP pattern: each endpoint loads /api/bootstrap.php, uses json_in() / json_out(), member session for admin, prepared PDO statements. The cron is the only thing that ever talks to Google News / Flipboard / RSSHub.
02 / Database tables
-- sql/news_tables.sql
news_articles
id INT AUTO_INCREMENT PRIMARY KEY
slug VARCHAR(120) UNIQUE NOT NULL -- e.g. 'daily-token-15-minutes'
kicker VARCHAR(80) NOT NULL
title VARCHAR(200) NOT NULL
byline VARCHAR(120) NOT NULL -- 'By tinyfiles.tv'
published_at DATE NOT NULL
read_minutes TINYINT NOT NULL
excerpt TEXT NOT NULL
body_html TEXT NOT NULL -- already-sanitized HTML, no scripts
tags VARCHAR(200) NOT NULL
status ENUM('draft','published') NOT NULL DEFAULT 'draft'
created_at DATETIME NOT NULL
updated_at DATETIME NOT NULL
news_feeds
id VARCHAR(20) PRIMARY KEY -- 'ap', 'reuters'
label VARCHAR(60) NOT NULL
feed_url TEXT NOT NULL -- the Google News site: URL
backup_url TEXT NULL
active TINYINT NOT NULL DEFAULT 1
news_wire_cache
id INT AUTO_INCREMENT PRIMARY KEY
feed_id VARCHAR(20) NOT NULL -- 'ap' / 'reuters'
guid VARCHAR(300) NOT NULL -- dedupe key: RSS guid or link hash
title VARCHAR(300) NOT NULL
link TEXT NOT NULL -- original AP / Reuters URL
pub_date DATETIME NULL
description TEXT NULL -- stripped, truncated, no HTML
fetched_at DATETIME NOT NULL
UNIQUE KEY (feed_id, guid)
FOREIGN KEY (feed_id) REFERENCES news_feeds(id)
Authored and wire stay in separate tables on purpose. Authored is ours to edit. Wire is a cache of someone else’s headlines — we never edit the title, we only store it + link to the original.
03 / API endpoints & request flow
GET /api/news/list?tab=all&q=
- returns { "authored": [...], "wire": [...] }
- authored: published rows only, newest first, body_html included
- wire: from news_wire_cache, newest first, per feed
- q= filters both by title/excerpt/tags (server-side LIKE, escaped)
GET /api/news/article?id=3
- returns one authored article (for a future article page)
Cron: cron/fetch_wire.php (every 15 min)
1. For each active row in news_feeds, fetch feed_url server-side
(server-side = no CORS problem, no rss2json needed)
2. Parse RSS XML with SimpleXML / DOM, take <item> title/link/pubDate/description
3. Dedupe on (feed_id, guid). Insert only new items.
4. Trim: keep last 100 items per feed, delete older.
5. If primary feed fails, try backup_url, log failure, keep old cache.
Front-end (production):
- Page loads, calls /api/news/list — one request, no proxies.
- If /api/news/list fails, fall back to the hard-coded authored HTML
+ the live browser RSS path you see on this page today.
- Refresh button = re-call /api/news/list, not re-fetch Google News.
Why server-side fetch matters: browsers block fetch(https://news.google.com/rss/...) with CORS. A server cron doesn’t have that problem. It also lets us cache (so 1,000 readers don’t hammer Google News), dedupe, and strip bad HTML once instead of in every browser.
04 / Validation & rules
// Server owns the truth. Browser submits intent only. Authored (admin only): - title/excerpt/body required, slug unique, status draft/published only. - body_html: strip <script>, <iframe>, on* attributes, javascript: links. Allow only p, h3, a, b, i, ul, ol, li, code, blockquote. - byline is fixed to the logged-in author — never trust a byline from the form. Wire (cron only, never from browser): - Accept only http(s) links. Reject javascript:, data:, file:. - Title: plain text, max 300 chars, strip all tags. - Description: strip tags, truncate to 400 chars, no HTML. - Link must host-match: apnews.com / reuters.com / news.google.com (for the Google News wrapper). Anything else = drop + log. - pubDate must parse as a date, else store NULL, keep the item. - Never rewrite a wire title. If it needs a correction, link to the original.
05 / Adding a feed / article
-- Add a new wire source (no code change):
INSERT INTO news_feeds (id, label, feed_url) VALUES
('bbc', 'BBC', 'https://feeds.bbci.co.uk/news/rss.xml');
-- Publish an authored article:
INSERT INTO news_articles (slug, kicker, title, byline, published_at, excerpt, body_html, tags, status)
VALUES ('my-take', 'Opinion', 'My take', 'By tinyfiles.tv', '2026-11-18', '...', '<p>...</p>', 'Authored,Opinion', 'published');
New feed = one row in news_feeds, cron picks it up in the next 15-minute run. New authored article = one row, front-end list picks it up. No deploy for new copy.