feat: read subscriptions from an Atom feed when a page lists no dated links

This commit is contained in:
randogoth 2026-10-11 13:35:19 +03:00
parent b0b86ef5b1
commit fa51d5793d
9 changed files with 411 additions and 4 deletions

View file

@ -7,6 +7,7 @@ import 'package:mews/application/controllers/reader_settings_controller.dart';
import 'package:mews/application/controllers/subscriptions_controller.dart';
import 'package:mews/database/database.dart';
import 'package:mews/mews/document.dart';
import 'package:mews/mews/parser.dart';
import 'fake_fetch.dart';
@ -207,6 +208,90 @@ void main() {
expect(redraws, 2);
});
test('a site publishing only through a feed still reports posts', () async {
const site = 'https://feedonly.example/';
const feed = 'https://feedonly.example/feed.xml';
// A page with no dated links, but it declares a feed (spec 6.2).
const page = '''
<?xml version="1.0" encoding="UTF-8"?>
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<title>Feed Only</title>
<meta name="mews-profile" content="0.1" />
<link rel="alternate" type="application/atom+xml" href="feed.xml" />
</head>
<body><h1>Feed Only</h1><p>Nothing listed here.</p></body>
</html>
''';
String feedWith(List<(String, String)> entries) => '''
<feed xmlns="http://www.w3.org/2005/Atom">
${entries.map((e) => '<entry><title>${e.$2}</title><link rel="alternate" href="${e.$1}" /><updated>2026-10-08T09:30:00Z</updated></entry>').join()}
</feed>
''';
final feeds = {feed: feedWith(const [('old.html', 'An older post')])};
final subscriptions = SubscriptionsController(
db: db,
fetch: FakeFetch({site: page}, feeds: feeds),
);
await subscriptions.load();
await subscriptions.toggle(parseMewsPage(source: page, url: site));
// What the feed already listed is the back catalogue, not news.
expect(await db.unreadCount(), 0);
expect((await db.allSubscriptions()).single.atomHref, feed);
// A post added to the feed later is new.
feeds[feed] = feedWith(const [
('new.html', 'A newer post'),
('old.html', 'An older post'),
]);
await subscriptions.refreshAll();
expect(await db.unreadCount(), 1);
final entries = await db.riverEntries(unreadOnly: true);
expect(entries.single.title, 'A newer post');
// Checking again reports nothing, because an address identifies a post.
await subscriptions.refreshAll();
expect(await db.unreadCount(), 1);
});
test('a page listing dated links is not fetched twice', () async {
const site = 'https://both.example/';
const page = '''
<?xml version="1.0" encoding="UTF-8"?>
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<title>Both</title>
<meta name="mews-profile" content="0.1" />
<link rel="alternate" type="application/atom+xml" href="feed.xml" />
</head>
<body><h1>Both</h1><ul>
<li><a href="one.html">2026-10-08 The only post</a></li>
</ul></body>
</html>
''';
// The feed would answer, but a page that lists its own posts is enough,
// so nothing asks for it.
final subscriptions = SubscriptionsController(
db: db,
fetch: FakeFetch({site: page}, feeds: const {
'https://both.example/feed.xml':
'<feed xmlns="http://www.w3.org/2005/Atom"><entry><title>Duplicate</title>'
'<link rel="alternate" href="one.html" /><updated>2026-10-08T09:30:00Z</updated>'
'</entry></feed>',
}),
);
await subscriptions.load();
await subscriptions.toggle(parseMewsPage(source: page, url: site));
await subscriptions.refreshAll();
// One post, under the title the page gave it.
final entries = await db.riverEntries();
expect(entries, hasLength(1));
expect(entries.single.title, 'The only post');
});
test('caches pages for offline reopening', () async {
await db.saveCachedPage(CachedPagesCompanion.insert(
url: 'https://example.com/page.html',