Improve encoding detection for link cards (#30780)

This commit is contained in:
David Roetzel 2024-06-21 14:51:10 +02:00 committed by KMY
parent f60b7fe905
commit 650875a820
3 changed files with 36 additions and 5 deletions

View file

@ -28,6 +28,7 @@ RSpec.describe FetchLinkCardService do
stub_request(:get, 'http://example.com/sjis_with_wrong_charset').to_return(request_fixture('sjis_with_wrong_charset.txt'))
stub_request(:get, 'http://example.com/koi8-r').to_return(request_fixture('koi8-r.txt'))
stub_request(:get, 'http://example.com/windows-1251').to_return(request_fixture('windows-1251.txt'))
stub_request(:get, 'http://example.com/low_confidence_latin1').to_return(request_fixture('low_confidence_latin1.txt'))
Rails.cache.write('oembed_endpoint:example.com', oembed_cache) if oembed_cache
@ -150,6 +151,14 @@ RSpec.describe FetchLinkCardService do
end
end
context 'with a URL of a page in ISO-8859-1 encoding, that charlock_holmes cannot detect' do
let(:status) { Fabricate(:status, text: 'Check out http://example.com/low_confidence_latin1') }
it 'decodes the HTML' do
expect(status.preview_card.title).to eq("Tofu á l'orange")
end
end
context 'with a Japanese path URL' do
let(:status) { Fabricate(:status, text: 'テストhttp://example.com/日本語') }