Retry oEmbed cache after redirect

This commit is contained in:
Christian Schmidt 2024-09-11 16:59:05 +02:00
parent a27f7f4e56
commit 41118205be
4 changed files with 48 additions and 11 deletions

View file

@ -107,11 +107,10 @@ class FetchLinkCardService < BaseService
def attempt_oembed
service = FetchOEmbedService.new
url_domain = Addressable::URI.parse(@url).normalized_host
cached_endpoint = Rails.cache.read("oembed_endpoint:#{url_domain}")
embed = service.call(@url, cached_endpoint: cached_endpoint) unless cached_endpoint.nil?
embed = service.call(@url, use_cached_endpoint: true)
embed ||= service.call(@url, html: html) unless html.nil?
embed ||= service.call(@url, use_cached_endpoint: true) if @url != @original_url.to_s
return false if embed.nil?

View file

@ -10,7 +10,7 @@ class FetchOEmbedService
@url = url
@options = options
if @options[:cached_endpoint]
if @options[:use_cached_endpoint]
parse_cached_endpoint!
else
discover_endpoint!
@ -57,9 +57,9 @@ class FetchOEmbedService
end
def parse_cached_endpoint!
cached = @options[:cached_endpoint]
cached = Rails.cache.read(cache_key)
return if cached[:endpoint].nil? || cached[:format].nil?
return if cached.nil? || cached[:endpoint].nil? || cached[:format].nil?
@endpoint_url = Addressable::Template.new(cached[:endpoint]).expand(url: @url).to_s
@format = cached[:format]
@ -68,14 +68,16 @@ class FetchOEmbedService
def cache_endpoint!
return unless URL_REGEX.match?(@endpoint_url)
url_domain = Addressable::URI.parse(@url).normalized_host
endpoint_hash = {
endpoint: @endpoint_url.gsub(URL_REGEX, '={url}'),
format: @format,
}
Rails.cache.write("oembed_endpoint:#{url_domain}", endpoint_hash, expires_in: ENDPOINT_CACHE_EXPIRES_IN)
Rails.cache.write(cache_key, endpoint_hash, expires_in: ENDPOINT_CACHE_EXPIRES_IN)
end
def cache_key
"oembed_endpoint:#{Addressable::URI.parse(@url).normalized_host}"
end
def fetch!

View file

@ -13,6 +13,7 @@ RSpec.describe FetchLinkCardService do
stub_request(:get, 'http://example.com/not-found').to_return(status: 404, headers: { 'Content-Type' => 'text/html' }, body: html)
stub_request(:get, 'http://example.com/text').to_return(status: 404, headers: { 'Content-Type' => 'text/plain' }, body: 'Hello')
stub_request(:get, 'http://example.com/redirect').to_return(status: 302, headers: { 'Location' => 'http://example.com/html' })
stub_request(:get, 'http://example.net/redirect-to-other-domain').to_return(status: 302, headers: { 'Location' => 'http://example.com/html' })
stub_request(:get, 'http://example.com/redirect-to-404').to_return(status: 302, headers: { 'Location' => 'http://example.com/not-found' })
stub_request(:get, 'http://example.com/oembed?url=http://example.com/html').to_return(headers: { 'Content-Type' => 'application/json' }, body: '{ "version": "1.0", "type": "link", "title": "oEmbed title" }')
stub_request(:get, 'http://example.com/oembed?format=json&url=http://example.com/html').to_return(headers: { 'Content-Type' => 'application/json' }, body: '{ "version": "1.0", "type": "link", "title": "oEmbed title" }')
@ -264,6 +265,39 @@ RSpec.describe FetchLinkCardService do
end
end
context 'when oEmbed endpoint cache populated with redirect target' do
let(:status) { Fabricate(:status, text: 'http://example.net/redirect-to-other-domain') }
let(:oembed_cache) { { endpoint: 'http://example.com/oembed?format=json&url={url}', format: :json } }
it 'uses the cached oEmbed response' do
expect(a_request(:get, 'http://example.net/redirect-to-other-domain')).to have_been_made
expect(a_request(:get, 'http://example.com/oembed?url=http://example.com/html')).to have_been_made
end
it 'creates preview card' do
expect(status.preview_card).to_not be_nil
expect(status.preview_card.url).to eq 'http://example.com/html'
expect(status.preview_card.title).to eq 'oEmbed title'
end
end
context 'when oEmbed endpoint cache populated with redirect target but page contains no oEmbed tags' do
let(:status) { Fabricate(:status, text: 'http://example.net/redirect-to-other-domain') }
let(:html) { 'Please fill out CAPTCHA' }
let(:oembed_cache) { { endpoint: 'http://example.com/oembed?format=json&url={url}', format: :json } }
it 'uses the cached oEmbed response' do
expect(a_request(:get, 'http://example.net/redirect-to-other-domain')).to have_been_made
expect(a_request(:get, 'http://example.com/oembed?format=json&url=http://example.com/html')).to have_been_made
end
it 'creates preview card' do
expect(status.preview_card).to_not be_nil
expect(status.preview_card.url).to eq 'http://example.com/html'
expect(status.preview_card.title).to eq 'oEmbed title'
end
end
# If the original HTML URL for whatever reason (e.g. DOS protection) redirects to
# an error page, we can still use the cached oEmbed but should not use the
# redirect URL on the card.

View file

@ -160,10 +160,12 @@ RSpec.describe FetchOEmbedService do
headers: { 'Content-Type': 'text/html' },
body: request_fixture('oembed_json_empty.html')
)
Rails.cache.write('oembed_endpoint:www.youtube.com', { endpoint: 'http://www.youtube.com/oembed?format=json&url={url}', format: :json })
end
it 'returns new provider without fetching original URL first' do
subject.call('https://www.youtube.com/watch?v=dqwpQarrDwk', cached_endpoint: { endpoint: 'http://www.youtube.com/oembed?format=json&url={url}', format: :json })
subject.call('https://www.youtube.com/watch?v=dqwpQarrDwk', use_cached_endpoint: true)
expect(a_request(:get, 'https://www.youtube.com/watch?v=dqwpQarrDwk')).to_not have_been_made
expect(subject.endpoint_url).to eq 'http://www.youtube.com/oembed?format=json&url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DdqwpQarrDwk'
expect(subject.format).to eq :json