Do not pass unknown encoding names to nokogiri. (#30987)

This commit is contained in:
David Roetzel 2024-07-10 16:25:39 +02:00 committed by GitHub
parent 36592d10aa
commit 2ea9336b68
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 35 additions and 1 deletions

View File

@ -274,7 +274,7 @@ class LinkDetailsExtractor
end end
def detect_encoding_and_parse_document def detect_encoding_and_parse_document
[detect_encoding, nil, @html_charset].uniq.each do |encoding| [detect_encoding, nil, header_encoding].uniq.each do |encoding|
document = Nokogiri::HTML(@html, nil, encoding) document = Nokogiri::HTML(@html, nil, encoding)
return document if document.to_s.valid_encoding? return document if document.to_s.valid_encoding?
end end
@ -286,6 +286,13 @@ class LinkDetailsExtractor
guess&.fetch(:confidence, 0).to_i > 60 ? guess&.fetch(:encoding, nil) : nil guess&.fetch(:confidence, 0).to_i > 60 ? guess&.fetch(:encoding, nil) : nil
end end
def header_encoding
Encoding.find(@html_charset).name if @html_charset
rescue ArgumentError
# Encoding from HTTP header is not recognized by ruby
nil
end
def detector def detector
@detector ||= CharlockHolmes::EncodingDetector.new.tap do |detector| @detector ||= CharlockHolmes::EncodingDetector.new.tap do |detector|
detector.strip_tags = true detector.strip_tags = true

View File

@ -0,0 +1,18 @@
HTTP/1.1 200 OK
server: nginx
date: Thu, 13 Jun 2024 14:33:13 GMT
content-type: text/html; charset=utf8
content-length: 192
accept-ranges: bytes
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Webserver Configs R Us</title>
</head>
<body>
<h2>Welcome</h2>
<p>Sneaky non-UTF character: á</p>
</body>
</html>

View File

@ -32,6 +32,7 @@ RSpec.describe FetchLinkCardService do
stub_request(:get, 'http://example.com/aergerliche-umlaute').to_return(request_fixture('redirect_with_utf8_url.txt')) stub_request(:get, 'http://example.com/aergerliche-umlaute').to_return(request_fixture('redirect_with_utf8_url.txt'))
stub_request(:get, 'http://example.com/page_without_title').to_return(request_fixture('page_without_title.txt')) stub_request(:get, 'http://example.com/page_without_title').to_return(request_fixture('page_without_title.txt'))
stub_request(:get, 'http://example.com/long_canonical_url').to_return(request_fixture('long_canonical_url.txt')) stub_request(:get, 'http://example.com/long_canonical_url').to_return(request_fixture('long_canonical_url.txt'))
stub_request(:get, 'http://example.com/alternative_utf8_spelling_in_header').to_return(request_fixture('alternative_utf8_spelling_in_header.txt'))
Rails.cache.write('oembed_endpoint:example.com', oembed_cache) if oembed_cache Rails.cache.write('oembed_endpoint:example.com', oembed_cache) if oembed_cache
@ -292,6 +293,14 @@ RSpec.describe FetchLinkCardService do
expect(status.preview_card).to be_nil expect(status.preview_card).to be_nil
end end
end end
context 'with a URL where the `Content-Type` header uses `utf8` instead of `utf-8`' do
let(:status) { Fabricate(:status, text: 'test http://example.com/alternative_utf8_spelling_in_header') }
it 'does not create a preview card' do
expect(status.preview_card.title).to eq 'Webserver Configs R Us'
end
end
end end
context 'with a remote status' do context 'with a remote status' do