diff --git a/lib/onebox/helpers.rb b/lib/onebox/helpers.rb
index 6b8f4d3f251..10dd9fa0e63 100644
--- a/lib/onebox/helpers.rb
+++ b/lib/onebox/helpers.rb
@@ -36,7 +36,7 @@ module Onebox
# prefer canonical link
canonical_link = doc.at('//link[@rel="canonical"]/@href')
canonical_uri = Addressable::URI.parse(canonical_link)
- if canonical_link && "#{canonical_uri.host}#{canonical_uri.path}" != "#{uri.host}#{uri.path}"
+ if canonical_link && "#{canonical_uri.host}#{canonical_uri.path}" != "#{uri.host}#{uri.path}" && canonical_uri.host != "localhost"
response = (fetch_response(canonical_uri.to_s, headers: headers, body_cacher: body_cacher) rescue nil)
doc = Nokogiri::HTML(response) if response
end
diff --git a/spec/lib/onebox/helpers_spec.rb b/spec/lib/onebox/helpers_spec.rb
index 6532905f501..f24719ff20f 100644
--- a/spec/lib/onebox/helpers_spec.rb
+++ b/spec/lib/onebox/helpers_spec.rb
@@ -52,6 +52,23 @@ RSpec.describe Onebox::Helpers do
expect(described_class.fetch_html_doc(uri).to_s).to match("success")
end
+
+ context "canonical link" do
+ it "follows canonical link" do
+ uri = 'https://www.example.com'
+ stub_request(:get, uri).to_return(status: 200, body: "invalid
")
+ stub_request(:get, 'http://foobar.com').to_return(status: 200, body: "success
")
+
+ expect(described_class.fetch_html_doc(uri).to_s).to match("success")
+ end
+
+ it "does not follow canonical link pointing at localhost" do
+ uri = 'https://www.example.com'
+ stub_request(:get, uri).to_return(status: 200, body: "success
")
+
+ expect(described_class.fetch_html_doc(uri).to_s).to match("success")
+ end
+ end
end
describe "redirects" do