From b58db8f12eb19787ee3bd1ec8abab21027b3d4ef Mon Sep 17 00:00:00 2001
From: Eugen Rochko <eugen@zeonfederated.com>
Date: Fri, 25 Mar 2022 19:31:35 +0100
Subject: [PATCH] Add workaround for YouTube Shorts links (#17869)
* Add workaround for YouTube Shorts links
* Update link_details_extractor_spec.rb
---
app/lib/link_details_extractor.rb | 2 +-
spec/lib/link_details_extractor_spec.rb | 8 ++++++++
2 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/app/lib/link_details_extractor.rb b/app/lib/link_details_extractor.rb
index fabbd244df..b0c4e4f425 100644
--- a/app/lib/link_details_extractor.rb
+++ b/app/lib/link_details_extractor.rb
@@ -208,7 +208,7 @@ class LinkDetailsExtractor
end
def valid_url_or_nil(str, same_origin_only: false)
- return if str.blank?
+ return if str.blank? || str == 'null'
url = @original_url + Addressable::URI.parse(str)
diff --git a/spec/lib/link_details_extractor_spec.rb b/spec/lib/link_details_extractor_spec.rb
index 84bb4579ca..7ea867c614 100644
--- a/spec/lib/link_details_extractor_spec.rb
+++ b/spec/lib/link_details_extractor_spec.rb
@@ -25,6 +25,14 @@ RSpec.describe LinkDetailsExtractor do
expect(subject.canonical_url).to eq 'https://foo.com/article'
end
end
+
+ context 'when canonical URL is set to "null"' do
+ let(:html) { '<!doctype html><link rel="canonical" href="null" />' }
+
+ it 'ignores the canonical URLs' do
+ expect(subject.canonical_url).to eq original_url
+ end
+ end
end
context 'when structured data is present' do