require 'rails_helper'
RSpec.describe LinkDetailsExtractor do
subject { described_class.new(original_url, html, html_charset) }
let(:original_url) { '' }
let(:html) { '' }
let(:html_charset) { nil }
describe '#canonical_url' do
let(:original_url) { 'https://foo.com/article?bar=baz123' }
context 'when canonical URL points to another host' do
let(:html) { '' }
it 'ignores the canonical URLs' do
expect(subject.canonical_url).to eq original_url
end
end
context 'when canonical URL points to the same host' do
let(:html) { '' }
it 'ignores the canonical URLs' do
expect(subject.canonical_url).to eq 'https://foo.com/article'
end
end
context 'when canonical URL is set to "null"' do
let(:html) { '' }
it 'ignores the canonical URLs' do
expect(subject.canonical_url).to eq original_url
end
end
end
context 'when structured data is present' do
let(:original_url) { 'https://example.com/page.html' }
context 'and is wrapped in CDATA tags' do
let(:html) { <<~HTML }
HTML
describe '#title' do
it 'returns the title from structured data' do
expect(subject.title).to eq 'Foo'
end
end
describe '#description' do
it 'returns the description from structured data' do
expect(subject.description).to eq 'Bar'
end
end
describe '#provider_name' do
it 'returns the provider name from structured data' do
expect(subject.provider_name).to eq 'Baz'
end
end
describe '#author_name' do
it 'returns the author name from structured data' do
expect(subject.author_name).to eq 'Hoge'
end
end
end
context 'but the first tag is invalid JSON' do
let(:html) { <<~HTML }
HTML
describe '#title' do
it 'returns the title from structured data' do
expect(subject.title).to eq 'Foo'
end
end
describe '#description' do
it 'returns the description from structured data' do
expect(subject.description).to eq 'Bar'
end
end
describe '#provider_name' do
it 'returns the provider name from structured data' do
expect(subject.provider_name).to eq 'Baz'
end
end
describe '#author_name' do
it 'returns the author name from structured data' do
expect(subject.author_name).to eq 'Hoge'
end
end
end
end
end