In working on a new implementation, it began to look like tests were passing when they shouldn't have passed. I made a branch off main and confirmed that we weren't testing the order of elements, which is significant in HTML.
Once that was fixed, it turned out that our code for handling some of the variations was also passing tests that it shouldn't have been passing. A couple of more changes fixed that.
Finally, three tests were failing because the expected strings were incorrect, but the other issues mentioned above were hiding that the tests should have failed.
In the process of fixing, we changed to use Nokogiri's HTML instead of XML. And to make it clear, the assertion helper was changed in all the test files.
In working on a new implementation, it began to look like tests were passing when they shouldn't have passed. I made a branch off
main
and confirmed that we weren't testing the order of elements, which is significant in HTML.Once that was fixed, it turned out that our code for handling some of the variations was also passing tests that it shouldn't have been passing. A couple of more changes fixed that.
Finally, three tests were failing because the expected strings were incorrect, but the other issues mentioned above were hiding that the tests should have failed.
In the process of fixing, we changed to use Nokogiri's HTML instead of XML. And to make it clear, the assertion helper was changed in all the test files.
Fixes: #674