Validation broken in Genesis themes by Document::normalize_document_structure() #4104
Description
Bug Description
I was testing out a Genesis theme and I saw in the admin bar that it was ✅ valid AMP. When I clicked to validate the page, however, I unexpectedly saw two validation errors:
This made no sense to me and I was racking my brain trying to figure it out. It turns out to be that the link[rel=canonical] element was located in the body during validation, when for non-validation requests it is in the head as expected.
The issue turns out to be \Amp\AmpWP\Dom\Document::normalize_document_structure() introduced in #3758, namely that it is not currently written to account for HTML comments occuring before the doctype.
Normally, WordPress themes hard-code the <!DOCTYPE html>, <html>, and <head> tags. Genesis, however, uses genesis_doctype() to generate them. When a validation request is being performed, source stack comments are added before and after genesis_doctype() like so:
<!--amp-source-stack {"type":"theme","name":"genesis","file":"lib\/structure\/header.php","line":28,"function":"genesis_do_doctype","hook":"genesis_doctype","priority":10}--><!DOCTYPE html>
<html lang="en-US">
<head >
<meta charset="UTF-8" />
<!--/amp-source-stack {"type":"theme","name":"genesis","file":"lib\/structure\/header.php","line":28,"function":"genesis_do_doctype","hook":"genesis_doctype","priority":10}-->This has the effect of Document::normalize_document_structure() normalizing the document to start as:
<!DOCTYPE html><html><head></head><body><!--amp-source-stack {"type":"theme","name":"genesis","file":"lib\/structure\/header.php","line":28,"function":"genesis_do_doctype","hook":"genesis_doctype","priority":10}--><!DOCTYPE html>
<html lang="en-US">
<head >Expected Behaviour
The DOM used for validation should be normalized to be the same as the DOM used for rendering.
Steps to reproduce
- Activate Genesis.
- Enable Transitional or Standard mode in the AMP plugin.
- Validate a URL.
- Notice unexpected validation errors for
link[rel=canonical]or other tags.
Do not alter or remove anything below. The following sections will be managed by moderators only.
Acceptance criteria
- Validating a page generated by Genesis should not result in validation errors when there are none on the frontend.
- HTML comments that occur before and around the
<!DOCTYPE>,<html>, and<head>tags must be preserved so that validation errors can be properly sourced.
Implementation brief
- Perhaps capture the whitespace before/after the
<!DOCTYPE>,<html>, and<head>and ensure that they are retained after normalization.