XPaths in Optimization Detective are invalid due to being zero-based instead of one-based

[westonruter]

This is an example XPath generated in Optimization Detective:

/*[0][self::HTML]/*[1][self::BODY]

This is actually invalid because it is zero-based when XPath is one-based. It should rather be:

/*[1][self::HTML]/*[2][self::BODY]

In the end it doesn't impact the behavior in any way since it is exclusively using the XPath as a unique key to identify an element. It isn't actually using the XPath for any querying (although it could if the HTML API supported that). But I chose XPath since it is a standard way to identify the path to an element in a document. And this should be fixed to be valid XPath, which can have some debugging value since Chrome DevTools for example supports XPath in the elements panel.

[westonruter]

I didn't catch this as a bug because I was using DevTools to test for the XPath correctness, and it was selecting the expected node. But it was selecting the expected node because it was doing a full text search for the attribute value, not actually doing an XPath query.

[westonruter]

This is the problematic code:

performance/plugins/optimization-detective/class-od-html-tag-walker.php

Lines 338 to 344 in cb9b035

	public function get_xpath(): string {
	$xpath = '';
	foreach ( $this->get_breadcrumbs() as list( $tag_name, $index ) ) {
	$xpath .= sprintf( '/*[%d][self::%s]', $index, $tag_name );
	}
	return $xpath;
	}

I believe this is the change needed:

- $xpath .= sprintf( '/*[%d][self::%s]', $index, $tag_name );
+ $xpath .= sprintf( '/*[%d][self::%s]', $index + 1, $tag_name );

The tests will then need to be updated as well.