Create decoupled infrastructure for gtag tag

[aaemnnosttv]

Feature Description

GTag rendering should be lifted out of Analytics into a core location where it can be used by multiple modules, initially the Ads module in addition to Analytics.

Note: This work has largely been completed as part of the POC for the Consent Mode epic, please see 82d9f94

---

Do not alter or remove anything below. The following sections will be managed by moderators only.

Acceptance criteria

• The GTag rendering should be extracted from the Analytics_4\Web_Tag class into a reusable class in the Core namespace.
• It should be modified to allow multiple tag IDs to be queued along with their initial tag config.
• Additional gtag() commands should be able to be queued for insertion either before or after the gtag.js script is requested.

Implementation Brief

• Follow the GTag class implementation in the branch trh/decouple-gtag.
• Register an instance of the GTag class in Plugin, but don't make use of it in the Analytics_4\Web_Tag class as yet, it will be integrated there in Update Analytics to use new gtag infrastructure #8273.

Test Coverage

• Provide PHPUnit test coverage for the above changes.

QA Brief (ENG:QA)

• Validate general logic of the GTag class.
• Create a new instance of the class in a theme functions.php file, an mu-plugin, or even in the Plugin class of SK
  • Add a new dummy GTag, i.e $gtag->add_tag( 'GT-12345', array( 'foo' => 'bar' ) ); and validate that the gtag script is successfully output on the page, and that the src correctly reflects the chosen tag ID.
    • Validate that the applicable gtag( 'config' ) call is present for the tag in question.
  • Add a dummy command, i.e $gtag->add_command( 'foo', array('bar') ); and validate that the gtag command call is successfully output within the gtag snippet.
    • Play around with the $position function arg and ensure that the position of the command changes accordingly.
  • Add another tag via the add_tag() method, such as GT-54321.
    • Validate that the second tag is present, but not the main src of the snippet.
    • Validate that there are 2x gtag('config') calls, one for each tag ID.
    • Validate that inline tag configs are specific to their tags and in the correct gtag('config') function calls.

Changelog entry

• Add decoupled infrastructure for GTag.

[aaemnnosttv]

Thanks @techanvil – thinking ahead just a little bit, one thing that comes to mind re #8274 is that we want to do this for GTM as well which doesn't use an enqueued WP script the way GA does. All that to say that we might want to decouple the gtag infra for the consent ("before") from the rest of the tag so that this part can be output independently. I suppose this part could be implemented by the consent mode infra almost treating it as its own tag that would always be printed very early relative to other scripts. WDYT?

If we want to address this in this issue then it should probably have a point in the AC, otherwise I'm okay with this proceeding if we want to handle it in a later issue.

[techanvil]

Thanks @aaemnnosttv, good spot. I think we can implement that in #8264, but I've updated the POC to accommodate this in a slightly cheap and cheerful manner.

Actually this makes me wonder if we even need to decouple the GTag as part of this epic - at least, as things stand we don't have it specced out to enqueue any additional GTag consent commands, only to call gtag() on the fly in response to the wp_listen_for_consent_change event. Are we missing something whereby we should be enqueuing some additional events, or could we avoid this issue and implement it in the Ads Module epic instead? 🤔

[techanvil]

I'm moving this back to AC as we need to determine whether or not it's actually needed for the Consent Mode MVP.

[bethanylang]

@aaemnnosttv I see this is still set as blocking #8273, which is a Consent Mode issue. Does this mean that we'll need to complete this Ads Module issue ASAP in order to deliver Consent Mode at the end of Sprint 122 as planned?

[techanvil]

@bethanylang thanks for spotting this. With this issue removed from the CoMo epic, #8273 can also be removed.

I have removed that issue from CoMo and assigned both of these to @10upsimon for evaluation and reuse in the Ads Module epic.

[10upsimon]

Note to AC reviewer, this has largely been kept as is from the Consent Mode epic, having reviewed the POC at 82d9f94 this should be a "lift and shift" effort.

[tofumatt]

ACs here look good 👍🏻

Since there's an IB as well (which also looks good) I'll move this right to EB

IB ✅

[kuasha420]

QA:Eng ✔️

• Went through the QA brief and verified every point.

LGTM ✅