Blueprints: Rename importFile to importWxr, switch to humanmade/WordPress importer

[adamziel]

Problem addressed

The importFile step works by interacting with HTML form like a user would. Problem is, the WXR importer requires some interactions with the forms which we fake by adjusting the POST data before submitting the form. This approach causes a number of issues listed below.

More context: #1183.

Implementation

Refactors the importFile Blueprint step to:

• Be named importWxz. The old name continues to work, though.
• Use the humanmade/WordPress-Importer for importing content – it is
  more reliable than the official wordpress-importer and has a
  convenient PHP API.
• Drop WXZ support – it's a maintenance burden and doesn't add clear
  value.
• Remove dependency on DOMParser – this step should now work in wp-now

Testing instructions

1. Go to http://localhost:5400/website-server/?import-wxr=https://raw.githubusercontent.com/helgatheviking/Simple-User-Listing/3e38ea3eb3a057424fbae4305cba2fd9f73d896b/demo-content/demo-content.xml&php=8.0&wp=6.5&storage=none&php-extension-bundle=kitchen-sink
2. Confirm the "Simple User Listing Block" page was created
3. Confirm in wp-admin a bunch of user accounts were created
4. Confirm in wp-admin the WordPress Importer plugin was installed

Closes #1183
Closes #379
Closes #1158
Closes #1161

cc @dmsnell

[dmsnell]

seems fine. I may have missed this point

Remove dependency on DOMParser – this step should now work in wp-now

Is that a TODO still? or is this already part of this PR?

[flexseth]

importWxr looks good, just looking back at everything in the steps

• Be named importWxz. The old name continues to work, though.

• Use the humanmade/WordPress-Importer for importing content – it is
  more reliable than the official wordpress-importer and has a
  convenient PHP API.

[flexseth]

Interested to see the tools

[adamziel]

Is that a TODO still? or is this already part of this PR?

@dmsnell It's a part of this PR. import-wxr.ts no longer parses the HTML output using DOMParser() as import-file.ts did:

function DOM(response: PHPResponse) {
	return new DOMParser().parseFromString(response.text, 'text/html');
}

[adamziel]

I've removed the site-data.spec.ts because of a weird, seemingly intermittent error it produced that seemed unrelated to this PR. I'll restore it in a separate PR and let's discuss the failure there.

[flexseth]

Question @dmsnell and @adamziel - I was looking at the discussion about data liberation

WXR (WordPress eXtended RSS) has been around for a while - and it's done a good job. But there's certainly room for improvement. Especially when it comes to including media/attachments and certain other things.

On data liberation

Is WXR the right format?
Is WXZ the right format?
Do we currently have the right format?
Is this a good opportunity to create the right format?

Portability

Moving from one instance to another should be what Playground does.
Duplicating sites through different means

Exit plan

Format the data so the user can take it with them, wherever they go

Can we discuss? 😎

[adamziel]

@flexseth WXR is for moving content, as in posts, pages etc. It won't move custom database tables and has some problems with media files, though. I'd rather not spend time on expanding it. Site Transfer Protocol and Playground Snapshots will enable moving sites between Playgrounds and WordPress hosts.

[dmsnell]

Is WXR the right format?
Is WXZ the right format?
Do we currently have the right format?
Is this a good opportunity to create the right format?

@flexseth to add to what @adamziel wrote, I think these are great questions, but mostly unrelated to this PR. it would be nice to join the discussion in WordPress.org Slack in the #data-liberation channel.

Personal opinion time: I think it's valuable to do a proper job with our existing commitments while also exploring new ideas. This PR I think is working to ensure that we do well the things we say we do, while leaving the door open to other kinds of imports.

[flexseth]

@dmsnell sounds good! I'll link the PR in the docs on importing and exporting with Playground for future discussions!