WXR unnecessarily includes all post authors

[johnbillion]

If you're exporting just one post type (from Tools -> Export), the resulting WXR file still includes all authors from the posts table, regardless if they are the author of a post of the post type being exported. This can be particularly noticable for Pages (which are often only authored by a small number of different users on the site) and for custom post types.

This means when you subsequently import the file you'll be asked to map a complete list of all authors from the exported site, even if they have no posts in the WXR file being imported.

[johnbillion]

Patch to only include relevant post authors in the WXR export.

[jeremyfelt]

I really like this idea. It makes sense to export all of the data based on the selected parameters.

It may make more sense to reuse the $post_ids array so that the query for distinct post authors mirrors the results from the main posts query. Restricting it to post_type could cause the same issue with extra authors for some taxonomies, etc...

The updated patch attached uses the $post_ids array instead.

[jeremyfelt]

Updated patch using $post_ids

[johnbillion]

Even better!

[johnbillion]

#28450 was marked as a duplicate.

[khromov]

Glad to see this is getting some traction! Hoping for inclusion in 4.0!

[johnbillion]

In 28731:

Only include relevant post authors in WXR exports. Fixes #20206. Props jeremyfelt.

[nacin]

Not sure if this requires re-opening, but does this fully cover things like attachments (either attached to a post or as a featured image) that would also need to be (but may not yet be) included in such an export?

[johnbillion]

Good spot.

If you don't limit your export to a certain post type, attachments are included in the export and all author IDs are present and correct, so we're ok.

If you do limit your export to a certain post type, the featured image post is not included in the export and gets lost. This is covered in #17379.

[lkraav]

This is huge. Just had to say that out loud.