Allow `count_many_users_posts()` to be short-circuited

[ethitter]

For a few reasons, it should be possible to prevent count_many_users_posts() from performing its query.

The first scenario comes into play when a plugin, such as Co-Authors Plus, removes the default count column. The WP_Users_List_Table class calls the function regardless of the columns presence, meaning the counts are never used.

Another scenario arises when a user has many, many posts. In one example, a user that is alphabetically first has 65,000 posts, which even on a performance-oriented host can take several seconds to count. Since the counts aren't cached, the delay can become burdensome when navigating through the users list.

[audrasjb]

Hello and thanks for the ticket/patch,

First, the new filter has to be documented using a ​regular docblock comment.

But I'm also wondering if we should rather use a more generic function to determine whether the user has a lot of posts, quite similarly to wp_is_large_network() does.

[ethitter]

I'll add the documentation on Monday; I'm kicking myself for omitting it.

I can see adding a default callback to the filter, but given the first scenario I described, I still believe a filter is useful for situations where the counts are never used.

[ethitter]

Add documentation

[jigar bhanushali]

Hello @ethitter

I've uploaded the updated patch with additional improvements to the function.

I've updated the following line to better sanitize user IDs:

From:

$userlist = implode( ',', array_map( 'absint', $users ) );

To:

$userlist = implode( ',', wp_parse_id_list( $users ) );

I also added:

$count = array_fill_keys( $userlist, 0 );

This change eliminates the need for an extra foreach loop.

cc @audrasjb

https://core.trac.wordpress.org/ticket/63004

[audrasjb]

Thank you both for your patches. I added a PR based on them to add a missing docblock info and to publicly run GitHub Actions against this changeset.

[audrasjb]

Please note that there is one PHP Unit Test issue to fix:

1) Tests_User::test_count_many_users_posts
array_fill_keys() expects parameter 1 to be array, string given

/var/www/src/wp-includes/user.php:683
/var/www/tests/phpunit/tests/user.php:589
phpvfscomposer:///var/www/vendor/phpunit/phpunit/phpunit:97
/var/www/vendor/bin/phpunit:118

This looks like a relevant regression.

[jigar bhanushali]

@audrasjb Here is my updated patch and it seems not to create any regression issues.

[jigar bhanushali]

I have uploaded my latest patch as suggested by @peterwilsoncc

[jorbin]

There isn't much time before beta 1, but I think this is a nice small enhancement that is worth including in 6.8.

@audrasjb I took the liberty of pushing @jigar787's patch (with a minor change) to your branch rather than create another PR for same.

[peterwilsoncc]

The tests are passing in the ​associated pull request based on 63004.5.patch​.

I made a minor change and continued to use absint within the function to sanitize the user IDs to make sure that future changes to wp_parse_ids_list() don't have an inadvertent effect on escaping.

The unit test for return values and types are passing, as is manual testing.

Great Peter 👍

[peterwilsoncc]

In 59900:

Users: Add pre-flight filter to count_many_users_posts().

Introduces the filter pre_count_many_users_posts() to allow developers to bypass the function in favour of either avoiding counts or their own counting functionality.

Props audrasjb, ethitter, jigar-bhanushali, jorbin.
Fixes #63004.