Fix for issue #625 #628
Fix for issue #625 #628
Conversation
A previous change broke the regEx which then broke the SQL which then broke the front end.
hyperionjrw
changed the title
| function posts_where_filter( $where, $query ) { | ||
| global $wpdb; | ||
|
|
||
| if ( $query->is_author() ) { | ||
| $post_type = $query->query_vars['post_type']; | ||
| if ( 'any' === $post_type ) { |
There was a problem hiding this comment.
How come we are removing this line here? This will cause a syntax error and throw off the support for 'any' post types.
There was a problem hiding this comment.
I didn't think I was. 🤔
I think I may have messed up when selecting which bits to commit as I had a heap of white space changes show up due to my automatic tendency to run a code formatting shortcut after almost all edits and also doing this between other things.
|
|
||
| //If we have an ID but it's not a "real" ID that means that this isn't the first time the filter has fired and the object_id has already been replaced by a previous run of this filter. We therefore need to replace the 0 | ||
| // This happens when wp_query::get_posts() is run multiple times. | ||
| if ( false === get_user_by( 'id', $id ) ){ | ||
| $id = 0; | ||
| if ( $id !== '\d+' && false === get_user_by( 'id', $id ) ){ |
There was a problem hiding this comment.
The check for $id !== '\d+' is probably not necessary. I think we can just have $id assigned to '\d+' if get_user_by( 'id', $id ) returns false since $id will not equal '\d+' if $id ends up equating to get_queried_object_id() (which will not be '\d+').
There was a problem hiding this comment.
The check for the string was to avoid doing an unnecessary user query that's obviously going to fail. String check is far quicker than a db query.
| @@ -713,12 +717,12 @@ function posts_where_filter( $where, $query ) { | |||
| } | |||
| $terms_implode = rtrim( $terms_implode, ' OR' ); | |||
|
|
|||
| $id = is_author() ? get_queried_object_id() : '\d+'; | |||
| $id = is_author() && $query->is_main_query() ? get_queried_object_id() : '\d+'; | |||
There was a problem hiding this comment.
This fix works without the check for $query->is_main_query() so I think we don't need to add it for now.
There was a problem hiding this comment.
If you query a second user from author.php to show something like, people in same dept or other authors you may like, you won't get the correct posts, you'll only get those where the user is in the post_author column.
e.g. If I add $x = new WP_Query( array ( 'post_status' => 'publish', 'author' => 9099999999 ) ); to author.php after the main loop. Go to /?author=1 the ID that's passed to the regex is 1 not 9099999999 so our custom query is missing any posts where the queried user isn't in the post_author column.
There was a problem hiding this comment.
Thinking this through you also need this check to avoid further risk of WordPress throwing db errors. If your custom queried user ID contains the author page user ID somewhere before the 3rd char from the end then we'd end up with a split user ID and invalid SQL.
e.g Go to /?author=1 and sub-query for author 10123456. That'd produce regex that's looking through the the SQL for \s*(.*1.)/ which would match 1012 from our ID and put the new SQL in the middle of the ID and thus a db error would be thrown.
Added some comments and fixed a wonky commit.
Changing the order to match previous commit.
A previous change #624 broke the regEx which then broke the SQL see #625 for full details.