Here are the args for my query :
$args = array(
'post_type' => 'news',
'meta_query' => array(
array(
'key' => 'topics',
'value' => 'sports',
)
)
);
This works when topics
is a string, but not when it is an array. I’d like this query to work when topics
is for example array( 'sports', 'nonprofit', etc. )
Is there a way to build meta queries with arrays as a meta_key ?
Feeding the query an array of possible values
If the value in the database is a string and you want to feed the query several values:
Searching for a specific value in a serialized array of data
If the value in the database is an array of several topics and you want to search for a single topic within that array (Note that an array in the database can be retrieved as such, but lives in the database in serialized form, which is a string also):
Using ‘LIKE’ as the compare value isn’t as clear-cut an instruction as you might have hoped for, but it is the best option to go with.
Next to that, your only other option would be retrieving all posts that have the meta_key “topics” set and iterating over them manually or,in other words, check for the value within the loop and display the posts on said condition.
EDIT 2021
Though still valid, the above answer is 9 years old.
Have a look at @sMyle’s and at @Kaji’s excellent answers as well.
To go off of Johannes’ response, since it is a serialized array, if you happen to be storing something like user id’s (which was my case), you may need to handle it a little differently.
Post meta was being saved like:
So yes they are integers but through
update_post_meta
they were being saved as strings.So you’re actually doing a LIKE comparison with the serialized string version of what you’re looking for.
EDIT 2022:
One issue to be aware of when doing this if it’s NOT a string that is saved (saved as integer), and you use
serialize( intval( $value ) )
, you run into the potential issue of matching against the index instead of the value.For example:
Is saved like this:
If you use
serialize( intval( 1 ) )
in meta query, you will match oni:1;
which is the index, NOT the value, so be careful and if you control the codebase, always save as a string so you don’t match on the index.Another slight improvement from @sMyles’ answer.
I have had cases where IDs have been stored both as strings (such as when taken from a form input) and as integers (e.g.
update_post_meta($post_id, authorized_users', array(get_current_user_id()));
). This is kind of like the well-known issue withwp_set_object_terms()
where you can use term IDs to set the terms, but if you don’t cast them as integers first you stand about a 50% chance of creating new terms with those numbers as their names instead.This can result in them being stored quite differently in a serialized array, as can be seen from the excerpts of just such a case from my test site’s database:
Both of the above, when fed through
print_r()
will render asTo fix this, I made a slight tweak to the
meta_query
by adding arelation
and another version of the query that cast the value as an integer instead of a string.Here’s the final result:
EDIT: Just realized that this method could run the risk of collisions with array indexes, which could allow someone illicit access to materials if they’re not in the array, but their user ID appears as an index. As such, while this works if you have the issue discussed, better practice is to ensure that any values you want to search for are cast as strings prior to saving them so that you can use @sMyles’ method instead.
I stuggled with something similar today.
I have to query a ACF (Advanced Custom Fields) relationship field with multiple related users (array).
After updating the field via php the query didn’t work. After updating it via the ACF UI the query worked.
Problem was, that my php code set the relationship values to be int-values, the UI set it to string-values. To make sure both work I use this query now (fitted to the example here):
I would go for Johannes’s answer. However, I want to improve that because using that meta_query, you will meet a case like this
your value is
when you search
then the result will return both ‘sport’ and ‘sport2’.
To fix that, change the meta_query args into
It’s because the value is serialize in database, and each item will be separated by a semicolons. Thus, above args will work
If items in the value are number, you just need to remove the double quote ”