Compare Results of MySQL Query against array

Question

I'm using the following query to join an intermediate table with two others, to generate a list of threads which have tags associated with them.

SELECT th.thread_id, owner_id, message, TIME
    FROM threads th
        INNER JOIN tag_thread_xref ttx 
            ON th.thread_id = ttx.thread_id
        INNER JOIN tags t 
            ON ttx.tag_id = t.tag_id                                    
    ORDER BY TIME DESC                              
    LIMIT ?         

I can then use a WHERE clause to find threads which are associated with a particular tag I'm searching for. My problem though figuring out how to get a list of all rows in table 'threads' which are associated with every element in an array of tag_names. For example, if the user does a search for tags 'television' and 'how-i-met-your-mother', I want to return only threads with contain BOTH of those tags.

I'm inexperienced with MySQL, but it seems like I want to create a fourth temporary table which contains only one column; my list of tag_name's. I then want to add that table as a third INNER JOIN to my query to constrain the list. I've done that here:

CREATE TEMPORARY TABLE tmp_tag_list (tag_name VARCHAR(30));
INSERT INTO tmp_tag_list (tag_name) VALUES ('television');
INSERT INTO tmp_tag_list (tag_name) VALUES ('how-i-met-your-mother');

SELECT th.thread_id,owner_id,message,time
    FROM threads th                                             
        INNER JOIN tag_thread_xref ttx                                                  
            ON th.thread_id = ttx.thread_id                                     
        INNER JOIN tags t                                                   
            ON ttx.tag_id = t.tag_id
        INNER JOIN tmp_tag_list tmp
            ON t.tag_name = tmp.tag_name                                                                                    
    ORDER BY TIME DESC                                              
    LIMIT 10

I expect the left table to be a collection of all rows from 'threads' which have a 'tag_id' associated with them; all tagged threads, and the right table to be a list of the tags I want to filter by. So I would expect the resulting inner join table to be only threads which have the tags listed in the right table.

I have two problems:

1. This isn't the result I'm getting. Instead, this query returns all threads which have ANY of the tags listed, and some rows appear twice.

2. Even if I do get this to work as expected, I don't see how I can use PHP's prepared statements to dynamically create as many INSERT clauses as are required for my array; I would have to revert back to PHP's old, less secure method of MySQL queries, unless there is another way.

Can anyone offer input on either of these issues?

Answer 1

There is probably a better way to do this, but off the top of my head, you could select where it is either tag, group by thread_id and tag, and then select from that, group by thread_id, where the count of tags (which you grouped by, so will only have one result for each tag) is at least two. Something like:

SELECT th.thread_id, owner_id, message, time FROM 
   (SELECT th.thread_id,owner_id,message,time,tag_name
    FROM threads th                                             
    INNER JOIN tag_thread_xref ttx                                                  
    ON th.thread_id = ttx.thread_id                                     
    INNER JOIN tags t                                                   
    ON ttx.tag_id = t.tag_id
    WHERE tag_name IN (<?php echo implode(",",$sanitizedTagsArray); ?>)
    GROUP BY th.thread_id, tag_name) AS t
GROUP BY th,thread_id
HAVING count(tag_name) >= <?php echo count($sanitizedTagsArray); ?>

If you want to use PDO:

$questionmarks = str_repeat("?,", count($tags)-1) . "?";    
...
WHERE tag_name IN (<?php echo $questionmarks ?>)
...
HAVING count(tag_name) >= <?php echo count($tags); ?>
...
$stmt = $db->prepare($query);
$stmt->execute($tags);

And then you're totally safe