[dev] need help to understand content tagger getTags by tag-id.

[Michael J Rubinsky]

Quoting roman stachura <roman at stachura.ch>:

> Hi horde devs.
>
>
> I was working through the Content_Tagger class.
>
> The class is nicely structured. Well done.
>
> http://git.horde.org/co.php/content/lib/Tagger.php?rt=horde-git&ws=1&r=2644a8bcaadeff77e52a5dfcc80c827b50167e79
>
> I have a question about the function
> public function getTags($args){...}
>
> line 280 :
>
>
> } elseif (isset($args['tagId'])) {
>
> $radius = isset($args['limit']) ? (int)$args['limit'] :  
> $this->_defaultRadius;
>
> unset($args['limit'];
>
> $inner = $this->_db->addLimitOffset('
>    SELECT object_id
>    FROM ' . $this->_t('tagged') . '
>    WHERE tag_id = ' . (int)$args['tagId'],
>    array('limit' => $radius));
>
> $sql = $this->_db->addLimitOffset('
>    SELECT DISTINCT tagged2.tag_id AS tag_id, tag_name
>    FROM (' . $inner . ') tagged1
>    INNER JOIN ' . $this->_t('tagged') . ' tagged2 ON  
> tagged1.object_id = tagged2.object_id
>    INNER JOIN ' . $this->_t('tags') . ' t ON tagged2.tag_id = t.tag_id',
>    array('limit' => $args['limit']));
> } else {...
>
> What is going on here "...FROM ($inner)...?
> can someone post it in plain sql?

I don't have the code in front of me at the moment, so going from  
memory here. First, to answer your question:

"FROM ($inner) tagged1" is a sql subquery with a table alias. It  
allows the subquery to act as a table with a 'object_id' column. So,  
e.g., later in the query we join on "tagged1.object_id =  
tagged2.object_id" the tagged1 table is actually the subquery results  
presented as a table.

If I'm reading the code correctly, the idea of this code is to return  
tags that are present on objects where those objects are also tagged  
with $args['tagId']. The radius determines the max number of other  
objects to look at. For example: I have an object tagged with 'foo'  
and 'bar', and another object tagged with 'foo'. If I run this query  
with the tag_id for 'foo', it will return 'bar'. At least that's the  
idea ;)

That being said, it looks like something might be wrong here, since  
the limit value is unset() before it's used in the final query, might  
be causing some WARNING to be rasied due to non-existant key.

Are you having issues with this method? I still need to fill in some  
areas of the unit tests, so any information on failing behavior would  
be helpful.


-- 
mike

The Horde Project (www.horde.org)
mrubinsk at horde.org