Question

I am trying to use the aggregation function to display info in a chart. For this example, a document in the collection looks like this (excluding unnecessary fields for this query):

{
 'locid' : <someid>, #Reference to a city & state collection
 'collat' : <dateobj>, #a date object when this entry was saved 
 'pid' : <someid>, #Reference to a person collection 
 'pos' : <int> #Value I am interested in matching with location & date
 }

So I basically start with a pid. I use this as my first $match parameter to limit the amount of data that gets thrown into the pipeline.

array(
    '$match' => array(
         'pid' => new \MongoId($pid)
     )
),

So now that I have selected the correct pid, I tell it I only want/need certain fields:

array(
    '$project' => array(
        'pos' => 1,
        'collat' => 1,
        'locid' => 1
    )
),    

The second match is to say I only care about these locations right now ($ids contains an array of locid):

    array(
        '$match' => array(
            'locid' => array('$in' => $ids)
        )
    ),

And finally, I am saying group all the returned documents by collat and locid

array(
            '$group' => array(
                '_id' => array(
                    'locid' => '$locid',
                    'collat' => '$collat'
                )
            )
        )

While the query completes OK and returns data, I am not getting the pos field back, it is only returning the locid and collat.

Questions

1. Isn’t that what $project is for? I use it to tell the driver what fields I want returned?
2. Once I get the pos field returning as well, how can I tell the driver I only want the lowest value for each locid & collat combo pair? So say there are two entries for that date, location, and person: 4 & 8. I only care about pos=4

My end goal is to create a line chart with the X-Axis as the dates (from collat) and the Y-Axis will be the pos field, and each line will plot individual locid data.

Here is the entire parameters being sent to the aggregation driver.

$ops = array(
        array(
            '$match' => array(
                'pid' => new \MongoId($pid)
            )
        ),
        array(
            '$project' => array(
                'pos' => 1,
                'collat' => 1,
                'locid' => 1
            )
        ),
        array(
            '$match' => array(
                'locid' => array('$in' => $ids)
            )
        ),
        array(
            '$group' => array(
                '_id' => array(
                    'locid' => '$locid',
                    'collat' => '$collat'
                )
            )
        )
    );
    $out = $myCollection->aggregate($ops);

Update This is the way I got it to group & return pos without throwing an error. I need to spot check it though to make sure it’s actually returning the correct values though.

array(
            '$group' => array(
                '_id' => array(
                    'locid' => '$locid',
                    'collat' => '$collat'
                ),
                array('$min' => '$pos')
            )
        )

Answer 1

Aggregation query is like an SQL statement group by. You are telling {$group} what field(s) you want to ‘GROUP BY’ but you are not telling it how you want to aggregate the grouped information.

The {$group} you want is probably something like:

{$group : { _id : { locid: "$locid", collat: "$collat"},
            pos : {$min : "$pos"}
           }
 }