count on aggregate in mongodb

Question

this my query for aggregate in pymongo:

db.connection_log.aggregate([
    { '$match': {
        'login_time': {'$gte': datetime.datetime(2014, 5, 30, 6, 57)}
    }},
    { '$group': {
        '_id': {
            'username': '$username',
            'ras_id': '$ras_id',
            'user_id': '$user_id'
        }, 
        'total': { '$sum': '$type_details.in_bytes'}, 
        'total1': {'$sum': '$type_details.out_bytes'}
    }}, 
    { '$sort': {'total': 1, 'total1': 1}}
])

How to count all result in aggregate?

Answer 1

Add to the end of your aggregation pipeline:

$group: {
    _id:null,
    count:{
        $sum:1
    }
}

SQL to Aggregation Mapping Chart

Answer 2

Well if you really want your results with a total count combined then you can always just push the results into their own array:

result = db.connection_log.aggregate([
    { '$match': {
        'login_time': {'$gte': datetime.datetime(2014, 5, 30, 6, 57)}
    }},
    { '$group': {
        '_id': {
            'username': '$username',
            'ras_id': '$ras_id',
            'user_id': '$user_id'
        }, 
        'total': { '$sum': '$type_details.in_bytes'}, 
        'total1': {'$sum': '$type_details.out_bytes'}
    }}, 
    { '$sort': {'total': 1, 'total1': 1}},
    { '$group' {
        '_id': null,
        'results': { 
            '$push': {
               '_id': '$_id',
               'total': '$total',
               'total1': '$total1'
            }
        },
        'count': { '$sum': 1 }
    }}
])

And if you are using MongoDB 2.6 or greater you can just '$push': '$$ROOT' instead of actually specifying all of the document fields there.

But really, unless you are using MongoDB 2.6 and are explicitly asking for a cursor as a result, then that result is actually returned as an array already without adding an inner array for results with a count. So just get the length of the array, which in python is:

len(result)

If you are indeed using a cursor for a large result-set or otherwise using $limit and $skip to "page" results then you will need to do two queries with one just summarizing the "total count", but otherwise you just don't need to do this.