MongoDB Aggregate by fields exists - mongodb

Question

I need to perform a sum with the following collection's schema:
{
"_id" : "20160530/108107/31",
"metadata" : {
"date" : "2016-05-30",
"offer" : "108107",
"adv" : 31,
"update" : ISODate("2016-05-30T15:27:20.240Z")
},
"daily_unique" : 4,
"daily_gross" : 4,
"hourly" : {
"17" : {
"unique" : 4,
"gross" : 4
}
},
"publisher" : {
"738" : {
"daily_unique" : 3,
"daily_gross" : 3,
"hourly" : {
"17" : {
"unique" : 3,
"gross" : 3
}
}
},
"43" : {
"daily_unique" : 1,
"daily_gross" : 1,
"hourly" : {
"17" : {
"unique" : 1,
"gross" : 1
}
}
}
}
},
{
"_id" : "20160530/78220/59",
"metadata" : {
"date" : "2016-05-30",
"offer" : "78220",
"adv" : 59,
"update" : ISODate("2016-05-30T15:24:49.900Z")
},
"daily_unique" : 2,
"daily_gross" : 2,
"hourly" : {
"17" : {
"unique" : 2,
"gross" : 2)
}
},
"publisher" : {
"43" : {
"daily_unique" : 2,
"daily_gross" : 2,
"hourly" : {
"17" : {
"unique" : 2,
"gross" : 2
}
}
}
}
}
First document have data from publisher 738 and 43, but second have data only from 43.
So, when I want to sum all data from publisher 738, I need to sum all daily_gross, or daily_unique only if its present in the publisher, as in the first document.
I am trying some different approaches, with $exists and $cond, but not getting results
aggregate(
['$match' => ['metadata.date' => date('Y-m-d')]],
['$group' => [
'_id' => '$metadata.offer',
'daily_u' => ['$sum' => '$daily_unique']
],
])
which gives me
[
0 => [
'_id' => '108107'
'daily_u' => 4
]
1 => [
'_id' => '78220'
'daily_u' => 2
]
]
When I try to dive deep in publisher I cannot get the results I want:
aggregate(
['$match' => ['metadata.date' => date('Y-m-d')]],
['$group' => [
'_id' => '$metadata.offer',
'daily_u' => [
'$sum' => [
'$cond' => [
'if' => [
'publisher.738' => ['$exists' => true],
'then' => 1,
'else' => 0
]
]
]
],
]]
)
But cannot get daily by publisher.
It even gets complicated when I try to get hourly data.
Can anybody point me in the right direction?
Thanks in advance.