Group values by sub string in MongoDB - mongodb

Question

I have this documents in my collection :
{_id: "aaaaaaaa", email: "mail1#orange.fr"},
{_id: "bbbbbbbb", email: "mail2#orange.fr"},
{_id: "cccccccc", email: "mail3#orange.fr"},
{_id: "dddddddd", email: "mail4#gmail.com"},
{_id: "eeeeeeee", email: "mail5#gmail.com"},
{_id: "ffffffff", email: "mail6#yahoo.com"}
And i would like this result :
{
result: [
{domain: "orange.fr", count: 3},
{domain: "gmail.com", count: 2},
{domain: "yahoo.com", count: 1},
]
}
I'm not sure you can use the aggregator and $regex operator

Answer 1

Aggregation Framework
I don't believe that with the present document structure you can achieve the desired result by using the aggregation framework. If you stored the domain name in a separate field, it would have become trivial:
db.items.aggregate(
{
$group:
{
_id: "$emailDomain",
count: { $sum: 1 }
},
}
)
Map-Reduce
It's possible to implement what you want using a simple map-reduce aggregation. Naturally, the performance will not be good on large collections.
Query
db.emails.mapReduce(
function() {
if (this.email) {
var parts = this.email.split('#');
emit(parts[parts.length - 1], 1);
}
},
function(key, values) {
return Array.sum(values);
},
{
out: { inline: 1 }
}
)
Output
[
{
"_id" : "gmail.com",
"value" : 2
},
{
"_id" : "yahoo.com",
"value" : 1
},
{
"_id" : "orange.fr",
"value" : 3
}
]

Answer 2

Aggregation Framework
MongoDB 3.4(Released Nov 29, 2016) onwords in aggregation framework have many methods
[
{
$project: {
domain: {
$substr: ["$email", {
$indexOfBytes: ["$email", "#"]
}, {
$strLenBytes: "$email"
}]
}
},
{
$group: {
_id: '$domain',
count: {
$sum: 1
}
}
},
{
$sort: {
'count': -1
}
},
{
$group: {
_id: null,
result: {
$push: {
'domain': "$_id",
'count': '$count'
}
}
}
}
]
Results
{
_id: null,
result: [
{domain: "#orange.fr", count: 3},
{domain: "#gmail.com", count: 2},
{domain: "#yahoo.com", count: 1},
]
}