MongoDB query to break ties and remove duplicates - mongodb

Question

I have documents which have a Version, URL, and DateAdded field (among others but these are the relevant ones).
I'd like to find all documents where the Version is "5.5" and the DateAdded is less than or equal to January 1, 2013. That's pretty straightforward, but I also want the following behavior:
If two or more documents have the same URL, only return the one with the most recent DateAdded (provided, again, that is is less than or equal to January 1, 2013). It would be great if all of this could be expressed in a single query (but my main concern is performance).
I've been doing this last bit of filtering in my client code (outside of MongoDB) but this ends up being inefficient, not to mention inelegant.
I've also tried using Mongo's MapReduce functionality to accomplish the same thing but this is extremely slow, as it appears to copy much of my collection to another collection.
Is there a performant solution?

Answer 1

This should do the trick.
Example data:
db.foo.insert({ "_id" : ObjectId("528bd5bded29286a62959513"), "Version" : "5.3", "URL" : "foo.bar.com/asdfwoaef", "DateAdded" : ISODate("2012-10-05T00:00:00Z") })
db.foo.insert({ "_id" : ObjectId("528bd5e8ed29286a62959514"), "Version" : "5.6", "URL" : "foo.bar.com/asdfwoaef", "DateAdded" : ISODate("2012-12-05T00:00:00Z") })
db.foo.insert({ "_id" : ObjectId("528bd621ed29286a62959515"), "Version" : "5.5", "URL" : "foo.bar.com/aafoobbb", "DateAdded" : ISODate("2012-11-04T00:00:00Z") })
db.foo.insert({ "_id" : ObjectId("528bd629ed29286a62959516"), "Version" : "5.5", "URL" : "foo.bar.com/aafoobbb", "DateAdded" : ISODate("2012-11-05T00:00:00Z") })
db.foo.insert({ "_id" : ObjectId("528bd642ed29286a62959517"), "Version" : "5.5", "URL" : "foo.bar.com/aafoobbb", "DateAdded" : ISODate("2013-01-02T00:00:00Z") })
db.foo.insert({ "_id" : ObjectId("528bd744ed29286a62959518"), "Version" : "5.5", "URL" : "foo.bar.com/ccbarcc", "DateAdded" : ISODate("2013-01-02T00:00:00Z") })
db.foo.insert({ "_id" : ObjectId("528bd780ed29286a62959519"), "Version" : "5.5", "URL" : "foo.bar.com/ccbarcc", "DateAdded" : ISODate("2012-04-05T00:00:00Z") })
Pipeline:
pipeline = [
{
"$match" : {
"Version" : "5.5",
"DateAdded" : {
"$lt" : ISODate("2013-01-01T00:00:00Z")
}
}
},
{
"$sort" : {
"URL" : 1,
"DateAdded" : -1
}
},
{
"$group" : {
"_id" : "$URL",
"doc" : {
"$first" : {
"id" : "$_id",
"DateAdded" : "$DateAdded"
}
}
}
}
]
db.foo.aggregate(pipeline)
And here is the result:
{
"result" : [
{
"_id" : "foo.bar.com/ccbarcc",
"doc" : {
"id" : ObjectId("528bd780ed29286a62959519"),
"DateAdded" : ISODate("2012-04-05T00:00:00Z")
}
},
{
"_id" : "foo.bar.com/aafoobbb",
"doc" : {
"id" : ObjectId("528bd629ed29286a62959516"),
"DateAdded" : ISODate("2012-11-05T00:00:00Z")
}
}
],
"ok" : 1
}