Question

I’ve got some documents that look like:

{
  _id: 3, 
  key: 3, 
  stuff: "Some data"
}

Some documents also have a signUpDate

We can populate a collection for demo purposes like this:

for(i=1; i<=100000; i++){
if(i%3===0)
     db.numbers.insert({_id:i, key:i, stuff:"Some data", signUpDate: new Date()});
else
     db.numbers.insert({_id:i, key:i, stuff:"Some data"});
}

… so a third of the documents have a signUpDate

What I’m trying to do is create a map reduce function that takes all the documents, where signUpDate is not null, and insert them into a separate collection, ordered randomly

Is this possible?

Answer 1

Ok, here’s a solution that works:

Using mongoshell:

First, we populate our data:

for(i=1; i<=100000; i++){
if(i%3===0)
     db.numbers.insert({_id:i, key:i, stuff:"Some data", signUpDate: new Date()});
else
     db.numbers.insert({_id:i, key:i, stuff:"Some data"});
}

So now, we have a third of our data with a signUpDate.

Now, a super-simple mapreduce:

m = function() {emit(this._id, Math.random());}
r = function(key, values){}

db.numbers.mapReduce(m,r, {out: "randomlyOrdered", query: { signUpDate: { $ne : null } }});

Next, ensureIndex to speed up sorting:

db.randomlyOrdered.ensureIndex({"value":1});

Now, find the numbers (randomly sorted)

db.randomlyOrdered.find({}, {"_id":1}).sort({"value":1});