Question

I’m trying to do something similar to

select column, count(*) 
  from table
 group by column

in MongoDB.

I tried

db.table.group( {
    key: 'column',
    initial: {sum:0},
    reduce: function(doc, prev) { prev.sum += 1; } }
})

I wanted to use something like $sum, but it was not working.

For

db.table.group( {
    key: 'column'
} )

I’m getting

uncaught exception: group command failed: { "errmsg" : "$reduce has to be set", "ok" : 0 }

my mongoDB version is 2.0.8 (so I cannot use aggregation framework).

According to result I’m not getting the result for grouped column values, but result for

select count(*) from table

What am I doing wrong?

edit: (data added)

> db.table.find()
{ "_id" : ObjectId("50e180ce9449299428db83e8"), "column" : "a", "cnt" : 1 }
{ "_id" : ObjectId("50e180d09449299428db83e9"), "column" : "b", "cnt" : 2 }
{ "_id" : ObjectId("50e180d19449299428db83ea"), "column" : "c", "cnt" : 3 }
> db.table.group( { key: 'column', initial: { sum: 0}, reduce: function(doc, prev) { prev.sum += 1; } } )
[ { "sum" : 3 } ]

Maybe I’m using it in wrong way, but I expect

[ { "a": 1, "b": 1, "c": 1 } ]

Answer 1

The key should be an object, not just a field name and you have an extra trailing } on the reduce line.

Try this instead:

db.table.group({
    key: {column: 1},
    initial: {sum:0},
    reduce: function(doc, prev) { prev.sum += 1; }
})