Mongodb和MapReduce-结果丢失

Question

with this code:

var map = function(){

    emit(1, { read: this.read, important: this.important, count: 1 })
}

var reduce = function(key, vals) {

    var i = 0, r = 0, c = 0;

    vals.forEach(function(d) {
       r += !d.read ? 0 : 1;
       i += d.important ? 0 : 1
       c += d.count;
    });

    return { read: r, important: i, count: c }
}

db.ipdr.mapReduce(map, reduce, { out: "__mr_test", verbose: true, query: { liid: "40001" } });
db.getCollection("__mr_test").find();

I get incomplete count:

{
    "_id" : 1,
    "value" : {
        "read" : 1,
        "important" : 7,
        "count" : 7607
    }
}

"count" is OK, but "read" and "important" should be a lot higher. Am I missing something?

Answer 1

Instead of

r += !d.read ? 0 : 1;
i += d.important ? 0 : 1
c += d.count;

Shouldn't you be having

r += !d.read ? 0 : d.read;
i += d.important ? 0 : 1
c += d.count;

Answer 2

This is not a good usage of mapReduce. The aggregation framework uses native code which is much faster than JavaScript execution and can do the same job. Assuming "read" and "important" are logical values that is:

db.posts.aggregate([
    { "$sort": { "read": 1, "important": 1 } },
    { "$group": {
        "_id": null,
        "read": { "$sum": { "$cond": [ "$read", 1, 0 ] } },
        "important": { "$sum": { "$cond": [ "$important", 1, 0 ] } },
        "count": { "$sum": 1 }
    }}
])

So not only faster but much more simple to code.

Given the sample documents:

{ "read" : true,  "important" : false }
{ "read" : true,  "important" : false }
{ "read" : false, "important" : false }
{ "read" : false, "important" : true  }
{ "read" : true,  "important" : true  }

The result would be:

{ "_id" : null, "read" : 3, "important" : 2, "count" : 5 }