假设我的 MongoDB 架构如下所示:

{car_id: "...", owner_id: "..."}

这是一个多对多的关系。例如，数据可能如下所示:

+-----+----------+--------+
| _id | owner_id | car_id |
+-----+----------+--------+
|   1 |        1 |      1 |
|   2 |        1 |      2 |
|   3 |        1 |      3 |
|   4 |        2 |      1 |
|   5 |        2 |      2 |
|   6 |        3 |      4 |
|   7 |        3 |      5 |
|   8 |        3 |      6 |
|   9 |        3 |      7 |
|  10 |        1 |      1 | <-- not unique
+-----+----------+--------+

我想获取每个车主拥有的汽车数量。在 SQL 中，这可能看起来像:

SELECT owner_id, COUNT(*) AS cars_owned
FROM (SELECT owner_id FROM car_owners GROUP BY owner_id, car_id) AS t
GROUP BY owner_id;

在这种情况下，结果如下所示:

+----------+------------+
| owner_id | cars_owned |
+----------+------------+
|        1 |          3 |
|        2 |          2 |
|        3 |          4 |
+----------+------------+

我怎样才能通过聚合框架使用 MongoDB 来完成同样的事情？

最佳答案

为了容纳潜在的重复，您需要使用两个 $group 操作:

db.test.aggregate([
    { $group: {
        _id: { owner_id: '$owner_id', car_id: '$car_id' }
    }},
    { $group: {
        _id: '$_id.owner_id',
        cars_owned: { $sum: 1 }
    }},
    { $project: {
        _id: 0,
        owner_id: '$_id',
        cars_owned: 1
    }}]
    , function(err, result){
        console.log(result);
    }
);

给出格式如下的结果:

[ { cars_owned: 2, owner_id: 10 },
  { cars_owned: 1, owner_id: 11 } ]

https://stackoverflow.com/questions/13240039/