mongodb 聚合 $group 然后 $push 一个对象

Question

这是我的数据：

> db.bookmarks.find({"userId" : "56b9b74bf976ab70ff6b9999"}).pretty()
{
    "_id" : ObjectId("56c2210fee4a33579f4202dd"),
    "userId" : "56b9b74bf976ab70ff6b9999",
    "items" : [
        {
            "itemId" : "28",
            "timestamp" : "2016-02-12T18:07:28Z"
        },
        {
            "itemId" : "29",
            "timestamp" : "2016-02-12T18:07:29Z"
        },
        {
            "itemId" : "30",
            "timestamp" : "2016-02-12T18:07:30Z"
        },
        {
            "itemId" : "31",
            "timestamp" : "2016-02-12T18:07:31Z"
        },
        {
            "itemId" : "32",
            "timestamp" : "2016-02-12T18:07:32Z"
        },
        {
            "itemId" : "33",
            "timestamp" : "2016-02-12T18:07:33Z"
        },
        {
            "itemId" : "34",
            "timestamp" : "2016-02-12T18:07:34Z"
        }
    ]
}

我想要类似的东西（其实我希望 _id 也可以变成 userId）：

{ 
    "_id" : "56b9b74bf976ab70ff6b9999", 
    "items" : [ 
        { "itemId": "32", "timestamp": "2016-02-12T18:07:32Z" },
        { "itemId": "31", "timestamp": "2016-02-12T18:07:31Z" },
        { "itemId": "30", "timestamp": "2016-02-12T18:07:30Z" }
    ] 
}

我现在拥有的：

> db.bookmarks.aggregate(
...     { $match: { "userId" : "56b9b74bf976ab70ff6b9999" } },
...     { $unwind: '$items' },
...     { $sort: { 'items.timestamp': -1} },
...     { $skip: 2 },
...     { $limit: 3},
...     { $group: { '_id': '$userId' , items: { $push: '$items.itemId' } } }
... ).pretty()
{ "_id" : "56b9b74bf976ab70ff6b9999", "items" : [ "32", "31", "30" ] }

我尝试在 mongo 中阅读文档并发现我可以 $push，但不知何故我找不到推送此类对象的方法，该对象未在整个对象的任何地方定义。我也想有时间戳..但我不知道我应该如何修改 $group （或其他人？？）这样做。感谢您的帮助！

Answer 1

我在 MongoDB 3.2.1 shell 中测试过的这段代码应该会为您提供所需的输出格式：

> db.bookmarks.aggregate( 
{ "$match" : { "userId" : "Ursula" } }, 
{ "$unwind" : "$items" }, 
{ "$sort" : { "items.timestamp" : -1 } }, 
{ "$skip" : 2 }, 
{ "$limit" : 3 }, 
{ "$group" : { "_id" : "$userId", items: { "$push" :  { "myPlace" : "$items.itemId", "myStamp" : "$items.timestamp" } } } }  ).pretty()

运行上面会产生这个输出：

{
    "_id" : "Ursula",
    "items" : [
        {
            "myPlace" : "52",
            "myStamp" : ISODate("2016-02-13T18:07:32Z")
        },
        {
            "myPlace" : "51",
            "myStamp" : ISODate("2016-02-13T18:07:31Z")
        },
        {
            "myPlace" : "50",
            "myStamp" : ISODate("2016-02-13T18:07:30Z")
        }
    ]
}

在 MongoDB 版本 3.2.x 中，您还可以在聚合管道的最后阶段使用 $out 运算符，并将聚合查询的输出写入集合。这是我使用的代码：

> db.bookmarks.aggregate( 
{ "$match" : { "userId" : "Ursula" } }, 
{ "$unwind" : "$items" }, 
{ "$sort" : { "items.timestamp" : -1 } }, 
{ "$skip" : 2 }, 
{ "$limit" : 3 }, 
{ "$group" : { "_id" : "$userId", items: { "$push" :  { "myPlace" : "$items.itemId", "myStamp" : "$items.timestamp" } } } }, 
{ "$out" : "ursula" } ) 

这给了我一个名为“ursula”的集合：

> show collections
ursula

我可以查询该集合：

> db.ursula.find().pretty()
{
    "_id" : "Ursula",
    "items" : [
        {
            "myPlace" : "52",
            "myStamp" : ISODate("2016-02-13T18:07:32Z")
        },
        {
            "myPlace" : "51",
            "myStamp" : ISODate("2016-02-13T18:07:31Z")
        },
        {
            "myPlace" : "50",
            "myStamp" : ISODate("2016-02-13T18:07:30Z")
        }
    ]
}
> 

最后，这是我在聚合查询中使用的输入文档。您可以将此文档与我编写聚合查询的方式进行比较，以了解我是如何构建新项目数组的。

> db.bookmarks.find( { "userId" : "Ursula" } ).pretty()
{
    "_id" : ObjectId("56c240ed55f2f6004dc3b25c"),
    "userId" : "Ursula",
    "items" : [
        {
            "itemId" : "48",
            "timestamp" : ISODate("2016-02-13T18:07:28Z")
        },
        {
            "itemId" : "49",
            "timestamp" : ISODate("2016-02-13T18:07:29Z")
        },
        {
            "itemId" : "50",
            "timestamp" : ISODate("2016-02-13T18:07:30Z")
        },
        {
            "itemId" : "51",
            "timestamp" : ISODate("2016-02-13T18:07:31Z")
        },
        {
            "itemId" : "52",
            "timestamp" : ISODate("2016-02-13T18:07:32Z")
        },
        {
            "itemId" : "53",
            "timestamp" : ISODate("2016-02-13T18:07:33Z")
        },
        {
            "itemId" : "54",
            "timestamp" : ISODate("2016-02-13T18:07:34Z")
        }
    ]
}