MongoDB $lookup on nested document，限制和统计检索到的数据

Question

我想计算每个子房间的用户未读取的所有通知（“A”、“B”、“C”等）。 考虑到可能是房间 Collections 中的数百万个通知文档和数百个子房间元素，我需要对其进行限制。 出于这个原因，我限制了前 100 个元素的 $lookup，然后检查该通知是否已被用户读取。 我使用 $lookup 中的文档 (roomId) 完成了此操作，但我无法使用子文档 (subRoom.id) 执行此操作。

通知集合使用（roomId：1，时间戳：-1）的复合索引

Notifications Collection： （id对应通知id，roomId是Rooms集合的链接）

[{
    "_id": "XXX",
    "id": "1",
    "read": ["A", "B", "C"],
    "roomId": "c1d87a4c-231d-4cc8-8438-35cf21ed7fc5",
    "content": "XXX",
    "timestamp": { "$date": "2021-12-31T22:50:53.000Z"  }
},{
    "_id": "XXX",
    "id": "2",
    "read": ["C"],
    "roomId": "c1d87a4c-231d-4cc8-8438-35cf21ed7fc5",
    "content": "XXX",
    "timestamp": { "$date": "2021-12-31T22:50:53.000Z" }
},
...
]

客房收藏：

[{
    "_id": "XXX"
    "subRoom": [{
        "id": "c1d87a4c-231d-4cc8-8438-35cf21ed7fc5",
        "image": "XXX",
        "name": "XXX"
    }, {
        "id": "c2d5081e-0cf1-4e69-937d-be357da1d104",
        "image": "XXX",
        "name": "XXX"
    }, {
        "id": "530c2c02-26e8-441c-af39-c5232dfe1f73",
        "image": "XXX",
        "name": "XXX"
    }],
    "id": "453a6458-6545-4842-8946-05f49efea216",
    "name": "XXX",
},
...
]

使用 roomId 代替 subRoom.id 的代码：

   { $lookup: {
      from: "notifications",
      let: { "id": "$id" },
      pipeline: [
        { $match: {
          $expr: 
              { $eq: [ "$roomId", "$$id" ] }
        }},
        { $limit: 100},
        { $project: {_id: 0, read: 1}}
      ],
      as: "messages"
    }},
   { $project: {_id: 0, id: 1, notRead: {
     $size: {
       $filter: {
          input: "$notifications",
            cond: {
              $not: {
                 $in: [
                   "A",
                   "$$this.read"
                  ]
                }
             }
          }
       }
    },
 }

使用 subRoom.id 的代码不起作用：

 { $lookup: {
      from: "notifications",
      let: { "id": "$subRoom.id" },
      pipeline: [
        { $match: {
          $expr: 
              { $eq: [ "$roomId", "$$id" ] }
        }},
        { $limit: 100},
        { $project: {_id: 0, read: 1}}
      ],
      as: "messages"
    }},
    {
      $addFields: {
        items: {
          $map: {
            input: { $zip: { inputs: ["$subRoom", "$messages"] } },
            in: { $mergeObjects: "$$this" },
          },
        },
      },
    },
   .
   . projection
   .

预期结果：

 [{
        "_id": "XXX"
        "subRoom": [{
            "id": "c1d87a4c-231d-4cc8-8438-35cf21ed7fc5",
            "notRead": 50 //e.g
        }, {
            "id": "c2d5081e-0cf1-4e69-937d-be357da1d104",
            "notRead": 35 //e.g
        }, {
            "id": "530c2c02-26e8-441c-af39-c5232dfe1f73",
            "image": "XXX",
            "notRead": 5 //e.g
        }],
        "id": "453a6458-6545-4842-8946-05f49efea216",
        "name": "XXX",
    },
    ...
] 

最后也是非常重要的一点，我想要一个可以用大数据完成的可扩展解决方案。

非常感谢您提前。

Answer 1

• $unwind解构subRoom数组，保留 null 和空数组属性
• $lookup使用管道收集通知， let id传递给管道，检查roomId的条件，用户不应该阅读通知
• $group by null 并计算未读通知总数
• $addFields使用$sum获取通知计数
• $group by _id并在结果中使用必填字段重建subRoom数组

db.rooms.aggregate([
  {
    $unwind: {
      path: "$subRoom",
      preserveNullAndEmptyArrays: true
    }
  },
  {
    $lookup: {
      from: "nitifications",
      let: { id: "$subRoom.id" },
      pipeline: [
        {
          $match: {
            $and: [
              { $expr: { $eq: ["$$id", "$roomId"] } },
              { read: { $ne: "A" } }
            ]
          }
        },
        {
          $group: {
            _id: null,
            count: { $sum: 1 }
          }
        }
      ],
      as: "subRoom.notRead"
    }
  },
  {
    $addFields: {
      "subRoom.notRead": { $sum: "$subRoom.notRead.count" }
    }
  },
  {
    $group: {
      _id: "$_id",
      name: { $first: "$name" },
      id: { $first: "$id" },
      subRoom: { $push: "$subRoom" }
    }
  }
])

操场

---

不使用$unwind阶段的第二个选项，

• $lookup使用管道收集通知， let id传递给管道，检查roomId的条件，用户不应该阅读通知
• $group by null 并计算未读通知总数
• $map迭代subRoom数组的循环
• $filter从查找通知计数迭代返回结果循环并获取当前subRoom文档
• $let声明一个变量n并将上面的过滤结果分配给它并从count返回$sum
• $mergeObjects合并 subRoom 的当前subRoom和新字段notRead

db.rooms.aggregate([
  {
    $lookup: {
      from: "nitifications",
      let: { id: "$subRoom.id" },
      pipeline: [
        {
          $match: {
            $and: [
              { $expr: { $in: ["$roomId", "$$id"] } },
              { read: { $ne: "A" } }
            ]
          }
        },
        {
          $group: {
            _id: "$roomId",
            count: { $sum: 1 }
          }
        }
      ],
      as: "notRead"
    }
  },
  {
    $project: {
      id: 1,
      name: 1,
      subRoom: {
        $map: {
          input: "$subRoom",
          as: "s",
          in: {
            $mergeObjects: [
              "$$s",
              {
                notRead: {
                  $let: {
                    vars: {
                      n: {
                        $filter: {
                          input: "$notRead",
                          cond: { $eq: ["$$this._id", "$$s.id"] }
                        }
                      }
                    },
                    in: { $sum: "$$n.count" }
                  }
                }
              }
            ]
          }
        }
      }
    }
  }
])

操场