如何根据mongodb聚合中上一个文档的值和下一个文档的值对值求和？

Question

我正在努力使用 mongodb 中的一些聚合函数。

假设我在两个集合中有一些这样的文档：

集合 A：

{index: "1", time: 10:30:00, value: 12}
{index: "2", time: 10:30:20, value: 11}
{index: "1", time: 10:30:32, value: 15}
{index: "2", time: 10:31:05, value: 13}
{index: "1", time: 10:31:40, value: 14}
{index: "2", time: 10:31:43, value: 10}

集合 B：

{index: "1", time: 10:30:10, cost: 10}
{index: "2", time: 10:30:15, cost: 12}
{index: "1", time: 10:30:30, cost: 8}
{index: "2", time: 10:30:50, cost: 9}
{index: "1", time: 10:31:10, cost: 10}
{index: "2", time: 10:31:23, cost: 11}

我想返回这个：

{index: "1", value: 12, totalCost: 18}
{index: "1", value: 15, totalCost: 10}
{index: "1", value: 14, totalCost: 0}
{index: "2", value: 11, totalCost: 9}
{index: "2", value: 13, totalCost: 9}
{index: "2", value: 10, totalCost: 11}

结果的{index: "1", value: 12, totalCost: 18}意思是：

根据集合A的index=“1”的文档，时间在10:30:00到10:30:32之间，查询集合B的文档index=“1”，时间在10:30:00到10:30:00到10:30:32 然后将其成本加总为totalCost。

mongodb聚合可以实现吗？谢谢！

Answer 1

好吧，我们需要使用几个运算符。

顺便说一句：请提供有效的期望，因为您基于样本数据的结果是错误的

{index: "1", value: 12, totalCost: 18} -> OK
{index: "1", value: 14, totalCost: 0}  -> OK
{index: "1", value: 15, totalCost: 10} -> OK
{index: "2", value: 10, totalCost: 11} -> KO (No B document greater than 10:31:43)
{index: "2", value: 11, totalCost: 9}  -> OK
{index: "2", value: 13, totalCost: 9}  -> KO (For 10:31:05 - 10:31:43, matchs only {index: "2", time: 10:31:23, cost: 11})

流水线说明

1. $graphLookup 用于为给定的index 值构造层次结构。
2. 我们需要从层次结构中删除：
   • 与父节点具有相同_id 的节点
   • 所有低于父级time的子级
   • 取最接近的time 与父母的time 值
3. 我们加入收藏B并采取项目：
   • index 与 A 文档相同
   • B 的 time 在 A 的父 time 和 A 的子 time 之间
4. 转换成想要的结果，求和totalCost

---

db.A.aggregate([
  {
    $graphLookup: {
      from: "A",
      startWith: "$index",
      connectFromField: "index",
      connectToField: "index",
      as: "range"
    }
  },
  {
    $addFields: {
      range: {
        $reduce: {
          input: "$range",
          initialValue: {
            index: "$index",
            time: "23:59:59",
            value: 0
          },
          in: {
            $cond: [
              {
                $and: [
                  {
                    $ne: [ "$_id", "$$this._id" ]
                  },
                  {
                    $gt: [ "$$this.time", "$time" ]
                  },
                  {
                    $lt: [ "$$this.time", "$$value.time" ]
                  }
                ]
              },
              "$$this",
              "$$value"
            ]
          }
        }
      }
    }
  },
  {
    $lookup: {
      from: "B",
      let: {
        index: "$index",
        start: "$time",
        end: "$range.time"
      },
      pipeline: [
        {
          $match: {
            $expr: {
              $and: [
                {
                  $eq: [ "$$index", "$index" ]
                },
                {
                  $and: [
                    {
                      $gt: [ "$time", "$$start" ]
                    },
                    {
                      $lt: [ "$time", "$$end" ]
                    }
                  ]
                }
              ]
            }
          }
        }
      ],
      as: "linked"
    }
  },
  {
    $project: {
      _id: 0,
      index: 1,
      value: 1,
      totalCost: { $sum: "$linked.cost" }
    }
  },
  {
    $sort: {
      index: 1,
      value: 1
    }
  }
])

MongoPlayground