在 mongodb 中将多个组合并为一个聚合
Combine multiple groups in an aggregation in mongodb
如果我有这样的collection:
{
"store" : "XYZ",
"total" : 100
},
{
"store" : "XYZ",
"total" : 200
},
{
"store" : "ABC",
"total" : 300
},
{
"store" : "ABC",
"total" : 400
}
我可以通过聚合获得 collection 中的 $sum
个订单:
db.invoices.aggregate([{$group: { _id: null, total: { $sum: "$total"}}}])
{
"result": [{
"_id": null,
"total": 1000
}
],
"ok": 1
}
而且我可以获得 $sum
按商店分组的订单:
db.invoices.aggregate([{$group: { _id: "$store", total: { $sum: "$total"}}}])
{
"result": [{
"_id": "ABC",
"total": 700
}, {
"_id": "XYZ",
"total": 300
}
],
"ok": 1
}
但是我怎样才能在一个查询中做到这一点呢?
您可以汇总如下:
$group
由store
字段,计算出subtotal
.
$project
字段 doc
以保持 subtotal
组在接下来的
组.
$group
乘以null
,累计净额
代码:
db.invoices.aggregate([{
$group: {
"_id": "$store",
"subtotal": {
$sum: "$total"
}
}
}, {
$project: {
"doc": {
"_id": "$_id",
"total": "$subtotal"
}
}
}, {
$group: {
"_id": null,
"total": {
$sum: "$doc.total"
},
"result": {
$push: "$doc"
}
}
}, {
$project: {
"result": 1,
"_id": 0,
"total": 1
}
}
])
输出:
{
"total": 1000,
"result": [{
"_id": "ABC",
"total": 700
}, {
"_id": "XYZ",
"total": 300
}
]
}
另一种方法是使用 $facet
聚合阶段。
$facet
允许你做多个嵌套 sub-aggregations
在您的主要聚合中。
- 每个 sub-aggregation 都有自己的管道。
- 对于 sub-aggregation 的每个结果,我们定义另一个字段。
像这样,例如:
db.invoices.aggregate([
{
$facet: {
total: [
{
$group: {
_id: null,
total: { $sum: "$total"}
}
}
],
store_totals: [
{
$group: {
_id: "$store",
total: { $sum: "$total"}
}
}
]
}
},{
$unwind: "$total"
},{
$project: {
_id: 0,
total: "$total.total",
store_totals: "$store_totals"
}
}
]
@BatScream wrote, that an $unwind
stage might be costly. However
we're unwinding an array of length 1 here. So I'm curious which
approach is more efficient under which circumstances. If someone can
compare those with console.time()
, I'd be happy to include the
results.
输出
Should be the same as in the accepted answer.
如果我有这样的collection:
{
"store" : "XYZ",
"total" : 100
},
{
"store" : "XYZ",
"total" : 200
},
{
"store" : "ABC",
"total" : 300
},
{
"store" : "ABC",
"total" : 400
}
我可以通过聚合获得 collection 中的 $sum
个订单:
db.invoices.aggregate([{$group: { _id: null, total: { $sum: "$total"}}}])
{
"result": [{
"_id": null,
"total": 1000
}
],
"ok": 1
}
而且我可以获得 $sum
按商店分组的订单:
db.invoices.aggregate([{$group: { _id: "$store", total: { $sum: "$total"}}}])
{
"result": [{
"_id": "ABC",
"total": 700
}, {
"_id": "XYZ",
"total": 300
}
],
"ok": 1
}
但是我怎样才能在一个查询中做到这一点呢?
您可以汇总如下:
$group
由store
字段,计算出subtotal
.$project
字段doc
以保持subtotal
组在接下来的 组.$group
乘以null
,累计净额
代码:
db.invoices.aggregate([{
$group: {
"_id": "$store",
"subtotal": {
$sum: "$total"
}
}
}, {
$project: {
"doc": {
"_id": "$_id",
"total": "$subtotal"
}
}
}, {
$group: {
"_id": null,
"total": {
$sum: "$doc.total"
},
"result": {
$push: "$doc"
}
}
}, {
$project: {
"result": 1,
"_id": 0,
"total": 1
}
}
])
输出:
{
"total": 1000,
"result": [{
"_id": "ABC",
"total": 700
}, {
"_id": "XYZ",
"total": 300
}
]
}
另一种方法是使用 $facet
聚合阶段。
$facet
允许你做多个嵌套 sub-aggregations 在您的主要聚合中。- 每个 sub-aggregation 都有自己的管道。
- 对于 sub-aggregation 的每个结果,我们定义另一个字段。
像这样,例如:
db.invoices.aggregate([
{
$facet: {
total: [
{
$group: {
_id: null,
total: { $sum: "$total"}
}
}
],
store_totals: [
{
$group: {
_id: "$store",
total: { $sum: "$total"}
}
}
]
}
},{
$unwind: "$total"
},{
$project: {
_id: 0,
total: "$total.total",
store_totals: "$store_totals"
}
}
]
@BatScream wrote, that an
$unwind
stage might be costly. However we're unwinding an array of length 1 here. So I'm curious which approach is more efficient under which circumstances. If someone can compare those withconsole.time()
, I'd be happy to include the results.
输出
Should be the same as in the accepted answer.