Reputation: 231
Here's the structure part of my collection:
_id: ObjectId("W"),
names: [
{
number: 1,
list: ["A","B","C"]
},
{
number: 2,
list: ["B"]
},
{
number: 3,
list: ["A","C"]
}
...
],
...
I use this request:
db.publication.aggregate( [ { $match: { _id: ObjectId("54a1de90453d224e80f5fc60") } }, { $group: { _id: "$_id", SizeName: { $first: { $size: { $ifNull: [ "$names", [] ] } } }, names: { $first: "$names" } } } ] );
but I would now use $size in every documents of my names array.
Is it possible to get this result (where "sizeList" is the result of $size) :
_id: ObjectId("W"),
SizeName: 3,
names: [
{
SizeList: 3,
number: 1,
list: ["A","B","C"]
},
{
SizeList: 1,
number: 2,
list: ["B"]
},
{
SizeList: 2,
number: 3,
list: ["A","C"]
}
...
],
...
Upvotes: 0
Views: 74
Reputation: 151092
All you really want here is a $project
stage and use $map
to alter the contents of the array members:
db.names.aggregate([
{ "$project": {
"SizeName": { "$size": "$names" },
"names": { "$map": {
"input": "$names",
"as": "el",
"in": {
"SizeList": { "$size": "$$el.list" },
"number": "$$el.number",
"list": "$$el.list"
}
}}
}}
])
You can alternately process with $unwind
and group it all back again, but that's kind of long winded when you have MongoDB 2.6 anyway.
Upvotes: 4
Reputation: 598
This isn't necessarily the most efficient way, but I think it does what you're aiming to do:
db.publication.aggregate( [
{ $unwind: '$names' },
{ $unwind: '$names.list' },
{ $group: {
_id: { _id: "$_id", number: "$names.number" },
SizeList: { $sum: 1 },
list: { $push: "$names.list" }
}
},
{ $group: {
_id: "$_id._id",
names: {
$push: {
number: "$_id.number",
list: "$list",
SizeList: "$SizeList"
}
},
SizeName: {$sum: 1}
}
}
]);
Upvotes: 0