how to unwind more than one array in MongoDB Aggregation - mongodb

Question

This is what my documents look like
{
"_id" : ObjectId("584149cafda90a8b18cdfcc1"),
"uid" : "583eaa7df4def0ec5a520d19",
"surid" : "58414631ec5ed099538929b8",
"createdat" : ISODate("2016-12-02T10:15:38.382Z"),
"response" : [
{
"qid" : "649975800",
"que" : "Which is your favourite color ?",
"ans" : [
"red",
"yellow"
]
},
{
"qid" : "309541969",
"que" : "which is your favourite fruits ? ",
"ans" : [
"apple",
"orange"
]
}
]
}
/* 2 */
{
"_id" : ObjectId("58414a28fda90a8b18cdfcc7"),
"uid" : "57ff2141b893ba1a2e89ef57",
"surid" : "58414631ec5ed099538929b8",
"createdat" : ISODate("2016-12-02T10:17:12.800Z"),
"response" : [
{
"qid" : "649975800",
"que" : "Which is your favourite color ?",
"ans" : "red"
},
{
"qid" : "309541969",
"que" : "which is your favourite fruits ? ",
"ans" : "banana"
}
]
}
/* 3 */
{
"_id" : ObjectId("58414a52fda90a8b18cdfcd1"),
"uid" : "57b300678c9f14d7555b668e",
"surid" : "58414631ec5ed099538929b8",
"createdat" : ISODate("2016-12-02T10:17:54.869Z"),
"response" : [
{
"qid" : "649975800",
"que" : "Which is your favourite color ?",
"ans" : "red"
},
{
"qid" : "309541969",
"que" : "which is your favourite fruits ? ",
"ans" : "banana"
}
]
}
This is what I need:
{
"que" : "Which is your favourite color ?",
"ans" :{red:3, yellow:1}
},
{
"que" : "which is your favourite fruits ? ",
"ans":{apple:1, orange:1, banana:3}
}
I want to this result with mongodb aggregation using unique surid and with separate answer.
it's all about to the feedback result of the user data.

Answer 1

Because you won't know the values for the embedded ans array in advance, the proposed desired output won't be feasible since it assumes you know the values. A much better and faster approach would be to get the output as an embedded counts document like:
{
"ques": "Which is your favourite color ?",
"counts": [
{ "value": "red", "count": 3 },
{ "value": "yellow", "count": 1 }
]
},
{
"ques": "which is your favourite fruits ?",
"counts": [
{ "value": "apple", "count": 1 },
{ "value": "orange", "count": 1 },
{ "value": "banana", "count": 3 }
]
}
which can be achieved by running this aggregate operation:
db.collection.aggregate([
{ "$unwind": "$response" },
{ "$unwind": "$response.ans" },
{
"$group": {
"_id": {
"surid": "$surid",
"ans": "$response.ans"
},
"ques": { "$first": "$reponse.que" },
"count": { "$sum": 1 }
}
},
{
"$group": {
"_id": "$_id.surid",
"ques": { "$first": "$ques" },
"counts": {
"$push": {
"value": "$_id.ans",
"count": "$count"
}
}
}
}
])
However, if the values are static and known in advance, then take advantage of the $cond operator in the $group stage to evaluate the counts based on the "response.ans" field, something like the following:
db.collection.aggregate([
{ "$unwind": "$response" },
{ "$unwind": "$response.ans" },
{
"$group": {
"_id": "$surid",
"ques": { "$first": "$reponse.que" },
"red": {
"$sum": {
"$cond": [ { "$eq": [ "$response.ans", "red" ] }, 1, 0 ]
}
},
"yellow": {
"$sum": {
"$cond": [ { "$eq": [ "$response.ans", "yellow" ] }, 1, 0 ]
}
},
"apple": {
"$sum": {
"$cond": [ { "$eq": [ "$response.ans", "apple" ] }, 1, 0 ]
}
},
"orange": {
"$sum": {
"$cond": [ { "$eq": [ "$response.ans", "orange" ] }, 1, 0 ]
}
},
"banana": {
"$sum": {
"$cond": [ { "$eq": [ "$response.ans", "banana" ] }, 1, 0 ]
}
}
}
}
])