Self joining query in MongoDB - mongodb

Question

We are exploring the possibility of migrating from relational database to MongoDB and having difficulty dealing with a query, the data schema is like this:
data_store:
id,
userId,
study,
form,
formData: {
element1: value1,
element2: value2,
.....
}
formData is a json array of dynamic element:value pairs, different form has a different list of pre-defined elements.
The requirement is by given 2 forms in a study, we need to display all their elements' data in one row for the same user, i.e. we need to join by userId and display the formData in a flat structure.
Also one user may have multiple data entry for the same form, so if a user has 3 entries for form A and 4 entries for form B, we expect that there are 12 rows for them in the result.
Sample data:
id: "id1",
user: "user1",
study: "study",
form: "f1",
formData: [
{ "e11": "value1" },
{ "e12": "value2" },
{ "e13": "value3" }
]
,
id: "id2",
user: "user1",
study: "study",
form: "f1",
formData: [
{ "e11": "value4" },
{ "e12": "value5" },
{ "e13": "value6" }
]
,
id: "id3",
user: "user1",
study: "study",
form: "f2",
formData: [
{ "e21": "value7" },
{ "e22": "value8" }
]
,
id: "id4",
user: "user1",
study: "study1",
form: "f2",
formData: [
{ "e21": "value9" },
{ "e22": "value10" }
]
,
id: "id2",
user: "user2",
study: "study1",
form: "f2",
formData: [
{ "e21": "value11" },
{ "e22": "value12" }
The expected result for the sample data above is:
Row#
study
user
f1.e11
f1.e12
f1.e13
f2.e21
f2.e22
1
study
user1
value1
value2
value3
value7
value8
2
study
user1
value1
value2
value3
value9
value10
3
study
user1
value4
value5
value6
value7
value8
4
study
user1
value4
value5
value6
value9
value10
5
study
user2
value11
value12
A similar query in relational database may be like this:
select t1.*, t2.*
from data_store t1, data_store t2
where t1.form = 'f1'
and t2.form = 'f2'
and t1.userId = t2.userId;
I am having difficulty of converting it into MongoDB query, any one who can shed some light will be greatly appreciated.

Answer 1

You can use the below Aggregation Query to get the desired result.
db.collection.aggregate([
{
$group: {
_id: {
"user": "$user",
"form": "$form",
},
"formData": {
"$push": {
"$concatArrays": [
"$formData",
[
{
"study": "$study"
}
]
]
},
},
}
},
{
$group: {
_id: {
"user": "$_id.user",
},
formData: {
"$push": "$formData"
}
}
},
{
$project: {
"formData": {
"$map": {
"input": {
"$arrayElemAt": [
"$formData",
0
]
},
"as": "f1",
"in": {
"$cond": {
"if": {
"$gt": [
{
"$size": "$formData"
},
1
]
},
"then": {
"$map": {
"input": {
"$arrayElemAt": [
"$formData",
1
]
},
"as": "f2",
"in": {
"$concatArrays": [
"$$f1",
"$$f2"
],
},
},
},
"else": [
"$$f1"
]
},
},
},
},
}
},
{
$unwind: {
path: "$formData",
}
},
{
$unwind: {
path: "$formData",
}
},
{
$replaceRoot: {
newRoot: {
"$mergeObjects": [
{
"user": "$_id.user"
},
{
"$reduce": {
"input": "$formData",
"initialValue": {},
"in": {
"$mergeObjects": [
"$$this",
"$$value"
],
}
},
},
],
}
}
}
])
Mongo Playground Working Sample
Let me know if you want an explanation of each stage.
Limitations:
This will work only if there are two forms of studies f1 and f2.
I am working on making this dynamic and will update this answer once done.
EDIT:
Make use of this updated query which will work dynamically.
db.collection.aggregate([
{
"$match": {
"form": {
"$in": [
"f1",
"f2"
]
},
},
},
{
"$group": {
"_id": "$user",
"formData": {
"$push": {
"$mergeObjects": [
{
"$reduce": {
"input": "$formData",
"initialValue": {},
"in": {
"$mergeObjects": [
"$$value",
{
"$arrayToObject": {
"$map": {
"input": {
"$objectToArray": "$$this"
},
"as": "formValue",
"in": {
"k": {
"$concat": [
"$form",
"-",
"$$formValue.k"
]
},
"v": "$$formValue.v"
}
},
},
},
]
},
}
},
{
"study": "$study",
"user": "$user",
"form": "$form"
},
],
},
},
},
},
{
"$project": {
"formData": {
"$cond": {
"if": {
"$gt": [
{
"$size": "$formData"
},
1
]
},
"then": {
$reduce: {
input: {
$range: [
0,
{
$size: "$formData"
}
]
},
initialValue: [],
in: {
$concatArrays: [
"$$value",
{
$let: {
vars: {
i: "$$this"
},
in: {
$map: {
input: {
$range: [
{
$add: [
1,
"$$i"
]
},
{
$size: "$formData"
}
]
},
in: [
{
"$let": {
"vars": {
"arrayElem1": {
$arrayElemAt: [
"$formData",
"$$i"
]
},
"arrayElem2": {
$arrayElemAt: [
"$formData",
"$$this"
]
},
},
"in": {
"$cond": {
"if": {
"$and": [
{
"$ne": [
"$$arrayElem1.form",
"$$arrayElem2.form"
]
},
// {
// "$eq": [
// "$$arrayElem1.user",
// "$$arrayElem2.user"
// ]
// },
],
},
"then": {
"$mergeObjects": [
"$$arrayElem2",
"$$arrayElem1",
],
},
"else": "$$REMOVE",
},
},
}
}
]
}
}
}
}
]
}
}
},
"else": [
"$formData"
]
},
},
},
},
{
"$project": {
"formData": {
"$reduce": {
"input": "$formData",
"initialValue": [],
"in": {
"$concatArrays": [
"$$value",
{
"$cond": {
"if": {
"$ne": [
"$$this",
[
null
]
]
},
"then": "$$this",
"else": []
}
}
]
}
}
}
}
},
{
"$unwind": "$formData"
},
{
"$replaceRoot": {
"newRoot": "$formData"
}
},
])
Mongo Playground Working Sample