MongoDb Aggregation (SQL UNION style) - mongodb

Question

I need some help/advice on how to replicate some SQL behaviour in MongoDB.
Specifically, given this collection:
{
"_id" : ObjectId("577ebc0660084921141a7857"),
"tournament" : "Wimbledon",
"player1" : "Agassi",
"player2" : "Lendl",
"sets" : [{
"score1" : 6,
"score2" : 4,
"tiebreak" : false
}, {
"score1" : 7,
"score2" : 6,
"tiebreak" : true
}, {
"score1" : 7,
"score2" : 6,
"tiebreak" : true
}]
}
{
"_id" : ObjectId("577ebc3560084921141a7858"),
"tournament" : "Wimbledon",
"player1" : "Ivanisevic",
"player2" : "McEnroe",
"sets" : [{
"score1" : 4,
"score2" : 6,
"tiebreak" : false
}, {
"score1" : 3,
"score2" : 6,
"tiebreak" : false
}, {
"score1" : 6,
"score2" : 4,
"tiebreak" : false
}]
}
{
"_id" : ObjectId("577ebc7560084921141a7859"),
"tournament" : "Roland Garros",
"player1" : "Navratilova",
"player2" : "Graf",
"sets" : [{
"score1" : 5,
"score2" : 7,
"tiebreak" : false
}, {
"score1" : 6,
"score2" : 3,
"tiebreak" : false
}, {
"score1" : 7,
"score2" : 7,
"tiebreak" : true
}, {
"score1" : 7,
"score2" : 5,
"tiebreak" : false
}]
}
And these two distinct aggregations:
1) Aggregation ALFA: this aggregation is purposely strange, in the sense that it is designed to find all matches where at least 1 tiebreak is true but only show sets where tiebreak is false. Please don't consider the logic of it, it is crafted to allow full freedom to the user.
{
$match: {
"tournament": "Wimbledon",
"sets.tiebreak": true
}
},
{
$project: {
"tournament": 1,
"player1": 1,
"sets": {
$filter: {
input: "$sets",
as: "set",
cond: {
$eq: ["$$set.tiebreak", false]
}
}
}
}
}
2) Aggregation BETA: this aggregation is purposely strange, in the sense that it is designed to find all matches where at least 1 tiebreak is false but only show sets where tiebreak is true. Please don't consider the logic of it, it is crafted to allow full freedom to the user. Please note that player1 is hidden from the results.
{
$match: {
"tournament": "Roland Garros",
"sets.tiebreak": false
}
},
{
$project: {
"tournament": 1,
"sets": {
$filter: {
input: "$sets",
as: "set",
cond: {
$eq: ["$$set.tiebreak", true]
}
}
}
}
}
Now suppose that these two aggregations purpose is to delimit what part of the database a user can see, in the sense that those two queries delimit all the documents (and details) that are visible to the user. This is similar to 2 sql views that user has rights to access.
I need/want to try to rewrite the previous distinct aggregations in only one. Can this be achieved?
It is mandatory to keep all restriction that were set in Aggregation A & B, without loosing any control on data and without leaking and data that was not available in query A or B.
Specifically, matches in Wimbledon can only be seen if they had at least one set which ended with a tiebreak. Player1 field CAN be seen. Single sets must be hidden if they did not end with a tiebreak and hidden otherwise. If needed, it is acceptable, but not desirable, to not see player1 at all.
Conversely, matches in Roland Garros can be seen only if they had at least one set which ended without a tie break. Player1 field MUST be hidden. Single sets must be seen if they ended with a tiebreak and hidden otherwise.
Again, the purpose is to UNION the two aggregations while keeping the limits imposed by the two aggregations.
MongoDB is version 3.5, can be upgraded to unstable releases if needed.

Answer 1

here's my two cents for the issue:
if you wish to avoid empty sets when
a "Wimbledon" doc has all true tibreaks,
or "Roland Garros" has all false tiebreaks
you may reshape the query:
...
{
$and: [{
"sets.tiebreak": true,
}, {
"sets.tiebreak": false
}],
$or: [{
"tournament": "Wimbledon"
}, {
"tournament": "Roland Garros"
}]
}
...
and use it in:
aggregate pipeline http://pastebin.com/cM6mNsuC
mapReduce (if performance is no a big issue..) http://pastebin.com/MShihSQL