MongoDB match in pipeline with multiple condition - mongodb

Question

I have next data:
table1
[
{
"_id" : ObjectId("5ef3611fc3e39a4891c479d9"),
"bid" : ObjectId("db08c671b5174f49886ca8de"),
},
{
"_id" : ObjectId("5ef30da4c3e39a4891c479d8"),
"bid" : ObjectId("db08c671b5174f49886ca8de"),
}
]
table2:
[
{
"_id" : ObjectId("5ef3626fc3e39a4891c479da"),
"t1_id" : ObjectId("5ef30da4c3e39a4891c479d8"),
"bid" : ObjectId("db08c671b5174f49886ca8de")
}
]
I have next SQL query with two conditions on join
SELECT table1.* FROM table1
LEFT JOIN table2 t2 ON(t2.bid=table1.bid AND t2.t1_id=table1._id)
and try to adapt it to mongoDB.
I came with next query
db.getCollection("table1").aggregate(
[
{
"$project" : {
"_id" : NumberInt(0),
"leads" : "$$ROOT"
}
},
{
"$lookup" : {
"from" : "table2",
"as" : "t2",
"let" : {
"bid" : "$bid",
"t1_id" : "$t1_id"
},
"pipeline" : [
{
"$match" : {
"$expr" : {
"$and" : [
{ "$eq" : [ "$table1._id","$$t1_id"]},
{ "$eq" : [ "$table1.bid","$$bid"]}
]
}
}
}
]
}
},
{
"$unwind" : {
"path" : "$ps",
"preserveNullAndEmptyArrays" : true
}
}
],
);
I got next result
{
"table1" : {
"_id" : ObjectId("5ef30da4c3e39a4891c479d8"),
"bid" : ObjectId("db08c671b5174f49886ca8de"),
},
"t2" : [
{
"_id" : ObjectId("5ef3626fc3e39a4891c479da"),
"t1_id" : ObjectId("5ef30da4c3e39a4891c479d8"),
"bid" : ObjectId("db08c671b5174f49886ca8de")
}
]
},
{
"table1" : {
"_id" : ObjectId("5ef3611fc3e39a4891c479d9"),
"bid" : ObjectId("db08c671b5174f49886ca8de"),
},
"t2" : [
{
"_id" : ObjectId("5ef3626fc3e39a4891c479da"),
"t1_id" : ObjectId("5ef30da4c3e39a4891c479d8"),
"bid" : ObjectId("db08c671b5174f49886ca8de")
}
]
}
And I can`t understand, why second record from table1 matched with t2, though t2.t1_id != table1._id.
Can you help me find a reason of such results and fix mongo query?

Answer 1

let parameter of $lookup takes an expression from the "local" table (t1), which I assume $t1_id doesn't exist
as parameter corresponds to the output result field of the lookup, not an alias.
In the pipeline, expressions starting with $$ are variables declared in the let parameter. expressions starting with $ are expressions from the "foreign" table (t2)
So, to directly translate the $lookup pipeline stage from your SQL statement, it should be
{
"$lookup" : {
"from" : "table2",
"as" : "t2",
"let" : {
"bid" : "$bid",
"t1_id" : "$_id" // should be $_id, as from the SQL you want table1._id, not table1.t1_id
},
"pipeline" : [
{
"$match" : {
"$expr" : {
"$and" : [
{ "$eq" : [ "$t1_id", "$$t1_id"] }, // $t1_id is from t2, $$t1_id is from variable in let parameter
{ "$eq" : [ "$bid", "$$bid"] } // $bid is from t2, $$bid is from variable in let parameter
]
}
}
}
]
}
}