Collection:
[
{ _id: \"Foo\", flag1: false, flag2: true, flag3: false },
{ _id: \"Bar\", flag1: true, flag2: false, flag3: true }
]
External functions don't work with the aggregation framework. Everything is parsed to BSON on input, so no JavaScript or anything else is allowed. This is all basically processed from BSON "operator" definition to native C++ code implementation so it is really fast.
What this comes down to is "converting" your expected logic to what the aggregation framework can process. There are in fact "logical" operators such as $or and $and that work in this context:
db.collection.aggregate([
{ "$project": {
"_id": 1,
"status": {
"$cond": [
{ "$or": [
// Your first set of rules requires "false" for "flag1" or
// "flag2" and "true" for "flag3"
{ "$and": [
{ "$not": [
{ "$or": [ "$flag1", "$flag2" ] },
]},
"$flag3"
]},
// Your second set of rules requires "true" for "flag1" or
// "flag2" and "false" for "flag3"
{ "$and": [
{ "$or": [ "$flag1", "$flag2" ] },
{ "$not": [ "$flag3" ] }
]},
]},
"ok",
"broken"
]
}
}}
])
So no external functions, just implement the logic with the operators that the aggregation framework supplies. In addition to the basic logical implementations there is $not to "reverse" the ligic and $cond which acts as a "ternary" in order to provide a different result from true/false evaluation.