db.chat.find().pretty().limit(3)
{
\"_id\" : ObjectId(\"593921425ccc8150f35e7662\"),
\"user1\" : 1,
\"user2\" : 2,
\"messages\" : [
{
The main thing here is the aggregation $slice to get the last element from the array,
db.chat.aggregate([
{ "$match": { "user1": 1, "messages.capty": "B" } },
{ "$redact": {
"$cond": {
"if": {
"$eq": [ { "$slice": [ "$messages.capty", -1 ] }, ["B"] ]
},
"then": "$$KEEP",
"else": "$$PRUNE"
}
}},
{ "$project": {
"user2": 1,
"body": {
"$arrayElemAt": [
{ "$map": {
"input": {
"$filter": {
"input": { "$slice": [ "$messages",-1 ] },
"as": "m",
"cond": { "$eq": [ "$$m.capty", "B" ] }
}
},
"as": "m",
"in": "$$m.body"
}},
0
]
}
}}
])
I'm actually being "extra safe" in the $project stage with the $filter but it all basically the same.
First the query selects the documents, we cannot actually say at this point to "only" match the last element of the array but we want to filter documents that do not have the condition on the array at all.
The $redact is the actual thing that looks at the "last" array entry and tests the value of the field. We can notate just the field from the array by $messages.capty
which returns just an array of those items. Here we then $slice or even $arrayElemAt if you want to get the last value, being the index of -1
.
At this point we only "filtered" the "documents" which do not match the condition. The final $project stage takes the last element of the array, checks it matches the condition ( which it should by the earlier stages ), extracts the value of "body"
and turns the single array content into just the plain value.
You can alternately forego the "carefulness" and simply just grab the last array element since $redact
should have done it's job:
db.chat.aggregate([
{ "$match": { "user1": 1, "messages.capty": "B" } },
{ "$redact": {
"$cond": {
"if": {
"$eq": [ { "$arrayElemAt": [ "$messages.capty", -1 ] }, "B" ]
},
"then": "$$KEEP",
"else": "$$PRUNE"
}
}},
{ "$project": {
"user2": 1,
"body": {
"$arrayElemAt": [ "$messages.body", -1 ]
}
}}
])
The whole thing really breaks down to "match the possible documents with a query" and then "compare and extract the last element with $slice or $arrayElemAt".
Results are:
{
"_id" : ObjectId("593921425ccc8150f35e7663"),
"user2" : 3,
"body" : "hiii 23"
}
{
"_id" : ObjectId("593921425ccc8150f35e7664"),
"user2" : 4,
"body" : "hiii 24"
}