projecting a computed field from embedded documents returns wrong result

[slingshotvfx]

When projecting a computed field from an embedded/nested document, mongomock returns the embedded object instead of the computed result.

Example document:

    document = {
        "a": "test_a",
        "embedded": {
            "embedded_key": "test_embedded",
        },
    }

querying that with projection={"embedded": "$embedded.embedded_key"} results in:

mongomock:

{'embedded': {'embedded_key': 'test_embedded'}}

the expected result is:

{'embedded': 'test_embedded'}

This bug is possibly related to/a dupe of #835

Full reproducible example:

import mongomock
import pymongo


def main(client: mongomock.MongoClient):
    collection = client["db_test"]["users"]

    document = {
        "a": "test_a",
        "embedded": {
            "embedded_key": "test_embedded",
        },
    }

    result = collection.insert_one(document)

    projection = {"embedded": "$embedded.embedded_key"}

    db_doc = collection.find_one({"_id": result.inserted_id}, projection=projection)
    client.close()
    return db_doc


if __name__ == "__main__":
    MONGODB_URI = "<your url here>"
    client = pymongo.MongoClient(host=MONGODB_URI)
    pymongo_res = main(client)

    client = mongomock.MongoClient()
    mongomock_res = main(client)

    print(pymongo_res)
    print(mongomock_res)
    assert pymongo_res == mongomock_res