location object expected, location array not in correct format

mongoengine 0.19.1
os: OS X

I have a field for geo point called location as below inside of geoip field:

"geoip" : { "continent" : "asia", "country" : "China", "country_code" : "CN", "latitude" : 1.01211, "longitude" : 0.71283, "timezone" : "Asia/Pacific", "city" : "", "area_code" : "", "location" : { "type" : "Point", "coordinates" : [ -0.15261409999993703, 51.4428321 ] } }

Now getting indexes displays that I have a 2dsphere index:

{
		"v" : 2,
		"key" : {
			"geoip.location" : "2dsphere"
		},
		"name" : "loc_idx",
		"ns" : "sample_db.users",
		"background" : false,
		"2dsphereIndexVersion" : 3
	}

Now my query in mongoengine is as follow:

[{'$geoNear': {'near': {'type': 'Point',
    'coordinates': ['-0.15261409999993703', 51.4428321]},
   'distanceField': 'geoip.location',
   'query': {'$or': [{'name': {'$regex': '.*domain.com'}},
     {'name': '172.16.131.12'}]}}}]

But when I run the below command I get an error:

result = list(Users.objects.aggregate(*geo_pipe))
pymongo.errors.OperationFailure: location object expected, location array not in correct format

Interesting part is that the same query works on mongo shell. Is this a bug?

Start by reproducing the aggregate call with the shown geo_pipe and compare it with the working Mongo shell query. Inspect how the aggregate pipeline is passed through MongoEngine, especially the $geoNear near coordinates, and use the differing behavior to isolate the discrepancy. Done means the Python aggregate accepts the same valid location format as the shell query, with a regression test covering it.