Nest Aggregation with dynamic fields - elasticsearch - elasticsearch

Question

Is it possible to use nest to create buckets with keywords/fields that are not strongly typed?
Due to the nature of this project. I do not have any root objects to pass in.
Below is an example.
var result = client.Search<PortalDoc>(s => s
.Aggregations(a => a
.Terms("agg_objecttype", t => t.Field(l => "CUSTOM_FIELD_HERE"))
)
);

Answer 1

string implicitly convert to Field, so you can pass a string for any field name
var result = client.Search<PortalDoc>(s => s
.Aggregations(a => a
.Terms("agg_objecttype", t => t
.Field("CUSTOM_FIELD_HERE")
)
)
);

Answer 2

Yes, something like that is possible. Looks here for my solution using nested fields. It allows to do all the operations on "dynamic" fields, but with somewhat greater efforts (nested fields are tougher to manipulate with). The gist has some proofs for search, but I implemented aggregations as well.
curl -XPOST localhost:9200/something -d '{
"mappings" : {
"live" : {
"_source" : { "enabled" : true },
"dynamic" : false,
"properties" : {
"customFields" : {
"type" : "nested",
"properties" : {
"fieldName" : { "type" : "string", "index" : "not_analyzed" },
"stringValue": {
"type" : "string",
"fields" : {
"raw" : { "type" : "string", "index" : "not_analyzed" }
}
},
"integerValue": { "type" : "long" },
"floatValue": { "type" : "double" },
"datetimeValue": { "type" : "date", "format" : "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd" },
"booleanValue": { "type" : "boolean" }
}
}
}
}
}
}'
Search should be done in the same nested query using AND, and aggregations should be done in nested aggregation.
I made it for dynamic fields, but it probably can be adjusted for something else. I doubt there can be more flexibility with searchable/aggregatable fields due to principles how indices work.