Dynamic mapping geopoint doesn't work

[fedcecca]

I don't know if it is a bug or it is a my mistake.
I'm using elasticsearch river to import a Mongo DB collection into elasticsearch. The collection have a geo point field (position.loc).

If I try to query using the geo point field using the following query:

{

 "query" : {
        "match_all" : {}
    }, 
   "filter" : {
        "geo_distance" : {
            "distance" : "12km",
            "position.loc" : [45.07085, 7.684341]
        }
    }

}

I receive the following error message:

Caused by: org.elasticsearch.index.query.QueryParsingException: [places2_idx] failed to find geo_point field [position.loc]

When I query the mapping I receive the following info about position.loc field 👍
"position": {
"properties": {
"loc": {
"properties": {
"coordinates": {
"type": "double"
},
"index": {
"type": "string"
},
"type": {
"type": "string"
}
}
}
}
}

I think that the mapping is incorrect. How can I force river to map the field correctly?

Thanks in advance.

[richardwilly98]

You should define a customer mapping. See #105.
There are also few geopoint closed issues: #218 (comment)

[fedcecca]

thanks you for your answer,
I have a complex document JSon structure. How I can customize the mapping? I have to specify all field into document structure or I can overide only the position.loc mapping. Do you have any exemples? I have created the index (and MongoDB mapping) using the service:
POST http://:9200/_river/mongodb/_meta
{
"type": "mongodb",
"mongodb": {
"db": "<MOGO_DB> ",
"collection": "<COLLECTION_NAME>"
},
"index": {
"name": "<COLLECTION_idx",
"type": "places"
}
}

Do I have to create before a type "places" for the "index" and after run the same command (POST http://:9200/_river/mongodb/_meta)?

Thanks in advance.