Adding fields on demand to a Document

say I have this documents:

from mongoengine import Document, EmbeddedDocument,  fields
import datetime

class EmbeddedColumn(EmbeddedDocument):
    created = fields.DateTimeField(default=datetime.datetime.now)


class Dattum(Document):
    datasource_id = fields.IntField(required=True)
    date_modified = fields.DateTimeField(default=datetime.datetime.now)
    point = fields.GeoPointField()
    columns = fields.ListField(fields.EmbeddedDocumentField(EmbeddedColumn))

At runtime i need to add some field to some instances according to a series of queries:

for row in csv_attach:
    dato = Dattum(datasource_id=datasource.pk)

    for column in columns:
        col_dict = model_to_dict(column)
        col_dict.pop('id')
        ecol = EmbeddedColumn(**col_dict)
        dato.columns.append(ecol)

        if ecol.geodata_type=='point':
            local_search = gmaps.local_search('%s %s' %(ecol.value, region))
            results = local_search['responseData']['results']
            result_len = 

            if len(results) == 1:
                result = results[0]
                #dato.point(result['lat'], result['lng'])
                dato.geojson = geojson.Point(dato.point)

    dato.save()

W开发者_如何学编程hen I retrieve some Dattum I see that it has the proper columns, but the no columns has fields from model_to_dict(column) and dato have no attribute geojson.

May be I'm asking too much magic to mongoengine, may be there is a proper way to tackle this. Any pointer?

I agree with dcrosta - use a DictField instead of an EmbeddedDocument - you have no validation on the created date, other than that you get a flexible store for your data, the same as an EmbeddedField.

This feature is not yet available in mongoengine. Although there seems to be dicussion regarding this issue in github