I want to use bulkloader to download all entities in a model with some self-defined Property.
If I define a model like this,
class MyType:
def __init__(self, arg):
self.name = arg['name']
self.id = arg['id']
class MyProperty(db.Property):
def get_value_for_datastore(self, instance):
val = super(MyProperty, self).get_value_for_datastore(instance)
if type(val) == dict:
val = MyType(val)
return pickle.dumps(val)
def make_value_from_datastore(self, val):
return None if val is None else pickle.loads(str(val))
class MyModel(db.Model):
info = MyProperty()
then how can I download MyModel
using the bulkloader such that there will not be un-pickled value in the file? I think I should define the export_transform
for info
in bulkloader.yaml, but I don't know what it should be like.
transformers:
- kind: MyModel
connector: csv
property_map:
- property: __key__
external_name: log_id
export_transform: transform.key_id_or_name_as_string
- property: info
external_name: info
export_transform: ### HERE ###
I've seen transform.py but still have no idea about how it works. Please tell my any method that can solve my problem. Thanks.
Okay I'm answering my own questions...
I still don't know why
pickle
doesn't work, but after changing to usesimplejson
instead ofpickle
, I can successfully exportMyProperty
in the designated format.The bulkloader.yaml may look like this.
And in myutils.py
load_info
may look like this.