Appengine-Fixture-Loader 0.1.5

Appengine fixture loader

appengine-fixture-loader

A simple way to load Django-like fixtures into the local development datastore, originally intended to be used by testable_appengine.

Installing

For the less adventurous, Appengine-Fixture-Loader is available on PyPI at https://pypi.python.org/pypi/Appengine-Fixture-Loader.

Single-kind loads

Let’s say you have a model like this:

class Person(ndb.Model):
    """Our sample class"""
    first_name = ndb.StringProperty()
    last_name = ndb.StringProperty()
    born = ndb.DateTimeProperty()
    userid = ndb.IntegerProperty()
    thermostat_set_to = ndb.FloatProperty()
    snores = ndb.BooleanProperty()
    started_school = ndb.DateProperty()
    sleeptime = ndb.TimeProperty()
    favorite_movies = ndb.JsonProperty()
    processed = ndb.BooleanProperty(default=False)

If you want to load a data file like this:

[
    {
        "__id__": "jdoe",
        "born": "1968-03-03T00:00:00",
        "first_name": "John",
        "last_name": "Doe",
        "favorite_movies": [
            "2001",
            "The Day The Earth Stood Still (1951)"
        ],
        "snores": false,
        "sleeptime": "23:00",
        "started_school": "1974-02-15",
        "thermostat_set_to": 18.34,
        "userid": 1
    },

...

    {
        "born": "1980-05-25T00:00:00",
        "first_name": "Bob",
        "last_name": "Schneier",
        "favorite_movies": [
            "2001",
            "Superman"
        ],
        "snores": true,
        "sleeptime": "22:00",
        "started_school": "1985-08-01",
        "thermostat_set_to": 18.34,
        "userid": -5
    }
]

All you need to do is to:

from appengine_fixture_loader.loader import load_fixture

and then:

loaded_data = load_fixture('tests/persons.json', kind = Person)

In our example, loaded_data will contain a list of already persisted Person models you can then manipulate and persist again.

The __id__ attribute, when defined, will save the object with that given id. In our case, the key to the first object defined will be a ndb.Key(‘Person’, ‘jdoe’). The key may be defined on an object by object base - where the __id__ parameter is omitted, an automatic id will be generated - the key to the second one will be something like ndb.Key(‘Person’, 1).

Multi-kind loads

It’s convenient to be able to load multiple kinds of objects from a single file. For those cases, we provide a simple way to identify the kind of object being loaded and to provide a set of models to use when loading the objects.

Consider our original example model:

class Person(ndb.Model):
    """Our sample class"""
    first_name = ndb.StringProperty()
    last_name = ndb.StringProperty()
    born = ndb.DateTimeProperty()
    userid = ndb.IntegerProperty()
    thermostat_set_to = ndb.FloatProperty()
    snores = ndb.BooleanProperty()
    started_school = ndb.DateProperty()
    sleeptime = ndb.TimeProperty()
    favorite_movies = ndb.JsonProperty()
    processed = ndb.BooleanProperty(default=False)

and let’s add a second one:

class Dog(ndb.Model):
    """Another sample class"""
    name = ndb.StringProperty()

Now, if we wanted to make a single file load objects of the two kinds, we’d need to use the __kind__ attribute in the JSON:

[
    {
        "__kind__": "Person",
        "born": "1968-03-03T00:00:00",
        "first_name": "John",
        "last_name": "Doe",
        "favorite_movies": [
            "2001",
            "The Day The Earth Stood Still (1951)"
        ],
        "snores": false,
        "sleeptime": "23:00",
        "started_school": "1974-02-15",
        "thermostat_set_to": 18.34,
        "userid": 1
    },
    {
        "__kind__": "Dog",
        "name": "Fido"
    }
]

And, to load the file, we’d have to:

from appengine_fixture_loader.loader import load_fixture

and:

loaded_data = load_fixture('tests/persons_and_dogs.json',
                           kinds={'Person': Person, 'Dog': Dog})

will result in a list of Persons and Dogs (in this case, one person and one dog).

Multi-kind, multi-level loads

Anther common case is having hierarchies of entities that you want to reconstruct for your tests.

Using slightly modified versions of our example classes:

class Person(ndb.Model):
    """Our sample class"""
    first_name = ndb.StringProperty()
    last_name = ndb.StringProperty()
    born = ndb.DateTimeProperty()
    userid = ndb.IntegerProperty()
    thermostat_set_to = ndb.FloatProperty()
    snores = ndb.BooleanProperty()
    started_school = ndb.DateProperty()
    sleeptime = ndb.TimeProperty()
    favorite_movies = ndb.JsonProperty()
    processed = ndb.BooleanProperty(default=False)
    appropriate_adult = ndb.KeyProperty()

and:

class Dog(ndb.Model):
    """Another sample class"""
    name = ndb.StringProperty()
    processed = ndb.BooleanProperty(default=False)
    owner = ndb.KeyProperty()

And using __children__[attribute_name]__ like meta-attributes, as in:

[
    {
        "__kind__": "Person",
        "born": "1968-03-03T00:00:00",
        "first_name": "John",
        "last_name": "Doe",

        ...

        "__children__appropriate_adult__": [
            {
                "__kind__": "Person",
                "born": "1970-04-27T00:00:00",

                ...

                "__children__appropriate_adult__": [
                    {
                        "__kind__": "Person",
                        "born": "1980-05-25T00:00:00",
                        "first_name": "Bob",

                        ...

                        "userid": 3
                    }
                ]
            }
        ]
    },
    {
        "__kind__": "Person",
        "born": "1999-09-19T00:00:00",
        "first_name": "Alice",

        ...

        "__children__appropriate_adult__": [
            {
                "__kind__": "Person",

                ...

                "__children__owner__": [
                    {
                        "__kind__": "Dog",
                        "name": "Fido"
                    }
                ]
            }
        ]
    }
]

you can reconstruct entire entity trees for your tests.

Parent/Ancestor-based relationships with automatic keys

It’s also possible to set the parent by using the __children__ attribute.

For our example classes, importing:

[
    {
        "__kind__": "Person",
        "first_name": "Alice",

        ...

        "__children__": [
            {
                "__kind__": "Person",
                "first_name": "Bob",
                ...

                "__children__owner__": [
                    {
                        "__kind__": "Dog",
                        "name": "Fido"
                    }
                ]
            }
        ]
    }
]

should be equivalent to:

alice = Person(first_name='Alice')
alice.put()
bob = Person(first_name='Bob', parent=alice)
bob.put()
fido = Dog(name='Fido', parent=bob)
fido.put()

You can then retrieve fido with:

fido = Dog.query(ancestor=alice.key).get()

Development

There are two recommended ways to work on this codebase. If you want to keep one and only one App Engine SDK install, you may clone the repository and run the tests by:

$ PYTHONPATH=path/to/appengine/library python setup.py test

Alternatively, this project contains code and support files derived from the testable_appengine project. Testable_appengine was conceived to make it easier to write (and run) tests for Google App Engine applications and to hook your application to Travis CI. In essence, it creates a virtualenv and downloads the most up-to-date SDK and other support tools into it. To use it, you run make. Calling make help will give you a quick list of available make targets:

$ make venv
(lots of output)
$ source .env/bin/activate
(.env) $ nosetests
(hopefully not that much output)

History

0.1.0 (2014-10-13)

• First release on GitHub.

0.1.1 (2014-12-4)

• Add support for multi-kind JSON files

0.1.2 (2014-12-4)

• Minor fixes

0.1.3 (2014-12-5)

• Added support for PropertyKey-based child entities

0.1.4 (2015-2-4)

• Fixed bug in which post-processor was called on every property change

• Added section on development to README.rst

0.1.5 (2015-2-11)

• Added __children__ support

• Added manual key definition through the __id__ attribute