This is an active fork of the original Eve-Mongoengine
Differences from the original repo:
- compatible with latest eve release.
- automatically integrate eve hooks with mongoengine methods. Inspired by a fork https://github.com/liuq/eve-mongoengine
- added capability to skip certain fields during schema construction.
- added capability to customize resource name.
Eve-Mongoengine is an Eve extension, which enables Mongoengine ODM models to be used as eve schema. If you use mongoengine in your application and simultaneously want to use eve, instead of writing schema again in cerberus format, you can use this extension, which takes your mongoengine models and auto-transforms it into creberus schema.
Official documentation: http://eve-mongoengine.readthedocs.org/en/latest/
Simple installation using pip:
pip install eve-mongoengine2
It loads all dependencies as well.
For development use virtualenv and editable copy of repisotory:
pip install -e git+https://github.com/wangsha/eve-mongoengine#egg=eve-mongoengine
- Auto-generated schema out of your mongoengine models
- Every operation goes through mongoengine -> you do not loose your mongoengine hooks
- Support for most of mongoengine fields (see Limitations for more info)
- Support for your user-defined fields (as far as they are derived from some Mongoengine's non-base field)
- Mongoengine validation layer not disconnected - use it as you wish
- Partial support for eve's media - you can use
FileField
for this purpose (again see Limitations for more info)
import mongoengine
from eve import Eve
from eve_mongoengine import EveMongoengine
# create some dummy model class
class Person(mongoengine.Document):
name = mongoengine.StringField()
age = mongoengine.IntField()
# default eve settings
my_settings = {
'MONGO_HOST': 'localhost',
'MONGO_PORT': 27017,
'MONGO_DBNAME': 'eve_mongoengine_test'
'DOMAIN': {'eve-mongoengine': {}} # sadly this is needed for eve
}
# init application
app = Eve(settings=my_settings)
# init extension
ext = EveMongoengine(app)
# register model to eve
ext.add_model(Person)
# let's roll
app.run()
Now the name of resource will be lowercase name of given class, in this example it will be
person
, so the request could be /person/
.
Or, if you are setting up your data before Eve is initialized, as is the case with application factories:
import mongoengine
from eve import Eve
from eve_mongoengine import EveMongoengine
ext = EveMongoengine()
...
# init application
app = Eve(settings=my_settings)
# init extension
ext.init_app(app)
...
If you want to use the name of model class "as is", use option lowercase=False
in add_model()
method:
ext.add_model(Person, lowercase=False)
Then you will have to ask the server for /Person/
URL.
In add_model()
method you can add every possible parameter into resource settings.
Even if you want to overwrite some settings, which generates eve-mongoengine under the hood,
you can overwrite it this way:
ext.add_model(Person, # model or models
resource_methods=['GET'], # allow only GET
cache_control="max-age=600; must-revalidate") # set max-age
When you register more than one model at time, you need to encapsulate all models into list:
ext.add_model([Person, Car, House, Dog])
Skip sensitive fields and eve hooks integration
You can mark the model with eve_exclude_fields
to skip certain model fields during schema construction. If you define eve hooks in mongoengine document, it will be automatically integrated.
class SensitiveInfoDoc(Document):
eve_exclude_fields = ["password"]
username = StringField()
password = StringField()
@staticmethod
def on_fetched_item(response):
response["extra_field"] = "a"
@staticmethod
def on_fetched_resource(response):
for item in response["_items"]:
item["extra_field"] = "a"
HTTP Methods
By default, all HTTP methods are allowed for registered classes:
- resource methods:
GET, POST, DELETE
- item methods:
GET, PATCH, PUT, DELETE
By default, eve validates against cerberus schema. Because mongoengine has larger scale
of validation possiblities, there are some cases, when cerberus is not enough. Eve-Mongoengine
comes with fancy solution: all errors, which are catchable by cerberus, are catched by cerberus
and mongoengine ones are catched by custom validator and returned in cerberus error format.
Example of this case could be mongoengine's URLField
, which does not have it's cerberus
opposie. In this case, if you fill in wrong URL, you get mongoengine error message. Let's see
an example with internet resource as a model:
class Resource(Document):
url = URLField()
author = StringField()
And then if you make POST request with wrong URL:
$ curl -d '{"url": "not-an-url", "author": "John"}' -H 'Content-Type: application/json' http://my-eve-server/resource
The response will contain
{"_status": "ERR", "_issues": {'url': "ValidationError (Resource:None) (Invalid URL: not-an-url: ['url'])"}}
Because Eve contains default functionality, which maintains fields 'updated' and 'created',
there has to be special hacky way how to do it in mongoengine too. At the time of initializing
EveMongoengine
extension, all registered mongoengine classes get two new fields: 'updated'
and 'created', both type mongoengine.DateTimeField
(of course field names are taken from config
values LAST_UPDATED
and DATE_CREATED
. This is is the only way how to ensure, that
Eve will have these fields avaliable for storing it's information about entity.
So please, do not be surprised, that there are two more fields in your model class:
class Person(mongoengine.Document):
name = mongoengine.StringField()
age = mongoengine.IntField()
app = Eve()
ext = EveMongoengine(app)
ext.add_model(Person)
Person._fields.keys() # equals ['name', 'age', 'updated', 'created']
If you already have these fields in your model, Eve will probably scream at you, that it's not possible to have these fields in schema.
No auto-updating LAST_UPDATED
field without http request
If you update your document using mongoengine model (i.e. by calling save()
, the updated
field
will NOT be automatically updated to current time. If you want this behavior, please implement
the hook yourself. Example:
class HawkeyDoc(Document):
# document with save() hooked
a = StringField(required=True)
b = StringField()
c = ReferenceField(SimpleDoc, reverse_delete_rule=CASCADE)
created_at = DateTimeField(required=True)
updated_at = DateTimeField(required=True)
def validate(self, clean=True):
now = get_utc_time()
if not self.created_at:
self.created_at = now
self.updated_at = now
return super().validate(clean)
Warning: Be aware, that when using QuerySet.update()
method, LAST_UPDATED
field WILL NOT
be updated!
Every insert/update (POST, PUT) goes through mongoengine's Document.save()
method,
but PATCH method uses as default method atomic mongoengine.QuerySet.update_one()
.
So if you have some hook bound to save()
method, you loose it in this way.
But you have an option to use save()
method in PATCH
requests in exchange
for one database fetch, so it is relatively slower. The same applies to insertion and deletion. If
you want to use this feature, set this options in data layer::
app = Eve()
ext = EveMongoengine(app)
#: this switches from using QuerySet.update_one() to Document.save()
app.data.mongoengine_options['use_document_save_for_patch'] = True
#: this switches from using cls.objects.insert() to Document.save() for each document
app.data.mongoengine_options['use_document_save_for_insert'] = True
#: this switches from using cls.objects(filter).delete() to Document.delete() for each document
app.data.mongoengine_options['use_document_delete_for_delete'] = True
ext.add_model(Person)
If you want to skip register sub-resources, set REGISTER_SUB_RESOURCE=True
in settings.
- You have to give Eve some dummy domain to shut him up. Without this he will complain about empty domain.
- You cannot use mongoengine's custom
primary_key
(because of Eve). - Cannot use
GenericEmbeddedDocumentField and SequenceField
. - You can use FileField (tested) and ImageField (not tested yet), but operation with files handles Eve's GridFS layer, not mongoengine's GridFSProxy!
- Tested only on python 3.9/3.8