This addon provides the basis to smoothly integrate the FastAPI framework into Odoo.
This integration allows you to use all the goodies from FastAPI to build custom APIs for your Odoo server based on standard Python type hints.
What is building an API?
An API is a set of functions that can be called from the outside world. The goal of an API is to provide a way to interact with your application from the outside world without having to know how it works internally. A common mistake when you are building an API is to expose all the internal functions of your application and therefore create a tight coupling between the outside world and your internal datamodel and business logic. This is not a good idea because it makes it very hard to change your internal datamodel and business logic without breaking the outside world.
When you are building an API, you define a contract between the outside world and your application. This contract is defined by the functions that you expose and the parameters that you accept. This contract is the API. When you change your internal datamodel and business logic, you can still keep the same API contract and therefore you don't break the outside world. Even if you change your implementation, as long as you keep the same API contract, the outside world will still work. This is the beauty of an API and this is why it is so important to design a good API.
A good API is designed to be stable and to be easy to use. It's designed to provide high-level functions related to a specific use case. It's designed to be easy to use by hiding the complexity of the internal datamodel and business logic. A common mistake when you are building an API is to expose all the internal functions of your application and let the oustide world deal with the complexity of your internal datamodel and business logic. Don't forget that on a transactional point of view, each call to an API function is a transaction. This means that if a specific use case requires multiple calls to your API, you should provide a single function that does all the work in a single transaction. This why APIs methods are called high-level and atomic functions.
Table of contents
- Usage
- What's building an API with fastapi?
- Dealing with the odoo environment
- The dependency injection mechanism
- The authentication mechanism
- Managing configuration parameters for your app
- Dealing with languages
- How to extend an existing app
- Managing security into the route handlers
- How to test your fastapi app
- Overall considerations when you develop an fastapi app
- Miscellaneous
- What's next?
- Known issues / Roadmap
- Changelog
- Bug Tracker
- Credits
FastAPI is a modern, fast (high-performance), web framework for building APIs with Python 3.7+ based on standard Python type hints. This addons let's you keep advantage of the fastapi framework and use it with Odoo.
Before you start, we must define some terms:
- App: A FastAPI app is a collection of routes, dependencies, and other components that can be used to build a web application.
- Router: A router is a collection of routes that can be mounted in an app.
- Route: A route is a mapping between an HTTP method and a path, and defines what should happen when the user requests that path.
- Dependency: A dependency is a callable that can be used to get some information from the user request, or to perform some actions before the request handler is called.
- Request: A request is an object that contains all the information sent by the user's browser as part of an HTTP request.
- Response: A response is an object that contains all the information that the user's browser needs to build the result page.
- Handler: A handler is a function that takes a request and returns a response.
- Middleware: A middleware is a function that takes a request and a handler, and returns a response.
The FastAPI framework is based on the following principles:
- Fast: Very high performance, on par with NodeJS and Go (thanks to Starlette and Pydantic). [One of the fastest Python frameworks available]
- Fast to code: Increase the speed to develop features by about 200% to 300%.
- Fewer bugs: Reduce about 40% of human (developer) induced errors.
- Intuitive: Great editor support. Completion everywhere. Less time debugging.
- Easy: Designed to be easy to use and learn. Less time reading docs.
- Short: Minimize code duplication. Multiple features from each parameter declaration. Fewer bugs.
- Robust: Get production-ready code. With automatic interactive documentation.
- Standards-based: Based on (and fully compatible with) the open standards for APIs: OpenAPI (previously known as Swagger) and JSON Schema.
- Open Source: FastAPI is fully open-source, under the MIT license.
The first step is to install the fastapi addon. You can do it with the following command:
$ pip install odoo-addon-fastapi
Once the addon is installed, you can start building your API. The first thing you need to do is to create a new addon that depends on 'fastapi'. For example, let's create an addon called my_demo_api.
Then, you need to declare your app by defining a model that inherits from 'fastapi.endpoint' and add your app name into the app field. For example:
from odoo import fields, models
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)The 'fastapi.endpoint' model is the base model for all the endpoints. An endpoint instance is the mount point for a fastapi app into Odoo. When you create a new endpoint, you can define the app that you want to mount in the 'app' field and the path where you want to mount it in the 'path' field.
figure:: static/description/endpoint_create.png
FastAPI Endpoint
Thanks to the 'fastapi.endpoint' model, you can create as many endpoints as you want and mount as many apps as you want in each endpoint. The endpoint is also the place where you can define configuration parameters for your app. A typical example is the authentication method that you want to use for your app when accessed at the endpoint path.
Now, you can create your first router. For that, you need to define a global variable into your fastapi_endpoint module called for example 'demo_api_router'
from fastapi import APIRouter
from odoo import fields, models
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)
# create a router
demo_api_router = APIRouter()To make your router available to your app, you need to add it to the list of routers returned by the _get_fastapi_routers method of your fastapi_endpoint model.
from fastapi import APIRouter
from odoo import api, fields, models
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)
def _get_fastapi_routers(self):
if self.app == "demo":
return [demo_api_router]
return super()._get_fastapi_routers()
# create a router
demo_api_router = APIRouter()Now, you can start adding routes to your router. For example, let's add a route that returns a list of partners.
from typing import Annotated
from fastapi import APIRouter
from pydantic import BaseModel
from odoo import api, fields, models
from odoo.api import Environment
from odoo.addons.fastapi.dependencies import odoo_env
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)
def _get_fastapi_routers(self):
if self.app == "demo":
return [demo_api_router]
return super()._get_fastapi_routers()
# create a router
demo_api_router = APIRouter()
class PartnerInfo(BaseModel):
name: str
email: str
@demo_api_router.get("/partners", response_model=list[PartnerInfo])
def get_partners(env: Annotated[Environment, Depends(odoo_env)]) -> list[PartnerInfo]:
return [
PartnerInfo(name=partner.name, email=partner.email)
for partner in env["res.partner"].search([])
]Now, you can start your Odoo server, install your addon and create a new endpoint instance for your app. Once it's done click on the docs url to access the interactive documentation of your app.
Before trying to test your app, you need to define on the endpoint instance the user that will be used to run the app. You can do it by setting the 'user_id' field. This information is the most important one because it's the basis for the security of your app. The user that you define in the endpoint instance will be used to run the app and to access the database. This means that the user will be able to access all the data that he has access to in Odoo. To ensure the security of your app, you should create a new user that will be used only to run your app and that will have no access to the database.
<record
id="my_demo_app_user"
model="res.users"
context="{'no_reset_password': True, 'no_reset_password': True}"
>
<field name="name">My Demo Endpoint User</field>
<field name="login">my_demo_app_user</field>
<field name="groups_id" eval="[(6, 0, [])]" />
</record>At the same time you should create a new group that will be used to define the access rights of the user that will run your app. This group should imply the predefined group 'FastAPI Endpoint Runner'. This group defines the minimum access rights that the user needs to:
- access the endpoint instance it belongs to
- access to its own user record
- access to the partner record that is linked to its user record
<record id="my_demo_app_group" model="res.groups">
<field name="name">My Demo Endpoint Group</field>
<field name="users" eval="[(4, ref('my_demo_app_user'))]" />
<field name="implied_ids" eval="[(4, ref('fastapi.group_fastapi_endpoint_runner'))]" />
</record>Now, you can test your app. You can do it by clicking on the 'Try it out' button of the route that you have defined. The result of the request will be displayed in the 'Response' section and contains the list of partners.
Note
The 'FastAPI Endpoint Runner' group ensures that the user cannot access any information others than the 3 ones mentioned above. This means that for every information that you want to access from your app, you need to create the proper ACLs and record rules. (see Managing security into the route handlers) It's a good practice to use a dedicated user into a specific group from the beginning of your project and in your tests. This will force you to define the proper security rules for your endoints.
The 'odoo.addons.fastapi.dependencies' module provides a set of functions that you can use to inject reusable dependencies into your routes. For example, the 'odoo_env' function returns the current odoo environment. You can use it to access the odoo models and the database from your route handlers.
from typing import Annotated
from odoo.api import Environment
from odoo.addons.fastapi.dependencies import odoo_env
@demo_api_router.get("/partners", response_model=list[PartnerInfo])
def get_partners(env: Annotated[Environment, Depends(odoo_env)]) -> list[PartnerInfo]:
return [
PartnerInfo(name=partner.name, email=partner.email)
for partner in env["res.partner"].search([])
]As you can see, you can use the 'Depends' function to inject the dependency into your route handler. The 'Depends' function is provided by the 'fastapi' framework. You can use it to inject any dependency into your route handler. As your handler is a python function, the only way to get access to the odoo environment is to inject it as a dependency. The fastapi addon provides a set of function that can be used as dependencies:
- 'odoo_env': Returns the current odoo environment.
- 'fastapi_endpoint': Returns the current fastapi endpoint model instance.
- 'authenticated_partner': Returns the authenticated partner.
- 'authenticated_partner_env': Returns the current odoo environment with the authenticated_partner_id into the context.
By default, the 'odoo_env' and 'fastapi_endpoint' dependencies are available without extra work.
Note
Even if 'odoo_env' and 'authenticated_partner_env' returns the current odoo environment, they are not the same. The 'odoo_env' dependency returns the environment without any modification while the 'authenticated_partner_env' adds the authenticated partner id into the context of the environment. As it will be explained in the section Managing security into the route handlers dedicated to the security, the presence of the authenticated partner id into the context is the key information that will allow you to enforce the security of your endpoint methods. As consequence, you should always use the 'authenticated_partner_env' dependency instead of the 'odoo_env' dependency for all the methods that are not public.
The 'odoo_env' dependency relies on a simple implementation that retrieves the current odoo environment from ContextVar variable initialized at the start of the request processing by the specific request dispatcher processing the fastapi requests.
The 'fastapi_endpoint' dependency relies on the 'dependency_overrides' mechanism provided by the 'fastapi' module. (see the fastapi documentation for more details about the dependency_overrides mechanism). If you take a look at the current implementation of the 'fastapi_endpoint' dependency, you will see that the method depends of two parameters: 'endpoint_id' and 'env'. Each of these parameters are dependencies themselves.
def fastapi_endpoint_id() -> int:
"""This method is overriden by default to make the fastapi.endpoint record
available for your endpoint method. To get the fastapi.endpoint record
in your method, you just need to add a dependency on the fastapi_endpoint method
defined below
"""
def fastapi_endpoint(
_id: Annotated[int, Depends(fastapi_endpoint_id)],
env: Annotated[Environment, Depends(odoo_env)],
) -> "FastapiEndpoint":
"""Return the fastapi.endpoint record"""
return env["fastapi.endpoint"].browse(_id)As you can see, one of these dependencies is the 'fastapi_endpoint_id' dependency and has no concrete implementation. This method is used as a contract that must be implemented/provided at the time the fastapi app is created. Here comes the power of the dependency_overrides mechanism.
If you take a look at the '_get_app' method of the 'FastapiEndpoint' model, you will see that the 'fastapi_endpoint_id' dependency is overriden by registering a specific method that returns the id of the current fastapi endpoint model instance for the original method.
def _get_app(self) -> FastAPI:
app = FastAPI(**self._prepare_fastapi_endpoint_params())
for router in self._get_fastapi_routers():
app.include_router(prefix=self.root_path, router=router)
app.dependency_overrides[dependencies.fastapi_endpoint_id] = partial(
lambda a: a, self.id
)This kind of mechanism is very powerful and allows you to inject any dependency into your route handlers and moreover, define an abstract dependency that can be used by any other addon and for which the implementation could depend on the endpoint configuration.
To make our app not tightly coupled with a specific authentication mechanism, we will use the 'authenticated_partner' dependency. As for the 'fastapi_endpoint' this dependency depends on an abstract dependency.
When you define a route handler, you can inject the 'authenticated_partner' dependency as a parameter of your route handler.
from odoo.addons.base.models.res_partner import Partner
@demo_api_router.get("/partners", response_model=list[PartnerInfo])
def get_partners(
env: Annotated[Environment, Depends(odoo_env)], partner: Annotated[Partner, Depends(authenticated_partner)]
) -> list[PartnerInfo]:
return [
PartnerInfo(name=partner.name, email=partner.email)
for partner in env["res.partner"].search([])
]At this stage, your handler is not tied to a specific authentication mechanism but only expects to get a partner as a dependency. Depending on your needs, you can implement different authentication mechanism available for your app. The fastapi addon provides a default authentication mechanism using the 'BasicAuth' method. This authentication mechanism is implemented in the 'odoo.addons.fastapi.dependencies' module and relies on functionalities provided by the 'fastapi.security' module.
def authenticated_partner(
env: Annotated[Environment, Depends(odoo_env)],
security: Annotated[HTTPBasicCredentials, Depends(HTTPBasic())],
) -> "res.partner":
"""Return the authenticated partner"""
partner = env["res.partner"].search(
[("email", "=", security.username)], limit=1
)
if not partner:
raise HTTPException(
status_code=status.HTTP_401_UNAUTHORIZED,
detail="Invalid authentication credentials",
headers={"WWW-Authenticate": "Basic"},
)
if not partner.check_password(security.password):
raise HTTPException(
status_code=status.HTTP_401_UNAUTHORIZED,
detail="Invalid authentication credentials",
headers={"WWW-Authenticate": "Basic"},
)
return partnerAs you can see, the 'authenticated_partner' dependency relies on the 'HTTPBasic' dependency provided by the 'fastapi.security' module. In this dummy implementation, we just check that the provided credentials can be used to authenticate a user in odoo. If the authentication is successful, we return the partner record linked to the authenticated user.
In some cases you could want to implement a more complex authentication mechanism that could rely on a token or a session. In this case, you can override the 'authenticated_partner' dependency by registering a specific method that returns the authenticated partner. Moreover, you can make it configurable on the fastapi endpoint model instance.
To do it, you just need to implement a specific method for each of your authentication mechanism and allows the user to select one of these methods when he creates a new fastapi endpoint. Let's say that we want to allow the authentication by using an api key or via basic auth. Since basic auth is already implemented, we will only implement the api key authentication mechanism.
from fastapi.security import APIKeyHeader
def api_key_based_authenticated_partner_impl(
api_key: Annotated[str, Depends(
APIKeyHeader(
name="api-key",
description="In this demo, you can use a user's login as api key.",
)
)],
env: Annotated[Environment, Depends(odoo_env)],
) -> Partner:
"""A dummy implementation that look for a user with the same login
as the provided api key
"""
partner = env["res.users"].search([("login", "=", api_key)], limit=1).partner_id
if not partner:
raise HTTPException(
status_code=status.HTTP_401_UNAUTHORIZED, detail="Incorrect API Key"
)
return partnerAs for the 'BasicAuth' authentication mechanism, we also rely on one of the native security dependency provided by the 'fastapi.security' module.
Now that we have an implementation for our two authentication mechanisms, we can allows the user to select one of these authentication mechanisms by adding a selection field on the fastapi endpoint model.
from odoo import fields, models
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)
demo_auth_method = fields.Selection(
selection=[("api_key", "Api Key"), ("http_basic", "HTTP Bacic")],
string="Authenciation method",
)Note
A good practice is to prefix specific configuration fields of your app with the name of your app. This will avoid conflicts with other app when the 'fastapi.endpoint' model is extended for other 'app'.
Now that we have a selection field that allows the user to select the authentication method, we can use the dependency override mechanism to provide the right implementation of the 'authenticated_partner' dependency when the app is instantiated.
from odoo.addons.fastapi.dependencies import authenticated_partner
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)
demo_auth_method = fields.Selection(
selection=[("api_key", "Api Key"), ("http_basic", "HTTP Bacic")],
string="Authenciation method",
)
def _get_app(self) -> FastAPI:
app = super()._get_app()
if self.app == "demo":
# Here we add the overrides to the authenticated_partner_impl method
# according to the authentication method configured on the demo app
if self.demo_auth_method == "http_basic":
authenticated_partner_impl_override = (
authenticated_partner_from_basic_auth_user
)
else:
authenticated_partner_impl_override = (
api_key_based_authenticated_partner_impl
)
app.dependency_overrides[
authenticated_partner_impl
] = authenticated_partner_impl_override
return appTo see how the dependency override mechanism works, you can take a look at the demo app provided by the fastapi addon. If you choose the app 'demo' in the fastapi endpoint form view, you will see that the authentication method is configurable. You can also see that depending on the authentication method configured on your fastapi endpoint, the documentation will change.
Note
At time of writing, the dependency override mechanism is not supported by the fastapi documentation generator. A fix has been proposed and is waiting to be merged. You can follow the progress of the fix on github
As we have seen in the previous section, you can add configuration fields on the fastapi endpoint model to allow the user to configure your app (as for any odoo model you extend). When you need to access these configuration fields in your route handlers, you can use the 'odoo.addons.fastapi.dependencies.fastapi_endpoint' dependency method to retrieve the 'fastapi.endpoint' record associated to the current request.
from pydantic import BaseModel, Field
from odoo.addons.fastapi.dependencies import fastapi_endpoint
class EndpointAppInfo(BaseModel):
id: str
name: str
app: str
auth_method: str = Field(alias="demo_auth_method")
root_path: str
model_config = ConfigDict(from_attributes=True)
@demo_api_router.get(
"/endpoint_app_info",
response_model=EndpointAppInfo,
dependencies=[Depends(authenticated_partner)],
)
async def endpoint_app_info(
endpoint: Annotated[FastapiEndpoint, Depends(fastapi_endpoint)],
) -> EndpointAppInfo:
"""Returns the current endpoint configuration"""
# This method show you how to get access to current endpoint configuration
# It also show you how you can specify a dependency to force the security
# even if the method doesn't require the authenticated partner as parameter
return EndpointAppInfo.model_validate(endpoint)Some of the configuration fields of the fastapi endpoint could impact the way the app is instantiated. For example, in the previous section, we have seen that the authentication method configured on the 'fastapi.endpoint' record is used in order to provide the right implementation of the 'authenticated_partner' when the app is instantiated. To ensure that the app is re-instantiated when an element of the configuration used in the instantiation of the app is modified, you must override the '_fastapi_app_fields' method to add the name of the fields that impact the instantiation of the app into the returned list.
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
app: str = fields.Selection(
selection_add=[("demo", "Demo Endpoint")], ondelete={"demo": "cascade"}
)
demo_auth_method = fields.Selection(
selection=[("api_key", "Api Key"), ("http_basic", "HTTP Bacic")],
string="Authenciation method",
)
@api.model
def _fastapi_app_fields(self) -> List[str]:
fields = super()._fastapi_app_fields()
fields.append("demo_auth_method")
return fieldsThe fastapi addon parses the Accept-Language header of the request to determine the language to use. This parsing is done by respecting the RFC 7231 specification. That means that the language is determined by the first language found in the header that is supported by odoo (with care of the priority order). If no language is found in the header, the odoo default language is used. This language is then used to initialize the Odoo's environment context used by the route handlers. All this makes the management of languages very easy. You don't have to worry about. This feature is also documented by default into the generated openapi documentation of your app to instruct the api consumers how to request a specific language.
When you develop a fastapi app, in a native python app it's not possible to extend an existing one. This limitation doesn't apply to the fastapi addon because the fastapi endpoint model is designed to be extended. However, the way to extend an existing app is not the same as the way to extend an odoo model.
First of all, it's important to keep in mind that when you define a route, you are actually defining a contract between the client and the server. This contract is defined by the route path, the method (GET, POST, PUT, DELETE, etc.), the parameters and the response. If you want to extend an existing app, you must ensure that the contract is not broken. Any change to the contract will respect the Liskov substitution principle. This means that the client should not be impacted by the change.
What does it mean in practice? It means that you can't change the route path or the method of an existing route. You can't change the name of a parameter or the type of a response. You can't add a new parameter or a new response. You can't remove a parameter or a response. If you want to change the contract, you must create a new route.
What can you change?
- You can change the implementation of the route handler.
- You can override the dependencies of the route handler.
- You can add a new route handler.
- You can extend the model used as parameter or as response of the route handler.
Let's see how to do that.
Let's say that you want to change the implementation of the route handler '/demo/echo'. Since a route handler is just a python method, it could seems a tedious task since we are not into a model method and therefore we can't take advantage of the Odoo inheritance mechanism.
However, the fastapi addon provides a way to do that. Thanks to the 'odoo_env' dependency method, you can access the current odoo environment. With this environment, you can access the registry and therefore the model you want to delegate the implementation to. If you want to change the implementation of the route handler '/demo/echo', the only thing you have to do is to inherit from the model where the implementation is defined and override the method 'echo'.
from pydantic import BaseModel
from fastapi import Depends, APIRouter
from odoo import models
from odoo.addons.fastapi.dependencies import odoo_env
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
def _get_fastapi_routers(self) -> List[APIRouter]:
routers = super()._get_fastapi_routers()
routers.append(demo_api_router)
return routers
demo_api_router = APIRouter()
@demo_api_router.get(
"/echo",
response_model=EchoResponse,
dependencies=[Depends(odoo_env)],
)
async def echo(
message: str,
odoo_env: Annotated[Environment, Depends(odoo_env)],
) -> EchoResponse:
"""Echo the message"""
return EchoResponse(message=odoo_env["demo.fastapi.endpoint"].echo(message))
class EchoResponse(BaseModel):
message: str
class DemoEndpoint(models.AbstractModel):
_name = "demo.fastapi.endpoint"
_description = "Demo Endpoint"
def echo(self, message: str) -> str:
return message
class DemoEndpointInherit(models.AbstractModel):
_inherit = "demo.fastapi.endpoint"
def echo(self, message: str) -> str:
return f"Hello {message}"Note
It's a good programming practice to implement the business logic outside the route handler. This way, you can easily test your business logic without having to test the route handler. In the example above, the business logic is implemented in the method 'echo' of the model 'demo.fastapi.endpoint'. The route handler just delegate the implementation to this method.
As you've previously seen, the dependency injection mechanism of fastapi is very powerful. By designing your route handler to rely on dependencies with a specific functional scope, you can easily change the implementation of the dependency without having to change the route handler. With such a design, you can even define abstract dependencies that must be implemented by the concrete application. This is the case of the 'authenticated_partner' dependency in our previous example. (you can find the implementation of this dependency in the file 'odoo/addons/fastapi/dependencies.py' and it's usage in the file 'odoo/addons/fastapi/models/fastapi_endpoint_demo.py')
Let's say that you want to add a new route handler '/demo/echo2'. You could be tempted to add this new route handler in your new addons by importing the router of the existing app and adding the new route handler to it.
from odoo.addons.fastapi.models.fastapi_endpoint_demo import demo_api_router
@demo_api_router.get(
"/echo2",
response_model=EchoResponse,
dependencies=[Depends(odoo_env)],
)
async def echo2(
message: str,
odoo_env: Annotated[Environment, Depends(odoo_env)],
) -> EchoResponse:
"""Echo the message"""
echo = odoo_env["demo.fastapi.endpoint"].echo2(message)
return EchoResponse(message=f"Echo2: {echo}")The problem with this approach is that you unconditionally add the new route handler to the existing app even if the app is called for a different database where your new addon is not installed.
The solution is to define a new router and to add it to the list of routers returned by the method '_get_fastapi_routers' of the model 'fastapi.endpoint' you are inheriting from into your new addon.
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
def _get_fastapi_routers(self) -> List[APIRouter]:
routers = super()._get_fastapi_routers()
if self.app == "demo":
routers.append(additional_demo_api_router)
return routers
additional_demo_api_router = APIRouter()
@additional_demo_api_router.get(
"/echo2",
response_model=EchoResponse,
dependencies=[Depends(odoo_env)],
)
async def echo2(
message: str,
odoo_env: Annotated[Environment, Depends(odoo_env)],
) -> EchoResponse:
"""Echo the message"""
echo = odoo_env["demo.fastapi.endpoint"].echo2(message)
return EchoResponse(message=f"Echo2: {echo}")In this way, the new router is added to the list of routers of your app only if the app is called for a database where your new addon is installed.
The fastapi python library uses the pydantic library to define the models. By default, once a model is defined, it's not possible to extend it. However, a companion python library called extendable_pydantic provides a way to use inheritance with pydantic models to extend an existing model. If used alone, it's your responsibility to instruct this library the list of extensions to apply to a model and the order to apply them. This is not very convenient. Fortunately, an dedicated odoo addon exists to make this process complete transparent. This addon is called odoo-addon-extendable-fastapi.
When you want to allow other addons to extend a pydantic model, you must first define the model as an extendable model by using a dedicated metaclass
from pydantic import BaseModel
from extendable_pydantic import ExtendableModelMeta
class Partner(BaseModel, metaclass=ExtendableModelMeta):
name = 0.1
model_config = ConfigDict(from_attributes=True)As any other pydantic model, you can now use this model as parameter or as response of a route handler. You can also use all the features of models defined with pydantic.
@demo_api_router.get(
"/partner",
response_model=Location,
dependencies=[Depends(authenticated_partner)],
)
async def partner(
partner: Annotated[ResPartner, Depends(authenticated_partner)],
) -> Partner:
"""Return the location"""
return Partner.model_validate(partner)If you need to add a new field into the model 'Partner', you can extend it in your new addon by defining a new model that inherits from the model 'Partner'.
from typing import Optional
from odoo.addons.fastapi.models.fastapi_endpoint_demo import Partner
class PartnerExtended(Partner, extends=Partner):
email: Optional[str]If your new addon is installed in a database, a call to the route handler '/demo/partner' will return a response with the new field 'email' if a value is provided by the odoo record.
{
"name": "John Doe",
"email": "[email protected]"
}If your new addon is not installed in a database, a call to the route handler '/demo/partner' will only return the name of the partner.
{
"name": "John Doe"
}Note
The liskov substitution principle has also to be respected. That means that if you extend a model, you must add new required fields or you must provide default values for the new optional fields.
By default the route handlers are processed using the user configured on the 'fastapi.endpoint' model instance. (default is the Public user). You have seen previously how to define a dependency that will be used to enforce the authentication of a partner. When a method depends on this dependency, the 'authenticated_partner_id' key is added to the context of the partner environment. (If you don't need the partner as dependency but need to get an environment with the authenticated user, you can use the dependency 'authenticated_partner_env' instead of 'authenticated_partner'.)
The fastapi addon extends the 'ir.rule' model to add into the evaluation context of the security rules the key 'authenticated_partner_id' that contains the id of the authenticated partner.
As briefly introduced in a previous section, a good practice when you develop a fastapi app and you want to protect your data in an efficient and traceable way is to:
- create a new user specific to the app but with any access rights.
- create a security group specific to the app and add the user to this group. (This group must implies the group 'AFastAPI Endpoint Runner' that give the minimal access rights)
- for each model you want to protect:
- add a 'ir.model.access' record for the model to allow read access to your model and add the group to the record.
- create a new 'ir.rule' record for the model that restricts the access to the records of the model to the authenticated partner by using the key 'authenticated_partner_id' in domain of the rule. (or to the user defined on the 'fastapi.endpoint' model instance if the method is public)
- add a dependency on the 'authenticated_partner' to your handlers when you need to access the authenticated partner or ensure that the service is called by an authenticated partner.
<record
id="my_demo_app_user"
model="res.users"
context="{'no_reset_password': True, 'no_reset_password': True}"
>
<field name="name">My Demo Endpoint User</field>
<field name="login">my_demo_app_user</field>
<field name="groups_id" eval="[(6, 0, [])]" />
</record>
<record id="my_demo_app_group" model="res.groups">
<field name="name">My Demo Endpoint Group</field>
<field name="users" eval="[(4, ref('my_demo_app_user'))]" />
<field name="implied_ids" eval="[(4, ref('group_fastapi_endpoint_runner'))]" />
</record>
<!-- acl for the model 'sale.order' -->
<record id="sale_order_demo_app_access" model="ir.model.access">
<field name="name">My Demo App: access to sale.order</field>
<field name="model_id" ref="model_sale_order"/>
<field name="group_id" ref="my_demo_app_group"/>
<field name="perm_read" eval="True"/>
<field name="perm_write" eval="False"/>
<field name="perm_create" eval="False"/>
<field name="perm_unlink" eval="False"/>
</record>
<!-- a record rule to allows the authenticated partner to access only its sale orders -->
<record id="demo_app_sale_order_rule" model="ir.rule">
<field name="name">Sale Order Rule</field>
<field name="model_id" ref="model_sale_order"/>
<field name="domain_force">[('partner_id', '=', authenticated_partner_id)]</field>
<field name="groups" eval="[(4, ref('my_demo_app_group'))]"/>
</record>Thanks to the starlette test client, it's possible to test your fastapi app in a very simple way. With the test client, you can call your route handlers as if they were real http endpoints. The test client is available in the 'fastapi.testclient' module.
Once again the dependency injection mechanism comes to the rescue by allowing you to inject into the test client specific implementations of the dependencies normally provided by the normal processing of the request by the fastapi app. (for example, you can inject a mock of the dependency 'authenticated_partner' to test the behavior of your route handlers when the partner is not authenticated, you can also inject a mock for the odoo_env etc...)
The fastapi addon provides a base class for the test cases that you can use to write your tests. This base class is 'odoo.fastapi.tests.common.FastAPITransactionCase'. This class mainly provides the method '_create_test_client' that you can use to create a test client for your fastapi app. This method encapsulates the creation of the test client and the injection of the dependencies. It also ensures that the odoo environment is make available into the context of the route handlers. This method is designed to be used when you need to test your app or when you need to test a specific router (It's therefore easy to defines tests for routers in an addon that doesn't provide a fastapi endpoint).
With this base class, writing a test for a route handler is as simple as:
from odoo.fastapi.tests.common import FastAPITransactionCase
from odoo.addons.fastapi import dependencies
from odoo.addons.fastapi.routers import demo_router
class FastAPIDemoCase(FastAPITransactionCase):
@classmethod
def setUpClass(cls) -> None:
super().setUpClass()
cls.default_fastapi_running_user = cls.env.ref("fastapi.my_demo_app_user")
cls.default_fastapi_authenticated_partner = cls.env["res.partner"].create({"name": "FastAPI Demo"})
def test_hello_world(self) -> None:
with self._create_test_client(router=demo_router) as test_client:
response: Response = test_client.get("/demo/")
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertDictEqual(response.json(), {"Hello": "World"})In the previous example, we created a test client for the demo_router. We could have created a test client for the whole app by not specifying the router but the app instead.
from odoo.fastapi.tests.common import FastAPITransactionCase
from odoo.addons.fastapi import dependencies
from odoo.addons.fastapi.routers import demo_router
class FastAPIDemoCase(FastAPITransactionCase):
@classmethod
def setUpClass(cls) -> None:
super().setUpClass()
cls.default_fastapi_running_user = cls.env.ref("fastapi.my_demo_app_user")
cls.default_fastapi_authenticated_partner = cls.env["res.partner"].create({"name": "FastAPI Demo"})
def test_hello_world(self) -> None:
demo_endpoint = self.env.ref("fastapi.fastapi_endpoint_demo")
with self._create_test_client(app=demo_endpoint._get_app()) as test_client:
response: Response = test_client.get(f"{demo_endpoint.root_path}/demo/")
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertDictEqual(response.json(), {"Hello": "World"})Developing a fastapi app requires to follow some good practices to ensure that the app is robust and easy to maintain. Here are some of them:
- A route handler must be as simple as possible. It must not contain any business logic. The business logic must be implemented into the service layer. The route handler must only call the service layer and return the result of the service layer. To ease extension on your business logic, your service layer can be implemented as an odoo abstract model that can be inherited by other addons.
- A route handler should not expose the internal data structure and api of Odoo. It should provide the api that is needed by the client. More widely, an app provides a set of services that address a set of use cases specific to a well defined functional domain. You must always keep in mind that your api will remain the same for a long time even if you upgrade your odoo version of modify your business logic.
- A route handler is a transactional unit of work. When you design your api you must ensure that the completeness of a use case is guaranteed by a single transaction. If you need to perform several transactions to complete a use case, you introduce a risk of inconsistency in your data or extra complexity in your client code.
- Properly handle the errors. The route handler must return a proper error response when an error occurs. The error response must be consistent with the rest of the api. The error response must be documented in the api documentation. By default, the 'odoo-addon-fastapi' module handles the common exception types defined in the 'odoo.exceptions' module and returns a proper error response with the corresponding http status code. An error in the route handler must always return an error response with a http status code different from 200. The error response must contain a human readable message that can be displayed to the user. The error response can also contain a machine readable code that can be used by the client to handle the error in a specific way.
- When you design your json document through the pydantic models, you must use the appropriate data types. For example, you must use the data type 'datetime.date' to represent a date and not a string. You must also properly define the constraints on the fields. For example, if a field is optional, you must use the data type 'typing.Optional'. pydantic provides everything you need to properly define your json document.
- Always use an appropriate pydantic model as request and/or response for your route handler. Constraints on the fields of the pydantic model must apply to the specific use case. For example, if your route handler is used to create a sale order, the pydantic model must not contain the field 'id' because the id of the sale order will be generated by the route handler. But if the id is required afterwords, the pydantic model for the response must contain the field 'id' as required.
- Uses descriptive property names in your json documents. For example, avoid the use of documents providing a flat list of key value pairs.
- Be consistent in the naming of your fields into your json documents. For example, if you use 'id' to represent the id of a sale order, you must use 'id' to represent the id of all the other objects.
- Be consistent in the naming style of your fields. Always prefer underscore to camel case.
- Always use plural for the name of the fields that contain a list of items. For example, if you have a field 'lines' that contains a list of sale order lines, you must use 'lines' and not 'line'.
- You can't expect that a client will provide you the identifier of a specific record in odoo (for example the id of a carrier) if you don't provide a specific route handler to retrieve the list of available records. Sometimes, the client must share with odoo the identity of a specific record to be able to perform an appropriate action specific to this record (for example, the processing of a payment is different for each payment acquirer). In this case, you must provide a specific attribute that allows both the client and odoo to identify the record. The field 'provider' on a payment acquirer allows you to identify a specific record in odoo. This kind of approach allows both the client and odoo to identify the record without having to rely on the id of the record. (This will ensure that the client will not break if the id of the record is changed in odoo for example when tests are run on an other database).
- Always use the same name for the same kind of object. For example, if you have a field 'lines' that contains a list of sale order lines, you must use the same name for the same kind of object in all the other json documents.
- Manage relations between objects in your json documents the same way. By default, you should return the id of the related object in the json document. But this is not always possible or convenient, so you can also return the related object in the json document. The main advantage of returning the id of the related object is that it allows you to avoid the n+1 problem . The main advantage of returning the related object in the json document is that it allows you to avoid an extra call to retrieve the related object. By keeping in mind the pros and cons of each approach, you can choose the best one for your use case. Once it's done, you must be consistent in the way you manage the relations of the same object.
- It's not always a good idea to name your fields into your json documents with the same name as the fields of the corresponding odoo model. For example, in your document representing a sale order, you must not use the name 'order_line' for the field that contains the list of sale order lines. The name 'order_line' in addition to being confusing and not consistent with the best practices, is not auto-descriptive. The name 'lines' is much better.
- Keep a defensive programming approach. If you provide a route handler that returns a list of records, you must ensure that the computation of the list is not too long or will not drain your server resources. For example, for search route handlers, you must ensure that the search is limited to a reasonable number of records by default.
- As a corollary of the previous point, a search handler must always use the pagination mechanism with a reasonable default page size. The result list must be enclosed in a json document that contains the count of records into the system matching your search criteria and the list of records for the given page and size.
- Use plural for the name of a service. For example, if you provide a service that allows you to manage the sale orders, you must use the name 'sale_orders' and not 'sale_order'.
- ... and many more.
We could write a book about the best practices to follow when you design your api but we will stop here. This list is the result of our experience at ACSONE SA/NV and it evolves over time. It's a kind of rescue kit that we would provide to a new developer that starts to design an api. This kit must be accompanied with the reading of some useful resources link like the REST Guidelines. On a technical level, the fastapi documentation provides a lot of useful information as well, with a lot of examples. Last but not least, the pydantic documentation is also very useful.
The 'odoo-addon-fastapi' module provides 2 useful piece of code to help you be consistent when writing a route handler for a search route.
- A dependency method to use to specify the pagination parameters in the same way for all the search route handlers: 'odoo.addons.fastapi.paging'.
- A PagedCollection pydantic model to use to return the result of a search route handler enclosed in a json document that contains the count of records.
from typing import Annotated
from pydantic import BaseModel
from odoo.api import Environment
from odoo.addons.fastapi.dependencies import paging, authenticated_partner_env
from odoo.addons.fastapi.schemas import PagedCollection, Paging
class SaleOrder(BaseModel):
id: int
name: str
model_config = ConfigDict(from_attributes=True)
@router.get(
"/sale_orders",
response_model=PagedCollection[SaleOrder],
response_model_exclude_unset=True,
)
def get_sale_orders(
paging: Annotated[Paging, Depends(paging)],
env: Annotated[Environment, Depends(authenticated_partner_env)],
) -> PagedCollection[SaleOrder]:
"""Get the list of sale orders."""
count = env["sale.order"].search_count([])
orders = env["sale.order"].search([], limit=paging.limit, offset=paging.offset)
return PagedCollection[SaleOrder](
count=count,
items=[SaleOrder.model_validate(order) for order in orders],
)Note
The 'odoo.addons.fastapi.schemas.Paging' and 'odoo.addons.fastapi.schemas.PagedCollection' pydantic models are not designed to be extended to not introduce a dependency between the 'odoo-addon-fastapi' module and the 'odoo-addon-extendable'
The error handling a very important topic in the design of the fastapi integration with odoo. It must ensure that the error messages are properly return to the client and that the transaction is properly roll backed. The 'fastapi' module provides a way to register custom error handlers. The 'odoo.addons.fastapi.error_handlers' module provides the default error handlers that are registered by default when a new instance of the 'FastAPI' class is created. When an app is initialized in 'fastapi.endpoint' model, the method _get_app_exception_handlers is called to get a dictionary of error handlers. This method is designed to be overridden in a custom module to provide custom error handlers. You can override the handler for a specific exception class or you can add a new handler for a new exception or even replace all the handlers by your own handlers. Whatever you do, you must ensure that the transaction is properly roll backed.
Some could argue that the error handling can't be extended since the error handlers are global method not defined in an odoo model. Since the method providing the the error handlers definitions is defined on the 'fastapi.endpoint' model, it's not a problem at all, you just need to think another way to do it that by inheritance.
A solution could be to develop you own error handler to be able to process the error and chain the call to the default error handler.
class MyCustomErrorHandler():
def __init__(self, next_handler):
self.next_handler = next_handler
def __call__(self, request: Request, exc: Exception) -> JSONResponse:
# do something with the error
response = self.next_handler(request, exc)
# do something with the response
return responseWith this solution, you can now register your custom error handler by overriding the method _get_app_exception_handlers in your custom module.
class FastapiEndpoint(models.Model):
_inherit = "fastapi.endpoint"
def _get_app_exception_handlers(
self,
) -> Dict[
Union[int, Type[Exception]],
Callable[[Request, Exception], Union[Response, Awaitable[Response]]],
]:
handlers = super()._get_app_exception_handlers()
access_error_handler = handlers.get(odoo.exceptions.AccessError)
handlers[odoo.exceptions.AccessError] = MyCustomErrorHandler(access_error_handler)
return handlersIn the previous example, we extend the error handler for the 'AccessError' exception for all the endpoints. You can do the same for a specific app by checking the 'app' field of the 'fastapi.endpoint' record before registering your custom error handler.
When you develop a new addon to expose an api with fastapi, it's a good practice to follow the same directory structure and naming convention for the files related to the api. It will help you to easily find the files related to the api and it will help the other developers to understand your code.
Here is the directory structure that we recommend. It's based on practices that are used in the python community when developing a fastapi app.
. ├── x_api │ ├── data │ │ ├── ... .xml │ ├── demo │ │ ├── ... .xml │ ├── i18n │ │ ├── ... .po │ ├── models │ │ ├── __init__.py │ │ ├── fastapi_endpoint.py # your app │ │ └── ... .py │ └── routers │ │ ├── __init__.py │ │ ├── items.py │ │ └── ... .py │ ├── schemas | schemas.py │ │ ├── __init__.py │ │ ├── my_model.py # pydantic model │ │ └── ... .py │ ├── security │ │ ├── ... .xml │ ├── views │ │ ├── ... .xml │ ├── __init__.py │ ├── __manifest__.py │ ├── dependencies.py # custom dependencies │ ├── error_handlers.py # custom error handlers
The 'models' directory contains the odoo models. When you define a new app, as for the others addons, you will add your new model inheriting from the 'fastapi.endpoint' model in this directory.
The 'routers' directory contains the fastapi routers. You will add your new routers in this directory. Each route starting with the same prefix should be grouped in the same file. For example, all the routes starting with '/items' should be defined in the 'items.py' file. The '__init__.py' file in this directory is used to import all the routers defined in the directory and create a global router that can be used in an app. For example, in your 'items.py' file, you will define a router like this:
router = APIRouter(tags=["items"]) router.get("/items", response_model=List[Item]) def list_items(): pass
In the '__init__.py' file, you will import the router and add it to the global router or your addon.
from fastapi import APIRouter from .items import router as items_router router = APIRouter() router.include_router(items_router)
The 'schemas.py' will be used to define the pydantic models. For complex APIs with a lot of models, it will be better to create a 'schemas' directory and split the models in different files. The '__init__.py' file in this directory will be used to import all the models defined in the directory. For example, in your 'my_model.py' file, you will define a model like this:
from pydantic import BaseModel class MyModel(BaseModel): name: str description: str = None
In the '__init__.py' file, you will import the model's classes from the files in the directory.
from .my_model import MyModel
This will allow to always import the models from the schemas module whatever the models are spread across different files or defined in the 'schemas.py' file.
from x_api_addon.schemas import MyModel
The 'dependencies.py' file contains the custom dependencies that you will use in your routers. For example, you can define a dependency to check the access rights of the user.
The 'error_handlers.py' file contains the custom error handlers that you will use in your routers. The 'odoo-addon-fastapi' module provides the default error handlers for the common odoo exceptions. Chance are that you will not need to define your own error handlers. But if you need to do it, you can define them in this file.
The 'odoo-addon-fastapi' module is still in its early stage of development. It will evolve over time to integrate your feedback and to provide the missing features. It's now up to you to try it and to provide your feedback.
The roadmap and known issues can be found on GitHub.
The FastAPI module provides an easy way to use WebSockets. Unfortunately, this support is not 'yet' available in the Odoo framework. The challenge is high because the integration of the fastapi is based on the use of a specific middleware that convert the WSGI request consumed by odoo to a ASGI request. The question is to know if it is also possible to develop the same kind of bridge for the WebSockets and to stream large responses.
Bugfixes
In case of exception in endpoint execution, close the database cursor after rollback.
This is to ensure that the retrying method in service/model.py does not try to flush data to the database. (#405)
Bugfixes
- When using the 'FastAPITransactionCase' class, allows to specify a specific override of the 'authenticated_partner_impl' method into the list of overrides to apply. Before this change, the 'authenticated_partner_impl' override given in the 'overrides' parameter was always overridden in the '_create_test_client' method of the 'FastAPITransactionCase' class. It's now only overridden if the 'authenticated_partner_impl' method is not already present in the list of overrides to apply and no specific partner is given. If a specific partner is given at same time of an override for the 'authenticated_partner_impl' method, an error is raised. (#396)
Bugfixes
Fix a typo in the Field declaration of the 'count' attribute of the 'PagedCollection' schema.
Misspelt parameter was triggering a deprecation warning due to recent versions of Pydantic seeing it as an arbitrary parameter. (#389)
Features
- The field total in the PagedCollection schema is replaced by the field count. The field total is now deprecated and will be removed in the next major version. This change is backward compatible. The json document returned will now contain both fields total and count with the same value. In your python code the field total, if used, will fill the field count with the same value. You are encouraged to use the field count instead of total and adapt your code accordingly. (#380)
Bugs are tracked on GitHub Issues. In case of trouble, please check there if your issue has already been reported. If you spotted it first, help us to smash it by providing a detailed and welcomed feedback.
Do not contact contributors directly about support or help with technical issues.
- ACSONE SA/NV
- Laurent Mignon <[email protected]>
This module is maintained by the OCA.
OCA, or the Odoo Community Association, is a nonprofit organization whose mission is to support the collaborative development of Odoo features and promote its widespread use.
Current maintainer:
This module is part of the OCA/rest-framework project on GitHub.
You are welcome to contribute. To learn how please visit https://odoo-community.org/page/Contribute.