Skip to content

TensorShareServer

The TensorShareServer schema defines the server configuration that will receive the tensors from any client.

It will be used by the TensorShareClient for sending tensors over the network.

Schema

The schema inherits from the pydantic.BaseModel class and has the following properties:

  • url: The URL of the server. This is the URL that the client will use to connect to the server. It should be a valid URL, including the protocol (e.g. http://localhost:5000 or https://example.com).

  • ping: The path to the health-check endpoint. This is the path that the client will use to check if the server is available. It should be a valid string, without the leading slash (e.g. ping or healthcheck). It will be ping by default.

  • receive_tensor: The path to the endpoint receiving the tensor. This is the path that the client will use to send the tensor to the server. It should be a valid string without the leading slash (e.g. receive_tensor or tensors). It will be receive_tensor by default.

  • response_model: The response model that the server will return to the client. It should be a subclass of BaseModel from Pydantic. It will be DefaultResponse by default, a simple model returning a message string.

Create a config - from_dict

The from_dict method creates a TensorShareServer config from a dictionary.

from tensorshare import DefaultResponse, TensorShareServer

config = TensorShareServer.from_dict(
    server_config={
        "url": "http://localhost:5000",      # required
        "ping": "ping",                      # optional
        "receive_tensor": "receive_tensor",  # optional
        "response_model": DefaultResponse,   # optional
    }
)
print(config)
# url=Url('http://localhost:5000/')
# ping=Url('http://localhost:5000/ping')
# receive_tensor=Url('http://localhost:5000/receive_tensor')
# response_model=<class 'tensorshare.schema.DefaultResponse'>

The url, ping and receive_tensor properties leverage the HttpUrl type from Pydantic. This means that they will be validated and normalized at runtime.

Use the config

Like any pydantic model, you can use the config as a standard Python class object.

config.url
>>> Url('http://localhost:5000/')

config.ping
>>> Url('http://localhost:5000/ping')

config.receive_tensor
>>> Url('http://localhost:5000/receive_tensor')

config.response_model
>>> <class 'tensorshare.schema.DefaultResponse'>

You can also get the URL as a string or some parts of it.

str(config.url)
>>> 'http://localhost:5000/'

config.url.scheme
>>> 'http'

config.url.host
>>> 'localhost'

config.url.port
>>> 5000

config.url.path
>>> '/'
Last update: 2023-08-20
Created: 2023-08-20