A RPC interface using aioredis and pydantic.
pip install aioredis-rpc
pydantic
is used to model complex objects which are transparently serialized
and packed into messages using msgpack
.
# Define Pydantic models
class FileData(BaseModel):
filename: str
data: bytes
Define a class using the @endpoint
decorator to specify which methods will be
accessible via rpc.
from redisrpc import endpoint
# Define an RPC class
class Dropbox:
files: Dict[str, FileData]
max_files: int
def __init__(self, max_files: int = 1000):
self.files = dict()
self.max_files = max_files
@endpoint
async def upload_file(self, file: FileData) -> int:
if len(self.files) >= self.max_files:
# Errors are propagated to the client-side
raise Exception('too many files')
self.files[file.name] = file
return len(file.data)
@endpoint
def get_file_names(self) -> List[str]:
return list(self.files.keys())
@endpoint
async def download_file(self, name: str) -> FileData:
return self.files[name]
Use the create_server
function to make an instance of your server-side rpc
class. The server instance will be assigned an rpc
attribute to access server
functions like connect
and disconnect
. Once connect
is called methods
decorated with @endpoint
will be invoked automatically by remote calls from
the client.
NOTE: The
RpcProvider.connect
method is non-blocking.
server = create_server(Dropbox, max_files=2)
# Returns once connected to redis
await server.rpc.connect(dsn="redis:https://localhost")
# Wait forever
while True:
await asyncio.sleep(1)
The create_client
function create a faux instance of the rpc class with only
the methods decorated by @endpoint
present. When these methods are called by
the client the function arguments are serialized and published to redis.
NOTE: If there are no subscribers to the redis channel then the client will throw a
RpcNotConnectedError
.
client = create_client(Dropbox)
await client.rpc.connect(dsn="redis:https://localhost")
Now that both ends are connected the @endpoint
decorated methods may be called
like they are accessing the actual class passed to create_client
.
file1 = FileData(name='file1', data=b'1234')
size = await client.upload_file(file1)