feat(ray): add help functions for ray model (#40)

Because

- provide support for constructing compatible `ray` model for `Instill
Model`

This commit

- add helper functions for constructing `ray` model

Resolves INS-2497
Resolves INS-2498

.pylint.ini

@@ -7,7 +7,7 @@ extension-pkg-whitelist=
 
 # Add files or directories to the blacklist. They should be base names, not
 # paths.
-ignore=CVS,protogen
+ignore=CVS,protogen,protobufs
 
 # Add files or directories matching the regex patterns to the blacklist. The
 # regex matches against base names, not paths.

instill/__init__.py

@@ -1,7 +1,17 @@
+# pylint: disable=no-name-in-module
 import os
 import sys
 from importlib.metadata import PackageNotFoundError, version
 
+from instill.helpers.protobufs.ray_pb2 import (
+    InferTensor,
+    ModelInferRequest,
+    ModelInferResponse,
+    ModelMetadataRequest,
+    ModelMetadataResponse,
+    ModelReadyRequest,
+    ModelReadyResponse,
+)
 from instill.utils.logger import Logger
 
 Logger.initialize()

instill/configuration/__init__.py

@@ -6,6 +6,9 @@
 import yaml
 from pydantic import BaseModel
 
+CLOUD_RAY_ADDRESS = "ray://core_ray_server:10001"
+CORE_RAY_ADDRESS = "ray://ray_server:10001"
+
 CONFIG_DIR = Path(
     os.getenv(
         "INSTILL_SYSTEM_CONFIG_PATH",

instill/helpers/protobufs/ray_pb2.pyi

@@ -0,0 +1,92 @@
+from google.protobuf.internal import containers as _containers
+from google.protobuf import descriptor as _descriptor
+from google.protobuf import message as _message
+from typing import ClassVar as _ClassVar, Iterable as _Iterable, Mapping as _Mapping, Optional as _Optional, Union as _Union
+
+DESCRIPTOR: _descriptor.FileDescriptor
+
+class ModelReadyRequest(_message.Message):
+    __slots__ = ["name", "version"]
+    NAME_FIELD_NUMBER: _ClassVar[int]
+    VERSION_FIELD_NUMBER: _ClassVar[int]
+    name: str
+    version: str
+    def __init__(self, name: _Optional[str] = ..., version: _Optional[str] = ...) -> None: ...
+
+class ModelReadyResponse(_message.Message):
+    __slots__ = ["ready"]
+    READY_FIELD_NUMBER: _ClassVar[int]
+    ready: bool
+    def __init__(self, ready: bool = ...) -> None: ...
+
+class ModelMetadataRequest(_message.Message):
+    __slots__ = ["name", "version"]
+    NAME_FIELD_NUMBER: _ClassVar[int]
+    VERSION_FIELD_NUMBER: _ClassVar[int]
+    name: str
+    version: str
+    def __init__(self, name: _Optional[str] = ..., version: _Optional[str] = ...) -> None: ...
+
+class InferTensor(_message.Message):
+    __slots__ = ["name", "datatype", "shape"]
+    NAME_FIELD_NUMBER: _ClassVar[int]
+    DATATYPE_FIELD_NUMBER: _ClassVar[int]
+    SHAPE_FIELD_NUMBER: _ClassVar[int]
+    name: str
+    datatype: str
+    shape: _containers.RepeatedScalarFieldContainer[int]
+    def __init__(self, name: _Optional[str] = ..., datatype: _Optional[str] = ..., shape: _Optional[_Iterable[int]] = ...) -> None: ...
+
+class ModelMetadataResponse(_message.Message):
+    __slots__ = ["name", "versions", "framework", "inputs", "outputs"]
+    class TensorMetadata(_message.Message):
+        __slots__ = ["name", "datatype", "shape"]
+        NAME_FIELD_NUMBER: _ClassVar[int]
+        DATATYPE_FIELD_NUMBER: _ClassVar[int]
+        SHAPE_FIELD_NUMBER: _ClassVar[int]
+        name: str
+        datatype: str
+        shape: _containers.RepeatedScalarFieldContainer[int]
+        def __init__(self, name: _Optional[str] = ..., datatype: _Optional[str] = ..., shape: _Optional[_Iterable[int]] = ...) -> None: ...
+    NAME_FIELD_NUMBER: _ClassVar[int]
+    VERSIONS_FIELD_NUMBER: _ClassVar[int]
+    FRAMEWORK_FIELD_NUMBER: _ClassVar[int]
+    INPUTS_FIELD_NUMBER: _ClassVar[int]
+    OUTPUTS_FIELD_NUMBER: _ClassVar[int]
+    name: str
+    versions: _containers.RepeatedScalarFieldContainer[str]
+    framework: str
+    inputs: _containers.RepeatedCompositeFieldContainer[ModelMetadataResponse.TensorMetadata]
+    outputs: _containers.RepeatedCompositeFieldContainer[ModelMetadataResponse.TensorMetadata]
+    def __init__(self, name: _Optional[str] = ..., versions: _Optional[_Iterable[str]] = ..., framework: _Optional[str] = ..., inputs: _Optional[_Iterable[_Union[ModelMetadataResponse.TensorMetadata, _Mapping]]] = ..., outputs: _Optional[_Iterable[_Union[ModelMetadataResponse.TensorMetadata, _Mapping]]] = ...) -> None: ...
+
+class ModelInferRequest(_message.Message):
+    __slots__ = ["model_name", "model_version", "inputs", "outputs", "raw_input_contents"]
+    class InferRequestedOutputTensor(_message.Message):
+        __slots__ = ["name"]
+        NAME_FIELD_NUMBER: _ClassVar[int]
+        name: str
+        def __init__(self, name: _Optional[str] = ...) -> None: ...
+    MODEL_NAME_FIELD_NUMBER: _ClassVar[int]
+    MODEL_VERSION_FIELD_NUMBER: _ClassVar[int]
+    INPUTS_FIELD_NUMBER: _ClassVar[int]
+    OUTPUTS_FIELD_NUMBER: _ClassVar[int]
+    RAW_INPUT_CONTENTS_FIELD_NUMBER: _ClassVar[int]
+    model_name: str
+    model_version: str
+    inputs: _containers.RepeatedCompositeFieldContainer[InferTensor]
+    outputs: _containers.RepeatedCompositeFieldContainer[ModelInferRequest.InferRequestedOutputTensor]
+    raw_input_contents: _containers.RepeatedScalarFieldContainer[bytes]
+    def __init__(self, model_name: _Optional[str] = ..., model_version: _Optional[str] = ..., inputs: _Optional[_Iterable[_Union[InferTensor, _Mapping]]] = ..., outputs: _Optional[_Iterable[_Union[ModelInferRequest.InferRequestedOutputTensor, _Mapping]]] = ..., raw_input_contents: _Optional[_Iterable[bytes]] = ...) -> None: ...
+
+class ModelInferResponse(_message.Message):
+    __slots__ = ["model_name", "model_version", "outputs", "raw_output_contents"]
+    MODEL_NAME_FIELD_NUMBER: _ClassVar[int]
+    MODEL_VERSION_FIELD_NUMBER: _ClassVar[int]
+    OUTPUTS_FIELD_NUMBER: _ClassVar[int]
+    RAW_OUTPUT_CONTENTS_FIELD_NUMBER: _ClassVar[int]
+    model_name: str
+    model_version: str
+    outputs: _containers.RepeatedCompositeFieldContainer[InferTensor]
+    raw_output_contents: _containers.RepeatedScalarFieldContainer[bytes]
+    def __init__(self, model_name: _Optional[str] = ..., model_version: _Optional[str] = ..., outputs: _Optional[_Iterable[_Union[InferTensor, _Mapping]]] = ..., raw_output_contents: _Optional[_Iterable[bytes]] = ...) -> None: ...

instill/helpers/protobufs/ray_pb2_grpc.py

@@ -0,0 +1,141 @@
+# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
+"""Client and server classes corresponding to protobuf-defined services."""
+import grpc
+
+import ray_pb2 as ray__pb2
+
+
+class RayServiceStub(object):
+    """Ray service for internal process
+    """
+
+    def __init__(self, channel):
+        """Constructor.
+
+        Args:
+            channel: A grpc.Channel.
+        """
+        self.ModelReady = channel.unary_unary(
+                '/ray.serve.RayService/ModelReady',
+                request_serializer=ray__pb2.ModelReadyRequest.SerializeToString,
+                response_deserializer=ray__pb2.ModelReadyResponse.FromString,
+                )
+        self.ModelMetadata = channel.unary_unary(
+                '/ray.serve.RayService/ModelMetadata',
+                request_serializer=ray__pb2.ModelMetadataRequest.SerializeToString,
+                response_deserializer=ray__pb2.ModelMetadataResponse.FromString,
+                )
+        self.ModelInfer = channel.unary_unary(
+                '/ray.serve.RayService/ModelInfer',
+                request_serializer=ray__pb2.ModelInferRequest.SerializeToString,
+                response_deserializer=ray__pb2.ModelInferResponse.FromString,
+                )
+
+
+class RayServiceServicer(object):
+    """Ray service for internal process
+    """
+
+    def ModelReady(self, request, context):
+        """ModelReady method receives a ModelReadyRequest message and
+        returns a ModelReadyResponse
+        """
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def ModelMetadata(self, request, context):
+        """ModelMetadata method receives a ModelMetadataRequest message and
+        returns a ModelMetadataResponse
+        """
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+    def ModelInfer(self, request, context):
+        """ModelInfer method receives a ModelInferRequest message and
+        returns a ModelInferResponse
+        """
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
+
+def add_RayServiceServicer_to_server(servicer, server):
+    rpc_method_handlers = {
+            'ModelReady': grpc.unary_unary_rpc_method_handler(
+                    servicer.ModelReady,
+                    request_deserializer=ray__pb2.ModelReadyRequest.FromString,
+                    response_serializer=ray__pb2.ModelReadyResponse.SerializeToString,
+            ),
+            'ModelMetadata': grpc.unary_unary_rpc_method_handler(
+                    servicer.ModelMetadata,
+                    request_deserializer=ray__pb2.ModelMetadataRequest.FromString,
+                    response_serializer=ray__pb2.ModelMetadataResponse.SerializeToString,
+            ),
+            'ModelInfer': grpc.unary_unary_rpc_method_handler(
+                    servicer.ModelInfer,
+                    request_deserializer=ray__pb2.ModelInferRequest.FromString,
+                    response_serializer=ray__pb2.ModelInferResponse.SerializeToString,
+            ),
+    }
+    generic_handler = grpc.method_handlers_generic_handler(
+            'ray.serve.RayService', rpc_method_handlers)
+    server.add_generic_rpc_handlers((generic_handler,))
+
+
+ # This class is part of an EXPERIMENTAL API.
+class RayService(object):
+    """Ray service for internal process
+    """
+
+    @staticmethod
+    def ModelReady(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/ray.serve.RayService/ModelReady',
+            ray__pb2.ModelReadyRequest.SerializeToString,
+            ray__pb2.ModelReadyResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def ModelMetadata(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/ray.serve.RayService/ModelMetadata',
+            ray__pb2.ModelMetadataRequest.SerializeToString,
+            ray__pb2.ModelMetadataResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def ModelInfer(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/ray.serve.RayService/ModelInfer',
+            ray__pb2.ModelInferRequest.SerializeToString,
+            ray__pb2.ModelInferResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)