fix: persistent styling between ruff and yapf (#279)

bentoml · Aug 30, 2023 · c9cef1d · c9cef1d
1 parent f678f71
commit c9cef1d
Show file tree

Hide file tree

Showing 145 changed files with 1,049 additions and 393 deletions.
diff --git a/cz.py b/cz.py
@@ -6,6 +6,7 @@
 import tokenize
 
 from tabulate import tabulate
+
 TOKEN_WHITELIST = [token.OP, token.NAME, token.NUMBER, token.STRING]
 
 def run_cz(dir: str, package: str):

diff --git a/examples/bentoml-demo/service.py b/examples/bentoml-demo/service.py
@@ -2,6 +2,7 @@
 
 import bentoml
 import openllm
+
 model = "dolly-v2"
 
 llm_config = openllm.AutoConfig.for_model(model)

diff --git a/examples/langchain-tools-demo/service.py b/examples/langchain-tools-demo/service.py
@@ -5,6 +5,7 @@
 
 import bentoml
 from bentoml.io import Text
+
 SAMPLE_INPUT = "What is the weather in San Francisco?"
 
 llm = OpenLLM(model_name="dolly-v2", model_id="databricks/dolly-v2-7b", embedded=False,)

diff --git a/openllm-client/src/openllm_client/__init__.py b/openllm-client/src/openllm_client/__init__.py
@@ -1,5 +1,9 @@
 from __future__ import annotations
 
 from . import benmin as benmin
-from ._base import BaseAsyncClient as BaseAsyncClient, BaseClient as BaseClient
-from .client import AsyncGrpcClient as AsyncGrpcClient, AsyncHTTPClient as AsyncHTTPClient, GrpcClient as GrpcClient, HTTPClient as HTTPClient
+from ._base import BaseAsyncClient as BaseAsyncClient
+from ._base import BaseClient as BaseClient
+from .client import AsyncGrpcClient as AsyncGrpcClient
+from .client import AsyncHTTPClient as AsyncHTTPClient
+from .client import GrpcClient as GrpcClient
+from .client import HTTPClient as HTTPClient
diff --git a/openllm-client/src/openllm_client/_base.py b/openllm-client/src/openllm_client/_base.py
@@ -4,6 +4,7 @@
 import functools
 import logging
 import typing as t
+
 from http import HTTPStatus
 from urllib.parse import urljoin
 
@@ -12,14 +13,23 @@
 import orjson
 
 import openllm_core
-from openllm_core._typing_compat import LiteralString, overload
-from openllm_core.utils import bentoml_cattr, ensure_exec_coro, is_transformers_available, is_transformers_supports_agent
 
-from .benmin import AsyncClient as AsyncBentoClient, Client as BentoClient
+from openllm_core._typing_compat import LiteralString
+from openllm_core._typing_compat import overload
+from openllm_core.utils import bentoml_cattr
+from openllm_core.utils import ensure_exec_coro
+from openllm_core.utils import is_transformers_available
+from openllm_core.utils import is_transformers_supports_agent
+
+from .benmin import AsyncClient as AsyncBentoClient
+from .benmin import Client as BentoClient
+
 if t.TYPE_CHECKING:
   import transformers
 
-  from openllm_core._typing_compat import DictStrAny, LiteralRuntime
+  from openllm_core._typing_compat import DictStrAny
+  from openllm_core._typing_compat import LiteralRuntime
+
 logger = logging.getLogger(__name__)
 
 @attr.define(slots=False, init=False)
@@ -204,7 +214,9 @@ async def ask_agent(self, task: str, *, return_code: bool = False, remote: bool
   async def _run_hf_agent(self, *args: t.Any, **kwargs: t.Any) -> t.Any:
     if not is_transformers_supports_agent(): raise RuntimeError('This version of transformers does not support agent.run. Make sure to upgrade to transformers>4.30.0')
     if len(args) > 1: raise ValueError("'args' should only take one positional argument.")
-    from transformers.tools.agents import clean_code_for_run, get_tool_creation_code, resolve_tools
+    from transformers.tools.agents import clean_code_for_run
+    from transformers.tools.agents import get_tool_creation_code
+    from transformers.tools.agents import resolve_tools
     from transformers.tools.python_interpreter import evaluate
 
     task = kwargs.pop('task', args[0])

diff --git a/openllm-client/src/openllm_client/benmin/__init__.py b/openllm-client/src/openllm_client/benmin/__init__.py
@@ -13,13 +13,16 @@
 """
 from __future__ import annotations
 import typing as t
+
 from abc import abstractmethod
 
 import attr
 import httpx
 
 import bentoml
-if t.TYPE_CHECKING: from bentoml._internal.service.inference_api import InferenceAPI
+
+if t.TYPE_CHECKING:
+  from bentoml._internal.service.inference_api import InferenceAPI
 
 __all__ = ['Client', 'AsyncClient']
 

diff --git a/openllm-client/src/openllm_client/benmin/_grpc.py b/openllm-client/src/openllm_client/benmin/_grpc.py
@@ -6,21 +6,31 @@
 import typing as t
 
 import bentoml
+
 from bentoml._internal.service.inference_api import InferenceAPI
-from bentoml.grpc.utils import import_generated_stubs, load_from_file
-from openllm_client.benmin import AsyncClient, Client
-from openllm_core._typing_compat import NotRequired, overload
-from openllm_core.utils import ensure_exec_coro, is_grpc_available, is_grpc_health_available
+from bentoml.grpc.utils import import_generated_stubs
+from bentoml.grpc.utils import load_from_file
+from openllm_client.benmin import AsyncClient
+from openllm_client.benmin import Client
+from openllm_core._typing_compat import NotRequired
+from openllm_core._typing_compat import overload
+from openllm_core.utils import ensure_exec_coro
+from openllm_core.utils import is_grpc_available
+from openllm_core.utils import is_grpc_health_available
+
 if not is_grpc_available() or not is_grpc_health_available(): raise ImportError("gRPC is required to use gRPC client. Install with 'pip install \"openllm-client[grpc]\"'.")
 import grpc
 import grpc_health.v1.health_pb2 as pb_health
 import grpc_health.v1.health_pb2_grpc as services_health
+
 from google.protobuf import json_format
 from grpc import aio
+
 pb, services = import_generated_stubs('v1')
 
 if t.TYPE_CHECKING:
   from bentoml.grpc.v1.service_pb2 import ServiceMetadataResponse
+
 logger = logging.getLogger(__name__)
 
 class ClientCredentials(t.TypedDict):

diff --git a/openllm-client/src/openllm_client/benmin/_http.py b/openllm-client/src/openllm_client/benmin/_http.py
@@ -5,6 +5,7 @@
 import time
 import typing as t
 import urllib.error
+
 from urllib.parse import urlparse
 
 import httpx
@@ -14,9 +15,12 @@
 import starlette.responses
 
 import bentoml
+
 from bentoml._internal.service.inference_api import InferenceAPI
-from openllm_client.benmin import AsyncClient, Client
+from openllm_client.benmin import AsyncClient
+from openllm_client.benmin import Client
 from openllm_core.utils import ensure_exec_coro
+
 logger = logging.getLogger(__name__)
 
 class HttpClient(Client):

diff --git a/openllm-client/src/openllm_client/client.py b/openllm-client/src/openllm_client/client.py
@@ -1,8 +1,11 @@
 from __future__ import annotations
 import logging
+
 from urllib.parse import urlparse
 
-from ._base import BaseAsyncClient, BaseClient
+from ._base import BaseAsyncClient
+from ._base import BaseClient
+
 logger = logging.getLogger(__name__)
 
 def process_http_address(self: AsyncHTTPClient | HTTPClient, address: str) -> None:

diff --git a/openllm-core/src/openllm_core/__init__.py b/openllm-core/src/openllm_core/__init__.py
@@ -1,33 +1,44 @@
 from __future__ import annotations
 
-from . import exceptions as exceptions, utils as utils
-from ._configuration import GenerationConfig as GenerationConfig, LLMConfig as LLMConfig, SamplingParams as SamplingParams
-from ._schema import EmbeddingsOutput as EmbeddingsOutput, GenerationInput as GenerationInput, GenerationOutput as GenerationOutput, HfAgentInput as HfAgentInput, MetadataOutput as MetadataOutput, unmarshal_vllm_outputs as unmarshal_vllm_outputs
-from ._strategies import AmdGpuResource as AmdGpuResource, CascadingResourceStrategy as CascadingResourceStrategy, LiteralResourceSpec as LiteralResourceSpec, NvidiaGpuResource as NvidiaGpuResource, available_resource_spec as available_resource_spec, get_resource as get_resource
-from .config import (
-    CONFIG_MAPPING as CONFIG_MAPPING,
-    CONFIG_MAPPING_NAMES as CONFIG_MAPPING_NAMES,
-    START_BAICHUAN_COMMAND_DOCSTRING as START_BAICHUAN_COMMAND_DOCSTRING,
-    START_CHATGLM_COMMAND_DOCSTRING as START_CHATGLM_COMMAND_DOCSTRING,
-    START_DOLLY_V2_COMMAND_DOCSTRING as START_DOLLY_V2_COMMAND_DOCSTRING,
-    START_FALCON_COMMAND_DOCSTRING as START_FALCON_COMMAND_DOCSTRING,
-    START_FLAN_T5_COMMAND_DOCSTRING as START_FLAN_T5_COMMAND_DOCSTRING,
-    START_GPT_NEOX_COMMAND_DOCSTRING as START_GPT_NEOX_COMMAND_DOCSTRING,
-    START_LLAMA_COMMAND_DOCSTRING as START_LLAMA_COMMAND_DOCSTRING,
-    START_MPT_COMMAND_DOCSTRING as START_MPT_COMMAND_DOCSTRING,
-    START_OPT_COMMAND_DOCSTRING as START_OPT_COMMAND_DOCSTRING,
-    START_STABLELM_COMMAND_DOCSTRING as START_STABLELM_COMMAND_DOCSTRING,
-    START_STARCODER_COMMAND_DOCSTRING as START_STARCODER_COMMAND_DOCSTRING,
-    AutoConfig as AutoConfig,
-    BaichuanConfig as BaichuanConfig,
-    ChatGLMConfig as ChatGLMConfig,
-    DollyV2Config as DollyV2Config,
-    FalconConfig as FalconConfig,
-    FlanT5Config as FlanT5Config,
-    GPTNeoXConfig as GPTNeoXConfig,
-    LlamaConfig as LlamaConfig,
-    MPTConfig as MPTConfig,
-    OPTConfig as OPTConfig,
-    StableLMConfig as StableLMConfig,
-    StarCoderConfig as StarCoderConfig,
-)
+from . import exceptions as exceptions
+from . import utils as utils
+from ._configuration import GenerationConfig as GenerationConfig
+from ._configuration import LLMConfig as LLMConfig
+from ._configuration import SamplingParams as SamplingParams
+from ._schema import EmbeddingsOutput as EmbeddingsOutput
+from ._schema import GenerationInput as GenerationInput
+from ._schema import GenerationOutput as GenerationOutput
+from ._schema import HfAgentInput as HfAgentInput
+from ._schema import MetadataOutput as MetadataOutput
+from ._schema import unmarshal_vllm_outputs as unmarshal_vllm_outputs
+from ._strategies import AmdGpuResource as AmdGpuResource
+from ._strategies import CascadingResourceStrategy as CascadingResourceStrategy
+from ._strategies import LiteralResourceSpec as LiteralResourceSpec
+from ._strategies import NvidiaGpuResource as NvidiaGpuResource
+from ._strategies import available_resource_spec as available_resource_spec
+from ._strategies import get_resource as get_resource
+from .config import CONFIG_MAPPING as CONFIG_MAPPING
+from .config import CONFIG_MAPPING_NAMES as CONFIG_MAPPING_NAMES
+from .config import START_BAICHUAN_COMMAND_DOCSTRING as START_BAICHUAN_COMMAND_DOCSTRING
+from .config import START_CHATGLM_COMMAND_DOCSTRING as START_CHATGLM_COMMAND_DOCSTRING
+from .config import START_DOLLY_V2_COMMAND_DOCSTRING as START_DOLLY_V2_COMMAND_DOCSTRING
+from .config import START_FALCON_COMMAND_DOCSTRING as START_FALCON_COMMAND_DOCSTRING
+from .config import START_FLAN_T5_COMMAND_DOCSTRING as START_FLAN_T5_COMMAND_DOCSTRING
+from .config import START_GPT_NEOX_COMMAND_DOCSTRING as START_GPT_NEOX_COMMAND_DOCSTRING
+from .config import START_LLAMA_COMMAND_DOCSTRING as START_LLAMA_COMMAND_DOCSTRING
+from .config import START_MPT_COMMAND_DOCSTRING as START_MPT_COMMAND_DOCSTRING
+from .config import START_OPT_COMMAND_DOCSTRING as START_OPT_COMMAND_DOCSTRING
+from .config import START_STABLELM_COMMAND_DOCSTRING as START_STABLELM_COMMAND_DOCSTRING
+from .config import START_STARCODER_COMMAND_DOCSTRING as START_STARCODER_COMMAND_DOCSTRING
+from .config import AutoConfig as AutoConfig
+from .config import BaichuanConfig as BaichuanConfig
+from .config import ChatGLMConfig as ChatGLMConfig
+from .config import DollyV2Config as DollyV2Config
+from .config import FalconConfig as FalconConfig
+from .config import FlanT5Config as FlanT5Config
+from .config import GPTNeoXConfig as GPTNeoXConfig
+from .config import LlamaConfig as LlamaConfig
+from .config import MPTConfig as MPTConfig
+from .config import OPTConfig as OPTConfig
+from .config import StableLMConfig as StableLMConfig
+from .config import StarCoderConfig as StarCoderConfig
diff --git a/openllm-core/src/openllm_core/_configuration.py b/openllm-core/src/openllm_core/_configuration.py
@@ -47,24 +47,50 @@ class GenerationConfig:
 import inflection
 import orjson
 
-# NOTE: Using internal API from attr here, since we are actually allowing subclass of openllm_core.LLMConfig to become 'attrs'-ish
 from attr._compat import set_closure_cell
-from attr._make import _CountingAttr, _make_init, _transform_attrs
-from cattr.gen import make_dict_structure_fn, make_dict_unstructure_fn, override
+from attr._make import _CountingAttr
+from attr._make import _make_init
+from attr._make import _transform_attrs
+from cattr.gen import make_dict_structure_fn
+from cattr.gen import make_dict_unstructure_fn
+from cattr.gen import override
 from deepmerge.merger import Merger
 
 import openllm_core
 
-from ._strategies import LiteralResourceSpec, available_resource_spec, resource_spec
-from ._typing_compat import AdapterType, AnyCallable, At, DictStrAny, ListStr, LiteralRuntime, LiteralString, NotRequired, Required, Self, overload
+from ._strategies import LiteralResourceSpec
+from ._strategies import available_resource_spec
+from ._strategies import resource_spec
+from ._typing_compat import AdapterType
+from ._typing_compat import AnyCallable
+from ._typing_compat import At
+from ._typing_compat import DictStrAny
+from ._typing_compat import ListStr
+from ._typing_compat import LiteralRuntime
+from ._typing_compat import LiteralString
+from ._typing_compat import NotRequired
+from ._typing_compat import Required
+from ._typing_compat import Self
+from ._typing_compat import overload
 from .exceptions import ForbiddenAttributeError
-from .utils import ENV_VARS_TRUE_VALUES, MYPY, LazyLoader, ReprMixin, bentoml_cattr, codegen, dantic, field_env_key, first_not_none, lenient_issubclass
+from .utils import ENV_VARS_TRUE_VALUES
+from .utils import MYPY
+from .utils import LazyLoader
+from .utils import ReprMixin
+from .utils import bentoml_cattr
+from .utils import codegen
+from .utils import dantic
+from .utils import field_env_key
+from .utils import first_not_none
+from .utils import lenient_issubclass
 from .utils.import_utils import BACKENDS_MAPPING
+
 if t.TYPE_CHECKING:
   import click
   import peft
   import transformers
   import vllm
+
   from transformers.generation.beam_constraints import Constraint
 else:
   Constraint = t.Any
@@ -781,6 +807,7 @@ def __attrs_init__(self, *args: t.Any, **attrs: t.Any) -> None:
     '''Optional tokenizer class for this given LLM. See Llama for example.'''
 
     # update-config-stubs.py: special stop
+
 class _ConfigBuilder:
   """A modified version of attrs internal _ClassBuilder, and should only be called within __init_subclass__ of LLMConfig.
 

diff --git a/openllm-core/src/openllm_core/_schema.py b/openllm-core/src/openllm_core/_schema.py
@@ -6,10 +6,13 @@
 import attr
 import inflection
 
-from openllm_core._configuration import GenerationConfig, LLMConfig
+from openllm_core._configuration import GenerationConfig
+from openllm_core._configuration import LLMConfig
 
 from .utils import bentoml_cattr
-if t.TYPE_CHECKING: import vllm
+
+if t.TYPE_CHECKING:
+  import vllm
 
 @attr.frozen(slots=True)
 class GenerationInput:

diff --git a/openllm-core/src/openllm_core/_strategies.py b/openllm-core/src/openllm_core/_strategies.py
@@ -13,11 +13,14 @@
 import psutil
 
 import bentoml
-from bentoml._internal.resource import get_resource, system_resources
+
+from bentoml._internal.resource import get_resource
+from bentoml._internal.resource import system_resources
 from bentoml._internal.runner.strategy import THREAD_ENVS
 
 from ._typing_compat import overload
-from .utils import DEBUG, ReprMixin
+from .utils import DEBUG
+from .utils import ReprMixin
 
 class DynResource(t.Protocol):
   resource_id: t.ClassVar[str]
@@ -98,10 +101,12 @@ def _from_system(cls: type[DynResource]) -> list[str]:
       # we don't want to use CLI because parsing is a pain.
       sys.path.append('/opt/rocm/libexec/rocm_smi')
       try:
-        from ctypes import byref, c_uint32
+        from ctypes import byref
+        from ctypes import c_uint32
 
         # refers to https://github.com/RadeonOpenCompute/rocm_smi_lib/blob/master/python_smi_tools/rsmiBindings.py
-        from rsmiBindings import rocmsmi, rsmi_status_t
+        from rsmiBindings import rocmsmi
+        from rsmiBindings import rsmi_status_t
 
         device_count = c_uint32(0)
         ret = rocmsmi.rsmi_num_monitor_devices(byref(device_count))
@@ -149,7 +154,11 @@ def _from_spec(cls: type[DynResource], spec: t.Any) -> list[str]:
     raise TypeError(f"'{cls.__name__}.from_spec' only supports parsing spec of type int, str, or list, got '{type(spec)}' instead.")
 
 def _raw_device_uuid_nvml() -> list[str] | None:
-  from ctypes import CDLL, byref, c_int, c_void_p, create_string_buffer
+  from ctypes import CDLL
+  from ctypes import byref
+  from ctypes import c_int
+  from ctypes import c_void_p
+  from ctypes import create_string_buffer
 
   try:
     nvml_h = CDLL('libnvidia-ml.so.1')