arkruntime 0.1.0__py3-none-any.whl

arkruntime/resources/batch/__init__.py

@@ -0,0 +1,6 @@
+# Copyright (c) 2026 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+
+from .batch import AsyncBatch, Batch
+
+__all__ = ["Batch", "AsyncBatch"]

arkruntime/resources/batch/_utils.py

@@ -0,0 +1,156 @@
+# Copyright (c) 2026 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+
+import asyncio
+import logging
+import time
+from datetime import datetime, timedelta
+from random import random
+from typing import Any, Awaitable, Callable, Optional, TypeVar
+
+import httpx
+
+from ..._constants import INITIAL_RETRY_DELAY, MAX_RETRY_DELAY
+from ..._exceptions import ArkAPIConnectionError, ArkAPIStatusError, ArkAPITimeoutError
+from ..._utils._model_breaker import ModelBreaker
+
+log: logging.Logger = logging.getLogger(__name__)
+
+
+def _calculate_retry_timeout(retry_times: int) -> float:
+    nbRetries = min(retry_times, MAX_RETRY_DELAY / INITIAL_RETRY_DELAY)
+    sleep_seconds = min(INITIAL_RETRY_DELAY * pow(2, nbRetries), MAX_RETRY_DELAY)
+    # Apply some jitter, plus-or-minus half a second.
+    jitter = 1 - 0.25 * random()
+    timeout = sleep_seconds * jitter
+    return timeout if timeout >= 0 else 0
+
+
+def _get_retry_after(response: httpx.Response) -> Optional[int]:
+    retry_after = response.headers.get("Retry-After")
+    if retry_after is not None:
+        if retry_after.isdigit():
+            return int(retry_after)
+    return None
+
+
+def _should_retry(response: httpx.Response) -> bool:
+    # Retry on request timeouts.
+    if response.status_code == 408:
+        return True
+
+    # Retry on lock timeouts.
+    if response.status_code == 409:
+        return True
+
+    # Retry on rate limits.
+    if response.status_code == 429:
+        return True
+
+    # Retry internal errors.
+    if response.status_code >= 500:
+        return True
+
+    return False
+
+
+def get_request_last_time(client: httpx.Client, timeout: Optional[Any] = None) -> datetime:
+    if timeout is None:
+        timeout = client.timeout
+    timeoutSeconds = 0
+    if isinstance(timeout, httpx.Timeout):
+        timeoutSeconds = timeout.read
+    elif isinstance(timeout, float):
+        timeoutSeconds = timeout
+    elif isinstance(timeout, int):
+        timeoutSeconds = timeout
+    else:
+        raise TypeError("timeout type {} is not supported".format(type(timeout)))
+    return datetime.now() + timedelta(seconds=timeoutSeconds)
+
+
+R = TypeVar("R")
+
+
+def with_batch_retry(
+    deadline: datetime,
+    breaker: ModelBreaker,
+    func: Callable[..., R],
+    *args,
+    **kwargs,
+) -> R:
+    retry_times = 0
+    while True:
+        breaker.wait()
+        if datetime.now() > deadline:
+            raise ArkAPITimeoutError(None, None)
+        try:
+            return func(*args, **kwargs)
+        except ArkAPIConnectionError:
+            waitTime = _calculate_retry_timeout(retry_times)
+            log.debug(
+                "Retry due to connection error, wait time: %is, retry times: %i",
+                waitTime,
+                retry_times,
+            )
+
+            if datetime.now() + timedelta(seconds=waitTime) > deadline:
+                raise ArkAPITimeoutError(None, None)
+            time.sleep(waitTime)
+        except ArkAPIStatusError as err:
+            retry_after = _get_retry_after(err.response)
+            log.debug(
+                "Got status error, retry after: %is, retry times: %i, error: %s",
+                retry_after,
+                retry_times,
+                err,
+            )
+
+            if retry_after is not None and retry_after > 0:
+                breaker.reset(retry_after)
+            if not _should_retry(err.response):
+                raise err
+
+        retry_times = retry_times + 1
+
+
+async def async_with_batch_retry(
+    deadline: datetime,
+    breaker: ModelBreaker,
+    func: Callable[..., Awaitable[R]],
+    *args,
+    **kwargs,
+) -> R:
+    retry_times = 0
+    while True:
+        await breaker.asyncwait()
+        if datetime.now() > deadline:
+            raise ArkAPITimeoutError(None, None)
+        try:
+            return await func(*args, **kwargs)
+        except ArkAPIConnectionError:
+            waitTime = _calculate_retry_timeout(retry_times)
+            log.debug(
+                "Retry due to connection error, wait time: %is, retry times: %i",
+                waitTime,
+                retry_times,
+            )
+
+            if datetime.now() + timedelta(seconds=waitTime) > deadline:
+                raise ArkAPITimeoutError(None, None)
+            await asyncio.sleep(waitTime)
+        except ArkAPIStatusError as err:
+            retry_after = _get_retry_after(err.response)
+            log.debug(
+                "Got status error, retry after: %is, retry times: %i, error: %s",
+                retry_after,
+                retry_times,
+                err,
+            )
+
+            if retry_after is not None and retry_after > 0:
+                breaker.reset(retry_after)
+            if not _should_retry(err.response):
+                raise err
+
+        retry_times = retry_times + 1

arkruntime/resources/batch/batch.py

@@ -0,0 +1,40 @@
+# Copyright (c) 2026 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+from ..._compat import cached_property
+from ..._resource import AsyncAPIResource, SyncAPIResource
+from .chat import AsyncChat, Chat
+from .embeddings import AsyncEmbeddings, Embeddings
+from .multimodal_embeddings import AsyncMultimodalEmbeddings, MultimodalEmbeddings
+
+__all__ = ["Batch", "AsyncBatch"]
+
+
+class Batch(SyncAPIResource):
+    @cached_property
+    def chat(self) -> Chat:
+        return Chat(self._client)
+
+    @cached_property
+    def embeddings(self) -> Embeddings:
+        return Embeddings(self._client)
+
+    @cached_property
+    def multimodal_embeddings(self) -> MultimodalEmbeddings:
+        return MultimodalEmbeddings(self._client)
+
+
+class AsyncBatch(AsyncAPIResource):
+    @cached_property
+    def chat(self) -> AsyncChat:
+        return AsyncChat(self._client)
+
+    @cached_property
+    def embeddings(self) -> AsyncEmbeddings:
+        return AsyncEmbeddings(self._client)
+
+    @cached_property
+    def multimodal_embeddings(self) -> AsyncMultimodalEmbeddings:
+        return AsyncMultimodalEmbeddings(self._client)

arkruntime/resources/batch/chat/__init__.py

@@ -0,0 +1,6 @@
+# Copyright (c) 2026 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+
+from .chat import AsyncChat, Chat
+
+__all__ = ["Chat", "AsyncChat"]

arkruntime/resources/batch/chat/chat.py

@@ -0,0 +1,22 @@
+# Copyright (c) 2026 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+from ...._compat import cached_property
+from ...._resource import AsyncAPIResource, SyncAPIResource
+from .completions import AsyncCompletions, Completions
+
+__all__ = ["Chat", "AsyncChat"]
+
+
+class Chat(SyncAPIResource):
+    @cached_property
+    def completions(self) -> Completions:
+        return Completions(self._client)
+
+
+class AsyncChat(AsyncAPIResource):
+    @cached_property
+    def completions(self) -> AsyncCompletions:
+        return AsyncCompletions(self._client)

arkruntime/resources/batch/chat/completions.py

@@ -0,0 +1,217 @@
+# Copyright (c) 2025 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+# License text: https://github.com/volcengine/ark-runtime-python/blob/main/LICENSE
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional
+
+import httpx
+
+from ...._base_client import make_request_options
+from ...._compat import cached_property
+from ...._resource import AsyncAPIResource, SyncAPIResource
+from ...._response import (
+    async_to_raw_response_wrapper,
+    to_raw_response_wrapper,
+)
+from ...._types import Body, Headers, Query
+from ...._utils._utils import async_with_sts_token, with_sts_token
+
+# AUTOGEN-START create-imports
+from ....types.chat.chat_completion_request_message_param import (
+    ChatCompletionRequestMessageParam,
+)
+from ....types.chat.chat_completion_response import ChatCompletionResponse
+from ....types.chat.chat_completion_response_format_param import (
+    ChatCompletionResponseFormatParam,
+)
+from ....types.chat.chat_completion_stop_param import ChatCompletionStopParam
+from ....types.chat.chat_completion_tool_choice_param import (
+    ChatCompletionToolChoiceParam,
+)
+from ....types.chat.chat_completion_tool_param import ChatCompletionToolParam
+from ....types.chat.reasoning_effort import ReasoningEffort
+from ....types.chat.request_service_tier import RequestServiceTier
+from ....types.chat.thinking_param import ThinkingParam
+from ...encryption import async_with_e2e_encryption, with_e2e_encryption
+
+# AUTOGEN-END create-imports
+from .._utils import async_with_batch_retry, get_request_last_time, with_batch_retry
+
+__all__ = [
+    "Completions",
+    "AsyncCompletions",
+    "CompletionsWithRawResponse",
+    "AsyncCompletionsWithRawResponse",
+]
+
+
+class Completions(SyncAPIResource):
+    @cached_property
+    def with_raw_response(self) -> CompletionsWithRawResponse:
+        return CompletionsWithRawResponse(self)
+
+    @with_sts_token
+    @with_e2e_encryption
+    def create(
+        self,
+        *,
+        messages: List[ChatCompletionRequestMessageParam],
+        model: str,
+        frequency_penalty: Optional[float] = None,
+        presence_penalty: Optional[float] = None,
+        repetition_penalty: Optional[float] = None,
+        logit_bias: Optional[Dict[str, float]] = None,
+        logprobs: Optional[bool] = None,
+        top_logprobs: Optional[int] = None,
+        max_tokens: Optional[int] = None,
+        max_completion_tokens: Optional[int] = None,
+        n: Optional[int] = None,
+        response_format: Optional[ChatCompletionResponseFormatParam] = None,
+        stop: Optional[ChatCompletionStopParam] = None,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        tool_choice: Optional[ChatCompletionToolChoiceParam] = None,
+        tools: Optional[List[ChatCompletionToolParam]] = None,
+        parallel_tool_calls: Optional[bool] = None,
+        user: Optional[str] = None,
+        service_tier: Optional[RequestServiceTier] = None,
+        reasoning_effort: Optional[ReasoningEffort] = None,
+        thinking: Optional[ThinkingParam] = None,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None = None,
+    ) -> ChatCompletionResponse:
+        deadline = get_request_last_time(self._client, timeout)
+        breaker = self._client.get_model_breaker(model)
+
+        return with_batch_retry(
+            deadline,
+            breaker,
+            self._post_without_retry,
+            "/batch/chat/completions",
+            body={
+                "messages": messages,
+                "model": model,
+                "frequency_penalty": frequency_penalty,
+                "presence_penalty": presence_penalty,
+                "repetition_penalty": repetition_penalty,
+                "logit_bias": logit_bias,
+                "logprobs": logprobs,
+                "top_logprobs": top_logprobs,
+                "max_tokens": max_tokens,
+                "max_completion_tokens": max_completion_tokens,
+                "n": n,
+                "response_format": response_format,
+                "stop": stop,
+                "temperature": temperature,
+                "top_p": top_p,
+                "tool_choice": tool_choice,
+                "tools": tools,
+                "parallel_tool_calls": parallel_tool_calls,
+                "user": user,
+                "service_tier": service_tier,
+                "reasoning_effort": reasoning_effort,
+                "thinking": thinking,
+            },
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+            ),
+            cast_to=ChatCompletionResponse,
+        )
+
+
+class AsyncCompletions(AsyncAPIResource):
+    @cached_property
+    def with_raw_response(self) -> AsyncCompletionsWithRawResponse:
+        return AsyncCompletionsWithRawResponse(self)
+
+    @async_with_sts_token
+    @async_with_e2e_encryption
+    async def create(
+        self,
+        *,
+        messages: List[ChatCompletionRequestMessageParam],
+        model: str,
+        frequency_penalty: Optional[float] = None,
+        presence_penalty: Optional[float] = None,
+        repetition_penalty: Optional[float] = None,
+        logit_bias: Optional[Dict[str, float]] = None,
+        logprobs: Optional[bool] = None,
+        top_logprobs: Optional[int] = None,
+        max_tokens: Optional[int] = None,
+        max_completion_tokens: Optional[int] = None,
+        n: Optional[int] = None,
+        response_format: Optional[ChatCompletionResponseFormatParam] = None,
+        stop: Optional[ChatCompletionStopParam] = None,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        tool_choice: Optional[ChatCompletionToolChoiceParam] = None,
+        tools: Optional[List[ChatCompletionToolParam]] = None,
+        parallel_tool_calls: Optional[bool] = None,
+        user: Optional[str] = None,
+        service_tier: Optional[RequestServiceTier] = None,
+        reasoning_effort: Optional[ReasoningEffort] = None,
+        thinking: Optional[ThinkingParam] = None,
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None = None,
+    ) -> ChatCompletionResponse:
+        deadline = get_request_last_time(self._client, timeout)
+        breaker = await self._client.get_model_breaker(model)
+
+        return await async_with_batch_retry(
+            deadline,
+            breaker,
+            self._post_without_retry,
+            "/batch/chat/completions",
+            body={
+                "messages": messages,
+                "model": model,
+                "frequency_penalty": frequency_penalty,
+                "presence_penalty": presence_penalty,
+                "repetition_penalty": repetition_penalty,
+                "logit_bias": logit_bias,
+                "logprobs": logprobs,
+                "top_logprobs": top_logprobs,
+                "max_tokens": max_tokens,
+                "max_completion_tokens": max_completion_tokens,
+                "n": n,
+                "response_format": response_format,
+                "stop": stop,
+                "temperature": temperature,
+                "top_p": top_p,
+                "tool_choice": tool_choice,
+                "tools": tools,
+                "parallel_tool_calls": parallel_tool_calls,
+                "user": user,
+                "service_tier": service_tier,
+                "reasoning_effort": reasoning_effort,
+                "thinking": thinking,
+            },
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+            ),
+            cast_to=ChatCompletionResponse,
+        )
+
+
+class CompletionsWithRawResponse:
+    def __init__(self, completions: Completions) -> None:
+        self._completions = completions
+        self.create = to_raw_response_wrapper(completions.create)
+
+
+class AsyncCompletionsWithRawResponse:
+    def __init__(self, completions: AsyncCompletions) -> None:
+        self._completions = completions
+        self.create = async_to_raw_response_wrapper(completions.create)

arkruntime/resources/batch/embeddings.py

@@ -0,0 +1,150 @@
+# Copyright (c) 2025 ByteDance Ltd. and/or its affiliates.
+# SPDX-License-Identifier: Apache-2.0
+# License text: https://github.com/volcengine/ark-runtime-python/blob/main/LICENSE
+
+from __future__ import annotations
+
+from typing import Optional
+
+import httpx
+
+from ..._base_client import make_request_options
+from ..._compat import cached_property
+from ..._resource import AsyncAPIResource, SyncAPIResource
+from ..._response import (
+    async_to_raw_response_wrapper,
+    to_raw_response_wrapper,
+)
+from ..._types import Body, Headers, Query
+from ..._utils import async_maybe_transform, maybe_transform
+from ..._utils._utils import async_with_sts_token, with_sts_token
+
+# AUTOGEN-START create-imports
+from ...types.embedding.embedding_input_param import EmbeddingInputParam
+from ...types.embedding.embedding_request_param import EmbeddingRequestParam
+from ...types.embedding.embedding_response import EmbeddingResponse
+from ...types.embedding.encoding_format import EncodingFormat
+
+# AUTOGEN-END create-imports
+from ._utils import async_with_batch_retry, get_request_last_time, with_batch_retry
+
+__all__ = [
+    "Embeddings",
+    "AsyncEmbeddings",
+    "EmbeddingsWithRawResponse",
+    "AsyncEmbeddingsWithRawResponse",
+]
+
+
+class Embeddings(SyncAPIResource):
+    @cached_property
+    def with_raw_response(self) -> EmbeddingsWithRawResponse:
+        return EmbeddingsWithRawResponse(self)
+
+    @with_sts_token
+    def create(
+        self,
+        *,
+        # AUTOGEN-START create-kwargs
+        model: str,
+        input: EmbeddingInputParam,
+        encoding_format: Optional[EncodingFormat] = None,
+        dimensions: Optional[int] = None,
+        user: Optional[str] = None,
+        # AUTOGEN-END create-kwargs
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None = None,
+    ) -> EmbeddingResponse:
+        deadline = get_request_last_time(self._client, timeout)
+        breaker = self._client.get_model_breaker(model)
+
+        return with_batch_retry(
+            deadline,
+            breaker,
+            self._post_without_retry,
+            "/batch/embeddings",
+            body=maybe_transform(
+                {
+                    # AUTOGEN-START create-body
+                    "model": model,
+                    "input": input,
+                    "encoding_format": encoding_format,
+                    "dimensions": dimensions,
+                    "user": user,
+                    # AUTOGEN-END create-body
+                },
+                EmbeddingRequestParam,
+            ),
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+            ),
+            cast_to=EmbeddingResponse,
+        )
+
+
+class AsyncEmbeddings(AsyncAPIResource):
+    @cached_property
+    def with_raw_response(self) -> AsyncEmbeddingsWithRawResponse:
+        return AsyncEmbeddingsWithRawResponse(self)
+
+    @async_with_sts_token
+    async def create(
+        self,
+        *,
+        # AUTOGEN-START create-kwargs
+        model: str,
+        input: EmbeddingInputParam,
+        encoding_format: Optional[EncodingFormat] = None,
+        dimensions: Optional[int] = None,
+        user: Optional[str] = None,
+        # AUTOGEN-END create-kwargs
+        extra_headers: Headers | None = None,
+        extra_query: Query | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None = None,
+    ) -> EmbeddingResponse:
+        deadline = get_request_last_time(self._client, timeout)
+        breaker = await self._client.get_model_breaker(model)
+
+        return await async_with_batch_retry(
+            deadline,
+            breaker,
+            self._post_without_retry,
+            "/batch/embeddings",
+            body=await async_maybe_transform(
+                {
+                    # AUTOGEN-START create-body
+                    "model": model,
+                    "input": input,
+                    "encoding_format": encoding_format,
+                    "dimensions": dimensions,
+                    "user": user,
+                    # AUTOGEN-END create-body
+                },
+                EmbeddingRequestParam,
+            ),
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+            ),
+            cast_to=EmbeddingResponse,
+        )
+
+
+class EmbeddingsWithRawResponse:
+    def __init__(self, embeddings: Embeddings) -> None:
+        self._embeddings = embeddings
+        self.create = to_raw_response_wrapper(embeddings.create)
+
+
+class AsyncEmbeddingsWithRawResponse:
+    def __init__(self, embeddings: AsyncEmbeddings) -> None:
+        self._embeddings = embeddings
+        self.create = async_to_raw_response_wrapper(embeddings.create)