main
  1# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
  2
  3from __future__ import annotations
  4
  5from typing import Union, Iterable, Optional
  6from typing_extensions import Literal
  7
  8import httpx
  9
 10from ... import _legacy_response
 11from ..._types import Body, Omit, Query, Headers, NotGiven, omit, not_given
 12from ..._utils import maybe_transform, async_maybe_transform
 13from ..._compat import cached_property
 14from ..._resource import SyncAPIResource, AsyncAPIResource
 15from ..._response import to_streamed_response_wrapper, async_to_streamed_response_wrapper
 16from ..._base_client import make_request_options
 17from ...types.responses import input_token_count_params
 18from ...types.responses.tool_param import ToolParam
 19from ...types.shared_params.reasoning import Reasoning
 20from ...types.responses.response_input_item_param import ResponseInputItemParam
 21from ...types.responses.input_token_count_response import InputTokenCountResponse
 22
 23__all__ = ["InputTokens", "AsyncInputTokens"]
 24
 25
 26class InputTokens(SyncAPIResource):
 27    @cached_property
 28    def with_raw_response(self) -> InputTokensWithRawResponse:
 29        """
 30        This property can be used as a prefix for any HTTP method call to return
 31        the raw response object instead of the parsed content.
 32
 33        For more information, see https://www.github.com/openai/openai-python#accessing-raw-response-data-eg-headers
 34        """
 35        return InputTokensWithRawResponse(self)
 36
 37    @cached_property
 38    def with_streaming_response(self) -> InputTokensWithStreamingResponse:
 39        """
 40        An alternative to `.with_raw_response` that doesn't eagerly read the response body.
 41
 42        For more information, see https://www.github.com/openai/openai-python#with_streaming_response
 43        """
 44        return InputTokensWithStreamingResponse(self)
 45
 46    def count(
 47        self,
 48        *,
 49        conversation: Optional[input_token_count_params.Conversation] | Omit = omit,
 50        input: Union[str, Iterable[ResponseInputItemParam], None] | Omit = omit,
 51        instructions: Optional[str] | Omit = omit,
 52        model: Optional[str] | Omit = omit,
 53        parallel_tool_calls: Optional[bool] | Omit = omit,
 54        previous_response_id: Optional[str] | Omit = omit,
 55        reasoning: Optional[Reasoning] | Omit = omit,
 56        text: Optional[input_token_count_params.Text] | Omit = omit,
 57        tool_choice: Optional[input_token_count_params.ToolChoice] | Omit = omit,
 58        tools: Optional[Iterable[ToolParam]] | Omit = omit,
 59        truncation: Literal["auto", "disabled"] | Omit = omit,
 60        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
 61        # The extra values given here take precedence over values defined on the client or passed to this method.
 62        extra_headers: Headers | None = None,
 63        extra_query: Query | None = None,
 64        extra_body: Body | None = None,
 65        timeout: float | httpx.Timeout | None | NotGiven = not_given,
 66    ) -> InputTokenCountResponse:
 67        """
 68        Get input token counts
 69
 70        Args:
 71          conversation: The conversation that this response belongs to. Items from this conversation are
 72              prepended to `input_items` for this response request. Input items and output
 73              items from this response are automatically added to this conversation after this
 74              response completes.
 75
 76          input: Text, image, or file inputs to the model, used to generate a response
 77
 78          instructions: A system (or developer) message inserted into the model's context. When used
 79              along with `previous_response_id`, the instructions from a previous response
 80              will not be carried over to the next response. This makes it simple to swap out
 81              system (or developer) messages in new responses.
 82
 83          model: Model ID used to generate the response, like `gpt-4o` or `o3`. OpenAI offers a
 84              wide range of models with different capabilities, performance characteristics,
 85              and price points. Refer to the
 86              [model guide](https://platform.openai.com/docs/models) to browse and compare
 87              available models.
 88
 89          parallel_tool_calls: Whether to allow the model to run tool calls in parallel.
 90
 91          previous_response_id: The unique ID of the previous response to the model. Use this to create
 92              multi-turn conversations. Learn more about
 93              [conversation state](https://platform.openai.com/docs/guides/conversation-state).
 94              Cannot be used in conjunction with `conversation`.
 95
 96          reasoning: **gpt-5 and o-series models only** Configuration options for
 97              [reasoning models](https://platform.openai.com/docs/guides/reasoning).
 98
 99          text: Configuration options for a text response from the model. Can be plain text or
100              structured JSON data. Learn more:
101
102              - [Text inputs and outputs](https://platform.openai.com/docs/guides/text)
103              - [Structured Outputs](https://platform.openai.com/docs/guides/structured-outputs)
104
105          tool_choice: How the model should select which tool (or tools) to use when generating a
106              response. See the `tools` parameter to see how to specify which tools the model
107              can call.
108
109          tools: An array of tools the model may call while generating a response. You can
110              specify which tool to use by setting the `tool_choice` parameter.
111
112          truncation: The truncation strategy to use for the model response. - `auto`: If the input to
113              this Response exceeds the model's context window size, the model will truncate
114              the response to fit the context window by dropping items from the beginning of
115              the conversation. - `disabled` (default): If the input size will exceed the
116              context window size for a model, the request will fail with a 400 error.
117
118          extra_headers: Send extra headers
119
120          extra_query: Add additional query parameters to the request
121
122          extra_body: Add additional JSON properties to the request
123
124          timeout: Override the client-level default timeout for this request, in seconds
125        """
126        return self._post(
127            "/responses/input_tokens",
128            body=maybe_transform(
129                {
130                    "conversation": conversation,
131                    "input": input,
132                    "instructions": instructions,
133                    "model": model,
134                    "parallel_tool_calls": parallel_tool_calls,
135                    "previous_response_id": previous_response_id,
136                    "reasoning": reasoning,
137                    "text": text,
138                    "tool_choice": tool_choice,
139                    "tools": tools,
140                    "truncation": truncation,
141                },
142                input_token_count_params.InputTokenCountParams,
143            ),
144            options=make_request_options(
145                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
146            ),
147            cast_to=InputTokenCountResponse,
148        )
149
150
151class AsyncInputTokens(AsyncAPIResource):
152    @cached_property
153    def with_raw_response(self) -> AsyncInputTokensWithRawResponse:
154        """
155        This property can be used as a prefix for any HTTP method call to return
156        the raw response object instead of the parsed content.
157
158        For more information, see https://www.github.com/openai/openai-python#accessing-raw-response-data-eg-headers
159        """
160        return AsyncInputTokensWithRawResponse(self)
161
162    @cached_property
163    def with_streaming_response(self) -> AsyncInputTokensWithStreamingResponse:
164        """
165        An alternative to `.with_raw_response` that doesn't eagerly read the response body.
166
167        For more information, see https://www.github.com/openai/openai-python#with_streaming_response
168        """
169        return AsyncInputTokensWithStreamingResponse(self)
170
171    async def count(
172        self,
173        *,
174        conversation: Optional[input_token_count_params.Conversation] | Omit = omit,
175        input: Union[str, Iterable[ResponseInputItemParam], None] | Omit = omit,
176        instructions: Optional[str] | Omit = omit,
177        model: Optional[str] | Omit = omit,
178        parallel_tool_calls: Optional[bool] | Omit = omit,
179        previous_response_id: Optional[str] | Omit = omit,
180        reasoning: Optional[Reasoning] | Omit = omit,
181        text: Optional[input_token_count_params.Text] | Omit = omit,
182        tool_choice: Optional[input_token_count_params.ToolChoice] | Omit = omit,
183        tools: Optional[Iterable[ToolParam]] | Omit = omit,
184        truncation: Literal["auto", "disabled"] | Omit = omit,
185        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
186        # The extra values given here take precedence over values defined on the client or passed to this method.
187        extra_headers: Headers | None = None,
188        extra_query: Query | None = None,
189        extra_body: Body | None = None,
190        timeout: float | httpx.Timeout | None | NotGiven = not_given,
191    ) -> InputTokenCountResponse:
192        """
193        Get input token counts
194
195        Args:
196          conversation: The conversation that this response belongs to. Items from this conversation are
197              prepended to `input_items` for this response request. Input items and output
198              items from this response are automatically added to this conversation after this
199              response completes.
200
201          input: Text, image, or file inputs to the model, used to generate a response
202
203          instructions: A system (or developer) message inserted into the model's context. When used
204              along with `previous_response_id`, the instructions from a previous response
205              will not be carried over to the next response. This makes it simple to swap out
206              system (or developer) messages in new responses.
207
208          model: Model ID used to generate the response, like `gpt-4o` or `o3`. OpenAI offers a
209              wide range of models with different capabilities, performance characteristics,
210              and price points. Refer to the
211              [model guide](https://platform.openai.com/docs/models) to browse and compare
212              available models.
213
214          parallel_tool_calls: Whether to allow the model to run tool calls in parallel.
215
216          previous_response_id: The unique ID of the previous response to the model. Use this to create
217              multi-turn conversations. Learn more about
218              [conversation state](https://platform.openai.com/docs/guides/conversation-state).
219              Cannot be used in conjunction with `conversation`.
220
221          reasoning: **gpt-5 and o-series models only** Configuration options for
222              [reasoning models](https://platform.openai.com/docs/guides/reasoning).
223
224          text: Configuration options for a text response from the model. Can be plain text or
225              structured JSON data. Learn more:
226
227              - [Text inputs and outputs](https://platform.openai.com/docs/guides/text)
228              - [Structured Outputs](https://platform.openai.com/docs/guides/structured-outputs)
229
230          tool_choice: How the model should select which tool (or tools) to use when generating a
231              response. See the `tools` parameter to see how to specify which tools the model
232              can call.
233
234          tools: An array of tools the model may call while generating a response. You can
235              specify which tool to use by setting the `tool_choice` parameter.
236
237          truncation: The truncation strategy to use for the model response. - `auto`: If the input to
238              this Response exceeds the model's context window size, the model will truncate
239              the response to fit the context window by dropping items from the beginning of
240              the conversation. - `disabled` (default): If the input size will exceed the
241              context window size for a model, the request will fail with a 400 error.
242
243          extra_headers: Send extra headers
244
245          extra_query: Add additional query parameters to the request
246
247          extra_body: Add additional JSON properties to the request
248
249          timeout: Override the client-level default timeout for this request, in seconds
250        """
251        return await self._post(
252            "/responses/input_tokens",
253            body=await async_maybe_transform(
254                {
255                    "conversation": conversation,
256                    "input": input,
257                    "instructions": instructions,
258                    "model": model,
259                    "parallel_tool_calls": parallel_tool_calls,
260                    "previous_response_id": previous_response_id,
261                    "reasoning": reasoning,
262                    "text": text,
263                    "tool_choice": tool_choice,
264                    "tools": tools,
265                    "truncation": truncation,
266                },
267                input_token_count_params.InputTokenCountParams,
268            ),
269            options=make_request_options(
270                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
271            ),
272            cast_to=InputTokenCountResponse,
273        )
274
275
276class InputTokensWithRawResponse:
277    def __init__(self, input_tokens: InputTokens) -> None:
278        self._input_tokens = input_tokens
279
280        self.count = _legacy_response.to_raw_response_wrapper(
281            input_tokens.count,
282        )
283
284
285class AsyncInputTokensWithRawResponse:
286    def __init__(self, input_tokens: AsyncInputTokens) -> None:
287        self._input_tokens = input_tokens
288
289        self.count = _legacy_response.async_to_raw_response_wrapper(
290            input_tokens.count,
291        )
292
293
294class InputTokensWithStreamingResponse:
295    def __init__(self, input_tokens: InputTokens) -> None:
296        self._input_tokens = input_tokens
297
298        self.count = to_streamed_response_wrapper(
299            input_tokens.count,
300        )
301
302
303class AsyncInputTokensWithStreamingResponse:
304    def __init__(self, input_tokens: AsyncInputTokens) -> None:
305        self._input_tokens = input_tokens
306
307        self.count = async_to_streamed_response_wrapper(
308            input_tokens.count,
309        )