OpenAiGateway

Build the request body for the OpenAI Responses API.

Build the request body for a stateful Responses API continuation.

Extract the latest tool results for a stateful continuation request.

Merge shared Responses API options onto the given request body.

Map a tool choice to the OpenAI Responses tool_choice shape.

Determine if OpenAI should receive full stateless conversation history.

Determine if the model supports encrypted reasoning content.

Build the text format options for structured output.

Create an HTTP client with the provider's request options.

Select a boot-registered connection without retaining dynamic provider names.

Get an HTTP client for the OpenAI API.

Get the base URL for the OpenAI-compatible API.

Process streamed response events and return the completed step.

Generate a lowercase UUID v7 for use as a stream event ID.

Generate text for a single Responses API step.

Stream text for a single Responses API step.

Build the request body for the current text generation step.

Derive a filename for a nameless document from its MIME type.

Map the given Hypervel attachments to OpenAI content parts.

Determine if the given uploaded file is an image.

Map the given Hypervel messages to OpenAI Responses API input format.

Map a user message to OpenAI format.

Map an assistant message to OpenAI format.

Remove file_search_call items, which the API resolves by id and so rejects when store is false.

Map a tool call to a function_call input item, keeping the item id only when OpenAI issued it and its reasoning survived.

Map a tool result message to OpenAI format.

Map the given tools to OpenAI function definitions.

Ensure hosted tool search is not used while response storage is disabled.

Map a regular tool to an OpenAI function definition.

Map a provider tool to an OpenAI provider tool definition.

Map a code execution tool to an OpenAI code interpreter definition.

Map a file search tool to an OpenAI file search definition.

Map a web search tool to an OpenAI web search definition.

Decode a structured output payload, tolerating markdown code fences.

Join reasoning blocks, separating each block with a blank line.

Validate the OpenAI response data.

Parse the OpenAI response data into a single step response.

Extract the text content from the output array.

Extract the reasoning text from the output array.

Extract citations from the output array.

Extract the ordered response output for full-history replay.

Extract the provider-hosted tool items from the output array.

Extract usage data from the response.

Extract usage data from an image generation response.

Extract and map the finish reason from the response.

Map tool calls with their associated reasoning blocks.

Execute a callback with failoverable error handling.

Get the short Retry-After delay of a rate limited or overloaded response, if it has one.

Determine if the response is a 402, or a 429 insufficient credits or quota error rather than a rate limit.

Get the status codes that indicate a provider is transiently unavailable and the request should fail over.

Get the patterns used to detect insufficient credits or quota errors.

Get the error codes or types that indicate insufficient credits or quota on a 429 response.

Return embedding values in input order without accepting ambiguous indices.

Parse an SSE stream body into decoded JSON data objects.

Read one line without waiting for a larger body buffer to fill.

Determine the appropriate filename for the audio file based on its MIME type.

Create an OpenAI gateway instance.

Generate an image.

Send an image generation request.

Send an image edit request with attachments.

Generate audio from the given text.

Generate text from the given audio.

Generate embeddings for the given inputs.