@spec extract_content_ids(SnakeBridge.Ref.t(), [integer()], keyword()) ::
  {:ok, [integer()]} | {:error, Snakepit.Error.t()}

Extract content token ids from the input_ids.

Parameters: input_ids: list[int]

The input_ids of the model output.

Parameters

• input_ids (list(integer()))

Returns

• list(integer())

@spec extract_reasoning(SnakeBridge.Ref.t(), String.t(), term(), keyword()) ::
  {:ok, {term(), term()}} | {:error, Snakepit.Error.t()}

Extract reasoning content from a complete model-generated string.

Used for non-streaming responses where we have the entire model response available before sending to the client.

Parameters: model_output: str

The model-generated string to extract reasoning content from.

request: ChatCompletionRequest

The request object that was used to generate the model_output.

Parameters

• model_output (String.t())
• request (term())

Returns

• {term(), term()}

@spec extract_reasoning_streaming(
  SnakeBridge.Ref.t(),
  String.t(),
  String.t(),
  String.t(),
  term(),
  term(),
  term(),
  keyword()
) :: {:ok, term()} | {:error, Snakepit.Error.t()}

Instance method that should be implemented for extracting reasoning

from an incomplete response; for use when handling reasoning calls and streaming. Has to be an instance method because it requires state - the current tokens/diffs, but also the information about what has previously been parsed and extracted (see constructor)

Parameters

• previous_text (String.t())
• current_text (String.t())
• delta_text (String.t())
• previous_token_ids (term())
• current_token_ids (term())
• delta_token_ids (term())

Returns

• term()

@spec is_reasoning_end(SnakeBridge.Ref.t(), [integer()], keyword()) ::
  {:ok, boolean()} | {:error, Snakepit.Error.t()}

Check if the reasoning content ends in the input_ids.

It is used in structured engines like xgrammar to check if the reasoning content ends in the model output.

Parameters: input_ids: list[int]

The input_ids of the model output.

Parameters

• input_ids (list(integer()))

Returns

• boolean()

@spec is_reasoning_end_streaming(
  SnakeBridge.Ref.t(),
  [integer()],
  [integer()],
  keyword()
) ::
  {:ok, boolean()} | {:error, Snakepit.Error.t()}

Check if the reasoning content ends in the input_ids on a

decode step.

It is used in structured engines like xgrammar to check if the reasoning content ends in the model output during a decode step. input_ids the entire model output and delta_ids are the last few computed tokens of the model output (like during a decode step).

Parameters: input_ids: list[int]

The entire model output.

delta_ids: list[int]

The last few computed tokens of the model output at the current decode step.

Parameters

• input_ids (list(integer()))
• delta_ids (list(integer()))

Returns

• boolean()

@spec new(
  term(),
  keyword()
) :: {:ok, SnakeBridge.Ref.t()} | {:error, Snakepit.Error.t()}

Initialize self. See help(type(self)) for accurate signature.

Parameters

• tokenizer (term())

@spec prepare_structured_tag(SnakeBridge.Ref.t(), term(), term(), keyword()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

Instance method that is implemented for preparing the structured tag

Otherwise, None is returned

Parameters

• original_tag (term())
• tool_server (term())

Returns

• term()