Packages
Extract structured data from text using LLMs with source grounding. Maps every extraction back to exact byte positions in the source. Supports Claude, OpenAI, and Gemini providers. Elixir port of google/langextract.
Current section
Files
Jump to
Current section
Files
lib/lang_extract/provider/response.ex
defmodule LangExtract.Provider.Response do
@moduledoc """
A provider's successful inference response: the raw text plus token
usage when the API reported it.
The success value of `c:LangExtract.Provider.infer/2`. `usage` is `nil`
when the response carried no usage block (some proxies and local models
omit it); when present it carries `:input_tokens` and `:output_tokens`.
"""
@type usage :: %{input_tokens: non_neg_integer(), output_tokens: non_neg_integer()}
@type t :: %__MODULE__{text: String.t(), usage: usage() | nil}
@enforce_keys [:text]
defstruct [:text, :usage]
end