Beancount.Normalizer (beancount_ex v0.6.0)

Normalizes raw engine output into a stable, engine-independent structure.

Normalization is what makes Beancount usable as an oracle: two different engines (the CLI today, a native engine tomorrow) can be compared on their normalized output rather than on incidental, backend-specific text such as temporary file paths.

The normalized map has the shape:

%{
  status: :ok | :error,
  errors: [%{line: non_neg_integer() | nil, message: String.t()}]
}

Summary

Types

error()

t()

Functions

normalize(exit_status, stdout, stderr, source_path \\ nil)

Normalize captured stdout/stderr for a given exit status.

Types

error()

@type error() :: %{line: non_neg_integer() | nil, message: String.t()}

t()

@type t() :: %{status: Beancount.Result.status(), errors: [error()]}

Functions

normalize(exit_status, stdout, stderr, source_path \\ nil)

@spec normalize(non_neg_integer() | nil, binary(), binary(), binary() | nil) :: t()

Normalize captured stdout/stderr for a given exit status.

source_path, when provided, is stripped from error lines so that the normalized output does not depend on temporary file locations and stays deterministic.

Examples

iex> Beancount.Normalizer.normalize(0, "", "")
%{status: :ok, errors: []}

iex> Beancount.Normalizer.normalize(1, "", "/tmp/x.bean:3: Invalid")
%{status: :error, errors: [%{line: 3, message: "Invalid"}]}