Parser configuration for Gumbo.parse/2.
Five fields map one-to-one onto gumbo's GumboOptions; :dirty is ours and
controls scheduler dispatch rather than the parse itself. new/1 validates
every key against gumbo's documented ranges and rejects anything else with
{:error, {:invalid_option, key, value}} before any work happens.
Summary
Types
tab_stop, stop_on_first_error, max_errors, max_attributes,
max_tree_depth, fragment_context and fragment_namespace are GumboOptions
fields; :dirty
is this library's scheduler policy. fragment_context is nil (no fragment,
gumbo's GUMBO_TAG_LAST) or a tag atom. tab_stop, max_errors,
max_attributes and max_tree_depth are bounded by C int range, so a value
that passes new/1 can never fail the NIF decode. max_errors defaults to
100 and max_attributes to 1024 rather
than to gumbo's "no cap"; see the comments on the defaults for why, and pass -1
to either for no cap. max_tree_depth counts the parser's open elements, so an
implied <html> counts and a fragment passes 401 to mean 400 levels of content,
as the reference implementation's Ruby layer does; it defaults to
-1, because the stack budget is what bounds depth here.
Functions
The workload size at and above which dispatch/2 picks :dirty.
The single scheduler policy for parse/2 and materialize/2: :dirty when
forced, :normal when forbidden, and :dirty when :auto and the workload
reaches dirty_threshold/0.
Returns the default options.
Validates opts against gumbo's documented ranges and merges them over the
defaults.
Converts the struct to the map the NIF decodes.
Types
@type t() :: %Gumbo.Options{ dirty: :auto | boolean(), fragment_context: nil | Gumbo.Tag.t(), fragment_namespace: Gumbo.Types.namespace(), max_attributes: -1..2_147_483_647, max_errors: -1..2_147_483_647, max_tree_depth: -1..2_147_483_647, stop_on_first_error: boolean(), tab_stop: 1..2_147_483_647 }
tab_stop, stop_on_first_error, max_errors, max_attributes,
max_tree_depth, fragment_context and fragment_namespace are GumboOptions
fields; :dirty
is this library's scheduler policy. fragment_context is nil (no fragment,
gumbo's GUMBO_TAG_LAST) or a tag atom. tab_stop, max_errors,
max_attributes and max_tree_depth are bounded by C int range, so a value
that passes new/1 can never fail the NIF decode. max_errors defaults to
100 and max_attributes to 1024 rather
than to gumbo's "no cap"; see the comments on the defaults for why, and pass -1
to either for no cap. max_tree_depth counts the parser's open elements, so an
implied <html> counts and a fragment passes 401 to mean 400 levels of content,
as the reference implementation's Ruby layer does; it defaults to
-1, because the stack budget is what bounds depth here.
Functions
@spec dirty_threshold() :: 20480
The workload size at and above which dispatch/2 picks :dirty.
Examples
iex> Gumbo.Options.dirty_threshold()
20_480@spec dispatch(t(), non_neg_integer()) :: :normal | :dirty
The single scheduler policy for parse/2 and materialize/2: :dirty when
forced, :normal when forbidden, and :dirty when :auto and the workload
reaches dirty_threshold/0.
Examples
iex> Gumbo.Options.dispatch(Gumbo.Options.new(), 20_479)
:normal
iex> Gumbo.Options.dispatch(Gumbo.Options.new(), 20_480)
:dirty
iex> {:ok, forced} = Gumbo.Options.new(dirty: true)
iex> Gumbo.Options.dispatch(forced, 1)
:dirty@spec new() :: t()
Returns the default options.
Examples
iex> Gumbo.Options.new().tab_stop
8Validates opts against gumbo's documented ranges and merges them over the
defaults.
An unknown key or an out-of-range value returns
{:error, {:invalid_option, key, value}}.
Examples
iex> {:ok, options} = Gumbo.Options.new(max_errors: 3)
iex> options.max_errors
3
iex> Gumbo.Options.new(bogus: 1)
{:error, {:invalid_option, :bogus, 1}}Converts the struct to the map the NIF decodes.
Every field is present, so the NIF can decode strictly.
Examples
iex> Gumbo.Options.to_nif(Gumbo.Options.new()) |> Map.keys() |> Enum.sort()
[:dirty, :fragment_context, :fragment_namespace, :max_attributes, :max_errors, :max_tree_depth, :stop_on_first_error, :tab_stop]