Struct NDArrayCollector 

pub struct NDArrayCollector<F> { /* private fields */ }

NDArrayCollector is a Collector which transforms frame-based replay data into a 2-dimensional array of type ndarray::Array2, where each element is of a specified floating point type.

It’s initialized with collections of FeatureAdder instances which extract global, player independent features for each frame, and PlayerFeatureAdder, which add player specific features for each frame.

It’s main entrypoint is Self::get_meta_and_ndarray, which provides ndarray::Array2 along with column headers and replay metadata.

Implementations

impl<F> NDArrayCollector<F>

pub fn new( feature_adders: FeatureAdders<F>, player_feature_adders: PlayerFeatureAdders<F>, ) -> Self

Creates a new instance of NDArrayCollector.

Arguments

• feature_adders - A vector of Arc<dyn FeatureAdder<F>>, each implementing the FeatureAdder trait. These are used to add global features to the replay data.

• player_feature_adders - A vector of Arc<dyn PlayerFeatureAdder<F>>, each implementing the PlayerFeatureAdder trait. These are used to add player-specific features to the replay data.

Returns

A new NDArrayCollector instance. This instance is initialized with empty data, no replay metadata and zero frames added.

pub fn get_column_headers(&self) -> NDArrayColumnHeaders

Returns the column headers of the 2-dimensional array produced by the NDArrayCollector.

Returns

An instance of NDArrayColumnHeaders representing the column headers in the collected data.

pub fn get_ndarray(self) -> SubtrActorResult<Array2<F>>

This function consumes the NDArrayCollector instance and returns the data collected as an ndarray::Array2.

Returns

A SubtrActorResult containing the collected data as an ndarray::Array2.

This method is a shorthand for calling Self::get_meta_and_ndarray and discarding the replay metadata and headers.

pub fn get_meta_and_ndarray( self, ) -> SubtrActorResult<(ReplayMetaWithHeaders, Array2<F>)>

Consumes the NDArrayCollector and returns the collected features as a 2D ndarray, along with replay metadata and headers.

Returns

A SubtrActorResult containing a tuple:

• ReplayMetaWithHeaders: The replay metadata along with the headers for each column in the ndarray.
• ndarray::Array2<F>: The collected features as a 2D ndarray.

pub fn process_and_get_meta_and_headers( &mut self, replay: &Replay, ) -> SubtrActorResult<ReplayMetaWithHeaders>

Processes a boxcars::Replay and returns its metadata along with column headers.

This method first processes the replay using a ReplayProcessor. It then updates the replay_meta field if it’s not already set, and returns a clone of the replay_meta field along with column headers of the data.

Arguments

• replay: A reference to the boxcars::Replay to process.

Returns

A SubtrActorResult containing a ReplayMetaWithHeaders that includes the metadata of the replay and column headers.

impl NDArrayCollector<f32>

pub fn from_strings( fa_names: &[&str], pfa_names: &[&str], ) -> SubtrActorResult<Self>

Trait Implementations

impl<F> Collector for NDArrayCollector<F>

fn process_frame( &mut self, processor: &ReplayProcessor<'_>, frame: &Frame, frame_number: usize, current_time: f32, ) -> SubtrActorResult<TimeAdvance>

Process a single frame from a replay.

fn process_replay(self, replay: &Replay) -> SubtrActorResult<Self>

Process an entire replay.

impl<F: TryFrom<f32> + Send + Sync + 'static> Default for NDArrayCollector<F>

where <F as TryFrom<f32>>::Error: Debug,

fn default() -> Self

Returns the “default value” for a type.