Struct Engine 

pub struct Engine<'a, S, D>
where
    S: ScopeData,
    D: HanjaDictionary + ?Sized,
{ /* private fields */ }

Stateful hanja conversion engine for chunked token streams.

Engine is the low-level streaming surface. Call Engine::push_token for each incoming token and then Engine::finish once the upstream reader is exhausted. When the dictionary reports a maximum word length, text chunks are buffered only at the tail so dictionary matches can cross chunk boundaries without requiring the whole document in memory. A trailing fallback hanja run is also kept buffered until a non-convertible boundary or EOF so render modes that expose annotation spans match one-shot conversion. Dictionaries with an unknown maximum keep hanja-containing text until a non-convertible boundary or EOF so long custom entries remain observable.

Implementations

impl<'a, S, D> Engine<'a, S, D>

where S: ScopeData, D: HanjaDictionary + ?Sized,

pub fn new(dictionary: &'a D) -> Self

Creates a streaming engine with default options.

pub fn with_options(dictionary: &'a D, options: EngineOptions) -> Self

Creates a streaming engine with explicit conversion options.

pub fn push_token(&mut self, token: InputToken<S>) -> Vec<OutputToken<S>>

Pushes one input token and returns output tokens that are now safe to emit.

pub fn flush(&mut self) -> Vec<OutputToken<S>>

Flushes all pending text without ending the engine.

pub fn finish(self) -> Vec<OutputToken<S>>

Finishes the stream and returns every remaining output token.

pub fn buffered_chars(&self) -> usize

Returns the number of Unicode scalar values currently buffered.