Struct LanguageToolEngine 

pub struct LanguageToolEngine { /* private fields */ }

Implementations

impl LanguageToolEngine

pub fn new(config: &LanguageToolConfig) -> Self

Trait Implementations

impl Engine for LanguageToolEngine

fn check_many<'life0, 'life1, 'life2, 'async_trait>( &'life0 mut self, texts: &'life1 [String], language_id: &'life2 str, ) -> Pin<Box<dyn Future<Output = Vec<Result<Vec<Diagnostic>>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Pack the prose ranges into as few requests as the size limit allows, and overlap those.

Both halves matter, and the first more than the second. LanguageTool charges about 8 ms per request before it reads a byte, so a document sent one range at a time pays that flat cost once per range — for a 36 kB Typst file, 109 times, which is most of the wall clock. Packing to [LanguageToolConfig::max_request_bytes] cuts that to ten requests. What remains is round-trip latency, and that is what the concurrency limit hides; requests are capped at max_concurrent_requests in flight so a shared server is not swamped.

fn name(&self) -> &'static str

fn check<'life0, 'life1, 'life2, 'async_trait>( &'life0 mut self, text: &'life1 str, language_id: &'life2 str, ) -> Pin<Box<dyn Future<Output = Result<Vec<Diagnostic>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

fn as_external(&self) -> Option<&ExternalEngine>

Downcast hooks for the two engines whose markup support is configured.

fn as_wasm(&self) -> Option<&WasmEngine>

fn supported_languages(&self) -> Vec<String>

BCP-47 tags this engine handles. Empty means every language.