Grammar parsing

Parse using the active phrase grammar

Phrase readers resolve the active grammar with longest matching phrase identity. This is useful when punctuation has aliases or a language extends its tokens. Comparing canonical phrase identity avoids treating a quoted delimiter as syntax.

recurloop

import language-kit
let DemoGrammar = phrase { dictionary = true }
let install_demo_grammar = phrase {
    type = <phrase-types:elaborate>
    action = fn (state:Context*, called:Phrase*) -> void {
        LanguageKit:define_token(state, context:phrase:find(state, "DemoGrammar"), "+", 4)
    }
}
install_demo_grammar
let DemoGrammar:"⊕" = <DemoGrammar:"+">
let check_alias = phrase {
    type = <phrase-types:elaborate>
    action = fn (state:Context*, called:Phrase*) -> void {
        let grammar = context:phrase:find(state, "DemoGrammar")
        let reader = LanguageKit:SliceReader:new_phrases(state, grammar, "⊕", 3, 0)
        if !reader { context:diagnostic:error(state, "reader allocation failed"); return }
        defer reader.destroy()
        if !reader.is("+") { context:diagnostic:error(state, "alias did not resolve") }
    }
}
check_alias

The slice length is in UTF-8 bytes: ⊕ occupies three. is checks without consuming. take_canonical consumes and returns an owned canonical token spelling, which must be freed. new_phrases requires a valid grammar identifier; null is not an implicit default grammar.

Choose the matching source boundary

Use Reader:new_phrases to consume the live source instead. A SliceReader:new_source supports lookahead over the current source origin; its cursor is independent. Keep any source refill or cursor-commit policy explicit in a syntax action.