Custom grammars

@lexiconlang/grammar is a Tracery-compatible context-free grammar engine with a TypeScript tagged-template DSL. JSON grammars from Tracery are mostly source-compatible; the TS form is type-checked and ergonomic.

TS template DSL

import { grammar, t } from "@lexiconlang/grammar";

const spell = grammar({
  start:   t`${"prefix.cap"} ${"element.cap"} ${"form.cap"}`,
  prefix:  ["lesser", "greater", "true", "binding"],
  element: ["fire", "frost", "shadow", "iron"],
  form:    ["bolt", "ward", "veil", "lash"],
});

spell.generate(ctx); // "Greater Frost Ward"

${"symbol.modifier"} looks up a symbol in the grammar and pipes its output through built-in modifiers.

JSON form (Tracery compatible)

{
  "start":   "#prefix.cap# #element.cap# #form.cap#",
  "prefix":  ["lesser", "greater", "true", "binding"],
  "element": ["fire", "frost", "shadow", "iron"],
  "form":    ["bolt", "ward", "veil", "lash"]
}

Both forms compile to the same AST.

Built-in modifiers

Modifier	Effect
.cap	Capitalize first letter
.upper	UPPERCASE
.lower	lowercase
.s	Pluralize (English heuristic)
.a	Prepend "a"/"an"
.title	Title Case
.trim	Trim whitespace
... 9 more	See package reference

Cross-pack symbol resolution

Symbols can reference generators in other plugins via namespace:

const spell = grammar({
  start: t`${"prefix.cap"} ${"markov:elven"}-${"element.cap"}`,
});

#markov:elven# resolves through the registry to a Markov-trained elvish syllable generator. Strategies inter-operate.

When to use grammars vs. other primitives

Need	Use
Pick from N weighted options	weightedList
Structured templates with recursion	grammar
Phonotactically coherent name parts	markov
Names that decompose into morpheme meanings	@lexiconlang/language
Object types with multiple fields	compose