@danielsimonjr/mathts-expression

Expression parser & evaluator for MathTS. The package root also re-exports the MathNode and StringOptions types used by scalarToNode / scalarToTex in @danielsimonjr/mathts-functions.

The expression engine for MathTS: a parser (string → AST), an AST node set, a compiler, and an evaluator. Includes a sandbox (getSafeProperty / getSafeMethod) for safe property access.

Limits on untrusted strings: max length 1,000,000 characters, max parenthesis nesting 125 (and 500 nested unary, power, assignment or conditional rules), max compile depth 512, which evaluate also applies before validation (see expression/tests/security/limits.test.ts and docs/api/expression.md).

Install

npm install @danielsimonjr/mathts-expression

What it provides

• createParse / createParserClass / createParser — parse to an AST.
• 16 AST node constructors (createConstantNode, createOperatorNode, …).
• compile / createEvaluate / compileExpression — evaluate expressions.
• createScalarBuilder — build a scalar node tree (number, symbol, op, call, from). ln lowers to natural log and log lowers to log10. A formula string is rejected. Evaluate with evaluateScalar from @danielsimonjr/mathts-functions.
• scalarCanonical / scalarCanonicalKey / scalarSize / scalarDepth / scalarComplexity / scalarSymbols — structural toolkit for an existing ScalarNode. The canonical key flattens + and *, rewrites - and /, and does not fold 2^3 to 8.
• enumerateScalars / scalarGrammarAccepts — deterministic size-ordered enumeration of scalar trees from a grammar, with deduplication by scalarCanonicalKey. It is a generator that returns an EnumerationStop reason (exhausted-space, candidate-limit, time-limit, cancelled). budget.maxCandidates, budget.maxWallClockMs, and an AbortSignal end it; the wall clock and the signal are checked every 128 candidates considered, so a grammar that yields few trees still stops promptly.

import { createScalarBuilder, scalarEquals } from '@danielsimonjr/mathts-expression';

const b = createScalarBuilder();
const aa = b.op('*', [b.symbol('a'), b.symbol('a')]);
const aSquared = b.op('^', [b.symbol('a'), b.number(2)]);
console.log(scalarEquals(aa, aSquared)); // true

• getStringSubset / setStringSubset — string indexing used by accessor/assign and by functions subset (the barrel is the only public entry).
• AST serializers on every node: toString(), toTex() (LaTeX), toHTML(), and toMathML() (MathML).
• Focused re-exports are also published: @danielsimonjr/mathts-{parser,ast,evaluator}.

Serializing an AST

Every node serializes to several formats. toMathML() returns a MathML fragment (like toTex() returns a LaTeX fragment, without delimiters); wrap it in a <math> element with mathMLDocument(node) to render. MathML is typeset natively by modern browsers — no external dependencies.

import { mathMLDocument, mathMLError } from '@danielsimonjr/mathts-expression';

const node = parse('c = 1 / sqrt(eps0 * mu0)'); // parse from the functions package
node.toString(); // 'c = 1 / sqrt(eps0 * mu0)'
node.toTex(); // ' c=\\frac{1}{\\sqrt{ eps0\\cdot mu0}}'
node.toMathML(); // '<mrow><mi>c</mi><mo>=</mo><mfrac>…</mfrac></mrow>'
mathMLDocument(node); // '<math xmlns="…" display="block">…</math>'  (renderable)

mathMLError(src) produces a <math><merror> element for a parse failure.

License

MIT (c) Daniel Simon Jr.