Boolean Logic, Bug Fixes & Doc Improvements

Added #

• Boolean Algebra: Full support for boolean operators (\land, \lor, \neg, \implies, etc.) and comparisons (>, <, =, \ge, \le). Comparisons returning BooleanResult are strictly type-checked.
• Documentation: Added dedicated Boolean Logic guide.
• Infrastructure: Migrated documentation site from Tailwind CSS to UnoCSS.

Fixed #

• Caching Bug: Fixed a critical hash collision issue in CacheManager where 0 > 1 would incorrectly return true due to colliding with 1 > 0.
• Regression: Resolved a regression causing LateInitializationError in CacheManager.

Real-Only Mode for Graphing Applications

Added #

• Interval Arithmetics: Added support for interval arithmetics.
• Real-Only Mode: Added realOnly parameter to Texpr, Evaluator, and EvaluationVisitor classes. When enabled, operations that would produce complex numbers (like sqrt(-1) or ln(-1)) return NaN instead. This provides Desmos-like behavior for graphing applications.

// Default: complex number support
final texpr = Texpr();
texpr.evaluate(r'\sqrt{-1}'); // Returns ComplexResult(0 + 1i)

// Real-only mode: NaN for undefined real operations
final texprGraph = Texpr(realOnly: true);
texprGraph.evaluate(r'\sqrt{-1}'); // Returns NumericResult(NaN)

Changed #

• FunctionRegistry: Refactored to support real-only aware function handlers alongside standard handlers.

• isValid() and validate(): Use parse() with try/catch instead. These methods will be removed in 1.0.0.

Bug Fixes & UX Improvements

• Parsing: Fixed strict syntax parsing for integral bounds (e.g., \int_0^1 is now supported).
• Evaluation: Fixed evaluateNumeric behavior for complex numbers; results are now handled correctly when using evaluate().
• JSON Export: Fixed VectorExpr JSON schema to use components key, matching the AST definition.

Added #

• LaTeX Commands: Implemented support for common academic symbols:
  • Arrows: \mapsto, \Rightarrow, \Leftarrow, \Leftrightarrow.
  • Relation & Set Operators: \approx, \propto, \cup, \cap, \setminus, \subset, \subseteq, \supset, \supseteq.
  • Quantifiers: \forall and \exists (parsed as variables for syntax tolerance).
  • Decorations: \dot, \ddot, and \bar functions for physics and statistical notation.

Changed #

• Breaking Change: Renamed LatexException to TexprException (and LatexParserException -> ParserException, etc.) to better align with the library name.
• Error Messages: Significantly improved parser error clarity. Error messages now use readable symbols (e.g., got: '*') instead of internal enum names (e.g., got: multiply).
• Parser: Improved syntax tolerance for known commands on implicit multiplication (e.g., sin(x)).
• Documentation: Updated feature guides, notation references, and example code.

Draft Release: Benchmarks, WASM, and Documentation Overhaul.

• WASM Support: Added initial WebAssembly compilation targets and examples (wasm/).
• Benchmarks: Benchmarking suite added (benchmark/).
• Documentation: Updates to API docs and feature guides.
• Testing: Expanded test coverage for complex features and edge cases.

Initial release of TeXpr.

This is a complete rebranding and evolution of the legacy latex_math_evaluator project. It serves as a engine for parsing, evaluating, and analyzing LaTeX math expressions.

✨ Features #

• Core Parsing:

• Full recursive descent parser for standard mathematical LaTeX.

• Generates a structured Abstract Syntax Tree.

• Supports implicit multiplication, variable binding, and custom functions.

• Evaluation Engine:

  • Numeric: Arithmetic, Trig, Log, Roots, Factorials.
  • Complex Numbers: Full support for i, Euler's identity, and complex transcendental functions.
  • Matrix: Determinants, Inverses, Transposition, and basic arithmetic.
  • Calculus: Numerical integration (Simpson's Rule) and Symbolic differentiation (Product/Chain rules).

• Interop & Export:

  • JSON: Export AST to JSON for external tooling.
  • MathML (Alpha): Export to standard web-compatible MathML.
  • SymPy (Alpha): Generate Python/SymPy code from Dart objects.

• Advanced Features:

  • Optimization: Multi-layer caching (L1-L4) with LRU/LFU policies.
  • Analysis: Piecewise differentiation and symbolic simplification.
  • Diagnostics: Levenshtein-based suggestions for syntax errors.

(Note: This project builds upon 1,800+ tests and architecture from the deprecated latex_math_evaluator v0.2.0)