hacklight

Hacklight

A lightweight, pragmatic syntax highlighter that's good enough™.

What It Is

Hacklight is a simple tokenizer/syntax highlighter that aims to correctly parse ~95% of real-world code in the 8 most popular languages. It's designed to be:

• Small - Single file, minimal dependencies
• Fast - Simple regex-based tokenization
• Browser-friendly - Works anywhere JavaScript runs
• Pragmatic - Handles the common cases well, doesn't sweat the edge cases

What It Isn't

This is not a perfect parser. It will get some things wrong, especially:

• Complex nested template literals
• Exotic regex patterns
• Language-specific edge cases
• Mixed-language contexts beyond HTML/CSS/JS

If you need 100% accuracy, use a proper AST parser. If you need something that works well for 95% of code you'll encounter and fits in a single file, this might be for you.

Supported Languages

• JavaScript (including JSX, modern ES2024 features)
• TypeScript
• HTML (with embedded CSS/JS)
• CSS (including CSS variables, at-rules)
• Python
• Java
• C/C++
• Go

Usage

const { tokenize } = createTokenizer('auto');  // auto-detects HTML or JS
const tokens = tokenize(sourceCode);

// Convert to HTML with syntax highlighting
const html = tokensToHtml(tokens);

Token Types

• keyword - Language keywords (if, for, class, etc.)
• identifier - Variables, function names, etc.
• string - String literals
• number - Numeric literals
• comment - Comments
• operator - Operators (+, -, =, etc.)
• punctuation - Brackets, semicolons, etc.
• regex - Regular expressions (JS)
• html_tag - HTML tags
• attr_name / attr_bool - HTML attributes
• css_selector - CSS selectors
• css_variable - CSS custom properties
• css_at - CSS at-rules (@media, etc.)
• error_string - Unterminated strings
• whitespace / newline - Formatting

Philosophy

Perfect is the enemy of good. This tokenizer makes practical trade-offs:

• Speed over correctness - Uses regex instead of full parsing
• Simplicity over completeness - One file, minimal state machine
• Common cases over edge cases - Handles typical code patterns well
• Pragmatism over purity - Some heuristics and "good enough" decisions

Known Limitations

• Regex literals vs division operators use heuristics that can be fooled
• Template literal interpolations are tokenized as one string
• HTML/CSS/JS context switching is simplified
• No semantic understanding (can't distinguish types from variables)
• Some exotic syntax constructs may tokenize incorrectly

License

MIT

Contributing

This is intentionally kept simple. Bug fixes welcome, but feature additions that add complexity will likely be declined. The goal is to stay small and "good enough".