A generic AST tree walker supporting enter/exit phases and actions like Skip, Stop, Replace, ReplaceSkip, and ReplaceStop. It traverses a flat array of AstNode's using parent pointers, allowing efficient modifications without recursion. The pattern is reusable in any project needing to walk and transform tree structures, particularly CSS or HTML ASTs.

A utility that splits strings by a separator while respecting nested brackets, parentheses, and quotes. It also handles CSS-like `var()` fallbacks. This is a classic implementation of a bracket-aware splitter that can be extracted and used in any text parsing context (e.g., parsing CSS values, configuration strings, or template syntax).

Defines the `Machine` trait with `reset()` and `next(cursor) -> MachineState` for building linear-time tokenizers. Accompanied by `BracketStack` and `Cursor` types, this pattern drives all Rust-based extractors (utilities, variants, CSS variables, etc.). The generics-based type-state pattern (using PhantomData) allows zero-cost abstraction and is reusable for any character-level or byte-level parsing task.

An incremental gitignore matcher that caches directory ignore rules to avoid re-parsing parent `.gitignore` files on each query. It supports depth limits, hidden files, file size filters, and symlink handling. The `CachedDir` enum internally caches root ignores and per-directory overrides. This is a highly optimized, self-contained utility for file exclusion that could be reused in any file scanning or linting tool.

A recursive-descent parser for CSS function values, handling words, functions, separators, strings, and escape sequences. It produces a flat array of tokens that can be modified and re-serialized. This is a clean, test-covered implementation of a CSS value parser that can be extracted for use in other CSS processing or linting tools.