Home

C# library to build and manipulate abstract syntax trees when writing compilers.

As part of my Oakley project to create a compiler, and its associated OakAsm project to create an assembler, I needed to represent abstract syntax trees in C#. This library was created so that I could share the code between those two projects.

Installation

dotnet add package MrKWatkins.Ast

Nodes and Trees

Trees are built from a self-generic base node type of your own. Nodes expose their children as a rich collection type and can be navigated by parent, sibling, ancestor and descendent, with several strategies for walking the whole tree.

Properties

PropertyNode<TNode> stores node state in a keyed Properties collection rather than in fields. That gives copyable nodes, structural equality by value, and single or multiple valued properties, all behind normal C# properties on your node types.

Messages

Errors, warnings and informational messages are attached to the nodes they apply to, then formatted for output — optionally with the offending source line highlighted.

Source Positions

Nodes can record where they came from with a SourcePosition. Text and binary files are both supported, and positions can be combined to cover the source of an entire subtree.

Lexing

SourceReader reads a text file character by character while tracking lines and columns, producing Token<TKind>s. TokenReader<TKind> then feeds those tokens to a parser, with lookahead, rewinding and error recovery.

Listeners

Listeners walk a tree and are notified as nodes are reached, with access to a context object to accumulate results. They are the lightweight option, best suited to building something new from a tree.

Processing

Processing runs a pipeline of stages over a tree, each stage containing one or more processors running serially or in parallel. It is best suited to mutating a tree, with replacers and validators for the two most common cases.

API Documentation

Reference documentation is generated from the release assembly:

Examples

Two worked examples live alongside the source:

Listeners builds a string representation of a tree using composite listeners.
Maths lexes, parses, reduces, validates, evaluates and compiles mathematical expressions.

Pull Requests

I'm not accepting pull requests at the current time; this project is tailored for some other projects of mine and I want to get them in a suitable state first.

Feel free to raise issues for bugs or suggestions, but I make no guarantees they will get looked at I'm afraid!

Use of AI

My general rule is I'll write the interesting bits and use AI for the boring bits. The main use of AI has been to generate the documentation and help with test coverage.

Licencing

Licensed under MIT.