Language Parsers

Revet uses Tree-sitter for fast, accurate AST parsing. Each language parser builds nodes and edges in the code dependency graph.

Supported languages

Language	Extensions	Graph features
Python	.py	Functions, classes, decorators, async, nested scopes
TypeScript / JS	.ts, .tsx, .js, .jsx	Classes, interfaces, generics, arrow functions, enums
Rust	.rs	Functions, structs, enums, traits, impl blocks
Go	.go	Functions, methods, structs, interfaces, goroutines
Java	.java	Classes, interfaces, records, enums, nested classes
C#	.cs	Classes, interfaces, records, structs, attributes, generics
Kotlin	.kt, .kts	Classes, objects, data classes, annotations, sealed classes
Ruby	.rb, .rake, .gemspec	Classes, modules, mixins, attr_accessors
PHP	.php	Classes, traits, enums, namespaces, attributes
Swift	.swift	Classes, structs, protocols, extensions, enums
C / C++	.c, .h, .cpp, .cc, .cxx, .hpp, .hxx	Functions, structs, C++ classes with inheritance, #include imports, macros, call graph

What parsers produce

Each parser creates:

• File node — one per source file, with language metadata
• Function / Method nodes — with parameter lists and line numbers
• Class / Struct / Trait nodes — with inheritance relationships
• Import nodes — with module specifiers for cross-file resolution
• Contains edges — file → symbol, class → method
• Imports edges — file → imported file (resolved cross-file)
• Calls edges — caller → callee (resolved cross-file)
• Inherits edges — subclass → superclass

Incremental parsing

Parsed file graphs are cached under .revet-cache/files/ keyed by content hash. On subsequent runs, only changed files are re-parsed by tree-sitter. Second runs on unchanged codebases are near-instant.

Adding a language

See Contributing for the step-by-step guide to adding a new language parser.