How Tree-sitter Turns TypeScript Source Code Into Structured, Queryable Facts
A developer began experimenting with Tree-sitter, a parsing library, to understand what an Abstract Syntax Tree (AST) reveals about TypeScript code that plain text search cannot. Using a small NestJS controller file as input, the developer parsed it with Tree-sitter and built a recursive tree printer to make the resulting syntax tree readable. Unlike text search, which finds character sequences without context, Tree-sitter identifies nodes by structural type — distinguishing a decorator from a comment or string containing the same characters. This allows developers to extract meaningful facts from code, such as which classes are controllers, which routes use guards, and which services a controller depends on. The experiment highlighted Tree-sitter's role as a structural foundation for editors and static analyzers, with the developer planning to next extract imports, classes, and decorators into normalized, queryable data.
