How to generate ASTs instead of CSTs?

[brando90]

What is your question?

I have simply type lambda calculus grammar. But when I use it and variations of it due to how parsers work I get a giant CST (compare to ASTs). I want to convert my CST to an AST with constructors Abs, Apply, Vars. Or directly from strings to ASTs. How does one do that?

If you're having trouble with your code or grammar

term : "λ" term ":" type "." term
    | "[" term term "]"
    | /[A-Za-z0-9_]+/
type : type "->" type  // "->"
    | /[A-Za-z0-9_]+/
// ignore tokens
IGNORE_TOKENS : "(" | ")"
%import common.WS
%ignore WS
%ignore IGNORE_TOKENS
// %import common.ESCAPED_STRING
// %import common.SIGNED_NUMBER

related:

• https://cs.stackexchange.com/questions/146612/how-does-one-parse-a-string-into-an-ast-abstract-syntax-tree-directly-instead
• How to generate ASTs instead of CSTs? #1053

[brando90]

from erezsh:

This might help: https://lark-parser.readthedocs.io/en/latest/examples/advanced/create_ast.html

Also useful to have a good grasp of the built-in tree construction option: https://lark-parser.readthedocs.io/en/latest/tree_construction.html

from: #1053

[brando90]

This might help: https://lark-parser.readthedocs.io/en/latest/examples/advanced/create_ast.html

Also useful to have a good grasp of the built-in tree construction option: https://lark-parser.readthedocs.io/en/latest/tree_construction.html

btw @erezsh, out of curiosity is CST -> AST a very standard PL (programming language) thing?

[erezsh]

Yeah, I'd say so. ASTs are the standard representation of code in compilers, and most parsers generate CSTs. Then a conversion is unavoidable.