Yoho 🔥: A toy compiler written in mojo

Hello Mojocians. I'm writing a toy compiler in mojo. https://github.com/YichengDWu/yoho

GitHub

GitHub - YichengDWu/yoho: A compiler written in Mojo 🔥 and generate...

A compiler written in Mojo 🔥 and generates RISC-V assembly - YichengDWu/yoho

9 Replies

quick question - how did you generate the parser in Mojo ? is there some kind of parser generator or just use recursive descent ?

Ethan•3mo ago

Yes there's a parser generator, see https://github.com/YichengDWu/yoho/blob/main/src/grammar/parser_generator.mojo. The algorithm is called Packrat parsing and very similar to recursive decent. When I need to update the grammar I modify the test/calc.gram file and run 'make parser' defined in Makefile.

ModularBot•3mo ago

Congrats @Ethan, you just advanced to level 6!

Muthu (முத்து)•3mo ago

thanks, thats awesome to know!

curiosity.fan•3mo ago

@Ethan what language does your toy compiler compile ?

Ethan•3mo ago

I want it to compile mojo so that the tokenizer and parser can be something useful to the mojo community. But to be clear it will only compile a subset of mojo like C.

sora•3mo ago

You could introduce a match function, replace raise with abort so you can remove raises (almost) everywhere. The errors are never caught anyways, and it doesn't make sense to have funcs returning optionals also raise.

struct TokenGenerator:
    ...

    @staticmethod
    fn `match`[reg: StringLiteral](s: String, pos: Int) -> Optional[String]:
        try:
            var re = Python.import_module("re")
            var name = re.compile(reg)
            return str(name.`match`(s, pos).group(0))
        except e:
            return None

struct TokenGenerator:
    ...

    @staticmethod
    fn `match`[reg: StringLiteral](s: String, pos: Int) -> Optional[String]:
        try:
            var re = Python.import_module("re")
            var name = re.compile(reg)
            return str(name.`match`(s, pos).group(0))
        except e:
            return None

Ethan•3mo ago

I do want it to raise a TokenError when, for example, there is unmatch parens. Nice suggestion tho.

sora•3mo ago

If your error is not resumable, you could use abort for error reporting I guess

Gaming

Programming

Yoho 🔥: A toy compiler written in mojo