indentation-based blocks

[w1kee]

hi, i am making a parser and i want to have significant indentation in my language. i have seen the other posts about indentation, but i'm not sure how to do something like this.
example of language:

foo
    bar

i want to be able to parse this example like this:

file = { "foo" ~ startblock ~ "bar" ~ endblock }
...

is this possible, if yes, how?

[tomtau]

see e.g. https://github.com/thomas9911/python_parser/blob/master/src/grammar.pest#L32 ?

[KSXGitHub]

This doesn't work. It cannot parse two subblocks inside a parent block.

[alensiljak]

I think you need to explicitly mark the NEWLINE after "foo" and add some indentation before "bar".
I have something similar for Ledger postings here.

[simonbuchan]

This seems to work for me:

start = _ { SOI ~ block* ~ EOI }

block = { id ~ nl ~ body? }
body = _ { indent_start ~ block ~ (indent ~ block)* ~ indent_end }

indent = _ { PEEK[..] }
indent_start = _ { PEEK[..] ~ PUSH(" "+) }
indent_end = _ { DROP }

nl = _ { (" "* ~ NEWLINE)+ | EOI }
id = { ASCII_ALPHA+ }

With input:

a
  b
   c
  d
  
  e
     f

the editor gives:

- block
  - id: "a"
  - block
    - id: "b"
    - block > id: "c"
  - block > id: "d"
  - block
    - id: "e"
    - block
      - id: "f"
      - EOI: ""
- EOI: ""

(don't know why the share button doesn't work for me?)

[KSXGitHub]

What does [..] in PEEK[..] do?

[LeoDog896]

It's covered in https://docs.rs/pest/latest/pest/#advanced-peeking - it reads the tabs bottom to top, as intended in an indentation schema.

[KSXGitHub]

Damn. Documentation is all over the place. I expected the pest book to contain all information I need.