error details are misaligned when using tabs

[srenatus]

Actual Behavior

With demo.rego as

package demo

foo {
	as
}

where the line with as starts with a tab character, and I run opa run demo.rego, I find:

error: 1 error occurred during loading: demo.rego:5: rego_parse_error: no match found
                as
          ^

Expected Behavior

The pointer ^ should respect the tabs or remove them -- right now, it's position is calculated without accounting for the width of the characters that make up the offset here.

[tsandall]

Thanks for filing this. I noticed it a while ago but failed to open a ticket 😅

In this case, the error is reported by the parser. The ^ is generated here: https://github.com/open-policy-agent/opa/blob/master/ast/parser_ext.go#L752 (not in check.go).

I was suspicious of the location you linked to but upon reflection, I don't think there's a problem. The padding is constructed based on the String representation of the ref -- which will not contain tabs.

I haven't thought about the best way to resolve the issue in parser_ext.go. Perhaps we could construct the padding with a mix of spaces & tabs?

[srenatus]

I was suspicious of the location you linked to but upon reflection, I don't think there's a problem. The padding is constructed based on the String representation of the ref -- which will not contain tabs.

Yup, indeed, I was pointing at the wrong thing; I've too noticed that when I was attempting to fix this 😄

I think the real issue is somewhere in here, as there's nothing accounting for the fact that a char could be "wider" than 1 char when printed...

Maybe this could be fixed bluntly by a pre-processing step that replaces tabs by n spaces (for some n) before feeding it into the parser? 🤔