rego_parse_error with unexpected / confusing message and location

[anderseknert]

If you can spot the error in the below function — good eyes! In fact, even the Rego parser fails to identify the location here, which is the call to time.diff and the incorrect parens used on the next last line.

within_time_slot(slot) {
    slot_split := split(slot, ":")
	
    slot_hour := to_number(slot_split[0])
    slot_minute := to_number(slot_split[1])
    
    now_rfc3339 := time.parse_rfc3339_ns(to_rfc_3339(datetime[0], datetime[1], datetime[2], datetime[3], datetime[4]))
    slot_rfc3339 := time.parse_rfc3339_ns(to_rfc_3339(datetime[0], datetime[1], datetime[2], slot_hour, slot_minute))
    
    # The date has already been determined - check only diff in hours and minutes
    [_, _, _, diff_hours, diff_minutes, _] := time.diff(now_rfc3339), slot_rfc3339))
    
    diff_hours * 60 + diff_minutes < 90
}

The error message reported from this however is:

1 error occurred: policy.rego:32: rego_parse_error: unexpected assign token: non-terminated set
	    slot_split := split(slot, ":")
	               ^

Full playground example here: https://play.openpolicyagent.org/p/IeLICFb2TD

[philipaconrad]

I'd be interested in working on this issue and a few others like it once I clear out more of my work backlog. Better error reporting is always useful!

[philipaconrad]

I'm adding a close-to-minimal test case to this thread that exercises the same confusing behavior:

package test

p {
    x := 5
    y := contains("a"), "b")
}

Error message:

1 error occurred: policy.rego:4: rego_parse_error: unexpected assign token: non-terminated set
	    x := 5
	      ^

EDIT: This case can actually be reduced further:

p {
    y := contains("a"), "b")
}