Indent-based syntax is relatively simple to parse. You basically need two pieces of state: are you in indent-sensitive mode (not inside a literal, not inside a

Indent-based syntax is relatively simple to parse. You basically need two pieces of state: are you in indent-sensitive mode (not inside a literal, not inside a parenthesized expression), and what indentation did the previous line have. Then you can easily issue INDENT and DEDENT tokens, which work exactly like "{" and "}". The actual Python parser does issue these tokens. Actually Haskell has both indent-based and curlies-based syntax, and curlies freely replace indentation, and vice versa (but only as pairs).