Tokenize a source reading Python code as unicode strings. This has the same API as tokenize(), except that it expects the *readline* callable to return str objects instead of bytes.
tokenize/generate_tokens
generate_tokens