URI Parse: IndexError: list index out of range about arpeggio HOT 3 CLOSED

You've hit an interesting case of nested empty parses. See unit test in the commit. There was a subtle bug triggered if nothing is matched in nested parses consisting of OrderedChoice and ZeroOrMore.

The authority rule if completely optional (indirectly), meaning that it will match empty strings. I'm not sure is this what you really want, or what RFC suggest?

Get the current master version. It parses now but you will get empty parse tree for illegal inputs. That is not what you want I guess. It is usually a good idea to terminate your grammar root rule with EOF match. That means that you must be at the end of your input when done parsing. If that is not the case than your have superfluous characters in the input and NoMatch should be raised.

def authority():
    return Optional(userinfo, '@'), host, Optional(':', port), EOF

from arpeggio.

I want to implement the full grammar of several RFCs so it's not possible to append EOF to authority. I also wondered how authority [1], host [2] and reg-name [3] are defined. Anyway, glad to see it's solved in the current master.

Kind regards!

[1] https://tools.ietf.org/html/rfc3986#section-3.2
[2] https://tools.ietf.org/html/rfc3986#section-3.2.2
[3] https://tools.ietf.org/html/rfc3986#page-21

from arpeggio.

Ah, right. authority is not your grammar root rule. Anyway, when you finish your grammar your top level rule should be something like:

def grammar(): return URI, EOF
def URI(): return ....

Now, your real RFC grammar starts from URI but your root rule used in ParserPython will include EOF at the end to signify that you are interested in parsing the whole input always. This will prevent successful parses for URIs that have superfluous characters at the end.

from arpeggio.