BUG: <foo>[0].<bar> raises ValueError

[TimScheckenbach]

Using the persons.fan and the constraint str(<start>[0].<last_name>..<lowercase_letter>) == "c"; raises a ValueError, but this should be valid.

[addisoncrump]

How would this work? Is this a forall or exists constraint on lowercase letter?

[andreas-zeller]

Fixed in andreas branch. Added support for .. and [; improved symbol recognition.

[andreas-zeller]

Output is Locisg Eccccc,50 as expected

[TimScheckenbach]

But this now only checks if any child of <start> has a child <last_name>, or am I missing something? I'd rather expected that Fandango should check if the first child of <start> has <last_name>. Now str(<start>[0].<digit>) == "1", doesn't raise an error, even though this shouldn't be a valid constraint, because <person_name> doesn't have a child <digit>.

[TimScheckenbach]

How would this work? Is this a forall or exists constraint on lowercase letter?

It's a forall constraint on lowercase letters in the last name.

[andreas-zeller]

Yes, <parent>[...].<symbol> is now handled like <parent>..<symbol>, i.e. simply checking whether there is a <symbol> at the end of the chain. We could also interpret the actual [...] contents here, but slices ([a:b]) and chains ([0][1][2]) make this somewhat risky to code correctly. I have left a TODO comment in the parse.py code.