[Jython-dev] CSV: Method Parsing

[Paul D. <pau...@ds...>]

Hi Guys,

I have been spending what little spare time I have looking at the csv 
module.  I'm afraid I haven't got very far yet as I am currently smack 
bang in the middle of a very steep learning curve!  I am making progress 
however and am still as keen as ever!

Currently, I am trying to understand the correct way to parse the 
arguments for csv.reader().  The method's definition from the python 
library docs is:

reader(csvfile[, dialect='excel'[, fmtparam]])

As csvfile is mandatory I thought it would be okay to specify it 
explicitly in my method signature, then use optional arguments for the 
rest.  With a little help from the Wiki FAQ I ended up with this:

public static PyObject reader(PyObject iterableObject, PyObject[] args, 
String[] keywords) {
}

But here is where I get confused because now Jython *EXPECTS* 3 
arguments which is not how Python behaves:

 >>> import csv
 >>> csv.reader()
Traceback (most recent call last):
   File "<stdin>", line 1, in ?
TypeError:  expected at least 1 arguments, got 0

jython:
 >>> import csv
 >>> csv.reader()
Traceback (innermost last):
   File "<console>", line 1, in ?
TypeError: reader(): expected 3 args; got 0
 >>>

So to make csv.reader() behave the same as python, I'd need to pass an 
array of PyObjects into reader() then use ArgParser and do my own type 
checking and error handling:

public static PyObject reader(PyObject[] args) {
    //Use ArgParser and my own error handling to throw if there isn't
    //at least one argument + other checks.
}

I guess it's easier to explicitly specify arguments when the module 
hides behind a python wrapper then you already know the number of 
arguments the wrapper uses.  I can't do this for csv as the reader() 
method is exposed directly through the csv.py wrapper so it must support 
dynamic arguments.

Does all of this make sense or am I getting it completely wrong?

Cheers,
Paul.