namedcsv.py

''' create object using numpy to read CSV. creates one variable per column
so you can use f.col_name instead of f.data['col_name']

f = NamedCSV(filename)
f.data # all data in the file in an np recarray
f.mass  
f.data['mass'] # same as previous line

f.extract(f.mass > 180)
'''

class NamedCSV(object):
    def __init__(self, name, **kwargs):
        self.data = np.genfromtxt(name, delimiter=',', names=True, unpack=True, **kwargs)
        self.unextract()

    @property
    def names(self):
        """ names of all columns"""
        return self.data.dtype.names


    def extract(self, match):
        """match is a boolean array used to select out elements in each of the
        fields.

        Example:
        b = NamedCSV('foo.csv')
        match = b.col1 > 0
        b.extract(match)

        Now all the fields (not just col1) only have values when col1 > 0.
        self.data is uneffected, so, in general len(self.data) != len(self.col_name)
        after the call.

        Call unextract() to undo this operation
        """

        for field in self.data.dtype.names:
            setattr(self, field, self.data[field][match])

    def unextract(self):
        for field in self.data.dtype.names:
            setattr(self, field, self.data[field])