case insensitive bool parsing

[changhiskhan]

From mailing list:

This is something that surprised me.
When reading from csv the values "True" and "False" are converted to bool, while "TRUE" and "FALSE"' to object
Using a converter {...: np.bool} I could take it into account.
Is it useful to take more possibilities into account knowing that there are a lot of possibilies True/False, Yes/No, ....?

In [19]: data = '''A;B
....: True;False
....: False;True'''
In [20]: df = pd.read_csv(StringIO(data), sep=';')
In [22]: df.dtypes
Out[22]:
A bool
B bool

In [23]: data = '''A;B
....: TRUE;FALSE
....: FALSE;TRUE'''

In [24]: df = pd.read_csv(StringIO(data), sep=';')

In [25]: df.dtypes
Out[25]:
A object
B object

In [27]: df = pd.read_csv(StringIO(data), sep=';', converters={'A':np.bool})

In [28]: df.dtypes
Out[28]:
A bool
B object

[moleary]

Hi,

I'm new to pandas, but I'd like to get involved. Is this something I could take a crack at?

Mark

[changhiskhan]

Absolutely. If you haven't already done so, a good starting point would be to read the pandas developer page, the pandas examples on read_csv etc, and then dive into the pandas.io.parsers module.
We look forward to your pull request!

[moleary]

Okay I'll take a look at it. Thanks.