Parsing of **12/31/2014** date in csv fails

[ranjanprj]

The parsing of following date format in csv is failing
12/31/2014

This I'm guessing is due to following lines in
dateparser.py ->

def create_date_formats(day_first=True):
    """generate combinations of time and date
    formats with different delimeters
    """

    if day_first:
        date_formats = ['dd/mm/yyyy', 'dd/mm/yy', 'yyyy/mm/dd']
        python_date_formats = ['%d/%m/%Y', '%d/%m/%y', '%Y/%m/%d']
    else:
        date_formats = ['mm/dd/yyyy', 'mm/dd/yy', 'yyyy/mm/dd']
        python_date_formats = ['%m/%d/%Y', '%m/%d/%y', '%Y/%m/%d']

The day_first is always true so it never takes mm/dd/yyyy format.
Request to have a look.

Regards

[ThrawnCA]

It's actually a default value for day_first, not forced. But yes, the code currently uses the default. What would you have it do instead?

[dragondave]

If there's a spreadsheet which contains strings which are unambiguously m/d/y and no/very few strings that are unambigously d/m/y, then it'd be reasonable to parse it as m/d/y by default.
(anything with a first number <13 and a second number >12/<32 is unambiguously m/d/y)
That'd mean parsing the sheet twice, though, and would lead to issues if there were only sometimes unambiguous m/d/y dates across a range of consistently m/d/y sheets.

There's also no obvious way to override messytable's behaviour; whilst I suspect something like messytables.types.DateType.formats = messytables.dateparser.create_date_formats(day_first=False) would work, it's not a nice, public, documented override; being able to configure it somewhere at a high level would make sense. Not sure where the right place would be.