John Machin wrote: > [EMAIL PROTECTED] wrote: > > > > >>> excel_date = 38938.0 > > > > >>> python_date = datetime.date(1900, 1, 1) + > > > > >>> datetime.timedelta(days=excel_date) > > > > >>> python_date > > > > datetime.date(2006, 8, 11) > > > > > > Err, that's the wrong answer, isn't it? Perhaps it shoud be > > > datetime.date(1900, 1, 29)? > > > > Actually was about to post same solution and got same results. (BTW > > Simon, the OP date is Aug 9th, 2006). Scratched head and googled for > > excel date calculations... found this bug where it treats 1900 as leap > > year incorrectly: > > http://www.ozgrid.com/Excel/ExcelDateandTimes.htm > > > > Plus it treats 1 jan 1900 as day 1, not 0 so just subtract 2 in the > > calc: > > >>>python_date = datetime.date(1900, 1, 1) + > > >>>datetime.timedelta(days=excel_date - 2) > > >>> python_date > > datetime.date(2006, 8, 9) > > > > ... and 2006-08-09 is only the correct answer if the spreadsheet was, > when last saved, using the 1900 ("Windows") date system, not the 1904 > ("Macintosh") date system.
John, Just for me own curiosity, is this Excel concept of date numbers same as the OLE concept (see http://msdn2.microsoft.com/en-us/library/82ab7w69.aspx or search "MFC DATE" on MSDN). I put in some test cases for conversion code here: http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/496683 But would be interested to add any additional info on variations on this theme. Cheers, Giles -- http://mail.python.org/mailman/listinfo/python-list
