Saving list of many python variables into excel sheet while simultaneously keeping variable types defined?

Question

It's possible with xlsxwriter to save variables to existing excel files and read them after, though the problem is that the variables are stored as strings in my excel file.

Let's say I have a list of many different variables with various types (pd.datetimerange, pd.df, np.arrays, etc.), if I save them to the excel file the variable type would be lost.

Also the Python script is called from my Excel file so I can't change anything in it without writing a VBA script. Which would temporarily close my workbook, dump the chars (with say pickle strings) and reopen it.

Is it possible to retrieve the types from the excel file without writing a parsing function first (which would take the excel string and yield me with the equivalent variable type)?

Answer 1

As per your comment, you can get eval to correctly process the symbols that are local to some module by passing the appropriate dict of locals into eval, along with your string. Here's a workable solution:

import pandas as pd

def getlocals(obj, lcls=None):
    if lcls is None: lcls = dict(locals().items())

    objlcls = {k:v for k,v in obj.__dict__.items() if not k.startswith('_')}
    lcls.update(objlcls)

    return lcls

x = "[123,DatetimeIndex(['2018-12-04','2018-12-05', '2018-12-06'],dtype='datetime64[ns]', freq='D')]"
lcls = getlocals(pd)

result = eval(x, globals(), lcls)
print(result)

Output:

[123, DatetimeIndex(['2018-12-04', '2018-12-05', '2018-12-06'], dtype='datetime64[ns]', freq='D')]

As a Responsible Person, it is also my duty to warn you that using eval for your application is ridiculously unsafe. There are many discussions of the dangers of eval, and none of them suggest there's a way to completely mitigate those dangers. Be careful if you choose to use this code.