True alternative to exec() in python

Question

So, I have been trying to execute a concatenated string in Python 3 and it worked. This function would eventually go into a (while) loop, where the loop counts up to the current month.

def all_installs_to_dataframe(month):
 '''Imports csv's from different months to pandas dataframes'''
 command = ["dataset = pd.read_csv('/path/file2017", "_overview.csv', sep=',', header=0, encoding='utf-16')"]
 return (command[0] + month + command[1])
exec(all_installs_to_dataframe("05"))

However, on my extensive research on Stackoverflow it seems like exec() is not liked (except for trustworthy input). Hence, I would like to know how to do this better? I found that many are fond of dictionaries, I tried this and it would give me a string as well:

dict = {'start': "dataset = pd.read_csv('/path/file2017", 'month': "05", 'end': "_overview.csv', sep=',', header=0, encoding='utf-16')"}
exec(dict['start']+dict['month']+dict['end'])

Which I would have to exec() again. So, how could I do this without using the exec() statement?

@nico: I initially had this loop over all my files (with datetime library):

month_date = 3
while month_date - 1<date.today().month:
 dataset = "dataset_%s" % (month_date)
 function = "pd.read_csv('/path/file2017%s_overview.csv', sep=',', header=0, encoding='utf-16')" % (month_date)
 dataset = function
 month_date += 1

Then I realized that I get a string and it won't work. So I am using the exec() function now. But everywhere I read I shouldn't use it, so how could I circumvent the exec() function in this loop?

• nico: Is there a reason you're not just sending the csv path directly? Are the commands diverse enough that this will not work? I'm trying to better understand what you want to do.

Answer 1

Why not executing the command in the function and compose the filename using str.format instead of using exec ?

def all_installs_to_dataframe(month):
 '''Imports csv's from different months to pandas dataframes'''
 dataset = pd.read_csv('/path/file2017{}_overview.csv'.format(month), sep=',', header=0, encoding='utf-16')

• ccasimiro9444: Merci for your answer Jean-Francoise Fabre. I tested it and this works perfectly fine. +1, when I get enough reputation.
• Jean-François Fabre: @ccasimiro9444 you can accept the answer already. That doesn't require any privileges :)
• ccasimiro9444: Accepted it, I am new to this :)
• ccasimiro9444: oh yeah! :D and +1 for your great answer