将字符串作为数据帧访问并在 python 循环中迭代地重新定义 object

Question

for i in dataframe_list:
    i=eval(i)
    for num in range(1,len(dataframe_list)):
       
       for column in [column for column in eval(i).columns if column not in eval(dataframe[num]):
             eval(i)= eval(i).withcolumn(column, lit=none)
       for column in [column for column in dataframe.columns if column not in dataframe2]:
             eval(dataframe[num])= eval(dataframe[num]).withcolumn(column,lit=none)

Dataframe_list is a list of the names of the dataframes. The problem is python doesn't recognise the strings as the object. So I use the eval() function. On the first with column containing the eval(i).withcolumn statement, I get an error saying saying eval must contain a string, bytecode or code object. To my knowledge i is the index of the list and is clearly a string. Can anyone help me get this to work? eval can't be used in a function/function call. I've tried exec() etc. How do I do this. Just need to iteratively redefine the dataframe. Can't do that if there is an eval on each side. Python says each i in dataframe_list is a pyspark dataframe but when i run the code either i or dataframe_list[z] comes up as a str that has no attribute columns. ?? even though z is i+1 index for accessing the list entries....so if dataframe_list[i] is a df then dataframe_list[z] must also be one.... Any ideas?

Answer 1

You have already done eval(i) in the outer for loop so there is no need to do eval(i) in the nested for loop again. Also, I believe eval(dataframe[num]) should be eval(dataframe_list[num]).columns ?

for i in dataframe_list:
    i=eval(i)
    for num in range(1,len(dataframe_list)):
       # now use i not eval(i) like this...
       # i = i.withcolumn(column, F.lit(None))

Answer 2

I decided to create another list with data frame types exclusively instead of evaluating the initial list. So i added a loop:

for i in dataframestr:
    i=eval(i)
    newdataframelist.append(i)

This gets around the issue of eval.