I am using dictionary comprehension to compare each dataframe from within a dictionary to its corresponding dataframe in locals(). For some reason, I am getting a key error when accessing the locals() dataframe from within the dictionary comprehension expression.

Here is the setup:

imort numpy as np
imort pandas as pd
df=pd.DataFrame(np.random.randint(3,size=[2,2]),columns=['A','B'])
   #    A  B
   # 0  1  2
   # 1  2  0

# Create dictionary of datafame(s) (only 1 here)
OldDFs={'df':df}

Here is the failure when my dictionary comprehension expression refers to a locals() dataframe:

#####################################################
#          This is the "key test expression"
#####################################################
{ ky:locals()[ky] for (ky,OldDF1) in OldDFs.items() }
   # Console message: KeyError 'df'

The key ky is of the correct type:

{ ky:type(ky) for (ky,OldDF1) in OldDFs.items() } # {'df': str}

It works outside of dictionary comprehension:

locals()['df'] # Prints "df"

The dictionary comprehension expression works if I refer to a dictionary other than locals():

{ ky:OldDFs[ky] for (ky,OldDF1) in OldDFs.items() } # Prints "df"

What is wrong with the “key test expression” above?

Background: In the actual situation, OldDFs consists of many dataframes read from a pickle file. The actual test I want to perform is

{ ky:OldDF1.equals(locals()[ky])
   for (ky,OldDF1) in OldDFs.items() }