Replace function in dataframe removes more than expected

Question

Suppose I have the following csv file:

Column1;Column2
AB; Test

Which has a NUL byte character in it:

I want to read it in with pandas, manipulate it and export it again. I want to remove NUL byte characters, my code is as follows:

import pandas as pd
df=pd.read_csv(r'C:foldertest.csv', sep=";", encoding="utf-8")
df=df.replace('x00','', regex=True) 
df.to_csv(r'C:foldertestoutput.csv', sep=";", index=False, encoding="utf-8")

I expect that only the NUL byte character is removed, however, more is removed:

So it removes the whole "cell", also the "Test". But I don’t want this.

Why, where is my mistake? How can I only remove the NUL byte character?

Asked By: PSt

||

Source

Answer 1

The issue is not coming from the replace, but rather from the fact that the NULL character is handled as a line terminator by read_csv.

What you can do it manually remove the NULL before passing it to read_csv:

import io

f = io.StringIO(open(r'C:foldertest.csv').read().replace('x00', ''))

df = pd.read_csv(f, sep=";", encoding="utf-8")

Output:

  Column1 Column2
0      AB    Test

Alternatively, if you just want to remove the NULL without pandas:

with open(r'C:foldertest.csv', 'r') as f1, open(r'C:foldertest_clean.csv', 'w') as f2:
    for line in f1:
        f2.write(line.replace('x00', ''))

Answered By: mozway

Replace function in dataframe removes more than expected

Question:

Answers: