Use regex pattern to replace numbers followed by a substring or numbers followed by a space and then substring

user_10583056 0
Viewed 768

For a column in a pandas dataframe, I want to remove any number either immediately followed by "gb" or "mb" or with a space in-between, in it's entirety. I.e. remove strings such as "500 gb" and "500mb". 

    Column_To_Fix
0   coolblue 100gb
1   connector 500 mb for thing
2   5gb for user
3   load 800 mb
4   1000 add-on
5   20 gb

The function below only works for row 0 and row 2, not sure how to add in the space requirement for the pattern:

pat = '(^|\s)\d+(gb|mb)($|\s)'
df['Column_To_Fix'].str.lower().replace(pat, ' ', regex=True)

Desired Output:

    Column_To_Fix
0   coolblue
1   connector for thing
2   for user
3   load
4   1000 add-on
5
regexp-replace pandas python regex
2 Answers

Try this pattern 

pat = '\d+ *(gb|mb)'

df['Column_To_Fix'].str.lower().str.replace(pat, ' ')

Out[462]:
0               coolblue
1    connector   for thing
2                 for user
3                   load
4              1000 add-on
5
Name: Column_To_Fix, dtype: object

If you prefer series.replace

df['Column_To_Fix'].str.lower().replace(pat, ' ', regex=True)
user_101892140

I had assumed the text was (no line numbers):

coolblue 100gb
connector 500 mb for thing
5gb for user
load 800 mb
1000 add-on
20 gb

and that the desired result (which maintains proper alignment and spacing) was:

coolblue
connector for thing
for user
load
1000 add-on

with an empty string on the last line. That can be achieved by replacing matches of the following regular expression with empty strings (using re.sub).

r'(?:^\d+ ?[gm]b | \d+ ?[gm]b(?= |$))'

Demo

user_2569700
Related