how to look back values of previous row for select cases only in pandas

mluser :

My data looks like below.

   col_1  col_2
      1        1
      1        1
      p        0
      1        1
      n        2
      n        2
      p        0
      p        0

I want to calculate values in col_2 from col_1. The logic that i want to apply is: When col_1 value ='p', replace value in col_2 from the previous row's value of col_2 and other values in col_2 does not change for any other value of col_1 and the final expected output is as following:

   col_1  col_2
      1        1
      1        1
      p        **1**
      1        1
      n        2
      n        2
      p        **2**
      p        **2**

I am calculating these columns in addition to others based on a date in the assign() function. This is the only step that I am not able to figure out. Because in this one I am looking at previous row's value; so shift() could work but I only need to look for previous value for col_2 when col_1 ='p'. For the time being I am doing it via for loop that gives me the flexibility to look back 1 row and check/replace value. Because of for loop this is not an efficient solution.

Do you know how to avoid a for loop and do it in more pandas way ?

Quang Hoang :

You can use mask then ffill:

df['col_2'] = df['col_2'].mask(df['col_1']=='p').ffill()