Mutate rows based on maching to user defined strings that works universally

Question

I have a data like this

 clas=c("CD_1","X.2_2","K$2_3","12k3_4",".A_5","xy_6")
 df <- data.frame(clas)
> df
    clas
1   CD_1
2  X.2_2
3  K$2_3
4 12k3_4
5   .A_5
6   xy_6

and I would like to change some rows that match this condition

if the strings after _ are 4,5 and 6 replace the strings before the _ with string B. So the output should like this;

    clas
1   CD_1
2  X.2_2
3  K$2_3
4 12kB_4
5   .B_5
6   xB_6

Thanks!

EDIT::

SO If I have data like this:

    clas
1   CD_1
2  X.2_2
3  K$2_3
4 12k3_4
5   .A_5
6  xy_11

Then applying your solution,

df %>% mutate(clas = str_replace(clas, "(.)(_[4511])", "B\\2"))

    clas
1   CB_1
2  X.2_2
3  K$2_3
4 12kB_4
5   .B_5
6  xB_11

But I only want to match 11 not 1. How can we do that ?

Answer 1

library(dplyr)
library(stringr)

clas <- c("CD_1","X.2_2","K$2_3","12k3_4",".A_5","xy_6")
df <- data.frame(clas)

df %>% mutate(clas = str_replace(clas, "(.)(_[456])", "B\\2"))

Here putting the matching pattern creates a match with 3 groups, the first containing the whole expression match ._[456], the second containing the . part and the third containing the _[456] part.

\\2 accesses the third group (0 indexing) and so you replace the whole pattern ._[456] with B followed by whatever matched _[456] where [456] is a character matching any of the options inside the brackets.

EDIT:

Each character inside of [] is treated individually, so [1111] is no different from [1] because that pattern only matches a single character that is either a 1 or 1 or 1 or 1. Instead you need to use | so you have (.)(_[45]|_11). This matches _4 or _5 or _11 in the second pattern group. Also if you want to match 1-9 but not 11 or 15 you need to use (.)(_[45])$ where $ is the end-of-string indicator. Go look at the cheatsheet and test these out on RegExr.