Remove part of a string from second field

Shell Programming and Scripting
1

I have a file in below format (pipe delimited):

1234__abc|John__abc|xyz
3345__abc|Kate__abc|xyz
55344|Linda__abc|xyz
33434|Murray|xyz

I want to remove any occurence of "__abc" in the second field of this file.
I did some research and found a way to replace the entire second field with another string:

sed 's/^\([^|]*\)|[^|]*|/\1|9999|/'

But I am not able to remove the "__abc" alone in the second field. Any help to do this would me much appreciated.

2 
$ cat fil
1234__abc|John__abc|xyz
3345__abc|Kate__abc|xyz
55344|Linda__abc|xyz
33434|Murray|xyz

Output:

$ awk -F"|" '{sub("__abc","",$2);}1' OFS="|" fil
1234__abc|John|xyz
3345__abc|Kate|xyz
55344|Linda|xyz
33434|Murray|xyz

Guru.

1 Like
3

Thanks a lot. That worked.

4

well if you have already touched sed, you were very close.

sed 's/__abc//2'  file

will give you what you need. I guess the missing part was the "2", right?

5

@sk1418 :

No, your statement would remove the second "__abc" found.
So you would miss the __abc occurrence that appear in the third line because it is the first occurrence in the line.

$ cat tst
1234__abc|John__abc|xyz
3345__abc|Kate__abc|xyz
55344|Linda__abc|xyz
33434|Murray|xyz

$ sed 's/__abc//2' tst
1234__abc|John|xyz
3345__abc|Kate|xyz
55344|Linda__abc|xyz
33434|Murray|xyz

But you could go with this instead (assuming the whole file having the same formatting than the given example) :

$ cat tst
1234__abc|John__abc|xyz
3345__abc|Kate__abc|xyz
55344|Linda__abc|xyz
33434|Murray|xyz

$ sed 's/|\(.*\)__abc|/|\1|/' tst
1234__abc|John|xyz
3345__abc|Kate|xyz
55344|Linda|xyz
33434|Murray|xyz
6

thanks for pointing this out. I didn't notice the special 3rd line. :wink:

your sed 's/|\(.*\)__abc|/|\1|/' tst works great for this example. however also not so generic.

e.g.

kent$  cat a
1234__abc|John__abc|xyz
3345__abc|Kate__abc|xyz
55344|Linda__abc|xyz__abc|xx
33434|Murray|xyz

yours:
kent$  sed 's/|\(.*\)__abc|/|\1|/' a
1234__abc|John|xyz
3345__abc|Kate|xyz
55344|Linda__abc|xyz|xx
33434|Murray|xyz

I made one with sed, it works, however don't know if it is the best solution with sed.

kent$  sed -r 's/\|/\x034/2;s/__abc\x034/|/;s/\x034/|/' a
1234__abc|John|xyz
3345__abc|Kate|xyz
55344|Linda|xyz__abc|xx
33434|Murray|xyz