如何使用sed或awk处理指定列的内容？

Question

I have such data in text.

2017-08-07 733 AA1(10.7.21.51) AllUsers 631 K:N　
2017-08-07 733 AA1(10.7.21.51) AllUsers 631(Peter) 1:N　
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208(Lucy) 2:C
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 K:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 99999(Kate) 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631(Peter) 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631(Peter) 2:C
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999(Kate) T:U　
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999(Kate) 3:U　

There are 6 columns in this text.Use space to separate each column of data.

I want to deal with the 5th column of data.

Some data contains names in the 5th column. There are more parentheses outside the data containing names. There are only numbers in the columns without names. This number is the employee number. I just want to take out the numbers in the 5th column, not name. I want this effect.

2017-08-07 733 AA1(10.7.21.51) AllUsers 631 K:N　
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 2:C
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 K:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 99999 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999 T:U　
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999 3:U　

I use this command to process data.

 cat mytextfile|sed 's/(/ /g' > resultfile 

But the 3rd column has also been modified, because the 3rd column also contains parentheses. In fact, I just want to process the 5th column.

What should I do with sed or awk?

Answer 1

Using sed (simple)

To remove all parens that contain only letters, try:

$ sed 's/([[:alpha:]]*)//' myfile
2017-08-07　733 AA1(10.7.21.51) AllUsers 631 K:N　
2017-08-07　733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07　733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07　733 AA1(10.7.21.51) AllUsers 2208 2:C
2017-08-07　733 AA1(10.7.21.51) AllUsers 2208 K:C
2017-08-07　189 AA2(10.7.4.54) AllUsers 99999 2:C
2017-08-07　189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07　189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07　733 AA3(10.7.21.51) AllUsers 99999 T:U　
2017-08-07　733 AA3(10.7.21.51) AllUsers 99999 3:U　

([[:alpha:]]*) matches ( followed by zero or more alphabetic characters followed by ) . s/([[:alpha:]]*)// locates those characters and replaces them with an empty string.

Using sed (improved)

This removes parenthesized expressions of alphabetic characters from the fifth field and only the fifth field:

$ sed -E 's/(([^[:space:]]+[[:space:]]+){4}[^[:space:]]*)\([[:alpha:]]*\)/\1/' myfile
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 K:N
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 2:C
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 K:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 99999 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999 T:U
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999 3:U

Using awk

To remove any parenthesized expression in the fifth field:

$ awk -F'[[:space:]]+' '{gsub(/\(.*\)/, "", $5)} 1' myfile
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 K:N
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N
2017-08-07 733 AA1(10.7.21.51) AllUsers 631 1:N
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 2:C
2017-08-07 733 AA1(10.7.21.51) AllUsers 2208 K:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 99999 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07 189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999 T:U
2017-08-07 733 AA3(10.7.21.51) AllUsers 99999 3:U

How it works:

1. -F'[[:space:]]+'

   This tells awk to use any sequence of unicode-recognized whitespace as the field separator. (The default is only to recognize sequences of blank, tab, and newline as field separators.)

2. gsub(/\\(.*\\)/, "", $5)

   This looks in the fifth field, $5 , for any parenthesized expression, \\(.*\\) , and replaces it with the empty string "" .

3. 1

   This is shorthand which tells awk to print the line.

Answer 2

In your example's specific case, the user names follow only digits, which is different than in column 3, where the IP address in parentheses follows letters and digits. You can leverage this to your advantage:

$ sed 's/\( [0-9][0-9]*\)([^)]*)/\1/g' mytextfile 
2017-08-07　733 AA1(10.7.21.51) AllUsers 631 K:N　
2017-08-07　733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07　733 AA1(10.7.21.51) AllUsers 631 1:N　
2017-08-07　733 AA1(10.7.21.51) AllUsers 2208 2:C
2017-08-07　733 AA1(10.7.21.51) AllUsers 2208 K:C
2017-08-07　189 AA2(10.7.4.54) AllUsers 99999 2:C
2017-08-07　189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07　189 AA2(10.7.4.54) AllUsers 631 2:C
2017-08-07　733 AA3(10.7.21.51) AllUsers 99999 T:U　
2017-08-07　733 AA3(10.7.21.51) AllUsers 99999 3:U　

That sed line says to capture a space followed by one or more numbers, followed by parentheses with anything inside, replacing all of that with the value captured.