Myujth

I have conditions in a file, that has around 10 thousend lines. As shown below.

COLUMN_NAME NOT IN 1234534
COLUMN_NAME1 NOT IN 34252
COLUMN_NAME_2 not in 67496575
COLUMN_NAME NOT in 1234534
foo COLUMN_NAME NOT IN (1234534,453535) rest of the line
COLUMN_NAME NOT IN 1234534,453535
columnsd not in (23123124232,6464777) rest on the line
COLUMN_NAME NOT IN 1234534
COLUMN_NAME NOT IN 1234534
fdfsdf COLUMN_NAME not in 1234534
COLUMN_NAME not in 1234534
column NOT IN (6764577,434545)
COLUMN_NAME not in (1234534)

I want to use the sed command to add the brackets after the IN clause. I want to replace the conditions above result in output like below.

COLUMN_NAME NOT IN (1234534)
COLUMN_NAME1 NOT IN (3422)
COLUMN_NAME_2 not in (67496575)
COLUMN_NAME NOT in (1234534)
COLUMN_NAME NOT IN (1234534,453535) rest of the line
COLUMN_NAME NOT IN (1234534,453535)
columnsd not in (23123124232,6464777) rest on the line
COLUMN_NAME NOT IN (1234534)
COLUMN_NAME NOT IN (1234534)
fdfsdf COLUMN_NAME not in (1234534)
COLUMN_NAME not in (1234534)
column NOT IN (6764577,434545)
COLUMN_NAME not in (1234534)

This would be what you are looking for

sed -i .bak 's/(sins+)([^() ]+)/1(2)/i' file

Explanation

s/regexp/replacement/flags tries to match input line against regexp, and if it manages, replaces the matched part with replacement.

1. 
   

   regexp part:

   
   
   
   • 
     (sins+) (ERE: (sins+)) matches a space followed by in plus one or more spaces, and keeps the matched portion in capture group 1,
   
   
   • 
     ([^() ]+) (ERE: ([^() ]+)) matches a group of non-space, non-parenthesis chars and keeps it in capture group 2,
   
   


2. 
   

   replacement part:

   
   
   
   • 
     1(2) expands up to <capture group 1>(<capture group 2>),
   
   


3. 
   

   flags part:

   
   
   
   • 
     i makes the matching case-insensitive.
   
   

Lets take it step by step.

I guess the best rule would be:

• IN (case insensetive) IN
  


• followed by: a number [0-9]+
  


• optionally followed by multiple (*): comma and number ,[0-9]
  

This would result in the regex:
IN[0-9]+(,[0-9]+)*

Next step is to add mandatory (s) and optional (s*) whitespace:

sINs*[0-9]+s*(,s*[0-9]+s*)*

Now to replace sINs* [0-9]+s*(,s*[0-9]+s*)* with 1(2) it you need capture groups.

(sINs*)([0-9]+s*(,s*[0-9]+s*)*)

Now the regex has to go inside s//1(2)/i (the i means case insensitive)

s/(sINs*)([0-9]+s*(,s*[0-9]+s*)*)/1(2)/i

And at last we have a shell command, including a backup in case of mistakes.

sed -i.bak 's/(sINs*)([0-9]+s*(,s*[0-9]+s*)*)/1(2)/i' filename

I tested it on your example data and the output is:

COLUMN_NAME NOT IN (1234534)
COLUMN_NAME1 NOT IN (34252)
COLUMN_NAME_2 not in (67496575)
COLUMN_NAME NOT in (1234534)
foo COLUMN_NAME NOT IN (1234534,453535) rest of the line
COLUMN_NAME NOT IN (1234534,453535)
columnsd not in (23123124232,6464777) rest on the line
COLUMN_NAME NOT IN (1234534)
COLUMN_NAME NOT IN (1234534)
fdfsdf COLUMN_NAME not in (1234534)
COLUMN_NAME not in (1234534)
column NOT IN (6764577,434545)
COLUMN_NAME not in (1234534)