mjhjmtu

I have a csv file with below data

Input

"01","5","Male","[""No.**"",""**kg""]","","","" 
"02","6","FeMale","[""No.""]","","",""

I would like to replace the "," string which is between the square brackets but not the data outside the square brackets.

Output:

"01","5","Male","[""No.**,**kg""]","","","" 
"02","6","FeMale","[""No.""]","","",""

I tried below command, but it does replace data outside the square brackets, but not for all lines.

sed '/[/,/]/s/"",""/,/'

sed '/[/,/]/s/"",""/,/'

This would look for a line with a [, then a line with a ], and between those lines, replace the first "","" encountered on each line. It doesn't really look at where the [..] are within a line.

As a zero-order try, something like this:

$ cat x
"01","5","Male","[""No.**"",""**kg""]","","","" 
$ sed 's/([.*)"",""(.*])/1,2/g' x
"01","5","Male","[""No.**,**kg""]","","","" 

The pattern matches a [, anything, "","", anything and a ], while capturing all but the "","" so it can put the pieces back together.

This will break for stuff like [..],"","",[..] (where the brackets close before a "","" is seen, the pattern searches for a following ]) and [.."","".."",""..] (with multiple "","" sequences inside the brackets, only one is removed).

Slightly more genarally with Perl, though this is a horrible substitution-within-substitution trick. You should probably use a proper parser instead:

$ cat y
no removal here: [...],"","",[...]
double removal here: [ "","" "","" ]
[""remove"",""here""],""not"",""here"",[""also"",""here""]

$ perl -pe 'sub x $a = shift; $a =~ s/"",""/,/g; return $a;
 s/([.*?])/ x($1) /eg ' y
no removal here: [...],"","",[...]
double removal here: [ , , ]
[""remove,here""],""not"",""here"",[""also,here""]

(.*? is a non-greedy match, it stops as soon as it can, that is, at the first ] in this case.)

sed '/"Male/s/"",""/,/1' filename

Output:

"01","5","Male","[""No.**,**kg""]","","",""
"02","6","FeMale","[""No.""]","","",""