CODEWITHSUNDEEP
regexlinuxsed
dr4gonwidj4j44
dr4gonwidj4j44

Reputation: 11

SED to delete C Program comments

I need to delete the comment line in a C program with sed in linux, assuming that each comment line contains the start and end tokens without any other statements before and after.

For example, the code below:

/* a comment line in a C program */
printf("It is /* NOT a comment line */\n");
x = 5; /* This is an assignment, not a comment line */
[TAB][SPACE] /* another empty comment line here */
/* another weird line, but not a comment line */ y = 0;

becomes

printf("It is /* NOT a comment line */\n");
x = 5; /* This is an assignment, not a comment line */
/* another weird line, but not a comment line */ y = 0;

I know that this regex 

^\s?\/\*.*\*\/$

matches the lines that I need to delete. However, the following command:

sed -i -e  's/^\s?\/\*.*\*\/$//g' filename

does not do the trick.

I am not too sure what I am doing wrong...

Thanks for your help.

Upvotes: 1

Views: 3782

Answers (3)

John1024
John1024

Reputation: 113994

This does it:

$ sed -e  '/^\s*\/\*.*\*\/$/d' file
printf("It is /* NOT a comment line */\n");
x = 5; /* This is an assignment, not a comment line */
/* another weird line, but not a comment line */ y = 0;

Notes:

  1. ^\s? matches zero or one spaces. It looks like you want to match zero or one or more spaces. So, we use instead ^\s*.

  2. Since you want to delete the lines rather than replace them with empty lines, the command to use is d for delete.

  3. It is not necessary to delimit a regex with /. We can use |, for example:

    sed -e  '\|^\s*/\*.*\*/$|d' file

    This eliminates the need to escape the /. Depending on how many times / appears in a regex, this may or may not be simpler and clearer.

Upvotes: 2

Ed Morton
Ed Morton

Reputation: 204558

This might be what you're looking for:

$ awk '{o=$0; gsub(/\*\//,"\n"); gsub(/\/\*[^\n]*\n/,"")} NF{print o}' file
printf("It is /* NOT a comment line */\n");
x = 5; /* This is an assignment, not a comment line */
/* another weird line, but not a comment line */ y = 0;
/* first comment */ non comment /* second comment */

The above was run on this input file:

$ cat file
/* a comment line in a C program */
printf("It is /* NOT a comment line */\n");
x = 5; /* This is an assignment, not a comment line */
     /* another empty comment line here */
/* another weird line, but not a comment line */ y = 0;
/* first comment */ non comment /* second comment */

and uses awk because once you're past a simple s/old/new/ everythings easier (and more efficient, more portable, etc.) with awk. The above will delete any empty lines - if that's a problem then update your sample input/output to include that but it's a easy fix.

Upvotes: 1

Joao Vitorino
Joao Vitorino

Reputation: 3266

What you are doing is replacing your regex with a empty string

sed -i -e  's/^\s?\/\*.*\*\/$//g' filename

that means

sed -i -'s/pattern_to_find/replacement/g' : g means the whole file.

What you need to do is delete the line with the regex

sed -i -e  '/^\s?\/\*.*\*\/$/d' filename

Upvotes: 0

Related Questions