How to UNCOMMENT a line that contains a specific string using Sed?

Question

The lines in the file :

-A INPUT -m state --state NEW -m tcp -p tcp --dport 2000 -j ACCEPT
-A INPUT -m state --state NEW -m tcp -p tcp --dport 2001 -j ACCEPT
-A INPUT -m state --state NEW -m tcp -p tcp --dport 2002 -j ACCEPT

to comment out let's say the line that contains

i can simply run this SED command:

sed -i '/ 2001 /s/^/#/' file

but now how do i revert back ?

as in uncomment that same line ?

i tried

sed -i '/ 2001 /s/^//' file

that does not work.

Why are you using this technique? The standard technique would be to use one line in the script and make the year number into an argument to the script. You could even default to, say, 2001 if no argument is provided (`${1:-2001}`), or use an environment variable (`${DEFAULT_YEAR:-2001}`), or both (`${1:-${DEFAULT_YEAR:-2001}}`). — Jonathan Leffler, Jul 22 '14 at 14:07

score 137 · Answer 1 · answered Dec 08 '14 at 09:37

137

Yes, to comment line containing specific string with sed, simply do:

sed -i '/<pattern>/s/^/#/g' file

And to uncomment it:

sed -i '/<pattern>/s/^#//g' file

In your case:

sed -i '/2001/s/^/#/g' file    (to comment out)
sed -i '/2001/s/^#//g' file    (to uncomment)

Option "g" at the end means global change. If you want to change only a single instance of pattern, just skip this.

answered Dec 08 '14 at 09:37

JNLK

1,688
2
12
10

1

add the `-e` argument for macOS and BSD – Jonas Feb 25 '17 at 10:55
15

To avoid commenting multiple times and taking into account multiple leading `#`, use the following commands: `sed -i '//s/^#*/#/g' file (to comment out)` and `sed -i '//s/^#*//g' file (to uncomment)` – Louis M May 02 '17 at 11:46
Thanks jnlk and @louis-m. This solution is what worked best for me and accounting for multiple # characters works perfectly. – grt3kl Jul 20 '18 at 13:35
5

And if you want to make sure there is no whitespace at the start of line (e. g. in `# This is commented out` you can use `sed -i '//s/^#*\s*//g' file` – smoothware Jan 03 '19 at 20:10
1

this should be the accepted answer IMHO, well explained. – Viktova May 09 '19 at 10:32
best answer. thk – acgbox Mar 16 '20 at 17:35
Why is it working? Where is `s/pattern/replacement/flags`???? :D – Dims Mar 24 '21 at 10:50
@smoothware this doesn't work if whitespace is before `#`. Instead use: `sed -i '//s/^[#[:space:]]*/#/g' file` to comment out, `sed -i '//s/^[#[:space:]]*//g' file` to uncomment – morrow Jan 25 '22 at 20:32
And use `sed -i_BKP-$(date -I) ...` to make a copy with a timestamp on the fly – Pablo Bianchi Jun 21 '23 at 20:59

Avinash Raj · Accepted Answer · 2014-07-22T14:05:37.660

56

Try this sed command,

sed -i '/^#.* 2001 /s/^#//' file

edited Jul 22 '14 at 14:05

answered Jul 22 '14 at 13:55

Avinash Raj

172,303
28
230
274

2

why `.` that will remove the first character even if its not a comment ? – Jul 22 '14 at 13:58
3

still needs to be fail proof. this answer needs corrections if it is to be chosen – Jul 22 '14 at 14:01
You can tighten the line-matching regex to `/^#.* 2001 /`, but what you've got is strictly within the remit of the question. – Jonathan Leffler Jul 22 '14 at 14:01
Make it fail proof. must uncomment only if it is commented. – Jul 22 '14 at 14:02
Oh, you're the op. So what's your actual file content? – Avinash Raj Jul 22 '14 at 14:04

score 6 · Answer 3 · edited May 23 '17 at 12:10

To complement @Avinash Raj's helpful answer with a more generic, POSIX-compliant solution.

Toggles commenting of lines that match a specifiable string that must occur as a separate word anywhere on the line.
The comment character (string) is also specifiable.

Note that the solution is awk-based, because a robust portable solution with sed is virtually impossible due to the limitations of POSIX' basic regular expressions.

awk -v commentId='#' -v word='2001' '
  $0 ~ "(^|[[:punct:][:space:]])" word "($|[[:punct:][:space:]])" { 
    if (match($0, "^[[:space:]]*" commentId))
      $0 = substr($0, RSTART + RLENGTH)
    else
      $0 = commentId $0
  } 
  { print }
  ' file > tmpfile.$$ && mv tmpfile.$$ file

(^|[[:punct:][:space:]]) and ($|[[:punct:][:space:]]) are the POSIX extended regex equivalents of the \< and \> word-boundary assertions known from other regex dialects.
Whitespace after the comment char is preserved, but not before it.
When prepending the comment char to a line, it is directly prepended, without whitespace.
Thus, if you only toggle comments with this solution, all whitespace is preserved.
POSIX awk doesn't offer in-place updating (neither does POSIX sed, incidentally), hence the output is first captured in a temporary file and that file then replaces the original on success.

score 2 · Answer 4 · answered Aug 13 '19 at 15:49

Quick example of how to comment and uncomment a line in a file.

Sample file :

umask 027
TMOUT=600

Lets now backup the file (just for laughs) and comment out and un comment:

# backup file (because we should always do this)
cp /etc/bash.bashrc /etc/bash.bashrc.$(date '+%Y-%m-%d,%H:%M:%S')

# original: TMOUT=600   , result :# TMOUT=600
sed -i '/[^#]/ s/\(^TMOUT=600.*$\)/#\ \1/' /etc/bash.bashrc

# original # TMOUT=600   ,result :TMOUT=600
sed -i '/^#.*TMOUT=600.*$/s/^#\ //' /etc/bash.bashrc

score 0 · Answer 5 · answered Jun 12 '19 at 09:15

0

For mac, which doesn't support standard sed parameters, this would remove the hashtag

sed -i "" "/.*#.*d\/docker-php-ext-xdebug\.ini.*/s/^#//g" docker-compose.yml

answered Jun 12 '19 at 09:15

max4ever

11,909
13
77
115

How to UNCOMMENT a line that contains a specific string using Sed?

5 Answers5

Linked