getting error "/bin/sed: -e expression #1, char 13: extra characters after command"

novicunix · 11-04-2014, 01:57 AM

Hi,
I am trying to delete a line from a file. This line contains '#' and '/' like below.
#element /abc/xyz/mtn...
When I am using a sed to delete the command, I am getting extra characters after command.
The command I used is as below.
/bin/sed -i '/#element /abc/xyz/mtn.../d' <file>

I tried using a different delimiter as below. The error is gone but the line is not deleted.
sed -i '/$$x$/d' <file> ( Here x="#element /abc/xyz/mtn...")

Please help me out
Thanks

NevemTeve · 11-04-2014, 03:53 AM

Please us [code] and [/code] tags:
[code]

Code:

sed '/^#element \/abc\/xyz\/mtn...$/d'

[/code]

novicunix · 11-04-2014, 05:05 AM

Hi,
Thanks for the reply. But I am using the string '#element /abc/xyz/mtn...' as variable and using that variable in the sed command.
What can I do in that case?

danielbmartin · 11-04-2014, 06:23 AM

Help us to help you. Provide a sample input file (10-15 lines will do). Construct a sample output file which corresponds to your sample input and post both samples here. With "Before and After" examples we can better understand your needs and also judge if our proposed solution fills those needs.

Daniel B. Martin

NevemTeve · 11-04-2014, 09:02 AM

I wouldn't use sed for this. How about awk?

Code:

FILTER='#element /abc/xyz/mtn...'
awk -vfilter="$FILTER" '{ if (filter != $0) print $0; }'

NevemTeve · 11-04-2014, 09:03 AM

I wouldn't use sed for this. How about awk?

Code:

FILTER='#element /abc/xyz/mtn...'
awk -vfilter="$FILTER" '{ if (filter != $0) print $0; }'

danielbmartin · 11-04-2014, 09:17 AM

Quote:

Originally Posted by NevemTeve

I wouldn't use sed for this. How about awk?

Also worth considering: grep -v.

Daniel B. Martin

NevemTeve · 11-04-2014, 09:43 AM

(The problem is that the 'filter' part might contain grep meta-characters)

ntubski · 11-04-2014, 11:30 AM

Quote:

Originally Posted by NevemTeve

(The problem is that the 'filter' part might contain grep meta-characters)

grep -Fv

NevemTeve · 11-04-2014, 11:40 AM

(The problem is that the whole line should match as in ^filter$)

ntubski · 11-04-2014, 01:08 PM

Quote:

Originally Posted by NevemTeve

(The problem is that the whole line should match as in ^filter$)

grep -Fxv

rknichols · 11-04-2014, 01:35 PM

Quote:

Originally Posted by novicunix

Hi,
Thanks for the reply. But I am using the string '#element /abc/xyz/mtn...' as variable and using that variable in the sed command.
What can I do in that case?

You need to use a delimiter character that can never appear in the string. For example, if you know that the string can never contain a semicolon:

Code:

x="#element /abc/xyz/mtn..."
/bin/sed -i ";$x;d" <file>

If there is no character that you can rule out, then you would have to modify the search string to escape each instance of that character. For example, using "/" as the delimiter:

Code:

x="#element /abc/xyz/mtn..."
xmod="$(echo "$x" | sed 's;/;\\/;g')"
sed -i "/$xmod/d" <file>

You still have to worry about characters such as "." in your example that are special in a regular expression. You can escape those at the same time you are escaping the delimiter character:

Code:

xmod="$(echo "$x" | sed 's;[[/^$.*\\];\\&;g')"

(I think that covers all the cases needed for a basic regular expression.)