can SED/AWK take variable as Pattern?

[rogergili]

I´m trying to find a line and the following one with a pattern i´ve got in a variable...

i´ve tried with sed and awk, see for example the following code would return the line i want and the next one...

sed -n '/phone/,+1 p' file.txt

i would need something like ---- sed -n '/$myvariable/,+1 p' file.txt ----- but the problem is because of the inverted commas ' ' i cannot use $ character to indicate i want to search for the value of the variable...

how could i do it?? i would really really appreciate your help! i´m kindda desperate!!!

Many thanks to you all!

[cheapscotchron]

Try changing the single quotes to double quotes. That should work.
e.g. sed -n "$myvariable p" file.txt \

[rogergili]

I've been trying to change for double quotes in different ways, but that's all I get:

: sed -n "$fulltime p" $filename
No addresses allowed: 08:04:22.712 p

any suggestions?

[drl]

Hi.

An illustration:

#!/bin/bash -

# @(#) s1       Demonstrate double quotes around variable, bash, sed.

echo
echo "(Versions displayed with local utility \"version\")"
version >/dev/null 2>&1 && version "=o" $(_eat $0 $1) sed
set -o nounset
echo

FILE=${1-data1}
cat > $FILE <<EOF
first line
second line
a phone number is 555-1234
next line
last line
EOF

echo " Data file $FILE:"
cat $FILE

echo
echo " Results:"
myvariable="phone"
sed -n "/$myvariable/,+1 p" $FILE

exit 0

Producing:

% ./s1

(Versions displayed with local utility "version")
Linux 2.6.11-x1
GNU bash 2.05b.0
sed GNU sed version 4.1.2

 Data file data1:
first line
second line
a phone number is 555-1234
next line
last line

 Results:
a phone number is 555-1234
next line

Experiment and see man pages for details ... cheers, drl

[rogergili]

hi dlr and cheapscotchron,

i'm still getting:
sed: command garbled: /08:04:22.712/,+1 p

i think it's because i'm using a SunOS 5.10 that doesn't have the GNU included...

[cheapscotchron]

Should work under solaris.
Can you post your entire script? Are you setting the shell properly?
e.g. Notice the first line of drl's script...

#!/bin/bash

This invokes the borne (again) shell.

FYI.. this should also work under kshell (ksh). Syntax would be slightly different for cshell (csh).

[rogergili]

It could be a matter of invoking... as I'm not sure what to do at all!

Guys, many thanks for all, but I managed to do the same without using awk/sed as it's been a nightmare... If you're interested you've got it here:

tail +$((`grep -n $fulltime $filename | cut -d: -f1`)) $filename | head -n 2

that takes the patter $fulltime and searches it in $filename and then returns the first match of that and the following line...

I really appreciate your help! Thank you!

[drl]

Hi.

#!/bin/bash -

# @(#) s1       Demonstrate double quotes around variable, bash, sed.

echo
echo "(Versions displayed with local utility \"version\")"
version >/dev/null 2>&1 && version "=o" $(_eat $0 $1) sed nawk
set -o nounset
echo

FILE=${1-data1}
cat > $FILE <<EOF
first line
second line
a phone number is 555-1234
next line
last line
EOF

echo " Data file $FILE:"
cat $FILE

echo
echo " Results, nawk on Solaris:"
myvariable="phone"
nawk -v "v=$myvariable" '
BEGIN   { print "v = " v }
$0 ~ v  { print ; outside = NR + 1; next }
NR == outside   { print ; exit }
' $FILE

echo
echo " sed on Solaris:"
sed -n "/$myvariable/{p
n
p
}
" $FILE

exit 0

Producing:

$ ./s1

(Versions displayed with local utility "version")
SunOS 5.10
GNU bash 3.00.16
sed - ( /usr/xpg4/bin/sed Aug 9 2005 )
nawk - ( /usr/bin/nawk Jan 8 2007 )

 Data file data1:
first line
second line
a phone number is 555-1234
next line
last line

 Results, nawk on Solaris:
v = phone
a phone number is 555-1234
next line

 sed on Solaris:
a phone number is 555-1234
next line

See man pages, etc. ... cheers, drl

( Edits 1,2: corrected sequence on sed Solaris )