Deleting a PARTICULAR word from a file using shell script - shell

Question

Hi I am having a problem in deleting a particular set of words from a file using Shell script. Here goes my problem,
My file: group.dat
Sample lines
ADMIN
ADMINISTRATION
ADMINISTRATOR
My Script: groupdel.sh
#!/bin/sh
groupf="<pathtofile>/group.dat"
tmp="<pathtofile>/te"
delgrp()
{
echo "Enter the group to be deleted"
read gname
echo "-------------------"
for gn in `cat $groupf`
do
if [ "$gname" = "$gn" ]
then
sed -e "s/$gname//g" $groupf > $tmp&&mv $tmp $groupf
echo "deleted group"
cat $groupf
exit 1
fi
done
}
echo "Welcome to Group delete wizard"
delgrp
Output:
Enter the group to be deleted
ADMIN
deleted group
ISTRATION
ISTRATOR
Problem: My problem is I dont want the script to delete ADMINISTRATION or ADMINISTRATOR but to delete only ADMIN, any help how to achieve it.
Thanks in Advance

Answer 1

#!/bin/sh
groupf="<pathtofile>/group.dat"
tmp="<pathtofile>/te"
delgrp()
{
echo "Enter the group to be deleted"
read gname
echo "-------------------"
sed -e "/^$gname[[:blank:]]/d" "$groupf" > "$tmp" && mv "$tmp" "$groupf"
echo "deleted group $gname"
cat "$groupf"
return 0
}
echo "Welcome to Group delete wizard"
delgrp
Assuming that the group name is at the beginning of the line and there are other things on the line and you want to delete the whole line, use the regular expression and command as shown.
There's no need for a loop since sed will iterate over the lines of the file for free.
You should return from a function rather than exit from it. Zero means success. One indicates an error or failure.
Always quote variable names that contain filenames.

Answer 2

If the file is one group per line, and the group name is the only thing on the line, use anchors in your regular expression:
s/^$gname:.*//g
If you have Perl installed, you can probably simplify this a bit with something like this:
if grep -q "^${gname}:" $groupf ; then
perl -ni -e "print unless /^${gname}:/" $groupf
echo "Group deleted."
else
echo "No such group $gname."
fi
Or even
grep -v "^${gname}:" $groupf > $tmp && \
cp -f $tmp $groupf && rm -f $tmp
which will copy all lines except the matching one to the temporary file, and then copy the tempfile over the original file, replacing it.
Note that I suggest using a cp rather than a mv in order to retain the permissions of the original file; mv will result in the edited file having permissions set according to your umask with no concern for the original permissions.
So, for the complete script:
#!/bin/sh
groupf="<pathtofile>/group.dat"
tmp="<pathtofile>/te"
delgrp()
{
echo -n "Enter the group to be deleted: "
read gname
echo "-------------------"
if grep -q "^${gname}:" $groupf ; then
grep -v "^${gname}:" $groupf > $tmp
cp -f $tmp $groupf
rm -f $tmp
else
echo "No such group '$gname'"
fi
}
echo "Welcome to Group delete wizard"
delgrp
That should work reliably.

Answer 3

You can use \W to denote the start and end of a word, if they are separated properly:
sed -e "s/\(\W\)$gname\(\W\)/\1\2/g" $groupf > $tmp&&mv $tmp $groupf

Answer 4

Awk is a readable alternative to sed:
awk -v to_delete="$gname" -F: '$1 == to_delete {next} {print}'

Answer 5

Why you don't use sed ?
sed 's/^word$//g'
Also you can use regex to specify multiple words
sed 's/word1|word2//g'
I didn't try this, but this is what you need. Just take a look on Internet on the sed syntax.
Regards