tempa=$(echo $y | sed 's/&/_/g')

here, I am trying to replace all instances of the character "&" with "_"

I've tried:
tempa=$(echo $y | sed 's/\&/_/g')

and

tempa=$(echo $y | sed 's/"&"/_/g')

and a few other random shots in the dark. I can't seem to escape the &

Thanks in advance.

paul

Personally, I found I had to escape the & just to be able to assign it, after that it worked:

t1=xx\&yy
echo $t1|sed 's/&/_/'
xx_yy
echo $t1
xx&yy

If I didn't escape it during the assignment this happens:

t1=xx&yy
[1] 9893
bash: yy: command not found
[1]+ Done t1=xx

HTH

thanks so much....

and my appologies for maybe not explaining this well enough

I am getting my variable $y from files in a directory... like this:

cd /somedirectory
for * in y

so I'm not sure how to escape the character in the declaration...

where I am looping through all files in somedirectory that may be poorly named. I am using this to easily replace spaces with underscores, but I cannot replace the "&" symbol.

What I have are filenames that are like this:

somefile name
apples & oranges

and I want to use sed to rename those files:

somefilename
apples_and_oranges

thanks again,

Paul

This might do for the & part...

yes, that's what I'm doing.

The problem is, the "&" is special to the sed function and I can't escape it. For any other character, what you suggest works. See this:

# echo "bubb&les" | sed -e 's/&/_/g'
bubb_les

# echo "bubb&les" | sed -e 's/b/_/g'
_u__&les

so it just ignores the "&"

...yea, still stuck on this one.

By your first example, sed is removing the & just fine. In the second example, you only asked it to remove the b so the & is not going to be removed in that case.

So you want to replace spaces and '&' chars: to do what you want, you need a separate sed cmd for each char you want to replace becuase you are replacing them with different replacement chars.
Like this:

t1='apples & oranges'
t2=`echo $t1|sed 's/&/and/'`
echo $t2
apples and oranges
t3=`echo $t2|sed 's/ /_/g'`
echo $t3
apples_and_oranges

thanks to everyone for the help.

here's one for someone else who wants to do this later. Could be done more efficently I'm sure, but this seems to work great.

This bash script replaces spaces with underscores and removes the '&' character from directories and files recursively from the current directory down.

Example:
# find ./
./blah & blah1
./blah & blah1/blah & blah1
./blah & blah1/blah & blah1/blah & blah4
./blah & blah1/blah & blah1/blah & blah3
./blah & blah1/blah & blah1/blah & blah2
./blah & blah1/blah & blah1/blah & blah2/bo & bo1
./blah & blah1/blah & blah1/blah & blah2/bo & bo2
./blah & blah1/blah & blah1/blah & blah2/bo & bo3
./blah & blah1/blah & blah1/blah & blah1
./blah & blah1/blah & blah2
./blah & blah1/blah & blah3
./blah & blah1/blah & blah4
# sr
# ls
blah_blah1
# find ./
./blah_blah1
./blah_blah1/blah_blah2
./blah_blah1/blah_blah3
./blah_blah1/blah_blah4
./blah_blah1/blah_blah1
./blah_blah1/blah_blah1/blah_blah4
./blah_blah1/blah_blah1/blah_blah3
./blah_blah1/blah_blah1/blah_blah2
./blah_blah1/blah_blah1/blah_blah2/bo_bo2
./blah_blah1/blah_blah1/blah_blah2/bo_bo3
./blah_blah1/blah_blah1/blah_blah2/bo_bo1
./blah_blah1/blah_blah1/blah_blah1


the code:

#!/bin/sh
for i in *
do
if [ -d "$i" ] # if * is a directory
then
cd "$i" # descend into the directory
for y in *
do
# tempa=$(echo $y | sed 's/ /_/g')
tempa=$(echo $y | sed 's/ /_/g' | sed -e 's/\&//g' | sed 's/__/_/g')
if [ "$y" != "$tempa" ]
then
mv "$y" "$tempa"
fi
if [ -d "$tempa" ] # if this is also a directory, call this program
then
cd "$tempa"
sr; # this is the name of THIS program, must be in your PATH
cd ..
fi
done
cd ..
fi
# tempa=$(echo $i | sed 's/ /_/g')
tempa=$(echo $i | sed 's/ /_/g' | sed -e 's/\&//g' | sed 's/__/_/g')
if [ "$i" != "$tempa" ]
then
mv "$i" "$tempa"
fi
done

-Paul Rennix