Question

I´m building a bash script and i want to replace “${PRODUCT_NAME}” with a “WORD” for example.
The problem is “$”, in bash $ stands for an argument so when i run the script

echo "Replace ${PRODUCT_NAME} with $1 in document.txt"

for file in $(grep -il ${PRODUCT_NAME} ~/Documents/tmp/document.txt)   
do   
   sed -e "s/${PRODUCT_NAME}/$3/g" $file > /tmp/tempfile.tmp   
   mv /tmp/tempfile.tmp $file   
done   

the script recognizes the ${PRODUCT_NAME} as an argument and not as a word that i want to replace. How do i make it recognize that “${PRODUCT_NAME}” is a word and not an argument, so i can replace it?

Answer 1

First of all, note that your sed script actually needs to use \${PRODUCT_NAME} rather than ${PRODUCT_NAME} as the search-pattern, because in a POSIX Basic Regular Expression, $ means “end of line” rather than “dollar sign”. That out of the way . . .

In Bash, you can either quote a dollar-sign with a backslash:

echo "Replace \${PRODUCT_NAME} with $1 in document.txt"

for file in $(grep -il ${PRODUCT_NAME} ~/Documents/tmp/document.txt)
do
   sed -e "s/\\\${PRODUCT_NAME}/$3/g" $file > /tmp/tempfile.tmp
   mv /tmp/tempfile.tmp $file
done

or else use single-quotes (which are like double-quotes except that they don’t allow $-based expansion, and don’t handle \):

echo 'Replace ${PRODUCT_NAME} with '"$1 in document.txt"

for file in $(grep -il ${PRODUCT_NAME} ~/Documents/tmp/document.txt)
do
   sed -e 's/\${PRODUCT_NAME}/'"$3/g" $file > /tmp/tempfile.tmp
   mv /tmp/tempfile.tmp $file
done

(By the way, is it intentional that your echo line uses $1 while your sed line uses $3? It seems like they should both use the same parameter.)