I want to write a script to go into a unix directory and recursively change all the instances of a word in every file in every directory and sub directory.

Here's where I am now:

I can execute
find . -print | xargs grep "my_word"

to recuresively return all the files with my_word in it.

So, now I know which files contain my_word, and now I need to figure out how to write a script and go in and change every occurance of my_word to new_word.

Can anyone help?

sed:

find . -type f |\
while read file
do
      grep -q "my_word" && sed 's/my_word/new_word/g' $file > tmp.tmp && mv tmp.tmp $file
done

along the similar lines:

find . -type f |\
while read file
do
     { rm "${file}"; sed -e '/my_word/s/my_word/new_word/g' > "${file}"; } < "${file}"
done

two diffeent files with same size after zipping giving different file size

Pls don't hijack the threads of the others - start a new thread.

With some versions of grep and xargs:

grep -FZlR old .|xargs -0 perl -i.bck -pe's/old/new/g'

I wanted to thank those who responded to my question. I ended up using a derivation of this solution:

-------------
find . -type f |\
while read file
do
{ rm "${file}"; sed -e '/my_word/s/my_word/new_word/g' > "${file}"; } < "${file}"
done
-----------