I suspect my search/rename task might have been beyond the capabilities of this combination of commands – or at least it was beyond the capabilities of me! I was looking to replace HTML quote characters (') with the literal ‘ symbol throughout files in a hierarchy of directories. This caused me NO END of issues. It seemed the major problems came from the fact that my filenames had both whitespace and literal quotes in them – which cause issues for xargs. The literal quote I was using in my sed string might have been an issue as well, but I think that escaping that made it OK.

I fussed around for half an hour trying to find a combination of escape characters and command line arguments that would make things work but eventually I gave up and had to go looking for a more dummy-friendly solution. I found a GUI program called regexxer that was excellent – well worth checking out for people who can’t get the command-line solution to perform for them.

No real reason other than I like to use grep

That’s one of many great things about *NIX systems – there is always more than one way to solve a problem.

Thanks for your suggestion.

grep -lr -e ” * | xargs sed -i ‘s///g’

Why not:

find -type f | xargs sed -i ‘s///g’

This saves both grep and sed parsing the file. Pass all files to sed, and let it parse it and replace.

This one is meant to be integrated in Thunar file manager. But will also work on any other machine that has a desktop environment installed.

http://www.barrydegraaff.tk/files/Linux/scripts/thunar_scripts/search_and_replace.sh

usage:
chmod +x search_and_replace.sh
search_and_replace.sh /path/to/folder

Requires zenity, gxmessage, xargs, grep and sed installed.

I consider myself a bash newbie, so any enhancements are welcome, please notify me at
http://www.barrydegraaff.tk/index.php?page=_content/contact.html

Happy coding!

Barry

The character you put after ‘s’ (s being substitute command) determines delimiter you want to use. So you can always specify delimiter that is not contained in your search string to avoid errors, and provide better readability compared to escaping forward slashes. In your case to search for http://www.example.com/oldpage.html and replace it with http://www.example.com/newpage.html you could use ‘_’ as separator:

sed ‘s_http://www.example.com/oldpage.html_http://www.example.com/newpage.html_g’

grep -lr -e ” *.txt | xargs sed -i ‘s/plums/apples/g’ -R

but no luck.

Any help would be appreciated.

You would have to escape the / forward slashes with back slashes \ like so:

grep -lr -e 'apples' *.html | xargs sed -i 's/http:\/\/www.example.com\//http:\/\/www.64bitjungle.com\//g'

Or if you’re just using sed on a single file:

sed -i 's/http:\/\/www.example.com\//http:\/\/www.64bitjungle.com\//g' filename.txt

Hope that helps

in your example it says apples/oranges with / being the delimiter but an url has many //’s

how would i go about writing a log of what was edited

grep ‘searchString’ DirectoryToSearch –include=*.py -nr