Remove a list of words from filename

Question

I am trying to remove a list of specific words from all my files with a certain directory and replace them with nothing.

So:

This Awesome Content 720p BLAH FOO BANG OOO - 30.9.2013.mp4

Becomes:

This Awesome Content - 30.9.2013.mp4

Now the following works great for a single find and replace one word.

find path/to/folder/ -maxdepth 3 -name '*.*' -execdir bash -c 'mv -i "$1" "${1//foo/}"' bash {} \;

I have also tried just multiple finds but this seems like a very long way of doing it and i seem to run into issues this way.

Couple of issues i have:

• Want it to be case insensitive
• Need "${1//foo/}" to refer to a list
• Remove white spaces if greater than 1

Trying to run this as a bash script on a cronjob.

Unless there is a better to way to remove everything between "This Awesome Content" - "30.9.2013.mp4".

Much appreciated.

Answer 1

You can access a file's name as a variable using the 'echo' command. Once that is done, the most powerful way to make your desired changes is to use 'sed'. You can string together sed commands with the '-e' flag. As part of a for loop in bash, this line gives you a start. You can also use a line like this as part of your 'find' statement.

echo $fyle | sed -e 's/FOO//gI' -e 's/BANG//gI'

Once you have your desired file names, you can then move them back to the original name. Let me know if you need more specific instructions.

UPDATE: Here is a more complete solution. You'll have to tune the script to your own file names, etc.

for fyle in $(find . -name "*.*")
do 
   mv -i $fyle `echo $fyle | sed -e 's/FOO//gI' -e 's/BANG//gI' `
done

Finally, to replace more than one whitespace character with one whitespace character, you can add another sed command. Here is a working command:

echo "file    input.txt" | sed 's/  */ /g'

Answer 2

One way to do this will be to add an intermediate step where you generate a file with mv commands to achieve this, and then execute that file. I am assuming you have a words_file file containing words you don't want.

cd to the folder before starting

# Create list of valid <file>s in file_list, and list of "mv <file> " commmands
# in cmd_file
ls | grep -f words_file | tee file_list | sed 's/\(.*\)/mv "\1" /g' > cmd_file

# Create the sed statements using the words_file, store it to sed_commands
# Then, apply the sed commands to file_list
sed 's/\(.*\)/s\/\1\/\/g/g' words_file > sed_commands
sed -f sed_commands file_list > new_file_names

# Combine cmd_file and new_file_names to produce the full mv statements
paste cmd_file new_file_names > final_cmds

# To verify the commands
cat final_cmds

# Finally, execute it
sh final_cmds

This is what I could come up with, it avoids manually writing sed -e for each word. Not sure if there is a simpler way using common bash utilities. Of course, you can use perl or python and write it more concisely.

Edit: Simplified it, did away with eval and xargs.