Bulk rename files in bash

You’ve got a directory full of files with spaces and weird characters in the names. Maybe they came from a Windows box. Maybe someone named them like they were writing a Victorian novel. Either way, you need them cleaned up before you can use them in a script without losing your mind to quote escaping hell.

Here’s the thing: shell filenames with spaces are the enemy of automation. They break loops, they break pipes, they break your soul at 2 AM when a script silently fails because for file in * split on the space instead of the newline.

Let’s fix it.

Replace spaces with underscores

The simplest case. You want my file.txt to become my_file.txt.

Using find with rename:

find /tmp/ -depth -name "* *" -execdir rename 's/ /_/g' "{}" \;

This finds all files with spaces in /tmp/, then uses the rename command (Perl-based, available on most Linux systems) to replace each space with an underscore. The -depth flag makes sure subdirectories are processed before their parent, so renaming doesn’t mess up the directory tree.

If you prefer a shell loop instead:

for file in *; do
  mv "$file" "${file// /_}"
done

That ${file// /_} is bash parameter expansion—replace all spaces (the first //) with underscores (the second part after the slash). Dead simple, no external commands needed.

Replace spaces with hyphens

Some people prefer hyphens to underscores. Both work; pick your poison.

Using find:

find /tmp/ -depth -name "* *" -execdir rename 's/ /-/g' "{}" \;

With a shell loop:

for file in *; do
  mv "$file" "${file// /-}"
done

Same deal—parameter expansion, just with a hyphen instead of an underscore.

Remove spaces completely

Nuke them entirely.

find /tmp/ -depth -name "* *" -execdir rename 's/ //g' "{}" \;

Be careful with this one. It’ll mangle things like my file.txt into myfile.txt, which is fine, but if you’ve got my - file.txt it becomes my-file.txt… actually that works out. Still, test on a copy first.

Remove special characters

Sometimes it’s not just spaces—you’ve got apostrophes, parentheses, hyphens in weird places, all kinds of nonsense.

Here’s a loop that strips out punctuation entirely:

for file in *.mp3; do
  mv "$file" "${file//[[:punct:]]/}"
done

The [[:punct:]] character class matches anything that’s not alphanumeric or whitespace. WARNING: This will also eat the dot in your file extension (.mp3 becomes mp3). You probably don’t want that. Better approach:

for file in *.mp3; do
  name="${file%.*}"           # filename without extension
  ext="${file##*.}"           # just the extension
  clean="${name//[[:punct:]]/}" # strip punctuation from name only
  mv "$file" "${clean}.${ext}"
done

Now you’re only stripping punctuation from the filename part, leaving the extension alone.

Use rename for find-and-replace patterns

The rename command is a Perl-based tool that’s incredibly powerful for bulk renaming. It’s much safer than building shell loops because it does a dry-run by default (with the -n flag) so you can see what it’s going to do before it does it.

Remove underscores:

rename -n 's/_//g' *

(Add -n to see what it’ll do without actually renaming. Remove -n to commit the changes.)

Remove a specific substring:

rename 's/2012mp3/2012/' *.mp3

Replace a dash in the middle of a filename:

rename 's/1234/12-34/' *.mp3

Remove the first N characters:

for file in *.mp3; do
  temp="${file:1}"  # skip first character (use 2 for two chars, etc.)
  mv "$file" "$temp"
done

Using perl for advanced patterns

If you need something more powerful, Perl gives you the full regex arsenal:

ls *.bed | perl -ne 'use File::Copy; chomp; $old=$_; s/\s+/_/g; move($old,$_);'

This reads filenames from ls, replaces any whitespace (including multiple spaces) with a single underscore, and moves the file. Perl’s s/\s+/_/g is more flexible than bash’s ${var// /} because \s+ matches any consecutive whitespace, not just single spaces.

The safe way: always test first

Before you unleash a batch rename on a production directory:

# Create a test directory
mkdir test_dir
cp my_files/* test_dir/

# Run your rename command there
cd test_dir
for file in *; do
  mv "$file" "${file// /_}"
done

# Look at the results
ls -la

# If it looks good, do it for real
cd ..
for file in my_files/*; do
  mv "$file" "${file// /_}"
done

It’s tedious, but it beats explaining to your team why all the filenames got corrupted.

Pick your tool

rename command: Best for complex patterns, has a dry-run mode (-n), works recursively.
Bash parameter expansion (${var//old/new}): Fastest, no dependencies, good for simple replacements.
find + rename: Best for deep directory trees, the -depth flag prevents path issues.
Perl: When you need regex superpowers and character classes like \s+ or [[:punct:]].