I have to rename a complete folder tree recursively so that no uppercase letter appears anywhere (it's C++ source code, but that shouldn't matter).
Bonus points for ignoring CVS and Subversion version control files/folders. The preferred way would be a shell script, since a shell should be available on any Linux box.
There were some valid arguments about details of the file renaming.
I think files with the same lowercase names should be overwritten; it's the user's problem. When checked out on a case-ignoring file system, it would overwrite the first one with the latter, too.
I would consider A-Z characters and transform them to a-z, everything else is just calling for problems (at least with source code).
The script would be needed to run a build on a Linux system, so I think changes to CVS or Subversion version control files should be omitted. After all, it's just a scratch checkout. Maybe an "export" is more appropriate.
Here's my suboptimal solution, using a Bash shell script:
#!/bin/bash
# First, rename all folders
for f in `find . -depth ! -name CVS -type d`; do
g=`dirname "$f"`/`basename "$f" | tr '[A-Z]' '[a-z]'`
if [ "xxx$f" != "xxx$g" ]; then
echo "Renaming folder $f"
mv -f "$f" "$g"
fi
done
# Now, rename all files
for f in `find . ! -type d`; do
g=`dirname "$f"`/`basename "$f" | tr '[A-Z]' '[a-z]'`
if [ "xxx$f" != "xxx$g" ]; then
echo "Renaming file $f"
mv -f "$f" "$g"
fi
done
Folders are all renamed correctly, and mv
isn't asking questions when permissions don't match, and CVS folders are not renamed (CVS control files inside that folder are still renamed, unfortunately).
Since "find -depth" and "find | sort -r" both return the folder list in a usable order for renaming, I preferred using "-depth" for searching folders.
I needed to do this on a Cygwin setup on Windows 7 and found that I got syntax errors with the suggestions from above that I tried (though I may have missed a working option). However, this solution straight from Ubuntu forums worked out of the can :-)
ls | while read upName; do loName=`echo "${upName}" | tr '[:upper:]' '[:lower:]'`; mv "$upName" "$loName"; done
(NB: I had previously replaced whitespace with underscores using:
for f in *\ *; do mv "$f" "${f// /_}"; done
)
Just simply try the following if you don't need to care about efficiency.
zip -r foo.zip foo/*
unzip -LL foo.zip
One-liner:
for F in K*; do NEWNAME=$(echo "$F" | tr '[:upper:]' '[:lower:]'); mv "$F" "$NEWNAME"; done
Note that this will convert only files/directories starting with letter K, so adjust accordingly.
Lengthy But "Works With No Surprises & No Installations"
This script handles filenames with spaces, quotes, other unusual characters and Unicode, works on case insensitive filesystems and most Unix-y environments that have bash and awk installed (i.e. almost all). It also reports collisions if any (leaving the filename in uppercase) and of course renames both files & directories and works recursively. Finally it's highly adaptable: you can tweak the find command to target the files/dirs you wish and you can tweak awk to do other name manipulations. Note that by "handles Unicode" I mean that it will indeed convert their case (not ignore them like answers that use tr
).
# adapt the following command _IF_ you want to deal with specific files/dirs
find . -depth -mindepth 1 -exec bash -c '
for file do
# adapt the awk command if you wish to rename to something other than lowercase
newname=$(dirname "$file")/$(basename "$file" | awk "{print tolower(\$0)}")
if [ "$file" != "$newname" ] ; then
# the extra step with the temp filename is for case-insensitive filesystems
if [ ! -e "$newname" ] && [ ! -e "$newname.lcrnm.tmp" ] ; then
mv -T "$file" "$newname.lcrnm.tmp" && mv -T "$newname.lcrnm.tmp" "$newname"
else
echo "ERROR: Name already exists: $newname"
fi
fi
done
' sh {} +
References
My script is based on these excellent answers:
https://unix.stackexchange.com/questions/9496/looping-through-files-with-spaces-in-the-names
( find YOURDIR -type d | sort -r;
find yourdir -type f ) |
grep -v /CVS | grep -v /SVN |
while read f; do mv -v $f `echo $f | tr '[A-Z]' '[a-z]'`; done
First rename the directories bottom up sort -r (where -depth is not available), then the files. Then grep -v /CVS instead of find ...-prune because it's simpler. For large directories, for f in ... can overflow some shell buffers. Use find ... | while read to avoid that.
And yes, this will clobber files which differ only in case...
Not portable, Zsh only, but pretty concise.
First, make sure zmv
is loaded.
autoload -U zmv
Also, make sure extendedglob
is on:
setopt extendedglob
Then use:
zmv '(**/)(*)~CVS~**/CVS' '${1}${(L)2}'
To recursively lowercase files and directories where the name is not CVS.
If you use Arch Linux, you can install rename
) package from AUR
that provides the renamexm
command as /usr/bin/renamexm
executable and a manual page along with it.
It is a really powerful tool to quickly rename files and directories.
rename -l Developers.mp3 # or --lowcase
rename -u developers.mp3 # or --upcase, long option
-R --recursive # directory and its children
-t --test # Dry run, output but don't rename
-o --owner # Change file owner as well to user specified
-v --verbose # Output what file is renamed and its new name
-s/str/str2 # Substitute string on pattern
--yes # Confirm all actions
You can fetch the sample Developers.mp3 file from here, if needed ;)
Using Larry Wall's filename fixer:
$op = shift or die $help;
chomp(@ARGV = <STDIN>) unless @ARGV;
for (@ARGV) {
$was = $_;
eval $op;
die $@ if $@;
rename($was,$_) unless $was eq $_;
}
It's as simple as
find | fix 'tr/A-Z/a-z/'
(where fix
is of course the script above)
In OS X, mv -f
shows "same file" error, so I rename twice:
for i in `find . -name "*" -type f |grep -e "[A-Z]"`; do j=`echo $i | tr '[A-Z]' '[a-z]' | sed s/\-1$//`; mv $i $i-1; mv $i-1 $j; done
This works on CentOS/Red Hat Linux or other distributions without the rename
Perl script:
for i in $( ls | grep [A-Z] ); do mv -i "$i" "`echo $i | tr 'A-Z' 'a-z'`"; done
Source: Rename all file names from uppercase to lowercase characters
(In some distributions the default rename
command comes from util-linux, and that is a different, incompatible tool.)
It is not exactly what the OP asked for, but what I was hoping to find on this page:
A "slugify" version for renaming files so they are similar to URLs (i.e. only include alphanumeric, dots, and dashes):
rename "s/[^a-zA-Z0-9\.]+/-/g" filename
Man, you guys/gals like to over complicate things... Use:
rename 'y/A-Z/a-z/' *
The simplest approach I found on Mac OS X was to use the rename package from http://plasmasturm.org/code/rename/:
brew install rename
rename --force --lower-case --nows *
--force Rename even when a file with the destination name already exists.
--lower-case Convert file names to all lower case.
--nows Replace all sequences of whitespace in the filename with single underscore characters.
The original question asked for ignoring SVN and CVS directories, which can be done by adding -prune to the find command. E.g to ignore CVS:
find . -name CVS -prune -o -exec mv '{}' `echo {} | tr '[A-Z]' '[a-z]'` \; -print
[edit] I tried this out, and embedding the lower-case translation inside the find didn't work for reasons I don't actually understand. So, amend this to:
$> cat > tolower
#!/bin/bash
mv $1 `echo $1 | tr '[:upper:]' '[:lower:]'`
^D
$> chmod u+x tolower
$> find . -name CVS -prune -o -exec tolower '{}' \;
Ian
Most of the answers above are dangerous, because they do not deal with names containing odd characters. Your safest bet for this kind of thing is to use find
's -print0
option, which will terminate filenames with ASCII NUL instead of \n.
Here is a script, which only alter files and not directory names so as not to confuse find
:
find . -type f -print0 | xargs -0n 1 bash -c \
's=$(dirname "$0")/$(basename "$0");
d=$(dirname "$0")/$(basename "$0"|tr "[A-Z]" "[a-z]"); mv -f "$s" "$d"'
I tested it, and it works with filenames containing spaces, all kinds of quotes, etc. This is important because if you run, as root, one of those other scripts on a tree that includes the file created by
touch \;\ echo\ hacker::0:0:hacker:\$\'\057\'root:\$\'\057\'bin\$\'\057\'bash
... well guess what ...
This works if you already have or set up the rename command (e.g. through brew install in Mac):
rename --lower-case --force somedir/*
for f in `find -depth`; do mv ${f} ${f,,} ; done
find -depth
prints each file and directory, with a directory's contents printed before the directory itself. ${f,,}
lowercases the file name.
Smaller still I quite like:
rename 'y/A-Z/a-z/' *
On case insensitive filesystems such as OS X's HFS+, you will want to add the -f
flag:
rename -f 'y/A-Z/a-z/' *
This works nicely on macOS too:
ruby -e "Dir['*'].each { |p| File.rename(p, p.downcase) }"
This is a small shell script that does what you requested:
root_directory="${1?-please specify parent directory}"
do_it () {
awk '{ lc= tolower($0); if (lc != $0) print "mv \"" $0 "\" \"" lc "\"" }' | sh
}
# first the folders
find "$root_directory" -depth -type d | do_it
find "$root_directory" ! -type d | do_it
Note the -depth action in the first find.
None of the solutions here worked for me because I was on a system that didn't have access to the perl rename script, plus some of the files included spaces. However, I found a variant that works:
find . -depth -exec sh -c '
t=${0%/*}/$(printf %s "${0##*/}" | tr "[:upper:]" "[:lower:]");
[ "$t" = "$0" ] || mv -i "$0" "$t"
' {} \;
Credit goes to "Gilles 'SO- stop being evil'", see this answer on the similar question "change entire directory tree to lower-case names" on the Unix & Linux StackExchange.
With MacOS,
Install the rename
package,
brew install rename
Use,
find . -iname "*.py" -type f | xargs -I% rename -c -f "%"
This command find all the files with a *.py
extension and converts the filenames to lower case.
`f` - forces a rename
For example,
$ find . -iname "*.py" -type f
./sample/Sample_File.py
./sample_file.py
$ find . -iname "*.py" -type f | xargs -I% rename -c -f "%"
$ find . -iname "*.py" -type f
./sample/sample_file.py
./sample_file.py
find . -depth -name '*[A-Z]*'|sed -n 's/\(.*\/\)\(.*\)/mv -n -v -T \1\2 \1\L\2/p'|sh
I haven't tried the more elaborate scripts mentioned here, but none of the single commandline versions worked for me on my Synology NAS. rename
is not available, and many of the variations of find
fail because it seems to stick to the older name of the already renamed path (eg, if it finds ./FOO
followed by ./FOO/BAR
, renaming ./FOO
to ./foo
will still continue to list ./FOO/BAR
even though that path is no longer valid). Above command worked for me without any issues.
What follows is an explanation of each part of the command:
find . -depth -name '*[A-Z]*'
This will find any file from the current directory (change .
to whatever directory you want to process), using a depth-first search (eg., it will list ./foo/bar
before ./foo
), but only for files that contain an uppercase character. The -name
filter only applies to the base file name, not the full path. So this will list ./FOO/BAR
but not ./FOO/bar
. This is ok, as we don't want to rename ./FOO/bar
. We want to rename ./FOO
though, but that one is listed later on (this is why -depth
is important).
This comand in itself is particularly useful to finding the files that you want to rename in the first place. Use this after the complete rename command to search for files that still haven't been replaced because of file name collisions or errors.
sed -n 's/\(.*\/\)\(.*\)/mv -n -v -T \1\2 \1\L\2/p'
This part reads the files outputted by find
and formats them in a mv
command using a regular expression. The -n
option stops sed
from printing the input, and the p
command in the search-and-replace regex outputs the replaced text.
The regex itself consists of two captures: the part up until the last / (which is the directory of the file), and the filename itself. The directory is left intact, but the filename is transformed to lowercase. So, if find
outputs ./FOO/BAR
, it will become mv -n -v -T ./FOO/BAR ./FOO/bar
. The -n
option of mv
makes sure existing lowercase files are not overwritten. The -v
option makes mv
output every change that it makes (or doesn't make - if ./FOO/bar
already exists, it outputs something like ./FOO/BAR -> ./FOO/BAR
, noting that no change has been made). The -T
is very important here - it treats the target file as a directory. This will make sure that ./FOO/BAR
isn't moved into ./FOO/bar
if that directory happens to exist.
Use this together with find
to generate a list of commands that will be executed (handy to verify what will be done without actually doing it)
sh
This pretty self-explanatory. It routes all the generated mv
commands to the shell interpreter. You can replace it with bash
or any shell of your liking.
Use typeset:
typeset -l new # Always lowercase
find $topPoint | # Not using xargs to make this more readable
while read old
do new="$old" # $new is a lowercase version of $old
mv "$old" "$new" # Quotes for those annoying embedded spaces
done
On Windows, emulations, like Git Bash, may fail because Windows isn't case-sensitive under the hood. For those, add a step that mv
's the file to another name first, like "$old.tmp", and then to $new
.
for f in `find`; do mv -v "$f" "`echo $f | tr '[A-Z]' '[a-z]'`"; done
I would reach for Python in this situation, to avoid optimistically assuming paths without spaces or slashes. I've also found that python2
tends to be installed in more places than rename
.
#!/usr/bin/env python2
import sys, os
def rename_dir(directory):
print('DEBUG: rename('+directory+')')
# Rename current directory if needed
os.rename(directory, directory.lower())
directory = directory.lower()
# Rename children
for fn in os.listdir(directory):
path = os.path.join(directory, fn)
os.rename(path, path.lower())
path = path.lower()
# Rename children within, if this child is a directory
if os.path.isdir(path):
rename_dir(path)
# Run program, using the first argument passed to this Python script as the name of the folder
rename_dir(sys.argv[1])
Source: Stackoverflow.com