I was recently put in charge of archiving alot of files onto my linux system, however the majority of the files have spaces in their names. This is causing my perl scripts to give me back tons of warning messages in regards to pattern matching errors.

I'd like to replace files with spaces in them with a '_'

Example: abbies file.txt --> abbies_file.txt

I'm hoping this is the cause of my perl scripts gagging.

However, I don't know how to do this...been looking around but I'm alittle nervous about wiping everything on accident. Any help?

Thanks!

Use quotes " for example:

mv "abbies file.txt" abbies_file.txt

I would love to but I have about 2000 files that I have to do this with ... that would take way too long! Thanks though!

Look for a recursive option.

Run it howevermany times as there are spaces in a single file name. If the maximum number of spaces you have in a filename is 5, run it 5 times.

Assuming that all the files are in one subdirectory, try:

---

#!/bin/bash

for FILE in `echo *.txt`
do
NEWFILE=`echo ${FILE} | sed 's/ /_/g'`
mv "${FILE}" ${NEWFILE}
done

---

(In case you have trouble reading it, the characters that surround the command that defines NEWFILE are backticks and the sed command is in single quotes.)

Hope this helps...

Rick

That wont work.

the 'for FILE in word' construct would break the filename apart because the shell interprets the space character as a delimiter.

You know... after I hit submit I was thinking that I should have made that loop begin with:

ls -1 *.txt | while read FILE

The rest of the script would be unchanged.

Nice catch. Thanks.

Rick

Sweet thanks! I'll give that a try first thing on monday! Can't believe the spaces in all these file names...windows users....

Thanks Again!! I'll let you know how I turn out!

Thanks for the help! Here's the script I ended using:

#!/bin/sh

find . -name '* *' | sort | while read FILE
do
NEWFILE=`echo ${FILE} | sed 's/ /_/g;`
mv "${FILE}" ${NEWFILE}
done

Seems to work great!

Thanks again!

NEWFILE=`echo ${FILE} | sed 's/ /_/g;`
^

I had to insert a (') single quote between the ( semicolon and (`) back-tick to make this work on my redhat 7.3 system. I am still getting errors for a few files, but I have not investigated these yet. Probably has something to do with (') single quotes or parenthesis in the file names.

Yep, that line as posted has a syntax error in it.
Heh, heh. You're renaming mp3 files aren't you. (Ran into this myself after downloading a bunch of mp3s from the NEARfest web site. There were only a dozen or so I did it by hand rather than fool around figuring out how to replace spaces as well as all the parentheses.)

*grin* yes, some of the files in questions were mp3/music files. I really want to force samba to rename these files as they are attempted to be written, but for now I am using above script in a cron job. I ended up doing the others by hand as well after an hour of not quite getting it last night after posting my reply!

thanks for the help though, I'm coming from the winworld, and my shell scripting (and use of certain utils like sed/awk) aren't quite up to par with my dos batch and vb scripting!

-Dave

This script works if all files are in the same directory. What if you have directories and files that are deeper?

ex: /this has spaces/this one too/file with spaces.doc

If you run this script above, it will change "this has spaces", and then when it gets to "this one too" the path will have changed to "this_has_spaces/this one too" and it cannot access that directory or file.

Any ideas?

This is tough, so far I'm looking at having to count the /'s to determine how deep the directory structure is and then minus a count of one to determine the end directory. I may be overshooting but i'll post an answer unless someone gets it before I do.

Ain't scripting fun? ;P