Make xargs handle filenames that contain spaces

Question

$ ls *mp3 | xargs mplayer  

Playing Lemon.  
File not found: \'Lemon\'  
Playing Tree.mp3.  
File not found: \'Tree.mp3\'  

Exiting... (End of file)  

Answer 1

ls | grep mp3 | sed -n "7p" | xargs -i mplayer {}

Note that in the command above, xargs will call mplayer anew for each file. This may be undesirable for mplayer, but may be okay for other targets.

Answer 2

Given the specific title of this post, here's my suggestion:

ls | grep ' ' | tr ' ' '<' | sed 's|<|\\ |g'

The idea is to convert blanks to any unique character, like '<', and then change that into '\ ', a backslash followed by a blank. You can then pipe that into any command you like, such as:

ls | grep ' ' | tr ' ' '<' | sed 's|<|\\ |g' | xargs -L1 GetFileInfo

The key here lies in the 'tr' and 'sed' commands; and you can use any character besides '<', such as '?' or even a tab-character.

Answer 3

The xargs utility reads space, tab, newline and end-of-file delimited strings from the standard input and executes utility with the strings as arguments.

You want to avoid using space as a delimiter. This can be done by changing the delimiter for xargs. According to the manual:

 -0      Change xargs to expect NUL (``\0'') characters as separators,
         instead of spaces and newlines.  This is expected to be used in
         concert with the -print0 function in find(1).

Such as:

 find . -name "*.mp3" -print0 | xargs -0 mplayer

To answer the question about playing the seventh mp3; it is simpler to run

 mplayer "$(ls *.mp3 | sed -n 7p)"

Answer 4

Alternative solutions can be helpful...

You can also add a null character to the end of your lines using Perl, then use the -0 option in xargs. Unlike the xargs -d '\n' (in approved answer) - this works everywhere, including OS X.

For example, to recursively list (execute, move, etc.) MPEG3 files which may contain spaces or other funny characters - I'd use:

find . | grep \.mp3 | perl -ne 'chop; print "$_\0"' | xargs -0  ls

(Note: For filtering, I prefer the easier-to-remember "| grep" syntax to "find's" --name arguments.)

Answer 5

It depends on (a) how attached you are to the number 7 as opposed to, say, Lemons, and (b) whether any of your file names contain newlines (and whether you're willing to rename them if they do).

There are many ways to deal with it, but some of them are:

mplayer Lemon*.mp3

find . -name 'Lemon*.mp3' -exec mplayer {} ';'

i=0
for mp3 in *.mp3
do
    i=$((i+1))
    [ $i = 7 ] && mplayer "$mp3"
done

for mp3 in *.mp3
do
    case "$mp3" in
    (Lemon*) mplayer "$mp3";;
    esac
done

i=0
find . -name *.mp3 |
while read mp3
do
    i=$((i+1))
    [ $i = 7 ] && mplayer "$mp3"
done

The read loop doesn't work if file names contain newlines; the others work correctly even with newlines in the names (let alone spaces). For my money, if you have file names containing a newline, you should rename the file without the newline. Using the double quotes around the file name is key to the loops working correctly.

If you have GNU find and GNU xargs (or FreeBSD (*BSD?), or Mac OS X), you can also use the -print0 and -0 options, as in:

find . -name 'Lemon*.mp3' -print0 | xargs -0 mplayer

This works regardless of the contents of the name (the only two characters that cannot appear in a file name are slash and NUL, and the slash causes no problems in a file path, so using NUL as the name delimiter covers everything). However, if you need to filter out the first 6 entries, you need a program that handles 'lines' ended by NUL instead of newline...and I'm not sure there are any.

The first is by far the simplest for the specific case on hand; however, it may not generalize to cover your other scenarios that you've not yet listed.

Answer 6

I know that I'm not answering the xargs question directly but it's worth mentioning find's -exec option.

Given the following file system:

[root@localhost bokeh]# tree --charset assci bands
bands
|-- Dream\ Theater
|-- King's\ X
|-- Megadeth
`-- Rush

0 directories, 4 files

The find command can be made to handle the space in Dream Theater and King's X. So, to find the drummers of each band using grep:

[root@localhost]# find bands/ -type f -exec grep Drums {} +
bands/Dream Theater:Drums:Mike Mangini
bands/Rush:Drums: Neil Peart
bands/King's X:Drums:Jerry Gaskill
bands/Megadeth:Drums:Dirk Verbeuren

In the -exec option {} stands for the filename including path. Note that you don't have to escape it or put it in quotes.

The difference between -exec's terminators (+ and \;) is that + groups as many file names that it can onto one command line. Whereas \; will execute the command for each file name.

So, find bands/ -type f -exec grep Drums {} + will result in:

grep Drums "bands/Dream Theater" "bands/Rush" "bands/King's X" "bands/Megadeth"

and find bands/ -type f -exec grep Drums {} \; will result in:

grep Drums "bands/Dream Theater"
grep Drums "bands/Rush"
grep Drums "bands/King's X"
grep Drums "bands/Megadeth"

In the case of grep this has the side effect of either printing the filename or not.

[root@localhost bokeh]# find bands/ -type f -exec grep Drums {} \;
Drums:Mike Mangini
Drums: Neil Peart
Drums:Jerry Gaskill
Drums:Dirk Verbeuren

[root@localhost bokeh]# find bands/ -type f -exec grep Drums {} +
bands/Dream Theater:Drums:Mike Mangini
bands/Rush:Drums: Neil Peart
bands/King's X:Drums:Jerry Gaskill
bands/Megadeth:Drums:Dirk Verbeuren

Of course, grep's options -h and -H will control whether or not the filename is printed regardless of how grep is called.

---

xargs

xargs can also control how man files are on the command line.

xargs by default groups all the arguments onto one line. In order to do the same thing that -exec \; does use xargs -l. Note that the -t option tells xargs to print the command before executing it.

[root@localhost bokeh]# find ./bands -type f  | xargs -d '\n' -l -t grep Drums
grep Drums ./bands/Dream Theater 
Drums:Mike Mangini
grep Drums ./bands/Rush 
Drums: Neil Peart
grep Drums ./bands/King's X 
Drums:Jerry Gaskill
grep Drums ./bands/Megadeth 
Drums:Dirk Verbeuren

See that the -l option tells xargs to execute grep for every filename.

Versus the default (i.e. no -l option):

[root@localhost bokeh]# find ./bands -type f  | xargs -d '\n'  -t grep Drums
grep Drums ./bands/Dream Theater ./bands/Rush ./bands/King's X ./bands/Megadeth 
./bands/Dream Theater:Drums:Mike Mangini
./bands/Rush:Drums: Neil Peart
./bands/King's X:Drums:Jerry Gaskill
./bands/Megadeth:Drums:Dirk Verbeuren

xargs has better control on how many files can be on the command line. Give the -l option the max number of files per command.

[root@localhost bokeh]# find ./bands -type f  | xargs -d '\n'  -l2 -t grep Drums
grep Drums ./bands/Dream Theater ./bands/Rush 
./bands/Dream Theater:Drums:Mike Mangini
./bands/Rush:Drums: Neil Peart
grep Drums ./bands/King's X ./bands/Megadeth 
./bands/King's X:Drums:Jerry Gaskill
./bands/Megadeth:Drums:Dirk Verbeuren
[root@localhost bokeh]# 

See that grep was executed with two filenames because of -l2.