Weblog entry #4 for atrixnet
See all weblog entries made by atrixnet.
#4
recursively remove broken symlinks in bash
Posted by atrixnet on Wed 24 Sep 2008 at 16:33
I'm sure there's some better way... but nonetheless, this works great:
for file in $(find . -type l); do if [[ "$(file $file|sed 's/.*broken.*/WOOT/')" == "WOOT" ]]; then echo "unlinking broken symlink '$file'"; unlink $file; fi; done;
How would/do you do it?
for file in $(find . -type l); do if [[ "$(file $file|sed 's/.*broken.*/WOOT/')" == "WOOT" ]]; then echo "unlinking broken symlink '$file'"; unlink $file; fi; done;
How would/do you do it?
Comments on this Entry
From the xutils-dev package (/usr/bin/cleanlinks):
$ whatis cleanlinks
cleanlinks (1) - remove dangling symbolic links and empty directories
$ whatis cleanlinks
cleanlinks (1) - remove dangling symbolic links and empty directories
[ Parent | Reply to this comment ]
To find broken links I use
With some more pipes you should be able to remove them...
( find / -type l 2>/dev/null |&n bsp;xargs file ) | grep 'broken sym bolic link'With some more pipes you should be able to remove them...
[ Parent | Reply to this comment ]
Posted by Anonymous (78.32.xx.xx) on Sat 27 Sep 2008 at 01:00
Totally cheating, but after searching for "exist" in the GNU find man page:
find -L . -type l -print0 | xargs -0 --no-run-if-empty rm
With -L (follow symbolic links), -type l matches against the type of the link's target (recursively following links if it's a symlink to a symlink) - unless that target does not exist, in which case it matches against the link.
(The -print0 and -0 options are specific to GNU find and xargs, and allow you to operate on files with bizarre names, e.g. containing newlines, by using the \0 (NUL) character as the line separator. There are only two characters not allowed in Unix filenames - NUL and '/'.)
If that's too subtle, here's a more long-winded solution:
find . | while read -r FILE; do
if ! test -e "$FILE"; then
rm "$FILE"
fi
done
This takes advantage of the fact that bash's test builtin dereferences symlinks (you could also write this as ! [ test -e "$FILE" ] if you prefer the "[" spelling), so test -e does not believe that a dangling symlink exists.
Both of these will also delete symlinks to (symlinks to...) dangling symlinks. If this is unacceptable, you probably want to do something involving stat(1) and readlink(1) - using file(1) in a script is rarely the right answer!
find -L . -type l -print0 | xargs -0 --no-run-if-empty rm
With -L (follow symbolic links), -type l matches against the type of the link's target (recursively following links if it's a symlink to a symlink) - unless that target does not exist, in which case it matches against the link.
(The -print0 and -0 options are specific to GNU find and xargs, and allow you to operate on files with bizarre names, e.g. containing newlines, by using the \0 (NUL) character as the line separator. There are only two characters not allowed in Unix filenames - NUL and '/'.)
If that's too subtle, here's a more long-winded solution:
find . | while read -r FILE; do
if ! test -e "$FILE"; then
rm "$FILE"
fi
done
This takes advantage of the fact that bash's test builtin dereferences symlinks (you could also write this as ! [ test -e "$FILE" ] if you prefer the "[" spelling), so test -e does not believe that a dangling symlink exists.
Both of these will also delete symlinks to (symlinks to...) dangling symlinks. If this is unacceptable, you probably want to do something involving stat(1) and readlink(1) - using file(1) in a script is rarely the right answer!
[ Parent | Reply to this comment ]