Find

find is one of those commands that can be quite arcane. Hopefully this article should clear things up with a brief list of basic uses.

Find Expressions

By File Type

Basic usage:

# find . -type [f - file|d - directory|l - links]

To find only normal files:

# find . -type f

To find both files and directories:

# find . -type f -o -type d

By filename

Basic usage:

# find . -name <term>     # name  - case sensitive
# find . -iname <term>    # iname - case insensitive name

The term you pass in must match the entire file name. You can use * as a wild card.

For regex matching, use -regex. The default regex used by find is emacs-regex. To define something else (like posix), use -regextype.

Examples

To find a file named exactly 'confession.txt':

# find . -name confession.txt -type f

To find a directory named 'backup' or 'Backup':

# find . -iname backup -type d

To find a file with the format NNN-NNNN.txt where N is an integer:

# find . -regextype posix-egrep -regex ".?\/?[0-9]{3}-[0-9]{4}.txt"

By time (created, access, modified)

Basic usage:

# find . -[a - accessed|c - created|m - modified][time|min] <time>

where time is in days when used with time, or minutes when used with min.

To find files modified within 3 minutes ago:

# find . -mmin -3

To find files created within 1 day ago:

# find . -cmin -1

By Ownership (groups or user)

Basic usage:

# find . -[user|group] <name or id>

1. find is able to find file owners by their username or user ID or by group name or group ID.

To find all files owned by leo:

# find . -user leo

Operators

You can mix and match the expressions discussed above with operators.

Operator	Description	Example	Notes
( )	Force precedence	( -type f -o type d ) -a -name "exception.txt"	Both brackets must be surrounded by at least one space. When scripting, you may need to escape the brackets as \( \).
!	not	! -name "exception.txt"	Can also use -not
-a	And	-type f -a -name "confession.txt"	This is the default operator and can be omitted. Can also use -and
-o	Or	-name "confession.txt" -o -name "murder.txt"	Can also use -or

Do things on match

Basic usage:

# find . -exec <command> {} \;

The {} is substituted with the matching filename. The \; at the end denotes the end of the command passed to -exec.

Move all .o files to /tmp/objects

# find . -iname "*.o" -exec mv {} /tmp/{} \;

The move command could be simplified so that the destination location is just a directory name (ie: -exec mv {} /tmp). The example above was done to show that you can use {} can be used more than once.

Convert all .php files into UTF-8 files

The following are two separate commands. Run them sequentially.

# find . -iname "*.php" -exec sh -c 'iconv -f cp1252 -t utf-8 {} > {}.utf8' \;
# for i in `(find . -name "*.utf8")`; do mv $i ${i/.utf8/}; done

Change the file mode for files to 644 and directories to 755

# find . -type f -exec chmod -v 644 {} \;
# find . -type d -exec chmod -v 755 {} \;

Finding bad permissions

You may also find files and directories without the permissions like so:

# find ! -perm 644 -type f -o ! -perm 755 -type d

Remove all broken symlinks

# find . -type l -! -exec test -e {} \; -print | while read i ; do rm  -v $i ; done

Grepping for lines on match

Suppose you want to find all .htaccess files and check if they have an AuthType directive.

# find . -name .htaccess -exec grep -oHne 'AuthType.*$' {} \;

Update file ownership after renumbering an account

After updating an account's uid, you may wish to update all files that were owned with the old uid. Do so in a parallelized fashion by running:

# cd /data/shared_dir
# ls | xargs -L 1 -P 6 -I '{{'  find {{ -user $OLDUID -exec chown $NEWUID:$NEWGID "{}" \;

Parallelizing Jobs

To run multiple jobs across all files found by find, use xargs.

Suppose you wanted to resize all pictures inside a directory using a resize script. To run this across all 8 CPUs:

# find . -iname '*.jpg' -print0 | xargs -0 -P 8 sh resize.sh

Note that we print the file using -print0 which delimits filenames with a null rather than newlines. xargs can then delimit using a null with the -0 flag.

For completeness's sake, here is the resize.sh script:

#!/bin/sh
for i in "$@" ; do
        File=`identify -format '%w %h %i\n' "$i" \
                | awk '$1 > 1920 || $2 > 1920 {sub(/^[^ ]* [^ ]* /, ""); print}' `

        if [ ! -z "$File" ] ; then
                echo "Resizing $File"
                mogrify -resize "1920x1920"  "$File"
        fi
done

Removing old files

Remove all files older than 100 days (such as to clear a /tmp directory).

$ find . -type f -mtime +100 -exec rm -v {} \;

Removing empty files and directories

$ find . -type d -empty -delete