cli helper. Docs

Docs. find

Bill Zimmerly (bill@zimmerly.com), Freelance Writer and Knowledge Engineer, Author

28 Mar 2006

Explore the vast terrain of the UNIX® file system with the find command. One of the most powerful and useful commands in the UNIX programmer's repertoire is find. All flavors of UNIX have file systems that can contain thousands of files of many different types. With so many choices, locating a specific file, or set of files, can be difficult. The find command makes this task easier in many ways.

There's nothing quite like the thrill of exploring, discovering new people, places, and things. The territory might change, but a few principles remain the same. One of those principles is to keep a written record of your journey; another is to know and use your tools.

The UNIX® operating system is much like a vast, uncharted wilderness. As you travel the terrain, you can pick up tools that assist you later. The find command is such a tool. The find command is capable of much more than simply locating files; it can automatically execute sequences of other UNIX commands, using the filenames found for input, as this article explains.

Find with few limits

All operating systems worth their salt have a tool to assist you in finding things. Unlike most of these tools, the UNIX find command can automatically perform many operations for you on the files it finds.

Standard find tools found in graphical user interfaces (GUIs) allow you to do a few common tasks with the files you find: You can mark them for cutting, copying, and pasting; you can move them to a new location; and you can open them with the program used to create them. These operations involve two or more steps and aren't automatic -- you find the files first, and then you use the GUI to mark them for the next operation. This approach is fine for many users, but the explorer wants more.

The UNIX find command can delete, copy, move, and execute files that it finds. In addition, with the -exec parameter, it can automatically run files through any sequence of UNIX commands you need. It can even ask you before it performs such operations on any file.

Simplify management of your file system

The UNIX find command, like most UNIX commands, has an intimidating array of options and switches that can discourage people from learning its depth -- but true explorers aren't intimidated just because the territory is vast. A good general principle goes a long way toward simplifying a complex topic. Start up an xterm, and try the following command:

$ find . -name *.gif -exec ls {} \;

The -exec parameter holds the real power. When a file is found that matches the search criteria, the -exec parameter defines what to do with the file. This example tells the computer to:

1. Search from the current directory on down, using the dot (.) just after find.
2. Locate all files that have a name ending in .gif (graphic files).
3. List all found files, using the ls command.

The -exec parameter requires further scrutiny. When a filename is found that matches the search criteria, the find command executes the ls {} string, substituting the filename and path for the {} text. If saturn.gif was found in the search, find would execute this command:

$ ls ./gif_files/space/solar_system/saturn.gif

The rest of the article builds on this general principle: Thoughtful use of the find command can make the management of UNIX file systems a much easier task. For example, the find command can process commands based on the type of file system where the file is found, if you use the -fstype parameter. And it's often useful to have the find command prompt you before it executes commands on a found file; you can tell it to do so by using the -ok parameter, as you'll see next.

Optional execution

An important alternative to the -exec parameter is -ok; it behaves the same as -exec, but it prompts you to see if you want to run the command on that file. Suppose you want to remove most of the .txt files in your home directory, but you wish to do it on a file-by-file basis. Delete operations like the UNIX rm command are dangerous, because it's possible to inadvertently delete files that are important when they're found by an automated process like find; you might want to scrutinize all the files the system finds before removing them.

The following command lists all the .txt files in your home directory. To delete the files, you must enter Y or y when the find command prompts you for action by listing the filename:

$ find $HOME/. -name *.txt -ok rm {} \;

Each file found is listed, and the system pauses for you to enter Y or y. If you press the Enter key, the system won't delete the file. Listing 1 shows some sample results:

< rm ... /home/bill/./.kde/share/apps/karm/karmdata.txt > ?
< rm ... /home/bill/./archives/LDDS.txt > ?
< rm ... /home/bill/./www/txt/textfile1.txt > ?
< rm ... /home/bill/./www/txt/faq.txt > ?
< rm ... /home/bill/./www/programs/MIKE.txt > ?
< rm ... /home/bill/./www/programs/EESTRING.txt > ?
.
.
.

After each question mark, the system paused; in this case, the Enter key was pressed to continue to the next file. (No files were removed.) The -ok parameter lets you control the automatic processing of each found file, adding a measure of safety to the danger of automatic file removal.

If too many files are involved for you to spend time with the -ok parameter, a good rule of thumb is to run the find command with -exec to list the files that would be deleted; then, after examining the list to be sure no important files will be deleted, run the command again, replacing ls with rm.

Both -exec and -ok are useful, and you must decide which works best for you in your current situation. Remember, safety first!

Use find creatively

You can perform myriad tasks with the find command. This section provides some examples of ways you can put find to work as you manage your file system.

To keep things simple, these examples avoid -exec commands that involve the piping of output from one command to another. However, you're free to use commands like these in a find's -exec clause.

Clean out temporary files

You can use find to clean directories and subdirectories of the temporary files generated during normal use, thereby saving disk space. To do so, use the following command:

$ find . \( -name a.out -o -name '*.o' -o -name 'core' \) -exec rm {} \;

File masks identifying the file types to be removed are located between the parentheses; each file mask is preceded by -name. This list can be extended to include any temporary file types you can come up with that need to be cleaned off the system. In the course of compiling and linking code, programmers and their tools generate file types like those shown in the example: a.out, *.o, and core. Other users have similar commonly generated temporary files and can edit the command accordingly, using file masks like *.tmp, *.junk, and so on. You might also find it useful to put the command into a script called clean, which you can execute whenever you need to clean a directory.

Copy a directory's contents

The find command lets you copy the entire contents of a directory while preserving the permissions, times, and ownership of every file and subdirectory. To do so, combine find and the cpio command, like this:

$ cd /path/to/source/dir

$ find . | cpio -pdumv /path/to/destination/dir

The cpio command is a copy command designed to copy files into and out of a cpio or tar archive, automatically preserving permissions, times, and ownership of files and subdirectories.

List the first lines of text files

Some people use the first line of every text file as a heading or description of the file's contents. A report that lists the filenames and first line of each text file can make sifting through several hundred text files a lot easier. The following command lists the first line in every text file in your home directory in a report, ready to be examined at your leisure with the less command:

$ find $HOME/. -name *.txt -exec head -n 1 -v {} \; > report.txt

$ less < report.txt

Maintain LOG and TMP file storage spaces

To maintain LOG and TMP file storage space for applications that generate a lot of these files, you can put the following commands into a cron job that runs daily:

$ find $LOGDIR -type d -mtime +0 -exec compress -r {} \;

$ find $LOGDIR -type d -mtime +5 -exec rm -f {} \;

The first command runs all the directories (-type d) found in the $LOGDIR directory wherein a file's data has been modified within the last 24 hours (-mtime +0) and compresses them (compress -r {}) to save disk space. The second command deletes them (rm -f {}) if they are more than a work-week old (-mtime +5), to increase the free space on the disk. In this way, the cron job automatically keeps the directories for a window of time that you specify.

Copy complex directory trees

If you want to copy complex directory trees from one machine to another while preserving copy permissions and the User ID and Group ID (UID and GID -- numbers used by the operating system to mark files for ownership purposes), and leaving user files alone, find and cpio once again come to the rescue:

$ cd /source/directory

$ find . -depth -print | cpio -o -O /target/directory

Find links that point to nothing

To find links that point to nothing, use the perl interpreter with find, like this:

$ find / -type l -print | perl -nle '-e || print';

This command starts at the topmost directory (/) and lists all links (-type l -print) that the perl interpreter determines point to nothing (-nle '-e || print') -- see the Resources section for more information regarding this tip from the Unix Guru Universe site. You can further pipe the output through the rm -f {} functionality if you want to delete the files. Perl is, of course, one of the many powerful interpretive language tools also found in most UNIX toolkits.

Locate and rename unprintable directories

It's possible in UNIX for an errant or malicious program to create a directory with unprintable characters. Locating and renaming these directories makes it easier to examine and remove them. To do so, you first include the -i switch of ls to get the directory's inode number. Then, use find to turn the inode number into a filename that can be renamed with the mv command:

$ ls -ail

$ find . -inum 211028 -exec mv {} newname.dir \;

List zero-length files

To list all zero-length files, use this command:

$ find . -empty -exec ls {} \;

After finding empty files, you might choose to delete them by replacing the ls command with the rm command.

Clearly, your use of the UNIX find command is limited only by your knowledge and creativity.

Conclusion

Exploring the vast terrain of the UNIX file system is easy with the find command. It is well worth your time to experiment with this command and see what works for you. As shown in the examples listed in this article, you can use find in many creative ways to make the exploration and management of your file system terrain easy, and fun.

Resources

• In Commands Reference, Volume 2, find Command, learn more about the find command specifically for the IBM® AIX® environment.
  
  
• The Unix Guru Universe (UGU) site is an excellent resource for UNIX.
  
  
• Read UHACC -- Presentations: using find for tips about how to speed up your searches with the find command by using the many options available.
  
  
• Linux.com | CLI Magic: GNU find describes the basics of using the find command. Note: CLI stands for Command Line Interface -- the anti-GUI!
  
  
• The author of Use free software within commercial UNIX (developerWorks, February 2006) points out how it's possible for UNIX commands, such as find or tar, to be functionally different between commercial versions. Read about how he advocates installing the GNU tools on your commercial version of UNIX for compatibility.