Like humans, most computers tend to gain weight as they age. OK, they don't really gain weight. But they do tend to gather stuff on their hard drives that consumes expensive disk space--space that could be put to better use. Worse, that trash fills up your backup media, consuming time and materials there, too.

While many tomes have been written about disk space recovery on an iSeries, there is also an open-source tool available that can help you hunt down and manage files on your Integrated File System (IFS)--whether they are used as part of your OS/400 software, as part of PASE, or as part of a Linux partition.

Find or Find File

Most of you have used a Windows machine and are aware of the Windows "Find File" utility, a handy tool to find misplaced files over local or network drives. Thus, when you move to a UNIX-like OS, you may overlook one of the most powerful tools available, the Find utility, because you think it's just a command-line version of the utility with which you are so familiar. Bzzzzt! Wrong!

Even if you have been using Linux for some time, you still may not have an appreciation for the power and elegance embodied in the Find utility. A quick look at the feature list shows that Find can locate files 1) by name or partial name, 2) by size or a range of sizes, 3) by date and time of last access, last modification, or last change, 4) by comparison with the date/time of another file, 5) by owner, 6) by file permission, 7) by file type (directory, link, etc.), or 8) by any combination of the preceding. And once Find locates a file matching your criteria, it can execute a command against it, making the appropriate file name substitutions as required.

Here's the format of the Find command:

Here, path is the directory from which Find begins it search, and expression is a group of operators passed to Find.

So that we can more easily discuss Find, I have created a bogus directory below, adding files and a directory to it. The files with the .java extension are, of course, Java programs. Files with .c or .cpp are C and C++ programs, respectively. And the lone file with the .sh extension is a shell script.

Let's start with a very simple example in which we want to produce a listing of all Java programs. The command and the results are shown below:

The first parameter to the Find command is the path (or paths) to search. In this case, we're in the directory where we want to begin the search. Thus, we'll use a period (.) to denote the current directory. After the directory come the operators. The first (-name) specifies the file(s) that we want to find. The second operator causes the file name currently under consideration to be printed. Notice that there is a Java program in the dir1 directory that Find located. This is because Find will traverse your directory tree, starting at the point you indicated.

Let's try the same request, this time swapping the order of the operators:

Huh? What happened here? This anomaly deserves an explanation. The operators to Find return logical values and are divided into three categories:

• Options, which affect how Find behaves and always return a true value
• Tests, which can return true or false
• Actions, the logical value of which depends on exit status of the action.

As it executes, Find traverses the directory (or directories) and for each file it finds, evaluates from left to right the expression (the list of operators) that you provide. It will continue the evaluation as long as it gets a true return value (indicating that the list of operators is logically ANDed together). Once a false value is returned, Find will cease evaluation and go on to the next file. In our successful example, the "-print" operator was executed only when the test "-name" returned a true value. Thus, only files matching our request were printed. If we put the "-print" operator before the test, it is always executed, producing the erroneous results.

Let's try a slightly more complicated example. This time, we want to find both Java and C++ programs. Here are the command and results:

As with any logical expression, you can use parentheses to force the order of evaluation. In this case, we would need the parentheses so that "-print" will be executed whenever we find either a Java or C++ program.

You may be wondering what the backslashes are for. No, it wasn't transmission line noise injected when I sent this article to MC Press. Instead, the backslashes are used to let the shell know that you want the parentheses passed on to the command instead of being used by the shell itself. Failure to so mark the parentheses (called "escaping") results in the following error:

Now that you have an idea how Find evaluates expressions, let's look at a series of simple examples that show it in use.

First, let's get a list of all files (-type f) that are greater than 512 K in size (-size +512k), followed by a list of files smaller than 512 K (-size -512k), followed by a list of files exactly 512 K. We'll use the "-ls" operator instead of the "-print" operator to indicate that we would like the output in the same format as given by the ls -dils command. The "-type f" is used to indicate only regular files, excluding things such as directories and symbolic links.

The difference between the comparisons was in the use (or lack thereof) of the plus sign (+) and the minus sign (-). The former means anything greater than, the latter means anything less than, and the absence of either means equal to.

What if we want a list of files that were modified over a year ago and are over 100 K in size?

The Find utility would be handy enough if all it did was produce listings (because you could use the output of Find as input to another command, as you'll see shortly). One of the operators available to Find is "-exec," which executes a command. Substituting "-print" for "-exec" in our first example yields the following:

The "-exec" operator is one whose logical value is based upon the return value of the command(s) executed. If the command was successful, then "-exec" will be true and Find will continue to evaluate the expression. Otherwise, the evaluation will stop. Once again, the backslash character is used--this time to escape the semicolon. Find considers anything between "-exec" and a semicolon to be the command string to execute. You need to convince the shell to pass the semicolon to Find, hence the need to escape it.

Using Echo within "-exec" is rather boring. What if we wanted to do something more utilitarian? How about if we save disk space by using the gzip utility on every Java source file that hasn't been modified for over a year? (The gzip utility is a file compressor that, by default, compresses a file and renames it with an extension of "gz.")

Take special note of the fact that the "-exec" operator is called each time a matching file is found. Sometimes, you want to deal with the complete list of files as a group. To do that, pipe the output to the command you want to use to operate on the list. For example, the tape archiver (tar) program allows you to save a list of files in a single file called an archive. In the previous example, we simply "gzipped" each matching file. What if we wanted to archive the files instead? That's simple:

The xargs utility builds and executes command lines from standard input. So the output of the Find command is reformed into this command:

Most of the time, you'll be better off using this form for efficiency's sake. Instead of spawning a process for each matching file, you'll only spawn one additional process for the list. Use "-exec" if you are interested in the outcome of the command you execute to effect the operation of Find or if the list you generate exceeds the maximum number of parameters that the command you wish to execute can handle.

Seek and Ye Shall Find

I've only touched on the simplest of uses for the Find utility. There are many operands available, but they vary by operating system. You'll need to check your documentation to determine what your vendor has added to the standard list. The GNU version included with RedHat 9 has 39 tests and a dozen supported actions. So the possibilities are endless.

The Find utility is supported both within the QSHELL and PASE environments on the iSeries, as well as on the various flavors of UNIX (and, of course, on Linux). So you can add it to your disk-cleaning arsenal on any of the platforms on which it is available.

Until next month, explore your IFS or Linux directories and see what you can find!

Barry L. Kline is a consultant and has been developing software on various DEC and IBM midrange platforms for over 20 years. Barry discovered Linux back in the days when it was necessary to download diskette images and source code from the Internet. Since then, he has installed Linux on hundreds of machines, where it functions as servers and workstations in iSeries and Windows networks. He recently co-authored the book Understanding Linux Web Hosting with Don Denoncourt. Barry can be reached at This email address is being protected from spambots. You need JavaScript enabled to view it..